Handbook of Elemental Speciation: Techniques and...

Handbook of Elemental Speciation:Techniques and Methodology

Handbook of Elemental Speciation: Techniques and Methodology R. Cornelis, H. Crews, J. Caruso and K. Heumann 2003 John Wiley & Sons, Ltd ISBN: 0-471-49214-0

Handbook of Elemental Speciation:Techniques and Methodology

Editor-in-chief

Rita CornelisGhent University, Belgium

Associate Editors

Joe CarusoUniversity of Cincinnati, USA

Helen CrewsCentral Science Laboratory, UK

Klaus HeumannJohannes Gutenberg-University Mainz, Germany

Copyright 2003 John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester,West Sussex PO19 8SQ, England

Telephone (+44) 1243 779777

Email (for orders and customer service enquiries): [email protected] our Home Page on www.wileyeurope.com or www.wiley.com

All Rights Reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form orby any means, electronic, mechanical, photocopying, recording, scanning or otherwise, except under the terms of the Copyright,Designs and Patents Act 1988 or under the terms of a licence issued by the Copyright Licensing Agency Ltd, 90 TottenhamCourt Road, London W1T 4LP, UK, without the permission in writing of the Publisher. Requests to the Publisher should beaddressed to the Permissions Department, John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex PO198SQ, England, or emailed to [email protected], or faxed to (+44) 1243 770620.

This publication is designed to provide accurate and authoritative information in regard to the subject matter covered. It is soldon the understanding that the Publisher is not engaged in rendering professional services. If professional advice or other expertassistance is required, the services of a competent professional should be sought.

Other Wiley Editorial Offices

John Wiley & Sons Inc., 111 River Street, Hoboken, NJ 07030, USA

Jossey-Bass, 989 Market Street, San Francisco, CA 94103-1741, USA

Wiley-VCH Verlag GmbH, Boschstr. 12, D-69469 Weinheim, Germany

John Wiley & Sons Australia Ltd, 33 Park Road, Milton, Queensland 4064, Australia

John Wiley & Sons (Asia) Pte Ltd, 2 Clementi Loop #02-01, Jin Xing Distripark, Singapore 129809

John Wiley & Sons Canada Ltd, 22 Worcester Road, Etobicoke, Ontario, Canada M9W 1L1

Wiley also publishes its books in a variety of electronic formats. Some content that appearsin print may not be available in electronic books.

Library of Congress Cataloging-in-Publication Data

Handbook of elemental speciation : techniques and methodology / edited by Rita Cornelis. . . [et al.].

p. cm.Includes bibliographical references and index.ISBN 0-471-49214-0

1. Speciation (Chemistry) – Technique. 2. Speciation (Chemistry) – Methodology. 3.Environmental chemistry – Technique. 4. Environmental chemistry – Methodology. I.Cornelis, Rita.

QD75.3 .H36 2003544–dc21

2002193376

British Library Cataloguing in Publication Data

A catalogue record for this book is available from the British Library

ISBN 0-471-49214-0

Typeset in 10/12pt Times by Laserwords Private Limited, Chennai, IndiaPrinted and bound in Great Britain by Antony Rowe Ltd, Chippenham, WiltshireThis book is printed on acid-free paper responsibly manufactured from sustainable forestryin which at least two trees are planted for each one used for paper production.

Contents

List of Contributors . . . . . . . . . . . . . . . vii

Preface . . . . . . . . . . . . . . . . . . . . . . . . . ix

Technical Abbreviationsand Acronyms . . . . . . . . . . . . . . . . . . . . xi

1 Introduction . . . . . . . . . . . . . . . . . . . . . 1

2 Sampling: Collection, Storage . . . . . . . 72.1 Sampling: Collection, Processing

and Storage of EnvironmentalSamples . . . . . . . . . . . . . . . . . . . 7

2.2 Sampling of Clinical Samples:Collection and Storage . . . . . . . . . 23

2.3 Food: Sampling with SpecialReference to Legislation,Uncertainty and Fitness forPurpose . . . . . . . . . . . . . . . . . . . . 47

2.4 Sampling: Collection,Storage – Occupational Health . . . 59

3 Sample Preparation . . . . . . . . . . . . . . . 733.1 Sample Treatment for Speciation

Analysis in Biological Samples . . 733.2 Sample Preparation Techniques for

Elemental Speciation Studies . . . . 953.3 Sample Preparation – Fractionation

(Sediments, Soils, Aerosols andFly Ashes) . . . . . . . . . . . . . . . . . 119

4 Separation Techniques . . . . . . . . . . . . . 1474.1 Liquid Chromatography . . . . . . . . 147

4.2 Gas Chromatography and OtherGas Based Methods . . . . . . . . . . . 163

4.3 Capillary Electrophoresis inSpeciation Analysis . . . . . . . . . . . 201

4.4 Gel Electrophoresis for SpeciationPurposes . . . . . . . . . . . . . . . . . . . 224

5 Detection . . . . . . . . . . . . . . . . . . . . . . 2415.1 Atomic Absorption and Atomic

Emission Spectrometry . . . . . . . . 2415.2 Flow Injection Atomic

Spectrometry for Speciation . . . . . 2615.3 Detection by ICP-Mass

Spectrometry . . . . . . . . . . . . . . . . 2815.4 Plasma Source Time-of-flight

Mass Spectrometry: a PowerfulTool for Elemental Speciation . . . 313

5.5 Glow Discharge Plasmas asTunable Sources for ElementalSpeciation . . . . . . . . . . . . . . . . . . 334

5.6 Electrospray Methods forElemental Speciation . . . . . . . . . . 356

5.7 Elemental Speciation byInductively Coupled Plasma-MassSpectrometry with HighResolution Instruments . . . . . . . . 378

5.8 On-line Elemental Speciation withFunctionalised Fused SilicaCapillaries in Combination withDIN-ICP-MS . . . . . . . . . . . . . . . . 417

5.9 Speciation Analysis byElectrochemical Methods . . . . . . . 427

5.10 Future Instrumental Developmentfor Speciation . . . . . . . . . . . . . . . 461

vi CONTENTS

5.11 Biosensors for Monitoring ofMetal Ions . . . . . . . . . . . . . . . . . 471

5.12 Possibilities Offered byRadiotracers for MethodDevelopment in ElementalSpeciation Analysis and forMetabolic and EnvironmentallyRelated Speciation Studies . . . . . . 484

6 Direct Speciation of Solids . . . . . . . . . 5056.1 Characterization of Individual

Aerosol Particles with SpecialReference to SpeciationTechniques . . . . . . . . . . . . . . . . . 505

6.2 Direct Speciation of Solids: X-rayAbsorption Fine StructureSpectroscopy for Species Analysisin Solid Samples . . . . . . . . . . . . . 526

7 Calibration . . . . . . . . . . . . . . . . . . . . . 5477.1 Calibration in Elemental

Speciation Analysis . . . . . . . . . . . 5477.2 Reference Materials . . . . . . . . . . . 563

8 Screening Methods for Semi-quantitative Speciation Analysis . . . . . . 591

9 Risk Assessments/Regulations . . . . . . . 6059.1 Environmental Risk Assessment

and the Bioavailability ofElemental Species . . . . . . . . . . . . 605

9.2 Speciation and Legislation . . . . . . 629

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . 635

List of Contributors

F. R. Abou-ShakraMicromass UK Ltd, Wythenshawe, UK

K. L. AckleyUniversity of Cincinnati, Cincinnati, USA

J. I. G. AlonsoUniversity of Oviedo, Oviedo, Spain

T. BergDanish Veterinary and Food Administration,Søborg, Denmark

J. BettmerJohannes Gutenberg-University, Mainz, Germany

I. BontideanLund University, Lund, Sweden

B. BouyssiereUniversite de Pau et des Pays de l’Adour, Pau,France

P. BreretonCentral Science Laboratory, York, UK

J. BuffleCABE, Sciences II, Geneva, Switzerland

C. CamaraUniversidad Complutense de Madrid, Madrid,Spain

J. A. CarusoUniversity of Cincinnati, Cincinnati, USA

H. ChassaigneEC, Joint Research Center, Institute for ReferenceMaterials and Measurements, Geel, Belgium

C. C. CheryGhent University, Ghent, Belgium

R. CornelisGhent University, Ghent, Belgium

H. CrewsCentral Science Laboratory, York, UK

E. CsoregiLund University, Lund, Sweden

K. de CremerGhent University, Ghent, Belgium

E. Dabek-ZlotorzynskaEnvironment Canada, Ottawa, Canada

J. H. DuffusThe Edinburgh Centre for Toxicology and theUniversity of Edinburgh, Edinburgh, UK

A. N. EatonMicromass UK Ltd, Wythenshawe, UK

H. EmonsEC, Joint Research Center, Institute of ReferenceMaterials and Measurements, Geel, Belgium

J. R. EncinarUniversity of Oviedo, Oviedo, Spain

viii LIST OF CONTRIBUTORS

M. E. FoulkesUniversity of Plymouth, Plymouth, UK

K. G. HeumannJohannes Gutenberg-University, Mainz, Germany

G. M. HieftjeIndiana University, Bloomington, USA

J. HlavayUniversity of Veszprem, Veszprem, Hungary

R. S. HoukIowa State University, Ames, USA

K. Keppel-JonesEnvironment Canada, Ottawa, Canada

G. KollenspergerUniversitat fur Bodenkultur Wien, Vienna, Austria

A. M. LeachIndiana University, Bloomington, USA

R. LobinskiUniversite de Pau et des Pays de l’Adour, Pau,France

D. M. McClenathanIndiana University, Bloomington, USA

R. K. MarcusClemson University, Clemson, USA

R. MacarthurCentral Science Laboratory, York, UK

B. MichalkeGSF National Research Center for Environmentand Health, Neuherberg, Germany

R. M. OlivasUniversidad Complutense de Madrid, Madrid,Spain

H. M. OrtnerTechnische Universitat Darmstadt, Darmstadt,Germany

K. PolyakUniversity of Veszprem, Veszprem, Hungary

M. Potin-GautierUniversite de Pau et des Pays de l’Adour, Pau,France

P. QuevauvillerEuropean Commission, DG Research, Brussels,Belgium

W. SchuhmannRuhr University Bochum, Bochum, Germany

J. SzpunarUniversite de Pau et des Pays de l’Adour, Pau,France

R. M. TownThe Queen’s University of Belfast, Belfast, North-ern Ireland

J. F. TysonUniversity of Massachusetts, Amherst, USA

F. VanhaeckeGhent University, Ghent, Belgium

E. WelterHASYLAB at DESY, Hamburg, Germany

C. ZhangTsinghua University, Beijing, China

X. ZhangTsinghua University, Beijing, China

Preface

The recognition of the fact that, in environmen-tal chemistry, occupational health, nutrition andmedicine, the chemical, biological and toxicolog-ical properties of an element are critically depen-dent on the form in which the element occurs in thesample has spurred a rapid development of an areaof analytical chemistry referred to as speciationanalysis. In contrast to its biological meaning, theterm speciation in chemistry refers to the distribu-tion of an element among defined chemical species,i.e. among specific forms of an element definedas to isotopic composition, electronic or oxidationstate, and/or complex or molecular structure.

The areas of speciation analysis have beenundergoing a continual evolution and devel-opment for the last 20 years. The area mostfrequently referred to is speciation of anthro-pogenic organometallic compounds and theproducts of their environmental degradation,such as methylmercury, alkyllead, butyl- andphenyltin compounds, and simple organoarsenicand organoselenium species. The presence of ametal(loid)–carbon covalent bond ensures a rea-sonable stability of the analyte(s) during samplepreparation. The volatility of the species allowsthe use of gas chromatography with its inherentadvantages, such as the high separation efficiencyand the absence of the condensed mobile phase,that enable a sensitive (down to the femtogramlevels) element-specific detection by atomic spec-troscopy. Much effort has been devoted by theEuropean Commission Measurement and TestingProgram to raise the standards of accuracy ofspeciation measurements in terms of appropriatecalibration and method validation using certifiedreference materials.

An insight into endogenous metal species inbiological systems has remained for a long timea challenge to the analyst. Indeed, millions ofyears of evolution have resulted in a great vari-ety of biological ligands with different functionsand a significant coordinating potential for traceelements. They include small organic acids, macro-cyclic chelating molecules, and macromolecules,such as proteins, DNA restriction fragments orpolysaccharides. The complexity and the usuallypoor understanding of the system (the majority oftrace element species with biological ligands havenot yet been discovered!) have been the majorobstacles on the way to the identification and char-acterization of the endogenous metal complexeswith biomolecules. Their generally poor volatil-ity in comparison with organometallic species callsfor separation techniques with a condensed mobilephase that negatively affects the separation effi-ciency and the detection limits.

A fundamental tool for speciation analysis hasbeen the combination of a chromatographic sepa-ration technique, which ensures that the analyticalcompound leaves the column unaccompanied byother species of the analyte element, with atomicspectrometry, permitting a sensitive and specificdetection of the target element. Recent impres-sive progress toward lower detection limits inICP MS, toward higher resolution in separa-tion techniques, especially capillary electrophore-sis and electrochromatography, and toward highersensitivity in electrospray mass spectrometry formolecule-specific detection at trace levels in com-plex matrices allows new frontiers to be crossed.Analytical techniques allowing direct speciation insolid samples are appearing.

x PREFACE

Speciation analysis is a rather complex task anda reference handbook on relevant techniques andmethodology has been awaited by all those withan interest in the role and measurement of elementspecies. These expectations are now fulfilled bythe Handbook of Elemental Speciation of whichthe first (of the announced two) volume is nowappearing. This first volume brings a collectionof chapters covering comprehensively differentaspects of procedures for speciation analysis at thedifferent levels starting from sample collection andstorage, through sample preparation approaches torender the species chromatographable, principlesof separation techniques used in speciation analy-sis, to the element-specific detection. This alreadyvery broad coverage of analytical techniques iscompleted by electrochemical methods, biosen-sors for metal ions, radioisotope techniques and

direct solid speciation techniques. Special concernis given to quality assurance and risk assessment,and speciation-relevant legislation.

Although each chapter is a stand-alone refer-ence, covering a given facet of elemental specia-tion analysis written by an expert in a given field,the editorial process has ensured the volume is anexcellent introductory text and reference handbookfor analytical chemists in academia, governmentlaboratories and industry, regulatory managers,biochemists, toxicologists, clinicians, environmen-tal scientists, and students of these disciplines.

Ryszard Lobinski

Pau, France, October 2002

Technical Abbreviations and Acronyms

Abbreviations

AAS atomic absorption spectrometryAC alternating currentAED atomic emission detectionAES atomic emission spectrometryAF atomic fluorescenceANOVA analysis of varianceCE capillary electrophoresisCGC capillary gas chromatographycpm counts per minuteCRM certified reference materialCVAAS cold vapour atomic absorption

spectrometryCZE capillary zone electrophoresisDC direct currentES electrosprayESI electrospray ionizationETAAS electrothermal atomic absorption

spectrometryETV electrothermal vaporizationEXAFS extended X-ray absorption fine

structure spectroscopyFAAS flame atomic absorption

spectrometryFID flame ionization detectorFIR far-infraredFPD flame photometric detectorFT Fourier transformFPLC fast protein liquid chromatographyGC gas chromatographyGD glow dischargeGLC gas–liquid chromatographyGSGD gas-sampling glow dischargeHGAAS hydride generation atomic

absorption spectrometry

HPLC high performance liquidchromatography

Hz HertzICP inductively coupled plasmai.d. internal diameterIEF isoelectric focusingIDMS isotope dilution mass

spectrometryINAA instrumental neutron activation

analysisIR infraredISFET ion-selective field effect transistorISE ion-selective electrodeITP isotachophoresisLA laser ablationLC liquid chromatographyLED light emitting diodeLOD limit of detectionLOQ limit of quantificationMAE microwave-assisted extractionsMIP microwave-induced plasmaMS mass spectrometryNIR near-infraredNMR nuclear magnetic resonanceo.d. outer diameterOES optical emission spectrometryPBS phosphate buffer salinePIXE particle/proton-induced X-ray

emissionQA quality assuranceQC quality controlQF quartz furnaceREE rare earth elementRPC reversed phase chromatographyRSD relative standard deviation

xii TECHNICAL ABBREVIATIONS AND ACRONYMS

SE standard errorSEM scanning electron microscopeSFC supercritical fluid chromatographySFE supercritical fluid extractionSFMS sector-field mass spectrometerSIMS secondary ion mass spectrometrySPME solid-phase micro-extractionTD thermal desorptionTEM transmission electron microscopeTIMS thermal ionization mass

spectrometryTOF time of flightUV ultravioletUV/VIS ultraviolet–visibleXAFS X-ray absorption fine structureXRD X-ray diffractionXRF X-ray fluorescence

Units

µg microgramsng nanogramspg picogramsfg femtogramsmL millilitresL litrescL centilitres

Symbols

M molecular massMr relative molecular massr correlation coefficients standard deviation of sampleσ population standard deviation

CHAPTER 1

Introduction

R. C. CornelisLaboratory for Analytical Chemistry, Ghent University, Belgium

H. M. CrewsCentral Science Laboratory, Sand Hutton, York, UK

J. A. CarusoUniversity of Cincinnati, Ohio, USA

K. G. HeumannInstitut fur Anorganische and Analytische Chemie, Mainz, Germany

1 Definition of Elemental Speciation and ofFractionation . . . . . . . . . . . . . . . . . . . . . . 2

2 Problems to be Solved . . . . . . . . . . . . . . . 2

3 Speciation Strategies . . . . . . . . . . . . . . . . . 34 References . . . . . . . . . . . . . . . . . . . . . . . . 5

‘Speciation’, a word borrowed from the biologicalsciences, has become a concept in analyticalchemistry, expressing the idea that the specificchemical forms of an element should be consideredindividually. The underlying reason for this isthat the characteristics of just one species ofan element may have such a radical impact onliving systems (even at extremely low levels)that the total element concentration becomes oflittle value in determining the impact of thetrace element. Dramatic examples are the speciesof tin and mercury, to name just these two.The inorganic forms of these elements are muchless toxic or even do not show toxic propertiesbut the alkylated forms are highly toxic. No

wonder analytical chemists had to study elementalspeciation and devise analytical techniques thatproduce qualitative and quantitative information onchemical compounds that affect the quality of life.

Before embarking on the definitions of ele-mental speciation and species, it may be inter-esting to give a short historical setting of thisemerging branch of analytical chemistry. Analyt-ical chemistry began as a science in the early19th century. A major milestone was the book byWilhelm Ostwald ‘Die Wissenschaftlichen Grund-lagen der analytischen Chemie’ (Scientific Fun-damentals of Analytical Chemistry) in 1894 [1].A personality who contributed substantially to thedevelopment of analytical chemistry and chemical


2 INTRODUCTION

analysis was Carl Remigus Fresenius. In 1841he published a very interesting book on qualita-tive chemical analysis [2]. It was followed overthe next 100 years by a series of standard workson qualitative and quantitative analysis by sev-eral generations of the Fresenius family and bythe publications by Treadwell [3, 4], Feigl [5] andKolthoff [6, 7], to name just these few. The interestremained largely focused on inorganic analyticalchemistry. The term ‘trace elements’ dates backto the early 20th century, in recognition of thefact that many elements occurred at such low con-centrations that their presence could only just bedetected. During the following 60 years all effortswere focused on total trace element concentrations.Scientists developed methods with increasing sen-sitivity. It was only in the early 1960s that ques-tions were raised concerning the chemical form ofthe trace elements and that the need for an ana-lytical methodology developed subsequently. Thisdevelopment has been growing exponentially tothe point that research on trace element analysistoday appears almost exclusively focused on traceelement species.

Extensive literature is available on the speci-ation and fractionation of elements. Newcomershave to absorb a wealth of highly specialised pub-lications and they miss the broader overview toguide them. This handbook aims to provide all thenecessary background and analytical informationfor the study of the speciation of elements.

The objective of this handbook is to presenta concise, critical, comprehensive and systematic(but not exhaustive), treatment of all aspectsof analytical elemental speciation analysis. Thegeneral level of the handbook makes it most usefulto the newcomers in the field, while it may beprofitably read by the analytical chemist alreadyexperienced in speciation analysis.

1 DEFINITION OF ELEMENTALSPECIATION AND OFFRACTIONATION

The International Union for Pure and AppliedChemistry (IUPAC) has defined elemental speci-ation in chemistry as follows:

(i) Chemical species. Chemical element: specificform of an element defined as to isotopiccomposition, electronic or oxidation state,and/or complex or molecular structure.

(ii) Speciation analysis. Analytical chemistry: ana-lytical activities of identifying and/or measur-ing the quantities of one or more individualchemical species in a sample.

(iii) Speciation of an element; speciation. Distri-bution of an element amongst defined chem-ical species in a system.

When elemental speciation is not feasible, the termfractionation is in use, being defined as follows:

(iv) Fractionation. Process of classification of ananalyte or a group of analytes from acertain sample according to physical (e.g.,size, solubility) or chemical (e.g., bonding,reactivity) properties.

As explained in the IUPAC paper [8], it isoften not possible to determine the concentrationsof the different chemical species that sum up tothe total concentration of an element in a givenmatrix. Often, chemical species present in a givensample are not stable enough to be determinedas such. During the procedure, the partitioning ofthe element among its species may be changed.For example, this can be caused by a change inpH necessitated by the analytical procedure, orby intrinsic properties of measurement methodsthat affect the equilibrium between species. Alsoin many cases the large number of individualspecies (e.g., in metal–humic acid complexes ormetal complexes in biological fluids) will make itimpossible to determine the exact speciation. Thepractice is then to identify various classes of theelemental species.

2 PROBLEMS TO BE SOLVED

While the incentive to embark on speciationand fractionation of elements is expanding, itbecomes more and more evident that the matterhas to be handled with great circumspection.Major questions include: What are the species

SPECIATION STRATEGIES 3

we want to measure? How should we sample thematerial and isolate the species without changingits composition? Can we detect very low amountsof the isolated species, which may represent onlya minute fraction of the total, already ultra-traceelement concentration? How do we calibrate thespecies, many of these not being available ascommercial compounds? How do we validatemethods of elemental analysis? All of thesequestions will be carefully dealt with in the firstvolume of the handbook. The second and thirdvolumes will extensively address elemental speciesof specific elements and the analysis of variousclasses of species.

Advances in instrumentation have been crucialto the development of elemental speciation. Therehas been a very good trend towards lower andlower detection limits in optical atomic spectrom-etry and mass spectrometry. This has allowed thebarrier between total element and element speciesto be crossed. While the limit of detection for manypolluting species is sufficient for their measure-ment in a major share of environmental samples,this is not yet the case for human, animal andperhaps plant samples at ‘background levels’. Thebackground concentration of elemental species ofanthropogenic origin was originally zero. Todaythey are present, because they have been and con-tinue to be distributed in a manner that affects thelife cycle. However, because we cannot measurethem in living systems it does not mean that theirpresence is harmless. At the same time it is alsohighly plausible that a certain background levelof these anthropogenic substances can be toler-ated without any adverse effect. In order to assessthe impact of low background levels of elementspecies we will have to develop separation anddetection techniques that surpass the performanceof the existing speciation methodology.

In the mean time research teams are veryresourceful in developing computer controlledautomated systems for elemental speciation analy-sis. These are or will become tremendous assetsfor routine analyses. Preferably systems shouldbe simple, robust, low cost and if at all pos-sible portable, to allow for fieldwork. Althoughcurrently limited it can be postulated that once

there are more regulatory or economic motives,this technology would develop rapidly. Moreover,sensors will play a major role in rapid detectionof elemental species. Simple screening methodswill be increasingly popular because they can pro-vide speedy and reliable tests to detect elementalspecies and give an estimate of their concentration.

Good laboratory practice and method validationare a must to produce precise and accurate results.To this end, it is evident that provision hasto be made for elemental species data in morecertified reference materials (CRMs) reporting onelemental species. This need will become evenmore acute once legislation becomes specificand cites elemental species instead of the totalconcentration of the element and its compounds,as is presently the general rule. This type oflegislation may be politically charged, becauseevery species carries a different health risk orbenefit. The toxicity may vary by several ordersof magnitude among species of the same element.This may lead to some confusing and dangerousconclusions. A product may be legally acceptableon the basis of total concentration but when thattotal consists of some very toxic species it mayconstitute a real hazard. The opposite may also betrue. This can be exemplified by the occurrence ofarsenic in food. Whereas the total arsenic in somefish derivatives, such as gelatine, often exceedsthe accepted limit, the product should not berejected, because the arsenic is mainly presentas arsenobetaine, a non-toxic arsenic species, asopposed to the toxic inorganic arsenic species.

3 SPECIATION STRATEGIES

‘Strategy’ signifies ‘a careful plan or method’or ‘the art of devising or employing plans orstratagems toward a goal’ [9]. Ideally scientistshope to learn everything about the elementalspecies they study: to start with its composition, itsmass, the bio- and environmental cycle, the stabil-ity of the species, its transformation, and the inter-actions with inert or living matter. This list is notexhaustive. The work involved to achieve this goalis, however, challenging, if not impossible to com-plete. Therefore a choice has to be made to identify

4 INTRODUCTION

the most important issues as elemental speciationstudies are pursued. A first group of compounds tobe studied very closely are those of anthropogenicorigin. Although they fulfil the requirements forwhich they were synthesised, unless they happento be synthetic by-products or waste, their long-term effect on the environment, including livingsystems, has often been ignored or misjudged. Oneof the most striking examples is the group of organ-otin compounds. They have surely proven to be themost effective marine anti-fouling agents, fungi-cides, insecticides, bacteriostats, PVC stabilisingagents, etc. However, the designers of these com-pounds never anticipated what the negative effectwould be on the environment. Their disturbingimpact on the life cycle of crustaceans constitutedthe first alarming event. In the mean time thesecomponents have become detectable, with concen-trations now increasing in fish products and evenin vegetables from certain areas. Little thought wasgiven that organotin compounds would be seriousendocrine disruptors, or that they would have sucha long half-life. Hence they will continue to bea burden on the environment and ultimately onmankind itself into the distant future. The organ-otin compounds are only a minute part of the totalamount of tin (mainly inert tin oxides) to be foundin contaminated areas. Determination of the totaltin concentration would surely not be appropriate.

In speciation studies, a lot of attention must bepaid to the stability. Species stability depends onthe matrix and on physical parameters, such astemperature, humidity, UV light, organic matter,etc. Next comes the isolation and purification of thespecies, the study of the possible transformationthrough the procedure, their characteristics andinteractions. New analytical procedures have to bedevised, including appropriate quantification andcalibration methodologies.

Besides the suspect elemental species of anthro-pogenic origin, there is the barely fathomabledomain of the species that developed along withlife on earth. For many elements nothing is reallyknown, or only a few uncertain facts can be stated.Whereas the total trace element concentration maybe static, the species may be highly dynamic. Theywill change continuously with respect to changes

in the surrounding environment, depending onchemical parameters such as pH value or concen-tration of potential ligands for complex formation,the physiological state of a cell, and state of healthof a living entity. Therefore, thermodynamic butalso kinetic stability of elemental species in theenvironment has to be taken into account. Unsta-ble species in the atmosphere are predominant andthis steady transformation requires special analyt-ical procedures. Although species in living cellscan be stable covalent compounds when the ele-ment forms the core of the molecule (such as Co invitamin B12), most elemental species exhibit verylow stability constants with their ligands. Thesecompounds are, however, very active in reachingthe target organs. This to say that a reliable speci-ation strategy will include stability criteria for thespecies and awareness of possible transformations.

Understanding the fate of the trace elementsin the life cycle is of paramount importance.When, through natural or anthropogenic activities,metal ions enter the environment and the livingsystems, only a small fraction will remain asthe free ion. The major share will be complexedwith either inorganic or organic ligands. Naturalmethylation of metal ions under specific conditionsis prevalent. The new species can be much moretoxic, as is the case with methylated mercury, orless toxic as in the case of arsenic. In the caseof mercury, the concentration of ionic mercuryin water may be very low (a few ng L−1) andthat of methylmercury only 1 % of total Hg.Unfortunately, this accumulates to mg kg−1 levelsin the top predators of the food chain, withmethylmercury making up 90 to 100 % of thetotal Hg concentration. Metal ions will also beincorporated into large molecular structures such ashumic substances. Elucidation of the many, often-labile species will take many more years. Traceelement speciation has become important in allfields of life, and concerns industry, academia,government and legislative bodies [10].

It is obvious that it would have been impossibleto accomplish the stated aims and objectives of thishandbook without the wholehearted cooperationof the distinguished authors who contributed the

REFERENCES 5

various chapters. To them we express our sincereappreciation and gratitude.

4 REFERENCES

1. Ostwald, W., Die Wissenschaftlichen Grundlagen deranalytischen Chemie (Scientific Fundamentals of Analyt-ical Chemistry), Engelmann, Leipzig, 1894 [citation inBaiulescu, G. E., Anal Lett., 33, 571 (2000)], 5th revisededition of the book published by Steinkopff, Dresden,1910.

2. Fresenius, C. R., Anleitung zur qualitativen chemischenAnalyse, 1841, followed by Anleitung zur quantitativenchemischen Analyse oder die Lehre von der Gewichts-bestimmung und Scheidung der in der Pharmacie, denKunsten, Gewerben und der Landwirthschaft haufigervorkommenden Korper in einfachen und zusammeng.Verbindungen (Introduction to the quantitative chemicalanalysis or the instruction of the determination of weightand the separation in pharmacy, arts, paints and fre-quently occurring agricultural compounds in their ele-mentary and correlated compounds), 4th edn. Vieweg,Braunschweig, 1859.

3. Treadwell, F. P., Kurzes Lehrbuch der analytischenChemie – Qualitative Analyse, two volumes, (Concise

Handbook of Analytical Chemistry – Qualitative Analy-sis), Deuticke, Leipzig, 1899.

4. Treadwell, F. P., translated by Hall, W. T., AnalyticalChemistry, Vol. I, Qualitative Analysis, Vol. II, Quanti-tative Analysis , John Wiley & Sons, Inc., New York,Chapman & Hall, London, 1932 and 1935 (original Ger-man text 1903 and 1904).

5. Feigl, F., translated by Oesper, R. E., Chemistry ofSpecific, Selective and Sensitive Reactions , AcademicPress, New York, 1949.

6. Kolthoff, I. M., Elving, P. J. and Sandell, E. B., Part I.Treatise on Analytical Chemistry, Theory and Practise,Vols 1–13, Interscience Publications, John Wiley & Sons,Inc., New York, 1959–1976.

7. Kolthoff, I. M. and Elving, P. J., Part II – Analyticalchemistry of the elements, Analytical chemistry of inor-ganic and organic compounds , Vols 1–17, IntersciencePublications, John Wiley & Sons, Inc., New York,1961–1980.

8. Templeton, D. M., Ariese, F., Cornelis, R., Danielsson,L. -G., Muntau, H., Van Leeuwen, H. P. and Lobin-ski, R., Pure Appl. Chem ., 72, 1453 (2000).

9. Webster’s Third New International Dictionary , 1976.10. Ebdon, L., Pitts, L., Cornelis, R., Crews, H. and Que-

vauviller, P., Trace Element Speciation for Environment,Food and Health , The Royal Society of Chemistry, Cam-bridge, UK, 2001.

CHAPTER 2

Sampling: Collection, Storage

2.1 Sampling: Collection, Processing and Storageof Environmental Samples

Hendrik EmonsEC Joint Research Center IRMM, Geel, Belgium

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 72 General Aspects of Environmental

Sampling . . . . . . . . . . . . . . . . . . . . . . . . . 83 Sampling for Speciation Analysis . . . . . . . 11

3.1 Air . . . . . . . . . . . . . . . . . . . . . . . . . . 113.2 Water . . . . . . . . . . . . . . . . . . . . . . . . 12

3.3 Biological material . . . . . . . . . . . . . . 133.4 Sediment and soil . . . . . . . . . . . . . . . 15

4 Sample Processing . . . . . . . . . . . . . . . . . . 165 Sample Storage . . . . . . . . . . . . . . . . . . . . 196 Further Challenges . . . . . . . . . . . . . . . . . . 217 References . . . . . . . . . . . . . . . . . . . . . . . . 21

1 INTRODUCTION

Public awareness and scientific understanding ofthe various compartments, processes and prob-lems in the environment have been significantlyimproved particularly in the last three decades. Toa large extent this can be attributed to the progressof environmental analysis. Analytical chemistryitself has gained from the needs of environmen-tal sciences, technology and legislation. It is nowwidely accepted that human activities which influ-ence the chemical composition of the environmenthave to be systematically controlled. Therefore,procedures for the analysis of an increasing num-ber of elements and chemical compounds in air,water, sediment and soil have been developed andthis is still going on. But modern environmen-tal observation has to provide more effect-related

information about the state of our environment andits changes with time. Therefore, it cannot be basedonly on investigations of abiotic environmentalsamples. Rather environmental studies and controlneed to take much more account of the situation inthe biosphere. This includes the transfer of contam-inants, mainly of anthropogenic origin, into plants,animals, and finally also into human beings. As aresult, biomonitoring plays an increasing role inmodern environmental observation programs. Forthat, selected biological organisms, called bioindi-cators, are used for the monitoring of pollutantseither by observation of phenomenological effects(loss of needles, discoloring of leaves, etc.) orby measurement of chemical compounds taken upby the specimens. The latter approach is basedon the chemical analysis of appropriate bioindi-cators and adds another dimension of complexity


8 SAMPLING: COLLECTION, STORAGE

to the sample matrices encountered in environ-mental analysis.

In general environmental analysis should con-tribute answers to the following questions:

• Which pollutants appear where, when and atwhich concentration?

• How mobile and stable are pollutants in eco-systems?

• Which transformations occur with originallyanthropogenic emissions in ecosystems?

• Where are pollutants accumulated or finallydeposited and in which chemical form?

• Which short- and long-term effects do they havewith respect to mankind and the environment?

All these questions are expanding both the chemi-cal nature and the concentration ranges of analytesto be included when considering environmentalstudies and control. For evaluating the necessity,toxicity, availability, distribution, transformationand fate of chemicals, the identity and concen-tration of chemical species [1] rather than thoseof total elements have to be studied in very com-plex systems and samples. Speciation analysis addsnew requirements to the usual boundary condi-tions in environmental analysis, in particular withrespect to analyte stability and preservation oforiginal chemical equilibria. Such aspects are alsopartially considered in the so-called ‘organic anal-ysis’ of compounds outside the POP (persistentorganic pollutants) group, but the thermodynamicand kinetic properties of different redox states orchemical complexes are adding new dimensions oflability and reactivity to analytical chemistry.

Sampling is always the first step of the totalanalytical process (Figure 2.1.1) and its designand implementation has a decisive influence onthe final analytical result. For the purpose of thischapter the term ‘sampling’ will include the collec-tion of specimens from the environment (accompa-nied in many cases by non-chemical operations foron-site sample preparation), followed by interme-diate storage and often by a mechanical processingof the collected material up to samples which areappropriate for the subsequent steps of chemical

analysis, namely analyte separation and determina-tion. In the following text, general aspects and spe-cific requirements for sampling, sample handlingand sample storage, in the speciation analysis ofenvironmental samples will be discussed.

2 GENERAL ASPECTS OFENVIRONMENTAL SAMPLING

The second step after the definition of the ana-lytical problem of interest consists of the care-ful selection and problem-specific design of thesampling procedure. For that one has to considernot only all relevant properties of the analyteof interest, the matrix and the chosen analyticaltechniques, but also a number of parameters thatare necessary for the final evaluation and assess-ment of the analytical data. Unfortunately, a largenumber of environmental analyses are still wastedbecause of insufficient sampling and sample han-dling strategies.

Obtaining representative samples is of utmostimportance. This includes accessing representativesampling sites for the purpose of the study, which

Problem Definition

Process Design

Sampling

Conservation

Sample Processing

Storage

AnalyticalSample Preparation

SpeciesIdentification

SpeciesQuantification

Data Evaluation

Assessment

Qua

lity

Ass

uran

ce/C

ontr

ol

Figure 2.1.1. General steps in the total analytical process.

GENERAL ASPECTS OF ENVIRONMENTAL SAMPLING 9

is often easier to achieve for the control of localemission sources than for the observation of larger‘normally’ exposed areas or even backgroundregions. One needs geographical, meteorologicaland biological data, and information about humanactivities for the site selection [2]. Sampling withina selected area is commonly planned by takinginto account a combination of pre-existing envi-ronmental knowledge, statistical approaches (gridsampling) [3] and financial considerations. Theheterogeneity of our environment on all scales,from ecosystems via populations and individualspecimens down to the molecular level createschallenging demands on sampling concepts. Fre-quently, one or more screening studies have toprecede the actual environmental sampling cam-paign in order to select the final sampling points.

Secondly, representativeness refers also to thekind of material to be sampled. The selectionof representative environmental specimens fromthe various environmental compartments, ecosys-tems, etc. depends mainly on the question (ground-water quality, fate of industrial emissions, foresthealth, etc.), but also on the available knowledgeabout key indicators of the environmental situa-tion. The frequently described ‘flow circles’ ofelements between the atmosphere, hydrosphere,biosphere, pedosphere and lithosphere are certainlynot sufficient for the proper design of samplingprocedures for chemical speciation. The physico-chemical properties of the target compounds haveto be considered in more detail to estimate thetransfer, transformation, deposition and accumu-lation of chemical species in environmental com-partments and specimens. In this respect it may beuseful to classify ‘chemical species’ from the pointof view of the nature of their primary interactionswith the surroundings, namely hydrophilic specieswith Coulomb forces or hydrogen bonding, andlipophilic species with hydrophobic interactions. Inaddition, characteristic pathways of species uptake,transport, accumulation and transformation withinthe biosphere, namely within food chains, shouldbe considered. By taking these factors into accountuseful sample sets of environmental indicators canbe composed which may significantly improvethe information content of environmental analysis.

Lake area

Air Air

Wet depositionWet deposition

Pigeon eggs

Tree leaves/needles(e.g., poplar/spruce)

Fish muscles

Zebra mussel tissue

Lake water

Sediment

Pigeon eggs

Tree leaves(e.g., poplar)

Lichen

Earthworms

Soil

(b)(a)

Urban area

Figure 2.1.2. Examples of environmental indicator sets (modi-fied from [31]): (a) limnic ecosystems: lake area; (b) terrestrialecosystems: urban area.

Examples for two types of ecosystems are illus-trated in Figure 2.1.2.

Another requirement for representativeness con-sists in the number of specimens which have tobe sampled for subsequent analysis. Ideally thisparameter should be calculated on the basis ofthe known variations in the chemical composi-tion of the sampled population which arise fromits geochemical/-physical or biochemical/-physicaldiversity and the variation in imissions in the stud-ied area. An additional uncertainty for biologicalspecimens originates from the natural heterogene-ity within organisms and even organs (see alsoSection 4). All this information is rarely knownin detail from screening investigations and there-fore, this lack limits the precision (and ofteneven the accuracy) of the environmental infor-mation decoded by the chemical analysis of thesampled material.

Representative (and reproducible) environmen-tal sampling has also to consider the variationsof climate/weather conditions, seasonal fluctua-tions of species concentrations in bioindicators andthe different exposure time of the samples. Anexample for the variation of the methylmercurycontent in mussel tissue is shown in Figure 2.1.3aand can be compared with the correspondingdata for total mercury (Figure 2.1.3b). Obviously,


10

9

8

7

6

510/92 12/92 02/93 04/93 06/93

month/year

(a)

Me-Hg [ng Hg/g]

08/93 10/93 12/93 02/94 04/94

60

50

40

30

2010/92 12/92 02/93 04/93 06/93

month/year

(b)

Hg [ng/g]

08/93 10/93 12/93 02/94 04/94

Figure 2.1.3. Seasonal variation of (a) methylmercury and (b) total mercury concentrations related to fresh mass in homogenatesof common mussel tissue from the German Wadden Sea between October 1992 and April 1994 (means and standard deviations).

not only it is necessary to document exten-sively all available parameters (location, time,climate, emission sources, population data suchas density, unusual damage etc.) in connectionwith environmental sampling campaigns, but alsostandardization of sample selection (and han-dling/manipulation) is recommended for obtaining

useful environmental information from the sam-pled material. Corresponding standard operatingprocedures have been developed for various objec-tives of environmental studies and should pro-vide also the basis for repetitive sampling (seebelow). But one should always keep in mind thatthe remaining variation in the representativeness

SAMPLING FOR SPECIATION ANALYSIS 11

of sampled material makes often a greatercontribution to the uncertainty in the finalenvironmental information than any other subse-quent analytical step.

In general the sampling procedure should bedeveloped in advance with as much care and detailas possible. Experts with the necessary knowledgein environmental sciences have to be consulted,but the analytical chemist should be also involvedfrom the beginning to avoid or at least reduce anysample manipulations which could influence thefinal analytical data.

3 SAMPLING FOR SPECIATIONANALYSIS

The key requirement of speciation analysis con-sists of the preservation of the species informationduring the whole analytical process (Figure 2.1.1).One could distinguish two principal strategies forachieving this goal: on the one hand, one may keepthe chemical species of interest unchanged dur-ing all critical steps of their analysis, and on theother hand, the species may be quantitatively trans-formed at an early stage into suitable derivativesfor further separation, accumulation and quantifica-tion methods. In practice there is usually a mixtureof both extremes, but we will consider primarilythe aspects of species-retaining sampling in the fol-lowing. Therefore, chemical stability and volatilityof the analytes of interest are of great importancefor the sampling procedures.

In addition, one has to take into accountthat speciation analysis of environmental samplesconstitutes, in most cases, trace or even ultratraceanalysis. Therefore, both contamination and lossof the analytes have to be avoided along thewhole sample pathway from the sampling site tothe analytical laboratory. All materials which arein contact with the sample must be checked inadvance as a possible source of contaminationor adsorption. Sampling devices with stainlesssteel surfaces which are often used for fieldsampling (knives, scalpels, drills, etc.) should beavoided in sampling for metal speciation. Carefulcleaning of all sampling tools is mandatory. Thiscan be done, for instance, by washing with

concentrated or 10 % nitric acid, depending uponthe material and the sample matrix, followedby several rinsing steps with distilled water.But the application of washing solutions whichactivate the adsorption sites at the surface ofthe tools (such as various concentrated acids forglassware) should be avoided to minimize specieslosses. The cleaning procedures for the samplingdevices as well as the protective measures for thesampling team (wearing of gloves, obeying safetyinstructions etc.) have to be integrated into theplanning and documentation of sampling.

3.1 Air

The collection of samples for ‘speciation analy-sis in the traditional meaning’ is mainly focusedon the investigation of metal(loid) compounds inthe gas phase. Hydrides, methylated and perme-thylated species of As, Hg, Pb, Se etc. have beensampled from terrestrial sources such as wastedeposits or from aquatic ecosystems such as oceanwater surfaces. The main techniques include theapplication of cryotrapping, solid adsorbent car-tridges, polymer bags or stainless steel canisters(with coated inner surfaces). The gas phase can betransferred into a pre-evacuated sampling device[4] or sucked with the help of a pump. In mostcases the air has to pass through a filter (often0.45 µm) to remove particles and aerosols up toa predefined size from the sample. The advan-tage of trapping techniques lies in the integratedaccumulation of the analyte and its partial sep-aration from other sample constituents. On theother hand, any sample treatment has to be care-fully validated with respect to possible changes ofthe original species pattern and contamination byused materials. Analyte loss can be caused by pho-tolytic or surface-catalyzed decomposition, hydrol-ysis, oxidation, adsorption on container walls orabsorption. Critical parameters during collectionand sample preservation include, in particular, tem-perature, light intensity, humidity, oxygen contentand aerosol concentration. For quantitative anal-ysis one has to adjust for any temperature andpressure variations during the transfer of the origi-nal gas phase into the sampling device and during


any subsequent operations because of the muchlarger influence of these parameters on analyteconcentrations in the gaseous state in compar-ison to condensed phases. Recently, the stabil-ity of various volatile As, Sb and Sn specieswas studied as a function of the temperatureand a reasonable recovery was obtained for mostof the species which were kept at 20 ◦C in thedark for 24 h [5].

At present the main problems occur in thequantitative sampling of the original air composi-tion under environmental on-site conditions and itspreservation. Therefore, the collection of the wholegas at the sampling location into containers, fol-lowed by its analysis within a relatively short time(about a day), seems to be the safest way at themoment. The sampling devices can be balloons,cylinders with inlet and outlet valves, canisters orbags. Special consideration should be given to theselected surface. Container walls coated with aninert polymer such as PTFE are recommended.Moreover, contamination by exhaust fumes fromthe pump has to be avoided. If the species of inter-est are stable enough, liquid absorbents, adsorbentcartridges or solid-phase microextraction can beapplied. All adsorption techniques offer the advan-tage of sampling larger amounts of air and inte-grated analyte preconcentration. The same is truefor the most widely used sampling approach forvolatile species at present, cryotrapping. Volatilemetal(loid) species have been sampled from urbanair [6], landfill gas [7] or gases from domes-tic waste deposits [8] with the help of a U-shaped glass trap, filled with a chromatographicpacking material (SP-2100 10 % on Supelcoport,60/80 mesh), and cooled by liquid nitrogen or amixture of acetone/liquid nitrogen. In most casesan empty cold trap has to be placed in front ofthe analyte trap for the removal of water vaporat −40 ◦C. Vacuum filling of stainless steel con-tainers is the official sampling method of the USEPA for volatile compounds in monitoring urbanair [9]. The sequential sampling of volatile Hgspecies using a noble metal trap in series with anactivated carbon trap has been reported [10]. Gasfrom a sewage sludge digester has been collected

in inert plastic Tedlar bags allowing sampling vol-umes of 10 L [5].

An almost unsolved problem is the validationof the sample integrity during all operations.Most of the compounds of interest for speciationstudies in the air are not very stable. For instance,some of the volatile metal(loid) species can betransformed by reactions with co-trapped reactiveair components such as ozone [11]. Therefore,the development of field-portable instrumentationfor sensitive, reliable, fast and efficient on-sitespeciation analysis will be necessary.

Other targets for the speciation analysis of airare metal compounds in the liquid or solid state,i.e. aerosols and dust particles. The latter samplesare collected on various filters (membranes ofcellulose or quartz, glass fibers) and furthertreated as other solid material (see Section 3.4).Aerosols are sampled by using impactors, filters,denuders, electrostatic separators etc. [12]. Theirpreservation with respect to species integrityis very difficult because of their usually highreactivity and it seems to be more promising toapply in situ methods for the direct on-site analysisof aerosol components.

3.2 Water

The sampling technique varies with the propertiesof the species and the water type of interest(groundwater, freshwater: river or lake, seawater,tap water, wastewater, interstitial water of soilor sediment phases, atmospheric precipitation:rain, snow etc.). It is also influenced by thelocation (open sea, inner city or somewhereelse) and the desired sampling depth belowthe water surface. Environmental waters are notas chemically homogeneous as water commonlyavailable in the analytical laboratory and theaspects of representativeness (see Section 2) andhomogeneity have to be obeyed for sampling.

Volatile analytes can be obtained by an on-line combination of purging the water with aninert gas such as helium followed by cryogenictrapping as described above. This approach hasbeen used, for instance, for the investigation ofmethyl and ethyl species of Se, Sn, Hg and Pb


in estuarine water [13]. One has to be extremelycareful to avoid contamination from the ambientair. Moreover, the formation of biofilms which canact as reaction sites for biomethylation on surfacesof the sampling devices has to be avoided.

Water sampling for speciation analysis shouldtake into account all precautions for trace analysisof such matrices (see for instance [14]). They rangefrom avoiding any contamination from the sam-pling vessel itself up to the selection of appropriatematerials for the sampling and storage devices.Any metal contact has to be avoided, also withinthe pumping and tubing system. Polycarbonate orpolyethylene bottles are recommended for mostof the metal species. Mercury species have to bekept in glass bottles [15]. Wet precipitation (rain,snow) is preferably collected with the help of auto-matic wet-only samplers [16, 17], which can alsobe used for sampling such specimens for mercuryspeciation [18].

But several of the routine operations in wateranalysis cannot be applied to all cases of specia-tion studies. On the one hand, side filtration (usu-ally with 0.45 µm membrane filters of celluloseor polycarbonate) or centrifugation is necessaryfor many purposes to remove bacteria and otherreactive nondissolved constituents from the watersample. On the other hand, such separation tech-niques should be checked not only as sources ofcontamination but also with respect to their influ-ence on original species distributions between thesolution phase and the interfaces between partic-ulate matter and water. The latter are prominentadsorption and reaction sites for many species ofinterest. The frequently recommended acidificationof water samples not only stabilizes metal ionsin solution by reducing their adsorption on con-tainer walls and bacterial activity, but also changesacid–base equilibria and coupled redox and com-plex formation equilibria, which would preventthe determination of such speciation patterns (seeChapter 5.9). Overall the validation of samplingtechniques with respect to species-retaining oper-ations has always to be performed with referenceto the target species and general approaches donot exist.

Preservation of original water samples can be amajor problem as discussed in Section 5. There-fore, a more promising approach for speciationanalysis of dissolved species consists of the appli-cation of in situ measurements as described inother chapters of this handbook (see for instanceChapter 5.9).

3.3 Biological material

Most of the speciation analysis in biologicalenvironmental samples has been directed to thedetermination of organometallic constituents andredox states of trace elements. For the design ofthe sample collection one has to take into accountthat the biosphere varies much more widely inits physical and chemical properties relevant tospecies distribution and transformation than theabiotic environmental media air, water, sedimentand soil. Therefore, only more general aspectswill be discussed here. Detailed sampling protocolshave to be developed separately for the specificstudies depending on the particular requirementsof the problem of interest.

Liquid samples from animals or plants (blood,urine, plant juices) have rarely been collected forspeciation analysis until now. The main reasonseems to be that environmental speciation studieswere focused on such specimens which accumulatethe compounds of interest and which are easilyavailable in larger amounts for analyzing theirultratrace constituents. In principle, biologicalfluids from the biosphere could be handled in acomparable manner to the corresponding humansamples (see other chapters of this handbook). Thiscommonly includes filtration or centrifugation ofthe fresh sample material followed by short-termstorage at −4 ◦C in the dark. Alternative techniquesare shock-freezing and preservation as a frozensample or lyophilizing and storage as a driedsample. Naturally, the latter method can only beapplied to chemical species which are very stable.Any addition of chemical preservatives, includingacidification, should be avoided.

‘Solid’ biological materials such as tissueshave attracted much more attention for spe-ciation analysis in recent years because they


are regarded as an important deposit of poten-tially hazardous compounds. The selection andidentification of the biological specimens in thenatural environment require appropriate biologi-cal and ecological knowledge. Moreover, a sci-entifically sound interpretation of the analyticaldata can only be performed if sufficient informa-tion about the environmental situation (exposurecharacteristics, population density and health, etc.)and ecological functions (trophic level, uptake andtransport routes for chemical compounds) as wellas relevant biological and biometric parameters(age, sex, variations of biological activity with day-time and season, surface area of leaves or needles,mass ratios of individual specimens and sampledorgans, etc.) of the studied organism are available.In addition, the sampling strategy has to take intoaccount not only the natural heterogeneity withina biological population, but also that within anindividual organism. To fulfill the requirements ofrepresentativeness sufficient numbers of individ-ual specimens should be sampled randomly withinthe selected area and for most studies they haveto be mixed and homogenized (see Section 4). Ithas also been shown that different parts of plants(for instance, algae) [19] accumulate many traceelements to a different extent which has to be con-sidered for the final sample composition. There-fore, detailed sampling procedures and protocolsare mandatory. They can be developed for thespecific purpose of the investigation on the basisof existing standards for long-term biomonitoringprograms such as environmental monitoring andspecimen banking [20–22].

During sample collection and further samplemanipulations one has to consider that the so-called solid biota are from a chemical point of viewvery fragile materials with significant water con-tent. Usually the amount of analyte in the chemi-cally complete biological sample (leaf, plant stem,liver, kidney, muscle tissue, egg etc.) is of interestand one has to preserve during sampling the totalchemical composition of the specimen as muchas possible. Therefore, the removal of the spec-imens from their natural environment should becarefully planned. Species transformation and loss

can already occur during collection at the sam-pling site. Degradation depends on the chemicalnature of the species and may be influenced by bio-chemical processes such as enzyme activity. Thisis usually more critical in animal organs than plantsamples. A key parameter for chemical reactionrates is the temperature. Consequently, one candiminish species transformations by decreasing thetemperature as much and as early as possible. Atpresent, shock-freezing of the desired samples inthe gas phase above liquid nitrogen seems to bethe safest technique and can be performed imme-diately at the sampling site. It offers the additionaladvantage of an inert gas atmosphere for the storedsamples. If this approach is not feasible within thespecific project a short-term preservation of thebiological material at −20 ◦C is recommended.

The first preparation steps with the sampleshave to be performed mostly at the sampling sitebefore freezing of the material. Plant samples fromnatural ecosystems are often modified by adheringmaterial such as dust, soil or sediment particles.A general rule does not exist for separating suchabiotic material and one has to decide which of thesurface-attached constituents can be considered asan integral part of the sample. Even gentle washingof freshly cut plant organs can partly removesome of the relevant compounds and extensivewashing procedures should be avoided in mostcases. Marine samples such as algae or musselscan be carefully cleaned of sediment by shakingthem in the surrounding water [20, 23].

Another operation of concern is the dissectionof target organs which should preferably beperformed immediately on the sampling site.Animal organs have to be extracted as fast aspossible to minimize species transformations. Alldissection tools must be selected with respect topossible contamination of the analytes of interest.Many speciation studies are aimed at metal(loid)compounds and therefore stainless steel knivesetc. should be avoided. One can use titaniumknives, tools which are coated with titanium nitrideor ceramic scalpels. In general all precautionsagainst contamination of the sample should becarefully carried out. Sample containers made frompolymeric material such as PTFE, polyethylene or


polycarbonate are often adequate. The samplingcrew has to wear protective gloves and thecontamination is further minimized by working asearly as possible on site in mobile laboratoriesequipped with clean-bench facilities. It has beenshown that the dissection of mussels, i.e. theseparation of mussel tissue from the shell, can beperformed after deep-freezing and cryostorage ofthe whole mussels in the laboratory [24].

Overall the on-site preparation and preservationof fresh biological material, i.e. samples whichstill include all liquid components (such as water)belonging originally to the target specimen, isrecommended for speciation analysis.

3.4 Sediment and soil

Most of the investigations which have beenreported under the heading ‘speciation’ of sed-iments, soils or related samples were actuallydirected to operationally defined fractionation [1]mainly by sequential extraction procedures. Thisis not within the scope of this chapter on chem-ical speciation in the sense of analyzing redoxstates and binding partners for the trace elementsof interest.

The most difficult and critical aspect of soil andsediment sampling is representativeness. Environ-mental specimens are very heterogeneous in theirchemical composition and an extensive screeningof each sampling site would be necessary for sci-entifically sound investigations. But usually thenumber and distances of the lateral and vertical(depth) sampling points are selected on the basis ofavailable geological information by applying sta-tistical models. Corresponding sampling grids andprocedures are described in the literature [25–27].

One has to take into account for the adap-tion of sampling concepts for speciation analysisthat larger amounts of such heterogeneous materialmust be collected in comparison to other envi-ronmental specimens for obtaining representativesamples. Therefore, appropriately sized samplingtools (shovel, corer, spoon, knife etc.) and contain-ers made from materials such as titanium, ceram-ics or plastics (e.g. polyethylene) are necessary.Depending on the problem of interest the soil can

be collected by initial separation of the mineral lay-ers or as intact depth profiles. The latter approach,which is also very common in the sampling of lakeor marine sediments, is achieved by using differentcorer types (with plastic tubes inside) depending onthe soil type, sampling depth and required samplemass. Recommendations for soil sampling can befound, for instance, in [28]. In general, large sam-ple constituents (>2 mm) and larger parts of plants(roots, branches) are manually removed from thecollected material.

Sediments can be collected with the help ofgrab or core samplers. Sediment traps are usedin dynamic flow systems such as rivers for sam-pling at least part of the suspended matter. On-siteoperations include the decantation of water, theremoval of particles larger than 2 mm and some-times a wet sieving of the sediment. There aredifferent approaches concerning the particle sizeseparation, but 20 or 63 µm are the most com-monly used filters. Particulate matter from aquaticecosystems is collected either by filtration (often0.45 µm) or by continuous flow centrifugation forprocessing larger sample volumes.

Special procedures for the species-retainingsampling of soil or sediment have not beendescribed and validated until now. The prepara-tion of harbor and coastal sediments as referencematerials for matrix-matched speciation analysis oftributyl tin has been reported [29, 30]. But theapplication of air drying and the described sam-ple processing (see Section 4) cannot be recom-mended for all other analytes of interest. Problemsalso arise from changes in the oxygen concen-tration during sampling, especially for originallyanoxic sediments. Even the definition of the sam-ple composition is difficult for speciation purposes,because the interfaces between solid particle andwater, biofilms and pore water can be promi-nent reaction sites for species transformations andit is almost impossible to preserve their originalstate during sampling at present. Therefore, thedevelopment of both in situ methods for specia-tion analysis (in particular for sediments) and newapproaches to species-retaining sampling proce-dures specifically designed for certain groups ofchemical species, will be necessary.


4 SAMPLE PROCESSING

Most of the solid materials collected from theenvironment have to be prepared by physicaloperations (grinding, drying, etc.) before any ana-lytical pretreatment in the narrow sense (extrac-tion etc.) can be applied. The influence of suchsample manipulations, called sample processing inthe following, on the total speciation analysis issometimes underestimated and should always becarefully controlled by the analyst with respect tocontamination, loss or transformation of the ana-lyte and relevant matrix modifications.

Usually the collected material has to be homog-enized and divided into chemically authenticaliquots for repetitive analysis. If one is notinterested in the ‘fractionation’ of the sampleconstituents [1] with respect to their surface attach-ment to different matrix components or the differ-entiation between inner- and extracellular speciesin biological specimens and comparable questions,grinding is performed for the purpose of homoge-nizing and creating large surface-to-volume ratiosfor subsequent species extractions.

For biological specimens sample processingwith the requirement of minimizing possiblechanges in their chemical composition can be

based on the developments and standards whichhave been acquired within the frame of environ-mental biomonitoring and specimen banking pro-grams [22, 31]. The immediate shock-freezing ofthe sampled material in the gas phase above liquidnitrogen shortly after its removal from the naturalenvironment on the sampling site and the furtherstorage at temperatures below −130 ◦C providesa raw material with very good mechanical prop-erties for crushing. During grinding the samplesshould be continuously cooled (preferably with liq-uid nitrogen) to avoid species transformation oreven loss [32] due to the local heating effects fromfriction. The cryogrinding of a broad range of bio-logical specimens, including fatty or keratin-richsamples (such as liver, hair) and fibrous material(plant shoots etc.), is possible by using a vibrat-ing rod mill (for instance, CryoPalla) operatedunder continuous cooling with liquid nitrogen [33].In many cases a pre-crushing of larger portions offrozen material has to be performed. The grind-ing at such low temperatures offers not only theadvantage of a diminished probability for chemi-cal transformations in the sample, but one can alsoachieve a fine powder with small particle sizeswithout sieving. This is shown for two differentbiological matrices in Figures 2.1.4 and 2.1.5. But

100

90

80

70

60

50

40

30

20

10

0

10

9

8

7

6

5

4

3

2

1

010 50

particle size [µm]

cum

ulat

ive

freq

uenc

y [%

]

freq

uenc

y [%

]

100 200 550

Figure 2.1.4. Particle size distribution after cryogrinding of pine shoots.

SAMPLE PROCESSING 17

100

90

80

70

60

50

40

30

20

10

0

10

9

8

7

6

5

4

3

2

1

010 50

particle size [µm]

cum

ulat

ive

freq

uenc

y [%

]

freq

uenc

y [%

]

100 200 550

Figure 2.1.5. Particle size distribution after cryogrinding of fish liver (bream).

soil samples can also be finely ground and homog-enized by this technique [34].

Such ground material cannot be used forfractionation between groups of species whichare differently immobilized at natural occurringsurfaces in a sample of interest, but it is well suitedfor the analysis of the chemical composition ofthe sample including speciation (see, for instance

[35, 36]). Moreover, the finely powdered materialis often very homogeneous [37] which simplifiesits proper aliquotation into small subsamples. Butthe natural heterogeneity of analyte distributionsin biological specimens can restrict the achievablehomogeneity even in such ground samples. Thisis demonstrated for the case of nickel in fishliver (bream) in Figure 2.1.6. The element (and

+

+

+

+

+

+

8.714.7 23.3 5.0

2.6 3.6

2.0

1.5

1.0

0.5

0.00 20 40 60

measurement no.

Ni [µg/g d.m.]

80 100 120

Figure 2.1.6. Nickel concentration related to dry mass in a finely cryoground homogenate of fish liver (bream) determined bysolid sampling–atomic absorption spectrometry with sample in-weights of about 680 µg.


consequently also the species) distribution in thepowder, which had a particle size histogram almostidentical to that shown in Figure 2.1.5, indicatedseveral microparticles with extremely high Nicontent. An estimation of the minimum samplemass required for repetitive analyte determinationswith an accepted uncertainty of 5 % accordingto [38] led to a sample in-weight of about6700 mg which is well above common analyticalparameters [37].

Other approaches for cryogenic sample grindingand homogenization include the use of planetary orball mills [30, 39]. For all techniques the transferof ultratrace amounts of surface material fromthe mill into the sample is unavoidable and thiscontamination has to be considered. The movingparts and the container walls should be made ofPTFE or preferably of titanium (which cannot bean analyte afterwards!) in order to minimize thecritical contamination.

Many speciation studies have been performedwith dried samples and most of the certified refer-ence materials available at present for the qualityassurance of species determinations in environ-mental matrices have been also dried. The BCRprocedures of sample processing for CRM 463/464Tuna Fish (methylmercury content) and CRM 627Tuna Fish (dimethylarsinic acid and arsenobetainecontents) included mincing of dissected fish mus-cles, freeze-drying, grinding in a ZrO2 ball mill,sieving and mixing. Comparable techniques wereapplied to the preparation of CRM 477 MusselTissue (content of butyl tins) [30]. The NISThas prepared a fresh (frozen) SRM 1974a MusselTissue (Mytilus edulis) certified for its methylmer-cury content by cryogenically ball milling dis-sected mussel tissue. This homogenized frozentissue was blended in an aluminum mixing drumand divided into aliquots. A part of the processedmaterial was prepared as SRM 2974 by subse-quent freeze-drying, blending and radiation ster-ilization (60Co) [40]. Two marine bivalve mollusctissue materials of NIST (SRM 2977 Mussel Tis-sue (Perna perna); SRM 1566b Oyster Tissue),which have been certified for their methylmercurycontent, were blended as partially thawed tissuesin a cutter mixer, afterwards frozen, freeze-dried,

powdered with the help of a jet-mill and radiationsterilized [41]. But one has to be cautious in apply-ing these or other procedures, which have beendeveloped for the preparation of reference materi-als, to projects directed to the analysis of originalspecies patterns in biological environmental sam-ples. The preservation of original species concen-trations of the collected specimens is not necessaryfor the further use of CRMs and aspects such ascost-efficient processing and long-term storage oflarge sample batches and the subsequent shipmentof stable subsamples all over the world influencesignificantly the selection of techniques. Therefore,sample manipulations such as sieving are accept-able for CRM production, but should be avoidedif one is interested in the speciation for the pur-pose of environmental monitoring or research. Thisalso holds true with respect to the adaption of pro-cessing procedures for soils and sediments. Forinstance, the BCR candidate reference materialsCRM 462 Coastal Sediment (organotin content)was air dried and, after temporary storage andtransportation at −20 ◦C, dried at 55 ◦C, sieved(1 mm), finely ground using a jet mill with a clas-sifier, sterilized at 120 ◦C, and homogenized in amixer [30]. Due to the lack of analytical methodsfor the direct speciation of organotin compounds insolids one cannot evaluate the species transforma-tions and losses during this sequence of processingsteps, but they certainly modify the original sampleto a larger extent.

The drying process can lead to sample changeswhich may significantly influence further specia-tion analysis. One should keep in mind that dryingcan remove more than just water molecules fromthe fresh sample and would consequently alter thegeneral chemical composition of the material. Forbiological samples prepared for speciation analy-sis, oven-drying should be avoided. But even gen-tle freeze-drying can change the species pattern inthe original material as was shown in the case ofmethyl- and butyltin species in fish muscles [42].One of the most significant limitations of usingdried materials for speciation analysis consists oftheir different extraction properties in comparisonto fresh (wet) samples. For example, the extrac-tion yield of arsenic species from fine powders

SAMPLE STORAGE 19

of algae (Fucus vesiculosus) and common mus-sel (Mytilus edulis) tissue by using methanol/waterdecreased from 95.5 and 97 % (wet samples) to70 and 75.5 %, respectively, by extracting driedaliquots. Obviously the freeze-drying not onlyremoves about 80 % (algae) and 95 % (mussel tis-sue) of the original sample mass as ‘water’, but italso reduces significantly the extractability of Asspecies [43].

In general, steps of sample manipulation shouldbe minimized, fresh and nonsieved material shouldbe analyzed if possible and the whole sampleprocessing for speciation analysis has to beevaluated for its influence on changes in theoriginal chemical composition.

5 SAMPLE STORAGE

The preservation of the chemical integrity of thesample has to be the main objective of storage.This concerns first of all the avoidance of loss andcontamination of the species of interest, but alsoof changes in sample mass or matrix propertiesrelevant for speciation analysis. The selection ofstorage methods for speciation of environmentalsamples depends on the type of sample material,the storage time, the size and number of samplesand the available financial resources. It usuallyalso has to be designed with respect to thetarget species because generalizations or stabilitypredictions cannot be obtained from studies aboutthe stability of specific species in a matrix undercertain conditions due to insufficient knowledgeabout the chemistry of trace element species insuch complex matrices. A major problem for mostof the chemical species consists of the lack ofanalytical methods to determine them directly ina solid sample. Therefore, the different storageapproaches for solids cannot be evaluated ina straightforward manner. Commonly long-termstudies are performed by using the analyticalprocedures which are based on the determinationof chemical species in the liquid or gaseousstate and which are available at the beginningof the stability investigation. Consequently, onehas to consider also uncertainties of long-termreproducibility of the analytical sample preparation

and of the methods for species separation andquantification in addition to aspects such ashomogeneity and contamination control.

Species transformations in closed vessels, i.e.reactions between sample constituents themselvesand/or with the inner container surfaces, can bestimulated by temperature, light, microbiologicalactivity or pH changes (in the case of liquids).The key storage parameter which has to be selectedcarefully is the temperature. In most cases it can beeasily adjusted and controlled. The storage of deep-frozen samples seems to be the safest approachfor many environmental samples and analytes. Butone has to develop and to apply species-retainingrethawing procedures and gentle methods for sub-sequent rehomogenization of the sample. The con-tainer material has to be selected with respect tothe storage temperature and possible contamina-tion. Polymer materials such as PTFE or polyethy-lene are frequently used, but storage at very lowtemperature such as in the gas phase above liquidnitrogen (<−150 ◦C) has to be carried out in othermaterials, for instance scintillation glass vials. Thecontainer size should be selected in order to mini-mize the ratio of container surface area to samplevolume and therefore species contamination andloss. The samples should be kept in the dark, or atleast in appropriately colored or wrapped bottles.One has to take into account that freeze-dried sam-ples are usually hygroscopic and have to be keptin a humidity-controlled environment. Recently, ithas been recommended to determine the optimalranges of water activity and water content of drybiological material, where hygroscopicity is lowwithout danger of product deterioration, and toadjust them for long-term storage [44]. Moreover,it may be necessary to reduce the access of oxy-gen to the storage medium (always advisable forliquid samples) [45] in order to avoid oxidationreactions. This is simultaneously achieved in thecase of liquid nitrogen cooling.

Air samples cannot be stored directly for futurespeciation analysis. The preservation of originalgaseous components in a cold trap is possible forseveral days in liquid nitrogen [46]. The long-term storage of gaseous species does not seem tobe feasible.


Liquid samples are often stored at 4 ◦C for alimited period. But bacterial activity and exchangereactions or adsorption/desorption at containerwalls can still be significant. Acidification fordiminishing the adsorption of species should beavoided because of the pH dependence of manyof the species equilibria in solution. Organometal-lic species may decompose under the influenceof light, microorganisms or suspended particles[47]. Systematic investigations have been per-formed into the stability of a number of speciesin aqueous solution (mainly calibrants) within theframe of BCR and SM&T feasibility studies fornew reference materials [30, 48]. For instance, itwas found that trimethyllead was stable in a BCRcandidate reference material ‘artificial rainwater’during storage in the dark at 4 ◦C for 12 months.But about 10 % of this species had decomposedafter 3 years [30]. One has to take into accountthat other solution components which are presentin real environmental waters can facilitate speciesdegradation. Therefore, conclusions from stabilityexperiments in pure aqueous standards cannot begeneralized for other water types. Species preser-vation can also be achieved by freezing liquidsamples. Using that approach tributyltin was stablein water for 3 months [49]. But one has to checkif the liquid–solid–liquid phase transitions changethe species pattern of interest in the samples ashas been indicated for selenomethionine and Sb(V) [50]. The lack of stabilizing interactions withsolid matrix components (such as biological tis-sues) can induce more easily transformations formany of the dissolved species (ions, metal–ligandcomplexes) during drastic temperature changes andphase transitions.

Biological solid samples can be preserved withrespect to their chemical composition at very lowtemperatures in the dark for many years [31].To prevent any microbiological activity the stor-age temperature should be below −130 ◦C. Thiscan be easily and efficiently achieved by stor-ing the samples in special cryocontainers in thegas phase above liquid nitrogen. Such methodshave been used for more than 20 years in envi-ronmental specimen banking. But one has totake into account that freezing and thawing may

induce sample changes such as alterations of pro-tein structures or chemical interactions betweenmetal ions and biomacromolecules, complexes, etc.Other approaches developed for the preservationof organometallic species in biological referencematerials [30] included the removal of most ofthe water in the original sample to reduce bacte-rial activity (but see Section 4). Dry mussel tis-sue did not show any further transformation ofbutyltin species during dark storage at −20 ◦Cfor 44 months, whereas phenyltins were not sta-ble. The concentrations of methylmercury andtwo arsenic species (arsenobetaine, dimethylarsinicacid) did not change during storage experimentswith dry fish tissues in brown bottles between−20 ◦C and 40 ◦C for 12 months and 9 months,respectively [30]. But stability studies of chemi-cal species in biological samples have only beenperformed until now for a limited number ofanalytes and matrices. Unfortunately, the relevantthermodynamic and kinetic properties for most ofthe chemical species in their natural biologicalmicroenvironment are not sufficiently known todesign a general strategy for the preservation oforiginal speciation patterns. At the moment rapiddeep-freezing and species analysis as fast as pos-sible seem to be advisable.

For sediments, soils and particulate matter thepreparation techniques for storage which are com-monly applied in trace element analysis such ascooling to 4 ◦C, freeze-drying, oven-drying or air-drying, cannot ensure the preservation of chemi-cal species. It has been shown within feasibilitystudies for the development of reference materi-als [30] that the tributyltin content of a harborsediment which was air-dried, sterilized at 80 ◦C,ground and sieved (75 µm) did not change dur-ing a storage period of 12 months at 20 ◦C. Butthis species degraded in a coastal sediment storedunder the same conditions [30]. Butyltins were suc-cessfully conserved in a sediment after freezingand lyophilization for at least 1 year whereas thetransformation of phenyltins could not be avoided[51]. Methylmercury was stabilized in a sedi-ment by γ -radiation which had destroyed bacte-rial activity [52]. But as discussed above, vali-dated preservation and storage methods for such

REFERENCES 21

complex samples in their original chemical com-position still have to be developed and deep-freezing of wet samples may be the safest approachat present.

6 FURTHER CHALLENGES

Speciation analysis requires samples with un-changed chemical composition, and appropriatesampling procedures for complex environmentalspecimens are still in the developing phase atpresent. New methodical approaches are necessaryfor species-retaining sample collection, preser-vation and chemical characterization. This con-cerns in particular the representative sampling ofvery heterogeneous matrices such as sedimentsor soils, the species-oriented sampling of biolog-ical material and the conservation of less stablespecies patterns. Speciation studies in biologicalorganisms have to take into account the natu-ral biological compartmentation, and therefore anincreasing number of investigations may even needthe sampling of single cells. In the future sam-pling also has to provide reproducibly study mate-rials with intact physical and chemical structuresincluding undisturbed interfaces for the investiga-tion of species- and matrix-specific interactions inreal microenvironments.

Progress with such challenges can only beexpected if the procedures for speciation analysisundergo the same rigorous quality assurance andcontrol as was developed for total element deter-minations mainly in the last 20 years and as is prac-ticed now by the laboratories with an appropriatesense of responsibility worldwide. For that matrix-matched certified reference materials (CRM) aremandatory [43]. The demand for ‘fresh’ (nondried)biological and abiotic environmental CRMs, whichare less manipulated than the presently availableones and which are certified for various speciespatterns, has to be fulfilled in the future in orderto establish speciation analysis as a scientificallysound analytical activity. Quality assurance for thewhole speciation procedure from sampling to dataevaluation has to be established with the empha-sis on controlling the trueness of the results with

respect to their information content concerning theoriginal environmental specimen, and not only inregard to the reproducibility of analytical data.

Moreover, fundamental chemical investigationsare necessary for the systematic design of com-plete procedures which allow the preservation ofthe original species information. They have to bedirected to the elucidation of the thermodynam-ics and kinetics of species states and processes incomplex natural systems such as biological cellcompartments, to the understanding and quantita-tive description of metabolic reactions under stressconditions which can take place during sample col-lection, and to all other processes which influencethe result of sampling.

7 REFERENCES

1. Templeton, D. M., Ariese, F., Cornelis, R., Danielsson,L. G., Muntau, H., van Leeuwen, H. P. and Lobinski, R.,Pure Appl. Chem., 72, 1453 (2000).

2. Lewis, R. A., Stein, N. and Lewis, C. W. (Eds), Environ-mental Specimen Banking and Monitoring as Related toBanking , Martinus Nijhoff, Boston, MA, 1984.

3. Guy, P., Sampling for Analytical Purposes , John Wiley &Sons, Ltd, Chichester, 1998.

4. Schweigkofler, M. and Niessner, R., Environ. Sci. Tech-nol., 33, 3680 (1999).

5. Haas, K. and Feldmann, J., Anal. Chem., 72, 4205 (2000).6. Pecheyran, C., Lalere, B. and Donard, O. F. X., Environ.

Sci. Technol., 34, 27 (2000).7. Feldmann, J., Koch, I. and Cullen, W. R., Analyst , 123,

815 (1998).8. Feldmann, J., Grumping, R. and Hirner, A. V., Frese-

nius’ J. Anal. Chem., 350, 228 (1994).9. Evans, G. F., Lumplein, T. A., Smith, D. L. and

Somerville, M. C., J. Air Waste Manag. Assoc., 42, 1319(1992).

10. Sommar, J., Feng, X. and Lindqvist, O., Appl. Organomet.Chem., 13, 441 (1999).

11. Szpunar, J., Bouyssiere, B. and Lobinski, R., Samplepreparation techniques for elemental speciation stud-ies, in Elemental Speciation. New Approaches for TraceMetal Analysis , Caruso, J. A., Sutton, K. L. and Ack-ley, K. L. (Eds), Chapter 2, Elsevier, Amsterdam, 2000,pp. 7–40.

12. Klockow, D., Kaiser, R. D., Kossowski, J., Larjava, K.,Reith, J. and Siemens, V., Metal speciation in fluegases, work place atmospheres and precipitation, inMetal Speciation in the Environment , Broekaert, J. A. C.,Gucer, S. and Adams, F. (Eds), Springer, Berlin, 1990,pp. 409–433.


13. Amouroux, D., Tessier, E., Pecheyran, C. and Donard,O. F. X., Anal. Chim. Acta, 377, 241 (1998).

14. Helmers, E., Sampling of sea and fresh water for theanalysis of trace elements, in Sampling and SamplePreparation , Stoeppler, M. (Ed.), Chapter 4, Springer,Berlin, 1997, pp. 26–42.

15. Ahmed, R. and Stoeppler, M., Analyst , 111, 1371 (1986).16. Klockow, D., Fresenius’ Z. Anal. Chem., 326, 2880

(1985).17. Gromping, A. H. J., Ostapczuk, P. and Emons, H., Che-

mosphere, 34, 2227 (1997).18. Augustin-Castro, B., Untersuchungen zum Quecksilber-

kreislauf in der Umwelt, Ph.D. Thesis, University ofEssen, Berichte des Forschungszentrums Julich 3694,Julich (1999).

19. Amer, H., Emons, H. and Ostapczuk, P., Chemosphere,34, 2123 (1997).

20. Umweltbundesamt (Ed.), Verfahrensrichtlinien fur Probe-nahme, Transport, Lagerung und Chemische Charakter-isierung von Umwelt- und Humanorgan-Proben , ErichSchmidt Verlag, Berlin, 1996.

21. Giege, B., Odsjo, T., Barikmo, J., Hirvi, J.-P., Petersen, H.and Petersen, A. E., Manual for the Nordic Countries.Nordic Environmental Specimen Bank – Methods in Usein ESB , TemaNord 543, ISBN 92-9120-662-8, 1995.

22. Becker, P. R., Wise, S. A., Thorsteinson, L., Koster, B. J.and Rowles, T., Chemosphere, 34, 1889 (1997).

23. Amer, H. A., Ostapczuk, P. and Emons, H., J. Environ.Monit., 1, 97 (1999).

24. Schladot, J. D. and Backhaus, F., The common mussel asmarine bioindicator for the environmental specimen bankof the Federal Republic of Germany, in Specimen Bank-ing , Rossbach, M., Schladot, J. D. and Ostapczuk, P.(Eds), Chapter 4.2, Springer, Berlin, 1992, pp. 75–87.

25. Kateman, G., Chemometrics – Sampling Strategies ,Springer, Berlin, 1987.

26. Guy, P. M., Sampling of Heterogeneous and DynamicMaterial Systems , Elsevier, Amsterdam, 1992.

27. Einax, J. W., Zwanziger, H. W. and Geiss, S., Chemo-metrics in Environmental Analysis , VCH, Weinheim,1997.

28. del Castro, P. and Breder, R., Soils and soil solutions, inSampling and Sample Preparation , Stoeppler, M. (Ed.),Chapter 5, Springer, Berlin, 1997, pp. 43–56.

29. Quevauviller, Ph., Method Performance Studies for Spe-ciation Analysis , The Royal Society of Chemistry, Cam-bridge, 1998.

30. Quevauviller, Ph. and Maier, E. A., Interlaboratory Stud-ies and Certified Reference Materials for EnvironmentalAnalysis. The BCR Approach , Elsevier, Amsterdam, 1999.

31. Emons, H., Schladot, J. D. and Schwuger, M. J., Chemo-sphere, 34, 1875 (1997).

32. Lambrecht, S., Emons, H., Matschullat, J. and Ross-bach, M., in preparation.

33. Koglin, D., Backhaus, F. and Schladot, J. D., Chemo-sphere, 34, 2041 (1997).

34. Arunachalam, J., Emons, H., Krasnodebska, B. andMohl, C., Sci. Total Environ., 181, 147 (1996).

35. Shawky, S. and Emons, H., Chemosphere, 36, 523 (1998).36. Jakubowski, N., Stuewer, D., Klockow, D., Thomas, C.

and Emons, H., J. Anal. At. Spectrom., 16, 135 (2001).37. Rossbach, M., Giernich, G. and Emons, H., J. Environ.

Monit., 3, 330 (2001).38. Pauwels, J. and Vandecasteele, C., Fresenius’ J. Anal.

Chem., 345, 121 (1993).39. Schladot, J. D. and Backhaus, F., Collection, preparation

and long-term storage of marine samples, in Samplingand Sample Preparation , Stoeppler, M. (Ed.), Chapter 7,Springer, Berlin, 1997, pp. 74–87.

40. Donais, M. K., Saraswati, R., Mackey, E., Demiralp, R.,Porter, B., Vangel, M., Levenson, M., Mandic, V., Aze-mard, S., Horvat, M., May, K., Emons, H. and Wise, S.,Fresenius’ J. Anal. Chem., 358, 424 (1997).

41. Tutschku, S., Schantz, M. M., Horvat, M., Logar, M.,Akagi, H., Emons, H., Levenson, M. and Wise, S., Fre-senius’ J. Anal. Chem., 369, 364 (2001).

42. Shawky, S., Emons, H. and Durbeck, H. W., Anal. Com-mun., 33, 107 (1996).

43. Emons, H., Fresenius’ J. Anal. Chem., 370, 115 (2001).44. Ruckold, S., Grobecker, K. H. and Isengard, H.-D., Fre-

senius’ J. Anal. Chem., 370, 189 (2001).45. Gomez-Ariza, J. L., Morales, E., Sanchez-Rodas, D. and

Giraldez, I., Trends Anal. Chem., 19, 200 (2000).46. Pecheyran, C., Quetel, C. R., Lecuyer, F. M. M. and

Donard, O. F. X., Anal. Chem., 70, 2639 (1998).47. van Cleuvenbergen, R., Dirkx, W., Quevauviller, P. and

Adams, F., Int. J. Environ. Anal. Chem., 47, 21 (1992).48. Quevauviller, Ph., de la Calle-Guntinas, M. B., Maier,

E. A. and Camara, C., Mikrochim. Acta, 118, 131(1995).

49. Valkirs, A. O., Seligman, P. F., Olson, G. J., Brinckman,F. E., Matthias, C. L. and Bellama, J. M., Analyst , 112,17 (1987).

50. Lindemann, T., Prange, A., Dannecker, W. and Neid-hart, B., Fresenius’ J. Anal. Chem., 368, 214 (2000).

51. Gomez-Ariza, J. L., Morales, E., Beltran, R., Giraldez, I.and Ruiz-Benitez, M., Quim. Anal., 13, S76 (1994).

52. Quevauviller, P., Fortunati, G. U., Filippelli, M., Baldi, F.,Bianchi, M. and Muntau, H., Appl. Organomet. Chem., 10,537 (1996).

2.2 Sampling of Clinical Samples:Collection and Storage

Koen De CremerLaboratory for Analytical Chemistry, Ghent University, Belgium

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 232 Presampling Steps . . . . . . . . . . . . . . . . . . 24

2.1 Collection and storage of sampleinformation . . . . . . . . . . . . . . . . . . 24

2.2 Cleaning and evaluation of theinstruments . . . . . . . . . . . . . . . . . . . 24

3 Collection and Storage of Blood . . . . . . . . 274 Collection and Storage of Urine . . . . . . . . 295 Collection and Storage of Tissues . . . . . . . 306 Microdialysis . . . . . . . . . . . . . . . . . . . . . . 317 Specific Precautions for Some Elements 33

7.1 Aluminum 337.2 Antimony . . . . . . . . . . . . . . . . . . . . 347.3 Arsenic . . . . . . . . . . . . . . . . . . . . . 347.4 Cadmium . . . . . . . . . . . . . . . . . . . . 357.5 Chromium . . . . . . . . . . . . . . . . . . . 357.6 Cobalt . . . . . . . . . . . . . . . . . . . . . . 36

7.7 Copper . . . . . . . . . . . . . . . . . . . . . . 367.8 Lead . . . . . . . . . . . . . . . . . . . . . . . 367.9 Lithium . . . . . . . . . . . . . . . . . . . . . 37

7.10 Manganese . . . . . . . . . . . . . . . . . . . 377.11 Mercury . . . . . . . . . . . . . . . . . . . . . 387.12 Nickel . . . . . . . . . . . . . . . . . . . . . . 397.13 Selenium . . . . . . . . . . . . . . . . . . . . 397.14 Tin . . . . . . . . . . . . . . . . . . . . . . . . . 407.15 Zinc . . . . . . . . . . . . . . . . . . . . . . . . 40

8 Influence of pH, Salt Molarity andAcetonitrile Concentration on a SelectedMetal–Protein Complex, i.e. theVanadate(V)–Transferrin Complex . . . . . . 41

9 Concluding Remarks . . . . . . . . . . . . . . . . . 4310 Acknowledgements . . . . . . . . . . . . . . . . . . 4411 References . . . . . . . . . . . . . . . . . . . . . . . . 44

1 INTRODUCTION

Accuracy, quality assurance, quality control, re-peatability, reproducibility, true value, validation,uncertainty of the measurement, . . . and so manymore concepts have become common language inany analytical laboratory. Indeed, these parametersshould be carefully considered during an analysisbut it is not a given fact that when these crite-ria are fulfilled the results will be perfect. Thiswill depend strongly on the quality of preanalyt-ical steps such as sampling, sample handling andconservation of the measured species. When ana-lyzing total concentrations of an element, factors

such as contamination and conservation are alreadyconsidered to be very important. When, however,there is a need for speciation between differentcomplexes of an element in a certain matrix, con-servation of the species becomes of paramountimportance. It is no longer sufficient to bring thetotal amount of an element into solution beforemeasurement. In addition, throughout the proce-dure, the sample should contain only the elementalspecies present in the original matrix and, more-over, in the original ratio. So, before separatingand analyzing the sample it is necessary to definethe conditions under which the different species tobe analyzed remain stable. In practice, this means


mostly that the conditions during separation canvary only slightly from those occurring in theanalyzed matrix. This will be demonstrated usingthe stability for the vanadium–transferrin complexunder different conditions (pH, salt concentration,acetonitrile concentration, . . .) that are commonlyused during chromatography experiments. In thischapter we will first consider the presamplingprocedures. Afterwards we will describe the rec-ommended sampling and storage procedures forblood, urine and tissues. Then a more detaileddescription of contamination hazards and storageconditions for specific trace elements will be givenwith a discussion of preservation of vanadiumspecies at the end.

2 PRESAMPLING STEPS

2.1 Collection and storage of sampleinformation

Before starting the collection of samples it isimportant to define a priori the exact goal of theexperiment. This is the only reliable way to drawup a questionnaire about the donor and the sampleto provide all the relevant information needed toevaluate the final results. Such a questionnaire maycome in very handy when rather odd (outlying)results are obtained. The length of the question-naire depends on the kind of element and on theaim of the study. For example, when looking fordose–effect or dose–response relationships, someparameters are of more importance than they mightbe for the establishment of reference values. Ineither case, information about the identity of thesample, the time, place and method of sampling,and possible additives, as well as about the personwho carried out the sampling, should be gathered.Also a timed record of the solid food and liquidintake may be necessary to interpret certain results.More specific information on the content of such aquestionnaire, in particular for measuring referencevalues, is given in the article of Cornelis et al. [1].In the case of speciation analysis, information onparameters such as pH and conductivity shouldalso be collected at the time of sampling and,maybe even more important, on the addition of

chemicals during sampling. If possible, additivesshould be avoided.

During the set-up of a biological experiment,researchers should also be aware of a fundamentalcharacteristic of all forms of life called biologicalrhythmicity, which up to now has more oftenthan not been overlooked. It may be interestingto cite here the words by Burns et al. [2]: ‘dailystatistically significant fluctuations occur in allof the normal biological variables studied in theexperimental animals and humans. However, manyresearchers are not aware of the negative impactbiological rhythmicity can have on experimentaldesign and/or data interpretation’. One of the mostcommon pitfalls consists of data transfer fromthe diurnally inactive laboratory animal to thediurnally active human. This can be avoided toa certain extent by reversing the 12 : 12 hourslight : dark cycle from the laboratory animals(spread over 10–14 days) so that their active period(dark) coincides with the active period of theresearcher during daylight. Other pitfalls concernthe frequency of sampling, synchronization of thelaboratory animals, shifting of a certain rhythmand plotting data on an ‘hours after treatment’basis versus a ‘time of day’ basis. For experimentsdealing with, e.g. enzymes, these thoughts shouldbe certainly borne in mind.

2.2 Cleaning and evaluationof the instruments

Careful selection of the instruments during sam-pling and storage will reduce greatly the risk ofcontamination. A collection tube completely freeof trace element contaminants does not exist andneither is there a ‘standard’ collection tube thatpossesses ideal characteristics for all the trace ele-ments. For example, when doing speciation analy-sis of aluminum, collection tubes or instrumentsmade of glass cannot be used. Plastics with acadmium-based softener or zinc-doped stoppersare to be avoided when analyzing for cadmium andzinc, respectively. For each element these instru-ments should be evaluated separately. While somemetals, including cobalt, copper, iron and sele-nium present no significant risk of contamination,

PRESAMPLING STEPS 25

others, such as aluminum, cadmium, chromium,manganese and nickel pose quite a significantrisk [3]. They warrant attention with respect notonly to the sampling itself but also to the selec-tion of the material to be used. Table 2.2.1 givesan overview of the presence of trace elements inlaboratory ware [4]. This table can be used to get afirst impression of the possible contamination haz-ards that might occur for the trace element it isdesired to study. Evaluation of the containers orvessels and instruments can be done by differentmethods, but they all have some drawbacks.

A first, rather elaborate, way is to determinethe concentration of the element in the collectiontube and instruments. If available, a very suitabletechnique is NAA (neutron activation analysis).Another option consists of analyzing a certifiedsample, e.g. blood or plasma, after storage inthe container for a length of time (e.g. 5 daysat −20 ◦C), typical for the study, and containinga very low concentration of the element. If thecertified value is reproduced the contaminationcan be considered negligible. This implies that theanalytical procedure is under control, which againcan only be confirmed by the analysis of a bloodsample certified for about the same concentrationsof the trace element as found in the real samples.

This latter includes a major drawback for thismethod, because there are few elements certifiedin blood samples at the low concentration intervalneeded for this kind of study. Another, morecommon, possibility is to rinse the containers withmild acid (0.03 mol L−1 HNO3 or HCl), EDTA,or with a solution that includes the major ligands(amino acids and peptides) that are present inbiological matrices, and to analyze the leachingsolution. Table 2.2.2 gives a summary of theleaching of trace elements from laboratory ware inthe presence of 6 M HCl or 9 M HNO3. Table 2.2.3lists the concentration range of trace elements inultra pure acids [4]. In a way, rinsing with acid isirrelevant because the pH is too acidic comparedto biological matrices. Compilation of a solutionwith possible biological ligands is also not asstraightforward as it should be. To exclude therisk of contamination completely, a very elaboratecleaning procedure can be used as described byVersieck and Cornelis [5]. This procedure consistsof washing laboratory ware as follows: wash withdistilled water; soak for 2 days in 30 % H2O2;rinse with Milli-Q water; boil for 8 h in a 1 : 1(v/v%) mixture of 65 % nitric acid and 96 %sulfuric acid, both of Suprapur analytical grade;rinse with Milli-Q water; boil twice for 8 h in

Table 2.2.1. Trace elements in laboratory ware. Reprinted from Talanta, Vol. 29, Kosta, Contaminationas limiting . . ., pp. 985–992, 1982, with permission from Elsevier Science.

Material Concentration range [mg kg−1]

100 10–0.1 0.1–0.01 0.01–0.001

Polyethylene and Na, Zn, Ca, K, Br, Fe, Pb, Mn, Al, Sn, Cu, Sb,polypropylene (Al, Ti)a Cl, Si, Sr Se, I Co, HgPVC Na, Snb , Br, Pb, Sn, As, Sb –

Al, Ca Cd, Zn, MgTeflon K, Na Cl, Na, Al, W Fe, Cu, Mn, Cs, Co

Cr, NiPolycarbonate Cl, Br Al, Fe Co, Cr, Cu, –

Mn, Ni, PbGlass Al, K, Mg, Fe, Pb, B, Zn, Sb, Rb, La, Au, Sc, Tl, U, Y,

Mn, Sr Cu, Rb, Ti, Ga, (As, Co)c Inc

(Cr, Zn)c

Silica – Cld , Fe, K Br, Ni, Cu, Sb, Se, Th,Sb, Cr Mo, Cd, Mn,

Co, As, Cs,Ag

aAl and Ti high in low-pressure polyethylene (used as catalyst).bHeavy metal compounds used alternatively as stabilizers in certain types of PVC.cNot certified in the NBS reference material 617; determined by the cited author.d Chlorine only high in synthetic quartz.


Table 2.2.2. Trace elements leached by hydrochloric acid and nitric acid from plastic containers after 1 week of contact(in ng cm−2). Reprinted from Talanta, Vol. 29, Kosta, Contamination as limiting . . ., pp. 985–992, 1982, with permission fromElsevier Science.

Material 6 M HCl 9 M HNO3

10 10–1 1 10 10–1 1

Polyethylene Na, Pb, Al Tl, Cr, Zn Ca, Sn, Cu, K, Ca Na, Fe, Se, Cu Cr, Mg, Pb(HP) Mg, Ba, Ni, Cr,

Cd, Sr, SePolyethylene Ca Zn, Na, Al, Fe, Sn, Co, Ni, Pb, Na Zn, Fe, K, Pb, Mg, Ca, Cu,(LP) Cu, K, Ba Mg, Se, Cd, Sr Ni, Sn, Sr Se, Cr, Cd,

Te, Ag, BaPolycarbonate Fe, Ca, Sn, Na, Cu, Cr, Cd, Mg, Te, Se, Ni, Sr – Al, Na, Fe, Cu, Zn, Ni,

Pb Ba, Al Ca, K Se, Cr, Cd,Pb

Teflon Fe Cu, Zn, Cr, Al, Cd, Se, Ni Ca, Fe Mg, Al, Na, Zn, Cr, Te, Cd,Ba, Ca, Na, Pb, Pb, Ba, Cu, Ni, Sr, SeTe, K, Mg, Sn, Sr K, Sn

Elements are ranked in order of decreasing amounts introduced into the acid.

Table 2.2.3. Trace elements in ultrapure acids. Reprinted from Talanta, Vol. 29, Kosta, Contamination as limiting. . .,pp. 985–992, 1982, with permission from Elsevier Science.

Range (µg L−1) HCl HF HNO3 H2SO4 HClO4

1 Al, Si, S B, Si, P, S Si, S, K Ca, Co, Cu, K, Cr, Fe, NaMg, Na, (Se)

1–0.1 Na, Mg, P, Na, Al, Ti, Ca, Al, Ti, Fe, Na, P, Mn, Ni, Sn, Ni, Sn, Br, K,Ca, Fe K, Fe, Cu, Zn Mg, Ca, B, Cu, Cr Sr, Tl Pb, Tl, Zn

0.1–0.01 B, Ti, V, Cu, Cr, Mn, Co, Zr, Zn, Ni, Ba, Pb, – Cd, SrZn, Sn, Ba Cd, Pb Cd

<0.01 Mn, Co, Cr, Cd, Ba Mn, Co, V – –Ni, Zr, Cd, Pb

Milli-Q water; rinse with Milli-Q water, and finallysteam-clean for 6–8 h with Milli-Q water. Theequipment is then dried upside down on a Teflonfoil or polypropylene tray at 55 ◦C in an especiallyreserved oven. Afterwards, the instruments arestored in an airtight plastic transport container untiluse. All steps are carried out in clean laboratoryconditions (class 100) except boiling for 8 h inthe 1 : 1 mixture of concentrated acids. Other highpurity items are cleaned in a similar way. Whensteam cleaning is not possible, e.g., when the itemsare too small, they are boiled for an additional6–8 h in Milli-Q water. A shorter method consistsof using 0.5 % HNO3 or 1 % EDTA with a finalrinse in distilled water [6]. These procedures areundoubtedly too elaborate and too expensive forroutine laboratory work in hospitals and are notnecessary for all trace elements. Some authorseven consider these procedures not necessaryfor short periods of storage (24 h). They tested

several types of tubes, cleaned and uncleaned.The only significant contamination was an increaseof 6 % in the aluminum content of uncleanedheparinized tubes [7]. Besides, when evacuatedtube systems (ETS, ‘vacutainers’) are used forsample collection, and this is the case in the greatmajority of the hospitals, a washing procedurecannot be applied. Therefore, one of the previousdescribed techniques should be used to evaluatethis kind of container for their metal content atregular time intervals. This can be done by rinsingwith EDTA or with 10 % HCl [3]. The criterionof acceptability is a maximum contamination levelequivalent to 1 % of the normal level in thedecontaminating reagent used [3]. If occasionallyany other material is to be used, immersionfor 24 h in 20 % HCl at 50 ◦C, followed bythree to four rinsings with demineralized wateris recommended. Pineau et al. [3] describe someregulations for routine work in hospitals which

COLLECTION AND STORAGE OF BLOOD 27

should be effective in eliminating most of thepolluting factors: (1) avoid as much as possible,ordinary cloth and cotton clothing, the use ofcosmetics that are rich in zinc and aluminum,and metallic articles of jewelry; (2) during thesampling and analysis, the doors and windowsof the room should be kept closed, the airconditioning turned off and no smoking should beallowed; (3) no corrosion may be present and donot use any concentrated volatile acid (e.g. HCl) inthe work room; (4) wash the floor frequently withdistilled water, but not during analysis; (5) keep allreagents, containers and tubes in a dust-free area;(6) keep out all metallic objects.

A rather elegant way to circumvent mostproblems of contamination (as far as the metal isconcerned) is the use of a radiotracer. However,contamination of exogenous ligands still can occurand for ethical reasons the use of radiotracersis not always possible for experiments withhuman subjects. For nonhuman experiments andwhen the nuclear facilities are available, the useof a radiotracer should be considered becausethis offers some major advantages, e.g. for thedetection of very low concentrations of the metalspecies [8, 9].

In conclusion, there is no general solution toexclude the risk of contamination, and the gravityof this problem depends strongly on the elementthat is under study. To minimize the risk ofcontamination some recommendations are offeredbut it is up to the researcher’s clear mind toevaluate the potential risk for contamination andto take all the precautions needed.

3 COLLECTION AND STORAGEOF BLOOD

In recent literature, one is aware of the importanceof a reliable methodology to obtain a represen-tative blood sample. Therefore, sample collectionguidelines are drawn up in different analytical andmedicinal fields [1, 10, 11]. For the collection ofblood, ETS systems are the most widely useddrawing systems, although syringe systems havealso been used. A newer sampling technique, i.e.microdialysis, will be discussed simultaneously for

blood, urine and tissues in a separate paragraph.The use of syringe systems should be avoided inthe case of trace element analysis because theycan cause significant contamination, usually origi-nating from the upper part of the plunger made ofrubber. In particular, lead and manganese are sen-sitive to this kind of contamination. In addition, thetransfer of blood into a tube for centrifugation rep-resents another potential risk of contamination [3].The area of skin from where the blood will be col-lected is cleaned with Milli-Q water and ethanoland allowed to dry by evaporation. When deal-ing with experimental animals, these will in mostcases first be anesthetized with diethyl ether. It isbetter not to use narcotics that need to be injectedinto the body, because they can have a poten-tial affinity for the metal under study. When thepatient or the animal is continuously connected toa catheter provided with a ‘lock’ (dialysis treat-ments) one should aspirate (not infuse) the contentof this lock before sampling to avoid contamina-tion from heparin or other constituents that werepresent in the lock. Collection from sites near afunctioning graft, fistula or active intravenous lineshould be avoided [10]. After putting on gloves(powder free, otherwise can contain e.g. zinc),insertion of the needle into the vein can proceed.The pressure of stasis should be low. This can beaccomplished with the patient supine. If the veinsare not clearly visible a tourniquet can be applied,but no longer than 1 min in order to avoid hemo-concentration of the blood sample. Needles arecolor coded by gauge size and the larger the gaugesize, the smaller the needle. Sizes 19 through 23are most commonly used. Selection of a suitablegauge size is important because too small a gaugemay damage the blood cells and increase the riskof hemolysis. For the same reason the speed ofthe blood flow into the ETS tubes should be con-trolled. If this speed is too high the red blood cellsmay be damaged when they hit the tube wall, caus-ing hemolysis. The use of a stainless steel needlefor the collection of blood is generally not suitablewhen examining trace metals. Analysis of bloodillustrates that the highest contamination occursin the first 20 ml of blood sampled, especially foriron, chromium, nickel, cobalt and manganese. For


the measurement of these elements the alternativeis the use of a polypropylene intravenous can-nula, mounted on a trocar. Propylene and Tefloncatheters also do not induce contamination whenanalyzing cesium, molybdenum and vanadium ifa sufficient quantity of blood is withdrawn [12].Advocating siliconized needles for trace elementdeterminations can be very misleading, becausesome types of needles are only siliconized onthe outside. If so, the hazard of contamination ofthe blood sample remains a serious possibility. Thepotential contamination of the needle can be testedin the same manner as for the laboratory ware,e.g. by comparing the elemental concentrations inblood that has and has not been in contact with theneedle [1].

In general, polypropylene or Teflon cathetersshould preferably be used for venipuncture, andif this is not possible, siliconized needles canbe applied. One should bear in mind that theseneedles can release a little aluminum, chromiumand nickel. The use of stainless steel needles inhospitals is considered unrealistic because it wouldrequire an unacceptably large quantity of blood(>20 ml) to minimize contamination.

To collect the blood, there is a choice betweenopen systems and the previous mentioned evac-uated systems. For open systems it appears thatpolyethylene tubes provided with stoppers offerthe best guarantees for most trace elements. Ifstoppered polyethylene tubes are not available,polystyrene tubes may be used [3]. In the case ofthe evacuated systems, tubes produced before 1985showed potential contamination of chromium, iron,nickel and zinc originating from the rubber stop-pers. Tubes equipped with siliconized stoppers(special trace element tubes) show less contamina-tion, e.g., for zinc [13], but it is suggested that, forother difficult elements such as aluminum, cobalt,chromium, nickel and manganese, vacuum tubes,even those especially developed for trace elementanalysis, should not be used. They can be used,with caution for lead. Vacuum tubes containing aserum separator (e.g. a gel) are generally rejectedfrom trace element analysis [3].

The use of an anticoagulant is very problematic,as most anticoagulants are either polyanions (e.g.

heparin) or metal chelators (e.g. EDTA or citrate)and therefore have a high affinity for metals. Forspeciation research this is disadvantageous in twoways. Firstly, because of their great affinity formost metals, these ligands can bind metal ionsoriginating from the wall of the container or anyother exogenous metal ion and thus contaminatethe sample to a greater extent. Therefore, thiscontamination hazard must be evaluated for eachparticular element under investigation and theblank value must be reported for each batch ofanticoagulant. Secondly, metal complexes presentin the original sample, but with a lower stabilityconstant, will be destroyed through addition ofthese anticoagulants. In this way, the results willnot reflect the original composition of the sample.Therefore, the use of anticoagulants (and otheradditives) should be avoided as much as possiblein order to preserve the original condition of thesample. So, speciation of metal species in serumsamples is to be preferred to speciation in plasmasamples. ETS tubes with a red stopper contain noadditives (Vacutainer, Becton Dickinson), whiletubes with a green or lavender stopper containrespectively heparin and EDTA. The use of tubeswithout additives is strongly recommended.

When feasible, in the case of adults, the firstmilliliters of blood will serve to rinse the needleand will not be used for trace element speciation,but kept apart, e.g. for clinical analyses. The bloodis allowed to clot spontaneously and the samplesare transported to a clean laboratory. After clot-ting, the serum is separated from the erythrocytesby centrifugation at 2500 rpm for 20 min. After thefirst spin, serum is pipetted into other clean tubesand centrifuged for a second time at 2500 rpm for15 min to remove remaining blood cells. Finally,the serum is pipetted in clean tubes and stored at<5 ◦C (short periods only) or frozen at −20 ◦C orless (plastic tubes only) [5]. Pipet tips used to aspi-rate serum samples should be made of propyleneor polycarbonate. These tips have been shown torelease cadmium, iron, nickel, chromium, molyb-denum, palladium and mercury and should there-fore be rinsed with 10 % HCl and demineralizedwater and tested for absence of contamination [3].If possible, analysis of the samples should occur

COLLECTION AND STORAGE OF URINE 29

on fresh material within a few days after sampling.If not, samples should be deep-frozen until analy-sis immediately after sampling. Standardization ofthe clotting and separation procedures and avoid-ance of hemolysis are important. Whole bloodsamples that are stored at −20 ◦C (>24 h) will behemolyzed [14]. Hemolysis can result in increasedconcentrations of some elements, e.g. lead, man-ganese and zinc, in serum due to release of theseelements out of the packed cells where these arepresent in a much higher concentration. Thereforeserum is separated from blood cells before freez-ing and storage. The degree of hemolysis shouldbe assessed by measuring hemoglobin concentra-tions in serum samples and a criterion for rejectionof hemolyzed samples should be put forward.

4 COLLECTION AND STORAGEOF URINE

Guidelines also exist for the collection of urine [1].Collection of urine poses more problems than tak-ing blood samples because of the many potentialsources of contamination, either in the environ-ment (e.g. occupational medicine in a pollutedenvironment) or in the method of collection andthe container used [3]. In our laboratory we triedto collect urine of rats directly from the bladderwith the use of the ETS system after anesthetiz-ing and before dissecting the animals. The pur-pose of this technique was to minimize the riskof contamination. However, this technique wasnot successful because most of the time the blad-der turned out to be empty. This was probablydue to a reflex of the rats to empty their blad-der in stress situations. We solved this problem byusing metabolic cages for the separate collectionof urine and feces. By doing this, the chance forcontamination seriously increases. Dust or otherforeign material in the neighborhood can fall intothe collection vessels. We partially circumventedthis problem by restricting the sampling intervalsto periods of 1 h and by the use of a radiotracer.When sampling urine of humans, a more rigorouscollection method should be applied. Depending onthe purpose and the circumstances of the measure-ment, sampling can request only one spot sample

or the collection period can be extended to 24 h.All time intervals in between can be used. Someauthors favor successive and separate collectionsin order to limit contamination. Afterwards, thesesamples are mixed together in the laboratory [3].On the other hand, short collection intervals (with-out mixing) can give additional information abouta possible rhythmicity of an element’s behavior. Inmorning urine, the element concentration is oftenrelatively high. In case it is impossible to col-lect 24 h samples, some authors suggest to use thismorning urine and to correlate the concentrationsof the elements to the creatinine elimination. Thiswould compensate for the effects of dilution [15].

Urine should be voided directly into an acid-washed polyethylene container that can be closedwith an airtight lid. The subject should be instructedto minimize contamination of the sample by avoid-ing contact with the inside of the container or lidand to close the container immediately after voidingthe urine. In between sample sessions the containershould be wrapped into a polyethylene bag. In gen-eral it is recommended that measurement of theparameters given in Table 2.2.4 are included. Whensome samples show values outside the expectedrange this should be mentioned in the sample report.

After collection of the sample it is advisableto divide the urine into subsamples (differentaliquots) after vigorous shaking for a few minutes.The samples should be kept at 4 ◦C in the refrigera-tor (short period) or frozen at −20 ◦C. The stabilityof the different metal compounds is species depen-dent (see below). Normally, precipitation of saltsand organic compounds occurs resulting in copre-cipitation of several trace elements. Often, urinesamples are stored in the presence of a preserva-tive (0.03 mol L−1 HCl or HNO3, sulfamic acid,

Table 2.2.4. Recommended parameters to bemeasured for urine collection.

Parameter Test/range

Sugar Negative by strip testProteinuria Negative by strip testUTI (Urinary

tract infection)Negative for nitrite

producing bacteriaUrinary density 1.012–1.030 (reject

outlying samples)Creatinine 7–17 mmol 24 h−1


Triton X-100). This should, if possible, be avoidedin speciation analysis because many metal–proteincomplexes are only stable in a well-defined pHinterval (mostly around physiological pH). Addi-tion of an acid will destabilize the metal–proteincomplex. In one of the following paragraphs it willbe clearly demonstrated that it is essential in speci-ation analysis to maintain as much as possible thephysiological conditions during storage and subse-quent steps in order to avoid artifacts. However,in the case of aluminum and mercury, it has beenshown that, even in frozen samples, the concen-tration rapidly decreases during the first few daysif no additive is present [16]. In these cases, if anadditive is necessary, both its influence on the sta-bility of the present species and its purity shouldbe checked.

5 COLLECTION AND STORAGEOF TISSUES

For collection of tissues there is an even greaterrisk of contamination when compared with the col-lection of biological fluids such as serum or urine.This is due to the increased amount of handling(cutting, cleaning, homogenization) needed to sam-ple a tissue. A possible solution is microdialysis(see below). Extensive research has been carriedout to assess the possible adventitious addition oftrace elements to the sample. It was found thatcontamination of biopsies taken from the liver bymeans of needle aspiration techniques was muchhigher compared with biopsies taken with surgicalblades. For iron, manganese, copper, zinc, cobalt,chromium and nickel significant additions weredetected with the needle aspiration technique, insome cases higher than the natural levels of the ele-ments in the sample [12]. Potential additions fromsteel scalpels were also examined. It was observedthat chromium contamination of frozen muscle tis-sue samples is about ten times greater than forfresh samples because of the greater friction in cut-ting. Addition of manganese, antimony and tung-sten was also found [17]. It is also important toadd that other steps in obtaining and handling biop-sies, e.g. taking them with a pair of tweezers, may

introduce additional errors. In our laboratory, wesample tissues using a surgical blade or a pairof scissors. After dissection of the animal, organsare rigorously cleaned by removing remnants offat and connective tissue with a pair of tweezersand a pair of scissors. Depending on the elementthese instruments can be made out of stainless steelor plastic. Afterwards, the tissues are washed fivetimes with a 0.9 % saline solution at physiolog-ical pH (7–7.5). Finally, organs and tissues arestored in plastic vessels with tightly fitting screwcaps in a refrigerator at 4 ◦C for very short periods(hours) or in the deep freezer at −20 ◦C for longerperiods. Tissues spoil at temperatures even slightlyabove freezing point because chemical reactions(e.g. enzymatic activity) continue to occur withintissues after the death of an animal. This can evenhappen in a frozen tissue, unless kept extremelycold [18]. Therefore it is very important to per-form the analysis as quickly as possible or to

Glassvessel

Teflonpestle

Figure 2.2.1. Evelhjem–Potter homogenizator. Minced tissueis placed in the glass vessel and buffer is added. The tissue ismildly homogenized by moving the Teflon pestle up and downin the glass vessel.

MICRODIALYSIS 31

store the tissue in the deepfreezer after sampling.Lyophilization of tissues can be another option, butits influence on the stability has to be checked [19].At the time of analysis, whole organs (spleen, kid-ney, testes) or parts of an organ (liver) are first cutinto small pieces. In our laboratory these piecesare further homogenized by means of an Evel-hjem–Potter homogenizator (Figure 2.2.1). Thissimple instrument is made of a glass vessel (like acold-finger) and a Teflon pestle of slightly smallerdiameter. Tissue and some buffer are placed inthe glass receptacle. By moving the Teflon pes-tle up and down in the glass vessel the tissuecells are crushed between the pestle and the glasswall. By repeating this handling for a few minutesthe tissue is homogenized in a very mild manner.Afterwards you can obtain any cell fraction (mem-branes, mitochondria, lysosomes, cytosol, . . .) bydifferential centrifugation. Because of the frictionwith the glass wall during homogenization, pos-sible contamination arising from this glass wallshould be checked a priori for the element underinvestigation.

6 MICRODIALYSIS

Microdialysis is a relatively new technique that hasbeen applied extensively in neurosciences to mea-sure neurochemicals in vivo [20–22]. Recently theuse of this technique has been extended to otherbiological fields. In theory, microdialysis mimicsthe passive function of a capillary blood vessel.A small diameter probe (different geometries andsizes available) containing a dialysis membraneis implanted into a tissue (in contact with theextracellular fluid) or a vessel that is continuouslyperfused with a suitable physiological fluid at alow flow rate with the aid of a pump. Molecularsubstances with a molecular weight smaller thanthe cut-off value of the dialysis membrane diffusealong a concentration gradient in or out the probe,depending on the relative concentration on bothsides of the membrane. If the concentration of theanalyte in the perfusion fluid is lower than that inthe tissue itself, the analytes move from the tissueinto the probe and consequently they are collected

in the perfusate and carried out of the body. Thistechnique offers many advantages compared toclassical sampling techniques, i.e. the withdrawalof blood samples or tissue fractions. Substancescan be continuously (dynamically) sampled with-out removing or altering the balance of body flu-ids. As a function of time, better resolutions areobtained, as sampling intervals are relatively short.Typical flow rates range from 0.5 to 10 µL min−1

with 2 µL min−1 being most commonly used. For acommon sampling interval of 5–10 min the sam-ple volume is restricted to 10–20 µL. To obtainoptimal results, sampling time and perfusion flowrate should be balanced. Also multiple microdial-ysis in the same animal is possible, e.g. by placingone probe in a tissue and another in a blood vessel.In this way the passage of a component from thetissue to the vessel or vice versa can be studied.Overall, fewer animals are needed to obtain time-dependent curves. Tissue areas as small as 1 mm3

can be sampled in either conscious (freely moving)or anesthetized animals. Sample recovery dependson a variety of factors such as sort of tissue, probedesign, flow rate, temperature, membrane surfaceand membrane cut-off value. Probes with a cut-off value ranging from 5 kDa up to 100 kDa areavailable. However, dialysis efficiency decreasesdramatically as the molecular weight of the analyteincreases. Because of the use of a semipermeablemembrane with a low cut-off value, proteins anddegradative enzymes are excluded from the sam-ple. In this way, sample degradation by enzymesis less common and because of the absence ofproteins there is no need for a sample clean-up:the sample can be directly injected into analyticalinstruments for on-line analysis.

The microdialysis probes are manufactured outof biocompatible components. This reduces therisk for contamination of metals in trace ele-ment analysis. Most materials are polycarbon-ate–ether polymer, cellulose (acetate) and poly-acrylonitrile for the membranes and fused silica,Teflon or polyethylether ketone for the tubing.The outer diameter of the probes ranges from250 to 500 µm. The dialysis membranes are usu-ally 1–10 mm long. Procedures for implantationof the probes into the tissues are provided by


the manufacturers. In most cases, this involvesthe insertion of a cannula through which a probeis placed. In general, the implantation techniquedepends on the durability of the probe and itsmembrane. Insertion of a probe causes increasedblood flow in the surrounding area. Therefore, afterinsertion an equilibration period should be allowedbefore starting the sampling. This recovery timedepends on the type of tissue, probe and degree oftrauma induced. In muscle, no histological changeswere observed less than 6 h after insertion. Morethan 6 h after implantation leukocyte infiltrationstarted and continued for 24 h without morphologi-cal changes. After 30 h, scar tissue starts to developaround the probe. This alters the diffusion rate andthe probe recovery. It can be concluded that forstudies up to 24 h, sampling in some tissues can bedone without changes in recovery. For long-term

studies, special consideration should be given toscar tissue formation. For sampling tumor tissue,there seem to be fewer problems concerning leuko-cytes infiltration or tissue changes.

Different probe designs are commercially avail-able. The design is specific for the tissue beingstudied. In Figure 2.2.2 some designs and their useare depicted. Linear probes are used to sampleperipheral tissues (dermal tissue, muscle, adi-pose tissue, liver and other organs) with minimaldamage [23]. Side by side probes and vascularprobes are best suited for intravenous sampling.For intracerebral implantations rigid concentricprobes are the model of choice. The advantageof loop probes is their larger surface area thatmakes it possible to enlarge the dialysis surfacewithout lengthening the probe too much. Thereforethese are best suited for sampling subcutaneous

(a)

Temporary Plug(protects probe during implant)

Membrane

Guide Fiber

Guide Fiber

(b)

Split IntroducerNeedle

VeinSealed Tubing

Probe

Pull and SplitWeb MeshSutured 10 Muscle

Figure 2.2.2. (a) The linear probe consists of a short length of hollow dialysis fiber attached to narrow-bore inlet and outlet tubes.An aqueous perfusion solution, which closely matches the ionic composition of the surrounding extracellular fluid, is pumpedthrough the probe at a constant flow rate. Low molecular weight analytes diffuse in or out of the probe lumen. Large moleculessuch as proteins or protein-bound analytes are excluded by the membrane. Molecules entering the lumen are swept away by theperfusion fluid. This dialysate is then collected for analysis. (Reproduced by permission of BAS, Inc.). (b) IV vascular probeswere designed for implantation into the rat jugular vein. They are also suitable for other soft tissues. Each probe includes asyringe needle and temporary cannula (split introducer) which aid placement. The thin-walled, plastic introducer slides over thesyringe needle, which is then used to pierce the vein. The needle is removed and replaced with an IV probe. A flexible wiremesh on the probe is sutured to the pectoral muscle. The cannula is then pulled out of the vein, leaving the probe behind [23].(Reproduced by permission of BAS, Inc.)

SPECIFIC PRECAUTIONS FOR SOME ELEMENTS 33

tissue or the peritoneal cavity. The researcher canchoose from these designs or develop their ownmicrodialysis probe fit for purpose. Custom-madeprobes can also be ordered by some manufacturers.

However, this new sampling method also hassome drawbacks. First of all, this sampling tech-nique requires sensitive analytical techniques todetect the low concentrations in the small sam-ple volumes. The technique is also less suited forhigh molecular weight molecules as they have aslow diffusion rate through the membrane. Themembrane itself can exhibit adsorption for somemolecules due to residual charges or membranecomposition that limit sample recovery. A minimalsurgical know-how is needed to implant the probein a correct and, for the animal (or human), safemanner. Damage of the peripheral tissue has to beavoided as much as possible. Most common effectsare direct trauma, circulatory effects (short term)and foreign body effects (longer term). All thesefacts can influence sample recovery. This bringsus to another great disadvantage, i.e. lack of quan-titative recovery. Although the amount of analytedoes not vary at a particular sampling site (in a lim-ited sampling interval), recovery varies from siteto site due to, e.g., differences in tissue volume.Also, within a tissue, there are differences in tis-sue density. There exist already some calibrationprocedures and, when used correctly, these yield agood approximation of the analyte concentration.However, there is still a debate going on aboutthe validity of these calibration procedures. Otherdoubts prevail about changes in tissue permeabil-ity induced by inserting the probe, the limitedsample area and possible changes in tissue tortu-osity at the probe location. If these problems aresolved, microdialysis will become a routine sam-pling method in the clinical research laboratory.

7 SPECIFIC PRECAUTIONSFOR SOME ELEMENTS

In this section the stability and storage of indi-vidual trace elements is considered. Gomez-Ariza et al. [24], Quevauviller et al. [25], Daset al. [26] and Cornelis et al. [1] have published

some excellent reviews during recent years wherethe reader may find additional information.

7.1 Aluminum

Because 8 % of the earth’s crust consists ofaluminum, exposure to this element is ubiquitous.This also implies that the risk for contaminationis real. Dust on the sample should be avoidedat all stages. Therefore sample separation shouldbe done in a clean room (class 100). Samplesshould be collected in a similar environment andfor venipuncture talc-free gloves should be worn.Water and reagents are other sources of aluminumcontamination. Blank values for aluminum shouldbe measured in these reagents. All glass andplastic ware must be thoroughly washed withacid or EDTA solutions and then checked fortheir contributions of aluminum to the sample.From a recent study it is clear that in Tefloncontainers there is no loss of aluminum for a(relatively short) storage period of 2 h [27]. Alsopolyethylene containers can be safely used [25].In the same report it is stated that the stabilizationand long-term storage of natural samples at 4 ◦Cwas difficult and unachievable in practice withoutsome type of pretreatment, e.g. addition of acid ora complexing agent, and that there is a need foradditional investigations for stabilization, e.g. bychemical buffering systems.

Because Al(III) is a ‘hard’ trivalent metalion, it binds strongly to oxygen-donor ligandssuch as citrate and phosphate. Therefore it isimportant that no additives containing this type ofligand are used during speciation experiments foraluminum. Serum fractionation studies show thatmost aluminum is protein bound, primarily to thetransport protein transferrin. Albumin appears toplay no role in serum transport [28]. For the lowmolecular weight fraction there is little agreement,although some reports indicate that citrate plays asignificant role [29]. In a more recent article thespeciation of aluminum in various biofluids andtissues is discussed [30].

In serum, aluminum levels of about 1–10 µg L−1

can be expected in healthy persons. In urine, levelsfluctuate around 10 µg L−1 for healthy persons [1].


7.2 Antimony

Antimony is a relative toxic element. Trivalentantimony is about ten times more toxic than pen-tavalent antimony. Organic forms of antimonyare less toxic than the inorganic forms. A sta-bility study showed that 25 µg L−1 antimony(III)can be stabilized at 40 ◦C for 12 months in anaqueous medium, lactic acid 0.1 mol L−1, or citricacid 0.05 mol L−1. These solutions were stored inpolyethylene bottles. Oxidation of antimony(III) toantimony(V) was likely to occur after 1 month ofstorage in an ascorbic acid solution (0.06 mol L−1)even when stored at 4 ◦C. After 3 months of stor-age antimony(III) oxidized in aqueous media andalready after 1 month in lactic acid at 25 ◦C. Betterresults were obtained in citric acid although it didnot completely prevent antimony(III) oxidation.Because the total antimony content (50 µg L−1)remained constant at 4 and 25 ◦C it was concludedthat polyethylene is suited to prevent adsorptiononto the vessel walls [31]. The use of weak acidmedia has to be avoided owing to the high insta-bility caused by the hydrolysis of antimony [24,31, 32]. Another preservation procedure is basedon solid-phase extraction using different solid sor-bents such as activated carbon or graphite [33].

Antimony is found in serum in concentrationsranging from 0.07 to 0.76 µg L−1 [34].

7.3 Arsenic

The collection of blood and urine for arsenic mea-surements is also very sensitive to contaminationfrom arsenic in reagents, dust and laboratory ware,so the same precautions should be taken as for alu-minum. However, if contamination occurs it willmost probably be in the form of inorganic arsenicand not organic arsenic which is metabolized, e.g.in animal tissues. In blood, arsenic is expectedto be stable at −20 ◦C. An extensive study onthe stability of common arsenic species such asarsenite [As(III)], arsenate [As(V)], monomethy-larsonic acid (MMA), dimethylarsinic acid (DMA)and arsenobetaine in urine shows that low tem-perature conditions (4 and −20 ◦C) are suitable

for the storage of samples for up to 2 months.For longer periods (4–8 months) the stability ofthe arsenic species was dependent on the urinematrix. Whereas the arsenic speciation in someurine samples was stable for 8 months at both 4 and−20 ◦C, other urine samples stored under identicalconditions showed substantial changes in the con-centration of arsenic(III), arsenic(V), MMA andDMA. The use of additives did not improve thestability of arsenic species in urine [35]. More-over, the addition of 0.1 mol L−1 HCl to urinesamples produced relative changes in inorganicarsenic(III) and arsenic(V) concentrations. Earlierstudies reported that 50 % of arsenite(III) was oxi-dized to arsenate(V) by dissolved air after 33 daysof storage [36]. Therefore, several acids have beenproposed for arsenic stabilization in water sam-ples, but as mentioned previously this should bemostly avoided in speciation analysis. A compari-son between storage of solutions at room temper-ature and 4 ◦C revealed that solutions of organicarsenic species were stable during long-time stor-age, while solutions of inorganic species wereonly stable during refrigerated storage [37]. Theformer BCR (Community Bureau of Reference,Measurements and Testing Programme) has alsoundertaken projects to evaluate the stability ofarsenic compounds [38]. The results showed thatpure solutions of arsenic(III), arsenic(V), MMA,DMA, arsenobetaine and arsenocholine were sta-ble for up to 1 year if they were stored in thedark and if the pH values were properly adjusted.No degradation was seen even if the temperatureincreased to 40 ◦C. However, degradation was seenwhen mixtures of arsenic species were prepared,with and without inorganic salts. Another studywith a mixture of arsenic species showed oxidationand methylation reactions [39]. After 4 months ofstorage at 20 ◦C in the presence of light, all arsen-ite(III) was converted to arsenate(V). An identi-cal solution stored in the dark at 40 ◦C showedonly slight degradation of arsenic(III) (due to lowmicrobiological activity). Similar experiments withmixtures of arsenic(V), DMA and arsenocholinestored at both 20 and 40 ◦C revealed the produc-tion of MMA after 2 months, while arsenic(III)was formed after 4 months at 20 ◦C. At 40 ◦C


the formation of arsenic(III) was observed after2 months and disappeared after 4 months. Whenthe mixture was stored at 4 ◦C no changes in thefate of the species were observed. In case of thestability of DMA and arsenobetaine in a freeze-dried tunafish, there were no changes seen in sta-bility for 9 months at either 20 or 40 ◦C [40].

Concentrations in serum of healthy persons varybetween 1 and 5 µg L−1, and this level dependson the level of seafood intake. In urine, valuesfor arsenic are around 10 µg L−1 for Europeancitizens. In Japan, however, concentrations can befive times higher [1].

7.4 Cadmium

Cadmium is often used in pigments and as a soft-ener in plastics. For this reason, sample contactwith colored stoppers and certain plastics duringsampling and processing must be avoided. Glassshould also be avoided. In a rainwater sample thatwas stored at 22 ◦C in a HDPE bottle, about 40 %of the dissolved cadmium was lost, probably dueto adsorption on the walls [27]. It has also beenshown that this loss happens during the first 30 minof storage. Storage in a Teflon bottle showed nolosses of cadmium. The person who performs thesampling should wear talc-free gloves. No specialneedle is required because stainless steel needlesdo not seem to release cadmium. Plastic syringesand test tubes should be cleaned and tested fortheir ability to release cadmium into the sample.Another important contamination source for cad-mium is smoking. Preferably, sample collectionand handling should be done by a nonsmokingperson.

Speciation of cadmium in liver indicates thatcadmium binds to different protein fractions(>400 kDa, 70 kDa and metallothionein, [41]. Inthe case of a freshly prepared solution cadmiumbinds to two different isoforms of rabbit metal-lothionein as is clearly visible in a CE-electrophe-rogram [42]. However, when the solution wasallowed to stand for 2 weeks without refrigerationit is obvious that the metallothionein has degradedas is apparent from the lack of well-defined indi-vidual components in a CE-electropherogram. For

free cadmium, the influence of pH has also beendemonstrated [42]. At pH values less than 8, morethan 6 mM of free cadmium can be present in solu-tion without significant hydroxide formation. AtpH values >8 the signal of free cadmium signifi-cantly decreases.

Cadmium concentrations in blood are generallyin the range of 0.1–2 µg L−1. Concentrations inurine are usually <1 µg L−1 [1].

7.5 Chromium

For the determination of chromium strict guide-lines should be applied in order to obtain reliabledata. In contrast to cadmium, no stainless steelneedle can be used in the case of chromium. Apropylene cannula is compulsory. The first 20 mlof blood cannot be used for chromium analy-ses. All tubes and plastic ware should be acidwashed before sample collection. Unwashed tubeswill invariably lead to a too high blank value forchromium. As sweat contains about ten times morechromium than does serum, it is important to avoidcontact of the sample with the skin.

Chromium(VI) is much more toxic than chro-mium(III) and of great concern in public health.The stability of chromium(III) and chromium(VI)in solution was thoroughly investigated as partof a BCR project [25, 43]. Significant losses ofchromium(VI) at 20 ◦C were observed in solu-tions stored in PTFE containers after the addi-tion of HCO3

−/H2CO3 buffer. The stability ofboth chromium(III) and chromium(VI) was sat-isfactory after 228 days of storage in quartzampoules at 5 ◦C at concentrations of 40 µg L−1

of chromium(III) and 10 µg L−1 of chromium(VI).In this case the addition of HCO3

−/H2CO3 bufferunder CO2 at pH 6.4 was necessary to avoidchromium(VI) reduction. A promising develop-ment shown in this project is the preparationof freeze-dried solutions containing chromium(III)and chromium(VI). The stability of these com-pounds was demonstrated over a period of 88 daysfor freeze-dried solutions stored at −20 ◦C afterreconstitution in HCO3

−/H2CO3 buffer underCO2 at pH 6.4. The stability at 20 ◦C has alsobeen verified [43]. The redox potential of the


chromium(VI)/chromium(III) system depends onthe pH of the medium. Chromium(VI) salts(dichromates) are very strong oxidants in suf-ficiently acid media. Their oxidizing strengthdecreases significantly with increasing pH whichmay explain their long-term stability when solu-tions are kept at a constant pH of 6.4 [25]. In arecent report, storage of water samples was donein the dark at 4 ◦C at neutral pH and the analysiswas done as fast as possible [44].

Chromium is known to be mainly bound totransferrin and albumin in serum [45]. In urine,serum and liver also a low molecular weight wasfound. In urine, this low molecular weight complexis the most abundant chromium species [46].

The chromium level in serum is about 0.1–0.2µg L−1. The concentration of chromium in urineof nonexposed individuals is below 1 µg L−1 [1].

7.6 Cobalt

No stainless steel needles can be used for thedetermination of cobalt values in blood. All thereceptacles should be acid washed. Cobalt isstable in blood for many years at −80 ◦C. Ifpossible, determination of cobalt should happen ina class 100 clean room. Additional contaminationcan arise from jewelry or dental prostheses. Itis of major importance to carefully wash theskin before sampling takes place. Because cobaltis an element with a short biological half-lifeit is important to know the exact time lapsebetween the beginning of the exposure and timeof sampling. A sample protocol with small timeintervals is advised.

Cobalt is an essential nutrient and a compo-nent of vitamin B12. Next to vitamin B12 thereare some cyanocobalamin analogs such as adeno-sylcobalamin, methylcobalamin and hydroxocobal-amin [47]. In urine cobalt is excreted both in theinorganic form and in organic forms [48]. The sta-bility of vitamin B12 is highest in the pH range4.5–5 [42]. When working at pH 9 it is likely thatcobalt is removed from the porphyrin structure inorder to form Co(OH)2. Oxidation of adenosyl-cobalamin in solution has also been reported [42].A solution of 1 mg L−1 in a deoxygenated buffer

that was allowed to stand at ambient temperaturefor 10 min showed already the oxidation productin its chromatogram. At 108 min after preparationthe majority of the adenosylcobalamin was in theform of the oxidized analog.

It appears that values of urine cobalt are inthe range 0.1–1 µg L−1 with values in serum andblood at the lower end of this range [49].

7.7 Copper

Recovery experiments with copper in aqueous solu-tion have shown that about 80 % of dissolved cop-per in HDPE containers is lost, probably due toadsorption on the walls of the container. This lossmainly occurs during the first 0.5 h of storage [27].In polypropylene containers the same effect has beendemonstrated. In Teflon containers, the concentra-tion of copper remained constant within the experi-mental error during the 2 h of measurement.

In serum, copper is bound to ceruloplasmin(160 kDa) and albumin (66 kDa) [26, 50]. In amore recent report, a small fraction of the copperis eluting in the dead volume of the column. It issuggested that this fraction consists of either freepositively charged copper ions or weakly boundcopper [51]. Copper ions added in vitro to a serumsample will result in an increase of the albuminbound fraction [52].

Copper serum concentrations range from 0.8 to1.4 mg L−1. In urine, copper concentrations around0.2 mg L−1 have been measured [1].

7.8 Lead

Sampling of blood for lead measurements can bedone using disposable sampling devices and con-tainers. However, the lead blank caused by chemi-cals and materials must be sufficiently below thelead concentration in serum in order to avoid mis-leading results. In the case of serum samples it is ofparamount importance to avoid hemolysis duringblood collection and subsequent serum separation.This is needed because 10 % of the lead concentra-tion in whole blood is situated in serum, while theremaining 90 % is concentrated in the packed cells.


This means that even marginal hemolysis will ele-vate the lead concentration in serum by a factorof 2 or more. Therefore it is necessary to estimatefor all serum samples the degree of hemolysis bymeasuring the hemoglobin concentration.

In serum, lead seems to elute in the fractionascribed to ceruloplasmin and ferritin [51, 53,54]. In a recent article, lead eluted in a fractionwith different chains of immunoglobulines andlow molecular weight components of serum, butno identification of the binding complex is putforward [51]. Lead peaks were seen only in serumof uremic patients.

Because of the environmental concern regard-ing organolead complexes, the stability of somealkyllead compounds in blood during storageat different temperatures has been studied [55].Spiked blood was stored prior to the analysisat room temperature, 4, −20 and −70 ◦C. Theresults showed that samples can be stored at 4 ◦Cfor 1 week, at −20 ◦C for 2 months and in adeepfreezer for at least 1 year. To secure the sta-bility of the species during transport, a polystyrenebox with dry ice was used, which preserved thespecies integrity for 2 days. The stabilities of theseorganolead compounds have also been investigatedin an aqueous solution [56]. Solutions containing500 ng L−1 of trimethyllead and triethyllead werestable for 3 months, but UV irradiation producedrapid decomposition of these lead species at con-centrations of 1.5 µg L−1 lead and 2.9 µg L−1 lead,respectively. Dimethyl- and diethyllead speciesdecomposed less rapidly under similar conditionsand degradation of trialkyllead solutions in day-light was also observed. Trimethyllead was fairlystable over a period of 12 months. It can be con-cluded that storage in the darkness of alkyllead andespecially of trialkyllead is advisable for preserva-tion of the species [24].

As for copper, lead in aqueous samples alsoshows a significant loss due to absorption on thewall of polypropylene or polyethylene bottles [27].In Teflon bottles, no loss of lead occurs.

The lead concentration in blood varies from 165to 296 µg L−1. The measured concentrations forlead depend on the year of sampling (use of leadedgasoline) and the sampled region [1].

7.9 Lithium

In the case of lithium no special precautions seemto be needed for the collection and storage ofsamples in relation to possible contamination. Awell-documented history of the sample is neces-sary for evaluation of the results. In the literature,no reports on speciation of lithium in serum orurine could be found. There are some reports on thedetermination of low concentrations in biologicalsamples and on the effect of lithium administrationduring the treatment of acute mania [57, 58].

In serum of healthy people, lithium concen-trations are around the 1 µg L−1 level. In urine,normal excretion is up to 60 µg over a period of24 h [57].

7.10 Manganese

Manganese is a very difficult element to measurein clinical samples because of the multiple pos-sibilities of contamination. One of the difficultiesis obtaining water pure enough for dilutions withvery low manganese content. The subsequent prob-lem is to conserve purified water, because con-tamination from ambient air or material occursquite rapidly. For example, the absorbance of theblank increased substantially after 2 to 3 h in anauto sampler vessel, even when the auto sam-pler was covered [59]. As with lead, most of themanganese in blood is concentrated in the packedcells, so hemolysis should be avoid. Contaminationoriginating from dust particles, can be eliminatedby working in clean room (class 100) conditions.Stainless steel needles cannot be used since theyleach manganese into the blood serum. There-fore the use of a Teflon cannula is recommended.Also all the vials and syringes need to be testedfor manganese contamination. As in the case ofchromium, sweat contains a lot of manganese andtherefore cleaning of the skin is compulsory. Eventap water can contain high amounts of manganese(>1 mg L−1).

In serum, the major fraction of manganeseco-elutes with a UV peak of unidentified serumcomponents, comprising different chains of immu-noglobulines and low molecular weight complexes.


A small amount also elutes from the column in thedead volume, probably as free (solvated) ions [51].In size-exclusion experiments, manganese wasfound in serum fractions corresponding to differentmolecular weight proteins, but no identificationwas done [60]. Recently, the first two mononuclearmanganese citrate complexes were synthesized inaqueous solution near physiological values [61].These manganese citrate species can be relevantto manganese speciation in biological media andpotentially related to the beneficial as well as toxiceffects of manganese on humans.

The manganese concentration in the serumof healthy persons is 0.5 µg L−1. In urine ofnonexposed persons the manganese concentrationamounts to 1 µg L−1 [1].

7.11 Mercury

For mercury, cleaning of the sample containers andchecking of their possible contribution to the mer-cury level in the sample is compulsory. It is gen-erally stated that water samples for mercury deter-mination should be stored in glass bottles and thatpolyethylene containers are considered unsuitable.Recently, 300 ml poly(ethylene terephthalate) con-tainers have been recommended for the samplingand storage of potable water [62]. Adding 0.5 mlof 20 % m/v potassium dichromate dissolved innitric acid prevented loss of mercury. In anotherreport on storage experiments in various contain-ers, it was shown that organomercury species werestable for at least 30 days in all containers, exceptthose made of polyethylene. Metallic mercury wasstable in all containers except those made of stain-less steel or polyethylene. Mercury(II) was rapidlylost from all containers except those made ofaluminum, which rapidly converted mercury(II)to metallic mercury, which was stable [63]. Ina recent report, the stability of methylmercury(highly toxic and an accumulator in the food chain)and inorganic mercury retained on yeast–silicagel microcolumns was tested and compared withthe stability of these species in solution [64]. Thecolumns were stored for 2 months at −20 ◦C, 4 ◦Cand room temperature. Methylmercury was found

to be stable in the columns over the 2 month periodat the three temperatures tested while the con-centration of inorganic mercury decreased after1 week of storage even at −20 ◦C. Formation ofmethylmercury and dimethylmercury from mer-cury(II) in the presence of trimethyllead, an abioticmethyl donor, has been observed. These processesbecame less significant when humic substanceswere added. Under these conditions methylmer-cury was preserved at a level of 1 ng L−1 at 4 ◦Cin the dark for 33 days [65]. Blood samples fortotal mercury determination can be stored for afew weeks in the refrigerator. Longer periodsrequire storage in the deepfreezer at −20 ◦C orbelow. However, stability of the individual mer-cury species should be tested for long storage peri-ods. The stability of methylmercury chloride andmercury chloride in aqueous solutions was stud-ied by the BCR at 0 ◦C and at room temperatureto validate analytical methods [25]. No detectableeffects of temperature were seen after storage for3 months in the dark. Significant losses of mercurywere observed after 100-fold dilution of the initialsolution. Methylmercury and inorganic mercury infish extract solutions were stable for 5 months at4 ◦C when stored in the dark [66]. Most data seemto indicate that darkness is necessary for preserva-tion of mercury species during storage in biologicalmatrices and that there is much less influence fromsurrounding temperature. For total mercury deter-mination in urine, acidification of the samples isrecommended (with nitric or acetic acid) to avoidmercury absorption by the container wall. Hence,in speciation analysis any addition of acid shouldbe avoided unless it is proven that is has no influ-ence on the metal species present in the matrix. Inthe case of mercury it is also important to avoidbacterial growth in the sample as this may reducesome mercury to volatile elemental mercury. Forall these reasons it is important to analyze the sam-ples as soon as possible after collection is done. Byextension, this also applies to all other elements.

The mercury concentration in serum of healthypersons is about 0.5 µg L−1. In urine of non-exposed individuals the mercury concentrationranges from 1 to 10 µg L−1. Speciation in urineindicated that inorganic mercury was the major


form of mercury excretion. Other results indicatedas well the presence of methylmercury andethylmercury [26, 67].

7.12 Nickel

Contamination is a major problem in studyingnickel in body fluids. Similar to chromium andmanganese, the concentration of nickel in sweatis several times higher than that in serum. Skinshould be washed carefully before collecting thesample. In the case of a smoking person, risk ofcontamination is even higher. Apparently, the useof a stainless steel needle does not compromise theresults when the first 3 ml of blood are discarded.However, all materials that come into contactwith the sample must be washed by an acidwashing procedure. Sample manipulations shouldbe carried out in a clean room (class 100). Theconcentration of nickel stored in Teflon containersremained constant over a period of 2 h (duration ofexperiment) [27].

Expected concentrations for nickel in serum andurine are respectively <0.3 µg L−1 and <3 µg L−1.A recent report indicated that workers in a nickelrefinery are heavily exposed because of the highnickel concentrations that were found in urine [68].Further speciation research is needed.

7.13 Selenium

For selenium the sampling procedures are essen-tially free of contamination problems. Standardequipment for sampling of body fluids can beused. In aqueous solution selenium may be presentas selenite(IV), selenate(VI), methylated seleniumand other forms of organic selenium such as sele-nium–cysteine and selenium–methionine, where itoccurs bound to proteins [25]. Variations in theconcentration of selenium species during storagehave been reviewed recently [69, 70]. Adsorptionand desorption phenomena were important at lowselenium concentrations found in environmentalsamples. Loss of selenium depends on pH, ionicstrength, container material and ratio of containersurface area per unit of volume. There was no

influence of light on inorganic selenium species.Selenium was released from Teflon, polyethyleneand polycarbonate containers by 50 % HNO3. Lessleaching was observed with HCl. In a projectof BCR solutions of 10 µg L−1 selenium(IV) and50 µg L−1 selenium(VI) were kept at pH 2 and 6both in the dark and exposed to light, at three dif-ferent temperatures (−20, 20 and 40 ◦C), in twotypes of containers (PTFE and polyethylene) [25].The effect of the chloride anion was also tested forits role in oxidizing selenium(IV) to selenium(VI)in basic solutions. Selenium(IV) and selenium(VI)remained stable for 2 months but longer storageperiods (6 months) resulted in a decrease in sele-nium(IV) concentration. Complete loss of sele-nium(IV) occurred after 12 months when sampleswere stored in polyethylene containers at pH 2in the absence and presence of chloride. Sele-nium(VI) was stable under those conditions forthe 12 months tested. The stability of selenium(IV)increased at pH 6, with 2 months being the max-imum storage time without risk of selenium(IV)loss. The presence of the chloride ion decreasedthe risk of losses in some cases. PTFE containersincreased selenium(IV) losses at pH 6, especially at10 µg L−1. Both species were stable at −20 ◦C forthe 12 months tested and losses of selenium(IV)and selenium(VI) were lower at 40 ◦C than atroom temperature [69]. Most of these findings areconfirmed in a recent report [70]. However, it isstated there that stability increases with decreas-ing temperature. The stability order for storagecontainers was given as Teflon > polyethylene >

polypropylene and for pH values was pH 2 >

pH 4 > pH 8. The stability of four volatile organicselenium species was also tested under differentstorage conditions. In a short-term stability test ofstudied Se species in urine it was observed thatafter 5 h of storage about 30 % selenite(IV) and60 % of SeCys were lost [71]. In a subsequentreport, the maximum allowed storage time was1 week [72].

Selenium concentrations in serum vary between0.04 and 0.16 mg L−1. In urine, the selenium con-centration is about 100 µg L−1. Both concentra-tions depend heavily on the selenium intake.


In serum, selenium was found to be distributedamong three different fractions with approximatemolecular weights of >600, 200 and 90 kDa [73].Up till now, three selenium-containing proteinshave been identified in the literature: selenoproteinP, glutathione peroxidase and albumin [74]. Forurine, several methods exist to determine theconcentration of trimethylselenonium ion (TMSe),selenite(IV) and total selenium [26, 75].

7.14 Tin

Apart from inorganic tin, several organotin com-pounds are found in nature: tributyltin (TBT),triphenyltin (TPT), dibutyltin (DBT), monobutyltin(MBT), diphenyltin (DPT) and monophenyltin(MPT) [19]. TPT and butyltin species were foundto be stable for 3 months in HCl-acidified waterstored in polyethylene bottles at 4 ◦C in the darkand for at least 20 days in brown glass bottlesstored at 25 ◦C [76]. In filtered seawater samples,acidified to pH < 2, TBT was found to be stablewhen the water was stored in the dark at 4 ◦C inPyrex bottles. The addition of an acid to preservethe butyltin concentrations was unnecessary whenpolycarbonate bottles were used for storage at 4 ◦Cin the dark. Under these conditions, butyltins werestable for 7 months, but TPT showed a reduc-tion from the first month of storage. Storage inPyrex bottles and acidification showed no improve-ment. A better preservation of TPT was obtainedin C-18 cartridges stored at room temperature andno changes in concentration were observed dur-ing the first 60 days. Under the same conditions,butyltins were stable for 7 months [19]. The sta-bility of organotin compounds was found to bedependent on the pretreatment applied for preser-vation. A significant reduction of the concentrationof butyltin was observed when the sample wasair-dried under the action of infrared radiation oroven-dried at 110 ◦C. Lyophilization or desiccationprocedures did not affect the stability of butyltinand phenyltin species. Four different storage pro-cedures were evaluated: (1) at 4 ◦C, (2) freezing,(3) lyophilization and storage in a refrigerator and(4) drying in a desiccator and storage at 4 ◦C. Thebutyltin compounds were stable for at least 1 year

for the four storage conditions, but a reduction inthe phenyltin content was observed after 3 months,freezing and lyophilization being the most reliableprocedures for conservation [77]. Also, to avoidlosses or changes of butyltin species in oysters andcockles, lyophilization has been shown to be a reli-able procedure [19]. Butyltin species were stableover 150 days. A higher stability was found in wetsamples stored at −20 ◦C in the dark. A reduc-tion of 14 % was observed in the TBT contentafter 270 days, followed by an increase in DBTconcentration. Afterwards, a general decrease inTBT and DBT levels and an increase in MBTlevels was observed. After 540 days, a generaldecrease of all the butyltin species was found. Thisbehavior confirms that TBT degrades by stepwisedebutylation to DBT, MBT and inorganic tin [19].Recently, a higher stability (44 months) of organ-otin compounds in mussels has been reported forlyophilized samples stored at −20 ◦C in the dark.Temperature, and especially light, affect the stabil-ity of butyltin species in this sample. Significantvariations were found in the butyltin content after3 and 6 months of storage at room temperature indaylight and in the dark, respectively. Phenyltincompounds showed a lower stability: a reductionof 30 % was reported for TPT after 12 months forsamples stored at −20 ◦C [78].

For tin in normal human serum, the followingvalues have been reported: 0.502 ± 0.096 ng mL−1

(mean) and 0.400–0.636 ng mL−1 (range) [79].

7.15 Zinc

In contrast to selenium, the hazard of contamina-tion by zinc during sample collection is very real.Main sources of contamination are collection vials,including the stopper. Careful acid washing of allglassware and plastic ware, followed by rinsing inpure water is recommended. For short-term stor-age of rainwater in HDPE and PTFE containersthere is no loss of zinc [27]. This is rather surpris-ing because HDPE containers have been reportedto contain 200 µg mL−1 of zinc as metallic impu-rities (catalyst). It is suggested that the observedstability of dissolved zinc in a snow sample isthe result of fortuitous balancing of the loss of

THE VANADATE(V)–TRANSFERRIN COMPLEX 41

zinc from the snow sample with the gain of zincresulting from it leaching out of the walls of thecontainer. It may also be due to the absence of anybiological transformation or lack of formation ofany colloidal or ion-exchangeable species likely tobe adsorbed onto the container surface [27]. Sim-ilar to lead and manganese, the zinc concentrationin packed cells is much higher than that in serum.This means that hemolysis should be assessed bymeasuring the hemoglobin concentration in serumafter separation. Moreover, zinc is an element thatshows pronounced diurnal changes, so it is impor-tant to pay attention at the time of sampling [80].In case of zinc contamination, there will be anincrease in the albumin bound fraction [52]. It hasalso been shown that the concentration of protein-bound serum zinc in human blood plasma variesdepending on whether the sample was taken froma patient who was standing or lying in a recumbentposition [81].

Zinc concentration in serum amounts to1 mg L−1. In urine, the zinc concentration variesbetween 50 and 1000 µg day−1.

In serum, zinc elutes in four different fractions.Some zinc elutes in the dead volume of ananion-exchange Mono Q column, probably asfree zinc ions. A part of zinc co-elutes withtransferrin and also in two other serum fractions,one between immunoglobulin G and transferrinand the other between transferrin and albumin. Itis suggested that the last fraction correspond to theα2-macroglobulin protein (720 kDa) [26, 51]. Thedistribution of zinc in the plasma of normal anduremic patients differs from each other [51].

8 INFLUENCE OF pH, SALTMOLARITY AND ACETONITRILECONCENTRATION ON A SELECTEDMETAL–PROTEIN COMPLEX, i.e. THEVANADATE(V)–TRANSFERRINCOMPLEX

To demonstrate the instability of some metal–pro-tein complexes, we use an example that we havebeen studying extensively in our laboratory, i.e. thevanadate(V)–transferrin complex [82]. Vanadium

is an element that has been far less studied inspeciation analysis in comparison to other elementssuch as arsenic and selenium. However, in recentyears the interest in vanadium has grown becauseof its potential medicinal use as a drug in diabetesand cancer treatment. At the start of our vanadiumspeciation project, we examined the stability ofthe vanadate–transferrin complex. This complexis the principal vanadium species in serum and isone of the strongest known vanadium complexes(log K = 6.5 m−1). However, in comparison withother metal–transferrin complexes, its stabilityconstant is rather low [log K(Fe3+) = 22.7 m−1

and log K(Al3+) = 12.9 m−1]. These constantsalready give an indication for the use of mildseparation techniques and conditions. The goal ofthese experiments was to outline the limits of thepH and salt molarity and acetonitrile concentra-tions that could be used during separation tech-niques without altering the vanadate–transferrincomplex. All these experiments were carried outby ultrafiltration. First we examined the influenceof pH. The result is shown in Figure 2.2.3. As canbe seen there is a strong pH dependency of thevanadate–transferrin complex. Around the phys-iological pH the percentage of vanadate that isbound to transferrin is high while on going toextreme acid media or extreme basic media thebinding is rapidly ruptured. This behavior can beexplained by the protonation and deprotonationof some amino acids at the protein binding site.Out of this picture it is clear that addition of anacid for preservation purposes, e.g. to urine, wouldlead to misleading results. This kind of study wasalso carried out for the vanadate–albumin com-plex (figure not shown). Here, addition of an acidleads to a higher binding capacity of albumin forvanadate at lower pH values, opposite to the vana-date–transferrin equilibrium.

Also addition of high amounts of salt (Fig-ure 2.2.4) to the sample induces rupture ofthe vanadate–transferrin binding. Salts are addedto the buffer during chromatographic techniquessuch as anion-exchange or hydrophobic inter-action chromatography. The salts are added togenerate a gradient, which governs the elutionbehavior of the analytes from the column. From


120

100

80

60

40

20

00 1 2 3 4 5 6 7 8 9 10 11 12 13 14

tran

sfer

rin b

ound

van

adiu

m-4

8 (%

)

pH

Figure 2.2.3. Stability of vanadium–transferrin complex as a function of pH (mean ± SD). (Reprinted from Fresenius’ Journal ofAnalytical Chemistry, Stability of vanadium(V) protein complexes during chromatography, Vol. 363, pp. 519–522, Figures 1–6,1999, by permission of Springer-Verlag GmbH & Co. KG.)

100

90

80

70

60

50

40

30

20

0

10

0 0.01 0.05 0.1 0.15 0.2 0.5 1 1.5 2

tran

sfer

rin b

ound

van

adiu

m-4

8 (%

)

NaCl (mol/L)

Figure 2.2.4. Stability of vanadium–transferrin complex as a function of NaCl concentration (mean ± SD). (Reprinted fromFresenius’ Journal of Analytical Chemistry, Stability of vanadium(V) protein complexes during chromatography, Vol. 363,pp. 519–522, Figures 1–6, 1999, by permission of Springer-Verlag GmbH & Co. KG.)

CONCLUDING REMARKS 43

100

0.1 0.2 0.3 0.4 0.5

tran

sfer

rin b

ound

van

adiu

m-4

8 (%

)

Cl-salt (mol/L)

LiCl NaCl Calcium chloride Ammonium chloride

80

60

40

20

0

Figure 2.2.5. Stability of vanadium–transferrin complex as a function of different chloride salts (mean ± SD). (Reprinted fromFresenius’ Journal of Analytical Chemistry, Stability of vanadium(V) protein complexes during chromatography, Vol. 363,pp. 519–522, Figures 1–6, 1999, by permission of Springer-Verlag GmbH & Co. KG.)

Figure 2.2.4 we can conclude that the use ofhydrophobic interaction chromatography (high saltconcentration at starting point) is not recom-mended for our kind of study. In anion-exchangechromatography, high salt concentrations are alsoused, but at the end of the chromatographic run.In this way, limited use of the technique canbe considered. We also noticed significant differ-ences between some kinds of salt (Figure 2.2.5).For sodium salts (sodium acetate, sodium bromide,sodium iodide and sodium chloride) no signifi-cant differences were found. However, chloridesalts (lithium chloride, sodium chloride, calciumchloride and ammonium chloride) exerted differentinfluences on the vanadate–transferrin binding. Incase of calcium chloride the reason probably willbe the double amount of chloride anions that arepresent in solution in comparison with the othermonochloride salts. For ammonium chloride, thiseffect originates from a shift in the pH after addi-tion of this salt.

Addition of high amounts of acetonitrile inthe buffer also negatively affects the vana-date–transferrin binding (Figure 2.2.6). This figure

shows that the use of high acetonitrile concentra-tions (as in reversed phase chromatography) forseparation of vanadate–protein complexes shouldbe avoided.

All these figures show that it is important toinvestigate the stability of the trace element com-plexes before embarking on separation procedures.If not, the chromatograms will not reflect the origi-nal distribution of the metal species present in yoursample, but only yield useless artifacts.

9 CONCLUDING REMARKS

The previous paragraphs indicate that a well-established sampling and storage protocol shouldbe established before starting speciation research.First of all, it is necessary to exclude all knownsources of contamination by evaluating and clean-ing all the needed instruments, laboratory ware andreagents. Therefore, some evaluation and clean-ing methods are suggested in the text. For certainexperiments one should also consider the possibleinfluence of the biological time on test subjects.For several trace elements specific precautions


100

0 1 5 10 15 20 25

acetonitrile in buffer (%)

tran

sfer

rin b

ound

van

adiu

m-4

8 (%

)

90

80

70

60

50

40

30

20

10

0

Figure 2.2.6. Stability of vanadium-transferrin complex as a function of acetonitrile concentration (mean ± SD). (Reprintedfrom Fresenius’ Journal of Analytical Chemistry, Stability of vanadium(V) protein complexes during chromatography, Vol. 363,pp. 519–522, Figures 1–6, 1999, by permission of Springer-Verlag GmbH & Co. KG.)

are listed. As for the sampling as such, there isa choice between different methods, each withits particular advantages and disadvantages. Apartfrom the classical sampling methods (evacuatedtubes or syringes for fluids, dissection for tissues),a more recent technique is mentioned, i.e. micro-dialysis. The advantages of this technique are itsdynamic and mild characteristics, which are ofutmost importance in trace element research. Thedisadvantages are the difficulties concerning detec-tion and quantification of the analytes.

In the last section of this chapter, it is empha-sized that storage and separation conditions formetal species cannot be varied infinitely. Theborders of their stability interval under separa-tion conditions should be outlined. This can bedone for large molecules by, e.g., ultrafiltrationexperiments. From the example for the vana-date–transferrin complex it is clear that use ofadditives (acids, preservatives, anticlotting agents,. . .) should be avoided as much as possible. There-fore serum is better suited for speciation purposesthan is plasma.

The main goal of this chapter was to givethe reader an overview of the most common

procedures and pitfalls encountered in speciationresearch. However, it is left to the reader’sjudgment to decide to what extent these rec-ommendations should be followed in his/herown research.

10 ACKNOWLEDGEMENTS

KDC is supported by a grant of the Fund forScientific Research-Flanders (Belgium) (FWO)

11 REFERENCES

1. Cornelis, R., Heinzow, B., Herber, R. F. M., Christensen,J. M., Paulsen, O. M., Sabbioni, E., Templeton, D. M.,Thomassen, Y., Vahter, M. and Vesterberg, O., PureAppl. Chem., 67, 1575 (1995).

2. Burns, E. R., Anat. Rec. (New Anat.), 261, 141 (2000).3. Pineau, A., Guillard, O., Chappuis, P., Arnaud, J. and

Zawislak, R., Crit. Rev. Clin. Lab. Sci., 30, 203 (1993).4. Kosta, L., Talanta , 29, 985 (1982).5. Versieck, J. and Cornelis, R., Trace Elements in Human

Plasma or Serum , CRC Press, Boca Raton, FL, 1989.6. Alcock, N. W., At. Spectrom., 5, 78 (1984).7. Schmitt, Y., J. Trace Elem. Electrolytes Health Dis., 1,

107 (1987).

REFERENCES 45

8. Cornelis, R., Analyst , 117, 583 (1992).9. Cornelis, R., J. Radioanal. Nucl. Chem., 244, 255 (2000).

10. Beto, J. A., Bansal, V. K. and Kahn, S., Adv. RenalReplace. Ther., 6, 85 (1999).

11. Procedures for the Collection of Diagnostic Blood Spec-imens by Venipuncture; Approved Standards (ed. 4),National Committee on Clinical Laboratory Standards,Wayne, PA, Publication H3-A4, 1998.

12. Versieck, J., Crit. Rev. Clin. Lab. Sci., 22, 97 (1985).13. de Haan, K. E. C., de Groot, C. J. and Boxma, H., Clin.

Chim. Acta , 170, 111 (1987).14. Nygren, O. and Nilsson, C., J. Anal. At. Spectrom., 2, 805

(1987).15. Neve, J. and Petetz, A., Expression of urinary selenium

levels in humans, in Selenium in Medicine and Biology ,Neve, J. and Favier, A. (Eds), Walter de Gruyter, Berlin,1988, pp. 189–192.

16. Boiteau, H. L. and Pineau, A., Mercury, in QuantitativeTrace Analysis of Biological Materials , McKenzie, H. A.and Smythe, L. E. (Eds), Elsevier, Amsterdam, 1988pp. 553–560.

17. Maletskos, C. J., Albertson, M. D., Fitzsimmons, J. C.,Masurekar, M. R. and Tang, C. W., Sampling and samplehandling of human tissue for activation analysis, in TraceSubstances in Environmental Health IV , Hemphill, D. D.(Ed.), University of Missouri, Columbia, MO, 1970,p. 367.

18. Uthe, J. F. and Chou, C. L., Sci. Total Environ., 71, 67(1988).

19. Gomez-Ariza, J. L., Giraldez, I., Morales, E., Ariese, F.,Cofino, W. and Quevauviller, P., J. Environ. Monit., 1,197 (1999).

20. Johansen, M. J., Newman, R. A. and Madden, T., Phar-macotherapy , 17, 464 (1997).

21. Sarre, S., Deleu, D., Van Belle, K., Ebinger, G. andMichotte, Y., Trends Anal. Chem., 12, 67 (1993).

22. Lunte, C. E., Scott, D. O. and Kissinger, P. T., Anal.Chem., 63, 773A (1991).

23. BioAnalyticalSystems, http://www.bioanalytical.com.24. Gomez-Ariza, J. L., Morales, E., Sanchez-Rodas, D. and

Giraldez, I., Trends Anal. Chem., 19, 200 (2000).25. Quevauviller, P., de la Calle-Guntinas, M. B., Maier,

E. A. and Camara, C., Mikrochim. Acta, 118, 131 (1995).26. Das, A. K., Chakraborty, R., Crevera, M. L. and de la

Guardia, M., Mikrochim. Acta, 122, 209 (1996).27. Sekaly, A. L. R., Chackrabarti, C. L., Back, M. H., Gre-

goire, D. C., Lu, J. Y. and Schroeder, W. H., Anal. Chim.Acta , 402, 223 (1999).

28. Harris, W. R., Berthon, G., Day, J. P., Exley, C., Flaten,T. P., Forbes, W. F., Kiss, T., Orvig, C. and Zatta, P. F.,J. Toxicol. Environ. Health , 48, 543 (1996).

29. Ohman, L. O. and Martin, R. B., Clin. Chem., 40, 598(1994).

30. Kiss, T., Jakusch, T., Kilyen, M., Kiss, E. and Lakatos, A.,Polyhedron , 19, 2389 (2000).

31. de la Calle-Guntinas, M. B., Madrid, Y. and Camara, C.,Fresenius’ J. Anal. Chem., 344, 27 (1992).

32. Smith, A. E., Analyst , 98, 65 (1973).33. Smichowski, P., Madrid, Y. and Camara, C., Fresenius’

J. Anal. Chem., 360, 623 (1998).34. Delves, C. E., Sieniawska, C. E., Fell, G. S., Lyon,

T. D. B., Dezateux, C., Cullen, A., Variend, S., Bon-ham, J. R. and Chantler, S. M., Analyst , 122, 1323(1997).

35. Feldmann, J., Lai, V. W. M., Cullen, W. R., Ma, M. S.,Lu, X. F. and Le, X. C., Clin. Chem., 45, 1988 (1999).

36. Al-Sibaai, A. A. and Fogg, A. G., Analyst , 98, 732(1973).

37. Jokai, Z., Hegoczki, J. and Fodor, P., Microchem. J., 59,117 (1998).

38. Lagarde, F., Asfari, Z., Leroy, M. J. F., Demesmay, C.,Olle, M., Lamotte, A., Leperchec, P. and Maier, E. A.,Fresenius’ J. Anal. Chem., 363, 12 (1999).

39. Demesmay, C., Olle, M. and Porthault, M., Fresenius’J. Anal. Chem., 348, 205 (1994).

40. Lagarde, F., Amran, M. B., Leroy, M. J. F., Demes-may, C., Olle, M., Lamotte, A., Munteau, H., Michel, P.,Thomas, P., Caroli, S., Larsen, E., Bonner, P., Rauret, G.,Foulkes, M., Howard, A., Griepink, B. and Maier, E. A.,Fresenius’ J. Anal. Chem., 363, 18 (1999).

41. Wang, J., Dreessen, D., Wiederin, D. R. and Houk, R. S.,Anal. Biochem., 288, 89 (2001).

42. Majidi, V. and Miller-Ihli, N. J., Analyst , 123, 809 (1998).43. Dyg, S., Cornelis, R., Quevauviller, P., Griepink, B. and

Christensen, J. M., Anal. Chim. Acta , 286, 297 (1994).44. Sacher, F., Raue, B., Klinger, J. and Brauch, H. J., Int. J.

Environm. Anal. Chem., 74, 191 (1999).45. Cornelis, R., Borguet, F., Dyg, S. and Griepink, B., Mikro-

chim. Acta , 109, 145 (1992).46. Feng, W. Y., Qian, Q. F., Ding, W. J. and Chai, Z. F.,

J. Radioanal. Nucl. Chem., 244, 321 (2000).47. Makarov, A. and Szpunar, J., J. Anal. At. Spectrom., 14,

1323 (1999).48. Gallorini, M., Edel, J., Pietra, R., Sabbioni, E. and

Mosconi, G., Sci. Total. Environm., 150, 153 (1994).49. Cornelis, R., Heinzow, B., Herber, R. F. M., Christensen,

J. M., Poulsen, O. M., Sabbioni, E., Templeton, D. M.,Thomassen, Y., Vahter, M. and Vesterberg, O., J. TraceElements Med. Biol., 10, 103 (1996).

50. Gardiner, P. E. and Ottaway, J. M., Anal. Chim. Acta.,124, 281 (1981).

51. Bayon, M. M., Cabezuelo, A. B. S., Gonzalez,E. B., Alonso, J. I. G. and Sanz-Medel, A., J. Anal. At.Spectrom., 14, 947 (1999).

52. Cornelis, R., Borguet, F. and De Kimpe, J., Anal. Chim.Acta., 283, 183 (1993).

53. Gercken, B. and Barnes, R. M., Anal. Chem., 63, 283(1991).

54. Owen, L. M. W., Crews, H. M., Hutton, R. C. andWalsh, A., J. Anal. At. Spectrom., 117, 649 (1992).

55. Nygren, O., Appl. Organomet. Chem., 8, 601 (1994).56. Van Cleuvenbergen, R., Dirkx, W., Quevauviller, P. and

Adams, F., Int. J. Environ. Anal. Chem., 47, 21 (1992).57. Sampson, B., J. Anal. At. Spectrom., 6, 115 (1991).


58. Keck, P. E., Strakowski, S. M., Hawkins, J. M., Dunaye-vich, E., Tugrul, K. C., Bennett, J. A. and McElroy, S. L.,Bipolar Disorders , 3, 68 (2001).

59. Neve, J. and Leclercq, N., Clin. Chem., 37, 723 (1991).60. Bratter, P., Ribas, E. and Schramel, P., Trace Elem. Anal.

Chem. Med. Biol., 6, 1 (1994).61. Matzapetakis, M., Karligiano, N., Bino, A., Dakanali, M.,

Raptopoulou, C. P., Tangoulis, V., Terzis, A., Giapintza-kis, J. and Salifoglou, A., Inorg. Chem., 39, 4044 (2000).

62. Copeland, D. D., Facer, M., Newton, R. and Walker,P. J., Analyst , 121, 173 (1996).

63. Bloom, N. S., Fresenius’ J. Anal. Chem., 366, 438 (2000).64. Perez, M. T., Madrid-Albarran, Y. and Camara, C., Fre-

senius’ J. Anal. Chem., 368, 471 (2000).65. Reinholdsson, F., Briche, C., Emteborg, H., Baxter, D. C.

and French, W., in CANAS 95 , Welz, R. (Ed.), Perkin-Elmer, Ueberlingen, 1995.

66. Quevauviller, P., Drabaek, I., Munteau, H. and Grie-pink, B., Appl. Organomet. Chem., 7, 413 (1993).

67. Shum, S. C. K., Pang, H. and Houk, R. S., Anal. Chem.,64, 2444 (1992).

68. Thomassen, Y., Nieboer, E., Ellingsen, D., Hetland, S.,Norseth, T., Odland, J. O., Romanova, N., Chernova, S.and Tchachtchine, V. P., J. Environ. Monit., 1, 15(1999).

69. Cobo, M. G., Palacios, M. A., Camara, C., Reis, F. andQuevauviller, P., Anal. Chim. Acta, 286, 371 (1994).

70. Gomez-Ariza, J. L., Pozas, J. A., Giraldez, I. andMoralez, E., Int. J. Environ. Anal. Chem., 74, 215 (1999).

71. Gomez, M. M., Gasparic, T., Palacios, M. A. andCamara, C., Anal. Chim. Acta, 374, 241 (1998).

72. Quijano, M. A., Gutierrez, A. M., Perez-Conde, M. C.and Camara, C., Talanta , 50, 165 (1999).

73. Bratter, P., Gercken, B., Tomiak, A. and Rosick, U.,in Proceedings of the Fifth International Workshop,Bratter, P. and Schramel, P. (Eds), de Gruyter, Berlin,1988.

74. Harrison, I., Littlejohn, D. and Fell, G. S., Analyst , 2,189 (1996).

75. Robberecht, H. J. and Deelstra, H. A., Talanta , 31, 497(1984).

76. Bergmann, K., Rohr, U. and Neidhart, B., Fresenius’J. Anal. Chem., 349, 815 (1994).

77. Gomez-Ariza, J. L., Morales, E., Beltran, R., Giraldez, I.and Ruiz-Benitez, M., Quim. Anal., 13, S76–S79 (1994).

78. Morabito, R., Munteau, H., Cofino, W. and Quevau-viller, P., J. Environ. Monit., 1, 75 (1999).

79. Versieck, J. and Vanballenberghe, L., Anal. Chem., 63,1143 (1991).

80. Dawson, J. B., Fresenius’ J. Anal. Chem., 324, 463(1986).

81. Behne, D., J. Clin. Chem. Clin. Biochem., 19, 115 (1981).82. De Cremer, K., De Kimpe, J. and Cornelis, R., Fresenius’

J. Anal. Chem., 363, 519 (1999).

2.3 Food: Sampling with Special Referenceto Legislation, Uncertainty and Fitness for Purpose

P. Brereton, Roy Macarthur and H. M. CrewsCentral Science Laboratory, Sand Hutton, York, UK

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 472 Sampling Targets and Methods of

Sampling . . . . . . . . . . . . . . . . . . . . . . . . . 483 Legislation and Standards . . . . . . . . . . . . . 49

3.1 Examples of relevant legislation . . . . 503.2 Codex and WTO . . . . . . . . . . . . . . . 50

4 Methods of Calculating the UncertaintyAssociated with Sampling . . . . . . . . . . . . . 504.1 Communication . . . . . . . . . . . . . . . . 514.2 Assessing sampling uncertainty . . . . . 514.3 Collaborative trial in sampling . . . . . 514.4 Combined sampling–analytical QA 52

5 Fitness for Purpose of Sampling andAnalysis . . . . . . . . . . . . . . . . . . . . . . . . . . 53

6 Example: The Measurement ofMolybdenum in Wheat . . . . . . . . . . . . . . . 546.1 Establishing QA parameters through

collaborative trial in sampling . . . . . . 546.2 Fitness for purpose of the

measurement of molybdenum inwheat . . . . . . . . . . . . . . . . . . . . . . . . 56

7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . 578 Acknowledgements . . . . . . . . . . . . . . . . . . 579 References . . . . . . . . . . . . . . . . . . . . . . . . 57

1 INTRODUCTION

An analytical sample typically consists of afew grams of homogenised food. The resultsof an analysis will tell us how much of aparticular species of element that few grams ofsample contained at the time it was analysed.However, it is hardly ever the case that it is theobjective of a measurement merely to find out thecomposition of the analytical sample. Typicallythe objective of a measurement is to determinethe level of those species within a food at thetime the sample was taken from the bulk. The‘bulk’ could be any body of food about whichinformation is required. For example, a batch offood product from a factory; a consignment of foodproduct for import; the agricultural product of aparticular region; or the diet of children within aparticular country.

If, as is nearly always the case, an estimate ofthe level of analyte in the bulk is the result requiredfrom the ‘analysis’ then it is important that theanalysis (determination of analyte in sample) andsampling (everything that happens to sample priorto analysis), are thought of as a whole ‘measure-ment process’ [1]. The quality of the measurementis affected by both analysis and sampling. In fact,it is typical for sampling to make a larger contri-bution to the uncertainty associated with the resultof a chemical measurement than analysis.

This chapter will examine, with special refer-ence to two trace elements (arsenic and molyb-denum), the effect of sampling in particular onresults produced by measurements. It will discussthe methods that are available (and those underdevelopment) for assessing the contribution madeby sampling to the uncertainty associated with a

48 FOOD: SAMPLING

measurement. It will show how to calculate theeffort that should be put into sampling in order toproduce samples that are fit for purpose, and formonitoring the quality of the sampling process.

2 SAMPLING TARGETSAND METHODS OF SAMPLING

The choice of plan for taking and preparingfood samples is dependent on the matrix, theanalyte(s), and the purpose to which results areto be put. The first two factors govern thequalitative features that should be selected forthe sampling plan (e.g. sampling tool, storageconditions, sample container). The third factorgoverns the quantitative performance requiredof the sampling plan (e.g. maximum acceptablebias, cost of sampling, sample size, numberof increments). For speciation measurements inparticular, the collection and storage of samplescan have a profound effect on the stability ofelement species.

For example, as part of an ongoing EU project,(G6RD-CT-2001-00 473/SEAS; personal commu-nication, Prof. Carmen Camara, University ofMadrid, 2002), which is investigating the sta-bility of a variety of elemental species in foodmatrices, Camara et al. have investigated arsenic

species in rice. They determined the homogene-ity for total arsenic and its species (As(III), As(V),DMA and MMA) in two sets (SEAS rice and NISTSRM 1568a) of rice samples. No problems werefound with the homogeneity results both withinand between the containers for both samples ofrice. They then tested the effect of storage for2 months of sets of 20 bottles of each rice at roomtemperature, at 4 ◦C and at −20 ◦C in the dark. Thestability at these temperatures was evaluated bycomparing the results at room temperature (about20 ◦C) and 4 ◦C with those obtained at −20 ◦C usedas reference at the measured time. The stability at−20 ◦C has been evaluated versus t = 0.

The uncertainty UT was calculated as:

UT = [(CV4 ◦C or 20 ◦C)2 + (CV−20 ◦C)2]1/2RT /100

(uncertainty with respect to reference at thesame time)

UT = [(CV−20 ◦C, t )2 + (CV−20 ◦C, t=0)

2]1/2RT /100

(uncertainty with respect to reference at t = 0)

RT = XT /Xreference

Table 2.3.1 shows the concentration in SEASrice of each arsenic species at the different times

Table 2.3.1. Stability study of SEAS rice sample along two months of storage at room temperature, 4 ◦C and −20 ◦C; whereRT = XT/X−20 ◦C and XT = mean of 20 replicates at room temperature, 4 ◦C and −20 ◦C.

As species T( ◦C) X ± SD (µg kg−1) RT −20 RT −20,t=0

t = 0 Month 1 Month 2 Month 1 Month 2 Month 1 Month 2

As(III) 20 88.3 ± 4.9 94.9 ± 2.2 1.08 1.02 1.88 2.024 81.9 ± 7.4 92.9 ± 2.3 1.00 1.00 1.75 1.98

−20 46.99 82.2 ± 7.4 93.2 ± 1.9 1.00 1.74 1.98±0.75

DMA 20 29.7 ± 3.6 27.5 ± 2.5 1.14 1.15 1.05 0.974 27.0 ± 1.6 24.9 ± 0.8 1.04 1.04 0.95 0.88

−20 28.33 26.0 ± 2.3 23.9 ± 0.6 1.00 0.92 0.85±1.11

MMA 20 n.d n.d 0.0 0.0 0.0 0.04 n.d n.d 0.0 0.0 0.0 0.0

−20 18.10 n.d n.d 0.0 0.0 0.0±1.7

As(V) 20 10.7 ± 3.9 n.d 0.97 0.0 0.44 0.04 10.5 ± 4.7 n.d 0.95 0.0 0.43 0.0

−20 24.45 11.0 ± 5.7 n.d 1.00 0.45 0.0±1.09

LEGISLATION AND STANDARDS 49

and temperatures. The ratios (RT ) for arsenicspecies, of the mean values of 20 measurementsmade at room temperature and 4 ◦C, versus themean value of 20 determinations made for samplesstored at −20 ◦C and versus time = 0 are given.For ideal stability, R is 1. Since the mean valuesof the different arsenic species at different timeswere not comparable, the values of UT are notincluded in Table 2.3.1.

The work indicated no instability in NIST SRM1568a but found that, for the SEAS rice, MMA andAs(V) were not stable under different conditions.No significant differences in the stability at thedifferent temperatures tested were detected and inall cases MMA and As(V) were completely lostby transformation into other As organic species.The sum of the content of the analysed As speciesin the stability study was constant during the first3 months, which meant that the total As contentremained constant. Also the results obtained afterthe first month were very similar and no furtherspecies interconversion was detected.

Camara et al. have postulated that the mostlikely reason for the instability of the arsenicspecies in the SEAS sample was the high humiditycontent (about 18 %) possibly leading to anaerobicactivity. In addition the NIST SRM 1568a rice hadbeen irradiated whilst the SEAS had not.

The reader is further directed to Chapter 2.1where extensive guidance about the collection andstorage of biological samples (which includes foodmatrices) is described. The author gives detailedapproaches to sample collection, processing andstorage that can be applied to the food sector.

Effective sampling will produce analytical sam-ples that are representative of the bulk. The meanlevel of the analyte in a large set of representa-tive samples will be equal to the mean level ofthe analyte in the bulk. Representative samples canalso be used to provide information on the spatial(or temporal) variation in the level of the analytethroughout the bulk.

Perfect sampling is not possible. Samples arenever wholly representative. The sampling processmay be biased in some way towards samplescontaining particular levels of analyte. Or it mayjust be the case that the wide variation in analyte

concentration throughout a bulk means that a widerange of results is generated by the analysis of thesamples and hence it is difficult to gain a preciseestimate of the mean concentration.

The sampling process may lead to the con-tamination of the sample with the species underinvestigation: for example if stainless steel sam-pling tools are used. On the other hand, speciesmay be adsorbed onto the surface of an inap-propriate container. Otherwise exemplary samplesmay become unrepresentative of the bulk if theyare stored at the ‘wrong’ temperature; or at the‘wrong’ humidity; or if they are exposed to light.Each matrix–analyte combination presents its ownchallenges and these are intensified when it iscrucial that species integrity is maintained. It isnot the object of this chapter to provide adviceon eliminating the effects of sampling on thespecies within a sample (good advice is givenin Chapter 2.1 that may be applied to food sam-ples), but rather to now provide guidance on howto assess the size of the effect that the samplingprocess has.

A well-designed sampling protocol will provideinformation about the extent to which results ofanalyses are representative, by giving an estimateof the size of the uncertainty associated with thesampling process and will allow an assessmentto be made of whether the sampling process is‘in control’. Even if there are good reasons forbelieving that a particular sampling process haslittle effect on the composition of a sample, ifthe size of the effect of the sampling processis not estimated then this belief remains aneducated guess.

3 LEGISLATION AND STANDARDS

Recent development of international standardsand legislation for sampling food products hasbeen driven by the need for the results ofmeasurements to be internationally comparable.Several international organisations such as theCodex Alimentarius Commission, the EuropeanUnion and the World Trade Organisation have beenresponsible for producing legislation and standardsrelevant to the sampling of foodstuffs.

50 FOOD: SAMPLING

The creation of ‘the single market’ within theEuropean Union gave increased the momentum ofthe harmonisation of national food laws withinthe Union and there is now a large body oflegislation referring to the sampling and analysisof foodstuffs.

3.1 Examples of relevant legislation

‘Official Control of Foodstuffs Directive’

Council Directive 85/591/EEC of 20 Decem-ber 1985 concerning the introduction of Com-munity methods of sampling and analysis forthe monitoring of foodstuffs intended for humanconsumption [2].

Sampling of product types, e.g.First Commission Directive 87/524/EEC of 6October 1987 laying down Community methods ofsampling for chemical analysis for the monitoringof preserved milk products [3]

Sampling for particular analytes, e.g.Commission Directive 2001/22/EC of 8 March2001 laying down the sampling methods and themethods of analysis for the official control of thelevels of lead, cadmium, mercury and 3-MCPD(3-monochloropropane-1,2-diol) in foodstuffs [4].

3.2 Codex and WTO

The Codex Alimentarius Commission is the maininternational body concerned with the setting offood standards. The body is jointly funded by theFood and Agriculture Organisation and the WorldHealth Organisation.

The Codex Committee on Methods of analysisand sampling has produced draft general guide-lines on sampling [5]. The guidelines have beenproduced to ensure that ‘fair and valid proceduresare used when food is tested for compliance with aparticular Codex commodity standard’. The guide-lines are addressed to member states in order toenable them to resolve trade disputes.

The World Trade Organisation (WTO) wascreated in 1995 to provide an organisation capableof updating the implementation of agreementsassociated with the General Agreement on Tariffs

and Trade. One of the Agreements that define theWTO rules has particular relevance to area of foodspecification and analysis and hence sampling:the Agreement of Sanitary and PhytosanitaryMeasures (SPS).

The SPS Agreement affirms that members ofthe WTO should be able to enforce measures toprotect human and animal health, and instructs thatSPS measures should be harmonised through theapplication of international standards guidelinesand recommendations. The Agreement appliesto all SPS measures ‘which may, directly orindirectly, affect trade’ and defines a sanitary orphytosanitary measure as (amongst other things)‘any measure applied to protect human or animallife or health . . . from risks arising from additives,contaminants, toxins . . . in foods, beverages orfeedstuffs’.

The Agreement defines harmonisation as ‘Theestablishment, recognition and application of com-mon sanitary and phytosanitary measures by dif-ferent members’. As part of the definition it defines‘international standards, guidelines and recommen-dations’ as ‘for food safety the standards, guide-lines and recommendations established by theCodex Alimentarius Commission relating to foodadditives, veterinary drug and pesticide residues,contaminants, methods of analysis and sampling,and codes and guidelines of hygienic practice’.

The wide scope of the SPS Agreement (director indirect effect on trade) and the specificationof the Codex Alimentarius Commission within theAgreement have (for the Members of the WTO)changed the nature of the Codex. Pre-WTO itwas a body that made recommendations that couldsafely be ignored by members who chose to doso; now members of the WTO are effectivelyobliged to make Codex standards the basis of theirnational controls.

4 METHODS OF CALCULATINGTHE UNCERTAINTY ASSOCIATEDWITH SAMPLING

The uncertainty associated with sampling has beenreferred to as ‘the uncertainty that dares notspeak its name’ [6] because its contribution to

UNCERTAINTY ASSOCIATED WITH SAMPLING 51

measurement uncertainty is often neglected. Theneglect arises from the lack of standard proce-dures for the quantitative assessment of samplingquality in the food sector; a perception that gain-ing information about sampling quality is diffi-cult and expensive; and a lack of communicationbetween samplers and analysts. Nevertheless in thefood sector, sampling uncertainty has been shownoften to make a larger contribution to measurementuncertainty than the uncertainty associated withanalysis (with the possible exception of homoge-nous liquid samples). Hence no realistic assess-ment of measurement uncertainty can be madewithout taking into account the contribution madeby sampling.

4.1 Communication

If an analyst is presented with a sample with noinformation on how that sample was producedthen, even if the analytical uncertainty is well char-acterised, they will be unable to produce a realis-tic estimate of the measurement uncertainty thatapplies to the concentration of analyte in the sam-pling target. Similarly, if customers are presentedwith the result of an analysis without an assessmentof the analytical uncertainty they will be unable todraw appropriate conclusions about even a well-characterised sample taken from a bulk.

4.2 Assessing sampling uncertainty

An essential starting point is to produce a well-characterised sample by following a prescribed (oragreed) sampling protocol. The protocol shouldhave been studied and characterised prior to its use(just as analytical methods should be validated).Methods for characterising the performance ofsampling protocols are under development [1,7–12]. The sample should then by analysedby a well-characterised method, for example amethod validated by collaborative trial [13] or bysingle laboratory validation following recognisedprotocols [14, 15].

A common statistical approach considers thatthe target consists of a large set of ‘normally

distributed’ samples, the mean concentration ofwhich is the true concentration of analyte in thesample. If several samples are taken and analysedthen the mean result provides an estimate of themean analyte concentration in the sampling target.The uncertainty associated with this estimate maybe calculated by:

confidence interval = mean result ± ts√n

where:t = value for the required confidence level given

in t-tables;s is the standard deviation displayed by

the results;n is the number of samples analysed.

However this approach is only valid if the ana-lytical method has been shown to be unbiased,and the measurements are carried out over severalbatches of analyses. Analysing the samples in asmall number of batches would produce results thatdo not represent analytical uncertainty, but some-thing closer to analytical repeatability. Althoughan estimate of the overall measurement uncertaintycan be gained, the contributions made by samplinguncertainty and analytical uncertainty cannot beseparated. This may lead to problems in identifyingthe causes of apparently large measurement uncer-tainties (i.e. quality failures). Also, if a reductionin measurement uncertainty is required, estimatesof the contributions made by sampling and anal-ysis are needed to identify the most economicalmethod of achieving the reduction (see Section 5).

4.3 Collaborative trial in sampling

Work in the geochemical and environmental sec-tors has been carried out to produce methods for theestimation of sampling uncertainty [7, 9, 10]. Meth-ods that rely upon a single sampler taking replicatesamples using a single protocol produce uncer-tainty estimates analogous to analytical repeata-bility. Methods that employ multiple samplers totake replicate samples using a single protocol pro-duce uncertainty estimates that are analogous tothe results of a collaborative trial of an analytical

52 FOOD: SAMPLING

method. They are referred to as collaborative trialsin sampling, and should be employed to provide avalid estimate of sampling uncertainty.

The collaborative trial in sampling is carriedout by a number of samplers (≥8) who, whilefollowing the sampling protocol under study, inde-pendently use different equipment to take dupli-cate samples from the sampling target. The dupli-cate samples must then be analysed in ran-dom order under repeatability conditions. NestedANOVA can then be employed to obtain estimatesof: the analytical repeatability standard deviation;the within-sampler repeatability standard deviation(s1); and the between-sampler standard deviation(s2) (Figure 2.3.1).

These standard deviations can be combinedin the usual way to produce an estimate of thesampling reproducibility standard deviation (saR).

saR =√

s21 + s2

2

For unbiased sampling methods the samplingreproducibility standard deviation represents thesampling uncertainty standard deviation, i.e. thecontribution made by sampling towards overall

measurement uncertainty. Once estimates for sam-pling repeatability and reproducibility have beenestablished they can be employed in a com-bined sampling–analytical QA system, and toproduce valid estimates for overall measurementuncertainty.

However, like collaborative trials for analyti-cal methods, collaborative trials in sampling areexpensive. A further problem is that they arevery specific (site, commodity, analyte) and aretherefore practicable only for use with bulkcommodities.

4.4 Combined sampling–analytical QA

A common element of analytical quality assuranceis the duplicate analysis of samples to check thatthe repeatability standard deviation associated withthe results of the analysis of a batch of samples isnot larger than the repeatability standard deviationassociated with the method of analysis. For amethod of analysis with a repeatability standarddeviation sr, the results of two analyses carried outunder repeatability conditions should rarely (95 %

Sampler 3

Sampler 8

Sampler 1

SampleA

Analysis

1A1

Sampler 2

SampleA

SampleB

Analysed under repeatability conditions in random order

SampleB

Analysis

1A2

Analysis

1B1

Analysis

1B2

Analysis

2A1

Analysis

2A2

Analysis

2B1

Analysis

2B2

Analysis

8A1

Analysis

8A2

Analysis

8B1

Analysis

2B2

Figure 2.3.1. Establishing sampling uncertainty.

FITNESS FOR PURPOSE OF SAMPLING AND ANALYSIS 53

confidence) differ by more than the repeatabilitylimit (r), where the repeatability limit is given by:

r = √8sr

This type of quality assurance can be extendedto sampling by the analysis of duplicate samples(taken under repeatability conditions). Several pro-tocols can be used to check the quality of sampling,but perhaps the simplest employs the duplicateanalysis of duplicate samples (Figure 2.3.2).

If duplicate samples are taken using a proto-col assessed as having a repeatability standarddeviation of s1, and each sample is analysed induplicate under repeatability conditions using amethod assessed as having a repeatability stan-dard deviation of sr, then the mean of the duplicateanalysis of each sample should rarely (95 % confi-

dence) be greater than√

4s2r + 8s2

1 . If the dupli-cate analyses fall within the repeatability limit(r) of each other (and any additional analyticalQA produces satisfactory results) then both sam-pling and analysis have been shown to be in con-trol. At this stage the sampling reproducibility and

standard uncertainty [16] associated with the anal-ysis (ua) can be combined to give an estimate ofthe uncertainty associated with the whole measure-ment (uc) by:

uc =√

sa2R + u2

a

5 FITNESS FOR PURPOSEOF SAMPLING AND ANALYSIS

‘Fitness for purpose’ of analytical measurementshas many definitions. One definition given byThompson and Fearn [17] is that a method that isfit for purpose results in lower total financial lossthan any other possible measurement method. Oneconsequence of this is that a method for making ameasurement that is sufficiently accurate is fit forpurpose only if there is no other method capableof achieving the same accuracy at a lower cost.As both sampling and analysis contribute to theuncertainty associated with a measurement and thecost of making a measurement, they both need to belooked at to assess a method’s fitness for purpose.

Sample A Sample B

Analysis A1 Analysis B2Analysis B1Analysis A2

Mean A Mean B

8 × s14 × srdifference <

8 × srdifference <

Sampling repeatability standard deviation = s1

Analytical repeatability standard deviation = sr

Measurement repeatability (singleanalysis of single sample) standard deviation = s1 + sr

+2 2

Figure 2.3.2. Combined sampling and analytical quality assurance.

54 FOOD: SAMPLING

If separate estimates of the uncertainty associ-ated with sampling and the uncertainty associatedwith analysis (and the cost of these procedures) areavailable, then it is possible to check whether theproportion of resources devoted to each of the pro-cedures is likely to lead to a measurement methodthat is fit for purpose.

For example, if a measurement is carried outby taking samples using a procedure with areproducibility (assessed for example during acollaborative trial in sampling) given by saR, andan analytical standard uncertainty given by ua

(assessed for example using methods describedin the Eurachem guide [16]); while the cost ofproducing one sample is £A, and the cost ofanalysing one sample is £B then the followingstatements can be made: If the cost of analysisand sampling is proportional to the ‘effort’ (twosamples cost twice as much to analyse as onesample etc.) then

AsaR2 = Ks Bua

2 = Ka

Ks and Ka are constants that reflect the cost atunit variance of sampling and analysis. The cost ofproducing samples with variance Vs and carryingout analyses with variance Va is given by

cos t = Ks

Vs+ Ka

Va

The variance associated with the measurementVm = Vs + Va, hence

cos t = Ks/Vs + Ka/(Vm − Vs).

At the minimum cost the gradient of this curve iszero. Hence at minimum cost:

Ka

(Vm − Vs)2− Ks

V 2s

= 0

Hence at minimum cost the optimum samplingvariance (V ′

s ) is given by:

V ′s = Vm(Ks − √

KsKa)

Ks − Ka

⇒ V ′s = Vm

√Ks(

√Ks − √

Ka)(√Ks − √

Ka) (√

Ks + √Ka

)

V ′s = Vm

√Ks√

Ks + √Ka

and the optimum analytical variance is given by

V ′a = Vm − V ′

s

The optimum expenditure on sampling is givenby L′

s = Ks/V ′s and the optimum expenditure on

analysis is given by L′a = Ka/V ′

aThe sum of the optimum expenditures (L′

a +L′

s) will be less than the sum of the original expen-ditures (La + Ls), while the overall measurementuncertainty will remain unchanged. If the valuesproduced for the optimum expenditures are sim-ilar to the current expenditure on sampling andanalysis then the measurement is likely to be ‘fitfor purpose’. If the sum of the optimum expen-ditures is much less then the current expenditureon sampling and analysis then the apportioning ofresources between sampling and analysis should bere-examined.

The data treatments used to provide estimatesof the uncertainty associated with sampling; forcombined analytical and sampling quality con-trol; and for the assessment of a measurementmethod fitness for purpose are all based on theassumption that ‘normal statistics’ describe uncer-tainties associated with sampling and analysis. Thisassumption is true (or cannot reasonably be tested)for many applications, although it is a good ideato employ the services of a professional statis-tician when designing experiments to assess theuncertainty associated with measurements so thatthis or any other assumptions used can be testedwhere possible.

6 EXAMPLE: THE MEASUREMENTOF MOLYBDENUM IN WHEAT

6.1 Establishing QA parameters throughcollaborative trial in sampling

The storage area was approximately 200 ft × 80 ftin total. The grain was stacked against the backwall of the barn and extended forwards towards theaccess area in the middle span of the barn. The pileof grain covered an area measuring approximately80 ft2 and was accessible only on its front slopingedge and top. The pile contained around 800 tonnes

MEASUREMENT OF MO IN WHEAT 55

and was in the region of 9 ft high. Each section ofthe stored grain contained a dryer on the top ofthe pile.

The sampling procedure adopted took generalaccount of a published standard method for thesampling of grain [18]. The standard emphasisesthe necessity of obtaining a properly representativesample and gives limited details described as ‘goodpractice’ which should be followed wheneverpracticable. It is noted that, as the composition ofthe lot is seldom uniform, a sufficient number ofincrements should be taken and carefully mixed togive a bulk (or composite) sample from which thelaboratory sample can be obtained. Information isonly given for a lot of up to 500 tonnes, in whichcase increments of 1 kg maximum to producea composite of 100 kg and in turn a laboratorysample of 5 kg is deemed to be appropriate. Largeror smaller laboratory samples are recognised to beappropriate in some cases.

An ‘in-house’ sampling scheme was normallyfollowed, in which a sampling spear (about 5 ftin length) containing approximately 100 g of grainwas used to take multiple increments to producea composite sample of around 600 g (or 5 kgif a very heterogeneously distributed analyte isbeing sampled).

For the purposes of the pilot sampling exer-cise, in the first instance, a professional samplerfollowed the normal sampling scheme and pro-duced a composite sample by taking eight samplespear increments at random within each sector.This corresponded to normal practice of taking aneight-increment composite sample per 50 tonnes ofgrain. Further sampling was then carried out byfour nonprofessional samplers. In each case thenonprofessional samplers took six sample incre-ments using the spear, to produce each compositesample. In order to take these six spear incre-ments randomly, use was made of the grid rep-resentation given in Figure 2.3.1. Working aroundthe grid a coin was tossed: heads represented across in the relevant area of the grid (take a sam-ple); if tails, no sample was taken. Some differ-ences in the ease (and therefore amount collected)and the manner of sampling were apparent amongthe nonprofessional samplers. In addition to the

samples taken by the four nonprofessional sam-plers one composite sample, again consisting ofsix incremental spear samples, was also taken torepresent the whole of the front sloping edge ofthe pile. Each of the composite grain sampleswas blended by hand on receipt and a labora-tory sample of approximately one-fifth the sizeof the composite sample was produced using asample splitter.

The laboratory samples were analysed byICPMS in random order under repeatabilityconditions to determine the concentration of traceelements. The concentration (dry matter basis) ofmolybdenum found in the samples is shown inTable 2.3.2.

A nested analysis of variance was used to deter-mine the contributions towards the overall variancefrom between-sampler variation in sampling (sam-pling reproducibility), within-sampler variation insampling (sampling repeatability) and analyticalrepeatability. The results of the analysis of vari-ance are shown in Table 2.3.3.

sr = √MSr = √

0.0002617 = 0.016 mg kg−1

s1 = √(MS1 − MSr)/2

Table 2.3.2. Results of measurements during collaborative trialin sampling.

Sampler Sample Analysis 1(mg Mo kg−1)

Analysis 2(mg Mo kg−1)

1 1 0.385 0.4032 0.424 0.435

2 3 0.469 0.4464 0.455 0.455

3 5 0.614 0.5766 0.494 0.511

4 7 0.531 0.4958 0.480 0.506

5 9 0.519 0.49610 0.487 0.492

Table 2.3.3. Sources of variation in sampling.

Source ofvariation

Degrees offreedom

Sum ofsquares

Mean squareestimate

Between-sampler (s2) 4 0.04301 0.01075Within-sampler (s1) 5 0.01055 0.002109Analytical (sr ) 10 0.002617 0.0002617Total 19 0.05618

56 FOOD: SAMPLING

= √(0.002109 − 0.0002617)/2

= 0.030 mg kg−1

s2 = √(MS2 − MS1)/4

= √(0.0107533 − 0.0021093)/4

= 0.046 mg kg−1

Standard deviation of variation due to sampling:

saR =√

s12 + s2

2 =√

0.0302 + 0.0462

= 0.055 mg kg−1

The uncertainty associated with the analyticalpart of the measurement process was studiedusing methods described in the Eurachem guideto measurement uncertainty [15]. The standarduncertainty associated with the measurement wasfound to be 0.058 mg kg−1.

Three conclusions can be drawn from this study:The first conclusion is that both sampling and

analysis make a significant contribution towardsthe measurement uncertainty. Although that state-ment sounds like it should always be true, if thestandard uncertainty (standard deviation) associ-ated with either sampling or (more commonly)analysis is less than 0.3 times the other component,then ignoring its contribution leads to a reduc-tion in the estimated measurement uncertainty ofless than 5 %. Therefore analyses carried out forthe purposes of quality assurance should produceresults that address both the analytical and sam-pling processes.

The second conclusion is that the standarduncertainty associated with results generated bythe measurement of molybdenum in wheat fromthis supplier is 0.080 mg kg−1

(√0.0552 + 0.0582

)

or 16 % for samples containing approximately0.5 mg kg−1 of the metal.

The third conclusion is that the results of theduplicate analysis of a single sample should liewithin 0.045 mg kg−1 or (

√8 × 0.016) or 9 % of

each other, and that the means of the results ofduplicate analysis of duplicate samples should liewithin 0.09 mg kg−1

(√4 × 0.0162 + 8 × 0.0302

)

or 18 % of each other. These results form the

basis of quality assurance tests for the precisionof sampling and analysis.

6.2 Fitness for purpose of themeasurement of molybdenum in wheat

The cost of taking a sample of wheat for theanalysis of molybdenum was £59.62. The cost ofanalysing a sample was £26.74. These costs werecombined with the analytical and sampling uncer-tainties to provide an estimate of the measure-ment’s fitness for purpose (Tables 2.3.4 and 2.3.5).

Vm = 0.0582 + 0.0552 = 0.00639

Ks = 59.62 × 0.0552 = 0.180

Ka = 26.74 × 0.0582 = 0.0900

V ′s = 0.00639 × √

0.180√0.180 + √

0.090= 0.00374

V ′a = 0.00639 − 0.00374 = 0.00265

L′s = 0.180/0.00374 = £48.16

L′a = 0.0900/0.00265 = £34.01

Figure 2.3.3 shows how the cost of the mea-surement varies with the proportion of the totalexpenditure devoted to sampling. It can be seenthat, in this case, the optimum expenditure onsampling and therefore on analysis is not very dif-ferent from the current expenditure. Therefore the

Table 2.3.4. Costs and uncertainties associated with the mea-surement of molybdenum in wheat.

Cost (£) Standard uncertainty(mg kg−1)

Analysis 26.74 0.058Sampling 59.62 0.055Measurement 86.36 0.080

Table 2.3.5. Optimum costs and uncertainties associated withthe measurement of molybdenum in wheat.

Optimum cost (£) Optimum standarduncertainty (mg kg−1)

Analysis 34.01 0.051Sampling 48.16 0.061Measurement 82.17 0.080

REFERENCES 57

200

0.3 0.4 0.5 0.6 0.7 0.8 0.9

Proportion of total cost spent on sampling

Cos

t of p

roce

ss

A

B

C

DE

180

160

140

120

100

80

60

40

20

0

Figure 2.3.3. Variation in cost of measurement with proportion of expenditure devoted to sampling: (A) cost of sampling, (B) costof analysis, (C) cost of measurement, (D) current division of expenditure between sampling and analysis, (E) optimum divisionof expenditure.

proportion of expenditure taken up by samplingand by analysis is about right and this aspect ofthe measurement is ‘fit for purpose’.

7 SUMMARY

The aim of a measurement carried out on somesample of food is to gain information about awider body of food rather than just informationabout the analytical sample. Thus, both analysisand sampling affect the quality of a measurement.

The need to produce comparable measurementscan be met in two ways: through the production ofinternational standards and international legislationon analysis and sampling by bodies such as thecodex Alimentarius Commission, the World TradeOrganisation and the European Union; and bythe use of statistical methods to validate theperformance of sampling methods.

Sampling methods can be validated by collab-orative trial. However, the process is expensiveand will not be suitable for all applications. Meth-ods for combined analytical and sampling qual-ity control are under development and should be

used (in combination with standard and/or vali-dated methods of sampling and analysis).

Information gained from the validation ofsampling and analytical methods can be usedto reduce the cost required to gain a givenmeasurement precision, as well as to provide theinformation required to validate the performanceof a measurement method.

8 ACKNOWLEDGEMENTS

The authors would like to acknowledge theUnited Kingdom Food Standards Agency whohave funded work in this area and been verysupportive of the studies.

9 REFERENCES

1. Ramsey, M. H., J. Anal. At. Spectrom., 13, 82 (1998).2. Council Directive 85/591/EEC of 20 December 1985

concerning the introduction of Community methods ofsampling and analysis for the monitoring of food-stuffs intended for human consumption, Off. J., L372, 0050–0052 (1985).

58 FOOD: SAMPLING

3. First Commission Directive 87/524/EEC of 6 October1987 laying down Community methods of sampling forchemical analysis for the monitoring of preserved milkproducts, Off. J., L 306, 0024–0031 (1987).

4. Commission Directive 2001/22/EC of 8 March 2001laying down the sampling methods and the methods ofanalysis for the official control of the levels of lead,cadmium, mercury and 3-MCPD in foodstuffs, Off. J.,L 077, 0014–0021 (2001).

5. Codex Committee on Methods of Analysis andSampling, Proposed draft general guidelines on sam-pling, ftp://ftp.fao.org:21/Codex/ccmas23/ma01 03e.pdf,2001.

6. Thompson, M., J. Environ. Monit., 1, 19 (1999).7. Ramsey, M. H., Argyraki, A. and Thompson, M., Ana-

lyst , 120, 1353 (1995).8. Thompson, M. and Ramsey, M. H., Analyst , 120, 261

(1995).

9. Ramsey, M. H., Argyraki, A. and Thompson, M., Ana-lyst , 120, 2309 (1995).

10. Argyraki, A., Ramsey, M. H. and Thompson, M., Ana-lyst , 120, 2799 (1995).

11. Ramsey, M. H., Analyst , 122, 1255 (1997).12. Thompson, M., Accreditation Qual. Assur., 3, 117 (1998).13. Thompson, M. and Wood, R., J. Assoc. Off. Anal. Chem.

Int., 76, 926 (1993).14. Fajgelj, A. and Ambrus, A. Principles and Practice of

Method Validation , Royal Society of Chemistry, 2000.15. Eurachem, The fitness for purpose of analytical methods:

a laboratory guide to method validation and related topics(1998), http://www.eurachem.ul.pt.

16. Eurachem, Quantifying Uncertainty in Analytical Mea-surement, 2nd edn (2000), http://www.eurachem.ul.pt.

17. Thompson, M. and Fearn, T., Analyst , 121, 275 (1996).18. British Standards Institution, Methods for sampling

cereals (as grain), BS4510, 1980.

2.4 Sampling: Collection,Storage–Occupational Health

Ewa Dabek-Zlotorzynska and Katherine Keppel-JonesEnvironment Canada, Ottawa, Canada

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 592 Collection of Airborne Particulate Matter 60

2.1 Filters . . . . . . . . . . . . . . . . . . . . . . . 602.2 Particulate sampling systems . . . . . . . 61

2.2.1 Samplers for ambient airparticulate matter . . . . . . . . . . 61

2.2.2 Samplers for particulates inworkplace atmospheres . . . . . 62

3 Collection of Volatile Compounds . . . . . . . 634 Sample Handling and Storage . . . . . . . . . . 64

5 Applications . . . . . . . . . . . . . . . . . . . . . . . 645.1 Lead . . . . . . . . . . . . . . . . . . . . . . . . 645.2 Chromium . . . . . . . . . . . . . . . . . . . . 655.3 Mercury . . . . . . . . . . . . . . . . . . . . . . 665.4 Platinum group metals . . . . . . . . . . . 675.5 Radionuclides . . . . . . . . . . . . . . . . . . 685.6 Metalloids . . . . . . . . . . . . . . . . . . . . 695.7 Miscellaneous applications . . . . . . . . 69

6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . 707 References . . . . . . . . . . . . . . . . . . . . . . . . 70

1 INTRODUCTION

The characterization of airborne pollution includ-ing metallics is of great importance to manyscientific fields, such as epidemiology and toxi-cology in health research [1–5]. In atmosphericsciences, it is important for modeling atmosphericprocesses and for environmental control purposes.Occupational health monitoring, on the other hand,relies on contaminant collection and subsequentcharacterization to evaluate health hazards forhumans exposed to pollutants in the workplace.There is wide international agreement that health-related sampling should be carried out based onhealth-related particle size-selective criteria. Threefractions have been identified: inhalable (parti-cles inhaled during breathing), thoracic (inhaledparticles that penetrate beyond the larynx), andrespirable (inhaled particles that penetrate to thealveolar region) [6, 7]. In addition, there is ample

evidence that the chemical speciation of trace met-als associated with particulate matter, and hencetheir bioavailability, depends on their oxidationstate and how the elements are partitioned amongthe various compounds.

The increasing awareness of problems associ-ated with airborne metallic pollutants has led to aneed for defining sampling procedures and adapt-ing existing analytical techniques to identify a widevariety of these compounds. Airborne contami-nants can be present in air as particulate matterin the form of liquids or solids; as gaseous mate-rial in the form of true gas or vapor; or in acombination of these forms. Most metals in theatmosphere are associated with airborne particu-late matter, though some trace atmospheric metalsexist also in gaseous form.

In the first part of this chapter, we will focusour attention on sampling techniques used forcollecting particulate and volatile pollutants. Total


and size-fractionating filter collectors such ascascade impactors and other collection devices areaddressed. However, the characteristics of thesesampling techniques and samplers are discussedonly briefly since more detailed descriptions can befound in several monographs [8–11]. An outlineof sample handling and storage completes thissection. The second part of the chapter willconsist of more detailed descriptions of samplingmethodologies for specific airborne metal species.

2 COLLECTION OF AIRBORNEPARTICULATE MATTER

Airborne particles are known as ‘particulate matter(PM)’ or simply ‘particles’. These particles arevery small solids and/or liquids that are producedby a variety of natural and man-made sources.Airborne particles vary widely in their chemicalcomposition and size. The size of particles mayrange from 0.005 µm to 100 µm in diameter. Thesuspended portion (total suspended particulates orTSP, i.e. found floating in air) is generally less than50 µm.

Suspended particulate matter is not a homo-geneous pollutant. In the atmosphere, it typicallyexhibits a bimodal size distribution with one peakin the range 0.1–2.5 µm, and a second peak in therange 2.5–50 µm. PM10 (particulates smaller than10 µm) include both a coarse fraction (PM10 – 2.5),which is generated mainly mechanically, and a finefraction (PM2.5), which corresponds to secondaryPM formed in the atmosphere by chemical reac-tions. PM2.5 are particles of 2.5 µm or less in diam-eter. The finer particles pose the greatest threatto human health because they can travel deepestinto the lungs. Depending on their size distribu-tion, these particulates may be transported up tothousands of kilometers from their source.

A particulate sampling train consists of the fol-lowing components: air inlet, particulate separa-tor or collecting device, air flow meter, flow ratecontrol valve, and air mover or pump. Of these,the most important component is the particulateseparator. The separator may consist of a singleelement (such as a filter or impinger), or there

may be two or more elements in a series (suchas a two-stage cyclone or multi-stage impactor)so as to characterize the particulates into differentsize ranges.

2.1 Filters

Sampling on filters is the most practical methodcurrently available to characterize the particle sizesand chemical composition of airborne particulates.In this case, the composition of bulk airborneparticulate samples is performed. Although theindividual character of the particulates is lostin this way, their size distribution can still bemeasured by bulk methods by using samplingdevices (e.g. cascade impactors and filters) thatcollect the particulates according to their size.Airborne particulates acquired by drawing ambientair through filter material remain deposited on thefilter, while all gases pass through the filter. Theaccurate measurement of flow rate and samplingtime or sample volume is as important as themeasurement of sample quantity.

Filter media of many different types and withmany different properties have been designed foror adapted to air sampling requirements. Theseinclude fibrous (e.g. glass, quartz), membrane (e.g.cellulose nitrate, polycarbonate, Teflon) and sin-tered (e.g. silver) filters. Lippmann [12] providesa comprehensive list of commercially availablefilter types, their sampling efficiencies and man-ufacturers. No single filter type is suitable for allapplications, and thus the choice of filter type fora given application depends greatly on the pro-posed analysis method for the collected sample.For example, membrane filters have the advantagethat they can retain particulates effectively on theirsurface (good for microscopy), whereas fibroustypes have the advantage of providing in-depthparticle collection and hence a high load-carryingcapacity (good for gravimetric assessment).

Filters are available in a range of dimensions(e.g. 25–100 mm diameter) and pore sizes (e.g.from 0.1 to 10 µm). General criteria which must beconsidered when selecting filter media are: (1) atleast 99 % collecting efficiency for particulates

COLLECTION OF AIRBORNE PARTICULATE MATTER 61

0.3 µm and greater in size, (2) low hygroscop-icity, since a hygroscopicity exceeding 1 mg perpiece leads to serious errors in weight concentra-tion measurement, and so an improper estimate ofthe environmental concentration, and (3) absenceof impurities that might interfere with the anal-ysis. As an example, glass fiber or Teflon fil-ters have been found to be unsuitable for thesampling of airborne dust with low platinumcontents [13]: only polycarbonate and cellulosenitrate gave blank values as low as 5 pg Pt pertotal filter.

Additionally, the filters should be mechanicallyand thermally stable and should not interactchemically with the deposit, even when subjectedto strong extraction solvent [8]. The chemicalrequirements of the filter depend on the natureof the proposed analysis. Spini et al. [14] havereported the reduction of Cr(VI) to Cr(III) whencellulose filters were extracted with alkalinesolution containing a known amount of Cr(VI).The same results were obtained by an aciddissolution (H2SO4) of the filters, which canbe explained by cellulose’s well-known reducingproperties. Therefore cellulose filters must beavoided for chromium speciation in airborneparticulates.

Teflon membrane and quartz fiber are the filtersmost commonly used for particulate chemical anal-yses, while cellulose fiber filters lend themselvesnicely to impregnation for absorbing gaseous pre-cursors, and etched polycarbonate membrane filtersare best suited for microscopic analyses.

2.2 Particulate sampling systems

A variety of PM samplers are available forcollecting particulates onto a filter. The flow ratethrough a sampler and filter is usually establishedaccording to specifications of the sampling methodand is held constant so that an accurate sampleand representative volume can be determined.Stationary PM samplers are used to evaluate bothoutdoor and indoor work environments. Personalbreathing zone samplers worn by workers areused to estimate their exposure to workplacepollutants.

2.2.1 Samplers for ambient airparticulate matter

The commercial market for PM sampling is largelydriven by the need to comply with U.S. ambientair quality standards. Since these standards specifymass concentration within a specific particle sizerange (PM10) the majority of the reference andequivalent samplers are designed for this purpose.

High-volume samplers are commonly used tocollect particulates with aerodynamic diametersless than 10 µm (termed PM10) at flow ratesof 1.1–1.4 m3 min−1. In these systems, PM iscollected on a large upwards-facing rectangularfilter, which is located inside a large, weatherproofhousing. Glass fiber and Hi-Vol samplers havebeen used for more than 50 years to measure airpollution, and thus procedures for these samplersare well established [8]. One drawback of theseunits is that frequent inlet cleaning is necessaryfor accurate size sampling.

For determining particle size distribution, im-pactors are commonly used. Particles are collectedon impaction plates depending on their size:coarser ones on plates in the upper stages, andsmaller ones on plates in the lower stages. Twodifferent impactor systems are used for size-fractionated PM collection: virtual impactors andcascade impactors. The physical principles of bothtypes are based on the mass moment of inertiacombined with the air resistance of particles, whichcause PM to be expelled from a sharply deviatedair stream. The resulting aerodynamic diameter(AD) of particles, which is different from thereal geometric diameter, therefore defines the sizedistribution. Impactors are easy to handle, arecommercially available and have well-defined cut-off characteristics.

If one is interested in a detailed size distribu-tion measurement of the airborne trace elements,cascade impactors are often used. They allow par-ticles to be sampled and segregated with respectto their size [15]. The cascade impactor consistsof a series of nozzles and coated impaction plates,called stages, with each stage collecting progres-sively smaller particles. Size segregation is depen-dent on sampler geometry and flow-rate. Cascadeimpactors can be quite variable in terms of stage


numbers and particle ‘cut-off’ (AD with 50 %collection efficiency) by changing the geometricdesign and the corresponding flow rate. However,artifacts like particle ‘blow-off’ or ‘bounce-off’from the collection substrate can appear [16].

Virtual impactors, on the other hand, are oftenconstructed for only two or three particle sizesand operate at relatively low flow rates. Theydiffer from cascade impactors in that the impactionplate is replaced by an opening that directs largerparticles to one sampling substrate, and the smallerparticles to another. This principle is used inthe commercially available PM10 dichotomous(dichot) sampler, which was the original samplerfor inhalable particles. This device uses a virtualimpactor to separate the fine (PM2.5) and coarse(PM10 –PM2.5) size fractions at a total flow rateof 16.7 L min−1. This is achieved by acceleratingparticles through a nozzle and then drawing 90 %of the flow stream off at right angles. The fineparticles follow the right-angle flow stream, whilethe larger particles continue toward the collectionnozzle. Particles are collected on 37 mm filters,which in this unit are generally Teflon or othermembrane filters. Separation of the two sizefractions minimizes potential interactions betweenthe more acidic fine fractions and the more basiccoarse fractions. Relative to cascade impactors, theeffects of particle bounce-off and re-entrainmentare reduced. However, this method requires thatpart of the total flow (typically 10 %) be drawnthrough the virtual surface. Therefore, correctionfactors must be applied to the coarse channel flowto account for contamination by the fine fraction.

More detailed descriptions of the above sam-pling devices and other samplers in current useare given in several monographs [9, 10].

2.2.2 Samplers for particulates in workplaceatmospheres

Sampling and analysis of work atmospheres aresimplified by two factors: (1) Industrial hygienistsusually know which contaminant or contaminantsare present in workroom air from the nature of theprocess plus a knowledge of raw materials, endproducts, and wastes. Therefore, identification of

workroom contaminants is rarely necessary and, asa rule, only quantification is required. (2) Usually,only a single contaminant of importance is presentin the workroom atmosphere and the absenceof obvious interfering substances permits greatsimplification of procedures [17].

One sampling device used is the cyclonesampler. Cyclones have found increasing use inrecent years as the first stage in two-stage samplersfor respirable mass dust exposure determinations.They use an impeller to impose a circulatorymotion upon air entering a cylindrical tube.General principles of centrifugal and gravitationalforces are used in the cyclone sampler to separatePM into various size fractions. By design, cyclonesused for respirable dust sampling are highlyefficient for removal of larger particles (i.e. greaterthan 10 µm) and are not efficient for particlesbelow about 2 µm. Cyclones are simple to operate,and can be very compact in design. As a result,cyclone samplers have been applied in a numberof personal monitoring applications.

Personal exposure monitors (PEMs) are sam-pling devices worn on the body to estimate an indi-vidual’s exposure to air pollution. As such, theyprovide a better representation of what individualsactually breathe than do fixed outdoor monitorsor even indoor monitors. PEM design is typi-cally based on cyclone or impactor samplers andcascade impactors have been identified as beingparticularly useful. Samplers for the inhalable frac-tions have already been proposed [18] and it isexpected that samplers for the thoracic and res-pirable fractions can be obtained with relativelysmall modifications to the design and/or mode ofoperation of existing respirable PM samplers (e.g.cyclones) [19]. In terms of filters, mixed-celluloseester membrane and polyvinyl chloride (PVC) fil-ters are often used for collecting metallic dusts forchemical analysis.

Among the currently available PEMs, the GSP(Strohlein GmbH, Kaarts, Germany) is one of themost precise with a low sampling bias [20]. Theclosed-face 37 mm filter cassette (SKC Inc., EightyFour, PA) which is widely used in the UnitedStates, undersamples 50 µm mass median aerody-namic diameter (MMAD) particle distribution by

COLLECTION OF VOLATILE COMPOUNDS 63

40 % and 12 µm MMAD particle distribution by10 %. Other samplers, including the IOM (Instituteof Occupational Medicine, Edinburgh, Scotland,UK) sampler (Negretti Automation, Aylesbury,UK, and SKC Inc., Eighty Four, PA), the ‘SevenHole Sampler’: (SKC Inc., Eighty Four, PA, andCasella Ltd., London, UK), the CIP10 samplers(Arelco, Fontenay-Sous-Bois, France), the PER-SPEC sampler (Lavoro e Ambiente, Bologna,Italy), and the 37 mm open-face filter cassette,do not always perform with the accuracy [20]required by the European Standards Organizationstandard on the assessments for measurement ofairborne particle concentration. Because of thelower flow rates and poorer flow controllers usedin these personal devices, they are believed to havepoorer precision than ambient samplers [21].

For further information, reviews of workplacemeasurements of coarse and fine PM areuseful [6, 7]. In addition, statistical samplingstrategies recommended in developing efficientprograms to monitor occupational exposures toairborne concentrations of chemical substances canbe found in the manual published by the NationalInstitute for Occupational Safety and Health [17].

3 COLLECTION OF VOLATILECOMPOUNDS

A variety of sampling methods and systemscan be employed for volatiles of interest. Ingeneral, methods for volatile compounds involvewhole air sampling or preconcentration of samplesusing liquid adsorbents, cryotrapping, adsorbentcartridges or impregnated surfaces [11]. Eachsampling method includes the following steps:(1) selection and preparation of sampling media,(2) the actual sampling process, (3) transport andstorage of the collected samples. The selectionof the optimal sampling method for targetcompounds (or a class of compounds) dependsgreatly on the physicochemical nature of thesecompounds and their expected concentration inair; sample volume must be compatible withthe sensitivity of the analysis method, and theexpected behavior of targeted compounds during

each step of the sampling process must be carefullyconsidered [11].

Sampling of whole air with containers ofdefined volume such as plastic bags (usually Teflonor Tedlar), cylinders with inlet and outlet valves,or stainless steel canisters is attractive becausethere is no risk of analyte breakthrough (unliketrapping), and no effect of moisture upon sampling.However, this method has two limitations: (1) thesample volume is limited to a few liters, which,for low compound concentrations encountered inair samples, may not be sufficient for analysispurposes; (2) sample stability during storage issometimes in doubt due to adsorption on (ordesorption from) container walls and chemicalreactions between compounds.

Thus, one of the most widely used methodsfor sampling of gaseous contaminants is precon-centration, either on a suitable solid adsorbent, bycryotrapping or, if the contaminant is reactive, inan absorbing solution contained in a bubbler orimpinger or coated on a solid porous support [11].Porous polymers, such as Tenax-GC, XAD resins,and polyurethane foams (PUF), have found wideapplication in gas sampling. Other types of sor-bents such as various types of charcoal, carbonmolecular sieves and other carbon-based sorbentsare also widely used, especially for more volatilecompounds. All solid adsorbents must be cleanedand tightly sealed prior to use. The cleaning pro-cedure depends on the type of adsorbent.

Cryogenic concentration of volatile compoundsis usually performed in an empty tube or atube filled with glass beads and cooled by liquidnitrogen. Usually, a second cryotrap is needed(cryofocusing) in order to allow narrow bands toenter the gas chromatography columns and thusto enhance resolution. However, the use of liquidnitrogen on-site, carrying a pump as well as thenecessary power supply, is not very convenient andthe large amounts of liquid nitrogen needed duringa sampling trip represent a major hazard duringtransport. In addition, plugging problems wereexperienced when sampling atmospheres with highlevels of humidity [22].

The application of diffusion denuders in com-bination with chemical analysis is another suitable


method for the determination of a variety of tracegases in the atmosphere [23]. In comparison totechniques using filter methods, the separationof reactive gas and particle phases by diffusiondenuder techniques is expected to be less subjectto artifacts.

Denuder systems can distinguish between com-pounds in gaseous and particulate form. In theirsimplest form, they consist of straight tubes theinside walls of which are coated with a suitableadsorber material acting as a sink for the gas ofinterest. Gases, due to their large diffusion veloc-ity compared to PM, can reach the walls of thetube and be adsorbed to the coating. After sam-pling for the desired period of time, the wet coatingis washed off and the determination is performedon the liquid extract. Unlike gases, however, PMpasses the denuder without touching the walls, asits diffusion is slow. In fact, denuders can be usedto trap gaseous components which could interferewith filter measurement. For example, the denudertechnique has been used to trap sulfur dioxide,which is responsible for the reduction of sampledCr(VI) on the filter surface [24].

4 SAMPLE HANDLINGAND STORAGE

Adequate contamination control is a prerequisite inall atmospheric trace element research, but is par-ticularly needed when the research is carried out inremote (or even semi-remote) areas. The samplingshould be carried out far enough from local sourcesto avoid contamination, and in atmospheric PMcollections it is strongly recommended that a sam-pling control device be used to monitor wind speedand direction and/or condensation nuclei counts.Furthermore, metal-free PM and deposition col-lectors should be employed, and they should bethoroughly cleaned with acid prior to use.

As important as contamination control duringsampling is the avoidance of contamination duringsample handling, storage and chemical analysis.It is therefore strongly advised to have a laminarflow clean bench in the field for all critical samplehandling (e.g. loading or unloading of filters). In

the laboratory, all critical manipulations should bedone at least on a clean bench, but for elementaldetermination at very low levels the use of a cleanroom is recommended. To minimize contaminationof filters, filter holders have been designed sothat filters can be loaded and unloaded in a cleanenvironment rather than in the field. Considerationshould be given to the material used to constructthe filter holder, particularly when measuringreactive components of particulate matter.

Particles can fall off filters when sampleshave large deposits and receive rough handlingduring movement from the field site to thelaboratory. Shorter sampling durations and lowerflow rates may be required to prevent overloadingin very polluted environments, especially those inwhich fugitive dust is a large contributor. Carefulhandling during transport will also minimize theloss of particles from the filter handling.

Sample integrity during storage is anotherimportant issue in atmospheric trace elementresearch. For particulate matter samples, some ele-ments may undergo changes (e.g. Cr(VI) reduc-tion) as a result of chemical reactions which takeplace on the collection substrate during samplestorage. Such losses may be minimized by storingthe samples in closed polypropylene vessels undera pure nitrogen atmosphere [25, 26]. For volatilemetal species, sample stability during cryogenicstorage has been evaluated and discussed [27].

5 APPLICATIONS

The aim of this section is to focus on varioussampling methods used in speciation studies ofairborne metals. However, many of these methodsare purely operational and do not specificallyidentify the metal species.

5.1 Lead

Environmental pollution from lead is a problemarising mainly from the use of tetraalkylleadcompounds as anti-knock additives in gasoline.Although this use is diminishing, the more stable

APPLICATIONS 65

forms, tri- and dialkyllead, are fairly persistent inthe environment. It has been shown that even inremote areas with no direct sources of emissionfrom traffic the concentrations of alkyllead speciesfound are correlated with the introduction andproliferation of motor vehicles. Other sources suchas mining, smelting and chemical production onlybecome major contributors in localized areas.

Most reported data concerning lead speciationin atmospheric samples are focused on thedetermination of tetraalkyllead (R4Pb) species inthe gas phase. However, it should be pointedout that R4Pb does not account for the totallead present in the gas phase. It has beensuggested that R4Pb decomposes in the atmosphereto produce vapor phase R3Pb+ and R2Pb2+species and inorganic lead PM [28]. A varietyof sampling techniques using solid adsorbentsoperated at ambient temperature or at extremelylow temperature (−40 ◦C or lower) have beenreported [28–32]. Due to the practical difficultiesincumbent in employing cold-trapping techniquesin the field, the use of solid adsorbents operatedat ambient temperature is recommended [28,31]. Among the solid adsorbents (PUF, activecharcoal, Amberlites, Chromosorb, Tenax andPoropak), only Tenax and Poropak gave recoveriesof more than 90 % for Et4Pb [32]. The morevolatile compounds, such as Me4Pb and Me3Pb+,were adsorbed less on Tenax than on Poropak.Due to the possibilities for R4Pb decomposition(particularly for tetraethyllead) in the presenceof ozone [33], a Teflon tubing pre-filter packedwith iron(II) sulfate crystals [31] is included inthe sampling train. The consensus opinion alsoindicates that air samples should be filteredto remove particulate matter, although concernhas been expressed that lead alkyls may beadsorbed onto the collected PM [31] leading to lowrecoveries.

Particulate alkyllead is generally found atpg Pb m−3 levels, some three orders of magni-tude below the concentration of vapor phase alkyl-lead compounds. Ionic alkyl lead species can betrapped in two water-filled gas bubblers connectedin series [34] after a pre-filter. This simple pro-cedure is reported to allow reasonable recoveries

of ionic species. Both inorganic lead and alkylleadparticulate species were effectively removed fromgas-phase tetraalkyllead compounds by a highlyefficient particle filter.

It should be noted that alkyllead species do notaccount for all the lead present in the gas phase.The largest fraction is inorganic lead, probablyparticulates of PbBrCl (not retained on a 0.45 µmfilter) derived from the combustion of antiknockingadditives in gasoline [28].

Particulate lead ambient air samples are oftencollected using high-volume samplers [35]. Expo-sure to lead occurs not only through inhalationbut also through ingestion of particles, whichmay be too large to be inhaled, especially byyoung children. Therefore, the high-volume sam-pler provides a more complete measure of expo-sure to airborne lead than the PM10 sampler, whichby design excludes particulates larger than res-pirable size.

5.2 Chromium

Cr(III) and Cr(VI) are the most common oxi-dation states of chromium in the environment.Whereas the Cr(III) species is essential for ani-mals and plants in trace concentrations, Cr(VI) istoxic and carcinogenic [36]. It has been speciallyclassified by the U.S. Environmental ProtectionAgency (EPA) as a group A inhalation carcino-gen [37]. Especially when occupational exposureis likely, inhalation is believed to be a majorhuman exposure pathway for Cr(VI). Occupa-tional exposure to airborne Cr(VI) has been asso-ciated with a number of work activities, includingmetal plating, welding, spray painting, leather tan-ning, dye and pigment manufacturing, and clean-ing of various parts prior to protective paint-ing, especially in the aircraft and automobileindustry [38].

Because of the relationship between chromiumtoxicity and oxidation state, the possibility ofchanging the valences of chromium due toreduction and/or oxidation must be eliminated.This has led to a need to define appropriatesampling, storage and analytical procedures.


The usual method of workplace monitoring ofparticulate chromium species is to collect PMsamples on filters for subsequent extraction andanalysis. Different filter materials have beenused during the last two decades, and due toreduction of Cr(VI) on the filters, inconsistentdata originating from this problem have beenreported [39]. Cellulose nitrate, cellulose acetate,PVC, PTFE and glass fiber filters have beenthe ones most widely used. Most methods forCr(VI) occupational monitoring (e.g. NIOSH [40],HSE [41], and OSHA [42]) specify PVC filtersfor sample collection, since other types of filtersare known to cause Cr(VI) reduction [14, 39, 43].However, as this reduction apparently occurs on aslow time scale, certain filter types such as glassfiber have showed reduction of Cr(VI) to a muchlesser extent than the others, so it is recommendedas a possible sampling material [14, 39, 44].

Inaccurate estimation of exposure to Cr(VI)is also caused by its instability during sampling,extraction or even sample preparation due to theenrichment of particles on the filter and thusenhanced contact with gaseous species, e.g. SO2,NOx , O3, and/or reaction on the filter with co-collected material, e.g. Fe(II), As(III)-containingcomponents. Due to the pH-dependent electro-chemical potential of the reduction of Cr(VI) toCr(III), losses of Cr(VI) are expected to increasewith decreasing pH of the filter surface. The useof the denuder technique to trap sulfur dioxide,which is responsible for the reduction of sampledCr(VI) on the filter surface, is proposed by Rohlingand Neidhart [24]. Losses of Cr(VI) between 16and 57 % depending on the composition of the PMand the sampling time were found. In addition, ifthe filters are not treated directly after sampling,storage in closed polypropylene vessels under apure nitrogen atmosphere is recommended [25,26]. Under these conditions the chromium speciesare stable for months [25].

An elegant solution to avoiding problems withCr(VI) instability is to analyze samples immedi-ately following collection, as recently proposedusing field-portable analytical methods [45–47]. Inone study, a portable spectrometer and a portablesolid phase extraction manifold were transported

to the field. An industrial hygiene survey wasconducted during sanding and spray painting oper-ations, and air samples were collected and analyzedfor Cr(VI) on site. Some area air samples were alsocollected above an electroplating bath. The resultsshowed this new method to be useful for bothenvironmental and industrial hygiene purposes.Another study developed a simple and readilyfield-adaptable system for automated continuousmeasurement of Cr(VI) in airborne particulate mat-ter. The system alternately collected the sample onone of two glass fiber filters. After 15 min of sam-ple collection on one filter, the sampling switchedover to the second filter. The freshly sampled filterwas washed for 8.5 min and the washings precon-centrated on an anion exchange minicolumn. Thewashed filter was dried with filtered hot air for thenext 6.5 min so that it was ready for sampling atthe end of the 15 min cycle [46].

5.3 Mercury

Mercury has long been identified as a poten-tial health and environmental hazard. Unlike mostother trace metals, the high volatility of mer-cury causes it to be present in the vapor form.In the atmosphere, the three main forms ofHg are: elemental Hg vapor (Hg0), which rep-resents more than 95 % of the global atmo-spheric mercury burden [48], reactive gas phaseHg (RGM) and particulate phase Hg (TPM). Ofthese three species, only Hg0 has been tentativelyidentified with spectroscopic methods [49] whilethe other two are operationally defined species,i.e. their chemical and physical structure can-not be exactly identified by experimental methodsbut are instead characterized by their propertiesand capability to be collected by different sam-pling equipment.

RGM is defined as water-soluble mercuryspecies with sufficiently high vapor pressure toexist in the gas phase. The reactive term refers tothe capability of stannous chloride to reduce thesespecies in aqueous solution without pretreatment.The most likely candidate for RGM species isHgCl2 and possibly other divalent mercury species.TPM consists of mercury bound or adsorbed to

APPLICATIONS 67

atmospheric particulate matter. Several differentcomponents are possible: Hg0 or RGM adsorbedto the particle surface, or divalent mercury specieschemically bound to the particle or integrated intothe particle itself. Another species of particularinterest is methylmercury (MeHg) due to its highcapacity to bioaccumulate in aquatic foodchainsand to its high toxicity. Since MeHg is only presentat low pg m−3 concentration levels in ambientair, it is not an important species for the overallatmospheric cycling of Hg.

Sampling and analysis of atmospheric Hg areoften done as total gaseous mercury (TGM) whichis a fraction operationally defined as speciespassing through a 0.45 µm filter or some othersimple filtration device such as quartz woolplugs, and which are collected on gold or othercollection material. TGM is mainly composedof elemental Hg vapor with minor fractions ofother volatile species such as HgCl2, CH3HgCl or(CH3)2Hg.

Gold-coated denuders were developed for re-moval of Hg vapor from air but have not beenapplied to air sampling [50]. Potassium chloride-coated tubular denuders followed by silver-coateddenuders have been used to collect HgCl2 (RGM)and elemental Hg emissions from incinerators [51]and for gaseous divalent mercury in ambientair [52–54].

For particulate Hg, a variety of different fil-ter methods using Teflon or quartz fiber filtershave been applied [55–57]. Recently, a collectiondevice based on small quartz fiber filters mountedin a quartz tube was designed [56]. The mer-cury collected on the filter can be released ther-mally, followed by gold trap amalgamation andcold vapor atomic absorbance spectrometry.

An excellent intercomparison for sampling andanalysis of atmospheric mercury species wasrecently reported [58]. Methods for sampling andanalysis of TGM, RGM and TPM were usedin parallel sampling over a period of 4 daysin Tuscany, June 1998. The results for thedifferent methods employed showed that TGMcompared well whereas RGM and TPM showeda somewhat higher variability. The relationshipsfor the measurement results of RGM and TPM

improved over the sampling period, indicating thatactivities at the sampling site during set-up andinitial sampling affected the results, in particularthe TPM results. Additional parallel sampling wasperformed for two of the TPM methods undermore controlled conditions, which yielded morecomparable results.

5.4 Platinum group metals

The emission of three metals of the platinum group(PGMs) (Pd, Pt, and Rh) from automobile catalyticconverters into the environment is of potential con-cern for human health. Platinum may affect thehealth of people through direct contact with plat-inum in dust, by inhalation of dust, and indirectlythrough the food chain. A maximum exposurelimit over 24 h of 2 µg m−3 of airborne, water-soluble platinum salts has been recommended bythe U.S. Occupational Safety and Health Adminis-tration. However, there are no data on the effects ofexposure to water-insoluble platinum compounds.Environmental concentrations of PGMs have notbeen shown to directly affect ecosystems or resultin direct health risks [59]. However, Pt was bioac-cumulated by rats exposed to a model substancewhich resembled Pt-containing particulates emit-ted by automobiles [60], and Pt was found to reactwith DNA [61].

There is not much literature available on thesubject of PGMs in air. Several types of filtershave been tested in order to avoid high PGMblank values [13, 62]. Some filter materials, e.g.glass fiber or Teflon, showed very high platinumcontents and were therefore unsuitable for thesampling of airborne dust with low platinumcontents. Only polycarbonate and cellulose nitrategave blank values as low as 5 pg Pt per totalfilter. Since knowledge regarding the particle sizedistribution of Pt metals in airborne particulatematter is critical to allow for a full assessment oftheir toxic potential and associated risk to humanhealth, there have been several impactor studiesdone to produce size-fractionation data on thisdistribution [13, 63].


5.5 Radionuclides

The radioactivity of PM has also attracted muchattention. Radioactivity in atmospheric PM hasthree major sources: emanations of radon andthoron gas, radioactive isotopes from cosmic rays,and anthropogenic radionuclides. Most of theseprocesses produce primary particulates, which laterattach to larger particulates by thermal diffusion.

The importance of speciation to characterizethe behavior of nuclides in the environment hasspawned a very large number of investigations,yet our knowledge is still limited and much morework is needed [64]. This unusual situation arisesfrom the difficulties in assessing radionuclides atvery low environmental concentrations and fromproblems related to the chemical (e.g. sorption,ion exchange) and physical (i.e. radioactive decay)instability of many species. Immediately after theirformation by nuclear reactions and/or radioactivedecay, radionuclides exist as single, highly kinet-ically and electronically excited atoms or ions,whose chemical behavior is very difficult to pre-dict and investigate. Very often it is a challengeto collect and analyze samples without chang-ing the identity of the real species. However, anadvantage of the determination of radionuclidesas compared to nonradioactive trace elements isthe practical absence of contamination during sam-pling and analysis.

Inhaled progeny of radon-222 are the mostimportant source of irradiation of the human respi-ratory tract and are clearly related to lung cancers,especially among miners of uranium and otherminerals [65]. As a natural product of the uraniumdecay series, 222Rn gas occurs in granitic and sim-ilar rocks and thus enters buildings or undergroundworkspaces constructed in these types of geologi-cal formations. Its own radioactive decay producessolid radon daughter elements which can attachto aerosols and reach the human respiratory tractin attached or unattached forms [66]. Like otherparticle-associated contaminants, radon’s healtheffects are heavily dependent on particle size dis-tribution, so size-specific sampling is critical. Oneoption is a modified cascade impactor known as alow pressure Andersen sampler, operated at lower

than standard pressures and thus able to selectmuch smaller diameter particles. In a study of col-lection substrates, stainless steel plates coated withsilicone grease provided the best size-selective per-formance with this sampler [67, 68]. For mea-surement of the alpha particles from the radonprogeny, a ZnS(Ag) scintillation counter was usedafter waiting 20 min for the decay of 218Po [67],which ensures that the appropriate species aretaken into account. Several devices which bothcollect and analyze air samples have also beendeveloped. The electrical low pressure impactor(ELPI) charges the aerosol particles and then mea-sures the resulting current at each impactor stageto produce size distribution data, though radioac-tivity can only be measured later after disassem-bly [69]. The portable RADON-check, however,concentrates the aerosol onto a grid and filter appa-ratus with in-line radiospectrometry, and can betuned to the aerosol-size retention function of thebronchi [70].

The radionuclide 129I has both cosmogenic andanthropogenic sources, and with a very long half-life (T1/2 = 15.7 × 106 year) will be a factor inpopulation dose estimates in spite of not being aradiological hazard at present. An efficient methodfor sampling iodine is to pump air through aTEDA-activated charcoal filter, which traps >99 %of the main forms of gaseous iodine, namely I2

and CH3I. Depending on the analysis method andthe data desired, the filter may need to be treatedimmediately to preserve its isotope ratios [71].

Filter collecting methods using impactors arecommonly utilized to characterize particulate radio-activity. Bondietti et al., using impactors to collectparticles in a certain size range, found 212Pb and214Pb mostly on PM with sizes <0.52 µm, with210Pb on particles of larger radii [72]. Among otherelements, 90Sr and 137Cs are found with smallerparticles, while 95Zr and 144Ce are often related tolarger ones [64].

Tritium is an anthropogenic contaminant ofconcern for environmental monitoring at nuclearfacilities because it is a byproduct of nuclearreactors. Ambient air concentrations of tritium(as HTO) are typically too low for practicalmeasurement using real-time devices; therefore

APPLICATIONS 69

adsorbents must generally be used to collect theatmospheric moisture, which is then analyzed forHTO. Both solid and liquid adsorbents can be usedto collect water vapor from ambient air, though thesolid types have several advantages. Compared toliquid adsorbent air sampling systems (bubblers orimpingers), solid adsorbents are more rugged (e.g.no breakable glass or evaporation problems) andallow for higher sampling rates and lower limitsof detection [73].

5.6 Metalloids

Volatile metalloid compounds have been identifiedin various anthropogenic gases such as domes-tic waste gas, landfill gas and sewage sludgedigester gas [22, 74, 75]. Most of these compoundsare thermodynamically unstable and thus undergochemical transformations. This feature is importantfor the choice of an appropriate sampling tech-nique. Possible phenomena causing analyte lossare diffusion, oxidation, hydrolysis, photodecom-position, adsorption, and heterogeneous surface-catalyzed breakdown [22].

So far, the method most used for samplingvolatile metalloid compounds like volatile metalshas been cryotrapping [74–81]. Feldmann andHirner [75] used chromatographic packing (SP-2100 10 % on Supelcoport −60/80 mesh) in a U-shaped glass tube immersed in liquid nitrogen asthe cryogenic liquid, for the sampling of variousmetalloids in landfill gas [79, 80] and in sewagegas [81]. Glass tubes packed with silanized glasswool have been used as sample collectors forcryofocusing of volatile metal species in urbanair [78].

The sampling of volatile metalloid compoundssuch as methylated, permethylated and hydridespecies of arsenic, antimony and tin has also beendescribed using Tedlar bags [22]. These bags arevery simple to use and a variety of differentvolumes can be sampled, from 1 to 100 L. Incontrast, stainless steel containers have a limitedvolume of a few liters, unless the sampled airis pressurized. To demonstrate the suitability ofTedlar bags for the sampling of such compounds,a series of stability tests were run using laboratory

synthetic and real samples analyzed periodicallyafter increasing periods of storage. The sampleswere stored in the dark at 20 and 50 ◦C. Basedon the results obtained, storage at 20 ◦C andanalysis carried out by the day after sampling wererecommended [22].

5.7 Miscellaneous applications

A number of studies have used size-selectivesampling and various pre-treatments of sam-ples to determine trace metal chemical forms orbinding forms, and thus obtain speciation infor-mation that can help to assess the relative bio-hazard of trace (and minor) elements associatedwith airborne particulates [82–94]. These methodsinclude: (1) separation of respirable airborne par-ticles from larger ones; (2) assessment of particlebioavailability through lung absorption by testingtheir solubility on air filters; and (3) assessmentof the probable state of chemical binding in oron the solid particles through sequential extrac-tion procedures. Different leaching solutions wereused in single and sequential extraction procedures.Although there is some controversy in the liter-ature over the use of this approach to chemicalspeciation [95], fractionation (‘operational specia-tion’) data from sequential extraction in combina-tion with particle size collection still provide usefulinformation on the source of the elements and ontheir potential bioavailability.

Such an approach was applied to determinevarious nickel species present in the airbornedusts of nickel-producing and nickel-using work-places [91, 92]. Nickel and some of its compoundsare identified as potential human carcinogens. Awet chemical procedure is described which appor-tions the airborne nickel species into four cat-egories: water soluble, ‘sulfidic’, ‘metallic’ and‘oxidic’ [91]. The suitability of various filter mate-rials for sampling airborne dusts for this fraction-ation study was also investigated. Sampling ofairborne dust is most frequently done with 37 mmmembrane filters (personal pumps) or with large203 × 254 mm sheets of fiber filters (high-volumesampling). If nickel is to be apportioned among


the various species in the sample, the filter mate-rial must be (a) chemically indifferent toward thesampled dust, (b) chemically resistant to bromineand methanol, and (c) readily wet-ashed to allowcomplete dissolution of the leached sample residue.Filter media that meet all the criteria are GelmanDM Metricel membranes and quartz fiber filters.Inferior for this purpose of fractionation are cellu-lose ester membranes (attacked by methanol), PVCand Teflon membranes (hydrophobic and difficultto destroy by acid digestion), and, most impor-tantly, glass fiber filters (surface alkalinity). Thenatural alkalinity of the glass fiber surface can reactwith soluble nickel compounds (normal salts) toform basic salts, which are only partly soluble. Toavoid undesirable secondary chemical changes inthe dust, quartz fiber filters should be used insteadof glass fiber filters [91].

In another study the procedure for speciation ofAs in size-fractionated filter-collected urban par-ticulate matter was described [93]. The sampleswere collected by Gent-type stacked filter unit andwere stored in plastic Petri dishes in a refrigeratorat 4 ◦C until analysis. A mild sequential proce-dure to differentiate As species between water-extractable, phosphate-extractable and refractoryform was utilized.

6 SUMMARY

This chapter has given a brief outline of samplingmethodology and samplers used for collecting air-borne particulate matter and volatile pollutantsfor epidemiological and occupational studies. Themain motivation for the sampling of ambientairborne particles remains the prediction and pre-vention of adverse health effects on humans.The evaluation of worker exposure to poten-tially hazardous agents including metallic contam-inants in the workplace is essential to establishingcause–effect relationships between an occupation-ally related illness and a specific agent(s).

There are a number of samplers that havebeen developed for epidemiological studies andto monitor individual exposures to health-relatedpollutants. Of all the airborne particulate matter

collection techniques, filter sampling is the mostversatile. With appropriate filter media, samplescan be collected in almost any form, quantityand state. Sample handling problems are usuallyminimal, and many analyses can be performeddirectly on the filter. Application of impactor-based selective sampling to obtain particle-sizedistribution of airborne particulate matter is ofinterest for the microcomposition and speciationof respirable submicron particles, which is ofparticular importance in environmental health.For sampling of gaseous contaminants such asvolatile metals and metalloids, cryotrapping andpreconcentration on solid adsorbents are widelyused methods.

A number of more specific methods havebeen developed to collect and determine varioustoxic metal species such as Cr(VI), Hg and Pbcompounds or platinum group metals in ambientair and workplace environments. In addition,various approaches (mainly operational) have beenproposed to estimate the behavior and fate ofatmospherically derived trace metals in orderto obtain information on their specific sourcesand reactions, and thus to assess their potentialhealth effects. These include appropriate selectivesampling methods in combination with varioussequential leaching procedures.

As links are established between workplaceexposure to pollutants and adverse health effects, itbecomes more and more important to develop andimplement reliable sampling and analysis methodsfor quantifying these pollutants.

7 REFERENCES

1. Dusseldorp, A., Kruize, H., Brunekreef, B., Hofsch-reuder, P., de Meerm, G. and van Oudvosrt, A. B., Am.J. Respir. Crit. Care Med., 152, 1932 (1995).

2. Prichard, R. J., Ghio, A. J., Lehmann, J. R., Win-sett, D. W., Park, P., Gilmour, M. I., Drher, K. L. andCosta, D. L., Inhal. Toxicol., 8, 457 (1996).

3. Samet, J. M., Stonehuerner, J., Reed, W., Devlin, R. B.,Dailey, L. A., Kennedy, T. P., Bromberg, P. A. and Ghio,A. J., Am. J. Physiol. (Lung Cell. Molec. Physiol.), 272,L426 (1997).

4. Smith, K. R. and Aust, A. E., Chem. Res. Tox., 10, 828(1997).

REFERENCES 71

5. Chao, C. C., Lund, L. F., Zinn, K. R. and Aust, A. E.,Arch. Biochem. Biophys., 314, 369 (1994).

6. Vincent, J. H., Analyst , 119, 13 (1994).7. Vincent, J. H., Analyst , 119, 19 (1994).8. Lodge, J. P. Jr., Methods of air sampling and analysis,

Lewis Publishers, Chelsea, MI, 1989.9. Watson, J. G. and Chow, J. V., Particle and gas mea-

surements on filters, in Environmental Sampling forTrace Analysis , Markert, B. (Ed.), VCH, Weinheim, 1994,pp. 125–161.

10. Mark, D., Atmospheric aerosol sampling, in AtmosphericParticles , Harrison, R. M. and van Grieken, R. (Eds),John Wiley & Sons Ltd, Chichester, 1998, pp. 29–94.

11. Zielinska, B. and Fujita, E., Organic gas sampling, inEnvironmental Sampling for Trace Analysis , Markert, B.(Ed.), VCH, Weinheim, 1994, pp. 163–184.

12. Lippmann, M., Filters and filter holders, in Air SamplingInstruments for Evaluation of Atmospheric Contaminants ,8th Edn, Cohen, B. and Hering, S. V. (Eds), AmericanConference of Government Industrial Hygienists, Cincin-nati OH, 1995.

13. Alt, F., Bambauer, A., Hoppstock, K., Mergler, B. andTolg, G., Fresenius’ J. Anal. Chem., 346, 693 (1993).

14. Spini, G., Profumo, A., Riolo, C., Beone, G. M. andZecca, E., Toxicol. Environ. Chem., 41, 209 (1994).

15. Lodge, J. P. and Chan, T. L., Cascade Impactor-Samplingand Data Analysis , American Industrial Hygiene Associ-ation, Akron, OH, 1986.

16. Markowski, G. R., Aerosol Sci. Technol., 7, 143 (1987).17. Leidel, N. A., Busch, K. A. and Lynch, J. R., Occupa-

tional Exposure Sampling Strategy Manual , U.S. Depart-ment of Health, Education, and Welfare, Cincinnati, OH,1977.

18. Mark, D. and Vincent, J. H., Ann. Occup. Hyg., 30, 89(1986).

19. Vincent, J. H., Tsai, P. J. and Warner, J. S., Analyst , 120,675 (1995).

20. Kenny, L. C., Aitken, R., Chalmers, C., Fabries, J. F.,Gonzales-Fernandez, E., Krombout, H., Liden, G.,Mark, D., Riediger, G. and Prodi, V., Ann. Occup. Hyg.,41, 135 (1997).

21. Wiener, R. W. and Rodes, C. E., Indoor aerosol andaerosol exposure, in Aerosol Measurement, Principles,Techniques, and Applications , Willeke, K. and Baron,P. A. (Eds), Van Nostrand Reinhold, New York, 1993,pp. 659–689.

22. Haas, K. and Feldmann, J., Anal. Chem., 72, 4205 (2000).23. Slanina, J. and Wyers, G. P., Fresenius’ J. Anal. Chem.,

350, 467 (1994).24. Rohling, O. and Neidhart, B., Fresenius’ J. Anal. Chem.,

351, 33 (1995).25. Dyg, S., Cornelis, R., Griepink, B. and Quevauviller, Ph.,

Anal. Chim. Acta, 286, 297 (1994).26. Nusko, R. and Heumann, K. G., Fresenius’ J. Anal.

Chem., 357, 1050 (1997).27. Amouroux, D., Tessier, E., Pecheyran, C. and Donard,

O. F. X., Anal. Chim. Acta, 377, 241 (1998).

28. Allen, A. G., Radojevic, M. and Harrison, R. M., Envi-ron. Sci. Technol., 22, 517 (1988).

29. Radziuk, B., Thomassen, Y., van Loon, J. C. and Chau,Y. K., Anal. Chim. Acta, 105, 255 (1979).

30. de Jonghe, W. R. A., Chakraborti, D. and Adams, F. C.,Anal. Chem., 52, 1974 (1980).

31. Hewitt, C. N. and Harrison, R. M., Anal. Chim. Acta,167, 277 (1985).

32. Nerin, C., Pons, B., Martinez, M. and Cach, J., Micro-chim. Acta , 112, 179 (1994).

33. Nielsen, T., Egsgaard, E., Larsen, E. and Schroll, G.,Anal. Chim. Acta, 124, 1 (1981).

34. Hewitt, C. N., Harrison, R. M. and Radojevic, M., Anal.Chim. Acta 188, 229 (1986).

35. Guidance for Siting Ambient Air Monitors Around Station-ary Lead Sources, Off. Air Qual. Plann. Stand., (Technol.Rep.): EPA-454/R-92-009, U.S. Environmental ProtectionAgency, Washington, DC, 1997.

36. International Agency for Research on Cancer, Mono-graphs on the Evaluation of the Carcinogenic Risks ofChemicals to Humans: Chromium, Nickel and Welding ,Vol. 49, IARC, Lyon, France, 1990, pp. 257–445.

37. Health Effects Assessment of Hexavalent Chromium , EPA540/1-86-019, U.S. Environmental Protection Agency,Washington, DC, 1984.

38. Criteria for a Recommended Standard, OccupationalExposure to Chromium (VI), National Institute for Occu-pational Safety and Health, Cincinnati, OH, 1975.

39. van der Wal, J. F., Ann. Occup. Hyg., 29, 377 (1985).40. Methods no. 7600 and 7604, in NIOSH Manual of Analyt-

ical Methods , 4th Edn, Eller, P. M. and Cassinelli, M. E.(Eds), National Institute for Occupational Safety andHealth, Cincinnati, OH, 1994.

41. Health and Safety Laboratory/HSE, Methods for Determi-nation of Hazardous Substances , MDHS Method no. 61,Health and Safety Executive, Sheffield, UK, 1998.

42. Method no. ID-215, in Hexavalent Chromium in Work-place Atmospheres , Occupational Safety and HealthAdministration, Salt Lake City, UT, 1998.

43. Sawarti, K., Ind. Health , 24, 111 (1986).44. Dyg, S., Anglov, T. and Christensen, J. M., Anal. Chim.

Acta , 286, 273 (1994).45. Wang, J., Ashley, K., Marlow, D., England, E. and Carl-

ton, G., Anal.Chem., 71, 1027 (1999).46. Marlow, D., Wang, J., Wise, T. J. and Ashley, K., Am.

Lab., 7, 26 (2000).47. Samanta, G., Boring, C. B. and Dasgupta, P. K., Anal.

Chem., 73, 2034 (2001).48. Ebinghaus, R., Hintelmann, H. and Wilken, R. D., Frese-

nius’ J. Anal. Chem., 350, 21 (1994).49. Edner, H., Faris, G. W., Sunesson, A. and Svanberg, S.,

Appl. Optics , 28, 921 (1989).50. Munthe, J., Xiao, Z., Schroeder, W. H. and Lindqvist, O.,

Atmos. Environ., 24A, 2271 (1991).51. Larjava, K., Latinen, T., Vahlman, T., Artmann, S.,

Siemens, V., Broekaert, J. A. C. and Klockow, D., Int. J.Anal. Chem., 149, 73 (1992).


52. Xiao, Z., Sommar, J., Wei, S. and Lindqvist, O., Frese-nius’ J. Anal. Chem., 358, 386 (1997).

53. Sommar, J., Feng, X., Gardfelt, K. and Lindqvist, O., J.Environ. Monitor., 1, 435 (1999).

54. Feng, X., Sommar, J., Gardfelt, K. and Lindqvist, O.,Fresenius’ J. Anal. Chem., 366, 423 (2000).

55. Keeler, G., Glinsorn, G. and Pirrone, N., Water Air SoilPollut., 80, 159 (1995).

56. Lu, J. Y., Schroeder, W. H., Berg, T., Munthe, J.,Schneeberger, D. and Schaedlich, F., Anal. Chem., 70,2403, (1998).

57. Berg, T., Bartnicki, J., Munthe, J., Lattila, H., Hreho-ruk, J. and Mazur, J., Atmos. Environ., 35, 2569 (2001).

58. Munthe, J., Wangberg, I., Pirrone, N., Iverfeldt, A., Fer-rara, R., Ebinghaus, R., Feng, X., Gardfeldt, K.,Keeler, G., Lanzillotta, E., Lindberg, S. E., Lu, J.,Mamane, Y., Prestbo, E., Schmolke, S., Schroeder, W. H.,Sommar, J., Sprovieri, F., Stevens, R. K., Stratton, W.,Tuncel, G. and Urba, A., Atmos. Environ., 35, 3007(2001).

59. Rosner, G. and Merget, R., in Anthropogenic PlatinumGroup Elements and Their Impact on Man and theEnvironment , Alt, F. and Zereini, F. (Eds), Springer,Berlin, 1999, pp 267–281.

60. Artelt, S. and Levsen, K., in Anthropogenic PlatinumGroup Elements and Their Impact on Man and theEnvironment , Alt, F. and Zereini, F. (Ed.), Springer-Verlag, Berlin, 1999, pp. 217–226.

61. Rosenberg, B., Van Camp, L. and Krigas, T., Nature,205, 698 (1965).

62. Rauch, S., Lu, M. and Morrison, G. M., Environ. Sci.Technol., 35, 595 (2001).

63. Zereini, F., Wiseman, C., Alt, F., Messerschmidt, J.,Muller, J. and Urban, H., Environ. Sci. Technol., 35, 1996(2001).

64. von Gunten, H. R. and Benes, P., Radiochim. Acta, 69, 1(1995).

65. Yu, K. N., Wong, B. T. Y., Law, J. Y. P., Lau, B. M. F.and Nikezic, D., Environ. Sci. Technol., 35, 2136 (2001).

66. Park, C., Progr. Phys. Geog., 17, 473 (1993).67. Yamasaki, K., Yamada, Y., Miyamoto, K. and Shimo, M.,

in Proceedings of IRPA-10 , International RadiologicalProtection Association, P-1b-42, Hiroshima, 2000.

68. Yamasaki, K., in Radon and Thoron in the HumanEnvironment , Katase, A. and Shimo, M. (Eds), WorldScientific Singapore, 1998, pp. 73–78.

69. Yamada, Y., Miyamoto, K., Tokonami, S., Shimo, M.and Yamasaki, K., in Proceedings of IRPA-10 , Inter-national Radiological Protection Association, P-1b-05,Hiroshima, 2000.

70. GRIMM Aerosol Technik GmbH and Co. KG, Ainring,Germany. http://www.grimm-aerosol.de.

71. Lopez-Gutierrez, J. M., Garcıa-Leon, M., Schnabel, C.,Schmidt, A., Michel, R., Synal, H.-A. and Suter, M.,Appl. Rad. Isot., 51, 315 (1999).

72. Bondietti, E. A., Papastefanou, C. and Rangarajan, C.,Aerodynamic size association of natural radioactivitywith ambient aerosols, in Radon and its Decay Products:Occurrence, Properties, and Health Effects , Hopke, P. K.,(Ed.), American Chemical Society, Washington, DC,1987, pp. 377–399.

73. Patton, G. W., Cooper, A. T. Jr. and Tinker, M. R., HealthPhys., 72, 397 (1997).

74. Feldmann, J., Grumping, R. and Hirner, A. V., Frese-nius’ J. Anal. Chem., 350, 228 (1994).

75. Feldmann, J. and Hirner, A. V., Int. J. Environ. Sci.Technol., 60, 339 (1995).

76. Pecheyran, C., Quetel, C. R., Lecuyer, F. M. M. andDonard, O. F. X., Anal. Chem., 70, 2639 (1998).

77. Jiang, S. G., Chakraborti, D. and Adams, F. C., Anal.Chim. Acta , 196, 271 (1987).

78. Pecheyran, C., Lalere, B. and Donard, O. F. X., Environ.Sci. Technol., 34, 27 (2000).

79. Feldmann, J. and Cullen, W. R., Environ. Sci. Technol.,31, 2125 (1997).

80. Feldmann, J., Koch, I. and Cullen, W. R., Analyst 123,815 (1998).

81. Feldmann, J. and Kleimann, J., Korresp. Abwasser 44, 99(1997).

82. Jervis, R. E., Krishnan, S. S., Ko, M. M., Vela, L. D.,Pringle, T. G., Chan, A. C. and Xing, L., Analyst , 120,651 (1995).

83. Dreetz, C. D. and Lund, W., Anal. Chim. Acta, 262, 299(1992).

84. Spokes, L. J., Jikells, T. D. and Lim, B., Geochim. Cos-mochim. Acta , 15, 3281–87 (1994).

85. Chester, R., Murphy, K. J. T. and Lin, F. J., Environ.Technol. Lett., 10, 887 (1989).

86. Brunori, C., Balzamo, S. and Morabito, R., Int. J. Envi-ron. Anal. Chem., 75, 19–31 (1999).

87. Polyak, K., Bodog, I. and Hlavay, J., Talanta , 41, 1151(1994).

88. Hlavay, J., Polyak, K., Bodog, I., Molnar, A. and Mes-zaros, E., Fresenius’ J. Anal. Chem., 354, 227 (1995).

89. Bodog, I., Polyak, K., Csikos-Hartyani, Z. and Hlavay, J.,Microchem. J., 54, 320 (1996).

90. Radlein, N. and Heumann, K. G., Fresenius’ J. Anal.Chem., 352, 748 (1995).

91. Zatka, V. J., Warner, J. S. and Maskery, D., Environ. Sci.Technol., 26, 138 (1992).

92. Fuchtjohann, L., Jakubowski, N., Gladtke, D., Bar-nowski, C., Klockow, D. and Broekaert, J. A., Fresenius’J. Anal. Chem., 366, 142 (2000).

93. Slejkovec, Z., Salma, I., van Elteren, J. T. and Zemplen-Papp, E., Fresenius’ J. Anal. Chem., 366, 830 (2000).

94. Hoffmann, P., Sinner, T., Dedik, A. N., Karandashev,V. K., Malyshev, A. A., Weber, S. and Ortner, H. M.,Fresenius’ J. Anal. Chem., 350, 34 (1994).

95. Bermond, A. P., Environ. Technol., 13, 1175 (1992).

CHAPTER 3

Sample Preparation

3.1 Sample Treatment for Speciation Analysisin Biological Samples

Riansares Munoz Olivas and Carmen CamaraUniversidad Complutense de Madrid, Spain

1 Introduction: Biological Samples ofInterest . . . . . . . . . . . . . . . . . . . . . . . . . . 741.1 Clinical samples . . . . . . . . . . . . . . . 741.2 Vegetables samples . . . . . . . . . . . . . 751.3 Nutritional samples . . . . . . . . . . . . . 75

2 Elements of Interest . . . . . . . . . . . . . . . . 762.1 Aluminium . . . . . . . . . . . . . . . . . . . 762.2 Arsenic . . . . . . . . . . . . . . . . . . . . . 762.3 Chromium . . . . . . . . . . . . . . . . . . . 772.4 Copper . . . . . . . . . . . . . . . . . . . . . . 772.5 Mercury . . . . . . . . . . . . . . . . . . . . . 772.6 Lead . . . . . . . . . . . . . . . . . . . . . . . 772.7 Selenium . . . . . . . . . . . . . . . . . . . . 772.8 Tin . . . . . . . . . . . . . . . . . . . . . . . . . 78

3 Sample Pretreatment Procedures . . . . . . . 783.1 General points of sampling and

storage . . . . . . . . . . . . . . . . . . . . . . 783.2 Clinical samples . . . . . . . . . . . . . . . 78

3.2.1 Release of the species out of thecells . . . . . . . . . . . . . . . . . . . . . 78

3.2.2 Selection of the group of species 793.2.3 Desalting . . . . . . . . . . . . . . . . . 793.2.4 Fractionating techniques . . . . . . 79

3.3 Nutritional samples . . . . . . . . . . . . . 794 Clean-Up . . . . . . . . . . . . . . . . . . . . . . . . 80

4.1 Elimination of lipids . . . . . . . . . . . . 804.2 Clean-up of extracts . . . . . . . . . . . . 80

5 Evaluation of Several ExtractionProcedures . . . . . . . . . . . . . . . . . . . . . . . 805.1 Aqueous extraction . . . . . . . . . . . . . 805.2 Simple solvent extraction . . . . . . . . 815.3 Enzymatic extraction . . . . . . . . . . . 815.4 Solid-phase extraction (SPE) . . . . . . 815.5 Supercritical fluid extraction (SFE) 825.6 Accelerated solvent extraction (ASE) 83

6 Derivatisation Techniques to GenerateVolatile Species . . . . . . . . . . . . . . . . . . . 836.1 Hydride generation (HG) . . . . . . . . 836.2 Cold vapour . . . . . . . . . . . . . . . . . . 876.3 Ethylation . . . . . . . . . . . . . . . . . . . 876.4 Grignard reactions . . . . . . . . . . . . . 876.5 Other methods . . . . . . . . . . . . . . . . 87

7 Preconcentration of the Species . . . . . . . . 877.1 Amalgam formation . . . . . . . . . . . . 887.2 Cold trap (CT) . . . . . . . . . . . . . . . . 887.3 High temperature trap . . . . . . . . . . . 887.4 Active charcoal retention . . . . . . . . 88

8 Separation and Identification Steps . . . . . 889 Accuracy of the Different Preparation

Steps: Need for Adequate CRMs . . . . . . . 899.1 Sources of error . . . . . . . . . . . . . . . 899.2 Relevance of CRMs . . . . . . . . . . . . 90

10 Trends and Perspectives . . . . . . . . . . . . . 9111 References . . . . . . . . . . . . . . . . . . . . . . . 92


74 SAMPLE PREPARATION

1 INTRODUCTION: BIOLOGICALSAMPLES OF INTEREST

There are different approaches in speciation anal-ysis depending on the problem or study that it isbeing undertaken:

• The most comprehensive one takes as its aim theidentification and quantification of all individualforms of the element.

• In some cases, especially for biological samples,some elements are strongly bound inside thetissues and the analyst is not able to obtain com-plete information about the real species presentin the sample. Furthermore, the extraction effi-ciency of the species detected depends on thesample treatment followed. As an illustrativeexample, the selenium species determined in theoyster tissue depend very much on the hydroly-sis conditions applied [1].

• Some analysis regards only a given species andthis is the object of determination. This is thecase for many elements with high differencein toxicity among its species (e.g. Cr(VI) andCr(III), Hg(II) and CH3Hg+, etc).

• Often it is necessary only to distinguish betweengroups of species without specifying the detailednature of the compounds (e.g. inorganic Hg/organic Hg).

• The term speciation is often used where thereis a distribution of an element among differ-ent physical fractions: dissolved or particulate(e.g. whole blood fractions: serum, plasma andblood cells).

There are many types of biological samples,and several subsets of this type may be identi-fied. Three general categories can be considered:(1) clinical, including animal and human tissue andbody fluids, organs, etc.; (2) vegetables, includingtissues of terrestrial and aquatic plants, dissolvedorganic material and algae; (3) nutritional, foodproducts and related substances.

1.1 Clinical samples

The need for information regarding the mobil-ity, storage, retention and toxicity of species

in a biological system is essential. Corneliset al. [2] have classified the elemental speciesfound in clinical samples regarding their toxicityas follows: (i) small organometallic molecules,generally contaminants of food, water or air.Some remain unchanged in the body (e.g. organ-otin, organolead compounds, methylmercury, . . .);(ii) biomarkers of contamination exposure by thetoxic species conversion into less toxic ones bythe body (e.g. inorganic arsenic species that canbe methylated by the organism); (iii) elementswith different oxidation states and thus differ-ent toxicity (e.g. Cr(III) and Cr(VI); (iv) traceelements that form a metal–ligand complexwith a low molecular weight compound, orbound to a protein (e.g. selenium in selenome-thionine, etc.); and (v) elements participatingin a biomolecule (e.g. Cu in ceruloplasmin).Szpunar has reported another useful classificationof metal species in an interesting review concern-ing bioinorganic speciation [3]. From this classifi-cation, six groups of species have been exposed:(a) molecules with a metal–carbon bond, e.g.selenoamino acids and organoarsenic compounds;(b) complexes with macrocyclic chelating agents,including tetrapyrroles, cobalamins and porphyrins;(c) complexes with nucleobases, such is thecase of many therapeutic drugs of Pt, Au, Ru,etc.; (d) metal complexes with proteins includingenzymes, e.g. metallothioneins which bind metalswith sulfur affinity (Cd, Cu, Zn); (e) complexeswith polysaccharides or glycoproteins, especiallyin plants; (f) metallodrugs, such as platinum orruthenium complexes known in cancer therapy,and gold compounds used as antiarthritic drugs asthe most significant examples.

The criteria governing the validity of a clini-cal sample required for analysis depends on theinformation needed. Then, definition of samplingand treatment should fit the final purpose of theanalysis, e.g. diagnosis of deficiency or toxicity,control of environmental or biological pollution,nutritional surveillance or forensic investigations.Speciation information from these samples can beused with two purposes: (i) to define and evaluatethe mechanisms of release of species into the bio-logical matrix and the degree of interaction with

INTRODUCTION 75

the environment; (ii) to know the absorption anddistribution mechanisms throughout the organism,their bio-availability, toxicity and excretion path-ways [4, 5].

Elemental speciation in this field still representsa great challenge due to the low concentrationof the species, their poor stability and the matrixinterference. In addition, some aspects have to besolved, such as understanding their role in physio-logical and pathological processes, the availabilityand practicability of analytical methods, and finallythe ethical considerations for sampling [5].

1.2 Vegetables samples

Most studies related to the analysis of plants havebeen concerned with the effects of elements onplant growth, element uptake, toxicity, transloca-tion and soil–plant relationship [6]. Many studieswith these biological samples come from the abil-ity of some plants to grow in the presence of highheavy metal concentrations. This growth is a resultof a variety of tolerance mechanisms, e.g. metalbinding at the cell wall, precipitation in vacuoles,and synthesis of metal binding compounds such asproteins, peptides, organic acids and phenol com-pounds [7]. One of the most abundant mechanismsof metal binding in plants seems to be the synthesisof phytochelatins and small peptides with high cys-teine content that have the ability to chelate heavymetals and then to reduce the concentration of freemetal ions in the cytosol. A detailed knowledge ofthe metal-binding characteristics of phytochelatinsis necessary to understand the role of these pep-tides for metal detoxification in plant systems.

1.3 Nutritional samples

Food science in the broadest sense can be extendedto include soil chemistry, plant uptake and, at theend of the food chain, the metabolic fate of ele-mental species when certain foods are consumedby humans or animals. Samples analysed are ofa great variety: from the relatively simple, e.g.fruits or vegetables, to more complex such as pro-cessed whole meals, diets etc. [8]. A number of

incidents have highlighted the need for elementalspeciation studies in food, one example being theMinamata Bay incident, in Japan [9], where fishaccumulated high amounts of the methylmercurydischarged into the bay by the Chisso Corporationfactory. This bio-accumulation led to a massivepoisoning of local people with fatal consequences.

The chemical form of an element in fooddetermines the mode of absorption in the intestineand the subsequent metabolism processes that mayoccur. The bio-availability of a species in food mayalso be measured and is defined as a measure ofthe proportion of the total amount of a nutrient thatis utilised for normal body functions [10].

The amount of a particular species absorbedinto the body after ingestion is difficult to mea-sure as the mechanisms of absorption, individualdifferences in requirements and metabolic con-trol often varies depending on the individual’sresponse [10]. International legislation for food isbased on the total element content and not the indi-vidual species, but some regulations and guidelinesare being introduced [11].

Reports of elemental speciation in all thesesamples have increased over the last few years.Speciation information may be used to determinethe fate of a trace element species in a biologicalsystem. In addition, the molecular forms that occurin a biological sample may also be monitored atthe cellular level, which is the key to understandingmany of the biotransformation processes in mostbiological systems.

Different steps are generally required for speci-ation analysis in biological samples: sample col-lection and handling, storage, evaluation of thespecies stability until analysis, sample pretreat-ment, complete species extraction from the matrix,clean-up procedures before separation, preconcen-tration and derivatisation as sensitivity improve-ment methods, etc. All these steps, from samplingto analysis mean that the integrity of the speciesis often a risk and they have been summarised inFigure 3.1.1.

This chapter aims to treat each step separately,discussing in depth the main considerations to takeinto account depending on the sample type and theinformation required.


Sampling and sample storage

Separation and species quantification

Sample treatment

Method validation

General and specific related problems of speciation in biological samples

Low analyte concentration

Several species for analyte

Complexity of the matrix

Stability of species

Cleanup

Extraction

Derivatisation Pre-concentration

Figure 3.1.1. Main problems of metal speciation in biological samples.

2 ELEMENTS OF INTEREST

It is interesting to have a look at the main chem-ical species of the elements that have primarilyconcerned in biospeciation.

2.1 Aluminium

The risk of aluminium in the organism hasbeen associated with several factors; these includea wide variety of uses, e.g. in packing andbuilding materials, paints, industrial exposures,hemodialysis treatment, etc. Additionally, free Almay compete with other metals, causing damagesin the activity of some enzymes, and even withDNA, probably by cross-linking the protein chains.Al is suspected of being involved in a number ofneurological disorders, e.g. Alzheimer disease, andsome liver diseases [12, 13]

Different Al species has been extensively stud-ied in serum and plasma of dialysis patients: theirconcentration in blood is very high due to theirreversible uptake of Al from dialysate solutionsand through the oral intake of Al-containing

medicines. The main compounds present in bloodare Al–transferrin, Al citrates and Al bicarbonates;they can be deposited in various parts of the bodywith a very long biological half-life. The species ofthis element have also been widely studied over thelast decades because of some Al–hydroxo com-plexes that are highly toxic for fish [6, 11, 14].

2.2 Arsenic

Arsenic is an element that raises much concernfrom both the environmental and human healthpoints of view. The different species of As pro-duce diverse toxicological effects in human, withinorganic forms being more toxic than organicones. For food control, the objective consists indiscerning the toxic inorganic forms from theharmless organo-arsenicals. Although foodstuffsyield total As concentrations below the legallimit, in some cases fish derivatives exceed thesevalues. However, the As present in the lattersamples is presumably arsenobetaine and so nottoxic. Biomonitoring of occupational exposure toarsenic needs to study different As species (As(III),

ELEMENTS OF INTEREST 77

As(V), MMA, DMA, As–betaine, As–choline,As–sugars) in a widerange of samples: serum,plasma, urine, tissues, . . . . It is important to quan-tify all these species in order to identify the causeof exposure and the metabolism of such speciesinto the body [14, 15].

2.3 Chromium

The difference in toxicity between Cr(III) andCr(VI) is well known, especially when they arepresent in living organisms [2]. As a consequence,Cr(III) is an essential trace element; conversely,Cr(VI) is known to be carcinogenic and mutagenic.For instance, if Cr(VI) is inhaled it is able tocross cell membranes and is transported by theblood cells, damaging them. Of particular concernis the regulation of total Cr level in the air(at the workplace) because of the high level ofdamage caused by this element when inhaled [14].Differentiation between those forms of chromiumis important for assessing food and water safety.

2.4 Copper

Copper is a common contaminant in drinkingwater, released by the piping for household taps.The corrosion caused by the Cu present in thewater results in a layer of CuO2 with smallamounts of other oxides. These species can also beconverted into different ones, and even bound toorganic matter. We still do not know much aboutthe toxicity of the different compounds that canresult from these transformations. In addition, theuse of pesticides and fungicides containing thiselement for many years has led to soil and watercontamination. Thus, speciation of Cu is crucial forunderstanding the mobility and risks to plants andconsequently to animals [14]. In contrast, copper isconsidered an essential element in animal nutrition:in most in vivo environments, copper is predomi-nantly associated with macromolecules, but thereare other low molecular mass (LMM) species thatcan traverse the cell membranes and are potentiallymore toxic [16].

2.5 Mercury

Mercury is probably the element most studied fromthe toxicological point of view. All forms areconsidered poisonous; however, the objective willconsist in discerning the highly toxic alkyl com-pounds (mostly methylmercury) from less toxicinorganic Hg forms. Some episodes of healthdamage by organomercury compounds are wellknown: in Minamata (Japan), MeHg contaminationcaused severe brain damage in infants whose moth-ers ingested contaminated fish during pregnancy.In Iraq, poisoning of human happened becauseof the consumption of wheat seeds treated withorganomercury insecticides. In general, exposureto organic Hg can cause brain damage to a devel-oping foetus since CH3Hg+ readily crosses theplacenta. It is very important then, from the nutri-tional and the health point of view, to developtechniques sensitive enough for the monitoring orrapid screening of Hg species in very varied matri-ces: food (mainly fish tissue), blood, urine, hair,brain tissue, etc. [14, 17].

2.6 Lead

Lead is considered to be a toxic element mainlydue to some neurological problems, renal dys-functions, hypertension and cancer, for whichthere is evidence for animals but not yet forhumans [18]. The most important group of speciesis the organolead compounds (RnPbX), which arequite toxic. The more substituted the organic chainthe higher the toxicity. Organolead compounds arevery labile and easy transformations take placebetween them. Lead (tetraalkyllead) has been usedas an additive in petrol for many years (it is nowforbidden in many Western countries) and this isone of the main reasons for concern about lead inbiological systems [19, 20].

2.7 Selenium

Selenium has been shown to be essential and toxic,depending on the concentration as well as on the


species present. The main essential role identifiedfor selenium has been as a component of theenzyme glutathione peroxidase (GSH-Px), one ofthe antioxidant defence systems of the body [21].Keshan disease is the most famous illness causedby a deficit on Se in the human and animal diet,due to the low content in soil and water in thatChinese area [22]. Food supplements containingSe–cysteine have been an object of study in recentyears because of their beneficial action in someareas of low selenium content in the soil [14].

The most common selenium species to be con-sidered are the inorganic selenium (Se(IV), Se(VI)),the biomethylated species (DMSe, DMDSe) andthe selenium amino acids (Se–cysteine, Se–methi-onine, Se–cystathionine, . . .) and other compoundswith higher molecular weight. Selenium specia-tion has also been undertaken in plants, blood andanimal tissues to monitor the occupational expo-sure to this element as well as problems related toits deficiency.

2.8 Tin

Butyltin and phenyltin compounds are very wellknown and studied because of their high toxicityfor aquatic life. Fortunately, levels of the mosttoxic compound, tributyltin (TBT) have been reg-ulated in most countries [14]. TBT has becomea common contaminant of fish and shellfish. Theregulation clearly establishes a daily intake thatit is still exceeded in many regions [23]. Nowa-days, it is important to measure the levels presentin the different tissues and body fluids in orderto clarify the metabolism and the transformationor detoxification mechanisms of the organism thathas ingested it.

3 SAMPLE PRETREATMENTPROCEDURES

3.1 General points of samplingand storage

The measurement of different trace element speciesrequires a sample preparation and special handling

that must be designed for each particular sample.Numerous steps in an analytical procedure increasethe risk of losses by adsorption to the containersor due to chemical instability. It enhances the pos-sible contamination from reagents and equipment,as well.

First of all, collection of biological samplesneeds a number of procedures because the majorproblem in such work is due to contaminationand losses of the trace elements. Some samplingprocedures have been presented in Chapter 2.2 ofthis book.

Concerning the storage of the samples, precau-tions such as full details on collection, freezing orlyophilisation of samples, addition of preservativeor/and additive, etc. must be done by the analystprior to the analysis as a routine. These proce-dures have already been treated in Chapter 2.2, ofthis book.

3.2 Clinical samples

The preliminary steps of pretreatment of clinicalsamples include, in many cases, the release ofspecies out of the cells, the selection of aparticular group of species, desalting procedures,and fractionating techniques.

3.2.1 Release of the species out of the cells

Whole blood is usually centrifuged to obtain serum,which is free of fibrinogen. Otherwise, an antico-agulant can be added to obtain a supernatant orplasma. Serum or plasma is homogenised by vor-texing and, at this point, it can be stored at −20 ◦Cor the analyst can proceed with the remaining stepsof the analysis [2]. In the case of tissues, the chem-ical speciation begins with the separation of thesoluble species from those bound to insoluble com-pounds. For this purpose, the tissue is subjected toan ultrasonic homogenisation in an isotonic buffer(pH = 7.4). The homogenate is centrifuged andthe supernatant, containing the soluble compounds,is subjected to different fractionating techniquesthat will be discussed later. Speciation of elementsbound to insoluble compounds that remain in the

SAMPLE PRETREATMENT PROCEDURES 79

precipitate cannot be pursued any further [2]. Thedistribution of trace elements in tissues can also beapproached at the subcellular level, among cytosol,mitochondria and nucleus. Pretreatment of urinesamples can be limited to a cleaning step by pass-ing it through a C18 cartridge and subsequentfiltration and appropriate dilution with water ordeproteination using ultrafiltration [24]. However,at room temperature, bacterial action can rapidlycause changes in the sample such as pH and for-mation of precipitates or change/interconversion ofspecies. Therefore, the pH is important to notewhen collecting urine samples.

3.2.2 Selection of the group of species

The first decision to make for elemental specia-tion in a clinical sample is whether low molecularmass (LMM) or high molecular mass (HMM) com-pounds are going to be investigated. The separationof both groups is generally done by ultrafiltrationyielding solutions free of protein [25]. The sepa-ration is characterised by the cut-off value of themembrane, this being the maximum molar massof the proteins able to pass through the pores.The ultrafiltrate is collected in the space beyondthe membrane and contains all components (witha molecular mass below the cut-off of the mem-brane) at the same concentration as in the originalsample. The most common problem encounteredwith ultrafiltration is the changes in concentra-tion of the LMM species because the filter canbe fouled during the filtration process.

3.2.3 Desalting

Desalting is necessary when the ionic strength ofthe solution does not accord with the conditionsrequired for chromatographic separations. Gelfiltration chromatography, with a fractionatingrange of 1–5 kDa, can be used for this purpose.

3.2.4 Fractionating techniques

First of all, the separation of LMM trace elementspecies can be performed by ion exchange chro-matography (Dowex resins) as they are charged

species. Before the chromatographic step, it isnecessary to remove the HMM compounds byultrafiltration to prevent clogging of the column.The pH of this ultrafiltrate must be carefullyadjusted to that of the mobile phase before injec-tion. Concerning the trace element species boundto proteins, they are usually separated by fast pro-tein liquid chromatography (FPLC), which coversion exchange, reversed phase, size exclusion andaffinity chromatography. The choice of a buffer aschromatographic eluent can be determinant for thetype of species analysed: if it contains salts with astrong affinity for the protein it may provoke somereplacement reaction [2].

3.3 Nutritional samples

Fresh foods may have to be prepared by usingstainless steel knives and/or plastic choppers,depending on the species of interest. Domesticblenders, coffee grinders and food processors fittedwith stainless steel, Ti or ceramic blades are veryeffective for homogenising food samples. Somespecial considerations must be taken depending onthe nature of the sample [8]:

• Liquid foods: milk, wine, fruit juice, etc. onlyneed to be shaken before subsampling; sparklingbeverages should be degassed in an ultrasonicbath or by bubbling with any inert gas; thosewines which form a precipitate (some red wines)need to be decanted before subsampling.

• Meat: inedible parts (skin, hair, etc.) shouldbe removed and discarded unless a particu-lar contamination study on these materials isrequired. Meat is homogenised using a blenderor equivalent.

• Fish: before subsampling it is necessary toproceed as for meat, avoiding useless parts, e.g.head, fins and larger bones. A food blender isthe common way to homogenise samples.

• Vegetables: gross surface contamination shouldbe removed from both root and leafy vegetablesby washing with pure water. Analysis of dietaryintakes may require peeling root vegetables ordiscarding the outer leaves of vegetables.


• Fruit: the outer skin or peel may be a usefulindicator of surface contamination, and it canbe of interest to analyse this separately from theedible portions.

• Whole meals or diets: when a mixture of ingre-dients is to be analysed there are two options:(i) representative samples of the relevant foodcan be subsampled; (ii) portions of the wholediet can be combined, homogenised and stored.

4 CLEAN-UP

Biological samples have a fairly complex matrixdue to the presence of thousands of compounds ofvery different nature: hydrocarbons, polysaccha-rides, lipids, amino acids, glycerides, etc. There-fore, clean-up procedures are necessary in orderto remove those compounds or groups of com-pounds useless for the purposes of analysis. Someof the most common clean-up methods are pre-sented here:

4.1 Elimination of lipids

Lipid decomposition is very fast in biologicalmaterials, and it is convenient to extract them com-pletely from the sample. The method of extrac-tion and purification of lipids must involve a mildtreatment to minimise their oxidative decompo-sition, which creates secondary compounds thatcould affect the stability of the species. One of thefastest and more efficient methods uses a mixtureof chloroform and methanol in such proportionsthat a miscible system is formed with the waterpresent in the tissue. Dilution with chloroform andwater separates the homogenate into two layers,the chloroform layer containing all the lipids, andthe methanolic layer containing all the nonlipids.Because of the high differences in humidity and/orlipid amount present in the sample, it is sometimesnecessary to dilute with distilled water. A purifiedlipid extract can be obtained by isolating the chlo-roform layer, if needed. The entire procedure canbe carried out in 10 min; it is efficient and freefrom deleterious manipulation [26].

4.2 Clean-up of extracts

It is also important to consider the clean-upneeded in many cases after the extraction step.Some extracts are ‘very dirty samples’ as a resultof a nonspecific extraction method. One of themost common procedures applied is by passingthe extract through an alumina column to retainthe high saline and other ionic compounds [27].The removal of the organic matrix from sampleswith a high organic content (like urine) can beachieved by using a simple treatment consisting ofsolid phase extraction with C18 cartridges [28]. Inaddition, this clean-up procedure has been foundto give a high stability for some compounds(selenium species) in this complex matrix.

5 EVALUATION OF SEVERALEXTRACTION PROCEDURES

Sample extraction procedures are often perceivedas bottlenecks in analytical methods. In the last fewyears, various attempts have been made to replaceclassical extraction techniques in order to reducethe volume of extraction solvents required and toshorten the sample preparation time. Several ofthese methods have been reported in the literature,using aqueous or organic solvents to solubilise thedifferent compounds; hydrolysis procedures (acid,alkaline and enzymatic hydrolysis); solid phaseextraction; and other enhanced techniques, suchas supercritical fluid extraction, accelerated solventextraction, etc.

5.1 Aqueous extraction

Different extraction procedures have been appliedfor Se speciation in samples like yeast, allium veg-etables and other plants. The first approach was asimple hot water extract; the second one was anacid hydrolysis with 0.1 M HCl in a methanolicmedium [29]. The extraction needs several hoursand the recovery and the selenium species profile isnot complete. In the case of arsenic, an ultrasound-assisted extraction with MeOH + HCl (50 % +

EVALUATION OF EXTRACTION PROCEDURES 81

10 %) has also been tested in environmental andbiological matrices with similar results to those forselenium [30].

5.2 Simple solvent extraction

Most organic solvent extraction methods serveonly to isolate species of interest from the majormatrix in order to eliminate interference with thesubsequent steps of the analytical method, that is,derivatisation, separation and detection. Samplesare commonly first digested with basic or acidreagents to completely solubilise the sample, thenneutralised prior to extraction with the appropri-ate solvent. Finally, the species are back-extractedfrom the organic solvent into an aqueous phase ifrequired by the analytical separation or detectiontechnique. As an illustrative example, some solventextraction procedures developed for mercury spe-ciation in biological samples include solubilisationby using KOH/MeOH or TMAH/MeOH (methano-lic tetramethylammonium hydroxide); extractionby adding dichloromethane, hexane, CH2Cl2 [31,32] or dithizone chelating reagent and chloroform[33]. The organic compounds are back-extractedinto the aqueous phase with sodium thiosul-fate solution buffered with ammonium acetate orwith potassium bromide/copper sulfate mixture tofavour the process. In all these cases, recoveriesof CH3Hg+ higher than 80 % are obtained whenapplied to different fish tissues and to sea plants.

5.3 Enzymatic extraction

Nonspecific enzymes are capable of breakingdown a wide range of proteins into their aminoacid components. In this way, enzymes can beemployed for digesting biological fluids, mainlyblood [34]. Depending on the enzyme and thedigestion conditions applied, this method can beapplied for elemental speciation. This approachhas been widely used for tin and seleniumspeciation in different biological materials [1, 35].The main enzymes used with extraction purposesinclude protease, lipase, trypsin, pepsin, pronase,

or mixtures of them. Some authors have found thatthe addition of additives like methanol or citricacid helps the solubilisation of some compounds(TBT in the case of tin) by complexation, andprevent bacterial growth during the incubationperiod, sometimes 48 h [36].

Generally, the use of enzymatic hydrolysis pro-cesses has shown better results in the release ofthe species of interest from solid biological sam-ples [29, 35–37]. A study performed by Morenoet al. [1] has evaluated the efficiency in both aque-ous soluble and solid fractions. The complexity ofsuch a process is clearly evidenced in Figure 3.1.2.Apart from this complexity, the use of enzymatichydrolysis for extraction in biological samples isvery convenient and widely applied.

5.4 Solid-phase extraction (SPE)

SPE was developed by Zhang and coworkersfor the extraction of pure organic compoundsfrom aqueous samples [38, 39]. In general, theanalyte from a relatively large volume of solutionis selectively retained by a solid reagent phase(by a variety of mechanisms) and then releasedinto a relatively small volume of eluent [30].Actually, it is being used for the extraction andpreconcentration of organometallic compounds,after a derivatisation into a volatile and apolarform, which easily evaporates from the aqueoussolution into the headspace of a gas chromatograph(GC). Then, it is adsorbed by the apolar phaseof the coated silica fibre of a SPME (solid phasemicro-extraction) device, establishing equilibriumof the analyte among the three phases: the aqueousphase, the headspace and the fibre coating. After10 min of exposure time, the fibre is inserted inthe GC injection port and the compounds thathave been collected are thermally desorbed forsubsequent analysis [40, 41]. The accuracy ofthis approach for biological samples has beentested by means of analysis of two referencematerials of fish tissue (NIES-11) and dogfishmuscle (NRC DORM-2). Another application ofSPE as an extraction/preconcentration method hasbeen the retention of inorganic lead complexed


ResidueTotal Se

Total Se

SeMetTMSe+

Total Se

10 kDa

10 KDa

10 KDa

10 kDa

Filtered

M.W

M.W.

M.W

M.W

HPLC

M.W.

Filtered

Total Se

HPLC

Filtered*

Total Se

Hydrolysed Total Se

E.H + 30 min., 14.000 g)

E.H+30 min. at 14.000 g

Total SeHydrolysed

Filtered*

Total Se

HPLC

HPLC

TMSe+

SeMet

TMSe+

+30 min sonication+30 min. 14.000 g

3 mL water

Supernatant Total Se

Hydrolysed

M.W

M.W.

M.W

M.W.

OYSTERTotal Se

E. H.+30 min. 14.000 g

*Stability study of the extracts

EH: Enzymatic hydrolysis

MW digestion: HNO3 + H2O2

TMSe+

SeMet

M.W.

Figure 3.1.2. Scheme of an enzymatic hydrolysis extraction procedure for selenium species in oyster tissue.

with DDC in a column of C18 silica, eluted withacetonitrile, or in a Chromosorb 102 column elutedwith methanol. This method was applied to fruitjuice [42].

5.5 Supercritical fluid extraction (SFE)

The properties of a supercritical fluid make thisextraction technique among the most interestingand promising to obtain extraction recoveriesclose to 100 %. The properties of supercriti-cal fluids that are attractive from an extraction

point of view include: (a) considerably great dif-fusion coefficients leading to efficient and rapidextractions; (b) low viscosity and absence ofsurface tension that facilitate pumping in theextraction process; and (c) density close to liquidsenabling the greater interactions on a molecularlevel necessary for the solubilisation. Temperatureand pressure changes, near the supercritical point,can affect the solubility by a factor of as muchas 100, or even 1000. Moreover, the use of fluidswith low critical temperature values (CO2, N2O,. . .) allows extractions under thermally mild con-ditions, protecting labile compounds [43].

DERIVATISATION TECHNIQUES TO GENERATE VOLATILE SPECIES 83

The main disadvantage is the fairly limitedability to study ionic or highly polar compounds[44]. This extraction technique has been extendedto supercritical fluid chromatography coupled toICP/MS, showing a high resolving power and100 % sample introduction efficiency. However,there are two main disadvantages: the necessityof using a heated transfer line to make possiblethe coupling, and the adverse effect to sensitivitycaused by the introduction of CO2 into theplasma [45].

5.6 Accelerated solvent extraction (ASE)

ASE is a new technique that uses organic solventsat high pressures and temperatures above theboiling point. A solid sample is enclosed in asample cartridge that is filled with an extractionfluid and used to statically extract the sample underelevated temperature (50–200 ◦C) and pressure(500–3000 psi) conditions for short periods oftime (5–10 min). Compressed gas is used topurge the sample extract from the cell into acollection vessel [46, 47]. Two main reasons areresponsible for the enhancement of ASE: (i) highersolubility of the analytes and better mass transferbetween the two phases, and (ii) disruption ofsurface equilibrium due to the high temperaturesand pressures applied (break-up of the strongsolute–matrix interactions, decrease of the solventviscosity allowing better penetration of matrixparticles, etc).

This technique usually requires the use offreeze-dried samples. In spite of the fact that ithas been widely used for extraction of organiccompounds, the use for metallic species determi-nation is very scarce and only a few papers havebeen published. Gallagher et al. [47, 48] have pro-posed this technique for the extraction of organo-arsenicals in seaweed. After extraction of arseniccompounds and solvent evaporation, the residuewas re-dissolved and prepared for HPLC by pass-ing it through a C18 cartridge for sample cleanup.It is important to consider the great influence ofthe matrix nature on the recovery of this techniqueas well as the particle size that depends on thenumber of cycles applied.

Examples of different extraction and othersample pretreatment strategies are summarised inTable 3.1.1.

The difficulties associated with species extrac-tion are illustrated in a recent intercomparisonexercise for selenium speciation in a white clovercertified reference material (BCR CRM 402) [49,50]. When comparing the results reported fromthis study, the extraction efficiencies differ con-siderably (from 15 to 75 %) and there is even noconsensus in terms of the species detected.

6 DERIVATISATION TECHNIQUESTO GENERATE VOLATILE SPECIES

As gas chromatography is often the method ofchoice in separation of species, the compoundsmust exhibit high volatility and thermal stability.The number of nonpolar organometallic speciesis fairly limited (tetraalkylleads, methylselenium,organomercury and naturally metalloporphyrincompounds). Then, derivatisation reactions mustbe used to transform the polar species intononpolar ones [19].

6.1 Hydride generation (HG)

The common method for derivatisation is the for-mation of volatile hydrides. This is applicable tocompounds of several elements as As, Sb, Bi,Ge, Pb, Se, Te, and Sn [4, 8, 51]. The inorganicforms of these elements react with sodium boro-hydride with formation of simple hydrides, but notall species react equally fast and in the same solu-tion conditions. This behaviour can in some casesbe used for the separation of the different species.Some of the alkyl derivatives also form volatilehydrides under similar conditions. The advantageof hydride generation is connected with their pre-concentration in a trap of liquid nitrogen; they arereleased by slowly raising the temperature. Thisgives a chance of simple separation, with sub-sequent determination of individual species [52].One of the biggest disadvantages of hydride gener-ation is the interference caused in the liquid phase.Interfering species can appear during the hydride


Tabl

e3.

1.1.

Dec

ompo

sitio

nan

dpr

etre

atm

ent

met

hods

for

spec

iatio

nin

biol

ogic

alsa

mpl

es.

Rep

rodu

ced

bype

rmis

sion

ofT

heR

oyal

Soci

ety

ofC

hem

istr

y.

Sam

ple

type

Com

poun

dde

term

ined

Sam

ple

trea

tmen

tSe

para

tion

dete

ctio

nR

efer

ence

Uri

neM

e 3Pb

+ ;E

t 3Pb

+A

djus

tpH

to10

with

NaO

H+

ED

TAto

com

plex

met

alio

nsan

dre

mov

epr

ecip

itate

byce

ntri

fuga

tion.

Pass

the

sam

ple

thou

ghsi

lica

gel

and

deso

rbM

e 3Pb

+an

dE

t 3Pb

+us

ing

acet

ate

buff

er(p

H=

3)in

10%

met

hano

l.A

ddbo

rate

buff

er(p

H=

7).

HPL

C70

HPL

C-I

CP/

MS

Sele

nium

spec

ies

Cle

an-u

p:C

18ca

rtri

dges

,pr

evio

usly

cond

ition

edw

ithM

eOH

+H

2O

.St

atio

nary

phas

e:Sp

heri

sorb

OD

Sco

lum

n28

Dilu

tion

into

aph

osph

ate

buff

er(p

H=

6).

Mob

ileph

ase:

phos

phat

ebu

ffer

HPL

C-I

CP/

MS

Sele

nite

and

sele

nate

Vol

unte

ers

supp

lem

ente

dw

ithSe

-Met

for

seve

ral

wee

ks.

Ani

on-e

xcha

nge

(Dio

nex

Ionp

acA

G11

-HC

)co

lum

n.E

lutio

nw

ith25

mM

NaO

Han

d2

%M

eOH

71

Sam

ple

dilu

tion

1+

1w

ithw

ater

.IE

C-E

TV

-AA

S/IE

C-D

AD

Hum

anse

rum

Mag

nesi

um–

prot

ein

com

plex

esPr

etre

atm

ent:

sam

ple

dilu

tion

two

fold

with

Tri

s-H

Cl

buff

eran

dsa

tura

ted

with

CO

2(p

H=

7.0

–7.

4).

Cle

an-u

pw

ithC

hele

x-10

0re

sins

.

Ani

onex

chan

geco

lum

n.E

lutio

nw

ithgr

adie

ntT

ris-

HC

lbu

ffer

(pH

=7.

4)/0

.2M

NaC

l72

SE-H

PLC

-IC

P/M

SPl

asm

a,ur

ine,

anim

altis

sues

Se,

Zn,

Cu,

and

Fesp

ecie

sB

lood

cent

rifu

ged

at12

00g

(2◦ C

)E

xtra

ctio

n:50

mM

Tri

s-H

Cl

(pH

=7.

4).

Hom

ogen

ised

inN

2at

mos

pher

e.C

entr

ifug

ed.

Mob

ileph

ase:

50m

MT

ris

HC

lbu

ffer

(pH

=7.

4)73

Blo

odM

e 4Pb

Hem

olys

eby

free

zing

at−2

0◦ C

(24

h)E

xtra

ctin

n-h

epta

ne(u

ltras

onic

bath

).C

entr

ifug

ean

dw

ithdr

awor

gani

cla

yer.

GC

74

Me 3

Pb+

Add

amm

oniu

mci

trat

e(p

H=

9)+

ED

TAan

dD

DT

C.

Ext

ract

twic

ew

ithpe

ntan

e.Se

para

tean

dco

mbi

neor

gani

cla

yers

.D

eriv

atis

eaf

ter

evap

orat

ion

with

2M

C4H

9M

gCl

inT

HF.

Add

pent

ane

and

1M

H2SO

4.

Sepa

rate

orga

nic

phas

esan

ddr

yw

ithan

hydr

ous

Na 2

SO4.

GC

74

HPL

C-A

AS;

HPL

C-I

CP/

MS

Plan

tsPh

ytoc

hela

tins

ED

TAin

wat

ercl

eanu

p+

filtr

atio

nT

ris-

HC

lbu

ffer

(pH

=8.

6)+

mer

capt

oeth

anol

(to

prev

ent

oxid

atio

nof

thio

lgr

oups

)D

isru

ptio

nof

cells

byso

nica

tion;

cent

rifu

gatio

n.C

ytos

olex

trac

tion

byfil

trat

ion

(0.2

2µ

m)

Rev

erse

dph

ase

colu

mn

Mob

ileph

ase:

amm

oniu

mac

etat

e(p

H=

7)7,

75

SeM

et,

Se-M

etC

yste

ine;

Se-C

ysth

atio

nine

Hot

wat

erex

trac

tion.

Enz

ymat

ichy

drol

ysis

with

10%

prot

ease

(sha

king

24h

atro

omte

mp.

).C

entr

ifug

atio

nan

dfil

trat

ion.

HPL

C76

Yea

st,

Whi

tecl

over

Se-C

yst,

Se-M

et,

Inor

gani

cSe

0.25

g+

wat

er/H

Cl

0.01

Mat

37◦ C

over

nigh

t+M

eOH

+C

HC

l 3+

H2O

shak

en(5

h)+

HC

l0.

01M

+pe

psin

/pro

nase

+lip

ase

stir

red

over

nigh

t.Fi

ltrat

ion

and

cent

rifu

gatio

n.

HPL

C-E

TAA

SPR

P-X

100

colu

mn

Mob

ileph

ase:

nick

elac

etat

e+

nick

elsu

lpha

te29

DERIVATISATION TECHNIQUES TO GENERATE VOLATILE SPECIES 85U

ltras

onic

atio

n(2

×30

min

.)in

MeO

H/H

2O

.Su

pern

atan

tfil

trat

ion

(0.4

5µ

m).

C18

cart

ridg

esan

dev

apor

atio

n.M

icro

bore

IEC

50

Yea

stSe

leni

umsp

ecie

sE

xtra

ctio

n:H

PLC

-IC

P/M

S•

Hot

wat

er(9

0◦ C

)•

10%

MeO

H+

0.2

MH

Cl

•30

mM

Tri

s-H

Cl(p

H=

7)+

0.1

mM

PMSF

•4

%D

rise

lase

in30

mM

Tri

s-H

Cl+

PMSF

•30

mM

Tri

s-H

Cl(p

H=

7)+

4%

SDS.

•20

mM

Am

mon

ium

phos

phat

e(p

H=

4)+

0.15

MN

aCl+

0.1

MPM

SF+

1m

ME

DTA

+5

%SD

S•

Am

mon

ium

phos

phat

e(p

H=

7.5)

+pr

onas

e+

lipas

ein

cuba

ted

16h

at37

◦ C•

25%

TM

AH

at60

◦ Cfo

r4

h.

•SE

Supe

rdex

-200

HR

colu

mn

(130

0kD

alim

it).

Elu

ent:

Tri

s-H

Cl

buff

er(p

H=

7)•

PRP-

X10

0co

lum

n.E

luen

t:A

mm

oniu

mph

osph

ate

buff

er•

Iner

tsil

OD

S-2

colu

mn.

Elu

ent:

0.1

%T

FA+

2%

MeO

H

77

HPL

CO

yste

rtis

sue

As-

Bet

,D

MA

,A

s-Su

gar,

As-

Res

idua

l

Ext

ract

ion:

focu

sed

mic

row

ave-

assi

sted

extr

actio

nw

ithM

eOH

/H2O

(1:1

)C

entr

ifug

atio

n;E

vapo

ratio

nto

dryn

ess;

Dilu

tion

with

wat

erup

to10

ml

Tobe

publ

ishe

d

Cle

an-u

p:60

0m

gC

18sn

ap-c

artr

idge

san

dfil

tere

d(0

.22µ

m)

nylo

nfil

ter

Ani

on-e

xcha

nge

colu

mn

Phos

phat

em

obile

phas

e(p

H=

6)T

BT

,D

BT

,M

BT

,T

PhT

,D

PhT

,M

PhT

Ext

ract

ion:

TM

AH

20%

(60

min

),ad

ditio

nof

HC

l(p

H=

7);

Add

ition

ofso

dium

Ac-

/HA

cbu

ffer

(pH

=4)

;E

xtra

ctio

nw

ithhe

xane

;D

eriv

atis

atio

nw

ithN

aBE

t 4

GC

-AE

D

TB

T,

DB

T,

MB

TE

xtra

ctio

n:m

etha

nol/H

Cl/t

ropo

lone

;So

nica

tion;

liqui

d–

liqui

dpa

rtiti

onin

gw

ithm

ethy

lene

chlo

ride

;ba

ck-e

xtra

ctio

nw

ithis

o-oc

tane

.G

rign

ard

deri

vatis

atio

n(p

enth

ylat

ion)

.C

lean

-up:

silic

aco

lum

n

GC

-MS

Oys

ter

tissu

eSe

-Met

,Se

-Cys

t 2,

TM

Se+ ,

Inor

g.Se

Frac

tiona

tion:

0.2

g+

Wat

er.

Cen

trif

uged

extr

actio

n:su

pern

atan

tan

dso

lidfr

actio

ns+

20m

gpr

otea

se+

Tri

s-H

Cl

0.1

M(p

H=

7.5)

.In

cuba

tion

(24

hat

37◦ C

).U

ltrafi

ltrat

ion

thro

ugh

10kD

afil

ter

HPL

C-I

CP/

MS

•A

nion

-exc

hang

ePR

P-X

100

colu

mn.

Mob

ileph

ase:

amm

oniu

mci

trat

e(p

H=

4.8)

•C

atio

n-ex

chan

gePR

P-X

200

colu

mn.

Mob

ileph

ase:

Pyri

dine

form

ate

(pH

=2.

8)

1

LC

-IC

P/M

SFi

shtis

sue

TB

T,

TPh

TE

xtra

ctio

n(C

O2

SFE

syst

em):

0.15

g+

MeO

Hat

100

◦ Can

d60

00ps

i.To

tal

time:

15m

in.

Mob

ileph

ase:

MeO

H/H

2O/A

c-(9

4:5

:1)

atpH

=6.

Ion

pair

ing:

pent

anes

ulfo

nate

4m

M.

78

Ext

ract

ion

(foc

used

mic

row

ave)

:O

rgan

otin

s•

0.2

g+

AcH

+N

aBE

t 4.

Tim

e:3

min

.•

0.2

g+

TM

AH

+A

cH(p

H=

5)+

isoo

ctan

e+

NaB

Et 4

.T

ime:

2m

in.

Cle

an-u

p:al

umin

aco

lum

n

MC

-GC

/MIP

-AE

S27

Shel

lfish

tissu

eM

BT

,D

BT

;T

BT

Ext

ract

ion:

0.2

gfr

eeze

-dri

ed+

buff

er(p

H=

7.5)

+10

mg

lypa

se+

10m

gpr

otea

se.

Incu

batio

nat

37◦ C

(4h)

.H

G-C

T-Q

F-A

AS

35

(con

tinu

edov

erle

af)


Tabl

e3.

1.1.

(con

tinu

ed)

Sam

ple

type

Com

poun

dde

term

ined

Sam

ple

trea

tmen

tSe

para

tion

dete

ctio

nR

efer

ence

Ars

enite

and

arse

nate

Sam

ples

free

zedr

ied

and

hom

ogen

ised

.To

tal

As:

mic

row

ave-

assi

sted

dist

illat

ion

(0.5

g+

KI+

asco

rbic

acid

+H

Cl

MW

-HG

-AA

S79

Ext

ract

ion:

0.5

g+

HC

lov

erni

ght.

Add

are

duci

ngag

ent+

chlo

rofo

rm(t

wic

e).

Bac

k-ex

trac

tin

1M

HC

l.C

entr

ifug

atio

nan

dfil

trat

ion.

Mar

ine

orga

nism

sA

rsen

icsp

ecie

sE

xtra

ctio

n:W

ater

/MeO

H(1

:1)

soni

cate

d(x

4);

filte

red;

puri

ficat

ion

with

diet

hyl

ethe

rph

enol

HPL

C-H

G-Q

FAA

S80

HPL

C-I

CP/

OE

S•

Ext

ract

ion:

met

hano

l/chl

orof

orm

(1:1

)•

Ext

ract

ion:

met

hano

l/wat

er(1

:1)

LC

-UV

-HG

-IC

P/O

ES

81

Cle

an-u

p:C

18ca

rtri

dges

LC

-UV

-HG

-IC

P/M

S82

Ani

on-e

xcha

nge

colu

mn.

Mob

ileph

ase:

phos

phat

ebu

ffer

sgr

adie

nt(p

H=

6)L

iver

Inor

gani

cse

leni

umE

xtra

ctio

n:50

mM

Tri

s-H

Cl

buff

er(p

H=

7)+

0.25

mM

gluc

ose

at25

◦ Cun

der

N2

atm

osph

ere.

Cen

trif

ugat

ion

at10

5.00

0g

at4

◦ C(1

h).

HPL

C83

Kid

ney

Se-C

yst,

Se-M

et,

Inor

gani

cSe

Ext

ract

ion:

HC

l0.

01M

+pr

onas

e+

lipas

est

irre

dov

erni

ght.

Filtr

atio

nan

dce

ntri

fuga

tion.

HPL

C-E

TAA

SPR

P-X

100

colu

mn

Mob

ileph

ase:

nick

elac

etat

e+

nick

elsu

lpha

te29

Hum

anm

ilkSe

-Cys

t,Se

-Met

,Se

-Cys

tam

ine

Cle

an-u

p:re

mov

alof

fat.

Prot

ein

prec

ipita

tion

byce

ntri

fuga

tion

at25

.840

gat

8◦ C

(30

min

)C

ZE

84

Food

supp

lem

ent

Se-M

etE

xtra

ctio

n:en

zym

olys

isw

ithpr

otea

seK

inN

aHC

O3

(pH

=7)

at37

◦ Cin

dark

ness

(20

h).

Filtr

atio

nw

ithny

lon

filte

rs(4

5m

m).

HPL

C85

Cer

eals

,M

ultiv

itam

ins

tabl

ets

Solu

ble/

inso

lubl

eco

pper

,zi

ncan

dir

onsp

ecie

s

•St

omac

hex

trac

tion:

enzy

mol

ysis

with

peps

in;

incu

batio

nfo

r4

hat

37◦ C

;pH

adju

stto

2.5

with

HC

l•

Inte

stin

eex

trac

tion:

enzy

mol

ysis

with

peps

in/p

ancr

eatin

/;in

cuba

tion

for

8h

at37

◦ C;

pHad

just

to7.

4w

ithN

aHC

O3

FI-I

CP/

MS

86

Cen

trif

ugat

ion

and

filtr

atio

nth

roug

h0.

45µ

mpo

resi

zefil

ters

PRECONCENTRATION OF THE SPECIES 87

formation or during their decomposition just beforethe detection step. In the first case, this is due toan inhibition of the hydride formation or a trans-formation into other compounds due to interactionswith other metals present in solution; in the secondcase, it can come from other hydride interactionsand/or analyte losses [51].

One possibility for hydride generation proce-dures is the use of an immobilised borohydridereagent. It has been described for arsenic andselenium [53, 54] in drinking water, where thederivatising reagent was immobilised on an anion-exchange resin and the hydride was generated onpassage of an acidified sample solution. The opti-mum tetrahydroborate concentration (0.05 %) wasconsiderably lower than typical values.

6.2 Cold vapour

This is similar to hydride generation. The dif-ference is the volatile product resultant from thereduction reaction: the volatile metal. It is onlyapplicable to Hg and Cd. It has been widely usedfor MeHg determination in several biological sam-ples due to the ability of this species to react withthe reductant borohydride [55, 56].

6.3 Ethylation

Another procedure of derivatisation is ethylationby using sodium tetraethylborate, applied for Sn,Se, Hg, Pb, . . . [32, 51, 57, 58]. Ethylation canbe accomplished in an aqueous medium; then,derivatisation and extraction frequently occur inthe same step. In addition, ethylation permits thesimultaneous derivatisation of organo -Sn, -Hg,and -Pb compounds. Nevertheless, liquid–liquidextraction remains necessary and organic solventsare required. Sometimes, there is no discriminationamong the different species of an element. This isthe case of Pb: Pb2+, Et3Pb+, or Et2Pb2+ all reactwith the reagent NaEt4B:

R3PbX + NaBEt4 ←−−→ R3PbEt + BEt3 + NaX

R2PbX2 + NaBEt4 ←−−→ R2PbEt2 + BEt3 + NaX

6.4 Grignard reactions

Derivatisation using Grignard reagents (alkyl orarylmagnesium chlorides) has a more universalcharacter. This has been applied for Sn, Hg, andPb speciation [59]. For example:

R3PbX + R′MgX ←−−→ R3PbR′ + MgX2

The main drawbacks of Grignard reagents aretheir atmospheric instability and their abilityto be hydrolysed in presence of water form-ing Mg(OH)2:

R3PbX + H2O ←−−→R′H + 12 Mg(OH)2

+ 12 MgX2

To avoid this problem, these reagents need tobe stabilised in ether and stored under an inertatmosphere. The water remaining is removed afterextraction with a complexing agent. Then, samplepreparation is laborious and time-consuming, andremoval of the excess of the derivatising reagentby water/acid addition is necessary to avoid highblank values.

6.5 Other methods

A very recent approach for derivatisation of aminoacids has been accomplished by esterificationof the carboxylic acid group using propan-2-ol,followed by the acylation of the amino groupwith trifluoroacetic acid. The derivatives are thenextracted into chloroform and analysed by GC-MS [60]. This new method has been applied toselenomethionine.

There are several other potentially applicableprocedures for speciation analysis, based on theformation of volatile chelates, such as trifluo-roacetylacetonates, dithiocarbamates [61]; volatileoxides and halides [68]; and carboniles. However,none of them has been reported for speciation anal-ysis in biological materials.

7 PRECONCENTRATIONOF THE SPECIES

The speciation analysis of trace elements oftenneeds a preconcentration step due to the very


low concentration of the compounds of interestin biological systems and the distribution of theselow total levels among several species. For thispurpose, metallic species can be trapped or retainedin solid adsorbents.

7.1 Amalgam formation

The best and most extensively used amalgamis that formed for mercury preconcentrationwith gold.

7.2 Cold trap (CT)

The analytes of interest are (after derivatisation)purged from the aqueous sample solution usingan inert gas. The gas stream is then dried, andthe species are cryotrapped in a fused silicacapillary column. Finally, sample introduction intothe chromatograph or the detector is accomplishedby slowly heating the trap, and the speciesare released depending on their volatilisationpoints [63]. A variation of this technique is thecapillary cold trap (CCT), in which the trap isreduced to a capillary column. This new designallows the preconcentration and separation of theanalytes in the same step. This technique has beenapplied to Sn, Pb and Hg speciation [51, 64].Based on this principle, an automatic speciationanalyser has been designed and employed formercury speciation in a fish reference material withexcellent results [51, 65].

7.3 High temperature trap

This consists of an electrothermal vaporisation sys-tem to raise the temperature during the collectionand the evaporation of the analytes. It is a very sen-sitive and precise method, and it has been used forarsenic determination in biological samples [66].

7.4 Active charcoal retention

This is a preconcentration technique mostlyemployed for trapping volatile chelates due to thenonpolar nature of such compounds [67].

8 SEPARATION ANDIDENTIFICATION STEPS

The hyphenation between chromatography andspectrometric detectors is the best approach forbiochemical speciation analysis. There are variousfactors to be overcome for a successful analy-sis [20]:

• The first requisite for a hyphenated technique isthat the species injected on a chromatographiccolumn leaves the column unchanged, andthat no artefact species are generated on thecolumn. The metal–ligand bound should bemuch stronger than the interaction of the metalor the ligand with the stationary phase.

• Secondly, the mobile phase must have a similarpH and composition to those of the samplematrix. Mobile or stationary phases competingwith or displacing ligands from the analyte metalcomplex should be avoided.

• The third point is that the chromatographictechnique should guarantee that the signalcorresponds to one particular species. Sincebiological samples are very complex mixturescontaining thousands of compounds, the use ofsuccessive separation techniques implying dif-ferent mechanisms may be required. Generally,metal complexes in biological macromoleculesare separated by ultrafiltration using filters (withmolecular cut-offs of 500, 5000 and 30 000 Da).Sometimes, a further purification of the com-pounds is necessary, using different techniques,such as hydrophobicity or electrical charge,affinity chromatography, etc. [2].

• Finally, the simplest method to identify thespecies of interest is by comparing their reten-tion times with that of a known standard. Thisapproach is fairly successful for organometal-lic species, but has not the same applicabilityfor biochemical compounds: standards are notavailable for the majority of the species present.Also, there is a probability of several specieshaving the same retention time. Therefore, ana-lytical techniques offering a characterisation ofthe metal complex or at least of the ligandare necessary, such as nuclear magnetic reso-nance (NMR), electrospray-mass spectrometry

ACCURACY OF PREPARATION STEPS 89

(ESI-MS), and liquid liquid mass spectrometry(LC-LC-MS) [20].

The possibilities for an on-line separationtechnique include gas chromatography, differenttypes of HPLC and electrophoresis. All separationtechniques can be coupled to specific detectors(atomic or molecular mass spectrometry) [3]. Thereader should refer to Chapter 4 of this book formore information about separation techniques andhyphenated detectors.

9 ACCURACY OF THE DIFFERENTPREPARATION STEPS: NEEDFOR ADEQUATE CRMs

Metal speciation in biological samples is muchmore complex that in any other common sampleand so the risk of error is significantly higher. Asit is shown in Figure 3.1.1 the specific problemsrelated to speciation in biological samples arisefrom the low concentration of the different speciesin a sample, their stability, the inherent difficultiesof sample treatment due to the matrix complexity,separation and quantification, among others.

9.1 Sources of error

Sources of error that are likely to occur in specia-tion analysis have been extensively reported overthe last few years [68]. The main errors occurringduring sample preparation can be summarised inthe following points:

• Sample extraction: extraction of chemicalspecies from biological samples is a speciallydifficult task and the way to perform it dependson the kind of information needed. For instance,to find out whether a metal is bound to a pro-tein of certain molecular mass or to determinethe oxidation state of an analyte requires a com-pletely different sample treatment.

A good assessment of quality assurance isthe verification of the recovery, which can beapplied by different ways. One of the mostwidely used is the mass balance in respect of

the total atom (amount of analyte found aftercomplete digestion) with respect to the sum ofthe species found, expressed as the amount ofanalyte as shown in the example of Figure 3.1.2.However, this is not always feasible becausefrequently not all the existing species in asample are correctly identified and/or quantifiedin the chromatograms (unknown peaks).

Spiking a sample with known amounts of thespecies of interest, leaving them to equilibrateand determining the species after extractionconstitutes a second way of testing recovery.However, results on recovery based on spikingare not always conclusive, especially in the caseof biological samples because it is very difficultto ensure that the spike is bound in the sameway as the analyte in the sample. For instance,if an animal tissue is spiked with Se methionine,this aminoacid is part of a protein in the sampleand it is obtained as Se-Met in solution after thesample has been hydrolysed.

A quantitative recovery of the spike doesnot necessary reflect a quantitative recoveryof the analyte from the sample. However, ifa quantitative recovery of the spike is notobtained, it implies that the extraction procedureis not adequate since it will certainly not beapplicable to naturally bound compounds.

Recovery must be tested using the standardaddition method. It is recommended to add thespike compounds to previously wetted material,allowing the mixture to stand for long enough(24 h) to reach a good equilibration. Freeze-driedmaterials have to be dried again before analysis.

• Species stability or transformation: is an impor-tant aspect to be considered during the extrac-tion process. Some authors [68] have reportedartificial formation of methyl mercury duringtreatment of samples containing high concen-tration of Hg(II). These findings are still thesubject of controversy, but this effect seems tobe insignificant for biological samples, in whicha large proportion of the organic mercury speciesoccurs in the form of methyl mercury. Any-how, transformation of species must always bechecked carefully. For instance, partial transfor-mation of Sb(III) to Sb(V) takes place when


Source of errors

Pre-concentration

Validation

Use of CRM Very few available

Traceability

Sample extraction

Extraction efficiency

DerivatizationPre-column

Post-column

Sample storage

Stability or transformation of the species

Interfering species

Recovery and

enhancement factor

Figure 3.1.3. Main sources of error during sample pretreatment.

solutions are exposed to air. Therefore, eachanalysis step should be handled with special caredepending on the lability of the target species.This risk of error can be evaluated by spikingthe sample with those species that can suffertransformation during the sample treatment orthe derivatisation process and then analysingthe extracts. This problem has been extensivelystudied for mercury speciation [68]. Only inthose cases in which it is established that speciestransformation does not occur during the extrac-tion procedure can the recovery be calculatedby spiking the extracts. However, the recov-ery assessment can be often overestimated andthis risk should be considered [62]. Taking intoaccount the main drawbacks of sample extrac-tion: low efficiency and species transformation,it is important to find a good compromise situa-tion between an acceptable recovery and preser-vation of species.

• Derivatisation: the errors involved are the deri-vatisation yield (often matrix dependent), whichis difficult to determine due to a lack ofappropriate high purity calibrates; the highnumber of steps before or after derivatisation(preconcentration, clean-up), which increases

the uncertainty of the analysis. As some ofthe reaction mechanisms are still not wellunderstood it is understandable that control of allthese factors is not an easy task and that the riskof errors increases with increasing the numberof steps involved in the analytical process.

• Preconcentration: the main problems encoun-tered in this step are related to the retention ofsome interfering species together with the ana-lytes of interest. A second factor to take intoaccount is the recovery of the species preconcen-trated. Ideally, it should be quantitative to assuregood accuracy. Finally, the sensitivity enhance-ment factor is a third point of relevance duringthis step.

Figure 3.1.3 shows the potential errors that canbe encountered during sample preparation andhandling.

9.2 Relevance of CRMs

The performance of the sample preparation appliedcan be validated in respect of accurate and precisespecies analysis by using extraction solutions

TRENDS AND PERSPECTIVES 91

Weighing of sample intake Species extraction

Calibration of balance? Efficiency losses, recovery? Stability of species

Clean-up

losses

Derivatization

EfficiencyDiscrimination of species

Stability of species

Separation

Standards, interferences?

Results

Figure 3.1.4. Different steps involved in a method of validation for speciation purposes.

of matrix CRMs. A key issue in comparingworldwide analytical results is their traceability,in other words when the results are achieved byan unbroken chain of calibrations connecting themeasurements process to the fundamental units.As has been previously commented, the chainfor speciation analysis in biological samples ismuch more complex than for total determinationsince the number of analytical steps involved ismuch higher.

The use of suitable CRM is very convenientfor validation of the different sample preparationsteps. When new procedures are applied to a CRMand the results are in good agreement with thecertified values (within their uncertainty) it canbe assumed that the analytical results obtained areaccurate: the method is then validated. However,it is well established that the use of CRMsfor method validation requires not only using aCRM of similar sample matrix but also of similarconcentrations to those of species of interest to bedetermined. These facts are extremely importantfor biological materials in which the extractionefficiency of the species is very much dependenton the matrix complexity or on the type ofcompounds in which the species are bound inthe sample.

The main problem in validating an analyticalmethod by using CRMs is the actual lack ofbiological CRMs available in the market forspeciation purposes, even though an importanteffort to improve the state of the art in this fieldis being carried out. This subject is discussed inChapter 7.2 of this book.

It is important to consider that CRMs allow theuser to link results with those of internationallyrecognised standards, enabling the accuracy of theresults to be verified at any desired time. There isan increasing demand for certified materials forchemical species. Nowadays the EU is makinga great effort to offer new biological referencematerials for speciation purposes and it is soonexpected to provide certified oyster tissue for Sn,As and Hg species, rice for As and Se and yeastfor Se-Met among others [69].

Figure 3.1.4 summarises the steps involvedduring an analytical method validation.

10 TRENDS AND PERSPECTIVES

• Improvement in the state of the art for sampletreatment, which still is the most limiting stepof speciation in biological samples.


• Synthesis of pure standards for the differentspecies to identify unknown species. Purecompounds such as arsenosugars or organic anti-mony species (except the trimethyl compound)are not available to the scientific community.

• Improvement of knowledge about species sta-bility in original samples and in the extractswill help in maintaining species integrity dur-ing the different steps involved in speciationmethodology.

• Concerning extraction procedures, the perspec-tives are focused on increasing the extractionefficiency of the species while the time of theprocedure is reduced. The use of microwaveradiation as an extraction technique is one ofthe directions that analytical chemists will haveto address in the future. Speciation methodshave to be simplified, and the different stepsinvolved (matrix disintegration, analyte extrac-tion, derivatisation and sometimes preconcen-tration) can in general be better controlled andenhanced under a microwave field, which willbe a key future issue in speciation analysis [6].The latest trend in microwave design is towardsa miniaturisation of magnetrons. Research isalso focused on reducing solvent consumption.Microwave-assisted procedures will allow oneto perform simultaneous extraction into a sol-vent. Its use will also improve the efficiency ofthe SPME process. This new approach has beenextensively developed and applied to determineorganic compounds such as pesticides and up tonow very few methods for speciation purposeshave been proposed.

• Development of new simple and compact instru-mentation to be implemented in laboratoriesto be used for routine speciation purposes.Instrumentation to be used for derivatisation,preconcentration and species separation havingan easy coupling to different detectors. Someinstruments have been developed based on theuse of capillaries and multicapillaries for coldtrapping in the preconcentration and speciationof mercury.

• Need of appropriate certified biological refer-ence materials, which are less manipulated thanthose currently available and which are certified

for species patterns must be fulfilled in the futureto enable speciation analysis as a scientificallysound analytical activity. Unchanged biologicalmaterial as CRMs and appropriate pure stan-dard compounds are indispensable for the fur-ther development of an accurate sample treat-ment with speciation purposes [69].

11 REFERENCES

1. Moreno, P., Quijano, M. A., Gutierrez, A. M., Perez-Conde, M. C. and Camara, C., J. Anal. At. Spectrom., 16,1044 (2001).

2. Cornelis, R., De Kimpe, J. and Zhang, X., Spectrochim.Acta B , 53, 187 (1998).

3. Szpunar, J., Analyst , 125, 963–988 (2000).4. Munoz-Olivas, R. and Camara, C., In Trace Element Spe-

ciation for the Environment, Food and Health , Ebdon, L.,Cornelis, R., Crews, H., Donard, O., Quevauviller, P. andPitts, L. (Eds) Royal Society of Chemistry, Cambridge,UK, 2002, p. 331.

5. Apostoli, P., Fresenius’ J. Anal. Chem., 363, 499 (1999).6. Sutton, K. L. and Heitkemper, D. T., In Comprehen-

sive Analytical Chemistry , Vol. XXXIII, Elemental Spe-ciation. New Approaches for Trace Element Analysis.Caruso, J. A., Sutton, K. L., Ackley, K. L. (Eds) Else-vier, Amsterdam, 2000, Chapter 14, pp. 501–530.

7. Leopold, I. and Gunther, D., Fresenius’ J. Anal. Chem.,359, 364 (1997).

8. Robb, P., Crews, H. M. and Baxter, M. J., In Induc-tively Coupled Plasma Spectrometry and its Applications ,Hill, S. J. (Ed.) Sheffield Academic Press, Sheffield, UK,1999.

9. Tucher, A., In Mad Cats and Dead Men at Minamata ,Earth Island Publishing, 1972, Chapter 1.

10. Fairweather-Tait, S. J., Fresenius’ J. Anal. Chem., 363,536 (1999).

11. Cornelis, R., Camara, C., Ebdon, L., Pitts, L., Welz, B.,Morabito, R., Donard, O., Crews, H., Larsen, E. H., Nei-dhart, B., Ariese, F., Rosenberg, E., Mathe, D., Morrison,G. M., Cordier, G., Adams, F., Van Doren, P., Mar-shall, J., Stojanik, B., Ekvall, A. and Quevauviller, P.,Fresenius’ J. Anal. Chem., 363, 435 (1999).

12. Massey, R. C. and Taylor, D., In Aluminium in Food andthe Environment , Royal Society of Chemistry, London,1988.

13. Viola, R. E., Morrison, J. F. and Cleland, W. W., Bio-chem., 19, 313 (1980).

14. Cornelis, R., Crews, H., Donard, O. F. X., Ebdon, L. andQuevauviller, Ph., Fresenius’ J. Anal. Chem., 370, 120(2001).

15. Gomez-Ariza, Morales, J. L., Sanchez-Rodas, E. and Gir-aldez, I., Trends Anal. Chem., 19, 200 (2000).

REFERENCES 93

16. Stephen Reid, R. and Attaelmannan, M. A., J. Inorg. Bio-chem., 69, 59 (1998).

17. Marsh, D. O., Myers, G. J. and Clarkson, T. W., Clin.Toxicol., 10, 1311 (1981).

18. World Health Organisation. In Elements in HumanNutrition and Health , WHO, Geneva, 1996.

19. Szpunar-Lobinska, J., Witte, C., Lobinski, R. and Adams,F. C., Fresenius’ J. Anal. Chem., 351, 351 (1995).

20. Lobinski, R. and Szpunar, J., Anal. Chim. Acta, 400, 321(1999).

21. Simonoff, M. and Simonoff, G., In Le Selenium et la Vie,Masson, Paris, 1991.

22. Gu, B. Q., Chin. Chim. Med., 96, 251 (1983).23. Belfroid, A. C., Puperhart, M. and Ariese, F., Mar. Pol-

lut. Bull., 40, 226 (2000).24. Larsen, E. H., Pritzl, G. and Hansen, S. H., J. Anal. At.

Spectrom., 8, 557 (1993).25. Rehberg, P. B., Acta Physio. Scand., 5, 305 (1943).26. Bligh, E. G. and Dyer, W. J., Can. J. Biochem. Physiol.,

37, 911 (1959).27. Rodriguez-Pereiro, I., Schmitt, V. O. and Lobinski, R.,

Anal. Chem., 69, 4799 (1997).28. Quijano, M. A., Gutierrez, A. M., Perez-Conde, M. C.

and Camara, C., Talanta , 50 (1), 165 (1999).29. Potin-Gautier, M., Gilon, N., Astruc, M., De Gregori, I.

and Pinochet, H., INT. j. Environ. Anal. Chem., 67, 15(1997).

30. Tyson, J. F., J. Anal. At. Spectrom., 14, 169 (1999).31. Ebdon, L., Foulkes, M. E., Le Roux, S. and Munoz-

Olivas, R., The Analyst , 127, 1108 (2002).32. Liang, L., Horvat, M., Cernichiari, E., Gelein, B. and

Balogh, S., Talanta , 43, 1883 (1996).33. Hintelmann, H. and Wilken, R. D., Appl. Organomet.

Chem., 8, 533 (1994).34. Abou-Shakra, F. R., Rayman, M. P., Ward, N. I., Hot-

ton, V. and Bastian, G., J. Anal. At. Spectrom., 12, 429(1997).

35. Pannier, F., Astruc, A. and Astruc, M., Anal. Chim. Acta,327, 287 (1996).

36. Ceulemans, M., Witte, C., Lobinski, R. and Adams, F. C.,Appl. Organomet. Chem., 8, 451 (1994).

37. Gilon, N., Astruc, A., Astruc, M. and Potin-Gautier, M.,Appl. Organomet. Chem., 9, 623 (1995).

38. Zhang, Z. and Pawliszyn, J., Anal. Chem., 65, 1843(1993).

39. Zhang, Z., Yang, M. J. and Pawliszyn, J., Anal. Chem.,66, 844A (1994).

40. Moens, L., De Smaele, T., Dams, R., Van Den Broeck, P.and Sandra, P., Anal. Chem., 69, 1604 (1997).

41. De Smaele, T., Moens, L., Sandra, P. and Dams, R.,Mikrochim. Acta, 130, 241 (1999).

42. Kradtap, S., MS Thesis, University of Massachussets,Amherst, MA, 1996.

43. Lee, M. L. and Markides, K. E., Chromatography Con-ferences, Provo, Ut, 1990.

44. Zoorob, G. K., McKiernan, J. W. and Caruso, J. A., Mi-krochim. Acta, 128, 145 (1998).

45. Carey, J. M., Vela, N. P. and Caruso, J. A., J. Anal. At.Spectrom., 7, 1173 (1992).

46. Richter, B. E., Jones, B. A., Ezzell, J. L. and Porter,N. L., Anal. Chem., 68, 1033 (1996).

47. Gallagher, P. A., Wei, X., Shoemaker, J. A., Brocknoff,C. A. and Creed, J. T., J. Anal. At. Spectrom., 14, 1829(1999).

48. Gallagher, P. A., Shoemaker, J. A., Wei, X., Brocknoff,C. A. and Creed, J. T., Fresenius’ J. Anal. Chem., 369,71 (2001).

49. A. Pedersen, G. and Larsen, E. H., Fresenius’ J. Anal.Chem., 358, 591 (1997).

50. Emteborg, H., Bordin, G. and Rodriguez, A. R., Analyst ,123, 245 (1998).

51. Dietz, C., Doctoral Thesis, 2001, University Complutenseof Madrid, p. 72.

52. Lobinski, R., Rodriguez-Pereiro, I., Chassaigne, H.,Wasik, A. and Szpunar, J., J. Anal. At. Spectrom., 13, 859(1998).

53. Tesfalidet, S. and Irgum, K., Anal. Chem., 61, 2079(1989).

54. Tesfalidet, S. and Irgum, K., Fresenius’ J. Anal. Chem.,341, 532 (1991).

55. Puk, R. and Weber, J. H., Anal. Chim. Acta, 292, 175(1994).

56. Tseng, C. M., de Diego, A., Martin, F. M., Amouroux, D.and Donard, O. F. X., J. Anal. At. Spectrom., 12, 743(1997).

57. Rapsomanikis, S., Donard, O. F. X. and Weber, J. H.,Anal. Chem., 58, 35 (1986).

58. Ashby, J., Clark, S. and Craig, P. J., J. Anal. At. Spec-trom., 3, 735 (1988).

59. Lobinski, R., Appl. Spectrosc., 51, 260A (1997).60. Vazquez-Pelaez, M., Montes-Bayon, M., Garcıa-Alonso,

J. I. and Sanz-Medel, A., J. Anal. At. Spectrom., 15, 1217(2000).

61. Rigin, V. I., Fresenius’ Z. Anal. Chem., 335, 15 (1989).62. Quevauviller, Ph., In Method Performance Studies for

Speciation Analysis 1988, The Royal Society of Chem-istry, Cambridge, UK.

63. Skogerboe, R. K., Dick, D. L., Pavlica, D. A. and Lichte,F. E., Anal. Chem., 47, 568 (1975).

64. Ceulemans, M. and Adams, F. C., J. Anal. At. Spectrom.,11, 201 (1996).

65. Dietz, C., Madrid, Y., Camara, C. and Quevauviller, Ph.,Anal. Chem., 72, 4178 (2000).

66. Shickling, C., Yang, J. and Broekaert, J. A., J. Anal. At.Spectrom., 11, 739 (1996).

67. de Pena, Y. P., Gallego, M. and Valcarcel, A., J. Anal.At. Spectrom., 9, 691 (1994).

68. Falter, R., Chemosphere, 39, 1075 (1999).69. Emons, H., Fresenius’ J. Anal. Chem , 370, 115 (2001).70. Blaszkewicz, M., Baumhoer, G. and Neidhart, B., Frese-

nius’ J. Anal. Chem., 325, 129 (1986).71. Gammelgaard, B. and Jons, O., J. Anal. At. Spectrom., 15,

945 (2000).


72. Godlevska-Zylkiewicz, B., Lesniewska, B. and Hulanicki,A., Anal. Chim. Acta, 358, 185 (1998).

73. Suzuki, K. T., Itoh, M. and Ohmichi, M., J. Chromatogr.B , 666, 13 (1995).

74. Nygren, O. and Nilson, C. A., J. Anal. At. Spectrom., 2,805 (1987).

75. Grill, E., Winnacker, E. L. and Zeuk, M. H., MethodsEnzymol., 205, 333 (1991).

76. Kotrebai, M., Birringuer, M., Tyson, J. F., Block, E. andUden, P. C., Analyst , 125, 71 (2000).

77. Casiot, C., Szpunar, J., Lobinski, R. and Potin-Gautier,M., J. Anal. At. Spectrom., 14, 645 (1999).

78. Kumar, U. T., Vela, N. P., Dorsey, J. G. and Caruso,J. A., J. Chromatogr. A, 655, 340 (1993).

79. Munoz, O., Velez, D., Cervera, M. L. and Montoro, R.,J. Anal. At. Spectrom., 14, 1607 (1999).

80. Heumann, K. G., Rottmann, L. and Vogl, J., J. Anal. At.Spectrom., 9, 1351 (1994).

81. Alberti, J., Rubio, R. and Rauret, G., Fresenius’ J. Anal.Chem., 351, 420 (1995).

82. Dagnac, T., Padro, A., Rubio, R. and Rauret, G., Talanta ,48, 763 (1999).

83. Kobayashi, M., Ogua, Y. and Suzuki, K. T., J. Chro-matogr. B , 760, 73 (2001).

84. Michalke, B., Fresenius’ J. Anal. Chem., 351, 670 (1995).85. B’Hymer, C. and Caruso, J. A., J. Anal. At. Spectrom.,

15, 1531 (2000).86. Dundar, M. S. and Haswell, S. J., Analyst , 120, 2085

(1995).

3.2 Sample Preparation Techniques for ElementalSpeciation Studies

Brice Bouyssiere, Joanna Szpunar, Martine Potin-Gautier and Ryszard LobinskiUniversite de Pau et des Pays de l’Adour, Pau, France

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 952 Sample Preparation for Organometallics 97

2.1 Separation of analytes from aqueousmatrices . . . . . . . . . . . . . . . . . . . . . . 992.1.1 Sorption (solid-phase

extraction) . . . . . . . . . . . . . . . 992.1.2 Solid-phase microextraction 992.1.3 Solvent extraction . . . . . . . . . 1012.1.4 Steam distillation . . . . . . . . . . 1022.1.5 Liquid–gas extraction (purge

and trap) . . . . . . . . . . . . . . . . 1022.2 Separation of alkylelement species

from solid samples . . . . . . . . . . . . . . 1032.2.1 Leaching methods . . . . . . . . . 1032.2.2 Solubilization of biological

samples prior to speciationanalysis . . . . . . . . . . . . . . . . . 103

2.2.3 Supercritical fluid extraction 1042.2.4 Microwave-assisted processes 104

2.3 Derivatization techniques for gaschromatography of organometallicspecies . . . . . . . . . . . . . . . . . . . . . . . 1052.3.1 Derivatization by hydride

generation . . . . . . . . . . . . . . . 105

2.3.2 Derivatization withtetraalkyl(aryl)borates . . . . . . 105

2.3.3 Derivatization with Grignardreagents . . . . . . . . . . . . . . . . 106

2.3.4 Other derivatizationtechniques . . . . . . . . . . . . . . . 106

2.4 Preconcentration and cleanup . . . . . . 1072.5 Automation of sample preparation and

GC sample introduction . . . . . . . . . . 1072.6 Organic samples (gas condensates,

shale oils, gasoline) . . . . . . . . . . . . . 1083 Sample Preparation for Organometalloid

Species: Arsenic and Selenium . . . . . . . . . 1093.1 Organoarsenic compounds . . . . . . . . 1093.2 Organoselenium compounds . . . . . . . 110

4 Sample Preparation in Speciation ofCoordination Complexes of Metals withBioligands . . . . . . . . . . . . . . . . . . . . . . . . 1114.1 Biological fluids . . . . . . . . . . . . . . . . 1114.2 Plant and animal tissues . . . . . . . . . . 112

5 Conclusions–Trends and Perspectives . . . . 1136 References . . . . . . . . . . . . . . . . . . . . . . . . 113

1 INTRODUCTION

Hyphenated (coupled, hybrid) techniques basedon the combination of a separation techniquewith atomic spectrometry have become stan-dard in speciation analysis because of their abil-ity to discriminate between the different forms

of an element [1–3]. These techniques usuallyshow an outstanding performance for standardsolutions but often fail when applied to a real-world sample. The successful speciation analysisof environmental and biological samples criticallydepends on the sample preparation, whose objec-tive is to isolate analyte species from the matrix,

96 SAMPLE PREPARATION FOR ELEMENTAL SPECIATION

convert them into chromatographable species andto preconcentrate them to match the detectionlimits of the chromatographic detection technique.Wet-chemical sample preparation methods, almostforgotten in total trace element analysis with theadvent of ICP AES and ICP MS [4], have a uniquerole to play in speciation analysis. Indeed, thelow analyte concentrations involved (often below0.1 µg kg−1 or 0.1 µg L−1), the fragility of the ana-lyte compounds and the need for preservation ofthe organometallic moiety throughout the proce-dure, and the strict requirements posed by thehyphenated system in terms of the sample volume,polarity and matrix acceptable by the instrumen-tal setup make sample preparation in speciationanalysis essential. Another distinct trend is automa-tion which implies the development of faster,possibly single-step, and efficient sample prepara-tion procedures.

The procedure for sample preparation dependson the analytical technique to be used and on

the sample type to be analyzed. The polarity(volatility) of the analyte determines the chromato-graphic technique to be chosen for the separationof the species prior to detection. The separationtechnique, in turn, sets the requirements for theanalyte solution resulting from the sample prepa-ration procedure. Gas chromatographic techniqueswould require a microliter volume of a nonpo-lar solvent free of suspended particles contain-ing volatile thermally stable species. An aqueoussolution of thermodynamically stable and kineti-cally inert species (organometallic or metal com-plexes) free of suspended particles will be requiredfor liquid chromatography-based techniques. Interms of analytes, three major areas can be dis-tinguished: speciation of redox forms, speciationof organometallics (containing a carbon–metal orcarbon–metalloid bond), and speciation of metalcomplexes. Figure 3.2.1 summarizes the principalanalytes that are subject of speciation analysis.Gas chromatography is the preferred separation

Li

H

Na

K

Rb

Cs

Fr

Be

Mg

Ca

Sr

Ba

Ra

Sc

Y

La

Ac

Ti

Zr

Hf

V

Nb

Ta

Cr

Mo

W

Mn

Tc

Re

Fe

Ru

Os

Co

Rh

Ir

Ni

Pb

Pt

Cu

Ag

Au

Zn

Cd

Hg

Ga

In

Tl

Ge

Sn

Pb

As

Sb

Bi

Se

Te

Po

Br

I

At

Kr

Xe

Rn

Al Si P S Cl Ar

B C N O F Ne

He

Redox states :

Se(IV)/Se(VI)

As(III)/As(V)

Cr(III)/Cr(VI)

Fe(II)/Fe(III)

Metallodrug metabolites :

Natural high molecularcomplexes :

metallothioneins,

phytochelatins,

transport proteins,

polysaccharides

Organic acidscomplexes :

tartrate

citrate

nicotianamine

Natural methylatedspecies :

HgMe2, HgMe+

GeMe2

Alkylmetals of anthropogenicorigin :

PbEt4, PbMe4, PbEt4-nn+, PbMe4-n

n+

SnBu3+, SnBu22+, SnBu3

+

SnPh3+, SnPh2

2+, SnPh3+

Natural organometalloid species :

organoarsenicals, arsenosugars,

seleno amino acids

selenoproteins

cisplatin, aurafin and theirprotein complexes, Tc-imaging agents

Figure 3.2.1. Species and fields of interest in speciation analysis.

SAMPLE PREPARATION FOR ORGANOMETALLICS 97

Table 3.2.1. Glossary of the most common steps in sample preparation for speciation analyses.

Filtration is used to separate particles in gases and aqueous samples. Usually a 0.45 µm filter is used but a0.2 µm filter is necessary prior to HPLC.

Solubilization is applied to bring biological materials into solution. It can be achieved by alkaline, acid or enzymatichydrolysis. Following the solubilization, an aqueous solution of species is obtained but the matrix isnot eliminated.

Leaching or solid–liquid extraction is applied to extract analyte species from solid samples (soil, sediment orbiological tissues). Leaching is the most popular method for speciation of metal complexes sinceleaching agents are pH neutral. Leaching is also the preferred methods for the analysis of soil andsediment samples that cannot be solubilized without the destruction of analytes.

Preconcentration is necessary to increase the concentration of the analyzed species in the solution introduced on achromatographic column in comparison with that present in an analyzed sample. The techniques usedinclude cryofocusing, gas solid extraction (trapping) (for analytes already in the gas–phase),solid-phase microextraction (SPME) for analytes in the gas or liquid phase, and liquid–solid extraction(sorption) for analytes in water. Preconcentration can be achieved by solvent extraction andevaporation of the leachate or extract.

Cleanup is the removal of the matrix components (fats, proteins, high boiling point hydrocarbons) that, ifco-introduced on a chromatographic column, would lead to destruction of the column or degradationof its separation properties. Cleanup is usually realized by low resolution chromatographic separationwith a mechanism different from that employed for the analytical separation (e.g. passing through aC18 column of polar species to be separated by anion-exchange)

Derivatization is the process of the controlled conversion of species originally present in a sample into forms withimproved chromatographic yield or separation coefficient. The most popular is derivatization of ionicor highly polar species into nonpolar species that can be readily separated by GC (e.g. Grignardderivatization).

technique for alkylelement species whereas liquidchromatography is predominantly used in all theother cases.

The choice of sample preparation procedure isalso determined by the matrix and the preconcen-tration factor that needs to be achieved to eliminatethe discrepancy between the concentration of theanalyte in the sample and the detection limits ofthe analytical setup. A sample preparation pro-cedure for speciation analysis usually requires anumber of steps (Table 3.2.1). They include fil-tration, preconcentration of analytes, solubilizationor leaching in the case of solid samples, cleanupprior to chromatography, and sometimes derivati-zation of analyte species in order to improve theirchromatographic behavior. It is essential that allthese steps are carried out with a maximum (quan-titative) efficiency and that the original speciesare not degraded. The number of steps necessaryand their duration affects the duration and tedious-ness of an analytical procedure and should be keptminimal. Only some energy-related samples (shaleoil, petroleum, gasoline) containing low polar ana-lytes can sometimes be analyzed directly, usuallyafter dilution.

This chapter discusses sample preparation meth-ods including the recovery of analytes from differ-ent matrices and the preparation of their solutionready to be injected onto a gas or liquid chromato-graphic column. Emphasis is put on proceduresthat require a minimum of time and a small numberof operations having in view the automation of thesample preparation process. Note that before carry-ing out any sample preparation procedure the issueof stability of chemical species in the sample to beanalyzed needs to be critically addressed [5].

2 SAMPLE PREPARATIONFOR ORGANOMETALLICS

Because of their easy conversion to volatileand thermostable species, organometallic speciesare usually determined by hyphenated techniquesbased on gas chromatography (GC). GC requiresan analyte species be presented as a nonpolarthermally stable compound in a nonpolar sol-vent or in a narrow (cryofocused) band. Thesample preparation procedure should thereforeinclude either a step of transfer of analytesfrom the aqueous solution (water sample, leachate


or solubilizate) into an organic solvent (solventextraction), or a step of formation of a narrowanalyte band on a chromatographic sorbent bypurge and trap or solid-phase microextraction.The derivatization step to yield a thermally stablespecies may either follow the solvent extractionstep (typically the case of Grignard derivatization)

or precede it (usually derivatization with NaBH4 orNaBEt4). An additional preconcentration step, bysorption or evaporation, is necessary where ultra-trace amounts of analytes are determined and/orpoorly sensitive detection techniques are used.Flow chart of sample preparation for speciation oforganometallic species is shown in Figure 3.2.2.

Chelation

Hg ,P

b ,Sn

Hg, P

b ,Sn

Hg

,Sn As, G

e, Sn

As, G

e, Sn

SnH

g, Pb, S

n

Non-polarorganometallic

compounds

Ionicorganometallic

compounds

Analyte

Complexing

Separation

Derivatization

Preconcentration

DetectionGC - element selective detection

CryotrappingIn-liner

enrichmentInert gaspurging

Grignardderivatization

Hydride generation

Alkylationwith

alkylborates

Liquid-liquidextraction

Solid-phaseextraction

SPME Purging withinert gas

Figure 3.2.2. Flow chart of the sample preparation procedure for speciation of organometallic species.


2.1 Separation of analytesfrom aqueous matrices

The choice of procedure depends on the matrix.A number of possibilities applicable to aqueoussamples include solid-phase extraction (sorption),solid-phase microextraction (SPME), liquid–gasextraction (purge and trap) and classical solvent(liquid–liquid) extraction. The separation of ana-lytes from water is usually combined with theirpreconcentration, and often with their derivatiza-tion. The procedures are classified according to theorder of extraction and derivatization steps. Whenanalytes are present at the ng L−1 level in relativelyclean water samples, in situ derivatization followedby an extraction technique with a high preconcen-tration factor (purge and trap, SPE or SPME) arepreferred. In the case of dirty matrices with higherconcentrations of analytes, the latter are usuallyextracted as nonpolar complexes (with DDTC,dithizone or tropolone) which are later derivatized,usually by means of a Grignard reaction.

2.1.1 Sorption (solid-phase extraction)

Solid-phase extraction (SPE), which is becom-ing increasingly popular for sample preparationin organic analysis, has found application in spe-ciation analysis for organometallics [6–11]. Theanalytes are extracted by sorption, eluted with asmall amount of an organic solvent and deriva-tized. Its advantages over liquid–liquid extractioninclude a higher enrichment factor, lower solventconsumption and risk of contamination, and theease of application to field sampling and automa-tion. Solid-phase extraction is particularly suitablefor the analyte preconcentration prior to HPLC; byusing an appropriate eluent the analytes can be sep-arated on the SPE cartridge itself or on a connectedHPLC column [12, 13].

Only filtered samples can be analyzed, whichmay constitute a considerable drawback. SPEcartridges in a variety of configurations and sizeshave been used but microcolumns proved to be thebest choice with respect to low dead volumes andlow consumption of the solvent [6]. C18 extractiondisks, reported to be well suited for the trace

enrichment of organics in environmental waters,did not show satisfactory retention of native di-and monobutyl(-phenyl) tin species [6].

Sorbents with immobilized chelating reagents,such as dithiocarbamate [7, 8], dithizone [12]or diphenylthiocarbazone [14], have widely beenused. A sampling and storage technique for spe-ciation analysis of lead and mercury in seawaterbased on the sorption on a column packed withdithiocarbamate resin has been proposed [11]. C18

minicolumns modified with sodium diethyldithio-carbamate have been used for the field samplingand preconcentration of mercury species in riverwater [13].

An interesting curiosity is the use of biomate-rials, e.g. dried yeast (Saccharomyces cerevisiae)cells, for enrichment of methylmercury from riverwater [9]. Another is the analytical potential offullerene, which was investigated as adsorbent fororganic and organometallic compounds from aque-ous solutions [15].

2.1.2 Solid-phase microextraction

Solid-phase microextraction (SPME) is a precon-centration technique based on the sorption of ana-lytes present in a liquid phase or, more often, in aheadspace gaseous phase, on a microfibre coatedwith a chromatographic sorbent [16, 17]. The fiberis incorporated in a microsyringe. The coating isexposed to the sample in order to extract the ana-lytes onto the coating. The fiber may be immersedin the solution or be placed in the headspace abovethe sample (Figure 3.2.3(a)). Once equilibrium isreached, the fiber is withdrawn into the needle forprotection against contamination (Figure 3.2.3(b)).It is then transferred either to a GC injector wherethe analytes are thermally desorbed or to an HPLCinterface where they are solubilized in the mobilephase used as solvent (Figure 3.2.3(c)).

The absorptive-type polydimethylsiloxane(PDMS) coatings are the most popular in speciationanalysis. This nonpolar phase extracts organometal-lic species such as methyl- and ethylmercurycompounds [18–20], alkylead species [21–23],butyl- and phenyltins [24, 25] and methylarseniccompounds [26, 27]. It is commercially available in


SPMEfiber holder

Needle guide

SPMEretractable fiber

a) Extraction b) Transfer c) Desorption

directextraction

headspace

extractionwith

membrane

GC injectoror

HPLC interface

Figure 3.2.3. Principle of solid-phase microextraction.

different thickness and so it can be applied to com-pounds with different volatilities and distributionconstants between the coated and liquid phases.PDMS-coated fibers are very rugged and can with-stand a temperature of GC injector of up to 300 ◦C.An adsorptive carboxen coating seems to be moreadapted to the analysis of volatile species such ashydrides of metals and metalloids [28].

SPME is a method that is emerging as animportant analytical tool for elemental speciationin environmental samples. This solvent-free tech-nique offers numerous advantages such as simplic-ity, the use of a small amount of liquid phase,low cost and the possibility of an on-line analyti-cal procedure. Interference problems induced bythe complexity of environmental matrices oftennoted in liquid–liquid extraction are limited byusing SPME.

A method based on SPME on a fiber coated(100 µm) with poly(dimethylsiloxane) was devel-oped for the determination of tetraethyllead inwater [22]. Speciation of alkyllead and inorganic

lead by derivatization with deuterium-labelledsodium tetraethylborate and SPME-GC/MS hasbeen studied [23]. A water sample has alsobeen subjected to headspace SPME of ethy-lated organometallic species onto PDMS-coatedfused silica fibers for sensitive simultaneousdetermination of organomercury, -lead and -tincompounds [21]. Methylmercury in river watersamples was derivatized in situ using NaBEt4, anda silica fiber coated with poly(dimethyl siloxane)was placed into the solution. Once equilibriumwas reached, the fiber was thermally desorbedinto a GC injector [29]. Methyltin chlorides andtetramethyltin were adsorbed from acetone solu-tion onto methylsilicone-coated fibres and trans-ferred to a headspace vial for equilibration withNaBH4 solution [30]. A number of applicationsconcerning organoarsenic species have been pub-lished [26, 27].

The high preconcentration factors, often com-bined with a derivatization step offered by thisnew tool, facilitate the chromatographic analysis


of species at trace and ultratrace levels as is oftenrequired in speciation analysis. Very low detec-tion limits can be reached. For example, quantifi-cation of butyl- and phenyltins at concentrationslower than 0.1 ng Sn L−1 is now possible by com-bined SPME GC-pulsed FPD or SPME GC-ICP-MS [24, 25, 31].

Applications of SPME to studies of elemen-tal speciation in different environmental andbiological matrices have been comprehensivelyreviewed by Mester et al. [32]. Selected appli-cations of SPME to analysis of organometallicspecies in environmental matrices are presented inTable 3.2.2.

2.1.3 Solvent extraction

Solvent (liquid–liquid) extraction can be directlyapplied to nonfiltered samples with complex matri-ces and allows the direct transfer of analytesinto a nonpolar organic solvent that can be ana-lyzed by GC. Nonpolar species, e.g. tetraalkyllead,are quantitatively extracted from water saturatedwith NaCl into a volume of hexane 20 timessmaller [43]. The same applies to some ‘ionicspecies’ with the marked covalent character (e.g.triphenyltin, tributyltin, triethyllead, methylmer-cury) but the use of water-immiscible solventswith polar character (toluene, ethylacetate) wasadvised. The properties of organometallic species

Table 3.2.2. Applications of SPME to environmental organometallic compounds analyses.

Compound Matrix Modea Fiber type Analytical techniqueb Reference

Tin compoundsMethyl- Seawater H PDMS, 100 µm Et-GC-FPD [30]Butyl- Sediment H PDMS, 100 µm Et-GC-ICP-MS [21]Butyl- Urine H PDMS, 100 µm Et-GC-MS-MS [33]Butyl- Sediment, sludge D PDMS-7, 100 µm Et-GC-AES [34]Butyl- Fresh, sea and

waste watersH Silica fiber Hy-GC-QSIL-FPD [35]

Pretreated by HFButyl- Sediment H PDMS, 100 µm Et-GC-(HC)-GD-

OES[42]

Phenyl-, hexyl- Mussel, potato H PDMS, 100 µm Et-GC-ICP-MS [31]

Lead compoundsMethyl- Sediment H PDMS, 100 µm Et-GC-ICP-MS [21]Methyl- Urine H PDMS, 100 µm Et-GC-MS-MS [33]Methyl-, ethyl- Sediment H PDMS, 100 µm Et-GC-(HC)-AES [42]Ethyl-, Pb2+ Blood, urine H PDMS, 100 µm Et-GC-FID [36]

PDMS/DVBEthyl- Waters, gasoline H PDMS, 100 µm nd-TD-QF-AAS [37]

Mercury compoundsMethyl- Sediment H PDMS, 100 µm Et-GC-ICP-MS [21]Methyl-, Hg2+ Urine H PDMS, 100 µm Et-GC-MS-MS [33]Methyl-, ethyl-, phenyl- Soil D PDMS, 100 µm nd-GC-MIP-AES [20]Methyl-, Hg2+ Waters, fish H,D PDMS, 100 µm Et-GC-MS, GC-FAS [19]Methyl- Gas condensate H PDMS, 100 µm nd-GC-MIP-AES [18]Methyl- Soil H PDMS, 100 µm Et-GC-MS [38]Methyl- Fish H, D PDMS/DVB,

65 µmnd-TD/ICP/MS [28]

Selenium compoundsSe(IV) Drinking water H, D PDMS, 100 µm Et-GC-MS [39]

Arsenic compoundsMethyl-, Phenyl- Water, soil D PA, 65 µm d-GC-MS [40]Methyl- Urine D PDMS, 100 µm d-GC-MS [26]Methyl-, organo- Drinking water In-tube SPME nd-HPLC-ESI-MS [41]

aH, headspace; D, immersion.bEt, ethylation; Hy, hydridation; nd, non derivatized; d, derivatization.


with a smaller number of organic substituents(e.g. monobutyltin) do not allow their quantita-tive extraction by any organic solvent. In order totransfer such compounds into an organic phase, theformation of extractable chelate complexes or non-polar covalent compounds in the aqueous solutionprior to extraction is necessary.

Diethyldithiocarbamate (DDTC) is the mostoften used reagent to form the chelate complexeswith organometallic species. The extraction isfollowed by a derivatization step, usually by aGrignard reaction. Extraction of the complexesof ionic organolead species with DDTC frompH 6–9 into hexane [44–49], organotin [50–52]and organomercury [53, 54] was found to givequantitative recovery. Dithiocarbamates are not aslight sensitive as dithizonates which makes thehandling easier and the procedure more reliable.The high selectivity of the hexane–tetramethyldi-thiocarbamate extraction system for ionic alkylleadsover Pb2+ facilitates greatly the determinationof these analytes in matrices containing highlevels of Pb2+ [55]. Inorganic interferents canbe efficiently masked with EDTA [45–47]. Fororganotin, tropolone has extensively been used [52,56–60] in addition to DDTC.

An alternative to chelate extraction is theformation of extractable nonpolar species in theaqueous phase by hydride generation [61, 62], ethy-lation with NaBEt4 [6, 63–66], propylation withNaBPr4 [67], butylation with NH4BBu4 [48, 49],or arylation with NaBPh4 [65, 68, 69]. The resul-tant species should have a boiling point exceedingby 20 ◦C that of the solvent to be used for theirextraction in order to enable the subsequent gaschromatographic separation. Also, it is importantthat the nucleophilic substituent be different fromorganic substituents of the species to be analyzed.For example, ethylation will fail for Et2Pb2+ andEt3Pb+ since it would lead to the identical species(Et4Pb) that is also the product of ethylation of Pb2+.

Preconcentration factors in solvent extractionare generally low (typically 1 : 50 up to 1 : 250).The methods are rapid, work well for less volatileanalytes and are relatively robust in terms ofcoping with interferences.

2.1.4 Steam distillation

Steam distillation has been evaluated as a tech-nique for the separation of methylmercury fromnatural water samples [70–73]. It was found togive higher and more reproducible recoveries thanother extraction techniques but under some condi-tions it could be responsible for artifactual methy-lation of Hg2+ [71]. The technique is rather slow;the addition of ammonium pyrrolidine dithiocarba-mate (APDC) was found to improve recovery (upto 85 % for seawater) and to eliminate the codis-tillation of inorganic mercury. The method wasfound to be free of artifacts and to be compara-ble to nitrogen-assisted distillation with the addedadvantage of increased samples throughput [72].

2.1.5 Liquid–gas extraction (purge and trap)

This technique consists of bubbling an aqueoussolution with an inert gas (nitrogen or helium)to extract the nonpolar volatile species into thegas phase. Some species can be extracted directly(e.g. tetraalkyllead, tetramethyltin, dimethylmer-cury) but others need to be converted into volatilehydrides or ethyl derivatives. Purge and trap meth-ods usually follow hydride generation, which isthe oldest but still the most popular method forvolatilization of As, Ge, Sn, Sb and of their methylspecies [74, 75]. Purging of less volatile tributyl-or triphenyltin hydrides is negatively affected bycondensation problems which may lead to non-quantitative recoveries of the analytes. In general,hydride generation is prone to matrix interfer-ences and hydrides are relatively reactive andtend to decompose when subject to harsh instru-mental conditions. A rapidly developing alterna-tive for the production of volatile derivatives isethylation using NaBEt4. It was developed forionic mercury [76–78], lead [77, 79], tin [77] andselenium [80] species and enjoys a continuouslyincreasing number of applications. Ethyl deriva-tives of butyl- and phenyltins are not sufficientlyvolatile to be efficiently purged. Furthermore, thistechnique is not selective for ethylmetal species.

The purged species are subject to the removalof moisture as in the case of analyses of gases,


are cryotrapped and released onto a packed orcapillary column. Capillary traps are becomingincreasingly popular because of the narrow bandof the analyte, compact size, and compatibilitywith capillary and multicapillary GC columns [77,81–83]. This method offers high preconcentrationfactors and allows the introduction of fairly cleansamples onto a GC column. Interferences, espe-cially with hydride generation, and the risk of con-densation with less volatile analytes are the majorshortcomings.

2.2 Separation of alkylelement speciesfrom solid samples

Solid samples of interest for speciation analysis oforganometallic species can be divided in two majorcategories: sediment and soils samples and bio-logical materials. Organometallic compounds areapparently not involved in mineralogical processesin sediment and soil and bind onto the surface.Therefore, the complete dissolution of the sam-ple prior to analysis is not considered necessary.In contrast, these compounds may be incorporatedin tissues of a living organism. Hence, solubiliza-tion of a biological material prior to separationof the analytes is mandatory even if in particularcases some success can be obtained by leaching.Once the analyte species are brought in solution,the latter is treated similarly to water samples asdescribed above.

2.2.1 Leaching methods

The basic approach to releasing organometalliccompounds from the sediment involves acid leach-ing (HCl, HBr, acetic acid) into an aqueous ormethanolic medium by sonication, stirring, shakingor Soxhlet extraction with an organic solvent [50,84–91]. In order to increase the extraction yieldthe addition of a complexing agent (tropolone,DDTC) is mandatory. Extraction recoveries oforganometallic compounds in environmental matri-ces have been critically discussed by Quevauvillerand Morabito [92].

A modification of the leaching procedures isdistillation of methylmercury. A sediment, soil or

biological tissue sample is suspended in an acidicsolution and the mixture is distilled at elevatedtemperature (ca. 180 ◦C) with nitrogen [93–97].The distillate is collected in an ice-cold container.A complexation reagent, e.g. pyrrolidinedithiocar-bamate, may be added to improve the extrac-tion recovery. Artifact formation of methylmercuryduring procedures for its extraction from environ-mental samples has been studied [98].

A number of leaching methods have been devel-oped for the recovery of butyl- and phenyltincompounds from biological tissues. The extractionefficiencies of 12 of these methods have been crit-ically compared [99]. Acidic conditions, togetherwith the use of tropolone and a polar organicsolvent were found to enhance the extractionefficiency, especially for mono- and disubsti-tuted compounds.

For soil and sediments, the need for a cumber-some sample preparation step is the basic weak-ness of the whole analytical procedure. Indeed,the majority of procedures reported have notonly been extremely time-consuming, taking from1 h to 2 days, but have also usually been ineffi-cient in terms of analyte recovery and, in gen-eral, unreliable. As shown by Chau et al. [100],only three out of ten sample preparation meth-ods described in the literature for the analysis ofsediments were able to recover more than 90 %of Bu3Sn+, whereas none of them was able torecover monobutyltin (BuSn3+) in a nonerraticand reproducible manner. In general, the morepolar the species to be extracted the lower isits recovery, the longer the leaching procedurenecessary and the higher the demand for accel-erated (supercritical fluid extraction, acceleratedsolvent extraction or microwave-assisted leach-ing) are going to be. The classical, supercriti-cal fluid and microwave-assisted techniques havebeen compared for the extraction of methylmer-cury from aquatic sediments [101].

2.2.2 Solubilization of biological samples priorto speciation analysis

A suitable digestion (hydrolysis) procedureshould allow for the complete destruction of


the matrix while the organometallic moietyremains unchanged. Three principal approacheshave been developed: (1) acid hydrolysis withHCl [102], (2) alkaline hydrolysis with methanolicNaOH [103, 104] or with tetramethylammoniumhydroxide (TMAH) [105, 106], and (3) enzymatichydrolysis [107–109].

The advantage of the total solubilization methodis the 100 % transfer of the species of interest intoan aqueous phase that is later subject to extractionand derivatization procedures. The metal–carbonbond seems to resist the degradation even inrelatively harsh conditions. However, the resultingsolutions are rich in dissolved organic matter andin salts formed during the neutralization of thesolution which may negatively affect extractionand derivatization efficiencies.

2.2.3 Supercritical fluid extraction

Substantial progress towards faster and potentiallyautomated speciation analysis of sediments wasoffered by supercritical fluid extraction (SFE) [89,110–115]. Equipment cost, however, is high, theextraction step still takes 10–50 min and therecoveries of many species are far from beingquantitative.

Analytical strategies involving SFE fromenvironmental matrices in tin, mercury, lead andarsenic speciation studies have been criticallydiscussed [116]. The method was found to besuccessful for all the analytes if derivatizationwas performed on the aqueous phase beforeextraction. Carbon dioxide modified with aceticacid was found to be the most suitable for theextraction of organotin compounds from sedimentand biota [116].

2.2.4 Microwave-assisted processes

Microwaves are high frequency (2.45 GHz) elec-tromagnetic waves which are strongly absorbedby polar molecules (e.g. water or mineral acids)and which interact only weakly with nonpolar sol-vents. Absorption results in dielectric heating; theheat appears in the core of the target sample.The well-known advantages of microwave heating

such as absence of inertia, rapidity of heating,efficiency and ease of automation have made itwidely used for accelerated extraction of polarcompounds into a nonpolar or weakly polar sol-vent. Many laboratory microwave ovens are com-mercially available.

The preservation of the organometallic moiety isthe prerequisite of a successful leaching/digestionprocedure prior to speciation analysis. It can beachieved by a careful optimization of the condi-tions of the microwave attack [117]. In contrastto common high temperature and pressure acidattack procedures, a focused low power microwavefield is preferred for extraction of organometallicspecies from the matrix. The carbon–metal bondsremain intact.

Microwave-assisted leaching has been shownnot only to reduce the time necessary for leachingof organometallic compounds from soil and sedi-ment samples but also to increase the recovery ofcompounds that are the most difficult to extract.The time necessary for the quantitative extrac-tion of organotin [64, 118–122], organolead [123]and methylmercury [78, 124] was reduced to2.5–3 min. The values obtained for the extractionrecoveries of monobutyltin from CRMs and can-didate CRMs by these techniques are among thehighest reported in the literature.

The microwave field can also accelerate leach-ing of analyte compounds from biomaterials [122,125]. When a suitable reagent is used (e.g. tetra-methylammonium hydroxide) the biological tis-sue can be solubilized within 2–3 min insteadof 1–4 h [64, 120, 124]. An even more attrac-tive alternative is the integration of the solubi-lization, derivatization and extraction steps into aone-step procedure. Hydrolysis with acetic acidcarried out in a low-power focused microwavefield in the presence of NaBEt4 and nonane hasbeen shown to shorten the sample preparationtime for the CGC-MIP AED determination oforganotin compounds in biological materials to3 min [126]. The issue of one-step microwave-assisted extraction–derivatization procedures hasbeen approached chemometrically for methylmer-cury in biological tissues [127].


2.3 Derivatization techniques for gaschromatography of organometallic species

A number of native organometallic compounds arevolatile enough to be separated by GC. They includetetraalkyllead species (MenEt4−nPb), methylsele-nium compounds (e.g. Me2Se, Me2Se2), someorganomercury compounds (MeHg+, Me2Hg) aswell as naturally occurring metalloporphyrins. Asindicated above they can either be readily purgedwith an inert gas or extracted into a nonpolar sol-vent and subsequently chromatographed by thermaldesorption, packed column or capillary GC.

The majority of organometallic species exist inquasi-ionic polar forms which have relatively highboiling points and often poor thermal stability. Tobe amenable to GC separation they must be con-verted to nonpolar, volatile and thermally stablespecies. The derivative chosen needs to retain thestructure of the element–carbon bonds to ensurethat the identity of the original moiety remainsconserved. The most common derivatization meth-ods include: (1) conversion of inorganic and smallorganometallic ions into volatile covalent com-pounds (hydrides, fully ethylated species) in aque-ous media; (2) conversion of larger alkylmetalcations: e.g. RnPb(4−n)+ with Grignard reagentsto saturated nonpolar species, and (3) conversionof ionic species to fairly volatile chelates (e.g.dithiocarbamate, trifluoroacetone) or other com-pounds. The three methods are fairly versatilein terms of organometallic species to be deriva-tized and the choice depends on the concen-tration of interest, the matrix and the samplethroughput required. Frequently, the derivativesare concentrated by cryotrapping or extractioninto an organic solvent prior to injection onto aGC column.

Derivatization (chemical modification) tech-niques for GC, HPLC and CZE in speciation anal-ysis have been reviewed [128].

2.3.1 Derivatization by hydride generation

Several elements (Hg, Ge, Sn, Pb, Se, Te, Sb,As, Bi, and Cd) can be transformed into volatilehydrides, forming the basis of their determination

[74, 75]. The usefulness of this procedure forspeciation analysis, however, is severely restrictedeither by the thermodynamic inability of somespecies to form hydrides, or by considerable kineticlimitations to hydride formation. Nevertheless,the technique is still essential for some classesof compounds. The chemical reaction of hydridegeneration, presented on the example of MeHg+determination is the following:

MeHg+ + NaBH4 −−−→ MeHgH + NaBH3+

Selenium, As, and Sb readily form hydridesonly in their lower oxidation states; the higherstates need to be reduced beforehand. Thus, allinorganic species of these elements form eventu-ally the same hydride (SeH2, AsH3, and SbH3,respectively) precluding simultaneous chromato-graphic speciation. Methyl- and dimethylarsonic(or stibonic) acid can be discriminated in one GCrun upon hydride generation producing volatileMeAsH2 and Me2AsH (or MeSbH2 and Me2SbH,respectively). Trialkyllead species form stablehydrides whereas dialkylleads are nonreactive.Mercury(II) and methylmercury, as well as germa-nium and methylgermanium species [74] can betransformed to gas chromatographable hydrides.Hydride generation has enjoyed the largest inter-est for organotin speciation analysis because of itscapability for the simultaneous determination ofionic methyl and butyl species in one chromato-graphic run [84, 108]. Hydride generation withNaBH4 is prone to interferences with transitionmetals which affect the reaction rate and analyticalprecision [75].

2.3.2 Derivatization with tetraalkyl(aryl)borates

The vulnerability of hydride generation to interfer-ences in real samples and the restricted versatilitycan, to a certain degree, be overcome by replac-ing NaBH4 by alkylborates. The most commonderivatization procedures rely on ethylation withsodium tetraethylborate (NaBEt4) which is watersoluble and fairly stable in aqueous media. Theuse of NaBEt4 in speciation analysis has beenreviewed [63].


The reaction of derivatization with tetraethyl-borate, presented for the example of MeHg+ isthe following:

MeHg+ + NaBEt4 −−−→ MeHgEt + NaBEt3+

Methyl-, butyl- and phenyltin compounds reactreadily with NaBEt4 to form thermally stablespecies that can be analyzed by GC. Whereasmethyltins can be purged upon derivatization, otherspecies need to be extracted because of their poorervolatility. All alkyllead species also readily reactbut only methyllead species can be unambiguouslydiscriminated as the derivatization of ethyl- andinorganic lead will lead to the formation ofthe same product, namely PbEt4. Methylmercuryand inorganic mercury can be determined inone run upon purge-and-trap preconcentration.Derivatization of various selenium species hasbeen demonstrated but showed poor potential forsimultaneous analysis. Nevertheless, Se(IV) can bedetermined selectively and free of interferences byits reaction with NaBEt4 [80].

In situ phenylation using sodium tetraphenylborate has been studied with some applications[65]. Other reagents, such as sodium tetra-propylborate for organotin or tetramethylammo-nium tetrabutylborate for organolead [48, 49].Some efforts concerning multielement and mul-tispecies derivatization using NaBEt4 have beenreported [77].

2.3.3 Derivatization with Grignard reagents

As mentioned above, hydride generation or ethy-lation with alkylborates fail for some species orin the cases of very complex matrices. An alter-native is derivatization with Grignard reagents,which is fairly versatile but requires an aqueous-free medium for the reaction to be carried out.In practice, it is applicable to extracts contain-ing complexes of an organometallic compoundwith dithizone, dithiocarbamates or tropolone.Ionic organometallic compounds are transformedto nonpolar species according to the equation (pre-sented on the example of MeHg+).

Grignard derivation of MeHg+:

MeHg+ + RMgX −−−→ MeHgR + Mg2+ + X−

R can be methyl, ethyl, propyl, buthyl or phenyland X can be Cl, Br or F.

Whereas Grignard derivatization still remainsthe primary method for lead speciation analy-sis [44–46, 129], its position has been gradu-ally eroded for organotin speciation in favor ofthe less cumbersome and time-consuming deriva-tization with NaBEt4 [50]. Other applications ofGrignard reagents, e.g. for the derivatization oforganomercury, have been limited to one researchgroup [53, 54]. An interesting curiosity is thedirect Grignard pentylation, demonstrated recentlyfor an organotin-contaminated lard sample [130].

Grignard reagents proposed for derivatizationin speciation analysis by GC with plasma sourcespectrometric detection have included methyl-,ethyl-, propyl-, butyl- and pentylmagnesium chlo-rides or bromides. Lower-alkyl magnesium saltsare generally preferred due to the smaller molec-ular mass and, hence, the higher volatility ofthe resulting species, which makes the GC sep-aration faster with less column carryover prob-lems associated with derivatized inorganic forms(which are often present in large excess). In addi-tion, the baseline is more stable and less Grig-nard reagent-related artefacts occur. Conversely,the low volatility of species derivatized by pentyl-magnesium chloride may facilitate concentrationby evaporation and, hence, a more efficient enrich-ment is achieved.

The unreacted Grignard reagent needs to bedestroyed prior to the injection of the derivatizedextract onto a column, which is achieved byshaking the organic phase with dilute H2SO4. Asa final step of the procedure, the organic phase isdried over, for example, anhydrous Na2SO4, andinjected onto the GC column.

2.3.4 Other derivatization techniques

The formation of volatile acetonates, trifluoroacet-onates and dithiocarbamates is a popular derivati-zation technique for inorganic GC [131]. Kineticrestrictions or the thermodynamic inability ofmany species to react, and the small differencesin retention times for the derivatized speciesof the same element make chelating agents of


limited importance as derivatization reagents forspeciation analysis.

Selenoaminoacids have been derivatized withisopropylchloroformate and bis(p-methoxyphenyl)selenoxide [132], with pyridine and ethyl chloro-formate [133] or silylated with bis(trimethylsilyl)-acetamide [134]. Selenomethionine forms volatilemethyl-seleno-cyanide with CNBr [135, 136]. Se-lective determination of selenium(IV) and sele-nium(VI) has been achieved using GC with flamephotometric detection (GC-FPD) and GC-MS afterderivatization of selenium(IV) with 4,5-dichloro-1,2-phenylenediamine [137].

Methods for the conversion of arsenic com-pounds to volatile and stable derivatives are basedon the reaction of monomethylarsonic acid anddimethylarsenic acid with thioglycolic acid methylester (TGM) [138, 139].

2.4 Preconcentration and cleanup

Some extraction techniques, such as sorption(solid-phase extraction) or purge-and-trap, offera high intrinsic preconcentration factor. Solventextraction methods have a common disadvantageof yielding a large volume of extract (usually about1–5 mL). It oversizes considerably the amountwhich can be introduced onto the capillary col-umn. The discrepancy between the concentrationof an analyte species in a sample and the detec-tor’s sensitivity is often increased by a dilutionfactor during the analysis of solid samples. There-fore, an additional preconcentration step (e.g. byevaporation) sometimes needs to be carried out onthe leachate or extract containing the analytes.

Purging the extracts with nitrogen or helium inprecalibrated tubes, Kuderna–Danish evaporationand rotary evaporation have been the methods ofchoice. Losses may occur in the preconcentrationof the derivatized species, especially for morevolatile Me3Pb+ species. Better recoveries areobtained when the solution of the organometallicchelates, which are less volatile than tetraalkylspecies, is preconcentrated prior to derivatization.Then, however, a minimum volume of 250 µL isrequired for easy handling during the derivatizationstep and removal of the unreacted Grignard

reagent, inducing a dilution factor of 1 : 250 forcapillary GC analysis.

The increasing sensitivity of element selectivedetectors for GC (sub-picogram absolute detectionlimits have become a standard) and the wideravailability of large volume injection techniqueshave recently contributed to the elimination of theneed for off-line enrichment.

The evaporative preconcentration and largevolume injection lead to a co-preconcentration ofother substances present in a sample and maybe detrimental to the chromatographic column inroutine analysis. Therefore a cleanup step of theextract is advised.

2.5 Automation of sample preparationand GC sample introduction

Recent advances in in situ derivatization, capillarycryotrapping and multicapillary GC have allowedthe development of an automated sample introduc-tion device for an atomic spectrometer for speci-ation analysis. The scheme of such an accessoryfor time-resolved introduction of element speciesinto an atomization/excitation/ionization sourceis shown in Figure 3.2.4. Organometallic com-pounds (alkyllead, butyltin, methylmercury) andsome ions, e.g. Pb2+, Hg2+, Se(IV), As(III), arevolatilized in situ by means of a suitable deriva-tization reaction (hydride generation or ethylationwith NaBEt4). The derivatives formed are purgedfrom the vessel and pass through a water scrubber(Nafion dryer) to a wide bore (0.53 mm) capil-lary trap where they are cryofocused at −100 ◦C.Then the trap is heated to release the species on amulticapillary capillary column. The analysis cycletakes less then 5 min. A number of applicationshave recently been described [81, 82, 140, 141]. Asimilar design with AAS detection has been suc-cessfully used by Dietz et al. [83].

The system developed shows two major advan-tages over the commercial purge-and-trap systemsand those described in the literature. It is a free-standing accessory including the separation step sono gas chromatograph is required because of theapplication of a multicapillary column. The secondone consists of using a 30 cm Nafion tube dryer


Cooling gas (N2)

3-way valve

1/16" tube

Cut-off valve

Splitter

Inside coil made of copperDewar with liquid nitrogen

tube Ø4 i.d.

1/4" tube

T-pipe

Oventemperature

control system

Trap temperaturecontrol system

Valvetemperature

control system

Purge and dryinggas inlet

Drying gas outlet

Pur

ge g

as

Detector

Waste

Purge vesselwith sample

Cooling chamber

Carrier gas inlet

Drying gas

Nafion drierSample inlet

Figure 3.2.4. Device for the time-resolved introduction of organometallic species into an atomic spectrometer.

which makes an external chiller redundant. Theresult is a compact accessory allowing for species-selective analysis of liquid and solid (with optionalmicrowave cavity) samples.

2.6 Organic samples (gas condensates,shale oils, gasoline)

This category includes various samples of whichthe most often analyzed are (1) fuels, for alkyl-lead [142, 143] and manganese carbonyl [144,145] compounds, (2) crude oils, for metallopor-phyrins [146–148], and natural gas condensatesfor mercury species [18, 149–151]. The analytesare volatile and thermally stable so that samplescan usually be injected directly onto a GC col-umn. Gasoline is often diluted ten times to avoidinterference with hydrocarbons and the resultingsaturation of the plasma.

The detection limits for metals in oils and gascondensates are high in comparison with thosefor other samples. This is due to the hydrocarbonmatrix which gives rise to background interferencewhen it enters into the plasma at the same

time as a species of interest. Moreover, carboncompounds can overload the plasma discharge,which has limited thermal energy, and hencereduce the excitation ability. The determinationof metalloporphyrins in crude oils is furthercomplicated by their low concentration levels,wide range of molecular weights, the complicatedisomerism and the complexity of the crude oilmatrix [150, 151]. Hence, a sample pretreatmentstep is often necessary.

In order to counteract the degradation ofGC performance due to the accumulation ofnonvolatile residues from the crude oil, removalof the asphaltene and pigment materials has beenrecommended. This also enables reduction of themaximum elution temperature for a sample, thusprotecting the column, reducing the bleed andstabilizing the detector [146]. Other benefits ofa sample pretreatment include group isolation ofmetalloporphyrins and nonporphyrins for GC andpreconcentration of metalloporphyrins for traceanalysis [146].

An elegant solution to avoid the interferencefrom hydrocarbons was proposed for speciation

SAMPLE PREPARATION FOR ORGANOMETALLOID SPECIES 109

analysis of mercury in gas condensates [18].Mercury-containing peaks are collected from thecolumn eluate by an amalgamation trap. Theyare subsequently released into the plasma as Hg0

in a flow of helium. Methylmercury and labileionic mercury can be extracted as complexes withcysteine into the aqueous phase to eliminate thematrix and then re-extracted into hexane prior toGC [149]. A direct approach to speciation analysisof mercury in gas condensates was recentlyproposed [150, 151].

3 SAMPLE PREPARATION FORORGANOMETALLOID SPECIES:ARSENIC AND SELENIUM

Arsenic and selenium are two metalloids witha very rich biochemistry. As a consequence, anumber of species containing As–C and Se–Cbonds are naturally formed by living organismsand require species-specific analytical methods.With the exception of simple alkylselenium species(Me2Se, Me2Se2, MeEtSe) and methylarsinic andmethylarsonic acids these compounds cannot bechromatographed in the gas phase and are usuallyanalysed by HPLC-ICP MS, which requires acustom designed sample preparation. Regardless ofthe sample preparation the sample to be introducedon the column should be in aqueous solutionpassing through a 0.2 µm filter.

3.1 Organoarsenic compounds

The species of interest include As(III), As(V),monoarsinic and dimethylarsonic acids, arsenobe-taine, arsenocholine and a number of arsenosug-ars in biological materials. An overview of thesetechniques can be found in two recent exten-sive reports [152, 153]. A freezing procedure hascommonly been used to preserve a biosample;arsenic speciation in fresh and defrosted sam-ples is carried out [154]. Arsenobetaine in sam-ple extracts that were stored at 4 ◦C for 9 monthsdecomposed to trimethylarsine oxide and twoother unidentified arsenic species [154, 155]. Urine

samples can be injected on a chromatographiccolumn directly (filtered) or after dilution withacidified acid [156–159]. Urine samples were col-lected in polycarbonate bottles and filtered througha 0.45 µm filter [160]. The storage procedureimplies the use of acid-washed PE bottles at−10 ◦C [161] but storage at 20 ◦C [160] has alsobeen reported. For solid materials the recoveryof organic arsenic from the matrix and separationfrom the matrix are prerequisites for a chromato-graphic analysis.

For foodstuff samples, defatting, e.g. by leach-ing with acetone, is the first step [162]. It is nec-essary to avoid generating an emulsion with thefat, which would make the subsequent clean-upmore difficult [163]. In addition, the efficiency ofthe subsequent methanol extraction step is appar-ently higher for defatted samples than for nonde-fatted ones [162]. The uncleaned samples generateproblems on the level of ICP MS cones and elec-trospray ion source.

Extraction of arsenobetaine, arsenocholine andarsenoribosides has usually been performed usingmethanol [162], methanol–chloroform–water ormethanol–water. A comparison study of thesemethods is available [164]. A methanol–watermixture is recommended for the dry tissueswhereas fresh samples can be efficiently leachedwith pure methanol. With CRM 422 (dried codmuscle certified for the total As content only, butwidely used in speciation studies) a precipitate offatty aspect during CH3OH–H2O extraction hasbeen observed [164]. Recoveries of 90 % for fishand 80 % for mussels have been achieved. Nodegradation of arsenobetaine to more toxic specieswas observed when an enzymic (trypsin) digestionprocedure was applied to the fish [165, 166]. Themethanolic extraction is typically repeated 2–3times followed by preconcentration of the extractby evaporation of methanol using a rotavaporator.

For fatty samples it is recommended toremove the lipids by shaking with diethylether orpetroleum [167]. Since some samples of seafoodproducts are prepared in oil and generally tendto have a high salt content an additional cleanupstep, e.g. on a strong cation exchanger [162], isrequired to eliminate the remains of liposoluble


compounds not extracted with acetone. The clean-up also avoids the need for periodically reversingand flushing the chromatography column in orderto overcome pressure buildup due to accumulationof material on the column.

There has been a surge of interest recently inthe use of microwave-assisted procedures for therecovery of organoarsenic compounds from bio-logical tissues [163, 168, 169]. A recent alterna-tive is accelerated solvent extraction (ASE). Itis based on performing static extractions at ele-vated temperatures and pressures. The pressurecan be programmed without the use of elevatedsolvent temperatures that could lead to decom-position of thermally unstable compounds. Theoptimization of extraction of organoarsenic com-pounds from seaweed by ASE has been dis-cussed [170, 171].

3.2 Organoselenium compounds

Since selenoaminoacids are soluble in water,leaching with hot water has been judged suffi-cient to recover selenium species not incorporatedinto larger molecules. The sample is homoge-nized with water, sonicated or heated, and ultra-centrifuged. The typical recovery of seleniumextracted in this way is ca. 10–15 % [172–176]but it may reach 100 % in the case of someselenized yeast. Selenocysteine and some otherselenoaminoacids are highly susceptible to oxida-tive degradation, because the selenol group hasa significantly lower oxidation potential than itssulfur counterpart. The carboxymethyl derivativehas been synthesized (by addition of iodoaceticacid) to stabilize selenocysteine and prevent itsdegradation [177].

The low yields of the aqueous leaching pro-cedure for some species and samples have ledsome workers to use more aggressive leachingmedia. A trade-off is always necessary betweenthe recovery of selenium from a solid matrix andthe preservation of the original selenium species.As shown by Casiot et al. [174] the addition ofSDS to a leaching mixture increases the yield ofSe by releasing selenoaminoacids bound in seleno-proteins. The recovery of selenoaminoacids can be

increased to above 95 % by degrading the speciesoriginally present with a mixture of proteolyticenzymes [175]. An example of the effect of samplepreparation procedure on the recovery and specia-tion for a garlic sample is shown in Figure 3.2.5.Similar results have been reported elsewhere foryeast [174].

Care is advised in the interpretation of literaturedata since the results depend on the sample

0

1.0

2.0

3.0

4.0

5.0

0 10 20 30 40

× 104

0

2.0

4.0

6.0× 103

solubilizationwith SDS(13.3 %)

0

0.2

0.4

0.6

0.8

1.0 × 104

leachingwith water(56.9 %)

0

1.0

2.0

Time, min

Sig

nal i

nten

sity

, cps

× 103

enzymolysiswith driselase

(17.1 %)

enzymolysis withlipase-protease

(5.3 %)

0 10 20 30 40

0 10 20 30 40

0 10 20 30 40

Figure 3.2.5. Effect of the sample preparation procedure onspeciation analysis for selenium in garlic.

SAMPLE PREPARATION FOR COMPLEXES OF METALS WITH BIOLIGANDS 111

preparation procedure. This applies in particularto the frequently used statement ‘the majorityof Se is present as selenomethionine’, describingthe result of a procedure involving an enzymicdigestion. Actually, selenomethionine usually con-stitutes a part of a larger stable selenoproteinthat was destroyed during the sample prepara-tion procedure.

Free selenoaminoacids can be separated byultrafiltration (in breast milk) [178] or dialysis(algal extract) [179]. A recent paper discusses theuse of crown ethers to eliminate salts from urinesamples prior to HPLC or CZE of organoseleniumcompounds [180].

In mammals, speciation of selenium is mostlyconcerned with the determination of the differ-ent Se-containing proteins [181–188]. The mostimportant seem to be selenoprotein P, a majorprotein which is sometimes used as a biochem-ical marker of selenium status [183], seleno-enzymes such as several glutathione peroxidasesand type 1 idothyronine de-iodinase [183, 186,188] and albumin [182, 186, 187]. More than 25Se-containing proteins or protein subunits havebeen detected in rat tissues labelled in vivo with75Se [188]. The sample preparation proceduresinclude equilibration with a physiological bufferand ultracentrifugation.

4 SAMPLE PREPARATION INSPECIATION OF COORDINATIONCOMPLEXES OF METALSWITH BIOLIGANDS

The previously discussed cases concerned organo-metallic and organometalloid species in which thecarbon–metal(metalloid) bond conferred speciesstability in terms of acid–base conditions inthe system. An increasingly important class ofmetallocompounds in speciation and fractionationstudies includes coordination complexes. Most ofthese species are sensitive to pH changes and othercoordination ligands present in the system, whichrequires particular attention during their recoveryfrom samples prior to liquid chromatographyor capillary zone electrophoresis. The majorbioligands include proteins and small anions with

specific functions, such as citrate, ATP, porphyrinsor cobalamins. The greatest interest is attached toessential elements, which include some transitionmetals such as iron, copper and zinc, and toxicmetals such as Al, Cr, Pb, Cd and Hg. Metalscan be an integral part of metalloproteins andmetalloenzymes, e.g. ferritin (Fe, Cu, Zn), β-amylase (Cu), alcohol dehydrogenase (Cd, Zn) andcarbonic anhydrase (Cu, Zn), or bind less firmly totransport proteins (albumin, transferrin).

4.1 Biological fluids

The most common body fluids include blood(subdivided by centrifugation into serum andpacked cells, and when an anti-coagulant isadded the blood is separated into plasma and redcells (erythrocytes)), amniotic fluid, breast milkand urine.

Sampling, sample preservation and preparationprior to chromatography are particularly criticalin clinical chemistry because of the low concen-trations involved (risk of contamination) the ther-modynamical instability of some species, and thecomplexity of the matrix [189–191].

Sample preparation of serum prior to HPLCincludes filtration of sample on a 0.45 µm or0.2 µm filter [192]. Erythrocytes were subjectedto three freeze–thaw cycles to lyse the cells [193,194], followed by a tenfold dilution with a bufferand centrifugation at 18 000 g to remove fragmentsof membranes, etc. An alternative procedure rec-ommended by Cornelis et al. [190] was based onmixing one part of packed cells with one partof toluene and 40 parts of ice-cold water, fol-lowed by centrifugation and 0.45 µm filtration ofthe lysate.

Blood cells need to be lysed to free their contentprior to chromatographic separation. The super-natant was further diluted. When low molecularweight compounds are of interest, ultrafiltra-tion on a 10 kDa filter is carried out [195–197].The filtrate is protein free and can be ana-lyzed, e.g. for metallodrug metabolites or por-phyrins. Breast milk should be centrifuged toremove fat; precipitation of casein with 1 Macetate is optional [198]. Dialysis and purification


by size-exclusion chromatography is required ifseparation by RP HPLC is to be undertaken [198].

Amniotic fluid is a urine-like fluid inhaled andswallowed by the human fetus. Some heavy met-als, e.g. Pb can cross the placenta and end up inamniotic fluid. Metal-binding ligands are importantin amniotic fluid because of the potential of beingtransporters to the neurological system [199]. Theamniotic fluid sample was centrifuged and thesupernatant was stored frozen at −20 ◦C priorto analysis.

4.2 Plant and animal tissues

Liver and kidney have been the most widelystudied organs because of their crucial function inthe metabolism of metals.

Washing cells in a Tris-HCl buffer (pH 8)containing 1 M EDTA to remove metal ionsreversibly bound to the cell wall has been rec-ommended [177]. In the majority of studies size-exclusion preparative chromatography [200–204]has been preferred to heat treatment [205–207] forthe isolation of the metallothionein (MT) fractionfrom the tissue cytosol. Guidelines for the prepara-tion of biological samples prior to quantification ofMTs were discussed with particular attention givento the care necessary to avoid oxidation [208].

Soluble extracts of tissues and cultured cellsare prepared by homogenizing tissue samples inan appropriate buffer [209]. Neutral buffers arenecessary for extracting MTs since Zn starts todissociate from the protein at pH 5. Cd and Cuare removed at lower pH values. A 10–50 mMTris-HCl buffer at pH 7.4–9 is the most commonchoice. For cytosols containing Cd-induced MTsdilution factors up to 10 have been used whereasfor those with natural MT levels equal amounts oftissue and buffer have been found suitable.

Metallothioneins are prone to oxidation duringisolation due to their high cystein content. Dur-ing oxidation disulfide bridges are formed andthe MTs either copolymerize or combine withother proteins to move into the high molecu-lar weight fraction. Since MTs may be oxidizedby, e.g. oxygen, Cu(I) or heme components, the

homogenization of tissues and subsequent isola-tion of MTs should be normally performed indeoxygenated buffers and/or in the presence ofa thiolic reducing agent [210]. β-Mercaptoethanolis added as antioxidant which additionally pre-vents formation of dimeric forms of MT [211].Other components added during homogenizationinclude 0.02 % NaN3 added as an antibacterialagent and phenylmethane-sulfonylfluoride which isadded as a protease inhibitor. The homogenizationstep is followed by centrifugation. The use of anrefrigerated ultracentrifuge (100 000 g) is stronglyrecommended.

As a result two fractions, a soluble one (cellsupernatant, cytosol) and a particulate one (cellmembranes and organelles), are obtained. Onlythe supernatant is analyzed for MTs. It is recom-mended that it is stored at −20 ◦C under nitrogenprior to analysis [211]. The extraction efficiencyby sucrose-Tris at pH 8 was 25–30 % [208]. Thepercentage of metals in cytosols is typically50–85 % and 30–57 % in the case of Zn and Cd,respectively [212, 213].

A heat treatment (at 60 ◦C for 15 min) of thecytosol extracts (especially of concentrated ones)is recommended to separate the high molecularweight fraction, which coagulates from the super-natant (containing MTs which are heat stable).Such a treatment reduces the protein load on theHPLC column not only improving the separa-tion of MT isoforms but also prolonging the col-umn lifetime. Enzymolysis in simulated gastric andgastrointestinal juice has been proposed for meatsamples [205].

Filtration of the cytosol (0.22 µm filter) beforeintroducing onto the chromatographic column isstrongly advised. A guard column should beinserted to protect the analytical column partic-ularly from effects of lipids, which otherwisedegrade the separation [205]. Any organic speciesthat adhere to the column can also bind inorganicspecies giving rise to anomalous peaks in subse-quent runs [205]. A new guard column was usedfor each injection to prevent adsorption by ligandswith a high affinity for cadmium that would oth-erwise interfere with subsequent injections [205].An extensive column cleanup was necessary [205].

REFERENCES 113

To avoid contamination of the analytical columnby trace elements, buffers should be cleaned byChelex-100 [205].

Higher molecular weight compounds, referredto as protein complexes with Cd or Ni, wereextracted into warm H2O. The extracts werecentrifuged and the proteins were fractionatedby successive filtration through membranes withmolecular weight cut-offs of 500, 5000 and30 000 Da [214–216] prior to off-line GF AAS.Guenther and coworkers [213, 217, 218] used lowpressure semi-preparative chromatography to sep-arate water-soluble species of Cd and Zn. The useof concentrated surfactants and dithiothreitol areimportant for the solubilization of high molecularweight proteins and metalloenzymes [219].

Water-soluble polysaccharide species withhigher molecular weights can be readily decom-posed by enzymic hydrolysis with a mixture ofpectinase and hemicellulase to release the dRG-II complex [220]. The same mixture was foundto be efficient for extracting the dRG-II–metalcomplexes from water-insoluble residue of veg-etables owing to the destruction of the pecticstructure [220].

Some attention has been paid to the analysisof enzymic digests of foodstuffs in the quest ofmolecular information to contribute to the knowl-edge on the bioavailability of some elements.Enzymolysis in simulated gastric and gastrointesti-nal juice has been proposed for meat [221]. Thesoluble fraction of the stomach and upper intesti-nal contents of a guinea pig on different diets wereinvestigated for the species of Al, Cu, Zn, Mn, Srand Rb. The effect of citrate on each of these ele-ments has also been assessed [221]. An enzymicdigest of bovine thyroglobulin has been analyzedfor iodine species [222].

5 CONCLUSIONS–TRENDSAND PERSPECTIVES

Novel derivatization reagents and instrumentaltechniques have recently rendered the samplepreparation prior to speciation of organometalliccompounds by hyphenated techniques faster and

easier to automate. The ultimate goal of thedevelopment of an automated speciation analyzerseems to be closer than ever. The critical issueremains the certainty that the species arriving atthe detector from the chromatographic columnis the one originally present in the sample. Itseems to be possible to control the stability ofspecies for which standards can be synthesized,such as anthropogenic organometallic pollutants.A lot, however, still remains to be done to controltransformations of metal coordination complexeswith bioligands during the sample preparation,especially because many of these species remainin complex chemical equilibria in a sample, whichare destroyed once a chemical or even a physical(dilution) operation is carried out on the sample.Particular care is advised regarding the speciesstability during sampling and storage. The recentdevelopment of isotope dilution analysis now givesa possibility of improving the accuracy duringthe quantification and the means of unambiguousdetermination of the species with a spike ofspecific compounds with a different isotopic ratio.

6 REFERENCES

1. Lobinski, R., Appl. Spectrosc., 51, 260A (1997).2. Quevauviller, P., Method Performance Studies for Spe-

ciation Analysis , Royal Society of Chemistry, Cam-bridge, 1998.

3. Caruso, J. A., Sutton, K. L. and Ackley, K. L., Elemen-tal Speciation. New Approaches for Trace Element Anal-ysis , Elsevier, Amsterdam, 2000.

4. Lobinski, R. and Marczenko, Z., Spectrochemical TraceAnalysis for Metals and Metalloids , Elsevier, Amster-dam, 1996.

5. Gomez-Ariza, J. L., Morales, E., Giraldez, I., Sanchez-Rodas, D. and Velasco, A., J. Chromatogr. A, 932, 211(2001).

6. Szpunar Lobinska, J., Ceulemans, M., Lobinski, R.and Adams, F. C., Anal. Chim. Acta., 278, 99 (1993).

7. Emteborg, H., Baxter, D. C., Sharp, M. and Frech, W.,Analyst , 69 (1995).

8. Lansens, P. and Baeyens, W., Anal. Chim. Acta., 228,93 (1990).

9. Perez Corona, T., Madrid Albarran, Y., Camara, C.and Beceiro, E., Spectrochim. Acta, Part B., 53B, 321(1998).

10. Jian, W. and McLeod, C. W., Talanta , 39, 1537 (1992).11. Johansson, M., Emteborg, H., Glad, B., Reinholdsson, F.

and Baxter, D. C., Fresenius’ J. Anal. Chem., 351, 461(1995).


12. Sanchez, D. M., Martin, R., Morante, R., Marin, J. andMunuera, M. L., Talanta , 52, 671 (2000).

13. Martinez Blanco, R., Tagle Villanueva, M., SanchezUria, J. E. and Sanz-Medel, A., Anal. Chim. Acta, 419,137 (2000).

14. Salih, B., Spectrochim. Acta B , 55, 1115 (2000).15. Ballesteros, E., Gallego, M. and Valcarcel, M., J. Chro-

matogr. A, 869, 101 (2000).16. Eisert, R. and Pawliszyn, J., Crit. Rev. Anal. Chem., 27,

103 (1997).17. Pawliszyn, J. B., Solid-Phase Microextraction: The-

ory and Practice Wiley VCH, Weinheim, Germany(1997).

18. Snell, J. P., Frech, W. and Thomassen, Y., Analyst ,1055 (1996).

19. Cai, Y., Monsalud, S., Furton, K. G., Jaffe, R. andJones, R. D., Appl. Organomet. Chem., 12, 565 (1998).

20. Mothes, S. and Wenrich, R., J. High Resol. Chro-matogr., 22, 181 (1999).

21. Moens, L., de Smaele, T., Dams, R., van den Broek, P.and Sandra, P., Anal. Chem., 69, 1604 (1997).

22. Gorecki, T. and Pawliszyn, J., Anal. Chem., 68, 3008(1996).

23. Yu, X. and Pawliszyn, J., Anal. Chem., 72, 1788(2000).

24. Aguerre, S., Bancon-Montigny, C., Lespes, G. andPotin-Gautier, M., Analyst , 125, 236 (2000).

25. Aguerre, S., Lespes, G., Desauziers, V. and Potin-Gautier, M., J. Anal. At. Spectrom., 16, 263 (2001).

26. Mester, Z., Vitanyi, G., Morabito, R. and Fodor, P., J.Chromatogr. A, 832, 183 (1999).

27. Mester, Z. and Pawliszyn, J., J. Chromatogr., 873, 129(2000).

28. Mester, Z., Lam, J., Sturgeon, R. and Pawliszyn, J., J.Anal. At. Spectrom., 15, 837 (2000).

29. Cai, Y. and Bayona, J. M., J. Chromatogr. A, 696, 113(1995).

30. Morcillo, Y., Cai, Y. and Bayona, J. M., J. High. Reso-lut. Chromatogr., 18, 767 (1995).

31. Vercauteren, J., De Meester, A., De Smaele, T., Van-haecke, F., Moens, L., Dams, R. and Sandra, P., J.Anal. At. Spectrom., 15, 651 (2000).

32. Mester, Z., Sturgeon, R. and Pawliszyn, J., Spectrochim.Acta , B56, 233 (2001).

33. Dunnemann, L., Hajimiragha, H. and Begerow, J., Fre-senius’ J. Anal. Chem., 363, 466 (1999).

34. Tutschku, S., Mothes, S. and Wennrich, R., Fresenius’J. Anal. Chem., 354, 587 (1996).

35. Jiang, G., Liu, J. and Yang, K., Anal. Chim. Acta, 421,67 (2000).

36. Yu, X., Yuan, H., Gorecki, T. and Pawliszyn, J., Anal.Chem., 71, 2998 (1999).

37. Fragueiro, M., Alava-Moreno, F., Lavilla, I. andBendicho, C., J. Anal. At. Spectrom., 15, 705(2000).

38. Beichert, A., Padberg, S. and Wenclawiak, B. W., Appl.Organomet. Chem., 14, 493 (2000).

39. Guidotti, M., J. AOC Intern., 83, 1082 (2000).40. Szostek, B. and Alstadt, J. H., J. Chromatogr. A, 807,

253 (1998).41. Wu, J., Mester, Z. and Pawlliszin, J., Anal. Chim. Acta,

424, 211 (2000).42. Orellana-Velado, N., Pereiro, R. and Sanz-Medel, A., J.

Anal. At. Spectrom., 16, 376 (2001).43. Radojevic, M., Allen, A., Rapsomanikis, S. and Harri-

son, R. M., Anal. Chem., 58, 658 (1986).44. Lobinski, R., Szpunar Lobinska, J., Adams, F. C.,

Teissedre, P. L. and Cabanis, J. C., J. Assoc. Off. Anal.Chem. Int., 76, 1262 (1993).

45. Lobinski, R., Boutron, C. F., Candelone, J. P., Hong, S.,Szpunar Lobinska, J. and Adams, F. C., Anal. Chem.,65, 2510 (1993).

46. Lobinski, R. and Adams, F. C., Anal. Chim. Acta, 262,285 (1992).

47. Heisterkamp, M., de Smaele, T., Candelone, J. P.,Moens, L., Dams, R. and Adams, F. C., J. Anal. At.Spectrom., 12, 1077 (1997).

48. Heisterkamp, M. and Adams, F. C., Fresenius’ J. Anal.Chem., 362, 489 (1998).

49. Bergmann, K. and Neidhart, B., Fresenius’ J. Anal.Chem., 356, 57 (1996).

50. Lobinski, R., Dirkx, W. M. R., Ceulemans, M. andAdams, F. C., Anal. Chem., 64, 159 (1992).

51. Ceulemans, M. and Adams, F. C., Anal. Chim. Acta,317, 161 (1995).

52. Bergmann, K., Roehr, U. and Neidhart, B., Fresenius’J. Anal. Chem., 349, 815 (1994).

53. Bulska, E., Baxter, D. C. and Frech, W., Anal. Chim.Acta., 12, 545 (1991).

54. Bulska, E., Emteborg, H., Baxter, D. C., Frech, W.,Ellingsen, D. and Thomassen, Y., Analyst , 117, 657(1992).

55. Blais, J. S. and Marshall, W. D., J. Environ. Qual., 15,255 (1986).

56. Gomez Ariza, J. L., Morales, E. and Ruiz Benitez, M.,Analyst , 641 (1992).

57. Harino, H., Fukushima, M. and Tanaka, M., Anal. Chim.Acta , 264, 91 (1992).

58. Forsyth, D. S., Weber, D. and Cleroux, C., Food Addit.Contam., 9, 161 (1992).

59. Forsyth, D. S., Weber, D. and Dalglish, K., J. AOACInt., 75, 964 (1992).

60. Gremm, T. J. and Frimmel, F. H., Water Res. 1992 , 26,1163 (1992).

61. Waldock, M. J., Mikrochim Acta., 109, 23 (1992).62. Dowling, T. M. and Uden, P. C., J. Chromatogr., 644,

153 (1993).63. Rapsomanikis, S., Analyst , 1429 (1994).64. Szpunar, J., Schmitt, V. O., Lobinski, R. and Monod,

J. L., J. Anal. At. Spectrom., 11, 193 (1996).

REFERENCES 115

65. Minganti, V., Capelli, R. and De Pellegrini, R., Frese-nius’ J. Anal. Chem., 351, 471 (1995).

66. Ceulemans, M., Lobinski, R., Dirkx, W. M. R. andAdams, F. C., Fresenius’ J. Anal. Chem., 347, 256(1993).

67. Heisterkamp, M. and Adams, F. C., J. Anal. At. Spec-trom., 14, 1307 (1999).

68. Hu, G. L. and Wang, X. R., Anal. Lett., 31, 1445(1998).

69. Cai, Y., Monsalud, S., Jaffe, R. and Jones, R. D., J.Chromatogr. A, 876, 147 (2000).

70. Bloom, N., Can. J. Fish Aquat. Sci., 46, 1131 (1989).71. Horvat, M., Bloom, N. S. and Liang, L., Anal Chim

Acta., 281, 135 (1993).72. Bowles, K. C. and Apte, S. C., Anal Chem., 70, 395

(1998).73. Bowles, K. C. and Apte, S. C., Anal. Chim. Acta, 419,

145 (2000).74. Gruter, U. M., Hitzke, M., Kresimon, J. and Hirner,

A. V., J. Chromatogr. A, 938, 225 (2001).75. Dedina, J. and Tsalev, D. L., Hydride-Generation

Atomic-Absorption Spectrometry, John Wiley, Chich-ester, 1995.


77. Ceulemans, M. and Adams, F. C., J. Anal. At. Spec-trom., 11, 201 (1996).

78. Tseng, C. M., de Diego, A., Pinaly, H., Amouroux, D.and Donard, O. F. X., J. Anal. At. Spectrom., 13, 755(1998).

79. Craig, P. J., Dewick, R. J. and van Elteren, J. T., Fre-senius’ J. Anal. Chem., 351, 467 (1995).

80. de la Calle Guntinas, M. B., Lobinski, R. and Adams,F. C., J. Anal. At. Spectrom., 10, 111 (1995).

81. Wasik, A., Pereiro, I. R., Dietz, C., Szpunar, J. andLobinski, R., Anal. Commun., 35, 331 (1998).

82. Wasik, A., Rodriguez Pereiro, I. and Lobinski, R., Spec-trochim. Acta, Part B , 53B, 867 (1998).

83. Dietz, C., Madrid, Y., Camara, C. and Quevauviller, P.,Anal. Chem., 72, 4178 (2000).

84. Segovia Garcia, E., Garcia Alonso, J. I. and SanzMedel, A., J. Mass Spectrom., 32, 542 (1997).

85. Pannier, F., Astruc, A. and Astruc, M., Anal. Chim.Acta., 287, 17 (1994).

86. Tutschku, S., Mothes, S. and Dittrich, K., J. Chro-matogr. A, 683, 269 (1994).

87. Liu, Y., Lopez Avila, V., Alcaraz, M. and Bec-kert, W. F., J. High Resolut. Chromatogr., 17, 527(1994).

88. Cai, Y., Rapsomanikis, S. and Andreae, M. O., Anal.Chim. Acta., 274, 243 (1993).

89. Cai, Y., Alzaga, R. and Bayona, J. M., Anal. Chem., 66,1161 (1994).

90. Cai, Y., Rapsomanikis, S. and Andreae, M. O., J. Anal.At. Spectrom., 8, 119 (1993).

91. Chau, Y. K., Yang, F. and Maguire, R. J., Anal. Chim.Acta., 320, 165 (1996).

92. Quevauviller, P. and Morabito, R., Trends Anal. Chem.,19, 86 (2000).

93. Hintelmann, H., Evans, R. D. and Villeneuve, J. Y., J.Anal. At. Spectrom., 10, 619 (1995).

94. Falter, R. and Ilgen, G., Fresenius’ J. Anal. Chem., 358,407 (1997).

95. Falter, R. and Ilgen, G., Fresenius’ J. Anal. Chem., 358,401 (1997).

96. Eiden, R., Falter, R., Augustin Castro, B. and Schoeler,H. F., Fresenius’ J. Anal. Chem., 357, 439 (1997).

97. Padberg, S., Burow, M. and Stoeppler, M., Fresenius’J. Anal. Chem., 346, 686 (1993).

98. Bloom, N. S., Colman, J. A. and Barber, L., Fresenius’J. Anal. Chem., 358, 371 (1997).

99. Pellegrino, C., Massanisso, P. and Morabito, R., TrendsAnal. Chem., 19, 97 (2000).

100. Zhang, S., Chau, Y. K. and Chau, Q. S. Y., Appl. Or-ganomet. Chem., 5, 431 (1991).

101. Lorenzo, R. A., Vazquez, M. J., Carro, A. M. andCela, R., Trends Anal. Chem., 18, 410 (1999).

102. Hardy, S. and Jones, P., J. Chromatogr. A, 791, 333(1997).

103. Fischer, R., Rapsomanikis, S. and Andreae, M. O.,Anal. Chem., 65, 763 (1993).

104. Rapsomanikis, S. and Craig, P. J., Anal. Chim. Acta.,248, 563 (1991).

105. Ceulemans, M., Witte, C., Lobinski, R. and Adams,F. C., Appl. Organomet. Chem., 8, 451 (1994).

106. Jimenez, M. S. and Sturgeon, R. E., J. Anal. At. Spec-trom., 12, 597 (1997).

107. Forsyth, D. S. and Iyengar, J. R., Appl. Organomet.Chem., 3, 211 (1989).

108. Pannier, F., Astruc, A. and Astruc, M., Anal. Chim.Acta., 327, 287 (1996).

109. Forsyth, D. S. and Iyengar, J. R., J. Assoc. Off. Anal.Chem., 72, 997 (1989).

110. Wai, C. M., Lin, Y., Brauer, R., Wang, S. and Beck-ert, W. F., Talanta , 40, 1325 (1993).

111. Vela, N. P. and Caruso, J. A., J. Anal. At. Spectrom., 11,1129 (1996).

112. Sun, Y. C., Mierzwa, J., Chung, Y. T. and Yang, M. H.,Anal. Commun., 34, 333 (1997).

113. Liu, Y., Lopez Avila, V., Alcaraz, M. and Bec-kert, W. F., J. High Resolut. Chromatogr., 16, 106(1993).

114. Holak, W., J. Assoc. Off. Anal. Chem. Int., 78, 1124(1995).

115. Emteborg, H., Bjorklund, E., Odman, F., Karlsson, L.,Mathiasson, L., Frech, W. and Baxter, D. C., Analyst ,19 (1996).

116. Bayona, J. M., Trends Anal. Chem., 19, 107 (2000).117. Szpunar, J., Schmitt, V. O., Donard, O. F. X. and

Lobinski, R., Trends Anal Chem., 15, 181 (1996).


118. Donard, O. F. X., Lalere, B., Martin, F. and Lobin-ski, R., Anal. Chem., 67, 4250 (1995).

119. Lalere, B., Szpunar, J., Budzinski, H., Garrigues, P. andDonard, O. F. X., Analyst , 2665 (1995).

120. Szpunar, J., Ceulemans, M., Schmitt, V. O., Adams,F. C. and Lobinski, R., Anal. Chim. Acta, 332, 225(1996).

121. Schmitt, V. O., Martin, F. M., Szpunar, J., Lobinski, R.and Donard, O. F. X., Spectra Anal., 25, 14 (1996).

122. Rodriguez, I., Santamarina, M., Bollain, M. H., Mejuto,M. C. and Cela, R., J. Chromatogr, A., 774, 379 (1997).

123. Rodriguez Pereiro, I., Schmitt, V. O. and Lobinski, R.,Anal. Chem., 69, 4799 (1997).

124. Tseng, C. M., de Diego, A., Martin, F. M. and Donard,O. F. X., J. Anal. At. Spectrom., 12, 629 (1997).

125. Shawky, S., Emons, H. and Duerbeck, H. W., Anal.Commun., 33, 107 (1996).

126. Rodriguez Pereiro, I., Schmitt, V. O., Szpunar, J.,Donard, O. F. X. and Lobinski, R., Anal. Chem., 68,4135 (1996).

127. Abuin, M., Carro, A. M. and Lorenzo, R. A., J. Chro-matogr. A, 889, 185 (2000).

128. Liu, W. P. and Lee, H. K., J. Chromatogr., 834, 45(1999).

129. Lobinski, R., Dirkx, W. M. R., Szpunar Lobinska, J.and Adams, F. C., Anal. Chim. Acta., 286, 381 (1994).

130. Jiang, G. B. and Zhou, Q. F., J. Chromatogr. A, 886,197 (2000).

131. Schwedt, G., Chromatographic Methods in InorganicAnalysis , Huthing, Heidelberg, 1981.

132. Kataoka, H., Miyanaga, Y. and Makita, M., J. Chro-matogr. A, 659, 481 (1994).

133. Cai, X. J., Block, E., Uden, P. C., Zhang, X., Quimby,B. D. and Sullivan, J. J., J. Agric. Food Chem., 43, 1754(1995).

134. Yasumoto, K., Suzuki, T. and Yoshida, M., J. Agric.Food Chem., 36, 463 (1988).

135. Ouyang, Z., Wu, J. and Xie, L., Anal. Biochem., 178,77 (1989).

136. Zheng, O. and Wu, J., Biomed. Chromatogr., 2, 258(1988).

137. Scerbo, R., de la Calle Guntinas, M. B., Brunori, C. andMorabito, R., Quim. Anal , 87 (1997).

138. Claussen, F. A., J. Chromatogr. Sci., 35, 568 (1997).139. Schoene, K., Steinhanses, J., Bruckert, H. J. and

Koenig, A., J. Chromatogr., 605, 257 (1992).140. Slaets, S., Adams, F., Rodriguez, I. and Lobinski, R., J.

Anal. At. Spectrom., 14, 851 (1999).141. Slaets, S. and Adams, F. C., Anal. Chim. Acta, 414, 141

(2000).142. Kim, A. W., Foulkes, M. E., Ebdon, L., Hill, S. J.,

Patience, R. L., Barwise, A. G. and Rowland, S. J., J.Anal. At. Spectrom., 7, 1147 (1992).

143. Rodriguez Pereiro, I. and Lobinski, R., J. Anal. At.Spectrom., 12, 1381 (1997).

144. Aue, W. A., Millier, B. and Sun, X. Y., Anal. Chem.,62, 2453 (1990).

145. Ombaba, J. M. and Barry, E. F., J. Chromatogr. A, 678,319 (1994).

146. Zeng, Y. and Uden, P. C., J. High Resolut. Chro-matogr., 17, 223 (1994).

147. Quimby, B. D., Dryden, P. C. and Sullivan, J. J., J.High. Resolut. Chromatogr., 14, 110 (1991).

148. Ebdon, L., Evans, E. H., Pretorius, W. G. and Row-land, S. J., J. Anal. At. Spectrom., 9, 939 (1994).

149. Tao, H., Murakami, T., Tominaga, M. and Miyazaki, A.,J. Anal. At. Spectrom., 13, 1085 (1998).

150. Frech, W., Baxter, D. C., Bakke, B., Snell, J. andThomassen, Y., Anal. Commun., 33, 7H (1996).

151. Snell, J., Qian, J., Johansson, M., Smit, K. andFrech, W., Analyst , 123, 905 (1998).

152. Burguera, M. and Burguera, J. L., Talanta , 44, 1581(1997).

153. Guerin, T., Astruc, A. and Astruc, M., Talanta , 50, 1(1999).

154. Le, S. X. C., Cullen, W. R. and Reimer, K. J., Environ.Sci. Technol., 28, 1598 (1994).

155. Muerer, A. J. L., Abildtrup, A., Poulsen, O. M. andChristensen, J. M., Analyst , 117, 677 (1992).

156. Larsen, E. H., Pritzl, G. and Hansen, S. H., J. Anal. At.Spectrom., 8, 557 (1993).

157. Ma, M. and Le, X. C., Clin. Chem., 44, 539 (1998).158. Tsalev, D. L., Sperling, M. and Welz, B., Analyst , 123,

1703 (1998).159. Zhang, X. R., Cornelis, R., De Kimpe, J., Mees, L. and

Lameire, N., Clin. Chem., 43, 406 (1997).160. Chana, B. S. and Smith, N. J., Anal. Chim. Acta, 197,

177 (1987).161. Heitkemper, D., Creed, J., Caruso, J. and Fricke, F. L.,

J. Anal. At. Spectrom., 4, 279 (1989).162. Ybanez, N., Velez, D., Tejedor, W. and Montoro, R., J.

Anal. At. Spectrom., 10, 459 (1995).163. Dagnac, T., Padro, A., Rubio, R. and Rauret, G.,

Talanta., 48, 763 (1999).164. Alberti, J., Rubio, R. and Rauret, G., Fresenius’ J. Anal.

Chem., 351, 420 (1995).165. Branch, S., Ebdon, L. and O’Neill, P., J. Anal. At.

Spectrom., 9, 33 (1994).166. Lamble, K. J. and Hill, S. J., Anal. Chim. Acta, 334,

261 (1996).167. Shiomi, K., Sugiyama, Y., Shimakura, K. and Naga-

shima, Y., Appl. Organometal. Chem., 9, 105 (1995).168. Ackley, K. L., B’Hymer, C., Sutton, K. L. and Caruso,

J. A., J. Anal. At. Spectrom., 14, 845 (1999).169. Helgesen, H. and Larsen, E. H., Analyst , 123, 791

(1998).

REFERENCES 117

170. McKiernan, J. W., Creed, J. T., Brockhoff, C. A., Caruso,J. A. and Lorenzana, R. M., J. Anal. At. Spectrom., 14,607 (1999).

171. Gallagher, P. A., Shoemaker, J. A., Wei, X. N.,Brockhoff-Schwegel, C. A. and Creed, J. T., Fresenius’J. Anal. Chem., 369, 71 (2001).

172. Bird, S. M., Ge, H., Uden, P. C., Tyson, J. F., Block, E.and Denoyer, E., J. Chromatogr. A, 789, 349 (1997).

173. Bird, S. M., Uden, P. C., Tyson, J. F., Block, E.and Denoyer, E., J. Anal. At. Spectrom., 12, 785(1997).

174. Casiot, C., Szpunar, J., Lobinski, R. and Potin Gau-tier, M., J. Anal. At. Spectrom., 14, 645 (1999).

175. Gilon, N., Astruc, A., Astruc, M. and Potin-Gautier, M.,Appl. Organomet. Chem., 9, 623 (1995).

176. Zheng, J., Goessler, W. and Kosmus, W., Trace Ele-ments Electrol., 15, 70 (1998).

177. Takatera, K., Osaki, N., Yamaguchi, H. and Watan-abe, T., Anal. Sci., 10, 567 (1994).

178. Michalke, B. and Schramel, P., J. Chromatogr. A, 807,71 (1998).

179. Fan, T. M., Lane, A. N., Martens, D. and Higashi, R. M.,Analyst , 123, 875 (1998).

180. Gammelgaard, B., Joens, O. and Bendahl, L., J. Anal.At. Spectrom., 16, 339 (2001).

181. Thomson, C. D., Analyst , 123, 827 (1998).182. Plecko, T., Nordmann, S., Rukgauer, M. and Kruse-

Jarres, J. D., Fresenius’ J. Anal. Chem., 363, 517(1999).

183. Persson Moschos, M., Huang, W., Srikumar, T. S.,Akesson, B. and Lindeberg, S., Analyst , 120, 833(1995).

184. Koyama, H., Kasanuma, Y., Kim, C., Ejima, A., Watan-abe, C., Nakatsuka, H. and Satoh, H., Tohoku J. Exp.Med., 178, 17 (1996).

185. Koyama, H., Omura, K., Ejima, A., Kasanuma, Y.,Watanabe, C. and Satoh, H., Anal. Biochem., 267, 84(1999).

186. Harrison, I., Littlejohn, D. and Fell, G. S., Analyst , 121,189 (1996).

187. Deagen, J. T., Butler, J. A., Zachara, B. A. and Whanger,P. D., Anal. Biochem., 208, 176 (1993).

188. Behne, D., Weiss Nowak, C., Kalcklosch, M., West-phal, C., Gessner, H. and Kyriakopoulos, A., Analyst ,120, 823 (1995).

189. Behne, D., Analyst , 117, 555 (1992).190. Cornelis, R., De Kimpe, J. and Zhang, X., Spectrochim.

Acta, Part B , 53B, 187 (1998).191. Sanz Medel, A., Spectrochim. Acta, Part B , 53B, 197

(1998).192. Van Landeghem, G. F., D’Haese, P. C., Lamberts, L. V.

and De Broe, M. E., Anal. Chem., 66, 216 (1994).

193. Bergdahl, I. A., Schuetz, A. and Grubb, A., J. Anal. At.Spectrom., 11, 735 (1996).

194. Gercken, B. and Barnes, R. M., Anal. Chem., 63, 283(1991).

195. Matz, S. G., Elder, R. C. and Tepperman, K., J. Anal.At. Spectrom., 4, 767 (1989).

196. Poon, G. K., Raynaud, F. I., Mistry, P., Odell, D. E.,Kelland, L. R., Harrap, K. R., Barnard, C. F. J. andMurrer, B. A., J. Chromatogr. A, 712, 61 (1995).

197. De Waal, W. A. J., Maessen, F. J. M. J. and Kraak, J.C., J. Chromatogr., 407, 253 (1987).

198. Makino, Y. and Nishimura, S., J. Chromatogr., BiomedAppl., 117, 346 (1992).

199. Hall, G. S., Zhu, E. G. and Martin, E. G., Anal. Com-mun., 26, 93 (1999).

200. Ang, S. G. and Wong, V. W. T., J. Liq. Chromatogr.,14, 2647 (1991).

201. Ang, S. G. and Wong, V. W. T., J. Chromatogr., 599,21 (1992).

202. Apostolova, M., Bontchev, P. R., Nachev, C. and Sir-akova, I., J. Chromatogr., Biomed. Appl. 29 Oct 1993 ,131, 191 (1993).

203. Van Beek, H. and Baars, A. J., J. Anal. At. Spectrosc.,11, 70 (1990).

204. Steinebach, O. M. and Wolterbeek, H. T., J. Chroma-togr., Biomed Appl., 130, 199 (1993).

205. Crews, H. M., Dean, J. R., Ebdon, L. and Massey, R.C., Analyst , 114, 895 (1989).

206. Mazzucotelli, A. and Rivaro, P., Microchem. J., 51, 231(1995).

207. Mazzucotelli, A., Viarengo, A., Canesi, L., Ponzano, E.and Rivaro, P., Analyst , 116, 605 (1991).

208. Suzuki, K. T. and Sato, M., Biomed. Res. Trace. Elem.,6, 51 (1995).

209. Polec, K., Perez-Calvo, M., Garcıa-Arribas, O.,Szpunar, J., Ribas-Ozonas, B. and Lobinski, R., J. Anal.At. Spectrom., 15, 1363 (2000).

210. High, K. A., Methven, B. A., McLaren, J. W., Siu,K. W. M., Wang, J., Klaverkamp, J. F. and Blais, J. S.,Fresenius’ J. Anal. Chem., 351, 393 (1995).

211. Van Beek, H. and Baars, A. J., J. Chromatogr., 442, 345(1988).

212. Jayawickreme, C. K. and Chatt, A., J. Radioanal. Nucl.Chem., 124, 257 (1988).

213. Guenther, K. and Waldner, H., Anal. Chim. Acta., 259,165 (1992).

214. Lange Hesse, K., Dunemann, L. and Schwedt, G., Fre-senius’ J. Anal. Chem., 339, 240 (1991).

215. Lange Hesse, K., Fresenius’ J. Anal. Chem., 350, 68(1994).

216. Lange Hesse, K., Dunemann, L. and Schwedt, G., Fre-senius’ J. Anal. Chem., 349, 460 (1994).


217. Guenther, K. and Von Bohlen, A., Spectrochim. Acta.,Part B , 46B, 1413 (1991).

218. Guenther, K., von Bohlen, A. and Strompen, C., Anal.Chim. Acta., 309, 327 (1995).

219. Leopold, I. and Fricke, B., Anal. Biochem., 252, 277(1997).

220. Szpunar, J., Pellerin, P., Makarov, A., Doco, T.,Williams, P. and Lobinski, R., J. Anal. At. Spectrom.,14, 639 (1999).

221. Owen, L. M. W., Crews, H. M., Massey, R. C. andBishop, N. J., Analyst , 120, 705 (1995).

222. Takatera, K. and Watanabe, T., Anal. Chem., 65, 759(1993).

3.3 Sample Preparation – Fractionation (Sediments,Soils, Aerosols, and Fly Ashes)

Jozsef Hlavay and Klara PolyakUniversity of Veszprem, Hungary

1 Introduction . . . . . . . . . . . . . . . . . . . . . . 1192 Subsampling, Storage, and Preparation of

Sediment Samples . . . . . . . . . . . . . . . . . . 1213 Subsampling, Storage, and Preparation of

Soil Samples . . . . . . . . . . . . . . . . . . . . . 1224 Subsampling, Storage, and Preparation of

Aerosol Samples . . . . . . . . . . . . . . . . . . . 1235 Subsampling, Storage, and Preparation of

Fly Ash Samples . . . . . . . . . . . . . . . . . . 1266 Sequential Extraction Techniques . . . . . . 1267 Standardized Sequential Extraction

Procedure Proposed by BCR . . . . . . . . . . 1288 Sequential Extraction Schemes Applied to

Sediment Samples . . . . . . . . . . . . . . . . . . 1299 Sequential Extraction Schemes Applied to

Soil Samples . . . . . . . . . . . . . . . . . . . . . 132

10 Sequential Extraction Schemes Applied toAerosol Samples . . . . . . . . . . . . . . . . . . . 134

11 Sequential Extraction Schemes Applied toFly Ash Samples . . . . . . . . . . . . . . . . . . 139

12 Principles and Application of Field-flowFractionation (FFF), Separation ofSuspended Particles into Specific SizeFractions . . . . . . . . . . . . . . . . . . . . . . . . 14012.1 Normal mode of FFF . . . . . . . . . . 14012.2 Thermal FFF (ThFFF) . . . . . . . . . . 14212.3 Flow FFF . . . . . . . . . . . . . . . . . . . 142

13 Discussion . . . . . . . . . . . . . . . . . . . . . . . 14414 Acknowledgements . . . . . . . . . . . . . . . . . 14415 References . . . . . . . . . . . . . . . . . . . . . . . 144

1 INTRODUCTION

Increasingly strict environmental regulations req-uire the development of new methods for analysisand ask for simple and meaningful tools to obtaininformation on toxic fractions of different mobilityand bioavailability in the solid phases. Objectivesof monitoring are to assess pollution effects on manand his environment, to identify possible sourcesand to establish relationships between pollutantconcentrations [1–6]. Thus, it is necessary toinvestigate and understand the mechanisms oftransport of trace elements and their complexesto understand their chemical cycles in nature. Anassessment of the impact of an element cannot be

based exclusively on its total concentration. It isoften not possible to determine the concentrationsof the different chemical species that summarizethe total concentration of an element in a givenmatrix. Concerning natural systems, the mobility,transport and partitioning of trace elements aredependent on the chemical form of the elements.Major variations of these characteristics are foundin time and space due to the dissipation andflux of energy and materials involved in thebiogeochemical processes, which determine thespeciation of the elements.

Chemical species present in a given sampleare often not stable enough to be determinedindividually. During the measurement process the

120 SAMPLE PREPARATION – FRACTIONATION

partitioning of the element among its species maybe changed. This behavior can be caused by, forexample, a change in pH necessitated by the ana-lytical procedure. Chemical extraction is usuallyemployed to assess operationally defined metalfractions, which can be related to chemical species,as well as to potentially mobile, bioavailable orecotoxic phases of a sample. It has been gener-ally accepted that the ecological effects of metals,e.g. their bioavailability, ecotoxicology and riskof groundwater contamination are related to suchmobile fractions rather than the total concentra-tion. The use of selective extraction methods todistinguish analytes that are immobilized in differ-ent phases of solids is also of particular interest ingeochemistry for location of mineral deposits.

Atmospheric speciation of metals has, untilrecently, received rather little attention because ofthe real difficulties in measuring even the totalconcentration of most metals in aerosol samples.Fractionation of trace metals in the atmosphere issomewhat different from that applied to speciationin aqueous media. This comes mainly from twoconsiderations: (i) the mechanism of interactionof the biosphere and the atmosphere, (ii) themechanism of transport in the atmosphere. Severalatmospheric parameters are controlled by aerosolparticles, and human health as well as the life ofaquatic and terrestrial ecosystems are also affectedby the toxic metal content of particles. The totalconcentration of the elements in the atmosphericaerosol particles can indicate the sources of thepollutant, while chemical fractionation can be usedto assess the different defined species present in theaerosol sample.

Fractionation is usually performed by a seque-nce of selective chemical extraction techniques,including the successive removal or dissolution ofthese phases and their associated metals. Fraction-ation has been defined as follows: the process ofclassification of an analyte or a group of analytesfrom a certain sample according to physical (e.g.,size, solubility) or chemical (e.g., bonding, reactiv-ity) properties [7]. Fractionation by size is carriedout by the separation of samples into different par-ticle size fractions usually during sampling. Theconcept of chemical leaching is based on the idea

that a particular chemical solvent is either selectivefor a particular phase or selective in its dissolu-tion process. Although a differentiated analysis isadvantageous over investigations of bulk chemistryof solids, verification studies indicate that there aremany problems associated with operational frac-tionation by partial dissolution techniques. Selec-tivity for a specific phase or binding form cannotbe expected for most of these procedures. There isno general agreement on the solutions preferred forthe various components in solids to be extracted,due mostly to the matrix effect involved in theheterogeneous chemical processes [8]. All factorshave to be critically considered when an extractantfor a specific investigation is chosen. Importantfactors are the aim of the study, the type of solidmaterial (sediment, aerosol, fly ash, sewage sludge,harbor mud, street dust, etc.) and the elementsof interest. Partial dissolution techniques shouldinclude reagents that are selective to only one ofthe various components significant in trace metalbinding. Whatever extraction procedure is selected,the validity of selective extraction results primarilydepends on the sample collection and preservationprior to analysis. There is a huge amount of liter-ature on specific research areas, in which appro-priate leaching protocols can be found for a givenproblem. Reviews exist on sediments and combus-tion wastes [9], trace metal speciation in soils [10],and aerosols [11] in the excellent book of Ure andDavidson [12]. The book gives many experimen-tal details as a basis of evaluation of experimentalconditions, so one can reliably find even validatedleaching methods for any special problem.

In many cases it is impossible to determine thelarge numbers of individual species. In practicevarious classes of species of an element canbe identified and the sum of its concentrationsin each fraction can be determined [13]. Suchfractionation is based on different properties ofthe chemical species, such as size, solubility,affinity, charge, and hydrophobicity. Fractionationmay simply involve an actual physical separation,e.g., filtration. Although a direct determination ofthe speciation of an element is often not possible,the available methods can still be applied toget valuable information. An evaluation of the

SEDIMENT SAMPLES 121

environmental impact of an element can sometimesbe done without the determination of the exactspecies and, notwithstanding, fractions that areonly operationally defined. It is also desirable tomeasure the total concentration of the element inorder to verify the mass balance.

The sequential leaching of metals bound to thespecific substrates have been worked out, e.g. forsediments, on the design and study of extractionschemes aiming to investigate the changes in theyearly life cycle of lakes or rivers [12]. Typicalsubstrates are [7]:

• carbonates of calcium, magnesium and iron,which dissolve upon a decrease in pH;

• iron and manganese compounds present inthe sediments and changing their adsorptioncapacities drastically according to the redoxconditions (presence/absence of O2), producingeither FeS or FeO(OH) liberating coprecipitatedor adsorbed metals at every change;

• organic matter present in sediments undergoingslow degradation, thus releasing the incorpo-rated metals;

• silicates and other refractory minerals, whichmight contain a high metal concentration, butwill not, under any environmental conditions,release them to the aquatic environment.

Measurement of such functionally and opera-tionally defined metal fractions will allow someforecasting of metal release from sediments undercertain conditions and represents, therefore, a valu-able tool in natural water management. Harmo-nization of methods should be done because thewide variety of procedures has led to lack ofcomparability of results even for single aquaticsystems. In this paper we present some typicalfractionation patterns for a variety of environmen-tally relevant trace elements in sediments, soils,aerosols and fly ashes as a result of a selec-tive literature survey and our own experiences.Among the solid environmental samples these fourtypes of material were chosen due to their impor-tance in environmental pollution and their effecton human health.

2 SUBSAMPLING, STORAGE,AND PREPARATION OF SEDIMENTSAMPLES

The sampling of environmental pollutants is dis-cussed in detail in Chapter 2.1. In the envi-ronmental sampling the act of sample removalfrom its natural environment can disturb stable ormetastable equilibria. Sampling uncertainty maycontain systematic and random components arisingfrom the sampling procedure. If the test portion isnot representative of the original material, it willnot be possible to relate the analytical result to theoriginal material, no matter how good the analyt-ical method is nor how carefully the analysis isperformed. Sampling errors cannot be controlledby the use of standards or reference materials. Thesame applies to the subsampling. Because of theheterogeneity and complex nature of solids, careshould be taken during sampling, subsampling,preparation, and analysis to minimize changes inspeciation due to changes in the environmentalconditions of the system.

For example, appropriate comparability amongoxide sediment samples collected at different timesand places from a given aquatic system andbetween different systems can be obtained mosteasily by analyzing the fine-grained fraction ofsediment. Some investigations have pointed toa relation between specific surface, grain sizefraction, and the speciation of trace elementsin sediments. Suspended particulate matter sam-pling is mainly carried out by filtration. Suchsamples are of limited utility for studies of thespeciation of elements in solids. In recent years,suspended sediment recovery by continuous-flowcentrifugation has commonly been used to obtainsufficient sample for speciation, up to a few gramsto carry out all the analysis: particle size dis-tribution, identification of mineralogical phases,total element content, and fractionation by sequen-tial extraction. Etcheber and Jouanneau [14] haveprovided a comparative study of suspended par-ticle matter separation by filtration, continuous-flow centrifugation, and shallow water sedimenttraps. Although particles were separated by densityrather than size, the continuous-flow centrifugation


technique was preferred due to its speed and highrecovery rate.

Sample preparation is one of the most importantsteps prior to analysis. The oxidized sediment layercontrols the exchange of trace elements betweensediment and overlying water in many aquaticenvironments. The underlying anoxic layer pro-vides an efficient natural immobilization processfor elements. Significant secondary release of par-ticulate metal pollutants can be obtained from theaccumulated metals as a result of processes suchas [9]:

• desorption from different substrates due toformation of soluble organic and inorganiccomplexes;

• post-depositional redistribution of trace metalsby oxidation and decomposition of organicmaterials due to microbiological activities;

• alteration of the solid–solution partitioning byearly diagenetic effects such as changing thesurface chemistry of oxyhydroxide minerals;

• dissolution of metal precipitates with reducedforms (metal sulfides) generally more insolublethan the oxidized form (surface complexes).

Transformations of metal forms during early dia-genesis have also been successfully studied bysequential leaching. However, many of these stud-ies did not consider that sample preservation tech-niques in trace element speciation studies of oxicsediments and sludges are different from thosewhich should be used for anoxic samples [15].Air and/or oven drying caused major changes insediment equilibrium by converting fractions rel-evant to trace element binding into highly unsta-ble and reactive forms [16]. Drying of sedimentshas also been reported to reduce the quantity ofFe extracted by techniques which remove amor-phous iron oxides (hydroxylamine, CH3COOH,pyrophosphate), suggesting an increase in theoxide crystallinity [17].

In practice, it is usually impossible to relatedata obtained from dried sediments to those thatexisted originally in the field. Such data may evenbe of limited value in comparing the bioavail-able concentrations of trace metals in samplescollected within the same environment. Bartlett

and James [16] found that manganese extractabil-ity changed as a function of storage time. Siev-ing and mixing in order to obtain a representativesample for bioavailability analysis may lead toprecise but inaccurate results. Wet storage of oxi-dized sediments is inadequate because of micro-bially induced changes from oxidizing to reducingconditions in the stored sediments. Extractabilityof the metal with the most insoluble sulfide (Cu)has been reported to decline rapidly during wetstorage [17]. Freezing is usually a suitable methodto minimize microbial activity. Freezing has beenfound to enhance the water solubility of metalsin the order of Mn (8–17 %) > Cu (7–15 %) >

Zn (6–12 %) > Fe (3–7 %) [17]. Storage of an-oxic sediments in a freezer was found to causechanges in the fractionation pattern of various met-als studied. It has been found that a double wallsealing concept, i.e. an inner plastic vial with thefrozen sediment contained under argon in an outerglass vial, is suitable. However, it seems to beimpossible to totally avoid changes in the in situchemical speciation of trace elements found innature, so samples should be extracted immediatelyto avoid changes in speciation during storage [16].

3 SUBSAMPLING, STORAGE, ANDPREPARATION OF SOIL SAMPLES

Seasonal [18, 19] and spatial variability [20, 21]are known to influence significantly the results ofsequential extraction schemes in soils. Soil man-agement practices (fertilizing, liming, sludge appli-cation) may cause significant seasonal changes inmobile fractions, but also natural seasonal vari-ation of extractable metals in extensively usedforest soils or undisturbed ecosystems may occuras well [22, 23]. Seasonal variation of extractablemetals is an inherent process that is at least assignificant as spatial variability [24]. Soil prop-erties may vary considerably on a micro-scaleof about 1–100 mm. Thus, metal solubility andextractability may be affected either directly bymicro-inhomogeneity of the total metal contentsor by simultaneous variation in soil properties(pH, cation exchange capacity (CEC), organic mat-ter, mineral composition and soil texture) [23].

AEROSOL SAMPLES 123

It has been concluded that the mobility ofmetals may frequently be underestimated whenassessed by chemical extraction of disturbed,homogenized, and sieved soil samples of well-aggregated acidic soils, particularly when anthro-pogenically polluted.

Sample preparation generally involves the fol-lowing steps: (1) drying or rewetting, (2) homo-genizing and sieving, (3) storage and, occasionally,(4) grinding. Usually, soil samples are air driedprior to extraction. Although changes in theextractability of some elements have already beenreported, this problem has only recently receivedmore attention [25, 26]. Air drying prior to extrac-tion is a standard procedure but leads to anincreased extractability of Fe and Mn, whereasother metals are more or less unaffected [25, 26].Several authors have identified possible mecha-nisms of these changes in metal extractability uponair drying. Drying of samples prior to the deter-mination of mobile metal fractions usually resultsin unrealistically large amounts of extractable Mn,Fe, Cu, and Zn, and underestimation of Ca, Mg,K, and probably Co, Ni, and V. The changes inextractability upon air drying are related to soilproperties, i.e. pH and organic matter content, andto the initial soil moisture conditions.

Although homogenizing and sieving are essen-tial steps in performing representative and repeat-able soil analysis, these procedures suffer fromsome drawbacks. The effects of structure disturb-ing soil sampling are obviously reinforced, thuscreating new surfaces for reactions with metals inthe solute phase, giving rise to adverse readsorp-tion or desorption processes during metal extrac-tion [27]. Homogenization of soil material fromdifferent horizons may result in erroneous changesin pH and carbonate content of the fine earth. As aconclusion, sample storage seems to be generallyless critical to the analysis of extractable metalfractions than does air drying, but it is likely toenhance the effects of air drying in the case ofredox-sensitive elements. Occasionally, soil sam-ples are ground prior to extraction. This procedurecauses physical breakdown of soil microaggre-gates, thus potentially altering the extractabilityof metals from soil samples [28]. The exposure

of fresh surfaces may, depending on soil proper-ties, increase the extractability of some metals butpotentially may also cause readsorption of metalsduring the batch process.

4 SUBSAMPLING, STORAGE,AND PREPARATION OF AEROSOLSAMPLES

Metals are transported in the atmosphere primarilyon aerosols which can be removed by wet and drydeposition process. The deposition, transport, andinhalation processes are controlled predominantlyby the size of the atmospheric aerosols. Thus, theprimary type of fractionation is the aerosol size dis-tribution. Once deposited, however, chemical spe-ciation in terms of both the dissolved/particulatedistribution of the metals in precipitation and theinorganic or organic complexes which the metalmay form, plays an important role in control-ling the environmental impact of atmosphericallydeposited metals.

The atmosphere is an important vector of globalmetal transport between regions, and globally,from land to sea, and from sea to land. Directatmospheric deposition makes only a minor con-tribution to the total metal contents of the litho-sphere because of the large reservoir of thesemetals in soils and rocks. The impact of the atmo-spheric inputs on lake biogeochemistry however,is strongly dependent on both the physical andchemical forms in which metals enter the naturalwater system. The transport processes are impor-tant both spatially and temporally and occur viaseveral atmospheric activities.

Aerosol particles are partly emitted into theatmosphere from sources on the surface (primaryparticles), while others come into being in the airby gas–particle conversion (secondary particles).These particle generally have sizes (diameters)<1 µm and are called fine particles. In contrast,surface dispersion creates particles of diameter>1 µm, termed coarse particles. Particles dueto combustion (fly ashes) can be found in bothsize intervals. The particles of different originare of different chemical composition and alsoof different chemical forms and physical state.


Particle size exercises a strong control on residencetime in the atmosphere and on particle dispersion.Large particles are rapidly removed near the sourceby gravitational sedimentation (typical residenttime in the atmosphere of hours to a few days),while small particles have a considerably longerresidence time in the atmosphere and are muchmore efficiently transported. Close to the sourcethe composition of the aerosol will be stronglyrelated to the parent material.

The chemical composition of the particulateemissions is less well known than the chemicalcomposition of the particles in ambient air. Inor-ganic minerals are mainly emitted from processes,particularly from cement, iron and steel industry,and as fly ash from coal combustion. Organic com-pounds and soot are mostly released from smallcombustion sources, mobile sources, and from pro-cesses associated with petroleum extraction andrefining. The ratio of organic to elementary carbonin these emissions is rather variable, from <1 in thecase of emissions from diesel engines, up to 5–10for low-calorific fuels such as lignite, peat and fire-wood. The anthropogenic emissions can be dividedin three main areas such as (i) incomplete com-bustion, formation of soot and associated organiccompounds (COC) (including small-scale residen-tial combustion, both solid and liquid fuels, and

internal combustion engines), (ii) fly ash and par-ticles from the fuel’s content of inorganic mineralmatter, (iii) industrial processes.

Particles can be generated from natural andanthropogenic sources. In general, it is esti-mated that the annual total amount of particlesfrom these sources is about 3000 million tonnesand 400 million tonnes, respectively [29]. Particlesfrom natural sources are overwhelmingly coarseparticles, from wind erosion, sea-spray formationand similar processes. Anthropogenic emissions,in contrast, contribute about 60 % to the total fineparticle mass in the atmosphere. Behind these esti-mates lie large uncertainties in terms of sourceassessment, speciation, and characterization of theatmospheric particles, not to mention the differ-ent lifetimes of particles. A rough estimate of theircontribution to particulate aerosol mass on a globalscale is given in Table 3.3.1.

These emission and formation figures are ratheruncertain. It should also be taken into account thatthe residence times of particles from the respectivesources are very variable. The difficulty in assess-ing the emissions of sea-salt particles from the seasurface, and of soil and desert dust from wind ero-sion, is partly due to the rapid sedimentation anddeposition of these particles. Coarse particles gen-erally have short residence times, typically of the

Table 3.3.1. Estimated contributions to the global atmospheric particulate mass(Tg = terragram) [29].

Source Annual emission or production (Tg year−1)

Range Best estimate

NaturalWind erosion 1000–3000 1500Sea salt 1000–10 000 1300Volcanoes 4–10 000 30Biological primary particles 26–80 50Forest fires 3–150 20Inorganic secondary particlesa 100–260 180Organic secondary particlesb 40–200 60

AnthropogenicDirect emissions 50–160 120Inorganic secondary particlesa 260–460 330Organic secondary particlesc 5–25 10

aOxidation of sulfur dioxide, reduced sulfur compounds and nitrogen dioxide, uptakeof ammonia.bMainly photochemical formation of particulate matter from isoprene and monoterpenes.cPhotochemical formation of particulate matter from anthropogenic emissions of VOCs.

AEROSOL SAMPLES 125

order of few days. Fine particles, such as the inor-ganic and organic secondary particles, have atmo-spheric residence times of 1–2 weeks.

The frequently used method of relating anelement in atmospheric aerosols to its source isto calculate enrichment factors (EF) by employingan indicator element. For crustal aerosols Alis normally used as the indicator element. Theenrichment factor can be calculated using theformula as

EF crust = (C(i)/C(Al))aerosol

(C(i)/C(Al))soil

Those elements that have EF values between 1 and10 are usually crustal in origin, while elementswith EF values in the range of 10–5000 aregenerally emitted by anthropogenic sources [30].

Aerosol sampling is usually carried out eitherby collection devices (impactors) or by fibrousor membrane filters. The operation principle ofimpactors is based on the fact that particles havemuch larger inertia than gas molecules, whichmakes their separation in a fluid in motion possible.Since aerosol particles may have different formsand density, impactor data are generally given foraerodynamic particle diameter. Usually cascadeimpactors are used which collect the particles indifferent size ranges. An important characteristicof a given impactor stage is the particle diameter(cut-off diameter) where the collection efficiencyis equal to 50 %. The main characteristics ofa Berner impactor [31], widely used in recentaerosol studies in Europe and in the USA, aresummarized in Table 3.3.2. The flow rate is1.9 m3 h−1 at 20 ◦C with an exhaust pressure of150 hPa. The impactor consists of eight stages,

Table 3.3.2. Main characteristics of the Berner-typecascade impactor [31].

Stage no. Cut-offdiameter

(µm)

Slitdiameter

(mm)

Numberof slits

9 16 15.9 18 8.0 5.0 87 4.0 2.7 136 2.0 1.2 365 1.0 0.70 534 0.50 0.60 303 0.25 0.42 312 0.125 0.30 631 0.0625 0.25 128

while a prestage (No. 9) excludes the sampling ofparticles with a diameter of 16 µm.

Aerosol sampling by filtration is based onthe passage of air by a pump through a filtersubstrate placed in a suitable filter holder. Fibrousfilters consist of mats of fibers made generallyof glass, quartz, or cellulose. Membrane filterscontain small pores of controlled size, and theyare usually composed of thin films of polymericmaterials [32]. By using any kind of filters, inthe absence of electric forces, larger and smallerparticles are captured from the air, pumped throughthe filter material, by impaction and diffusion,respectively. The characteristics of different typesof fibrous and membrane filters widely used forsampling aerosol particles for subsequent chemicalanalysis are summarized in Table 3.3.3.

The surface reactivity is important for chemicalanalysis, since some filters react with atmospherictrace gases resulting in sampling artifacts. Thefilters require pretreatment before sampling whichis usually done by acid wash to remove alkalinesites. Teflon, quartz, and Nuclepore filters have

Table 3.3.3. Properties of filters used for particulate sampling with a face velocity of 10 cm s−1 [32].Note that the efficiencies refer to particles with diameters above 0.03 µm.

Filter Composition Density(mg cm−2)

Surfacereactivity

Efficiency(%)

Teflon Polytetrafluoroethene 0.5 Neutral 99Whatman 41 Cellulose fiber 8.7 Neutral 58Whatman GF/C Glass fiber 5.2 Basic (pH 9) 99Gelman Quartz Quartz fiber 6.5 pH 7 98Nuclepore Polycarbonate 0.8 Neutral 93Millipore Cellulose acetate/nitrate 5.0 Neutral 99


been found to give the best substrates for chemicalanalysis. Teflon and Nuclepore filters are usedfor inorganic substances, while quartz filters arecommonly applied for sampling of organic species.However, in the latter case it should be takeninto account that quartz filters can adsorb organicvapors. This can be checked by using a quartzbackup filter in the air stream to correct theconcentration in the aerosol phase determined onthe first filter. It is suggested to use a second filterafter a Teflon filter which does not adsorb organicvapors [33]. Consequently, before sampling, itis necessary to desorb volatile organics fromthe quartz filters at high temperature. Aerosolsampling is usually carried out by filters with adiameter of several centimeters. So the samplingrate is typically some m3 day−1 which gives theamount of aerosols sufficient for the majority ofchemical analysis. However, in some measurementprograms a large mass of aerosols is needed andhigh volume of aerosols are sampled. The flow ratein high volume sampling is about 60 m3 h−1 withfilters of diameter around 25 cm or larger. Glassfiber filters can fulfill this requirement.

5 SUBSAMPLING, STORAGE, ANDPREPARATION OF FLY ASH SAMPLES

Fly ash is the fine particulate waste materialthat remains after incomplete combustion. It isproduced in massive quantities, mainly by fossilfuel-based power plants and waste incinerators.Fly ash contains high levels of potentially toxicchemicals, but these are often strongly boundto the particulate matrix and only a fractionposes an immediate threat. Various tests canbe used to determine the amounts that willleach from the fly ash matrix (for instancewith rainwater) and thus become available fortransport to soils, rivers, or groundwater andeventual uptake by organisms. For environmentalrisk assessment the potentially hazardous fraction,i.e. the fraction that can be leached from thematrix under typical environmental conditions,needs to be determined [34]. Submicrometer flyash particles may be emitted from power plantsmoke stacks in spite of electrostatic filters, and

thus enter the environment. Another reason forstudying leaching properties from fly ash is toevaluate its potential for re-use in constructionmaterials [35].

Depending on the aim of the measurement,leaching tests can be subdivided into relativelyquick compliance tests to verify whether a materialmeets certain legal criteria (e.g., the Europeanleaching test) [36] and more extensive materialcharacterizations (e.g., pH-stat tests at a range ofpH values) [37] or sequential leaching tests [38].The outcome of such measurements often dependsrather critically on the experimental conditionsand over the years several measurement protocolshave been developed by standardization agenciesin various countries. Unfortunately, the results ofleaching tests carried out in different countriescannot be compared as long as the test methodshave not been harmonized at the internationallevel. For the interpretation of leaching test resultsit is necessary to understand the leaching processesin terms of the physico-chemical properties of theelement species and the matrix. The results can beused to link metal leaching behavior to differenttypes of binding. Particle size of fly ashes is animportant parameter since it controls the specificsurface area of the combustion material which,evidently, is directly proportional to the amountof toxic metals adsorbed. Hence separation bysieving should be done before leaching procedures.Due to the heterogeneous nature of different flyash samples, a semi-homogeneous portion can beobtained by separating and using the close particlesize fractions. In some cases an agate ball is placedinto the bottle and, before removing a test portionfrom the sample, a strong shaking of the bottle forsome minutes is advised. Storage of fly ash sampleis usually carried out in a closed bottle.

6 SEQUENTIAL EXTRACTIONTECHNIQUES

Sequential extractions have been applied usinga series of extractants with increasing extractioncapacity, and several schemes have been developedto determine species in solid samples. Althoughinitially thought to distinguish some well-defined

SEQUENTIAL EXTRACTION TECHNIQUES 127

chemical forms of trace metals [39], they insteadaddress operationally defined fractions [40]. Theselectivity of many extractants is weak or not suf-ficiently understood, and it is questionable whetherspecific trace metal compounds actually exist andcan be selectively removed from multicomponentsystems. Due to varying extraction conditions, sim-ilar procedures may extract a significantly differ-ent amount of metals. Concentration, operationalpH, liquid/solid (L/S) ratio, and duration of theextraction process affect considerably the selectiv-ity of extractants. The conventional approach ofequilibration during a single extraction step is theshaking or stirring of the solid phase–extractantmixture. There is no general agreement on thesolutions preferred for the extraction of variouscomponents in solid samples, due mostly to thematrix effects involved in the heterogeneous chem-ical processes [8]. The aim of the study, the typeof the solid materials, and the elements of inter-est determine the most appropriate leaching solu-tions. Partial dissolution techniques should includereagents that are sensitive to only one of the vari-ous components significant in trace metal binding.

In sequential multiple extraction techniques,chemical extractants of various types are appliedsuccessively to the sample, each follow-up treat-ment being more drastic in chemical action ordifferent in nature from the previous one. Selec-tivity for a specific phase or binding form cannotbe expected for most of these procedures. In prac-tice, some major factors may influence the successin selective leaching of components, such as:

• the chemical properties of the leaching solu-tions chosen;

• experimental parameters;• the sequence of the individual steps;• specific matrix effects such as cross-contamina-

tion and readsorption;• heterogeneity, as well as physical associations

(e.g. coatings) of the various solid fractions.

All of these factors have to be critically consid-ered when an extractant for specific investigation ischosen. Fractions of sequential extraction schemescan be:

(i) Mobile, exchangeable elements: this fractionincludes the water-soluble and easily exchange-able (unspecifically adsorbed) metals and easilysoluble metallo-organic complexes. Most of therecommended protocols seek to first displace theexchangeable portion of metals as a separate entity.Chemicals used for this fraction fall commonly inone of the following groups [27]:

• water or highly diluted salt solutions (ionicstrength <0.01 M, e.g. MgCl2);

• neutral salt solutions without pH buffer capacity(e.g. CaCl2, NaNO3);

• salt solutions with pronounced pH buffer capac-ity (e.g. NH4OAc).

(ii) Elements bound to carbonates: to dissolvetrace elements bound on carbonates buffer solu-tions (e.g. HOAc/NaOAc; pH = 5) are commonlyused. Zeien and Brummer [40] have proposed todissolve carbonates by adding equivalent amountsof diluted HCl to 1 mol L−1 NH4OAc/HOAcbuffer, addressing specifically adsorbed and sur-face occluded trace element fractions of soils with>5 % m/m carbonates.

(iii) Elements bound to easily reducible fractions(bound to Fe/Mn oxides): NH2OH · HCl at pH 2is generally used but procedures differ in minoroperational details such as S/L ratios, treatmenttime, interstep washing procedure.

(iv) Elements bound to easily extractable organ-ics: NaOCl or Na4P2O7 are used mostly.

(v) Elements bound to moderately reducible ox-ides: NH2OH · HCl/HOAc or NH4Ox/HOx mix-tures are mainly applied.

(vi) Elements bound to oxide and sulfide frac-tions: H2O2/NH4OAc is used most frequently.

(vii) Elements bound to silicates (residual frac-tion): this fraction mainly contains crystallinebound trace metals and is most commonly dis-solved with concentrated acids and special diges-tion procedures, i.e. strong acid mixtures areapplied (HF/HClO4/HNO3) to leach all remain-ing metals.


As can be seen, a wide range of extractionprocedures is readily available for different metalsand variations of the extraction conditions areutilized due to varying solid sample composition.The following parameters have to be consideredwhen designing an adequate extraction procedure:

Extractants: chemical and physical interfer-ences both in extraction and analysis steps, respec-tively.

Extraction steps: selectivity, readsorption andredistribution processes. As extractants the follow-ing chemicals are commonly used:

• for cation changing conditions (e.g. NH4Cl);• for anion changing conditions (e.g. NaNO3);• acids (acidity increases with each extraction step

from unbuffered salt solutions (NH4NO3) tostrong acids (HNO3/HClO4)).

If the single extractants for the different stepsare chosen with respect to their ion exchangecapacity or reduction/oxidation capacity, each stephas to be designed individually following specialconsiderations [23].

Concentration of the chemicals: the efficiencyof an extractant to dissolve or desorb trace metalsfrom solid samples will usually be increasedwith increasing concentration or ionic strength.Thermodynamic laws predict the efficiency of anextractant to dissolve or desorb trace metals fromsolid samples.

Extraction pH: extractants with a large buffer-ing capacity or extractants without buffer capacitycan be used.

Liquid/solid (L/S) ratio and extraction capacity:the relative amount of extractant added to the solidsamples has various influences on the results. If,over a sufficiently wide L/S ratio, the capacityof the extractant to dissolve a metal fractionexceeds its total amount present in the solidsample, then the metal concentration in the extract(mg L−1 extract) will decrease with an increase inL/S ratio. However, the total amount (mg kg−1)extracted will be constant with increasing L/Sratio. Nevertheless, as sediments, aerosols, flyashes, and soils are multiphase/multicomponentsystems, dissolution of other compounds due to

the nonselectivity of the extractant may confusethis behavior.

Extraction time and batch processes: the effectof extraction time is related to the kinetics ofthe reactions between solid sample and leachant.Extractions may be predominantly based on eitherdesorption or dissolution reactions. For desorptionof metal cations from heterogeneous soil systemsSparks [41] has identified four rate-determiningsteps, namely (i) diffusion of the cations in the(free) bulk solution, (ii) film diffusion, (iii) particlediffusion, and (iv) the desorption reaction. Accord-ingly, the rates of most ion exchange reactions arefilm and/or particle diffusion controlled. Vigorousmixing, stirring, or shaking significantly influencesthese processes. Film diffusion usually predomi-nates with small particles, while particle diffusionis usually rate limiting for large particles. For min-eral dissolution, essentially three rate-controllingsteps have been identified, namely (i) transport ofsolute away from the dissolved minerals (trans-port controlled kinetics), (ii) surface reaction-controlled kinetics where ions are detached fromthe surface of minerals, and (iii) a combinationof both. Mechanical actions, e.g. stirring or shak-ing, increase the rate of transport-controlled reac-tions, while they do not affect surface-controlledreactions. Shaking and other batch processes mayenhance the dissolution of readily soluble saltseffectively, but are unlikely to affect the dissolu-tion rate of less soluble minerals.

Extraction temperature: within the normalrange of extraction temperatures (20–25 ◦C orroom temperature), the effect of temperature onextractability of the elements is usually small, butit has to be considered for interpretation of smalldifferences [42].

Finally, the whole procedure has to be opti-mized with regard to selectivity, simplicity, andreproducibility.

7 STANDARDIZED SEQUENTIALEXTRACTION PROCEDUREPROPOSED BY BCR

Sequential extraction schemes have been devel-oped during the past 30 years for the determina-tion of chemical binding forms of trace metals

EXTRACTION SCHEME FOR SEDIMENT SAMPLES 129

in sediment and soil samples. The lack of uni-formity of these schemes, however, did not allowthe results so far to be compared worldwide orthe procedures to be validated. Indeed, the resultsobtained by sequential extraction are operationallydefined, i.e. the ‘forms’ of metals are defined bythe procedure used for their determination. There-fore, the significance of the analytical results isrelated to the extraction scheme used. Anotherproblem was the lack of suitable reference materi-als. Thus, standardization of leaching and extrac-tion schemes was required with the preparation ofsediment and soil reference materials which werecertified for their contents of extractable trace ele-ment with single and sequential extraction proce-dures [43]. The Community Bureau of Reference(BCR, now Standards, Measurements, and TestingProgramme) has recently launched a programmeto harmonize sequential extraction schemes forthe determination of extractable trace metals insediments [44]. BCR has proposed a standard-ized three-stage extraction procedure (BCR EUR14763 EN), which was originally developed forthe analysis of heavy metals in sediments [45].This procedure is currently used and evaluatedalso as an extraction method for soils [46]. So far,the BCR procedure has been successfully appliedto a variety of sludge [47], sediment [48], andsoil [46] samples. The BCR scheme was recentlyused to certify the extractable trace element con-tents of a certified reference material (CRM 601,IRMM). Although this procedure offers a tool forobtaining comparable data, poor reproducibilityand problems with lack of selectivity have stillbeen reported [46, 49]. Various research groupshave used this technique and found partially dis-crepancies when applying the scheme. The sameextraction scheme has also been used for the deter-mination of extractable elements in soils [50].

8 SEQUENTIAL EXTRACTIONSCHEMES APPLIED TO SEDIMENTSAMPLES

Main mineralogical components of sediments,which are important in controlling their metalconcentrations, are hydrous oxides of iron and

manganese, organic matter, and clay. The degreeof interaction between sediment samples andextractant solutions can be altered by changesin experimental parameters such as reagent con-centration, final suspension pH, L/S ratios, tem-perature, contact time, and intensity of mixing.Recently, researchers have tended to use sim-ilar extraction protocols, mostly by adaptingor modifying the scheme of Tessier et al. [51].Salomons and Forstner [52] have used sequentialextraction techniques to determine the chemicalassociations of heavy metals with specific sed-imentary phases. They distinguished five majormechanisms for metal accumulation on sedimen-tary particles: (1) adsorptive bonding to fine-grained substances, (2) precipitation of discretemetal compounds, (3) coprecipitation of metalswith hydrous iron oxides and manganese oxidesand carbonates, (4) association with organic com-pounds, and (5) incorporation in crystalline mate-rial. It was pointed out that the standard extractionmethod should be relatively simple, in order tomake routine analysis of large numbers of sedi-ments possible.

Tessier et al. [53] have collected sediment sam-ples from streambeds in an undisturbed watershedin eastern Quebec (Gaspe Peninsula). The sedimentsamples were separated into eight distinct parti-cle size classes in the size range from 850 µmto <1 µm by wet sieving, gravity sedimentation,or centrifugation. Each sediment subsample wasthen subjected to a sequential extraction proce-dure designed to partition the particulate heavymetals into five fractions. It was one of the firstfractionation process by particle size and chemicalbonding for investigation of sediments. A simul-taneous sediment extraction procedure for lowcarbonate sediments, which partitions sediment-bound trace metals (Fe, Mn, Zn, Cu, and Cd)into easily reducible (associated with manganeseoxides), reducible (associated iron oxides) andalkaline extracted (bound to organic) metal wasinvestigated. This method was compared to thesequential extraction procedure based on the workof Tessier et al. [51]. Both methods showed goodagreement for the partitioning of Zn and Cdamong the easily reducible, reducible, and organic


components of the sediment. The two methods alsoshowed the same general distribution of Mn, Fe,and Cu among the three sediment components,although concentrations of metals recovered by thetwo methods differed; less Mn and Fe, and moreCu were recovered from sediments by sequentialextraction over the simultaneous procedure. Thelower recovery of Mn was, in part, attributed to theloss of this metal in the ‘in between’ reagent rinsesrequired in the sequential extraction procedure.Greater recovery of Cu by the sequential extrac-tion method might be due to the pretreatment ofthe sediment with strong reducing agents prior tothe step used for liberating organically bound met-als. Advantages of simultaneous extraction oversequential technique included rapid sample pro-cessing time (i.e. the treatment of 40 samples perday versus 40 samples in 3 days), and minimalsample manipulation.

Sediments are the ultimate sinks for pollutants.Before these sediments become part of the sedi-mentary record (deeply buried), they are able toinfluence the composition of surface waters. Thesediments can be divided in two sections: an oxicsurface layer and an anoxic sediment. In anoxicsystems when sulphide is present, Zn, Cd, andCu are likely to be present as sulfides. Remo-bilization of the deposited sediments is possiblewhen the overlying surface water changes (pH andcomplexing agent). In addition, changes in the sur-face water composition may enhance or preventthe removal of dissolved trace metals by partic-ulates and subsequent removal by sedimentation.Remobilization also occurs when sediments arebrought from an anoxic to an oxic environment astakes place during dredging and disposal on land.Salomons et al. [54] have reviewed the processesaffecting trace metals in deposited sediments. Thesediment–water system could be divided in threeparts: the oxic layer, the anoxic layer and theoxic–anoxic interface. Available data showed thattrace metals such as Cu, Zn, and Cd occurredas sulfides in marine and estuarine anoxic sed-iments. Calculations showed that organic com-plexation was unlikely and the dominant specieswere sulfide and bisulfide complexes. Cr and Aswere probably present as adsorbed species on the

sediments. Changes from an anoxic to an oxicenvironment, as occur during dredging and landdisposal of contaminated sediments, might causea mobilization of some trace metals. The chem-ical forms of many elements in the sedimentsof St. Gilla Lagoon (Sardinia, Italy) have beenevaluated [55]. Five fractions were separated fromsediments by sequential chemical extraction. Themetals in each fraction were determined by thetotal reflection X-ray fluorescence (TXRF) tech-nique. Both principal and trace element distribu-tions in the sequential phases were discussed interms of pollution sources, metal transport, anddeposition/redeposition in air-dried sediments. Theuse of a sequential extraction procedure couldbe an effective method for comparative studiesbetween natural and contaminated areas, as wellas between areas subjected to different chemicalstresses. The results showed that in the examinedarea the lithogeneous fraction was the most rele-vant for total metal content. However, under oxi-dizing conditions among the ‘mobile’ fractions, thereducible fraction proved to be the most importantsink for Zn and Pb, the oxidizable fraction wasonly relevant for Cu at almost natural level.

Availability of heavy metals depends greatly onthe properties of particle’s surface, on the kind ofstrength of the bond and external conditions suchas pH, Eh, salinity and concentration of organicand inorganic complexation agents. Most particlesurfaces have an electrical charge, in many casesa negative one. In solutions, an equivalent numberof ions of opposite charge will gather around theparticle, thus creating an electric double layer.The surface charge is strongly affected by pHand the composition of surface. Especially hydrousoxides of Fe, Al, Si, Mn and organic surfaces (e.g.functional amino and carboxyl groups) participatein the H+ transfer. Lattice defects of clay mineralsand the adsorption of ions also contribute tosurface charges. The sorption process can bephysical or chemical adsorption as well as sorptionby ion exchange. Physical adsorption on theexternal surface of a particle is based on the vander Waals forces or relatively weak ion–dipole ordipole–dipole interactions (about 1 kcal M−1).

EXTRACTION SCHEME FOR SEDIMENT SAMPLES 131

Two sequential extraction schemes (a modifiedTessier procedure [51] with five steps and thethree-step protocol developed by BCR [45]) wereapplied to four sediment samples with differentheavy metal contents [56]. The results obtained forpartitioning of Cd, Cr, Cu, Ni, Pb, and Zn showedthat the metal distributions with the two proce-dures were significantly different. With the three-step protocol significant amounts of all the heavymetals were extracted with the oxidizing reagent,whereas with the modified Tessier procedure thenonresidual metals were distributed between thesecond, third, and fourth fractions (CH3COOH-acetate buffer (pH = 5), reducing, and oxidiz-ing reagents, respectively). The residual fractionobtained applying the three-step procedure wasin general higher than that obtained using thefive-step procedure, except for Cd. The three-step sequential extraction protocol [45] has beenevaluated with regard to total recovery, repro-ducibility, selectivity of extractants, and extentof phase exchanges or redistribution of met-als during the extraction [57]. Model sedimentsof known composition were prepared consist-ing of humic acid and natural minerals such askaolin, quartzite, and ochre. It was shown thatthe chemistry of a metal could be a more impor-tant parameter than its actual phase location inthe sediment in determining its response to theextractants.

The reproducibility of Tessier’s extractions [51]and the total content of Cd, Cr, Cu, Fe, Mn, Pb, Zn,and Ca in river sediment have been evaluated [58].The accuracy of the dissolution procedures wasestimated using a reference material, BCR 145.None of the methods applied proved optimal forall the metals determined. The concentrations ofmetals extracted by the various reagents werecharacterized by good reproducibility on speciesbonded to the carbonates, to iron and manganeseoxides, and in the residual fraction; precision waslower in the other cases. The sequential procedurealso showed a satisfactory mass balance.

Sequential extraction procedures offer the ad-vantage of simulating, to a certain extent, the var-ious natural environmental conditions. Recently,

investigations on bottom sediments of Lake Bal-aton, Hungary, rivers in its catchment area, andharbors were extensively carried out in spring,summer and fall [59]. A modified sequentialextraction procedure of BCR [45] was applied asa four-step sequential leaching for determinationof the distribution of seven elements [59, 60].The fractions were (1) exchangeable metals andmetals bound to carbonate, (2) metals bound toFe/Mn oxides, (3) metals bound to organic mat-ter and sulfide, and (4) acid-soluble metals. Thecritically examined three-step sequential extrac-tion method [45] should be used to compare thedata produced by different laboratories, but thestrong acid-soluble fraction of elements could addmore valuable information. The four-step methodallowed a deeper understanding of the associa-tion of elements with the compounds of sediments.The sequential leaching protocol was sufficientlyrepeatable and reproducible for application infractionation studies. The amounts of elementsremoved correlated well with those determined bypseudo-total, acid digestion of sediment. It hasbeen found that elements concentrated mainly inthe acid-soluble fraction indicating no serious pol-lution in the lake sediment. Results were com-pared to the sediment quality values (SQVs) andsediment background values (SBVs) [61]. SQVswere summarized for 22 metals and metalloidsfrom different area in the USA, Canada, TheNetherlands, Norway, Australia, New Zealand, andChina. Globally, SQVs for metals and metalloidsvary over several orders of magnitude. RegionalSQVs can be useful as an initial step in a sedimenthazard/risk assessment. The SQVs are numerical,based on total dry weight concentrations of sedi-ments collected from more than 50 different sam-pling points all over the world (North America,Asia, Europe, and Australia). The SBVs for fresh-waters include background values summarized forlakes and streams and the global shale averagevalues. Data clearly showed that the average con-centrations of elements have been found to be lessthan the SQVs and other background data for soils.This means that the sediment is not polluted and,after removing from the bottom, its disposal on thesoil is feasible.


9 SEQUENTIAL EXTRACTIONSCHEMES APPLIED TO SOILSAMPLES

To assess metal mobility of trace elements insoils on different time scales and sampling sites,a wide range of extraction schemes have beenemployed [26, 27]. These methods vary with res-pect to the extraction conditions: chemical natureand concentration of leaching solutions, solu-tion/soil ratio, operational pH, and extractiontime [27]. If more than one leaching solution isused, differences occur due to variation in theextraction sequence. The most critical steps aresampling, sample preparation and the selectivityand accuracy of the leaching procedure [26]. Asfor total metal concentrations, spatial heterogene-ity, as well as seasonal variation of extractablemetal fractions may bias the results. The use ofcorrelation coefficients for choosing extractants forassessment of plant availability of elements needsconsideration. For extraction of the exchangeablefraction, almost all possible combinations of majorcations with either Cl−, NO3

− or acetate havebeen used, with concentrations ranging between0.05 and 1 mol L−1, and pH in the neutral range.The solution/soil ratios were varied from 4 : 1 to100 : 1, the extraction times between 30 min and24 h. In ideal systems, the relative exchangeabil-ity of trace metals is determined by the affin-ity of the exchanging cation for the solid phase.This affinity increased with increasing valencyand decreasing radius of the hydrated cation.Although heterogeneous soil systems may deviatefrom this ideal behavior, the selectivity of soilsfor cations was frequently observed to increaseaccording to Na < K < Mg < Ca. Consequently,under comparable conditions, e.g. concentration,extraction time, soil/solution ratio, the efficiencyof cations to exchange trace metals increasedaccording to Li < Na < K < Mg < Ca < Ba (<

La). Therefore, salts of Ca and Ba were regardedas most effective and selective agents in extract-ing exchangeable trace metals. Unfortunately, bothcations may cause serious background problems(interferences) during determination of Pb andother trace metals. Usually, this can be resolved

only by dilution of the extracts by >1 : 10 priorto measurement decreasing the detection limit bythe same ratio. For that reason, the use of eas-ily volatilizable salts, i.e. MgNO3 or NH4NO3 hasbeen proposed [40, 51]. Compared to 0.1 mol L−1

solutions of CaCl2 or BaCl2, 1 mol L−1 NH4NO3

in 2.5 : 1 ratio was found to extract about equalamounts of Al, Fe, Mn, Ni, Pb, V, and Zn, andCd was less efficiently extracted by 1 mol L−1

NH4NO3. This can be explained by a more effec-tive extraction with CaCl2 through the forma-tion of chlorocomplexes with Cd and Cu. Otheranions frequently used are either acetate or nitrate.At equal concentrations, the complexing abilityincreases in the order nitrate < chloride < acetate.The selectivity for extraction of the unspecificallysorbed (exchangeable) fraction should decrease inthe same order.

The specifically sorbed fraction is explicitlyaddressed by only a few methods. In addition toother differences, the wide range of cations usedsuggests that most methods do not address any‘specifically sorbed’ fraction, and, probably evendo not extract the same operationally defined sol-ubility group. To extract specifically sorbed tracemetals, Pb(NO3)2 seems to be most adequate, dueto its low pK (7.7) and large atomic radius, andit is effective in displacing other trace metals, i.e.Cd (pK = 10.1), Ni (pK = 9.9), Co (pK = 9.7),Zn (pK = 9.0) and Cu (pK = 7.7), with smalleratomic radius than Pb. Pb(NO3)2 was found toextract less metal than acetic acid, probably indi-cating that the latter was more specific [63]. Unfor-tunately, those trace metals that are constituentsof the extractants cannot be determined. There-fore, Zeien et al. [40]. proposed 1 mol L−1 NH4Acand 1 mol L−1 NH4NO3 in sequence to extractan operationally defined fraction under optimizedanalytical conditions by using only one cation(NH4

+) and decreasing the pH throughout theextraction sequence.

Among the extractants most frequently used todissolve trace metals bound to carbonates are acidssuch as HCl and acetic acid (pH = 3–3.5), buffersolutions of HAc/NaAc (pH = 5) and the bufferingcomplexing agent Na2EDTA at pH = 4.6. Any ofthese extractants seems to have some potential

EXTRACTION SCHEME FOR SOIL SAMPLES 133

to extract carbonates from soils, but is probablyneither effective for quantitative dissolution (i.e.CH3COOH buffers), nor selective (cold dilutedacids, i.e. HCl), or both, i.e. Na2EDTA [63].

For extraction of organically bound trace met-als, various approaches have been used, e.g. theirrelease by oxidation or dissolution of the organicmatter, or through addition of competing, e.g.complexing or chelating, ligands. Among the oxi-dizing extractants, H2O2, either purely or com-bined with HNO3 or NH4Ac, extracted moretrace metals from soils than NaOCl [63], i.e.from Fe/Mn oxides. K4P2O7 and Na4P2O7 havebeen reported to dissolve organic matter by dis-persion and to efficiently complex the releasedmetals [64]. Accordingly, the variation in extrac-tion parameters with concentrations between 0.1and 1 mol L−1, solution/soil ratios between 10 : 1and 100 : 1, and extraction times from 1 h to24 h indicate that results obtained by differ-ent procedures are hardly comparable and arelikely to extract non-organically bound trace met-als to a varying extent. K4P2O7 was found toextract more metals when used before Mn oxideextraction with NH2OH · HCl, while the latterextractant has little effect on the organicallybound fraction.

As an alternative to pyrophosphate salts, someprocedures employ NaOH [39] or NaOH/EDTAmixture to extract organically bound trace met-als by dissolution of organic matter. The selec-tivity of these methods is considered low, andthe extracted metals may precipitate as hydrox-ides [65]. As an alternative to destruction of theorganic ligands, organically bound trace metalsmay be extracted by competing synthetic chelates,e.g. EDTA or DTPA [46]. In sequential extrac-tions, EDTA or its ammonium salt [51] has beenless frequently used than the advantages wouldsuggest. As NH4EDTA, adjusted with NH4OH topH 4.6, has been reported to dissolve considerableamounts of amorphous sesquioxides, it may beless selective than some pyrophosphate methods.Nevertheless, it should be considered as an alterna-tive to extractants with alkaline pH, e.g. Na4P2O7,K4P2O7, NaOH, or NaOCl. Thus, NH4ETDA(pH 4.6) can be fitted in a sequence of extractants

with decreasing pH that is thought to increase theselectivity by minimizing adverse effects on eachsubsequent extraction step, i.e. readsorption or pre-cipitation of trace metal compounds [51]. More-over, the procedure is non-destructive to organicmatter and organo-mineral associations, thus cre-ating no new surfaces that may cause adsorp-tion of trace metals during subsequent extractionsteps as discussed by Beckett [27]. The dissolu-tion of amorphous sesquioxides is probably lim-ited by choosing a reasonable extraction sequence,extracting organically bound trace metals afterremoval of the most labile oxide fraction, e.g.the Mn oxides [62], and by a comparatively shortextraction time of 90 min, as proposed by Zeienand Brummer [40]. Accordingly, good correla-tions were found between organic carbon andNH4ETDA-extractable metal fractions, althoughthere was an evidence that EDTA extractants coulddissolve trace metals from amorphous sesquiox-ides. Among sesquioxides, the Mn oxides are mostsusceptible to changes in pE and pH. Therefore,trace metals bound to Mn oxides, i.e. Pb, maybe readily mobilized upon changed environmentalconditions, e.g. flooding. For that reason, this envi-ronmentally significant fraction is separated priorto Fe and Al oxides by most sequential extractionprocedures. Essentially, Mn oxides are extractedby reducing agents, e.g. NH2OH · HCl or hydro-quinone, either pure or mixed with NH4Ac, HAc,or diluted HNO3.

With higher concentrations (i.e. 0.25 mol L−1)and higher temperatures during extraction (i.e.50 ◦C–100 ◦C), NH2OH · HCl extracted consid-erable amounts of trace metals from sesquiox-ides with a wide range of crystallinities [27]. Asintended by Tessier [51]. 0.04 mol L−1 NH2OH ·HCl in 25 % HAc (pH = 2 at 85 ◦C for 5 h)actually should extract most of the sesquiox-ides, including the crystalline fractions. Zeienand Brummer [40] proposed 0.1 mol L−1 NH2OH ·HCl in 1 mol L−1 NH4Ac (pH 6, 30 min at 20 ◦C).Nevertheless, this procedure seemed to be com-parably selective, hence it dissolved on an aver-age 37 % (0.12 %–73.9 %) of total Mn, butonly 0.02 %–2.9 % of total Fe from a vari-ety of soils. A negative correlation between


Fe and Mn extracted by 0.1 mol L NH2OH ·HCl/1 mol L−1 NH4Ac (pH = 6) indicated thatonly for low levels of Mn oxides present in thesoil, this reagent dissolve some Fe oxides up to2.9 % of total Fe. Since NH2OH · HCl had lit-tle effect on the organically bound metal frac-tion, it should be applied prior to extractants likeK4P2O7, Na4P2O7, NH4EDTA [63]. Trace met-als bound to Fe and Al oxides were extractedeither in one step or were partitioned in twofractions, referred to as amorphous and crys-talline Fe oxides. Essentially, trace metals boundto amorphous Fe oxides were removed by var-ious modifications of acid oxalate solution inthe dark [27]. To extract either the total amountof Fe oxides, or the crystalline fraction subse-quent to removal of the amorphous Fe oxides,acid oxalate solutions were frequently employedeither under diffuse illumination or UV radi-ation at 20 ◦C–100 ◦C and solution/soil ratiosbetween 10 : 1 and 50 : 1 for 0.5 to 3 h. Theconcentrations of the (NH4)2C4O4 · H2O/H2C2O4

reagents were either 0.175 mol L−1/0.1 mol L−1 or0.2 mol L−1/0.2 mol L−1, occasionally used alongwith 0.1 mol L−1 ascorbic acid. This variety ofconditions and the pronounced effects of variedillumination and temperature on Fe extractabilitysuggest that hardly two procedures extract equalamounts of trace metals from soils [27]. Despitedifferences in the extraction parameters, most pro-cedures may fairly selectively remove the crys-talline Fe oxides when employed subsequent toextractions of Mn oxides, amorphous Fe oxidesand organic and carbonate fractions [27]. Uncer-tainties remain whether different extraction condi-tions may result in dissolution of varied amountsof trace elements from clay minerals.

Several commonly used extraction proceduresand the referred fractions are available in the lit-erature. The procedures contain in general theextraction steps as described previously. Slightor significant modifications of these most fre-quently used procedures are widely reported. Mostextraction procedures addressed a wide rangeof heavy metals but some extraction schemeswere developed for specific elements or groupsof elements.

10 SEQUENTIAL EXTRACTIONSCHEMES APPLIED TO AEROSOLSAMPLES

The information on aerosol chemical speciationindicates the mobility of the elements once theaerosol is mixed directly into natural waters orduring scavenging of the aerosol by wet deposi-tion. During mixing of an aerosol with aqueoussolutions it is the anthropogenic metals which arepreferentially released, with these having poten-tially the most harmful impact on the biologicalcommunity. Atmospheric aerosols have importantroles in the biogeochemistry and transportationof trace elements in the air. Direct atmosphericdeposition makes only a minor contribution to thetotal metal contents of the lithosphere because ofthe large reservoir of these metals in soils androcks. The aerosol particles influence the solarradiation transfer, cloud–aerosol interactions, andcontrol the optical, electrical, and radioactive prop-erties of the atmosphere. Aerosols sampled withinthe urban environment exhibit a greater solubilitythan aerosols with a crustal origin and this shouldbe kept in mind when interpreting the results ofsequential leaching.

Atmospheric removal occurs by dry depositionof aerosol particles to water, soil, buildings, orplants, or by wet deposition of aerosol particlesand gases in rain, fog, hail, and snow. Wetdeposition is a very important removal process forthose elements associated with small particles andwhich are predominantly anthropogenic in origin.Approximately 80 % of the atmospheric removalof elements such as Pb, Cd, Cu, Ni, and Zn tothe ocean takes place by wet deposition, whereas40 % of that process for Fe and Al occurs bydry deposition [11]. This is mostly size dependent,so size fractionation is an important control onremoval processes. Wet deposition provides amechanism by which the metals in aerosol particlescan be solubilized and the pH of rainwater is amajor control on metal solubility in precipitation.Rainwater pH is governed by a balance betweenthe concentration of acid and neutralizing speciespresent in solution. Usually the metal solubilityincreases as pH decreases. Before aerosol particlesare removed by precipitation they are cycled within

EXTRACTION SCHEME FOR AEROSOL SAMPLES 135

the atmosphere through clouds and subjected torepeated wetting and drying cycles before removal.

The atmospheric speciation of metals has sev-eral difficulties. Measurement of the total concen-tration of most metals in atmospheric samples ishampered by the lack of proper quality assuranceprograms. Speciation of trace metals in the atmo-sphere is different from those occurring in thehydrosphere. Therefore, forms of an element aredetermined by the mechanism of interaction of thebiosphere and the atmosphere, and the mechanismof transport in the atmosphere. For estimation ofthe elemental budget in the atmosphere dry andwet deposition rates have to be calculated. The drydeposition rate can be calculated from the resultsof elemental contents of atmospheric aerosols. Thedeposition velocity of the aerosol particles can bedetermined with following formula [66]:

v =

n∑

i=1

civi

n∑

i=1

ci

(1)

where v is the dry deposition velocity (cm s−1), ci

is the concentration of the ith element (ng m−3),and vi is the deposition velocity of the ith particle.The dry deposition rate (Dd, mg m−2 year−1) iscalculated with the following formula:

Dd = civ 0.315 (2)

where Dd is the dry deposition rate (mg m−2

year−1), ci is the concentration of the i th element(ng m−3), v is the dry deposition velocity (cm s−1),and 0.315 is a calculation factor.

Chester et al. [67] suggested a sequential le-aching scheme for the characterization of thesources and environmental mobility of trace metalsin the marine aerosols. The distribution of ele-ments can be reliably determined in three fractionsas environmentally mobile, bound to carbonatesand oxides (Fe, Mn oxides), and bound to organicmatter and silicates (environmentally immobile)fractions. Aerosols sampled within the urban envi-ronment usually exhibit a greater solubility thanaerosols with a crustal origin. Particular attention

has to be paid to distinguish between environ-mentally mobile and environmentally immobilefractions because these represent the two extrememodes by which the metals are bound to the solidmatrices. The interaction of trace metals in theaerosols with the other receiving spheres (hydro-sphere, lithosphere, biosphere) depends greatlyon the solubility of metals under environmentalconditions. Chester et al. [68] have demonstratedthat aerosol speciation data can be related to theextent of the solubility of an element in an aque-ous medium, with metals in the environmentallymobile form being most soluble, and can providea framework for assessing the reactivity of the ele-ments once they have been deposited at the surface.Lum et al. [69] provided data on the chemical spe-ciation of a number of elements in an aerosoldominated by pollutants and mainly generated athigh temperature by applying a sequential leachingscheme to samples of an Urban Particulate Mat-ter Standard Reference Material (SRM 1648). Forcharacterization of the crustal aerosols a five-stagesequential leaching technique was used for soil-sized aerosols [68]. Comparing the data derivedfrom the two studies reveals that the speciationsignatures for some elements differ considerably;aerosols collected in a polluted city generally con-tained more environmentally mobile fractions con-taining elements such as Pb, Cr, Zn, Cu, and Cd.The crustal aerosol samples consisted of a higherportion in the stable fractions for the all elementsstudied. The environmentally mobile/bound to sili-cates fractions can be interpreted with particle sur-face/particle matrix associations formed as a resultof high temperature anthropogenic processes andlow temperature crustal weathering processes.

The environmental mobility indicates the ex-change of metals adsorbed or condensed on thesurface of aerosols. For the separation of this frac-tion of the total metal content of an aerosol, themetals have to be displaced from the substrate byreversing the binding mechanisms without affect-ing metals held in other associations. The mostcommonly used reagents are solutions of either1 mol L−1 NH4OAc or MgCl2. The pH needs tobe sufficiently high that protons neither competenor react with other phases in the aerosols, but


not so high that hydroxides precipitate. A slightlymore acidic solution such as NH2OH · HCl andHOAc can be used to decompose carbonates lib-erating incorporated trace metals, and dissolvingany free heavy metal carbonates. This reducingchemical agent can release heavy metals associatedwith Fe/Mn oxides. After reduction of Mn(III) andMn(IV) to Mn(II), and Fe(III) to Fe(II), the speciesproduced are soluble, and metal ions bound to theoxy-compounds are liberated into solution. Finally,the organic and silicate fractions can be decom-posed using a combination of HNO3 and HF. SinceSiF4 is volatile it is lost from the solution, and themetal ions are released from the lattice.

Recently, fractionation by size and chemicalbonding on aerosols has been studied for the firsttime [70]. The sequential leaching technique hasbeen applied to filter-collected aerosols in eightparticle size ranges for determination of the dis-tribution patterns of elements. Particular atten-tion was paid to distinguish between the fine andcoarse particle size fractions, and the environ-mentally mobile and environmentally immobileportions. Among several elements Pb showed aunimodal distribution at a maximum of <1 µm,indicating unambiguously the single anthropogenicsource (traffic) (Figure 3.3.1). The pollution at thissite was around 29 ng m−3, this value being oneor two magnitudes lower, due to less traffic and

long-range transportation, than others found in dif-ferent major cities [11]. Emissions from vehicleexhausts dominated the lead contribution to theatmosphere, although smelting operations also con-tributed to this atmospheric lead load, emittingboth PbO and Pb0. The relative significance of leadsources in the atmosphere has recently changedworldwide as a result of the considerable declinein the use of leaded vehicle fuels. The actual leadcompounds in a particular aerosol will dependon the other constituents in the atmosphere andthe age of the aerosol. The predominant inorganicaerosol-phase lead species has been identified asPbSO4 · (NH4)2SO4 by XRD [70] and it is sug-gested that this species arises from transformationsof the primary emitted aerosol compounds dur-ing atmospheric transport. The distribution of Pbamong the three fractions was rather even, theenvironmentally mobile fraction being about 40 %.

The average geometric mean concentration ofCd was found to be <1 ng m−3 and unimodal dis-tribution pattern was obtained (Figure 3.3.2) [70].The concentration of Cd in aerosols depends con-siderably on the location, pollution sources, time,meteorological conditions, etc. and ranges from 1to 300 ng m−3 in major cities. Cadmium was mostlyassociated with the environmentally mobile frac-tions (50 %); smaller amounts of Cd compoundwere found in the fractions bound to carbonates

8

7

6

5

4

Pb,

ng/

m3

3

total

organic matter, silicate

carbonate, oxide

mobile

µm

Pb

2

1

0

0.08

8

0.18

0.35

0.71

1.4

2.8

5.7

11.3

Figure 3.3.1. Distribution of Pb in three fractions as a function of particle size [70].

EXTRACTION SCHEME FOR AEROSOL SAMPLES 137

0.16

0.18

0.2

0.14

0.12

0.1

0.08Cd,

ng/

m3

0.06

total

organic matter, silicate

carbonate, oxide

mobile

µm

Cd

0.04

0.02

0

0.08

8

0.18

0.35

0.71

1.4

2.8

5.7

11.3

Figure 3.3.2. Distribution of Cd in three fractions as a function of particle size [70].

or oxides (33 %) and bound to organic matter orsilicate (17 %). Lum et al. [69] have reported thatCd in an urban aerosol was found to be almostcompletely in exchangeable form. The data of thesequential leaching procedures showed somewhatdifferent patterns, due to the different experimen-tal conditions (leaching steps, reagents, extractiontime, temperature, etc.), and the origin of sam-ples. Nevertheless, Cd was mainly found in envi-ronmentally mobile form in fly ashes collected atmunicipal incineration sites, and upon disposal thetoxic metal would be transported into the receivingmedia. High enrichment factor (EF = 4665) valuesindicate that Cd was emitted from anthropogenicsources [70]. Cadmium liberated during combus-tion processes has been shown to occur in elementaland oxide forms, whereas emissions from refuseincineration were predominantly as CdCl2. Differ-ent metals were strongly fractionated between dif-ferent aerosol size fractions and this had importantimplications for all aspects of atmospheric transportfrom public health to global metal cycling.

There are several projects in which fraction-ation by particle size has been applied andonly the total concentrations of elements aredetermined. Recently, three size fractions ofparticulate matter (PM), fine particles (PM2.5),

(particle size <2.5 µm), coarse particles (PM2.5 – 10)and PM10, were measured at eight sampling sitesin four large Chinese cities during 1995 and1996 [71]. Annual means of PM10 concentrations,of which 52–75 % were PM2.5, ranged from 68to 273 µg m−3. Within each city, the urban sitehad higher annual means of all measured PM sizefractions. It was clearly demonstrated that the ele-ments were enriched more in fine particles thanin coarse ones. An air quality monitoring pro-gram in the Czech Republic has provided data forthe concentrations of aerosol and gas-phase pol-lutants [72]. Fine particulate matter (PM2.5) wascomposed mainly of organic carbon and sulfatewith smaller amounts of trace metals. Coarse par-ticle mass concentrations were typically between10 and 30 % of PM2.5 concentrations. The ambi-ent monitoring and the source characterization datawere used in receptor modeling calculations.

For evaluation of the atmospheric budget andthe environmental effects of trace metals on thebiosphere, the calculation of the dry and wet depo-sitions is of vast importance. The relative signif-icance of the two depositional processes variesbetween locations and is primarily a function ofthe rainfall intensity in that area. Wet depositionis a very important removal procedure for those


elements associated with small particles and whichare predominantly anthropogenic in origin. Wetdeposition rates are based upon the concentrationof trace metals in precipitation samples collectedby a wet-only sampler. Using the methods devel-oped for fractionation by chemical bonding, i.e.the three-stage sequential leaching protocols, thedry deposition of aerosols can further be dividedand a more reliable estimation can be performed.

In a recent survey the atmospheric bud-get was calculated for Lake Balaton, Hungary,using a monitoring system of 3 years [65]. Ithas been reported recently [66] that the ratioDdry/Dwet is significant, in particular, for Pband Zn and usually higher for the others (V,Cr, Ni, Cu, As). In the former study [65] ithas been found that, for elements such as Fe,Al, and Cu, the ratio of the two depositionsis opposite, i.e. dry deposition plays a moreimportant role in the pollution of the environ-ment (Table 3.3.4.). On the other hand, ratios ofDdry/Dwet clearly indicated that elements such Mn,As, Cd, Pb, and mainly Zn, were deposited inmuch higher amounts by wet deposition than bydry deposition.

The environmentally mobile fractions of totaldry deposition of elements were also calculatedand compared to the total wet deposition. Itwas obvious that the contribution of mobilefractions of the dry deposition to pollution was,except for Cu, minor. Copper compounds weremainly removed from the atmosphere by dry

deposition and one third of the amount of Cuwas environmentally mobile. However, if mobilefractions were compared to the total solubledeposition (Ddrymobile + Dwet), less than 50 % ofCu came only from dry deposition and much less(4–27 %) from all other elements. This means thatin the budget of soluble compounds wet depositionplays the more important role. In the case ofelements such as Al, Fe, and Cu, dry deposition hasbeen found to be the major source of pollution, butthe mobile parts of dry deposition played a minorrole compared to the total soluble deposition.

Furthermore, the soluble fractions of deposi-tions (Ddrymobile + Dwet) were compared to thetotal depositions (Ddry + Dwet). The water qual-ity of the lake has been influenced by the solublepart of the atmospheric depositions, and it hasbeen found that 85–94 % of toxic elements (Pb,Cd, Ni, Zn, and As) were dissolved in the water.The other portions of elements were stable com-pounds formed under natural environmental con-ditions, and after precipitation they settle to thebottom of the lake. So, metal compounds, sooneror later, become the part of the bottom sediment,since the fate of dissolved metals depends greatlyon the physical and chemical conditions of the bulkwater (pH, complex forming capacity, adsorptionon clays, quartz, organic matter, biological activi-ties, etc.). In long-term studies the concentrationsof the same elements of the lake water have beenfound to be very low, even lower than the stan-dards permit for drinking waters [65]. Sediments

Table 3.3.4. Total dry (Ddry, kg year−1) and wet (Dwet, kg year−1) deposition, sum of the dry and wet deposition (Ddry + Dwet,kg year−1), mobile fraction of the total dry deposition (Ddrymobile, kg year−1), sum of the mobile fraction of the dry and wetdeposition (Ddrymobile/Dwet kg year−1), ratio of mobile fraction of dry and wet deposition (Ddrymobile/Dwet, %) of elements insamples collected around the Lake Balaton [65].

Element TotalDdry

(kg year−1)

TotalDwet

(kg year−1)

Ddry +Dwet

(kg year−1)

TotalDdrymobil

(kg year−1)

Ddrymobil +Dwet

(kg year−1)

(Ddrymob ×Dwet)

100 (%)

[Ddrymob(Ddrymob + Dwet)]

×100 (%)

[(Ddrymob + Dwet)(Ddry + Dwet)]

×100 (%)

Al 7 980 5 670 13 650 2 131 7 801 37.6 27.3 57.1Fe 19 740 13 482 33 222 4 284 17 766 31.8 24.1 53.5Mn 6 960 17 634 24 594 1 134 18 768 6.4 6.0 76.3Pb 618 1 374 1 992 309 1 683 22.5 18.3 84.5Cd 12 54 66 5.4 59.4 10.0 9.1 90.0Cu 900 336 1236 300 636 89.3 47.2 51.5Ni 156 516 672 54 570 10.5 9.4 84.8Zn 618 5 604 6 222 214 5 818 3.8 3.7 93.5As 138 582 720 39 621 6.7 6.2 86.2

EXTRACTION SCHEME FOR FLY ASH SAMPLES 139

are recognized as a sink and reservoir for metals,metalloids, and other contaminants. Therefore, theaccumulation of metals in a lake can be followedby the analysis of bottom sediments.

11 SEQUENTIAL EXTRACTIONSCHEMES APPLIED TO FLY ASHSAMPLES

In recent years it has been recognized that thepotential toxicity of fly ashes is much more relatedto the leachable fraction of contaminants, since thetotal toxic amount of fly ash is not extractableunder natural environmental conditions. A lot ofleaching tests have been developed worldwide,each characterized by different aims (researchor regulatory controls) and different experimentalparameters (pH, leachant, stirring device, time ofextraction, etc.) [9]. Research groups have beenworking on the design and study of extractionprotocols aiming at the sequential solubilizationof metals bound to the specific substrates makingup fly ash and known to undergo changes in theyearly life cycle after deposition. Measurementof such functionally defined metal fractions willallow some forecasting of metal release from flyashes under certain conditions and constitutes,therefore, a valuable tool in disposal management.Harmonization and optimization of methods arenecessary because the wide variety of procedurescan lead to incomparability of results even forsingle aquatic systems. It is also desirable tomeasure the total concentration of the elementin order to verify the mass balance. Selectiveleaching of elements from power plant ashes bywater and ammonium acetate and various acids isfrequently carried out to determine the nature ofthe elements and potential mobility during storage[73]. There are some limitations to this procedure,e.g. minerals encased by organic matter or byother minerals may not come into contact withthe solvent and, therefore, will not be leached.Elemental interaction (i.e. precipitates, colloids,alkali halides or other water-soluble minerals)may have a relatively large effect, especiallyin water leaching. The water-soluble mineralsare mostly alkali halides. Elements soluble in

ammonium acetate are ion-exchangeable in nature,and are mostly associated with organic matter incoal, such as salts of organic acids, and withclay minerals. The elements removed by HClare associated with acid-soluble minerals suchas carbonates, metasulfides and oxides, and alsowith organic matter functional groups such ascarboxyl groups. Hydrofluoric acid is the principalsolvent of silicates, while disulfide minerals aremostly removed by nitric acid. A sequentialextraction scheme was applied to the determinationof binding forms of trace Cd in coal fly ashreference material, NBS 1633a [74]. The sum ofthe Cd present in the individual fractions showed agood agreement with the certified value of total Cdcontent. The extraction efficiency was >85 %. Theextractable Cd provides information concerningthe binding form of the element.

The chemical fractionation of As, Cr, and Niin milled coal, bottom ash, and ash collected byelectrostatic precipitator from a coal-fired powerplant was determined by a sequential leaching pro-cedure [73]. Deionized water, NH4OAc, and HClwere used as extracting agents and the leachatewas analyzed by ICP-AES. Arsenic in the milledcoal was mostly associated with organic matter,and 67 % of this arsenic was removed by NH4OAc.This element was totally removed from milledcoal after extraction with HCl. Both Ni and Crin this sample were extracted by HCl, indicatingthat water could mobilize Ni and Cr in an acidicenvironment. The chromium was leached by waterfrom fly ash as a result of the high pH of the water,which was induced during the leaching. Ammo-nium acetate removed Ni from bottom ash throughan ion exchange process. Austin and Newland [75]leached fly ashes from a power plant and a munic-ipal incinerator for 3 h with 0.1 mol L−1 HCl. Theyfound that Cd was rapidly removed from the ashparticles in the initial 5 min of leaching. A com-bined physical and chemical approach has beenproposed to quantify the relative concentrations ofelements in the aluminosilicate matrix and in thenonmatrix or surface material of coal fly.

Using several fly ash samples from differentsources, it was found that the difference in the


Table 3.3.5. Comparison of results for CW6 sample (n = 3, t (95 %) = 4.303 [76]).

Fractions Cu Zn Pb Cd V

(mg kg−1) (%) (mg kg−1) (%) (mg kg−1) (%) (mg kg−1) (%) (mg kg−1) (%)

1 1199 ± 59 59.6 12 046 ± 422 45.9 2355 ± 874 24.8 307 ± 10 74.2 1.1 ± 0.1 5.52 302 ± 2 15.0 7603 ± 1417 29.0 2295 ± 707 24.1 90 ± 3 21.7 2.4 ± 0.4 11.93 30 ± 13 1.5 1757 ± 522 6.7 418 ± 147 4.4 9 ± 6 2.2 2.7 ± 0.2 13.44 311 ± 5 15.5 2482 ± 758 9.4 1473 ± 80 15.5 5 ± 0.5 1.2 4.0 ± 0.1 19.85 170 ± 15 8.4 2379 ± 288 9.0 2975 ± 442 31.2 3 ± 1.2 0.7 10.0 ± 0.8 49.5�1–5 2012 ± 70 100 26 267 ± 2732 100 9516 ± 704 100 414 ± 29 100 20.2 ± 1.1 100Total 1911 ± 379 26 936 ± 4520 8345 ± 2922 428 ± 82 23.2 ± 16.7

partition depends considerably on the characteris-tics of the raw material and the operational con-ditions, i.e. combustion temperature, furnace, etc.So, each type of fly ash should be investigated sep-arately and this fact has to be taken into accountwhen a standard reference material is being pro-duced. A five-stage sequential leaching procedurehas been developed and applied to fly ash samplescollected at different emission sources [38]. Thesolvent leaching experiments together with solid-phase examinations carried out by X-ray powderdiffraction provided information on the possibleenvironmental impact of particle-associated pol-lutants. The particulate elements were partitionedinto five fractions: (1) exchangeable elements,(2) elements bound to carbonates, (3) elementsbound to Fe/Mn oxides, (4) elements bound to sul-phide compounds, and (5) elements bound to sili-cates, residual fraction. This procedure was opti-mized for analysis of a fly ash candidate referencematerial [76]. The total concentration of elementsof a candidate fly ash reference material was deter-mined in six laboratories and analytical proceduresfor the quantitative leaching of inorganic contami-nant were harmonized. The average concentrationwas calculated and compared to the sum of the fivefractions determined by the optimized procedure.Data are shown in Table 3.3.5.

Results of the total elemental analysis and thesequential leaching method were compared andgood agreement was found. The leachability ofthe metals proved to be different, so variousdistribution patterns have been achieved. Copperwas found mostly in the environmentally mobilefraction (60 %), while zinc was concentrated inthe exchangeable elements and elements bound to

carbonates fractions, (75 %). The sum of the frac-tions of sequential leaching can also give reliableinformation on the amounts of elements in fly ashsamples. Half of the amount of Pb was identifiedin environmentally mobile fractions (exchangeableelements and elements bound to carbonates), andabout 30 % was concentrated in a stable fraction.Results for Cd indicated a great environmentalconcern, as almost the total amount was accu-mulated into the exchangeable elements and ele-ments bound to carbonates fractions, and if thefly ash were disposed of Cd compounds could betransported into the receiving media. Agreementbetween the two methods (analysis of total concen-tration and the sequential leaching) is excellent, sothe optimised sequential leaching method can reli-ably be used. Certified reference materials play akey role in the determination of performance char-acteristics since they stand for the validation ofreliable methods. They also permit the compara-bility of results in different laboratories. If certifiedreference materials are lacking, the accuracy ofany sequential leaching procedure developed canbe controlled by an independent method.

12 PRINCIPLES AND APPLICATIONOF FIELD-FLOW FRACTIONATION(FFF), SEPARATION OF SUSPENDEDPARTICLES INTO SPECIFIC SIZEFRACTIONS

12.1 Normal mode of FFF

An effective way of fractionating is to use field-flow fractionation (FFF). The primary separationin FFF is effected by the action of an external

FIELD-FLOW FRACTIONATION 141

field [77]. The force induced by the field movesmolecules toward the wall of a channel andsince the molecules cannot penetrate the wall anexponential concentration distribution is built up.If the force affecting the molecules is constant, thethickness of the resulting concentration distribution(layer thickness) depends solely on the diffusioncoefficient of the molecule. Slowly diffusing largemolecules lie closer to the wall than smallermolecules having higher diffusion coefficients.The separation is amplified by the parabolic flowprofile of the liquid pumped through the channel(Figure 3.3.3).

As can be seen the highest flow velocity takesplace in the middle of the channel, while closerto the wall the flow is slowed down by thefrictional drag. Because of the flow profile, largemolecules residing close to the wall are also in

slower streamlines than the smaller ones and willtherefore be more retained than small molecules.This separation mechanism is called the normalmode of FFF.

If the sample particles have a radius of thesame magnitude as their layer thickness, theparticles will be in the contact with the wall. Noexponential concentration distribution will existand the velocity of the particle is determined onlyby its protrusion into the fast streamlines. Thecenter of a large particle is in faster streamlinesthan that of a smaller one, when the particles arein contact with the wall. The elution order is nowreversed compared with the normal mode of FFF,i.e. smaller particles elute after larger ones. Thisis the steric mode of FFF. Particles smaller than1 µm usually elute in normal mode and >1 µm insteric mode.

Injector

Normal mode

Flow profile

< 1 µm

> 1 µm

Accumulation wall

Steric mode

Field

Recorder

DetectorPump

Figure 3.3.3. Principle of field-flow fractionation [77]. Reproduced from ref. 77 by permission of Matti Jussila.


12.2 Thermal FFF (ThFFF)

The effect by which molecules are moved towardsthe channel wall in ThFFF is thermal diffu-sion. The upper wall of a thermal FFF chan-nel is heated by electrical cartridge heaters andthe lower wall is cooled by water circulation(Figure 3.3.4). The temperature gradient insidethe channel induces thermal diffusion of macro-molecules usually toward the cooled wall. Thethermal diffusion coefficient is relatively indepen-dent of molar mass but depends on the interactionsbetween the macromolecule and the carrier liq-uid. When any exponential concentration buildsup against the wall, thermal diffusion is counter-acted by ordinary diffusion. The high molar massselectivity of ThFFF is due to the strong effectof molar mass on the ordinary diffusion coeffi-cient. ThFFF is usually used in normal mode only.The most common application of ThFFF is themolar mass analysis of polymers soluble in organicsolvents. The use of water as a carrier is limitedby the poor thermal diffusion of polymers in aque-ous solutions.

12.3 Flow FFF

The wall elements of a flow FFF channel aremade of semipermeable ceramic frit. This allowsan additional flow to penetrate the channel per-pendicularly to the main flow (Figure 3.3.5). Allmolecules inside the channel are moved towardthe wall by this cross flow. To prevent the samplemolecules from being flushed out of the channel,the wall is covered by an ultrafiltration membranehaving a cut-off well below the molecular weightof the analyte. The low molar mass carrier can stillpenetrate the wall elements freely. Because all thesample molecules are affected uniformly by thecross flow, the retention in normal mode is deter-mined only by the diffusion coefficient (D) of thesample. In normal mode operation, D for the sam-ple can be evaluated from its retention time. Usingthe Stokes–Einstein relationship, D can be con-verted to the diameter and similarly the fractogramcan be turned into a particle size distribution. Insteric mode straight calculation of particle sizehowever, is not possible and calibration using stan-dards of known particle size is required. Flow FFF

Carrierin

Thermocouples

Channel

Bolt

Clamping plate

Thermal insulation

Hot blockElectric cartridgeheater

Carrier outlet

Mylar spacer

Polished surface

Cold block

Water circulation

Thermal insulation

Clamping plate

Figure 3.3.4. Construction of a thermal FFF channel [77]. Reproduced from ref. 77 by permission of Matti Jussila.

FIELD-FLOW FRACTIONATION 143

Plexiglassclamping block

Porous fritcoated with castcellulose acetate

Channel

Mylar spacer

Ultrafiltrationmembrane

Porous frit

Plexiglassclamping block

Main flow out

Cross flow in

Main flow in

Cross flow out

Figure 3.3.5. Construction of an FFF channel [77]. Reproduced from ref. 77 by permission of Matti Jussila.

is applicable to most particulate sample materialssuspended in water. The primary developer of thistechnique is Calvin Giddings and FFF has beenproven successful in the analysis of pharmaceu-ticals, biotechnology products, soils, and foods,among others. A summary of the new develop-ments has been published recently in a book enti-tled Field-Flow Fractionation Handbook [78].

ICP-MS was used for the quantitative measure-ment of trace elements in specific, submicrometersize-fraction particulates, separated by sedimenta-tion FFF [79]. Fractions were collected from theeluent of the FFF centrifuge and nebulized intoan argon ICP mass spectrometer. Measured ioncurrents were used to quantify the major, minor,and trace element composition of size-separated(Se) colloidal (<1 µm) particulates. This approachproved to be ideal for studying the chemistry ofclays and other colloidal material. The combina-tion of these two techniques (SeFFF-ICP-MS) has

the potential of providing a unique and impor-tant tool for the measurement of trace elementcontamination associated with colloidal suspendedmatter in environmental water. Using on-channelpreconcentration, the flow FFF coupled on-linewith ICP-MS has been applied to the study ofelement distributions in colloids for 28 elementsin natural water [80]. The technique was highlyflexible and applicable in various particle sizeranges. Furthermore, it has a relatively simple the-oretical background which gives the possibilityto compare theoretically obtained sizes with sizesor molecular weights obtained from calibrations.With optimized conditions, detection limits andreproducibility were sufficient for metal specia-tion in natural freshwater samples. SedimentationFFF coupled on line with ICP-MS has opened newpossibilities for studies of trace metal adsorptiononto natural colloids [81]. Major elements Al,Si, Fe, and Mn were determined simultaneouslytogether with trace elements Cs, Cd, Cu, Pb, Zn,


and La. The ability to relate metal uptake to thesize and composition of colloids is expected to leadto new insights into uptake processes and into thetransport and fate of trace metals in aquatic sys-tems. The FFF techniques need further refinementbut can be of great help in studies of element trans-port in rivers, estuaries, and aquifers.

13 DISCUSSION

Despite their limitations, sequential extractionschemes can provide a valuable tool to distinguishbetween trace metal fractions of different particlesize and solubility. These fractions are empiricallyrelated to mobility classes in different solidmatrices. The speciation of metals governs theiravailability to the biota and their potential tocontaminate the environment. Available forms ofmetals are not necessarily associated with oneparticular chemical species or a specific solidsample component. Speciation of trace elementsmay vary with time, depending on the solid-phasecomponents that are present, pH, and the numberand accessibility of adsorption sites. Soluble andexchangeable forms of metals will decrease withtime if there are other solid components presentthat can adsorb the metal more strongly and havefree sites that are accessible (e.g. hydrous oxide,organic matter).

The present state of knowledge on solid matterfractionation of trace elements is still somewhatunsatisfactory because the appropriate techniquesare only operationally defined and associatedwith conceptional and practical problems. Withrespect to estimating bioavailable element con-centrations, one such conceptional problem is theeffect of competition between binding sites onthe solid substrate and selective mechanisms ofmetal translocation by the different organismsinvolved. This situation cannot yet be improvedby more sophisticated analytical approaches tofractionation.

On the other hand, the usefulness of a differenti-ated approach, even if only operationally defined,to the interactive processes between water–biotaand solid phases has been clearly proven. Thepossible environmental implications, e.g. of land

disposal of waste material, of acid precipitation,of redox changes in subsoil, and of ingestion ofpolluted urban dust, can be qualitatively estimated,particularly when the physicochemical conditionsof the interacting compartments of the environ-ment are taken into consideration. The method ofsequential chemical extraction is the least sophis-ticated and most convenient technique availablefor a fractionation assessment. However, we mustbe certain that we fully understand what is hap-pening during extraction to minimize the risk ofproducing artifacts and choose standard proceduresto ensure that results are comparable. The pri-mary importance of proper sampling protocols hasbeen emphasized, since the sampling error cancause erroneous results even using highly sophis-ticated analytical methods and instruments. Thenumber of fractionation steps required dependson the purposes of the study. The BCR proto-cols give a simple guide for most of the solidsamples and the results can be compared betweendifferent laboratories. Geoscientists and environ-mental engineers extensively use results of chem-ical fractionation analysis and scientists have theresponsibility to show the pitfalls and limitations ofsequential extraction procedures developed. Decla-ration of the uncertainty of results is a must andgreatly improves the quality of these activities.

14 ACKNOWLEDGEMENTS

The authors are indebted to the OTKA T 029 250,AKP 2000-30 2,5 and the Balaton Secretariat of thePrime Minister’s Office for their financial support.

15 REFERENCES

1. Cabral, A. R. and Lefebvre, G., Water Air Soil Pollut.,102, 329 (1998).

2. Pichtel, J., Sawyerr, H. T. and Czarnowska, K., Environ.Pollut., 98, 169 (1997).

3. Hewitt, C. N. and Harrison, R. M., Monitoring, inUnderstanding our Environment , Hester R. E. (Ed.), TheRoyal Society of Chemistry, London, 1986, Chapter 1,pp. 1–10.

4. Mehra, A., Cordes, K. B., Chopra, S. and Fountain, D.,Chem. Spec. Bioav., 11, 57 (1999).

REFERENCES 145

5. Barona, A., Aranguit, I. and Elias, A., Chemosphere, 39,1911 (1999).

6. Schalscha, E. and Ahumada, I., Water Sci Technol., 37,251 (1998).

7. Templeton, D., Ariese, F., Cornelis, R., Muntau, H., vanLeeuwen, H. P., Danielsson, L.-G. and Lobinski, R., PureAppl. Chem. 72, 1453 (2000).

8. Martin, J. M., Nirel, P. and Thomas, A. J., Mar. Chem.,22, 313 (1987)

9. Kersten, M. and Forstner, U., Speciation of trace metalsin sediments and combustion waste, in Chemical Specia-tion in the Environment , Ure, A. M. and Davidson, C. M.(Eds), Blackie, London, 1995, Chapter 9, pp. 234–275.

10. Ritchie, G. S. P and Sposito, G., Speciation in soils,in Chemical Speciation in the Environment , Ure, A. M.and Davidson, C. M. (Eds), Blackie, London, 1995,Chapter 8, pp. 201–233.

11. Spokes, L. J. and Jickells, T. D., Speciation of metals inthe atmosphere, in Chemical Speciation in the Environ-ment , Ure, A. M. and Davidson, C. M. (Eds), Blackie,London, 1995, Chapter 6, pp. 137–168.

12. Ure, A. M. and Davidson, C. M., Chemical Speciation inthe Environment , Blackie, London, 1995.

13. Buffle, J., Wilkinson, K. J., Tercier, M. L. and Partha-sarathy, N., in Reviews on Analytical Chemistry, Euro-analysis IX , Palmisano, F., Sabbatini, L., Zambonin, P.G. (Eds), Societa Chimica Italiana, 1997, pp. 67–82.

14. Etcheber, H. and Jouanneau, J. M., Estuar. Coast. Mar.Sci., 11, 701 (1980).

15. Kersten, M. and Forstner, U., Mar. Chem., 22, 299(1987).

16. Bartlett, R. and James., B. Soil Sci. Soc. Am. J. 85, 721(1980).

17. Thomson, E. A., Luoma, S. N., Cain, D. J. and Johans-son, C., Water Air Soil Pollut., 14, 215 (1980).

18. Cuesta, P. A., McDowell, L. R., Kunkle, W. E., Bul-lock, F., Drew, A., Wilkinson, N. S. and Martin, F. G.,Commun. Soil Sci. Plant Anal., 24, 335 (1993).

19. Linehan, D. J., Sinclair, A. H. and Mitchell, M. C., J.Soil Sci., 40, 103 (1989).

20. Beckett, P. H. T. and Webster, R., Soils Fert., 34, 1(1971).

21. Webster, R., Adv. Soil Sci., 3, 1 (1985).22. Vaughn, C. E., Center, D. M. and Jones, M. B., Soil Sci.,

141, 43 (1986).23. Wenzel, W. W., Brandstetter, A., Pollak, M. A., Mentler,

A. and Blum, W. E. H., Seasonal changes of organicmatter, pH, nitrogen and some metals in forest topsoilsin Austria: A case study of two soils with and without alitter layer, in Environmental Impacts of Soil ComponentInteractions, Part II, Toxic Metals, Other Inorganics andMicrobial Activities , Huang, P. M., Berthelin, J., Bollag,J. M., McGill, W. B. and Page, A. L. (Eds), CRC Press,Boca Raton, FL, 1995, Chapter 8, pp. 85–95.

24. Hammer, R. D., O’Brien, R. G. and Lewis, R. J., Soil.Sci. Soc. Am. J., 51, 1320 (1987).

25. Jones, D. L. and Edwards, A. C., Commun. Soil Sci.Plant Anal. 24, 171 (1993).

26. Wenzel, W. W. and Blum, W. E. H., Assessment ofmetal mobility in soil-methodological problems, in MetalSpeciation and Contamination of Soil , Allen, H. E. andHuang C. P., (Eds), Lewis Baco Raton, FL, 1994,Chapter 9.

27. Beckett, P. H. T., Adv. Soil. Sci., 9, 143 (1989).28. Gilliam, F. S. and Richter, D. D., J. Soil Sci. 39, 209

(1988).29. Hinds, W. C., Aerosol Technology. Properties, Behavior,

and Measurement of Airborne Particles,, John Wiley &Sons, Inc., New York, 1999.

30. Chester, R., Lin, F. J. and Murphy, K. J. T., Environ.Technol. Lett., 10, 887 (1989).

31. Berner, A., Design principles of the AERAS low pressureimpactor, in Aerosols , Liu, B. Y. H., Piu, D. Y. H. andFissan, H. J. (Eds) Elsevier, Amsterdam, 1984.

32. Waldman, J. M., Munger, J. W. and Jacob, D. J., Mea-surement methods for atmospheric acidity and acid depo-sition, in Atmospheric Acidity, Sources, Consequencesand Abatement , Radojevic, M. and Hamson R. M. (Eds),Elsevier, London, 1992, pp. 205–243.

33. McDow, S. R. and Huntzicker, J. J., Atmos. Environ.,24A, 2563 (1990).

34. Reardon, E. J., Czank, C. A., Warren, C. J., Dayal, R.and Johnston, H. M., Waste Manage. Res., 13, 435 (1995).

35. Hamilton, K. L., Nelson, W. G. and Curley, J. L., Envi-ron. Toxicol. Chem., 12, 1919 (1993).

36. Compliance test for granular waste material, CEN Proto-col EN 12457–part C, 1996.

37. Characterization of waste – Leaching behavior tests – In-fluence of pH under steady state conditions – part 1 pH-static test, draft CEN/TC 292/WG6.

38. Bodog, I., Csikos-Hartyanyi, Zs. and Hlavay, J., Micro-chem. J., 54, 320 (1996).

39. Sposito, G., Lund, L. J. and Chang, A. C., Soil. Sci. Soc.Am. J., 46, 260 (1982).

40. Zeien, H. and Brummer, G. W., Mitteilgn Dtsch. Boden-kundl. Gesellsch , 59, 505 (1989).

41. Sparks, D. L., Kinetics of Soil Chemical Processes , Aca-demic Press, London, 1989.

42. McLaren, R. G., Lawson, D. M. and Swift, R. S., J. SoilSci., 37, 223 (1986).

43. Quevauviller, P., Analyst , 123, 1675 (1998).44. Quevauviller, P., Trends Anal. Chem., 17, 289 (1998).45. Ure, A. M., Quevauviller, P., Muntau, H. and

Griepink, B., Int. J. Environ. Anal. Chem., 51, 135 (1993).46. Davidson, C. M., Duncan, A. L., Littlejohn, D., Ure,

A. M. and Garden, L. M., Anal. Chim. Acta , 363, 45(1998).

47. Perez-Cid, B., Lavilla, I. and Bendicho, C., Analyst , 121,1479 (1996).

48. Mester, Z., Cremisini, C., Ghiara, E. and Morabito, R.,Anal. Chim. Acta, 359, 133 (1998).


49. Rauret, G., Sanchez, J. F., Sahuquille, A., Rubio, R.,Davidson, C. M., Ure, A. M. and Quevauviller, P., J.Environ. Monitor., 1, 57 (1999).

50. Davidson, C. M., Ferreira, P. C. S and Ure, A. M., Fre-senius’ J. Anal. Chem., 363, 446 (1999).

51. Tessier, A., Campbell, P. G. C. and Bisson, M., Anal.Chem., 51, 844 (1979).

52. Salomons, W. and Forstner, U., Environ. Technol. Lett.,1, 506, (1980).

53. Tessier, A., Campbell, P. G. C. and Bisson, M., J. Geo-chem. Explor., 16, 77 (1982).

54. Salomons, W., de Rooij, N. M., Kerdijk, H. and Bril, J.,Hydrobiologia , 149, 13 (1987).

55. Battiston, G. A., Gerbasi, R., Degetto, S. and Sbrig-nadello, G., Spectrochim. Acta , 458B, 217 (1993).

56. Lopez-Sanchez, J. F., Sahuquillo, A., Fiedler, H. D.,Rubio, R., Rauret, G., Muntau, H., Marin, P., Valladon,B. M., Polve, M. and Monaco, A., Anal. Chim. Acta, 342,91 (1997).

57. Coetzee, P. P., Gouws, K., Pluddemann, S., Yacoby, M.,Howell, S. and den Drijver, L., Water SA, 21, 51 (1995).

58. Accomasso, G. M., Zelano, V., Daniele, P. G., Gastaldi,D., Ginepro, M. and Ostacoli, G., Spectochim. Acta , 49a,1205 (1993).

59. Weisz, M., Polyak, K. and Hlavay, J., Microchem. J., 67,207 (2000).

60. Polyak, K. and Hlavay, J., Fresenius’ J. Anal. Chem. 363,587 (1999).

61. Chapman, P. M., Wang, F., Adams, W. J. and Green, A.,Environ. Sci. Technol. 33, 3937 (1999).

62. Miller, W. P., Martens, D. C. and Zelazny, L. W., SoilSci. Soc. Am. J. 50, 598 (1986).

63. Pickering, W. F., CRC Crit. Rev. in Anal. Chem., 11(81),233 (1981).

64. Bascomb, C. L., J. Soil Sci., 19, 251 (1968).65. Hlavay, J., Polyak, K. and Weisz, M., J. Environ. Moni-

tor., 3, 74 (2001).

66. Molnar, A, Meszaros, E., Polyak, K., Borbely-Kiss, I.,Koltay, E., Szabo, G. and Horvath, Z., Atmos. Environ.,29, 1821 (1995).

67. Chester, R., Lin, F. J. and Murphy, K. J. T, Environ.Technol. Lett., 10, 887 (1989).

68. Chester, R., Murphy, K. J. T, Towner, J. and Thomas, A.,Chem. Geol., 54, 1 (1986).

69. Lum, K. R., Betteridge, J. S. and Macdonald, R. R.,Environ. Technol. Lett., 3, 57 (1982).

70. Hlavay, J., Polyak, K., Molnar, A and Meszaros, E.,Analyst , 123, 859 (1998).

71. Wei, F., Teng, E., Wu, G., Hu, W., Wilson, W. E., Chap-man, R. S., Pau, J. C. and Zhang, J., Environ. Sci. Tech-nol. 33, 4188 (1999).

72. Pinto, J. P., Stevens, R. K., Willis, R. D., Kellogg, R.,Mamane, Y., Novak, J., Santroch, J., Benes, I., Lenicek,J. and Bures, V., Environ. Sci. Technol. 32, 843 (1998).

73. Goodarzi, F. and Huggins, F. E., J. Environ. Monitor., 3,1 (2001).

74. Petit, M. D. and Rucandio, M. I., Anal. Chim. Acta, 401,283 (1999).

75. Austin, D. E. and Newland, L. W., Chemosphere, 14, 41(1985).

76. Polyak, K. and Hlavay, J., Fresenius’ J. Anal. Chem., 371,838, (2001).

77. Field-Flow Fractionation, http://www.analytical.chem-istry.helsinki.fi/research/instruments/fff

78. Schimpf, M., Caldwell, K. and Giddings, J. C., Field-Flow Fractionation Handbook , John Wiley & Sons, Inc.,New York, 2000.

79. Taylor, H. E., Garbarino, J. R., Murphy, D. M. and Beck-ett, R., Anal. Chem., 64, 2036 (1992).

80. Hassellov, M., Lyven, B., Haraldsson, C. and Siri-nawin, W., Anal. Chem., 71, 3497 (1999).

81. Hassellov, M., Lyven, B. and Beckett, R., Environ. Sci.Technol ., 33, 4528 (1999).

CHAPTER 4

Separation Techniques

4.1 Liquid Chromatography

Kathryn L. Ackley and Joseph A. CarusoUniversity of Cincinnati, Ohio, USA

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 1471.1 Variables affecting a liquid

chromatography separation . . . . . . . . 1481.2 Characterizing chromatographic

separations . . . . . . . . . . . . . . . . . . . . 1492 Liquid Chromatographic Stationary

Phases . . . . . . . . . . . . . . . . . . . . . . . . . . . 1512.1 Normal phase chromatography

(NPC) . . . . . . . . . . . . . . . . . . . . . . . 1512.2 Reversed phase chromatography

(RPC) . . . . . . . . . . . . . . . . . . . . . . . 1512.3 Reversed phase ion pair

chromatography (IPC) . . . . . . . . . . . 1522.4 Micellar chromatography . . . . . . . . . 153

2.5 Ion exchange chromatography (IEC) 1532.6 Size exclusion chromatography (SEC) 1552.7 Chiral liquid chromatography . . . . . . 1552.8 Micro liquid chromatography . . . . . . 157

3 Selecting a Mobile Phase Suitable forLiquid Chromatography withElement-specific Detection . . . . . . . . . . . . 158

4 Detectors Used for Elemental Speciationwith Liquid Chromatography . . . . . . . . . . 158

5 Interfacing Liquid Chromatographywith Detectors Used for ElementalSpeciation . . . . . . . . . . . . . . . . . . . . . . . . 160

6 Conclusions and Future Directions . . . . . . 1607 References . . . . . . . . . . . . . . . . . . . . . . . . 161

1 INTRODUCTION

Separation techniques are key components inelemental speciation analyses. Many sensitiveelement-specific detectors exist that can rapidlyprovide total element information, but it is notuntil a separation technique is coupled withan element-specific detector that the variousforms of a particular element in a sample canbe determined. A variety of separation tech-niques have been employed in speciation analysesincluding gas chromatography (GC), liquid chro-

matography (LC), capillary electrophoresis (CE),and supercritical fluid chromatography (SFC). LC,however, has emerged as one of the most pop-ular separation techniques for elemental specia-tion analysis.

Liquid chromatographic separations are carriedout by introducing sample onto a chromatographiccolumn filled with a solid stationary phase whilea liquid mobile phase is continuously pumpedthrough the column. Early LC stationary phaseswere made up of a liquid stationary phase coatedonto a solid support, but modern LC stationary


148 SEPARATION TECHNIQUES

phases are usually comprised of chemically modi-fied silica or polymers. The analytes in the sampleinteract with both the stationary and the mobilephases while passing through the column, and theextent to which each of the analytes interacts withthese phases determines the length of time eachanalyte resides in the column. Thus, separation isachieved when each of the analytes interacts withthe phases to a different extent, and exits the col-umn at different times.

LC has a number of advantages that make itan attractive choice for speciation analyses. LC,unlike GC, is capable of separating non-volatilecompounds as well as those that decompose atelevated temperatures. LC is an extremely versatiletechnique since both the stationary phase andthe mobile phase may be altered to achievethe desired separation, and an enormous varietyof stationary phases are commercially available.Furthermore, separations can be further enhancedby the addition of additives to the mobile phase.Usually, minimum sample preparation is required,and LC systems are readily interfaced to element-specific detectors such as inductively coupledplasma-mass spectrometers (ICP-MS).

Liquid chromatography is the general termgiven to chromatographic separations with a liquidmobile phase, but the majority of LC separationsperformed currently are part of a subset of LCknown as high performance liquid chromatography(HPLC). HPLC columns differ from earlier LCcolumns because the stationary phase particleshave smaller diameters. Typical diameters forparticles in commercially available HPLC columnsare 3–5 µm. The mobile phase is pumped athigh pressure through the stationary phase. Theincreased pressure is a result of pumping aviscous liquid through a column packed withparticles of small diameter. Thus, HPLC is oftenerroneously referred to as ‘high pressure liquidchromatography’. The LC work described in thischapter will deal with HPLC separations unlessotherwise noted. HPLC separations can be furthersubdivided by the general type of stationary phasethat is used in the separation. The focus of thischapter is not to offer a comprehensive discussionof the theory of LC but rather to provide the reader

with an overview of the role of LC in elementalspeciation analyses.

1.1 Variables affecting a liquidchromatography separation

A variety of variables may be adjusted to optimizean LC separation. The nature of the chemicalspecies to be separated must first be considered.The interactions between an analyte and thestationary and mobile phases are based upon dipoleforces, electrostatic interactions, and dispersionforces. Thus, knowledge of the analyte’s structureis useful in predicting how it will behave during agiven separation. Generally, compounds that arequite different in polarity or chemical structurewill be much easier to separate than compoundsthat are more similar. (e.g. dibutyl tin chlorideand tetraphenyl tin would be easier to separatethan dibutyl tin chloride and tributyl tin chloride).The selection of the stationary phase to be usedin an LC separation is usually based upon thenature of the analytes that are to be separated.For example, a mixture of anionic species wouldtypically be separated using a column with ananion exchange stationary phase. The nature of thespecies to be separated may be altered prior to theseparation by derivitization to facilitate separationand/or detection [1, 2].

The mobile phase composition is a criticalvariable when optimizing a separation. The mobilephase may consist of one component such asmethanol, or it may be a mixture of solventsand/or aqueous solutions such as buffers. Thesample must be soluble in the mobile phase toprevent the precipitation of sample within thecolumn. Buffers may be added to the mobilephase to control pH, and other solutes such aschiral additives, ion pair reagents, and surfactantsmay be added to enhance a separation. Thecomposition of the mobile phase may also bealtered during the separation. Isocratic separationsare those in which the mobile phase composition isheld constant throughout the separation, but whengradient elution is used, the composition of themobile phase is changed during the course of theseparation. For example, a separation may begin

INTRODUCTION 149

140 C

B

A

D

20

40 328 12 16 20

Retention time/min

24 28

40Inte

nsity

(ar

bitr

ary

units

)

Con

cent

ratio

n of

met

hano

l (%

)

60

80

100

120

100

40

30

50

60

70

80

90

Figure 4.1.1. Separation of 5 ppm of: A, Pb(II); B, triethyllead chloride; and C, triphenyllead chloride; and 100 mg L−1 of D,tetraethyllead. A Nucleosil C18 column was used with a mobile phase of 8 mM sodium pentane sulfonate at pH 3. Gradientelution was used. The methanol concentration was 40–90 % over 10 min, held at 90 % methanol for 20 min. ICP-MS detectionwas used [12]. (Reproduced by permission of the Royal Society of Chemistry.)

with a mobile phase consisting of 100 % water.Over the course of the separation, the mobilephase composition may uniformly change over aperiod of 15 min from 100 % aqueous to 50 %aqueous and 50 % methanol. Gradient separationsare often utilized to decrease the time required toperform a separation or to improve resolution andpeak shapes. Figure 4.1.1 shows a chromatogramobtained during a gradient separation.

The flow rate of the mobile phase is also asignificant variable in the LC separation, for it hasa major effect upon the time required to completea separation. Increasing the flow rate will typicallydecrease the separation time, but it will increase thepressure inside the column, as more liquid is beingforced through the spaces in the stationary phase.One must also consider the method of detectionwhen adjusting the flow rate. Frequently HPLCsystems are interfaced to element-specific detectorsto perform speciation analyses, and the ability ofthe detector to accommodate the increased effluentflow rate must be considered.

Temperature may also impact an LC separation.Fluctuations in temperature can cause changes inan analyte’s retention time. Thus, HPLC columnsare frequently housed in water jackets or ovens to

keep the column temperature constant. Generally,increases in temperature cause a decrease inanalyte retention time because the mobile phaseviscosity is reduced and the rates of diffusionare increased.

1.2 Characterizing chromatographicseparations

A basic understanding of the way chromatographicseparations are characterized is helpful whenstudying LC separations utilized in speciationanalyses, and a cursory explanation of severalsignificant equations used to characterize LCseparations is presented here.

The term ‘retention time’ refers to the amountof time it takes for an analyte to pass throughthe column and reach the detector. An unknowncompound’s retention time in a chromatographicsystem may aid in the compound’s identification.For example, if a plant extract is being analyzedto determine the selenoamino acids present inthe sample, and a peak is observed that hasthe same retention time as selenomethionine, theanalyst knows that the sample either containsselenomethionine or another species with the same


retention time as selenomethionine. The unknownpeak, however, could not be attributed to any ofthe other selenoamino acids that have a differentretention time.

The capacity factor, usually denoted k′, isa dimensionless parameter that represents thenormalized retention time for an analyte to takeinto account nonchromatographic contributions tothe retention time such as the distance from thecolumn to the detector. The capacity factor iscalculated using equation (4.1.1)

k′ = (tr − to)

to(4.1.1)

where tr is the analyte’s retention time and to isthe time it takes for an analyte unretained by thestationary phase to pass through the column andreach the detector.

The selectivity of a separation, denoted α,refers to how well the chromatographic system isable to distinguish between two different analytes.The selectivity is the ratio of the two analytes’capacity factors.

α = k′1

k′2

(4.1.2)

The resolution, Rs, refers to the efficiency of aseparation. The resolution is calculated by dividingthe distance between two chromatographic peaksby the average of their widths at the base.

Rs = �tr12 (wt1 + wt2)

(4.1.3)

Analyte peaks in LC separations have Gaussianpeak shapes caused by longitudinal diffusion, dis-persive effects, and the fact that the partitioningof the analyte between the stationary phase andthe mobile phase is not an instantaneous process.A mathematical expression for the symmetry of achromatographic peak is the peak asymmetry fac-tor, As. The peak asymmetry factor is determinedby drawing a vertical line from the tallest pointof the peak to the base of the peak. At 10 % ofthe peak’s height, the width of the peak on theleft side of the line is measured along with thewidth of the peak on the right side. The ratio of

the peak ‘half-widths’ is called the peak asymme-try factor. ‘Fronting’ occurs when the front portionof the peak is much wider than the back portion ofthe peak. ‘Tailing’ occurs when the back portionof the peak is much wider than the front por-tion of the peak. Figure 4.1.2 provides examplesof these types of peaks. Ideally, peaks obtained inLC would have a peak asymmetry factor of 1 andwould be as narrow as possible.

Two related terms that are frequently used inLC are plate height, H , and plate count, N .These terms relate to a column’s chromatographicefficiency and were derived from theoretical workthat treated chromatographic columns as if theywere similar to distillation columns [3]. These

(a)

(b)

(c)

Figure 4.1.2. (a) Ideal chromatographic peak. (b) Chromato-graphic peak exhibiting ‘fronting’. (c) Chromatographic peakexhibiting ‘tailing’.

LIQUID CHROMATOGRAPHIC STATIONARY PHASES 151

terms are theoretical in nature, and there areno plates actually inside of a chromatographiccolumn. The plate height is determined by dividingthe length of the stationary phase, L, by theplate count.

H = L

N(4.1.4)

Plate count is frequently used to assess columnperformance. There are multiple ways to calculateN . One method is to determine the width of a peakat half its maximum height, W1/2, and substitutethat value into equation (4.1.5).

N = 5.54 t2r

W 21/2

(4.1.5)

Regardless of the method used to calculate N , forthis discussion it is sufficient to state that columnswith a large N value are more efficient than thosewith smaller N values.

2 LIQUID CHROMATOGRAPHICSTATIONARY PHASES

2.1 Normal phase chromatography(NPC)

Normal phase chromatographic systems are com-prised of a polar stationary phase and a nonpo-lar mobile phase such as hexane. When chromato-graphic systems consisting of a nonpolar stationaryphase and a polar mobile phase were later devel-oped, they were given the name ‘reversed phase’since they were the reverse of the established ‘nor-mal phase’ systems. Unmodified silica or aluminawere frequently used in early NPC work, but thepeak shapes obtained with these stationary phaseswere often broad with basic compounds exhibit-ing tailing and retention times that were difficult toreproduce. These problems were alleviated with thedevelopment of bonded stationary phases that have apolar functional group (cyano, diol, etc.) chemicallybonded to the silica. Analytes separate in NPC asthey are reversibly adsorbed by the polar functionalgroups of the stationary phase. One of the principaladvantages of this technique is that it allows analytesthat are insoluble in polar solvents to be separated.

NPC has seen limited use in the area ofelemental speciation. The major limitation of thistechnique is the nonpolar mobile phase, whichis incompatible with most elemental detectorsused in speciation analyses. Furthermore, manyanalytes that are separated using NPC may alsobe separated using reversed phase chromatography,which is compatible with elemental detectors.Nonetheless, examples of the use of NPC forelemental speciation can be found in the literature.Xu and Lesage used an aminopropyl column toseparate vanadyl and nickel petroporphyrins [4].The separation mechanisms were determined tobe hydrogen bonding as well as Van der Waalsinteractions between the petroporphyrins and theamino groups in the stationary phase. Hexane,toluene, and dichloromethane were componentsin the mobile phase and a fluorescence detectorwas used. NPC was used to separate organotinpesticides [5]. A column with a cyanopropyl-bonded silica stationary phase was used for theseparation. Separated analytes were subjected toUV photoconversion, post-column complexation,and fluorescence detection. The method was usedto measure triphenyltin acetate in water adjacentto a potato field sprayed by the pesticide.

2.2 Reversed phase chromatography(RPC)

RPC is one of the most widely used LC tech-niques in elemental speciation. It is used to sep-arate nonpolar and/or slightly polar species. Thepolar mobile phases used are typically aqueousor a mixture of water and an organic modifiersuch as methanol or acetonitrile. The stationaryphase is most commonly silica that has been mod-ified through silanization. The silanol–OH groupspresent on the silica surface are replaced by alkylchains creating a nonpolar stationary phase suitablefor reversed phase separations. Chains with 18, 8,or 2 carbon atoms are the most common. Separa-tion is based on the hydrophobicities of the specieswith compounds that are the most hydrophobiceluting the latest. Separations may be manipu-lated by changing a variety of variables includingthe stationary phase functional group, pH, ionic


strength, organic modifier(s) used in the mobilephase, and the gradient program. The pH of theeluent is an important parameter in RPC. In manyinstances, the pH of the mobile phase dictates ifa compound is protonated or deprotonated. Thisaffects the charge of the analyte and its retentionon the column. Traditionally, silica-based station-ary phases cannot tolerate eluent pHs below 2 orabove 7 since cleavage or hydrolysis of the station-ary phase may occur. Recently, column manufac-turers have produced silica-based reversed phasestationary phases that can tolerate eluent pHs of 2to 10. Reversed phase stationary phases may alsobe made of polymeric materials, but silica-basedcolumns are still the most commonly used.

RPC has been widely used in elemental specia-tion to separate organometallic species. It has beenutilized in the speciation of metal porphyrins [6],in speciation studies of platinum in chemother-apy drugs [7], in the determination of telluriumcompounds in wastewater [8], and the separa-tion of organotin compounds [9, 10]. Zoorob andCaruso [11] utilized a column with an octadecylstationary phase and a mobile phase consisting of80 % water and 20 % methanol to separate thechromium species in azo dye, see Figure 4.1.3.One of the dyes investigated was found to containuncomplexed and potentially bioavailable Cr(III).Many more examples of RPC used for elementalspeciation can be found in the literature. However,

01

50 100

Time (seconds)

150 200

6

5

4

3

Inte

nsity

(10

00 c

/s)

2

Cr DyeCr Dye

Cr(III)

Figure 4.1.3. Reversed phase separation of Acid Blue 193, acommonly used chromium azo dye. The mobile phase com-position was water–methanol (80 : 20, v/v). ICP-MS detectionwas used. [11] (Reprinted from Journal of Chromatography,Vol. 773, Zoorob and Caruso, Speciation of chromium dyes byhigh performance liquid . . ., pp. 157–162, 1997, with permis-sion from Elsevier Science.)

these selected examples illustrate the breadth ofanalytical problems to which this technique canbe applied.

2.3 Reversed phase ion pairchromatography (IPC)

Reversed phase ion pair chromatography is similarto RPC in that RPC stationary phases are used,but an ion pair reagent is added to the mobilephase. An ion pair reagent is a salt with a cationor anion having a polar head group and a nonpolartail. Examples of ion pair reagents include sodiumalkyl sulfonates, tetraalkyl ammonium salts, andtriethylalkyl ammonium salts. The concentration ofion pair reagent in the mobile phase is typicallybetween 0.001 and 0.005 M. IPC’s popularityresults from its ability to simultaneously separateanions, cations, and noncharged species.

The separation mechanism involved in IPC isnot totally understood. One widely held theory isthat an ionic analyte is electrostatically attractedto the charged end of the ion pair reagent, andan ‘ion pair’ is formed. The charge neutralizationcoupled with the nonpolar tail of the ion pairreagent causes the charged analyte to be retainedby the nonpolar stationary phase. Another theory isthat the hydrophobic portion of the ion pair reagentadsorbs to the stationary phase with the chargedportion of the ion pair reagent exposed to passinganalytes. Thus, a pseudo ion exchange stationaryphase is formed causing charged species that wouldnot have otherwise been retained on the columnto interact with the stationary phase. More thanlikely, both of these mechanisms occur during anIPC separation.

IPC has been used to speciate Pb(II), tri-ethyllead chloride, triphenyllead chloride, andtetraethyllead chloride [12]. Sodium pentane sul-fonate was the ion pair reagent. At an ion pairconcentration of 2 mM, the inorganic lead and thetriethyllead peaks overlapped significantly. At anion pair concentration of 8 mM, the inorganic leadand triethyl lead were nearly totally resolved (seeFigure 4.1.1). This separation was coupled withICP-MS detection for the analysis of a leadedfuel standard reference material. This separation


illustrates how IPC can successfully be used toseparate inorganic and organometallic species thatare neutral or charged.

Another example of the successful applica-tion of IPC is the work of Jiang and Houk [13].They developed separations for inorganic phos-phates, adenosine phosphates, inorganic sulfates,and amino acids. Separations were performed withreversed phase stationary phases, tetraalkylammo-nium salts as ion pairing reagents, and smallamounts of organic modifier (less than 5 %).

2.4 Micellar chromatography

Micellar chromatography is similar to IPC. A surfac-tant is added to the mobile phase at a concentrationabove the critical micelle concentration (CMC) sothat micelles are formed in the mobile phase. TheCMC is characteristic of each surfactant. Below thisconcentration, surfactant molecules tend to adsorb atsurfaces, as in IPC, to minimize solvophobic inter-actions. At surfactant concentrations greater thanthe CMC, the surfactant molecules come togetherto form micelles that have the hydrophobic por-tion of the surfactant oriented inward and the polar,hydrophilic portion oriented outward towards thepolar mobile phase. Figure 4.1.4 shows the struc-ture of a micelle. Micelles can be anionic, cationic,uncharged, or zwitterionic. Analytes may partition

Hydrophobic portion of surfactant molecule

Polar portion of surfactant molecule

Figure 4.1.4. Structure of the micelle.

inside of the stationary phase, mobile phase, andmicelles, making the separation of charged, neu-tral, hydrophobic, and hydrophilic species possible.Jimenez and Marina [14] have written a review ofretention modeling in micellar liquid chromatogra-phy. Hydrophobic analytes that may not be solublein the polar mobile phases used in RPC may besoluble in micellar chromatography eluents sincethey may partition into the interior of the hydropho-bic micelle.

Ding et al. [15] utilized micellar chromatogra-phy for the speciation of dimethyl arsenic acid(DMA), monomethyl arsonic acid (MMA), As(III),and As(V) in urine. Micellar chromatography wasselected because proteins found in the urine samplewere dissolved by the micelles, so they eluted inthe void volume. This was advantageous becauseclinical samples typically need to be deproteinizedto avoid proteins that are insoluble in RPC mobilephases from precipitating inside the LC column.Figure 4.1.5 shows a chromatogram of arsenicspecies present in urine. Another advantage of thisseparation was that the chloride ions in the sam-ple were separated from the arsenic species. AnICP-MS was used as the chromatographic detector,and chloride ions interfere with the determinationof arsenic by ICP-MS. By separating the chlo-ride ions chromatographically, the interference canbe eliminated.

Micellar chromatography has been success-fully used to separate alkyltin compounds usinga 0.1 M sodium dodecyl sulfate (SDS) micellarmobile phase and a C-18 stationary phase [16].Three surfactants were studied–SDS which isnegatively charged, dodecyltrimethylammoniumbromide which is positively charged, and poly-oxyethylene(23)dodecanol which is nonionic. SDSwas found to separate the analytes. This was notsurprising since the organotin species investigatedwere cationic and would have more electrostaticinteractions with an anionic surfactant.

2.5 Ion exchange chromatography (IEC)

IEC is used to separate free ions and easily ion-izable species. IEC is a commonly used techniquein speciation analyses because metallic species of


00

10

20

30

Cou

nts

Tho

usan

ds

40

5

Time (min)

10 15

As(V)

2

1

DMA

As(III)

Figure 4.1.5. Chromatogram of arsenic species in urine. Peaks 1 and 2 are different forms of chloride that were visible sinceICP-MS detection was used. DMA = dimethylarsinic acid. A Hamilton PRP-1 column was used. This is a reversed phasepolymeric column. The mobile phase consisted of 0.05 M cetyltrimethylammonium bromide (CTAB) and 10 % methanol. ThepH was 10.2, and the column temperature was 40 ◦C [15]. (Reprinted from Journal of Chromatography, Vol. 694, Ding et al.,Arsenic speciation by micellar chromatography, pp. 425–431, 1995, with permission from Elsevier Science.)

interest frequently occur in the ionized form. Thestationary phase typically consists of an ionic func-tional group such as a quaternary ammonium groupor a sulfonate group bonded to a substrate such as apolystyrene–divinylbenzene polymer or silica. IECis frequently subdivided into cation exchange chro-matography and anion exchange chromatographydepending upon the functional groups present inthe stationary phase. The ionic sites on the station-ary phase have the opposite charge of the analytesto be separated. Counter ions in the mobile phasemaintain electrical neutrality within the chromato-graphic column. Analytes having a charge thatis opposite of that of the charge bearing func-tional group will interact with the stationary phaseelectrostatically. The retention time of an ana-lyte increases with increasing electrostatic force.Mobile phases used in IEC separations are typi-cally aqueous solutions of inorganic salts.

IEC is frequently utilized to separate arsenicspecies. Wang et al. [17] used an anion exchangecolumn to separate arsenite and arsenate in coalfly ash extracts. ICP-MS detection was used, andchloride ions, which hinder the detection of arsenicby ICP-MS, were separated from the analytes of

interest, thus eliminating the interference. Anionexchange columns have been used to separatearsenic species in other matrices such as soil [18]and fish extracts [19]. Just a few of the many otherapplications of anion exchange chromatographiccolumns include the separation of bromate andbromide in drinking water [20], the separation ofinorganic and organic antimony species [21], andthe separation of Cr(III) and Cr(VI) species [22].

Cation exchange chromatography is also used inelemental speciation. Suyani et al. [23] used cationexchange chromatography to separate trimethyltinchloride, tributyltin chloride, and triphenyltinacetate. In some instances, an anion exchangecolumn and a cation exchange column areconnected in series to allow for the determinationof both anionic and cationic species. Terasahdeet al. [24] successfully utilized this technique toseparate six arsenic species. Gradient elution wasused and the ionic strength of the mobile phaseincreased and the pH decreased over the course ofthe separation. The three mobile phase componentswere dilute nitric acid, water, and carbonate buffer.

IEC has the advantage of using aqueous buffers,which makes this technique compatible with element


selective detectors such as an ICP mass spectrom-eter. IEC is also utilized in speciation analyses forthe purpose of sample cleanup prior to analysis andfor sample preconcentration.

2.6 Size exclusion chromatography (SEC)

SEC is used to separate macromolecules (MW >

2000 Da) such as proteins, synthetic polymers, andbiopolymers according to their size. The separa-tion mechanism is not based on chemical interac-tions as in other types of LC but rather on theability of an analyte to penetrate the pores ofthe stationary phase. Small molecules will pen-etrate the pores of the stationary phase greaterthan larger molecules, so large molecules areeluted before small molecules. The mobile phasedoes not play a large role in the separation,and it is usually selected based on its ability tosolubilize the analytes being separated. SEC isoften subdivided into gel permeation chromatog-raphy and gel filtration chromatography. Gel fil-tration chromatography refers to the separationof water-soluble macromolecules, and gel perme-ation chromatography refers to the separation ofmacromolecules that are soluble in organic sol-vent. The SEC system must be calibrated withmolecules of known molecular weight that havesimilar physical properties to those of the analytesof interest.

SEC is used in the field of elemental speciationto study such things as metalloproteins [25, 26]and metabolites of metal-containing drugs [27]. Itmay also be used to separate species of interestfrom interfering low molecular weight componentsof the sample matrix. SEC was used to separateprotein-bound copper in serum from sodium andphosphate ions that interfered with the determina-tion of copper by ICP-MS [28]. Crews et al. [29]used SEC to investigate the cadmium-containingproteins in pig kidney following cooking and invitro gastrointestinal digestion. They found thatmost of the soluble cadmium in the pig kidneywas associated with a metallothionein-like protein.Klueppel et al. [30] used SEC with ICP-MS detec-tion to study platinum metabolites in plants. Grasswas cultivated with a platinum-containing solution,

and the extracted platinum species having a molec-ular mass less than 10 kDa were separated usingSEC. Not all of the metabolites could be resolvedusing SEC, but the inorganic platinum complexeswere separated from the organoplatinum species.

2.7 Chiral liquid chromatography

Chiral molecules have stereoisomers that arenonsuperimposable mirror images of each other(see Figure 4.1.6). These types of stereoisomersare referred to as enantiomers. The most commontype of chiral molecule has a tetrahedral carbonatom with four different groups attached to it.However, molecules can be chiral even if theydo not have an asymmetric carbon so long asthey are nonsuperimposable mirror images of oneanother. Chiral separations can be very difficultto achieve since most of the physical propertiesof enantiomers are the same. Yet performingchiral separations is of particular importance whenstudying pharmaceuticals since biological systemstend to be chiral systems with one enantiomer ofa drug having a different biological effect than itschiral counterpart in vivo.

Several strategies have been employed to carryout chiral separations including the use of chiralstationary phases, chiral derivatizing agents toform diasteriomers, and the use of chiral mobilephase additives. A drawback to the use of chiralmobile phase additives for LC is that the mobilephase flow rate associated with LC necessitateslarge amounts of chiral additive to equilibratethe column and perform the separation. Thus, theuse of chiral additives to separate enantiomersis primarily used in capillary electrophoresis andthin layer chromatography. Derivatization of chiral

H

CHO

COOH

CH3

H

COH

HOOC

H3C

Mirror Plane

Figure 4.1.6. Two isomers of a chiral molecule.


molecules with a chiral derivatizing agent ofknown optical purity may be performed prior toseparation by LC. The advantage of this techniqueis that after derivatization, the separation may beperformed with commercially available reversedphase LC columns, which are much less expensivethan LC columns with chiral stationary phases.However, finding a suitable derivatizing agent canbe difficult, and method validation to confirm thatthe enantiomeric ratio obtained after derivatizationis the same as the enantiomeric ratio in the sampleis problematic.

The third alternative is the use of LC columnsthat have chiral selectors immobilized on thesurface of the stationary phase. Commonly usedchiral stationary phases include cellulosic andamylosic, macrocyclic antibiotic, chiral crownether, ligand exchange, cyclodextrin, protein, andPirkle phases [31]. Separation occurs when thechiral analytes form transient diastereomeric com-plexes with the chiral selectors on the stationaryphase. LC columns with chiral stationary phasesare expensive and not all enantiomers can be sep-arated with the stationary phases available. Thus,the number of papers appearing in the literaturediscussing the use of chiral LC for elemental

speciation is limited. As the number of stationaryphases available grows, the number of papers onthis topic is expected to increase as well.

Chiral chromatography has primarily been usedin the field of elemental speciation to sepa-rate chiral selenoamino acids. Mendez et al. [32]separated selenomethionine enantiomers using aß-cyclodextrin column. O-Phthalaldehyde and 2,3-naphthalenedicarboxaldehyde are fluoroionogenicderivatizing reagents that were used to derivatizethe selenomethionine enantiomers prior to sepa-ration. Fluorimetric and on-line hydride genera-tion ICP-MS detection were used to detect theseparated enantiomers. Chiral selenoamino acidshave also been separated using a chiral crownether stationary phase [33, 34]. A 0.10 M perchlo-rate mobile phase was used, and no precolumnderivatization was necessary. Figure 4.1.7 showsthe separation of the selenoamino acids. Thisseparation was used to determine the seleniumspecies present in nutritional supplements [33] andselenium-enriched samples [34]. A teicoplanin-based chiral stationary phase was also used to sep-arate selenomethionine enantiomers in selenizedyeast [35]. Teicoplanin is a macrocyclic antibioticused to achieve chiral separations. Vancomycin is

0

0

10 20 30

Time/min

40 50 60

60000

50000

40000

Res

pons

e/co

unts

s−1

30000

20000

10000

1

3

2

4

5 6

7

Figure 4.1.7. Separation of selenoamino acids using a chiral crown ether column. The mobile phase consisted of0.1 M HClO4 at pH 1. The flow rate was 0.5 mL min−1 for 35 min, then it was increased to 1.0 mL min−1. Peaks:1 = L-selenocystine; 2 = L-selenomethionine; 3 = meso-selenocystine; 4 = D-selenocystine; 5 = D-selenomethionine; 6 =L-selenoethionine; and 7 = D-selenoethionine [33]. (Reproduced by permission of the Royal Society of Chemistry.)


another macrocyclic antibiotic frequently used forthe same purpose.

2.8 Micro liquid chromatography

Standard LC columns usually have internal diame-ters of 4.6 mm. However, analysts are increasinglyfinding the use of columns with smaller internaldiameters to be advantageous. While no singleconvention exists for what defines a micro LC col-umn, guidelines have been suggested in a reviewarticle by Vissers et al. [36]. Chromatography, per-formed with columns having internal diameters of0.5–1.0 mm, is referred to as micro LC. CapillaryLC generally describes separation with columnshaving internal diameters of 100–500 µm, andnanoscale LC refers to columns with internal diam-eters of 10–100 µm. Throughout this chapter, theterm micro LC will collectively refer to micro LC,capillary LC, and nanoscale LC.

Performing separations with micro scale LCcolumns has the advantage of reduced solventconsumption resulting from much smaller flowrates. This may also be advantageous when inter-facing micro LC columns with atomic spectrom-eters that do not tolerate mobile phases withhigh concentrations of salts or organic modifierssince smaller mobile phase flow rates are used.Because of the reduced column volume, less sam-ple is injected onto the column. The reducedsample requirements make micro LC attractivefor analyses when limited amounts of sampleare available.

Micro LC columns are frequently made bypacking fused silica capillaries. Reversed phase,ion exchange, and ion pairing separations canall be carried out using micro LC columns.The small flow rate and high pressure dropper unit length of small diameter columns posechallenges when selecting/developing a solventdelivery system [37]. Reciprocating and syringepump systems capable of delivering mobile phasesat flow rates on the order of 50–150 µL min−1 arecommercially available [36]. For lower flow rates,split-flow techniques may be utilized.

Care must be taken when interfacing micro LCcolumns to the selected detector to minimize band

broadening. Analytes separated by the column exitthe column in a narrow band of the mobile phase.Analyte band broadening occurs as the separatedanalyte continues to diffuse in the mobile phaseadjacent to the analyte band. Band broadeningcauses wider peaks with decreased peak heightresulting in a decrease in resolution and an increasein detection limits. To help reduce this problem,tubing connecting the column to the detectorshould have an internal diameter at least as smallas the chromatographic column. Frequently, directinjection nebulization is employed to interfacemicro LC columns to detectors such as inductivelycoupled plasma-mass spectrometers to minimizethe band broadening that occurs when the columneffluent is nebulized into a spray chamber [38–40].The use of micro LC columns coupled to atomicspectrometers for trace elemental speciation hasbeen reviewed by Garraud et al. [41].

Pergantis et al. [42] used micro LC for the spe-ciation of arsenic animal feed additives. Threereversed phase columns were investigated, a con-ventional column with a 4.6 mm i.d., a microborecolumn with a 1 mm id, and a micro column witha 0.32 mm i.d. (The nomenclature used in ref. 42was used to refer to the columns rather than thenomenclature described earlier in this section ofthe chapter.) Graphite furnace-atomic absorptionspectroscopy (GF-AAS) was one of the detectionmethods used. Fractions were collected at 30 sintervals, and the fractions were analyzed by GF-AAS. When GF-AAS was used for the detection,the limit of detection for the micro LC columnwas three orders of magnitude lower than for theconventional column. The lower limit of detec-tion resulted from all of the micro LC fraction(5 µL) being injected into the graphite furnace.Only 2 % of the conventional LC fraction (500 µL)could be injected into the graphite furnace. How-ever, because the total sample volume for eachLC fraction was used in a single GF-AAS anal-ysis, duplicate measurements could not be made.Thus, the analysts found the microbore columnoffered the best compromise. They also foundthe flow rates with the micro LC column wereeasily accommodated by continuous flow-liquid


secondary ion–mass spectrometry and direct liq-uid introduction-mass spectrometry. This researchillustrates how micro LC separations can be moreadvantageous than separations with conventionalcolumns for some applications.

Other applications of micro LC include thespeciation of organolead compounds [38], leadand mercury compounds [39], and the speciationof chromium [40].

3 SELECTING A MOBILE PHASESUITABLE FOR LIQUIDCHROMATOGRAPHY WITHELEMENT-SPECIFIC DETECTION

When selecting a mobile phase, one must considerthe type of detector being utilized. Element-specific detection methods such as ICP-MS, FAA,and ICP-AES are coupled with LC to performelemental speciation analyses, and one must selecta mobile phase that will not only achieve thedesired separation but will also be compatible withthe detector being used.

Liquid sample is aspirated into the flame, in thecase of FAA, or the plasma, in the case of ICP-MSand ICP-AES, by a nebulizer. Mobile phases witha high salt content (>0.2 % total dissolved solid)can clog the nebulizer, and should be avoided [43].Organic solvents such as methanol and acetonitrileare frequently used in LC especially in RPC.However, special precautions must be taken whenintroducing organic solvent into an ICP. Organicsolvents cause plasma instability and may causehigh reflective powers or extinguish the plasmaall together. Soot may form on the samplerand skimmer cones of the ICP-MS. Even smallamounts of organic solvent may have negativeeffects on the detector. Olesik and Moore [44]reported that both atom and ion emission signalswere depressed when organic solvents (<2 % v/v)were present with ICP-AES. Despite this, the useof organic solvents cannot totally be eliminatedin LC separations since they are often necessaryto achieve a desired separation. Devices such aswater-cooled spray chambers and Peltier coolersmay be used to minimize the amount of solvent

reaching the plasma. Furthermore, plasmas maybe operated at increased RF powers to prevent theplasma from being extinguished.

Additional consideration must be made whenusing gradient elution. Gradient elution causesplasma or flame conditions to change during thecourse of the separation as the composition ofthe mobile phase pumped to the plasma/flamechanges. This effect is dependent upon the natureof the gradient and the solvents utilized. How-ever, analysts should be aware that gradientseparations may cause changes in the ioniza-tion/atomization source.

4 DETECTORS USED FORELEMENTAL SPECIATION WITHLIQUID CHROMATOGRAPHY

Element-specific detectors are the most commonlyused detectors for LC separations when performingspeciation analyses. UV and diode array detectors,which are typically used to detect separated organiccompounds, are not particularly useful in elementalspeciation analyses because the analytes of interestfrequently do not absorb UV light. Also, manycompounds from the sample matrix may absorbin the UV causing interferences.

Sensitive, element-specific detectors are neededfor elemental analysis. The sample concentrationof analytes of interest in biological and environ-mental samples is often in the µg L−1 or ng L−1

range, so detectors sensitive enough to detectlow levels of analyte are desirable. Methods ofdetection commonly used for elemental speciationanalyses include flame atomic absorption spec-trometry (FAAS), graphite furnace atomic absorp-tion spectrometry (GF-AAS), inductively coupledplasma-atomic emission spectrometry (ICP-AES),and inductively coupled plasma-mass spectrometry(ICP-MS).

FAAS is one of the most widely used methodsfor the analysis of single elements in samples.As methods were developed to perform elementalspeciation, FAAS was employed for element-specific detection. FAAS instruments have theadvantage of being inexpensive, relative to other

DETECTORS FOR ELEMENTAL SPECIATION WITH LC 159

detectors such as mass spectrometers, and they areavailable in most laboratories. Also, FAAS unitsrequire liquid sample be continuously nebulizedinto the flame making it compatible with LCsystems. However, the chief drawback of FAAS isthat it is not sensitive enough for most elementalspeciation applications. To improve sensitivity ofFAAS detection, work has been done to improvethe nebulization process.

GF-AAS is more sensitive than FAAS becausethe flame atomizer is replaced by an electrothermalatomizer which is often referred to as a graphitefurnace. A few microliters of sample are placedin the atomizer. The sample is first evaporatedat low temperature and then ashed at highertemperature in the atomizer. Then, the temperatureis rapidly increased to 2000–3000 ◦C, causingrapid atomization of the sample. The atomicabsorption is measured in the region above theheated surface [45]. Unlike FAAS, sample is notcontinuously aspirated into the graphite furnacemaking it difficult to interface this techniquewith LC. Typically when GF-AAS is used forLC detection, fractions of the LC effluent arecollected for subsequent analysis by GF-AAS. Thechromatograms obtained with LC-GF-AAS appearslightly different from chromatograms obtainedwhen using detectors that allow continuous sampleaspiration. Figure 4.1.8 shows an example of achromatogram obtained when a GF-AAS was usedfor LC detection.

Cold vapor atomization is a special atomizationtechnique used for the determination of mercury byAAS. Detection limits in the microgram per literrange can be realized with this technique. Mercuryis oxidized to Hg2+ with sulfuric and nitric acidfollowed by reduction back to metallic mercurywith SnCl2. Gas is bubbled through the reactionmixture sweeping the metallic mercury into anabsorption tube.

ICP-AES is frequently used for detection inelemental speciation analyses. The liquid samplestream is converted to an aerosol by the neb-ulizer. The aerosol then passes through a baf-fled spray chamber that allows only the smallestaerosol droplets to reach the plasma. Between 1and 5 % of the original sample actually reaches

0.0

20 4

Retention Time (min.)

Pea

k A

rea

6 8 10 12 14 16 18

20 4

Retention Time (min.)

6 8 10 12 14 16 18

0.3

0.05

0.10

0.15

0

0.1

0.2

0.3

0.2

0.1

I

II III

(A)

(B)

(C)

Figure 4.1.8. Chromatograms of separations obtained onreversed phase columns with GF-AAS detection. (a) Chroma-togram obtained using a conventional LC column withcollection of 500 µL fractions. (b) Chromatogram obtainedusing microbore LC column with collection of 40 µLfractions. (c) Chromatogram obtained using a micro LCcolumn with collection of 5 µL fractions. I = p-arsanilic acid;II = 3-nitro-4-hydroxyphenylarsonic acid; III = 4-nitrophenyl-arsonic acid [42]. (Reprinted from Journal of Chromatography,Vol. 764, Pergantis et al., Liquid chromatography and massspectrometry, pp. 211–222, 1997, with permission fromElsevier Science.)

the plasma. Detection limits for ICP-AES are typ-ically in the range of mg L−1 or high µg L−1.These detection limits are often not low enoughfor many elemental speciation applications. Toimprove the sensitivity of ICP-AES, work hasbeen done to improve sample transport efficien-cies. Direct injection nebulization and hydride gen-eration have been investigated to improve limitsof detection.

ICP-MS is probably the most widely used LCdetection method for elemental speciation analy-ses because of its superior sensitivity and because


it is easily interfaced to LC systems. The use ofan ICP-MS instrument as a detector for LC hasbeen reviewed by Sutton and Caruso [46] ICP-MSsystems are more expensive than FAAS and ICP-AES systems, but they offer detection limits inthe nanogram per liter range. The sample introduc-tion system for ICP-MS is nearly identical to thesample introduction system for ICP-AES, so likeICP-AES, ICP-MS suffers from poor sample trans-port efficiency. However, ICP-MS offers superiorlimits of detection because of the large signal-to-noise ratios that result from the use of a quadrupolemass analyzer. To improve sensitivity, work hasbeen done to try to improve the sample transportefficiency. Techniques such as direct injection neb-ulization, ultrasonic nebulization, and thermospraynebulization have been used, with some success,to improve sensitivity. Micronebulizers such as thehigh efficiency nebulizer and the microconcentricnebulizer have been used to improve sensitivitywhen introducing sample at very low flow rates.Pneumatic nebulizers such as the glass concentricnebulizer and the cross flow nebulizer still remainthe most commonly used nebulizers to introduceliquid sample into an ICP.

5 INTERFACING LIQUIDCHROMATOGRAPHY WITHDETECTORS USED FOR ELEMENTALSPECIATION

One of the advantages of utilizing LC for elementalspeciation analyses is the ease with which LCsystems can be interfaced to element-specificdetectors. In detectors based on ICP-MS, ICP-AES, and FAAS sample is continuously aspiratedinto the plasma at a flow rate comparable tothe flow rates typically utilized in LC analyses(0.5–2.0 mL min−1). Interfacing the LC system tothese detectors is easily achieved using a length ofinert tubing such as polyetherether ketone (PEEK)tubing to connect the end of the LC column withthe sample nebulizer. Care should be taken tominimize the length and internal diameter of thetubing to minimize band broadening.

Interfacing LC systems to detectors based onGF-AAS that introduce sample in discrete sample

volumes is more difficult. Fractions of the columneffluent may be collected for subsequent analysisby GF-AAS. Another approach is to completelyvolatilize the sample effluent prior to introductioninto the graphite furnace.

Hydride generation is a technique that has beenused to improve the sensitivity of FAAS, GF-AAS, ICP-AES, and ICP-MS in elemental specia-tion studies. If the species of interest are capableof forming volatile hydrides, the column efflu-ent may be subjected to reaction with sodiumborohydride converting the separated analytes intovolatile hydrides which may then be swept tothe flame, plasma, or graphite furnace. Only theelements arsenic, bismuth, germanium, lead, anti-mony, selenium, tin, and tellurium are capableof forming volatile hydrides. Hydride generationimproves a method’s sensitivity since the ana-lytes are separated from much of the samplematrix. It also allows the sample to be intro-duced to the detector in gaseous form so energyis not required to volatilize the sample. Sam-ple transport efficiencies are also dramaticallyimproved. The use of hydride generation in chem-ical speciation with atomic spectrometry has beenreviewed by Nakahara [47]. One limitation ofhydride generation is that all of the species ofinterest for a particular element may not formvolatile hydrides. For example, arsenate readilyforms a volatile hydride, but arsenobetaine andarsenosugars, which are important environmentalarsenic species, do not. To overcome this prob-lem, after the species are separated chromato-graphically, they may be decomposed on-line priorto hydride generation. Le et al. [48] used on-line microwave decomposition of the LC columneffluent to rapidly decompose organoarsenic com-pounds to arsenate. Once in the form of arse-nate, volatile hydrides were formed that weredetected by AAS.

6 CONCLUSIONS AND FUTUREDIRECTIONS

LC coupled with atomic spectrometric detectionwill continue to be one of the mainstays ofelemental speciation analysis. LC offers the analyst

REFERENCES 161

tremendous flexibility and a technique that isgenerally compatible with and easily interfaced toatomic spectrometric detectors. Future research islikely to be seen in the areas of chiral LC, SEC,and micro LC. As more chiral stationary phasesbecome commercially available, this technique willbe utilized to a greater extent. SEC has alreadybeen used significantly in the area of elementalspeciation, but as more focus is being placedon the role of metals in biological systems andan understanding of metalloproteins, SEC willbe utilized to an even greater extent. Micro LCwill become increasingly popular particularly asmicronebulization techniques improve. Interfacingmicro LC columns to atomic spectrometers hasposed challenges in the past, but advances inthe development of micronebulizers have madeinterfacing the two techniques easier. The reducedsolvent consumption and sample requirementsmake micro LC an attractive technique.

Early work in elemental speciation involvingLC utilized destructive detection methods that pro-vided elemental information about analytes but notstructural information. Coupling LC columns tomass spectrometers that utilize ‘soft’ ionizationtechniques allows the analyst to obtain structuralinformation. Examples of ‘soft’ ionization tech-niques include electrospray ionization and tunableplasmas. Frequently unidentified metal-containingspecies are found in environmental and clini-cal samples during a speciation analysis. Oftenthese species are unknown metabolites, and learn-ing their identities may help researchers under-stand how an organism metabolizes a drug orreacts to a pollutant. Because of this, the useof detectors capable of providing structural infor-mation will continue to increase. These detectorsare discussed in detail in subsequent chapters ofthis book.

It should be realized that frequently thereare multiple LC techniques capable of solvingan analytical problem, and rarely is there onlyone set of conditions capable of achieving aseparation. To understand this point, one need onlyto look at a recent review on the use of LCfor the speciation of organotin compounds [49].Dozens of chromatographic conditions have been

listed from the literature for the separation oforganotin species. The versatility offered by LCwill continue to make it one of the major separationtechniques used for elemental speciation in thedecade ahead.

7 REFERENCES

1. Toyo’oka, T. (Ed.), Modern Derivatization Methods forSeparation Sciences , John Wiley & Sons, Ltd, Chichester,1999.

2. Liu, W. and Lee, H. K., J. Chromatogr. A, 834, 45(1999).

3. Martin, A. J. P. and Synge, R. L. M., Biochem. J., 35,1358 (1941).

4. Xu, H. and Lesage, S., J. Chromatogr., 607, 139 (1992).5. Stab, J. A., Rozing, M. J. M., van Hattum, B., Cofino,

W. P. and Brinkman, U. A. T., J. Chromatogr., 609, 195(1992).

6. Rivaro, P. and Frache, R., Analyst , 122, 1069 (1997).7. Cairns, W. R. L., Ebdon, L. and Hill, S. J., Fresenius’ J.

Anal. Chem., 355, 202 (1996).8. Klinkenberg, H., van der Wal, S., Frusch, J., Terwint, L.

and Beeren, T., At. Spectrom., 11, 198 (1990).9. Rivas, C., Ebdon, L., Evans, E. H. and Hill, S. J., Appl.

Organomet. Chem., 10, 61 (1996).10. Dauchy, X., Cottier, R., Batel, A., Jeannot, R., Bor-

sier, M., Astruc, A. and Astruc, M., J. Chromatogr. Sci.,31, 416 (1993).

11. Zoorob, G. K. and Caruso, J. A., J. Chromatogr. A, 773,157 (1997).

12. AL-Rashdan, A., Vela, N. P., Caruso, J. A. and Heitkem-per, D. T., J. Anal. At. Spectrom., 7, 551 (1992).

13. Jiang, S. and Houk, R. S., Spectrochim. Acta , 43B, 405(1988).

14. Jimenez, O. and Marina, M. L., J. Chromatogr. A., 780,149 (1997).

15. Ding, H., Wang, J., Dorsey, J. G. and Caruso, J. A., J.Chromatogr. A., 694, 425 (1995).

16. Suyani, H., Heitkemper, D., Creed, J. and Caruso, J.,Appl. Spectrosc., 43, 962 (1989).

17. Wang, J., Tomlinson, M. J. and Caruso, J. A., J. Anal. At.Spectrom., 10, 601 (1995).

18. Thomas, P., Finnie, J. K. and Williams, J. G., J. Anal. At.Spectrom., 12, 1367 (1997).

19. Ackley, K. L., B’Hymer, C., Sutton, K. L. and Caruso,J. A., J. Anal. At. Spectrom., 14, 845 (1999).

20. Creed, J. T., Magnuson, M. L., Pfaff, J. D. and Brock-hoff, C., J. Chromatogr. A., 753, 261 (1996).

21. Ulrich, N., Fresenius’ J. Anal. Chem., 360, 797 (1998).22. Inoue, Y., Sakai, T. and Kumagai, H., J. Chromatogr. A,

706, 127 (1995).23. Suyani, H., Creed, J., Davidson, T. and Caruso, J., J.

Chromatogr. Sci., 27, 139 (1989).


24. Terasahde, P., Pantsar-Kallio, M. and Manninen, P. K. G.,J. Chromatograph. A, 750, 83 (1996).

25. Mason, A. Z., Storms, S. D. and Jenkins, K. D., Anal.Biochem., 186, 187 (1990).

26. Dean, J. R., Munro, S., Ebdon, L., Crews, H. M. andMassey, R. C., J. Anal. At. Spectrom., 2, 607 (1987).

27. Matz, S. G., Elder, R. C. and Tepperman, K., J. Anal. At.Spectrom., 4, 767 (1989).

28. Lyon, T. D. B. and Fell, G. S., J. Anal. At. Spectrom., 5,135 (1990).

29. Crews, H. M., Dean, J. R., Ebdon, L. and Massey, R. C.,Analyst , 114, 895 (1989).

30. Klueppel, D., Jakubowski, N., Messerschmidt, J.,Stuewer, D. and Klockow, D., J. Anal. At. Spectrom ., 13,255 (1998).

31. Stalcup, A. M., Chiral separations, in Kirk–Othmer Ency-clopedia of Chemical Technology , 4th edn, Supplement,John Wiley & Sons, Inc., New York, 1998, p. 133.

32. Mendez, S. P., Gonzalez, E. B., Fernandez Sanchez, M. L.and Sanz Medel, A., J. Anal. At. Spectrom., 13, 893 (1998).

33. Sutton, K. L., Ponce de Leon, C. A., Ackley, K. L., Sut-ton, R. M. C., Stalcup, A. M. and Caruso, J. A., Analyst ,125, 281 (2000).

34. Ponce de Leon, C. A., Sutton, K. L., Caruso, J. A. andUden, P. C. J. Anal. At. Spectrom., 15, 1103 (2000).

35. Mendez, S. P., Gonzalez, E. B. and Sanz Medel, A., J.Anal. At. Spectrom., 15 1109 (2000).

36. Vissers, J. P. C., Claessens, H. A. and Cramers, C. A., J.Chromatogr. A., 779, 1 (1997).

37. Yang, F. J. (Ed.), Microbore Column Chromatography, aUnified Approach , Marcel Dekker, New York, 1989.

38. Tangen, A., Trones, R., Greibrokk, T. and Lund, W., J.Anal. At. Spectrom., 12, 667 (1997).

39. Shum, S. C. K., Pang, H. and Houk, R. S., Anal. Chem.,64, 2444 (1992).

40. Powell, M. J., Boomer, D. W. and Wiederin, D. R., Anal.Chem., 67, 2474 (1995).

41. Garraud, H., Woller, A., Fodor, P. and Donard, O. F. X.,Analusis , 25, 25 (1997).

42. Pergantis, S. A., Cullen, W. R., Chow, D. T. and Eigen-dor, G. K., J. Chromatogr. A., 764, 211 (1997).

43. Zoorob, G. K., McKiernan, J. W. and Caruso, J. A.,Mikrochim. Acta, 128, 145 (1998).

44. Olesik, J. W. and Moore, A. W., Anal. Chem., 62, 840 (1990).45. Skoog, D. A., Holler, F. J. and Nieman, T. A., Principles

of Instrumental Analysis , 5th edn, Harcourt Brace CollegePublishers, Philadelphia, PA, 1998.

46. Sutton, K. L. and Caruso, J. A., J. Chromatogr. A, 856,243 (1999).

47. Nakahara, T., Bunseki Kagaku, 46, 513 (1997).48. Le, X., Cullen, W. R. and Reimer, K. J., Talanta , 41, 495

(1994).49. Harrington, C. F., Eigendorf, G. K. and Cullen, W. R.,

Appl. Organomet. Chem., 10, 339 (1996).

4.2 Gas Chromatography and Other GasBased Methods

J. Ignacio Garcıa Alonso and Jorge Ruiz EncinarUniversity of Oviedo, Spain

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 1632 Species which can be Analysed by Gas

Chromatography . . . . . . . . . . . . . . . . . . . . 1642.1 Volatile species . . . . . . . . . . . . . . . . 1642.2 Nonvolatile elemental species . . . . . . 1662.3 Derivatisation reactions . . . . . . . . . . . 166

2.3.1 Hydride generation . . . . . . . . 1662.3.2 Aqueous ethylation . . . . . . . . 1682.3.3 Aqueous propylation . . . . . . . 1692.3.4 Grignard derivatisation . . . . . . 1702.3.5 Other derivatisation

reactions . . . . . . . . . . . . . . . . 1712.3.6 Comparison of derivatisation

reactions . . . . . . . . . . . . . . . . 1722.4 Preconcentration and clean-up before

GC separation . . . . . . . . . . . . . . . . . 1723 Separation Techniques Used . . . . . . . . . . . 174

3.1 Cryogenic trapping and thermaldesorption in packed columns . . . . . . 174

3.2 Gas chromatography with packedcolumns . . . . . . . . . . . . . . . . . . . . . . 174

3.3 Gas chromatography with capillarycolumns . . . . . . . . . . . . . . . . . . . . . . 175

4 ICP-MS as Detector for GasChromatography . . . . . . . . . . . . . . . . . . . . 1764.1 Development of atomic gas

chromatography detectors forelemental speciation . . . . . . . . . . . . . 176

4.2 Interfaces for GC-ICP-MS . . . . . . . . 1784.3 Analytical characteristics of the

GC-ICP-MS coupling . . . . . . . . . . . . 1814.4 Isotope ratio measurements with

GC-ICP-MS . . . . . . . . . . . . . . . . . . . 1844.4.1 Applications of isotope ratio

measurements . . . . . . . . . . . . 1874.5 Comparison of different ICP-MS

instruments . . . . . . . . . . . . . . . . . . . 1885 Application of the GC-ICP-MS Coupling 190

5.1 Environmental applications . . . . . . . . 1905.2 Biological applications . . . . . . . . . . . 1925.3 Isotope dilution analysis with

GC-ICP-MS . . . . . . . . . . . . . . . . . . . 1925.4 Reference materials and quality

control . . . . . . . . . . . . . . . . . . . . . . . 1956 Conclusions . . . . . . . . . . . . . . . . . . . . . . . 1957 References . . . . . . . . . . . . . . . . . . . . . . . . 196

1 INTRODUCTION

There is a clear need today of measuring not only thetotal concentration of a certain element in the samplebut also the concentration of the different chemicalspecies in which this particular element may be dis-tributed. The ability of a certain element to formdifferent chemical species will be responsible for

its geochemical distribution in the environment, itsbio-availability and its toxicity towards different liv-ing organisms. Recently, the IUPAC [1] has definedthe terms ‘elemental speciation’ (the distribution ofdefined chemical species of an element in a system)and ‘chemical species’ (specific form of an elementdefined as to isotopic composition, electronic or oxi-dation state, and/or complex or molecular structure)


in trying to clarify those important concepts in mod-ern analytical chemistry.

The big impact of speciation in modern analyt-ical chemistry can be reflected not only in the vastamount of analytical methodologies developed inthe last decade but also by the fact that new regula-tions are being implemented in different countriesbased on these studies [2]. This means that inter-est in elemental speciation will increase even moreas these methodologies need to be implemented inroutine chemical laboratories and included in qual-ity assurance programs.

From an analytical point of view, speciationanalysis is not an easy task. From the samplingto the final measurement step we have to take intoaccount a series of problems which could make allour efforts worthless:

1. We need to keep the identity and concentrationof the different species in the sample unchangedthroughout the whole analytical procedure or, atleast, provide means to correct for any changesin speciation.

2. We need to separate and positively iden-tify the different chemical species present inthe sample.

3. We need to determine very low levels of thosespecies with adequate precision and accuracy.

4. We need to take into account matrix effectsfrom complex biological and environmentalsamples which could affect most pretreatmentprocesses used.

All the above mentioned factors make speciationanalysis a complex task requiring several analyticalsteps which may include extraction, clean-up,derivatisation, preconcentration, separation andfinal measurement procedures. Most publicationson speciation analysis have focused mainly onthe last two analytical steps: separation andmeasurement. The preferred techniques makeuse of a potent chromatographic procedure(gas chromatography or high performance liquidchromatography) coupled to sensitive and selective(or specific) atomic detection techniques such asatomic absorption (AAS), atomic emission (AES)and mass spectrometry (MS) in combination withflames or, more recently, plasmas (ICP or MIP).

In this chapter we will focus on the use of gaschromatography (GC) for speciation analysis bothfor volatile compounds per se and for those non-volatile compounds which can be derivatised andthen separated by GC. We will describe differ-ent modes of GC and those derivatisation methodswhich are currently employed for speciation analy-sis. Finally, and due to its current interest, we willreview the main characteristics and applicationsof the coupling of GC with inductively coupledplasma mass spectrometry (ICP-MS), discussingcurrent and future applications.

2 SPECIES WHICH CAN BEANALYSED BY GASCHROMATOGRAPHY

The basic conditions governing the suitability ofGC for the separation and detection of a givencompound are its volatility and thermal stability.Within the field of speciation very few compoundsfulfil those requirements directly and the analysthas to resort to chemical reactions to transformnonvolatile compounds (usually ionic) into volatilethermally stable compounds. Those reactions areknown as derivatisation reactions and may includehydride generation, alkylation, etc. Many derivati-sation reactions have been developed for GC andthat has broadened the field of application of GC tothe speciation of a large variety of compounds.

2.1 Volatile species

Within the field of elemental speciation sev-eral authors have identified volatile metalspecies in different samples. Feldmann andcoworkers [3, 4] have identified volatile speciessuch as dimethylmercury (Me2Hg), dimethylsele-nium (Me2Se), tetramethyltin (Me4Sn), trimethy-lantimony (Me3Sb), trimethylbismuth (Me3Bi),methylated arsines (MexAsHy , x + y = 3), dime-thyltellurum (Me2Te), tetraalkylated lead com-pounds (EtxMeyPb, x + y = 4) in sewage sludgegases and, more recently, molybdenum andtungsten hexacarboniles [5] (Mo(CO)6 andW(CO)6) in gases from municipal waste disposal

SPECIES WHICH CAN BE ANALYSED BY GC 165

sites. Most of these compounds are extremely toxicand their presence in these samples needs to bestudied. Unfortunately, the quantification of thesespecies is complex because of the lack of commer-cial standards or certified reference materials formost of these compounds. For that reason, Feld-mann [6] developed a semiquantitative methodol-ogy consisting in the addition of nebulised waterat the exit of the GC column in order to obtainthe same response from the volatile compound andfrom the aqueous standards of those elements.

Volatile tetraalkyllead compounds have beenused for years as antiknocking agents in gaso-line. These compounds were detected in air sam-ples at levels up to 14 ng Pb m−3 by sampling70 L of air in a cryogenic trap. The authors con-cluded that their presence in air was due to gasolineevaporation rather that to its incomplete combus-tion [7]. Tetraethyllead is the standard antiknockingagent but other authors have detected the presenceof mixed methyl-ethyl tetraalkyllead compoundsin gasoline by GC-ICP-MS [8, 9]. Recently, themeasurement of lead isotope ratios [10] in differ-ent organolead compounds present in gasoline andatmospheric particulate matter has been used tostudy lead pollution sources.

Volatile compounds of Se, Sn, Hg and Pb havebeen also detected in natural [11] and marine [12]waters. These studies demonstrated that the inter-action between anthropogenic and natural sourcesled to the formation of volatile metallic andorganometallic species in rivers, estuaries and thesea. Similar studies carried out by the same researchgroup, using cryogenic trapping, showed the pres-ence of volatile species of Hg, Sn, In, Ga, Se,P and As in air samples collected in urban andrural areas. The same authors detected volatile tinhydrides and methylated butyltin compounds in thebay of Arcachon and other polluted areas of France,Belgium and Holland [13]. The extremely low con-centration levels of these compounds in the samplesrequired special preconcentration techniques. In thiscase, cryogenic trapping was used after purging thecompounds from 1 L of water with He or passing100 L of air through adsorption columns immersedin liquid nitrogen. Thermal desorption GC-ICP-MS

was used to achieve both species separation and ade-quate limits of detection. Previous environmentalstudies on volatile species in the atmosphere usedroom temperature trapping on solid sorbents [14,15] and thermal desorption to the GC column [16],avoiding the use of organic solvents which couldcause contamination.

Recently [17] the coupling of GC to adouble focusing ICP-MS instrument allowed thespeciation of sulfur compounds in human salivaincubated under anaerobic conditions. The sulfurcompounds detected, H2S, CH3SH, (CH3)2S,(C2H5)2S, (CH3)2S2 and (C2H5)2S2, are generatedby anaerobic microorganisms in the mouth and canbe the cause of more than 80 % of the cases ofbad breath. The use of high resolution in the massspectrometer allowed the separation of the 32S peakfrom its polyatomic interference 16O2

+ and thedetection of up to eight volatile sulfur compoundswas facilitated by the use of a special GC column(SPB-1 Sulphur).

Other applications for the determination ofvolatile compounds, such as the detection oftetraalkyllead compounds in gasoline [18] orhalogenated compounds in macro-algae [19] are

Table 4.2.1. Volatile elemental species detected according tothe type of sample.

Sample type Detectedspecies

Techniqueused

Ref.

Gasoline Tetraalkylleadcompounds

GC-FAAS 18

Landfill gas As, Se, Sn, Sb, Te,Hg, Pb, Bi, Mo,W compounds

GC-ICP-MS 5

Air Tetraalkylleadcompounds

GC-AAS 14

Air Hg0, MeHgCl,Et2Hg, Me2Hg

GC-AFS 15

Air MeHgCl, Me2Hg GC-AED 16Fermented

salivaH2S, CH3SH,

(CH3)2S,(C2H5)2S,(CH3)2S2,(C2H5)2S2

GC-SF-ICP-MS 17

Macro-algae CH3I, CH2I2,CHBr3, CHBrCl2,CH2BrCl, C2H5I

PT-GC-AED 19

Estuarinewaters

Me2Se, Me2Se2,Me2Hg, Et2Hg,Me4Sn, Et4Sn,Me4Pb, Et4Pb

CT-GC-ICP-MS 11


worth mentioning here. Table 4.2.1 summarizesthe different volatile species detected according tothe type of sample.

2.2 Nonvolatile elemental species

Most elemental species are ionic or polar com-pounds which show high boiling points and lowthermal stability. Over the years, suitable derivati-sation reactions have been developed for their sep-aration and determination by GC. Those chemicalreactions were selected in order to produce com-pounds of low polarity, low boiling points and highthermal stability with adequate chemical yields.In many cases, these derivatisation reactions canbe coupled to a preconcentration and/or separationtechnique (e.g. liquid–liquid extraction), improv-ing the sensitivity and selectivity of the analyticalprocess. In the following pages we will review themost important derivatisation reactions for elemen-tal speciation using GC.

2.3 Derivatisation reactions

In all speciation steps we need to preserve therelevant metal–carbon bonds so we can identifythe original species present in the sample, andthis is especially true for derivatisation reactions.In that sense, all derivatisation reactions used,prior to GC separations, follow one of the fol-lowing processes: (i) conversion of inorganic ororganometallic ions into covalent volatile com-pounds in aqueous media (e.g. hydride genera-tion, ethylation), (ii) conversion of inorganic ororganometallic ions into covalent volatile com-pounds in organic media using Grignard reagents(e.g. butylation) and (iii) conversion of ionicspecies into stable volatile chelates (dithiocarba-mates, acetonates and trifluoroacetonates).

It is important to realize that all derivatisationprocedures can be affected by matrix effects in thesample and, hence, recovery studies have to becarried out for different sample types under theselected experimental conditions. Unfortunately,there is a lack of commercially availablederivatised analytes which makes the computing

of derivatisation recoveries difficult [20, 21].Most authors use internal standards to correct forderivatisation recoveries and analyte losses butthis is clearly not the best solution. Other authorsrecommend the use of isotopically enriched speciesand isotope dilution procedures which do notrequire known or quantitative recoveries in any ofthe extraction, clean-up and derivatisation steps: ifwe can ensure that isotopic equilibrium betweenthe original and spiked species in the sample isachieved, the recovery factors in any subsequentseparation and derivatisation procedure will notaffect the final concentration results because thefinal measurement is an isotope ratio that willbe constant and independent of the number ofmolecules or atoms isolated to measure thatratio. Another advantage of the use of enrichedisotopes is the fact that species transformations ordegradation can be detected using several enrichedisotopes [22].

2.3.1 Hydride generation

This well-known derivatisation reaction is usuallyapplied to small inorganic and organometallic ionswhich form highly volatile covalent compounds.Different species containing the elements As,Sb, Hg, Sn, Pb, Bi, Cd, Se, Te and Ge havebeen derivatised using this technique [23]. Thereaction takes place with the sample in acidicaqueous medium by adding sodium borohydrideboth as reductor and hydride source. Depending onthe species and the element considered we haveto optimise the concentration of reducing agentand the type and concentration of the acid used[24, 25]. Unfortunately, this procedure cannot beapplied to all species of the same element dueto thermodynamic and/or kinetic constraints orbecause of low stability of the hydrides formed.For example, this is the case for some alkylatedlead species [26] and the high oxidation statesof Se, As [27] and Sb. In the last case, thesehigh oxidation state species have to be reducedpreviously, with the consequent loss of speciationinformation.

On the other hand, the methylated species ofmercury [26] and germanium [28] form hydrides


easily and the metal–carbon bond is preservedduring the reaction which facilitates the identi-fication and quantification of the species. Simi-lar results were obtained for tin speciation: tinhydrides are readily formed and the tin–carbonbond is not broken during derivatisation. Therehave been numerous papers on tin speciation usinghydride generation before GC separation. In mostcases, hydride generation was coupled to a cryo-genic trapping step in a chromatographic columnimmersed in liquid nitrogen. After the trapping stepthe liquid nitrogen was removed and a nichromewire, wound around the column, was heated elec-trically to desorb the analytes and separate themin the GC column. AAS [29] and ICP-MS [30]have been used as suitable selective and sensitivedetection methods. The system used by SegoviaGarcıa et al. [30], which is basically an evolu-tion of that published previously by Donard et al.[29], is presented in Figure 4.2.1. Basically, agiven volume of sodium borohydride was addedto the sample with the help of a programmedperistaltic pump. A flow of helium carrier gastransported the hydrides formed from the samplecontainer to the GC column which was immersed

in liquid nitrogen. For ICP-MS work it was impor-tant to remove the excess of hydrogen formedbefore the ICP torch and that was accomplishedwith the help of a three-way valve connected afterthe GC column during the preconcentration step.After the given purge and trap time had elapsed,then (i) the liquid nitrogen was removed, (ii) thesolenoid valve just before the GC column and thethree-way valve after the column were switched,(iii) the heater of the column was connected and(iv) the ICP-MS acquisition sequence was initi-ated. This procedure allowed the on-line derivati-sation–preconcentration–separation–detection ofbutylated tin species with absolute detection limitsbetween 50 and 200 pg for AAS and between 4and 7 pg for ICP-MS.

Clark and Craig [31] and Sullivan et al. [32]suggested an interesting approach in which thehydride generation occurred in a special reac-tor packed with sodium borohydride and locatedinside the gas chromatograph. This ‘reactive GC’approach was applied successfully for tin specia-tion with a considerable reduction in analysis time.

However, hydride generation suffer from severematrix interferences when applied to real samples

Hydrogenpurge

Ar carrier

Solenoidvalve

GCColumn

solenoidvalve

SampleNaBH4He

ICPMS

liquid N2

“T” piece

Ni-Crwire

Figure 4.2.1. Hydride generation, cryogenic trapping, GC separation and ICP-MS detection for the speciation of butyltincompounds. Reprinted with permission from John Wiley & Sons, Ltd.


such as wastewaters, sediments and biota. Theseinterferences cause both signal suppression anddecreased precision in the measurements. Differ-ent authors have published studies on the effectof organic matter and transition metals on hydridegeneration yields [21, 33] but most studies havefocused on those matrix effects for butyltin specia-tion studies [34] due to the usually complex matri-ces in which those compounds are determined.Organic solvents, pesticides and other organiccompounds did not influence hydride reactionyields but humic substances caused low repro-ducibility due to the formation of foams dur-ing the reaction. More serious interferences werecaused by transition metals in the reaction mediumwhich decomposed the organometallic hydridesformed and sodium borohydride itself [26]. A pos-sible solution to this problem was the additionof L-cysteine to the reaction medium [33]. Thisreagent formed complexes with the transition met-als improving the sensitivity and reproducibility ofthe hydride generation reaction.

Other applications of hydride generation for ele-mental speciation in combination with GC havebeen published [35–38]. For example, the specia-tion of antimony in freshwater plant extracts wasdescribed by Dodd et al. [36]. These authors indi-cated that the problem of organostibine molecularrearrangement could be solved by adequate selec-tion of the experimental procedure. Table 4.2.2summarizes those and other applications of hydridegeneration in elemental speciation studies.

Table 4.2.2. Elemental species detected after hydride genera-tion and separation by GC.

Sample type Detected species Instrumentationused

Ref.

Fish TBT, DBT GC-FPD 32Estuarine

watersSn4+, MexSn4−x ,

BuxSn4−xGC-QF-AAS 29

Mussels MBT, DBT, TBT GC-QF-AAS 35Aquatic

plantsDifferent Sb

compoundsGC-MS 36

Natural andwastewaters

Ge4+, MeGe3+,Me2Ge2+

CT-ICP-MS 28

Harboursediment

MeSnCl3, MBT,DBT, TBT

CT-GC-QF-AAS

34

Soils As(III), As(V),MMA, DMA

GC-SF-ICP-MS

37

Seawater Hg+, MeHg+ GC-AFS 38

2.3.2 Aqueous ethylation

The ethylation reaction consist in the addition ofone or more ethyl groups to inorganic or alky-lated metal species to form the di- (Hg, Se), tri-(Bi) or tetraalkylated species (Sn, Pb) which arehydrophobic, volatile and thermally stable and,hence, suitable for GC separations. The ethyla-tion reaction is simple and quantitative (except forinorganic ions), can be performed in the aqueousphase and does not destroy existing metal–carbonbonds. There are two ways in which ethylation canbe performed: (i) using Grignard reagents [39, 40](ethyl magnesium bromide) or (ii) using sodiumtetraethylborate [41]. In the first case ethylation isperformed in an anhydrous organic phase, whilein the second case it is performed in the aque-ous phase. The use of Grignard reagents is lessprone to matrix interferences when compared toaqueous ethylation or hydride generation. Thisis because Grignard ethylation takes place in anorganic phase after the separation of the analytesfrom the matrix by liquid–liquid extraction. Also,De la Calle-Guntinas et al. [42] have reportedthat the derivatisation using Grignard reagents(ethylation and pentylation) provided higher reac-tion yields than aqueous ethylation using NaBEt4.The main disadvantage of Grignard ethylation isthe requirement of an anhydrous organic mediumfor the reaction to take place and that meansthat a previous liquid–liquid extraction proce-dure has to be applied. For the speciation oforganotin compounds this liquid–liquid extractionwas performed in a strong acidic medium [43],whereas organolead compounds were extracted inweak acid media [44]. Complexing agents used forextraction included tropolone [40, 45] (for organ-otin compounds), ditizone and diethyldithiocarba-mate (DDTC). The fact that the derivatisation takesplace in an inert atmosphere and that the excess ofGrignard reagent has to be destroyed after derivati-sation makes ethylation using Grignard reagentscumbersome and time consuming.

Aqueous ethylation using sodium tetraethylbo-rate combines the advantages of working in theaqueous phase of hydride generation (no previ-ous extraction into organic phase, speed, conve-nience of use) with the low matrix interferences of


Grignard reagents. De Diego et al. [46] comparedthe analytical performance characteristics of bothaqueous ethylation and hydride generation for mer-cury speciation in samples of high salt and transi-tion metal content. The results demonstrated thatthe presence of transition metals seriously affectedhydride generation while the content of chlorideinfluenced the ethylation yields. Other authors sug-gested that the two derivatisation techniques arecomplementary [47], as their performance dependson the type of sample measured. In general terms,ethylation using NaBEt4 fits well in the actual trendof speciation analysis: to minimize the number ofanalytical steps, and so reduce sample handling andanalysis time. This reduction in sample handlingand reagent usage, which is provided by aqueousethylation, leads to improved reproducibility andlower analytical errors (losses of analytes, contam-ination, etc).

Aqueous ethylation for trace metal speciationwas first proposed by Rapsomanikis et al. [48] in1986 for the speciation of methylated organoleadcompounds in waters using a home-made GC-AASinterface after purge and trap preconcentration.Later, this derivatisation procedure was applied forthe speciation of, mainly, Sn, [49, 50, 51] Hg [48,52, 53] and Pb [48, 54] compounds. An excellentreview on the use of NaBEt4 for trace metalspeciation was published by Rapsomanikis [55].In this review, the author explains the chemicalreactions that take place and the applications of thisreagent for trace metal speciation. We have also totake into account that, when analysing complexsamples, a large excess of NaBEt4 is usuallyrequired to compensate for the consumption of thereagent by other sample components [56].

In situ derivatisation using NaBEt4 has alsobeen applied to reduce analysis time [57]. In thiscase, a packed reactor inside the gas chromato-graph containing NaBEt4 was used in a similarway to that previously described for hydride gen-eration [31, 32]. Another alternative for shorteranalysis times consist in combining matrix extrac-tion, derivatisation with NaBEt4 and liquid–liquidextraction in a single analytical step [58]. All thisprocess can be performed in less than 3 minswith the use of focused microwave systems [59].

Table 4.2.3. Organometallic species which have been detectedafter derivatisation using sodium tetraethylborate.


Ref.

Waters Me2Pb2+, Me3Pb+ CT-GC-AAS 48Waters Hg, MeHg, EtHg GC-AFS/MS/

MIPAES60

Riversediment

MBT, DBT, TBT GC-QF-AAS 50

Fish Hg, MeHg GC-FAPES 53Sediments,

biota andwastewaters

MBT, DBT, TBT GC-FPD 61

Open-oceanseawater

MBT, DBT, TBT,MPhT, DPhT,TPhT

GC-ICP-MS(shield torch)

62

Sediment MBT, DBT, TBT GC-AAS/MS 57Fish and

marinesediments

MBT, DBT, TBT GC-MIP-AES 63

Other applications of aqueous ethylation have beendescribed [60, 61, 62, 63] and a summary of rel-evant applications of ethylation for elemental spe-ciation is presented in Table 4.2.3.

The combination of ethylation with NaBEt4 andsolid-phase microextraction (SPME) for the detec-tion of very low concentrations of organometal-lic compounds in waters is a rapidly growingfield [64]. We will expand on the use of SPME forextraction/injection in combination with GC laterin this chapter.

2.3.3 Aqueous propylation

The generalized use of NaBEt4 as a derivatisa-tion reagent for organotin speciation has not beenextended to other organometallic species of Pb orHg. This was limited by the fact that most Pbspecies of interest contained ethyl groups which,after derivatisation, would form the same species.For example, Et4Pb, Et3Pb+, Et2Pb2+ and Pb2+will all form Et4Pb after derivatisation with thecorresponding loss of speciation information. Thesame can be said of Et2Hg, EtHg+ and Hg2+,which would prevent the use of EtHg+ as internalstandard for the determination of both inorganicand methylmercury compounds. This problem canbe solved by resorting to a different alkylationreagent. So, Grignard reagents based on propyl


magnesium or butyl magnesium have been exten-sively used for organolead speciation [65, 66].

Another recent approach, which maintains theadvantages of easy handling and aqueous derivati-sation of NaBEt4, and can be applied for both Pband Hg speciation is the use of sodium tetrapropy-lborate (NaBPr4). The synthesis of this derivatisa-tion reagent was described in detail by De Smaeleet al. [67], who applied it to the simultaneousspeciation of Pb, Hg and Sn compounds in envi-ronmental samples. Later, other authors comparedboth derivatisation reagents for the speciation ofmercury compounds [68]. In this paper, the advan-tages of NaBPr4 over NaBEt4 for the determinationof both inorganic mercury and methylmercury con-sisted mainly in the ability to use ethylmercuryas internal standard for quantification after selec-tive GC-ICP-MS detection. The application of thisreagent to the detection of organolead compoundsin snow and road dust has been described [69].

The main drawback of the use of NaBPr4

is the low chemical stability of this reagent incomparison with its ethylated analogue. In spite ofthis, the use of aqueous propylation has increasedin recent years as the reagent can now be obtainedcommercially. Some of the applications describedto date are summarized in Table 4.2.4.

2.3.4 Grignard derivatisation

Grignard derivatisation reactions, which use alkyl-magnesium halides, can only be performed in awater-free organic phase and under inert atmo-sphere. That means that these reactions require amore complex experimental set-up but have theadvantage of quantitative derivatisation in mostcases. The organometallic species, usually presentin an aqueous phase, have to be extracted into

Table 4.2.4. Applications of tetrapropylborate to elementalspeciation.


Ref.

Snow androad dust

TML, DML, TEL,DEL

GC-MIP-AES 69

Fish Hg, MeHg GC-ICP-MS 68Sediments MBT, DBT, TBT GC-ICP-MS 67

an organic solvent prior to derivatisation and thisis done generally using chelating agents such astropolone, DDTC, etc. Different alkyl groups havebeen used including methyl, propyl, butyl, pentyland hexyl. However, the most generally used alkylgroup has been the butyl group.

Butylation reactions, using for example butyl-magnesium chloride, are interesting and usefulalternatives for the speciation of both Hg and Pbcompounds in biological [68, 70, 71] and environ-mental samples [72]. Radojevic et al. [73] com-pared butylation and propylation efficiencies fororganolead compounds using Grignard reagents.They concluded that propylation yields for dialky-lated species (R2Pb2+) were higher than butylationyields. Other authors have also described lowerbutylation yields for these species [44].

The extraction of organolead compounds intothe organic phase has been performed using DDTCat pH 9 in the presence of EDTA to preventthe co-extraction of inorganic lead. [70, 72, 74]In this way the extraction of inorganic lead,usually present in much higher concentration thanthe organolead compounds, is prevented. Thisis necessary as a large quantity of inorganiclead would consume the available butylationreagent, reducing the derivatisation yield for theother organolead compounds, and form hugeamounts of the late eluting Bu4Pb which couldcause memory effects due to tailing and columncontamination.

Butylation has also been evaluated for the spe-ciation of mercury compounds. Garcıa Fernandezet al. [68] compared aqueous ethylation andpropylation with Grignard butylation for the deter-mination of methylmercury in marine referencematerials (DOLT-2, NRCC, Canada). Both aque-ous propylation and Grignard butylation pro-vided similar results as ethylmercury could beused as internal standard for both derivatisationreactions.

The use of pentylmagnesium halides has beenmainly focused to the determination of organ-otin compounds in biological [45, 35, 75] andenvironmental samples [76]. However, this reagenthas been also applied with good results to thespeciation of both lead and mercury [77]. Stab


et al. [78] reported that derivatised organotinspecies were more prone to degradation duringthe destruction of the excess of Grignard reagentwhen the derivatisation alkyl group was smaller.So, larger alkyl groups, such as pentyl or hexyl pro-vided more stable species. Also, the lower volatil-ity of the pentylated species in comparison withthose ethylated or propylated would facilitate thepreconcentration of those species by evaporationof the solvent under an inert gas flow. Thereare two drawbacks which have been describedfor the pentylation of organotin compounds: first,the derivatised compounds would require higherelution temperatures and could condensate in theinterface between the gas chromatograph and thedetector [24]. Second, the derivatisation yields forthe pentylation of di- and trialkyltin compoundswere lower than when using shorter chain derivati-sation reagents [76].

Other Grignard reagents, such as methylmagne-sium [78, 79], propylmagnesium [80, 81] or hexyl-magnesium halides [82] have also been proposedfor elemental speciation. Cai et al. [82] appliedsupercritical fluid extraction (SFE) in combinationwith Grignard derivatisation. The main advantageof this approach is that extraction, derivatisationand clean-up can be performed in a single stepwithout the need for organic solvents. The tech-nique of SFE has been also applied for the extrac-tion of mercury [83], lead [84] and arsenic [85]compounds in environmental samples.

Some relevant applications of Grignard derivati-sation in elemental speciation are summarized inTable 4.2.5.

2.3.5 Other derivatisation reactions

Many other derivatisation reactions have beendescribed for elemental speciation. Of specialinterest are those derivatisation reactions whichcan be performed in the aqueous phase. Forexample, the use of sodium tetraphenylborate hasbeen proposed [60, 86] as an alternative to NaBEt4when information about EtHg+ is required togetherwith inorganic Hg and MeHg+.

The determination of sulfur- and selenium-containing amino acids is another field wherederivatisation reactions for GC separation areoften employed. Studies reported by Clausen andNielsen [87] demonstrated that L-selenomethionineis the Se species which can be more efficientlyabsorbed by the organisms and that explainsthe growing interest in the determination of Se-containing amino acids in foods and nutritionalsupplements. The derivatisation of amino acids forGC separation has been the subject of numerouspublications. In general, it is necessary to deriva-tise both the carboxylic acid and the amino groupto obtain thermally stable volatile compounds. Thisderivatisation can be performed in two sequen-tial steps [88] consisting in the esterification ofthe carboxylic acid in an acidic medium using analcohol (isopropanol or isobutanol) and the subse-quent acylation of the amino group using an anhy-dride (trifluoroacetic or heptafluorobutiric). Later,the direct derivatisation of both the carboxylic acidand the amino group using alkyl chloroformateswas proposed [89]. In our experience [90] the two-step procedure provided higher recoveries and acleaner GC-MS chromatogram while the one-step

Table 4.2.5. Applications of Grignard reagents in elemental speciation.

Sample type Derivatisation Extraction Detected species Instrument Ref.

Atmosphericparticulates

BuMgCl DDTC/hexane DML, DEL, MEL GC-ICP-MS 72

Rainwater PrMgCl DDTC/hexane TML, DEL GC-ICP-MS 80Human urine BuMgCl DDTC/pentane–hexane Pb2+, TML, TEL GC-MS 70Marine products EtMgBr NaCl–HCl/Et2O–hexane TBT, DBT, TPhT GC-FPD 39Mussels MeMgI NaCl–HCl/Et2O–hexane MBT, DBT, TBT,

MPhT, DPhT, TPhTGC-MIP-AES/MS 78

Fish tissue BuMgCl DDTC/toluene Hg, MeHg GC-ICP-MS 68Polyurethane

foamPrMgBr HCl/toluene DBT GC-FPD 81

Sediments HexMgBr SFE TBT, PDT, TPhT GC-FPD 82


Table 4.2.6. Analytical characteristics of some derivatisation reactions used for the speciation of organometallic compounds.

Characteristic Hydride generation Aqueous alkylation Grignard reagents

Reaction yield High for small ions but lower fortrialkylated species

High, except for monoalkylatedand inorganic species

Very high

Reproducibility Low due to foam formation High HighInterferences Severe in samples with high

organic or metallic contentLow matrix effects except in the

presence of chlorideNo interferences due to

matrix separationStability of the

reagentsHigh Decompose slowly with light and

humidityRequire an inert and dry

atmosphereEase of handling High due to the low reaction pH

(pH ∼= 2)High (pH ∼= 5) Very low due to extraction

into an organic phase,inert atmosphere anddestruction of the excessof reagent

Speed of reaction Instantaneous, allows on-linederivatisation

High, can be accelerated usingMW irradiation

Low, the global processrequires several reactionsteps

Limitations Only suitable for smallorganometallic species in simplematrices

NaBEt4 not suitable for someapplications, use of NaBPr4instead

None

Cost of reagents Low High High

procedure formed one main product and a secondsubproduct showing the selenium isotopic patternin the GC-MS spectrum.

2.3.6 Comparison of derivatisation reactions

The analytical characteristics of three derivati-sation reactions currently used for the specia-tion of organometallic compounds are comparedin Table 4.2.6. As can be observed in the table,the suitability of one or other reaction for agiven application will depend mainly on the typeof organometallic compounds analysed and onthe matrix. In general terms, aqueous alkylationreactions, when both tetraethyl- and tetrapropylbo-rate are available, seem to offer the best analyticalcharacteristics. This is also justified by the increas-ing number of publications on elemental speciationusing this type of derivatisation reaction.

2.4 Preconcentration and clean-up beforeGC separation

Elemental speciation in complex biological (humanserum, fish tissue, . . .) or environmental sam-ples (sediments, waste waters, . . .) requires, inmany cases, the use of different preconcentrationand/or clean-up procedures which could isolate the

species of interest from the sample matrix. Sampleswith high fat content, sulfur-containing com-pounds, etc. could seriously affect both derivati-sation and extraction yields and even prevent thedetection of the sought compounds due to spectralinterferences. Most preconcentration and clean-upprocedures make use of classical approaches whichare tedious, time consuming, require large quanti-ties of organic solvents and are error prone due tocontamination or analyte losses. The fact that moststandard analytical procedures make use of clas-sical clean-up and preconcentration methods hasprevented further developments in this field [91].However, in the last few years, many solid-phaseextraction (SPE) and solid-phase microextraction(SPME) procedures have been published whichprovided improved analytical performance charac-teristics comparable with classical methods.

Procedures involving SPE either with C18[40] or fluorisil [39] cartridges or filtration mem-branes [70] have been published. All these pro-cedures are normally applied before the finalderivatisation reaction. However, the clean-up ofthe organic extract prior to the final injection intothe gas chromatograph has also been applied using,for example, silica gel columns [32, 40].

The last revolution in the field of organicsample preparation, which can also be applied


for preconcentration and clean-up in elemen-tal speciation, has been the introduction ofSPME [92]. Since the pioneering work of Arthurand Pawliszyn [93] in 1990 the use of SPME pro-cedures in elemental speciation, and in organicanalysis in general, can be considered almost rou-tine nowadays. The advantages of SPME for ele-mental speciation arise from its simplicity: in a firststep the analytes are distributed between the fibreand the sample and, once equilibration has takenplace, the analytes are thermally desorbed in theGC injector block. The whole SPME procedureprovides several advantages in comparison withclassical preconcentration and clean-up methods:(i) it does not require the use of organic solvents,(ii) it is fast and simple to perform, (iii) it allowsin situ preconcentration, (iv) it can be automatised,(v) it can be selective by adequate selection of thefibre material, (vi) it has less breakthrough prob-lems than SPE, (vii) it requires low sample vol-umes because of its high preconcentration factorand (viii) it does not require further clean-up. Allthese characteristics make SPME an excellent sam-ple preparation technique.

In spite of the fact that SPME was first devel-oped for the analysis of organic compounds,applications in elemental speciation were quicklydeveloped [94–102] for the analysis of lead,mercury and tin species with excellent results.Most of these publications make use of nonpolarpoly(dimethylsiloxane) fibres which require priorderivatisation of ionic organometallic species. Inmost cases derivatisation is done using NaBEt4in the same solution where SPME is performed.However, derivatisation can be performed in

the fibre itself or in the GC injector afterpreconcentration [103].

The coupling of SPME with a gas chro-matograph is relatively simple using traditionalsplit/splitless injectors for capillary columns. Theonly modification required is the use of narrowerbore glass liners, which provide a linear flowof gas along the fibre surface [103] and whichfacilitate the desorption of the analytes in theinjector. The parameters which influence the reten-tion of the analytes in the fibre include: (i) thenature of the fibre itself, (ii) the mode of oper-ation (head space or immersed), (iii) the volumeof sample, (iv) its pH and ionic strength, (v) themode of stirring, (vi) adsorption temperature and(vii) adsorption time. After adequate optimisationof all these parameters, detection limits in the lowng L−1 range have been accomplished for Pb, Hgand Sn organometallic species [98]. Selected appli-cations of SPME to elemental speciation using GCare included in Table 4.2.7.

Stir bar sorptive extraction (SBSE) [104] isa new extraction procedure in which magneticstir bars are coated with poly(dimethylsiloxane)and act as both stirrer and extraction phase. Thisprocedure has been applied for elemental specia-tion showing that the extraction efficiency of stirbars is superior to that of SPME but they requirespecial desorption equipment and cryofocusing ofthe analytes prior to GC separation [105]. Allderivatisation procedures used with liquid–liquidextraction, SPE and SPME can be also usedwith SBSE, so we can expect an increase inthe use of SBSE for elemental speciation in thenear future.

Table 4.2.7. Applications of SPME for elemental speciation using GC as separation technique.

Type of sample Derivatisationreagent

Detected species Instrument Ref.

Sea water NaBH4 MBT, DBT, TBT GC-FPD 99River water, fish NaBEt4 MeHg GC-MS 94Mussels and potatoes NaBEt4 TPhT GC-ICP-MS 102Tap water NaBEt4 Pb2+, Et4Pb GC-ITMS/FID 96Natural and

wastewaters,sediments

NaBEt4 MBT, DBT, TBT, MPhT,DPhT, TPhT

GC-FPD 100

Waters andsediments

NaBEt4 TeET, TeBT, MBT, DBT,TBT

GC-FPD 101


3 SEPARATION TECHNIQUES USED

3.1 Cryogenic trapping and thermaldesorption in packed columns

This analytical methodology has been extensivelyemployed for the preconcentration and separationof both volatile (directly) and ionic (after derivati-sation) species. The basic idea consists in the purgeof the analytes, either derivatised or not, fromthe sample using a gentle flow of He and theirtrapping on a U-shaped glass column filled withchromatographic material and immersed in liquidnitrogen (−192 ◦C). After the trapping step, theliquid nitrogen is removed and the glass columnis heated using a nichrome wire [25]. The ana-lytes are then desorbed sequentially from the col-umn and carried to the detector by another flowof helium. This experimental set-up is depictedin Figure 4.2.1 for hydride generation coupled toICP-MS. The connections between the column andthe detector have to be as short as possible to min-imize dead volumes and peak broadening. Someauthors recommend also the heating of the connec-tion tubing between column and detector to pre-vent condensation of low volatility analytes and,hence, the existence of peak broadening effects.For the determination of volatile species in airor other gases it is also recommended to usea dry ice–acetone trap (−20 ◦C) to retain watervapour prior to the liquid nitrogen trap. The sep-aration of the different species is based moreon the differences in boiling points than on theproperties of the chromatographic packing used.Normally apolar sorbents, such as Chromosorb,are used.

As we have said before, this system has beenextensively used for the analysis of volatile speciesin gases (air, landfill gas, etc.) and waters allow-ing the separation of the different elemental speciesand the treatment of large volumes of sample. Incomparison with the use of solid sorbents at roomtemperature cryogenic trapping offers two distinctadvantages: (i) it is not selective, so many differentspecies of various elements can be trapped simul-taneously and (ii) unstable chemical species canbe preserved for long periods of time before the

desorption and analysis is performed. The lattercharacteristic is specially important for the analy-sis of gaseous samples [106]. In this way, differentvolatile species have been determined in naturalwaters [11, 13], landfill gases [5] and differentatmospheres [106]. The low limits of detectionreported using ICP-MS detection (down to fem-togram levels) indicate the degree of developmentreached by this methodology. The coupling of thecryogenic trap to ICP-MS instruments offer differ-ent possibilities. While Pecheyran et al. [106] use asupplementary flow of argon to support the plasma,Feldmann and Cullen [5] employ a T piece to mixthe helium carrier gas with a wet aerosol obtainedusing the ICP-MS nebuliser.

For the detection of nonvolatile species usingthis approach the only difference is the useof a reaction cell before the cryogenic trap.In this reaction cell the nonvolatile species arederivatised, converted into volatile species andtransported by a flow of helium to the cryogenictrap (see Figure 4.2.1). For derivatisation, hydridegeneration has been mainly employed [107, 108]but also aqueous ethylation has been used [48]with satisfactory results for lead, mercury andtin species. However, it was observed that thedetection of high molecular weight hydrides andethylated compounds suffered from condensationand memory effects [109].

An interesting development in this approach isthe possibilities of automation of the whole system[110]. For this purpose, a modified split/splitlessinjector using a programmed temperature vaporizer(PTV) was used to trap at −40 ◦C ethylatedmethyl- and butyltin species. To prevent bandbroadening, the whole chromatograph was cooledto preconcentrate the analytes at the top of thechromatographic column. The detection limits,using standard GC-MS instrumentation, were inthe low µg L−1 levels.

3.2 Gas chromatography withpacked columns

Packed column GC was the first chromatographictechnique to be applied for elemental speciation. It

SEPARATION TECHNIQUES USED 175

has important advantages over thermal desorptionfrom either cryogenic traps or solid sorbents:(i) the interaction with the chromatographic pack-ing influences the separation of the analytestogether with the differences in volatility, (ii) thecolumns are stable and can be used many times,(iii) the analytes can be introduced dissolved inorganic apolar solvents which allows the analy-sis of elemental species derivatised using Grignardreactions and (iv) the separation can be improvedby temperature programming improving the chro-matographic resolution.

However, packed column GC lacks the neces-sary resolution for the analysis of complex envi-ronmental and biological samples and cannot beroutinely applied unless a selective or specificelemental detector is used. For elemental specia-tion using atomic detectors most of the first sep-arations were described using this methodology.The packing material used was generally 5–10 %Carbowax or OV-101 supported on Chromosorb.These first studies will be discussed later in thischapter.

3.3 Gas chromatography withcapillary columns

The use of capillary columns provides improvedresolution as compared with packed or mega-bore columns. This resolution improvement canbe clearly observed in Figure 4.2.2 [109] whenmoving from a packed column to a megaborecolumn and finally to a capillary column for theseparation of the same analytes. Additionally, thereduction in the carrier gas flow rate from 15–20to 1–4 mL min−1 provides increased sensitivitydue to the lower dilution factor in the mobilephase. However, we have to take into account thatthis increased chromatographic resolution will alsodepend on the connections between the column andthe detector. For atomic detectors, adequate inter-faces have to be built which do not produce extraband broadening and, hence, losses of theoreticalplates in the coupling [109].

The main limiting factor of the use of capillarycolumns is their limited loading capacity. Typi-cally, 1–2 µL of sample are injected in the columnwhich, in most cases, represents a small percentage

0 1 3 5 7 9 11 0 3 6 9 12 3 5 7 9 11

1

12

4

2

34

5

67

8

9

1

23

4

5

6

789

5

6 7

8

9

Figure 4.2.2. Improvement in chromatographic resolution from a packed column (left), to a megabore column (center) and to acapillary column (right) [109]. (Reprinted with permission from Elsevier Science.)


of the total sample volume. In speciation analysisthis can be a problem as very low concentrationshave to be measured usually. Different alternativeshave been proposed to increase the sensitivity ofcapillary GC: (i) sample preconcentration by evap-oration to dryness and reconstitution in a lowersolvent volume, (ii) the use of electronic pressurecontrol systems in large volume injectors [111] and(iii) the preconcentration of the analytes in tem-perature programmed high volume injectors whichallow solvent purge at low temperature [62, 112].However, the first of those procedures causes lowreproducibility and the second is limited by thelarge amount of solvent reaching the detector.

The efficiency of capillary columns is increasedby decreasing the internal diameter of the col-umn, but the loading capacity decreases. Thisproblem can be solved by resorting to multicap-illary columns. Recently, multicapillary columnsconsisting in a bundle of ca. 1000 capillar-ies of 40 µm internal diameter have been com-mercialised [113]. The high efficiency of thesecolumns allowed the use of carrier gas flows ashigh as 100 mL min−1, reducing the analysis timeto one-tenth of that required for a standard capil-lary column without reduction in loading capacity.Another advantage of multicapillary columns istheir ability to work under isothermal conditionswhich simplifies the instrumentation used [114].Columns of this type were first applied for elemen-tal speciation by Rodriguez Pereiro and cowork-ers [115–119] using atomic detectors. The smalllength of these columns and the use of isothermalseparations allowed the construction of a miniatur-ized speciation instrument coupled to a MIP-AESdetector [120].

4 ICP-MS AS DETECTOR FOR GASCHROMATOGRAPHY

4.1 Development of atomic gaschromatography detectors forelemental speciation

Detector traditionally used in the analysis oforganic compounds by GC, such as the flame ion-isation detector (FID) or the thermal conductivity

detector (TCD), are not suitable for elemental spe-ciation due to their lack of selectivity and sensitiv-ity. The electron capture detector (ECD) has beenapplied for the detection of Pb [121], Sn [122],Hg [123], Se [124] and As [125] compounds butit cannot ensure specific detection. Atomic spec-troscopy detectors (AAS, AES, AFS and elementalMS) are, on the other hand, perfectly suited for ele-mental speciation analysis by GC due to their highelemental sensitivity and selectivity. The fact thatmost elemental species are measured in complexsamples at very low concentration levels requiresthe use of selective or, better, specific detectorswhich only respond to the element of interestwithout interferences from co-eluting compounds.The big advantage of the coupling of gas chro-matographs to atomic detectors is that the influenceof the matrix can be reduced or totally eliminatedallowing near-specific detection.

Atomic absorption spectroscopy (AAS) is con-sidered the most selective atomic spectroscopictechnique for trace element speciation [126]because of the so-called key and lock mechanism[127]: only the element of interest will be able toabsorb the radiation emitted by the hollow cathodelamp. This advantage was considered of utmostimportance in earlier speciation studies. However,there is a clear disadvantage to this approach: wewill be able to detect only those species whichcontain the element of interest. Nowadays it is con-sidered necessary to be able to detect selectivelyand simultaneously different elements if required.Complex environmental and biological samplesmay contain species of different elements whichcould be interrelated. In this sense, atomic emis-sion detectors, such as the flame photometric detec-tor (FPD) or the microwave induced plasma atomicemission detector (MIP-AED) have been consoli-dated as suitable detectors for elemental specia-tion with GC. The combination of MIP-AED withsimultaneous diode array detection is one of themost powerful combinations nowadays for elemen-tal speciation.

In the first years of speciation studies, AASestablished itself as the most popular GC detectiontechnique for elemental speciation. The first waydevised to couple those two techniques was to

ICP-MS AS DETECTOR FOR GC 177

introduce the gaseous eluent from the GC columninto the spray chamber of conventional AASinstruments using a short, heated transfer line.Later, to avoid dilution of the GC eluent, theanalytes were introduced directly into the burner.The first coupling of GC and AAS was describedby Kolb et al. [128] in 1966 for the determinationof organolead compounds in gasoline. Since then,many applications have been published for thespeciation of Pb [129], Cr [130], As, Ge, Sn andSe [131] and many others. In order to increasethe residence time of the species in the flameand, hence, improve the sensitivity, Ebdon et al.[132] used a ceramic tube inside the flame wherethe analytes were ‘trapped’. This idea was alsofollowed in other laboratories [133, 134] but thebig breakthrough came from the use of electricallyheated tubes where electrothermal atomisation wasachieved. These electrically heated atomizers werebuilt either from graphite [135–137] or quartz[50, 138–140] and achieved temperatures between1000 and 2000 ◦C. Because of the lack of flamegases which diluted the sample, sensitivities wereca. 100-fold better than these for conventionalflame AAS.

Emission in flames has been applied for ele-mental speciation using mainly standard FPDs. Thespeciation analysis of tin using a FPD is well rep-resented in the literature [141, 142] owing to itsgood sensitivity and selectivity for this element.Also, its widespread availability and ease of usemade it the detector of choice in many modernspeciation applications [61, 81, 143]. However, thegeneral use of emission detectors for elementalspeciation arrived with the development of analyt-ical plasmas. Its high sensitivity and the possibilityof detecting several elements simultaneously madeatomic emission detection using plasmas an idealtool for elemental speciation. Research on the cou-pling of microwave induced plasmas (MIP) andinductively coupled plasmas (ICP) to gas chro-matographs for elemental speciation started in par-allel with the development of AAS detectors.

The first description of an MIP used as ele-mental detector for GC was given by McCormacket al. [144] in 1965. In this application organiccompounds were detected selectively using atomic

emission. The coupling of an MIP to the GC is rel-atively simple using a heated transfer line whichtakes the column only millimetres from the plasma,so reducing the dead volume [145]. Another advan-tage is that the use of He as carrier gas in thechromatograph is compatible with the MIP as thisgas is the most used plasma gas. The only dis-advantage of this coupling is the need to bypassthe plasma when the solvent front elutes or toswitch on the plasma after the solvent front haseluted [146]. This is due to the low tolerance ofthe MIP towards organic solvents and any molec-ular gas in general. The first commercial GC-MIP-AED instrument was commercialised by HewlettPackard (now Agilent) in 1989. This commer-cial system was applied by many laboratories forthe speciation of Pb [147], Sn [148], Hg [149]and Se [150] in both environmental and biolog-ical samples. Also, many laboratories have per-formed their own ‘home made’ GC-MIP couplingswith excellent results [151, 152]. Another GC-MIPsystem, the ‘Automatic Speciation Analyser’ hasbeen described recently [120]. This system com-bines on-line sample preparation by microwaveassisted extraction/derivatisation, cryogenic trap-ping and multicapillary GC separation coupled toa MIP-AES detector to automate the whole speci-ation procedure.

More details on the use of AAS and AESfor elemental speciation studies are given inChapter 5.1 of this book.

The use of the ICP as an emission source forelemental analysis first and as an ion source forelemental mass spectrometry later boosted researchon the coupling of this source to GC for elementalspeciation. In comparison with the MIP, the ICPsource is more tolerant to molecular gases andprovides better atomisation efficiencies. However,when the ICP was operated as an emission source,it provided worse sensitivity than the MIP due tothe dilution of the GC eluent in the high carriergas flow rate of the ICP. Also, the ICP providedlower excitation capabilities than the MIP for thedetection of nonmetals, and some elements couldnot be detected at all due to the high plasmaemission background because of the entrainmentof air in the plasma (e.g. for H, C, O, N).


The development of the ICP as an ion sourcefor elemental mass spectrometry in the 1980sand the growing research on suitable GC-ICP-MSinterfaces resulted in a real alternative to the GC-MIP-AES detector for multielemental speciation.ICP-MS offers extremely high sensitivity andselectivity, true multielemental capabilities withfast scanning or simultaneous detectors and thepossibility of measuring isotope ratios. This lastcapability, in combination with isotope dilutionanalysis, is the base of the most powerfulanalytical technique for the validation of speciationmethodologies: speciated isotope dilution analysis.In the rest of this chapter we will focus onthe coupling of GC to ICP-MS for elementalspeciation studies.

4.2 Interfaces for GC-ICP-MS

In this section we will review the different interfacesdescribed for the GC-ICP-AES and GC-ICP-MScoupling. Because of the similarities between bothtypes of couplings, we will not distinguish betweenAES or MS detection.

The first interfaces were developed for the cou-pling of packed column GC to the ICP and wewill describe those first. In those first studies, theend of the chromatographic column was connectedto a T piece in which an additional flow of argonwas introduced (about 0.9 L min−1) to transport theanalytes to the plasma. This additional argon flowwas necessary to punch a hole in the central chan-nel of the plasma as the carrier gas flow from thegas chromatograph, 10–20 mL min−1, was insuffi-cient for this purpose. The addition of an extra gasflow for the coupling with an ICP is a constantfeature of most interfaces developed later. How-ever, this caused dilution of the analytes and lossesin sensitivity in comparison with MIP detectors.In these early studies different organic compoundswere evaluated and the detection capabilities ofthe GC-ICP-AES coupling compared with otherGC detectors [153–157]. Later, the end of thechromatographic column or the transfer line wasconnected directly to the torch, either at the base[158] or within the central channel [159] reducing

the problems of dilution of the analytes. Chong andHouk [159], using a stainless steel transfer line,found that dead volumes could be minimized byadequate positioning of the transfer line within thecentral channel of the torch. The direct coupling ofthe transfer line to the torch has the inconveniencethat the spray chamber and nebuliser have to beremoved for the GC coupling. Peters and Beau-chemin [160, 161] designed an interface whichsolved this problem. An automatic switching valveallowed the alternative use of the nebuliser or theGC interface without modifications to the instru-mental set-up. This interface also used a setupdescribed previously [162] which consisted in theaddition of an external sheathing argon flow whichprevented the dilution of the analytes and cen-tred the eluent in the central channel of the torch,improving their transport through the plasma.

At the beginning of the 1990s the first publica-tions on the coupling of capillary GC with ICP-MSwere described. Kim et al. [8, 163] used a novelinterface for the detection of Pb, Sn, Fe and Nicompounds. The interface consisted of a heatedaluminium bar with a longitudinal slit in whichthe capillary column was introduced. A heatingtape was used to maintain the high temperaturethroughout the interface. The last part of the chro-matographic column was inside a stainless steeltube and introduced in the central channel of thetorch almost to its end. The necessary argon make-up flow was introduced using a T piece. From thesame laboratory, an improved version of the inter-face was described in which the argon make-upgas was previously heated and cold spots in thesystem eliminated. This interface was applied forthe analysis of high boiling point compounds suchas Fe, Ni and V porphyrins [164, 165] with hightemperature capillary GC separation.

From the same period are the first paperspublished by Caruso’s group [166, 167] for thecoupling of low pressure ICP sources to GC for thedetection of brominated organic compounds. Thetransfer line used consisted of a heated stainlesssteel tube, isolated with glass fibre tape. A shortlength of a 0.25 mm i.d. fused silica column wasused to transfer the analytes to the torch. Aswith previous interfaces, a T piece was used to


introduce a small make-up flow of, in this case,N2 (at 100 mL min−1) needed to keep the plasma.The sampling cone was made of aluminium andmodified to be connected to the torch using an O-ring to keep the low pressure in the plasma. Theexclusion of air and the use of lower gas flowsresulted in lower polyatomic interferences in thedetection of S and P by ICP-MS. Later, Evanset al. [168] used a similar interface to study plasmaconditions in which both molecular and atomicinformation could be obtained. By modifying theforward RF power, the pressure in the plasmaand the mixture of plasma gases used, atomicinformation (with picogram detection limits) andmolecular spectra, similar to those provided byelectron impact, could be obtained. These are theso-called ‘tunable sources’, which would offerboth elemental (for quantification) and molecularinformation (for identification purposes).

Back in 1995 Prange and Jantzen [169]described a novel interface design based on aheated quartz transfer line in which the capillarycolumn was inserted all the way to the base ofthe plasma. For the first time, detection limits inthe femtogram range using ICP-MS detection wereobtained. Also, the simultaneous detection of dif-ferent Sn, Hg and Pb species was demonstrated.In this case, the argon make-up gas was previouslyheated to prevent peak broadening in the interface.

The suitability of a certain interface designfor GC-ICP-MS coupling can be indicated by thenumber of publications in which this interfaceis used after its first description. This adjectiveof ‘suitable’ can be applied then to the interfacedescribed by De Smaele et al. [170] also in 1995.This interface consisted of a heated transfer line2.5 m in length with three concentric tubes ofstainless steel, Teflon and fused silica respectively.Initially, the part of the transfer line inside theplasma torch was not heated and that causedcondensation of high molecular weight analytespresent in the samples in this part of the transferline. Later, after modifications to this part ofthe interface, the analytical characteristics of theinterface improved [171, 172].

Gallus and Heumann [173] coupled GC to ICP-MS using a relatively simple and inexpensive

interface. In this case, the capillary column wasconnected to a heated stainless steel transfer linethrough a six-way switching valve. The transferline was introduced in the torch up to about3 cm below the induction coil. By switching thevalve, gaseous standards could be introduced in theplasma using a flow cell. Both the flow cell and theswitching valve were inside the chromatographicoven. Mass bias, for isotope ratio work, could becorrected using this approach.

The interface described by Pritzl et al. [174]consisted of a special injector tube of 4 mmexternal diameter, constructed in stainless steel,which was attached to a heated transfer line(20–350 ◦C). A narrow tube, also in stainlesssteel, transversed the transfer line and injector tubealmost to the base of the plasma. The argon make-up gas, preheated in the oven, was introducedcoaxially with the column.

In the search for quantification of elementalspecies for which no standards were availabledifferent authors have designed interfaces in whichthe eluent from the column was mixed with anaqueous aerosol in the spray chamber [6] obtainingrelatively stable plasma conditions. By nebulisingaqueous standards of the element of interest asemiquantitative estimation of the concentration ofthe species in the samples was possible. Prohaskaet al. [37] used a similar approach for the couplingof a double focusing ICP-MS instrument to the gaschromatograph for the analysis of different arsenicspecies. The last part of the chromatographiccolumn was connected to the sample inlet of astandard concentric nebuliser and the nebuliseritself connected to the base of the torch by using apiece of PTFE tubing. Condensation of the analytesfrom the GC eluent in the gas phase would notaffect the separation and detection as the analyteswould be transported as an aerosol to the plasma.

A similar idea lies behind the interface describedby Montes Bayon et al. [9] depicted graphicallyin Figure 4.2.3. The last 10 cm of the column areinserted in a heated copper tube and connected toa Swagelok T piece. The make-up argon gas isintroduced through the side arm of the T pieceat room temperature and flows externally to the


Swagelok T

Capillary column

GC Oven

Metallic block

Cu tube

Steel tubeChromatographiccolumn

Argonmake-up

ICP-MS

Heater

Temp. sensor

Figure 4.2.3. GC-ICP-MS interface developed by MontesBayon et al. [9]. (Reproduced by permission of The RoyalSociety of Chemistry.)

heated copper tube inside the T piece. The con-densation of the analytes in the gas phase afterexiting the column would not cause band broaden-ing as the high velocity external argon flow wouldprevent deposition of the analytes in the transfertubing connecting to the torch. The authors useflexible room temperature PFA tubing for the con-nection between the interface and the torch. Themain advantage of this interface design in compari-son with those reported previously is its flexibilityand the use of room temperature transfer tubingfrom the interface to the torch. Further publicationsfrom the same research group [72, 175] demon-strated that peak profiles using this interface weresimilar to those obtained using standard GC detec-tors such as FID. Only slight band broadening wasobserved for high molecular weight species, suchas tetrabutyllead or tributylethyltin.

Krupp et al. [176, 177] were first to describethe coupling of a gas chromatograph to a multi-collector ICP-MS instrument for precise and accu-rate isotope ratio measurements. Their transferline consisted of a steel tube, electrically heated,inserted directly in the central channel of the torch.Using a T piece, a wet aerosol is introduced at the

Table 4.2.8. General characteristics of transfer lines describedfor capillary GC-ICP-MS coupling.

Year General characteristics Ref.

1992 Rigid transfer line in Al (φ = 2.5 cm).Four thermocouples for T control.Capillary column introduced almostto the plasma. Species with elevatedmolecular weight can be analysed.

164, 165

1993 Transfer line made of stainless steel(1 m, φ = 0.16 cm) heated at270 ◦C. Deactivated fused silicatube inside a water-cooled torch.Low pressure ICP.

167, 168

1995 Quartz transfer line heated at 240 ◦Cpermits the direct insertion of theanalytes in the plasma.

169

1995 Transfer line in deactivated silicainside an steel tube (2.5 m,φ = 0.31 cm). Ar make-up is heatedinside the GC. Flexible.

171, 172

1996 Transfer line in stainless steelconnected to the torch through asix-way valve inside the GC oven.Flow cell to introduce gaseousstandards.

173

1996 Special stainless steel injector(φ = 4 mm) to connect torch andheated steel transfer line(20–350 ◦C). Ar make-up gaspreviously heated inside GC ovenand added coaxially to the column.Relatively versatile.

174

1997 Heated PTFE tube (1 m, φ = 0.3 mm)to connect the capillary column tothe torch. Nebulised standards aremixed using a T piece. Standardlessquantification. Rigid.

6

1999 Capillary column introduced in thecentral channel of a concentricnebuliser which is connected to thetorch by a PTFE tube. Nonheated,rigid.

37

1999 Similar to standard GC detectors. Theeluent from the column is mixedwith the Ar make up gas in a Tpiece and connected to the plasmawith a piece of PFA tube. Flexible,PFA tube nonheated.

9, 72, 175

2001 Transfer line of heated steel tube. Armake-up, preheated in the GC oven,goes inside the transfer line. A Tpiece at the base of the torch is usedto introduce wet aerosols.

176, 177

base of the torch. This aerosol is used for the neb-ulisation of Tl required for mass bias correction intheir lead isotope ratio measurements. A summaryof the general characteristics of different GC-ICP-MS interfaces is given in Table 4.2.8.


4.3 Analytical characteristics of theGC-ICP-MS coupling

The coupling of a gas chromatograph to an ICP-MS instrument is not as simple and straightforwardas the use of the ICP-MS instrument as detectorfor liquid chromatography where the exit of thechromatographic column can be connected directlyto the nebuliser. As we have seen in the previouschapter special interfaces have to be constructedwhich increase the complexity of the coupling.However, there are several advantages whichmake GC-ICP-MS favourable in comparison withalternative HPLC-ICP-MS couplings [178]:

(i) High sample introduction efficiency. We canassume safely that 100 % of our sample willbe transported to the plasma. In comparisonwith the sample introduction efficiency oftypical nebuliser/spray chamber assemblies[2–3 %] used in HPLC coupled with ICP-MS,we can expect better sensitivity.

(ii) Efficient use of plasma energy. The ana-lytes are carried to the plasma only bymonoatomic gases, when He is used as carriergas, so plasma energy can be used more effi-ciently for atomisation and ionisation. Thereis no need to volatilise and atomise sol-vent molecules as it is for liquid sampleintroduction.

(iii) Separation of the solvent from the analytes.The use of the GC allows separation of thesolvent molecules from the analyte moleculesduring the chromatographic run. So, theanalytes arrive in the plasma only in thecompany of the GC carrier gas and the make-up argon gas. That means that, when theanalytes elute from the column, the plasmais stable and no distortion of the plasma bysolvent molecules is observed [160].

(iv) Stable plasma conditions during the chromato-graphic run. The chromatographic separationtakes place using temperature programmingand not by gradient elution as is typical ofHPLC separations. This means that the sensi-tivity remains constant during the separation,a stable baseline is obtained and, hence, lowdetection limits can be achieved.

(v) High chromatographic efficiency. In compar-ison with HPLC peak widths are much nar-rower and so signal to noise ratio is betterusing GC.

(vi) Low spectral interferences. The use of adry plasma and the absence of any solventlead to much less spectral interferences incomparison with solution nebulisation.

(vii) Low signal drift. The amount of materialreaching the sampler and skimmer conesis very small so there is less chance ofcone blocking.

All these characteristics make the GC-ICP-MScoupling ideal for ultratrace speciation analysisin comparison with alternative HPLC procedures.As an example, Table 4.2.9 compares absolutedetection limits published for the speciation oforganotin compounds using both HPLC [179] andGC [9, 62, 63, 180–185] as separation techniquesand using different detectors. When comparingdifferent GC detectors, the best detection limits areprovided by the GC-ICP-MS coupling which goesdown to the single femtogram range using shieldtorch instruments [62]. The comparison betweenHPLC-ICP-MS and GC-ICP-MS is also clear:more than three orders of magnitude of differencein sensitivity in two papers published by the sameauthors [9, 179].

Most publications using the GC-ICP-MS cou-pling are devoted to the speciation analysis of Sn,Hg, Pb, As and Se. Table 4.2.10 shows examplesof the detection limits which can be achieved with-out special preconcentration procedures for the

Table 4.2.9. Absolute detection limits (pg as Sn) published forthe detection of tin species using different detection techniques.

Technique Absolute detectionlimits (pg as Sn)

Ref.

GC-MS 0.5–1 180GC-QF-AAS 700–1200 181GC-ECD 1–50 182GC-MIP-AES 0.4 63GC-ICP-AES 25 183GC-FPD 3–160 184GC-FPD (quartz induced

lumines)0.8 185

GC-ICP-MS 0.05–0.1 9GC-ICP-MS (shield torch) 0.0007–0.0016 62HPLC-ICP-MS 200 179


Table 4.2.10. Absolute detection lim-its (pg) achieved by capillary GCcoupled to ICP-MS for the detectionof organometallic species of differentelements.

Element Range (pg) Ref.

Sn 0.05–0.1 9Hg 0.09–0.83 68Se 2.5 106Pb 0.002–0.009 72As 0.54 6Bi 0.09 6I, Br, Cl 16, 40, 158 168

detection of these and other species using capil-lary GC-ICP-MS. As can be observed, detectionlimits are at or below the picogram range for mostelements except for the halogens.

In spite of the extremely low limits of detectionoffered by standard GC-ICP-MS, different authorshave devised procedures for reducing even morethose detection limits. In this sense, preconcen-tration systems such as purge and trap [186] andlarge volume injectors [62] have provided relativedetection limits in the low ng L−1 to pg L−1 con-centration range. In combination with shield torch,Tao et al. [62] obtained sub-femtogram detectionlimits for butyl- and phenyltin compounds. Theyobserved that the use of a shield torch with adry plasma offered a 100-fold increase in sensi-tivity in comparison with standard plasma condi-tions. In this way, concentrations of organotin com-pounds in the low pg L−1 range were detected inopen-ocean seawaters. However, this tremendousincrease in sensitivity required ultraclean samplepreparation methodologies to be employed [62].For example, recent work carried out in our labo-ratory for the determination of butyltin compoundsin coastal seawater samples by GC-ICP-MS [187]has shown that blank values for both TBT, DBTand MBT reduced well below 1 ng kg−1 for 100 mlsamples when sample preparation was carried outunder clean room conditions (class 1000).

The optimisation of sensitivity using the GC-ICP-MS coupling has been approached from dif-ferent angles. It is clear that the optimum plasmaand ion lens conditions used for wet plasmas differgreatly from those under dry plasmas and, hence,

adequate optimisation strategies have to be imple-mented. To avoid time-consuming optimisations bysequential injections of standards in the system,many authors employ xenon traces (1 %) added tothe chromatographic carrier gas for optimisation.Gas flows, torch position, plasma and ion lens con-ditions are optimised by continuous monitoring ofthe 126Xe signal. De Smaele et al. demonstratedthe clear correlation between optimum conditionsfor Xe and those for organotin species [171]. Thisoptimisation strategy was also employed by dif-ferent authors [106, 119]. In other cases, volatilespecies of the element of interest are continuouslyfed to the plasma for optimisation, such as arsine[174] or mercury vapour [117] using a switchingvalve. Ruiz Encinar and coworkers [10, 175, 188]have applied a simple approach: they used poly-atomic argon species, m/z = 80 (40Ar2

+), only forthe optimisation of the ion lens after the optimi-sation of plasma conditions had previously beenperformed by nebulisation. These authors indicateddrastic changes in the optimum values of differention lenses of a quadrupole ICP-MS instrument bychanging from wet to dry plasma conditions.

The GC-ICP-MS interfaces described inTable 4.2.8 do not provide means of diverting theeluting solvent front from the plasma. This isdue to the robustness of the ICP in comparisonwith other plasmas and its high tolerance towardsorganic solvents. However, this mode of operationcan cause carbon deposits in the sampling conewhich, in turn, give rise to signal drift due topartial blocking of the sampling cone. Whenthis effect was observed, the addition of a smallflow of oxygen to the argon make-up gas (about20 mL min−1) and/or the increase of the RFpower eliminated these carbon deposits [63, 119].However, it has been observed recently [117] thatthe addition of oxygen caused a severe decrease inthe sensitivity for mercury which was correlatedwith the flow of O2 added. Many applicationpublished using GC-ICP-MS do not use an extraoxygen flow.

The selectivity of GC-ICP-MS for elementalspeciation is excellent. In comparison with thecomplex emission spectra of GC-MIP-AES or GC-ICP-AES, the simplicity of the mass spectra using


the ICP as an ion source is remarkable. In thissense, ICP-MS is very selective and the naturalisotope abundances of the elements give a clearindication of the presence of a given elementalspecies in the sample [189]. New spectral inter-ferences have not been described for the speci-ation of trace elements using GC-ICP-MS bothbecause of lack of formation of new polyatomicions and also because of the chromatographicseparation from the solvent peak. Organic com-pounds in the plasma decompose completely andno new polyatomic ions containing carbon havebeen described using GC-ICP-MS. The only poly-atomic species observed by GC-ICP-MS is 81BrH+which would interfere with 82Se in the specia-tion of selenoaminoacids. Bromine impurities inthe chloroform solvent gave a large solvent peakat mass 82 which was well separated from theselenium-containing aminoacids and hence did notinterfere [90]. Other well-known spectral interfer-ences in ICP-MS, such as 16O2

+ on 32S+, havealso been observed by GC-ICP-MS but their mag-nitude was much lower than when using aqueousnebulisation [17].

The multi-isotopic and multi-elemental capabil-ities of the ICP-MS have also been exploited forelemental speciation using GC-ICP-MS. The onlytrue limitation of this coupling is the fast transientchromatographic peaks, of ca. 2 s, which limit thenumber of isotopes which can be monitored simul-taneously in sequential instruments. If we assumethat we need at least ten data points to followaccurately a chromatographic peak and that thispeak lasts for ca. 2 s, the optimum total integrationtime for all isotopes selected will be about 200 ms.In sequential instruments, such as quadrupole sys-tems, this time will have to be divided between thenumber of selected isotopes reaching a practicallimit of about ten isotopes monitored simultane-ously (assuming that fast peak jumping and onlyone point per peak are used). This limit is set alsoby the fact that counting statistics worsen dramati-cally as the integration time per isotope is reducedto a few ms and this will also affect the achiev-able detection limit of the system. However, forsimultaneous instruments, such as time of flight

systems, the problem is the opposite. These instru-ments acquire ca. 20 000 full spectra per secondwhich can be accumulated up to the selected inte-gration time. If we assume that we can accumu-late spectra for a maximum of 200 ms, to obtainadequate GC peak profiles, and that we have todivide this integration time between the total num-ber of acquired data points in the mass spectrum(ca. 25 000 data points in the LECO ICP-TOF-MSinstrument), the total integration time per pointswill be about 0.008 ms. That will make countingstatistics very poor for GC-ICP-TOF-MS, but trulymulti-isotopic and multi-elemental.

In spite of these capabilities, most multi-elemental or multi-isotopic speciation studies pub-lished using GC-ICP-MS concentrate only on themeasurement of a few elements or a few iso-topes. For example, the simultaneous speciation ofHg, Sn and Pb has been performed successfully[97]. The combination of multicapillary GC withICP-MS allowed the determination of 11 speciesof these three elements in less than 2 min [119].Multi-isotopic studies have also been publishedusing both quadrupole and multicollector ICP-MS.In the first study [10] three Pb isotopes weremeasured for isotope ratios in the assessment oforganolead sources in airborne particulate matter.In the second study, [176, 177] all four Pb isotopesplus 202Hg and 203,205Tl were measured simultane-ously in ethylated NIST 981 lead isotopic standard.

Long-term stability of GC-ICP-MS systems hasbeen a subject of worry for most people working inthis field. Sensitivity drift is common in ICP-MSinstruments and this can be worsened by carbondeposits in the cones. For this reason the use ofinternal standards is mandatory for GC-ICP-MSwork. Different modes of internal standardizationhave been proposed for GC-ICP-MS:

(a) Methodological internal standards. As in anyother analytical technique, the complexity andlarge number of analytical steps involved in speci-ation analysis require the use of adequate internalstandards to compensate for errors in the wholespeciation process. These internal standards areadded to the sample at the beginning of the ana-lytical procedure to compensate for nonquantita-tive recoveries, low derivatisation yields, dilution


errors and so on. Obviously, these kind of internalstandards have to behave like the analytes withsimilar chemical and physical properties. Someauthors [63, 190] have indicated that, when thecorrection factors are elevated, it is recommendedto apply simultaneously standard additions to thesample. Examples of this type of internal stan-dards, which of course would also correct for driftin the GC-ICP-MS system, are tripropyltin (forbutyltin compounds) or ethylmercury (for inor-ganic and methylmercury). The use of isotopi-cally enriched elemental species [175, 188] wouldfall also in this category of methodological inter-nal standards.

(b) Instrumental internal standards. There are twogeneral procedures used to correct for instrumentalvariations in GC-ICP-MS. The first method con-sists in the addition to the final sample of a volatilecompound, not present in the sample, but con-taining the metal under study [191]. This way ofinternal standardization would compensate for bothsensitivity drift and injection precision. Examplesof this type of internal standards are tetraalkylatedSn [169] or Pb [72] compounds and dialkylatedHg compounds [68]. The second method consistsin the use of an impurity in the carrier gas (usu-ally Xe) for signal normalization. This mode ofinternal standardization will compensate only forinstrumental drift [51, 80, 102, 171, 172]. In manycases a combination of both modes of instrumentalcorrection is used [102, 171].

4.4 Isotope ratio measurements withGC-ICP-MS

The analytical characteristic which makes thecoupling of a gas chromatograph to the ICP-MSinstrument unique in comparison with the other GCdetectors indicated in Table 4.2.9 is the possibilityof direct measurement of elemental isotope ratios.We have indicated in the previous chapter thatthe ideal internal standard would have to possesschemical and physical properties similar to theanalytes. It is clear that another isotope of thesame element would be the ideal internal standardand, hence, the measurement of elemental isotope

ratios by GC-ICP-MS will open the way forthe application of isotope dilution methodologiesin elemental speciation. Another advantage ofthe use of isotope ratios is the improvement inmeasurement precision. For example, injectionuncertainties of ca. 6–7 % were reduced to 2–3 %using Xe impurities in the carrier gas [80]. Asimilar injection uncertainty, in peak area mode[10, 175], was reduced to 0.5–1.5 % by measuringisotope ratios.

The first publication on the measurement ofisotope ratios by GC-ICP-MS appeared in 1987[159]. In that publication Chong and Houk mea-sured isotope ratios for B, C, N, Si, S, Cl and Brin different organic compounds with instrumentalprecisions ranging between 0.37 and 18 %. Theyconcluded that this technique could be appliedto study reaction mechanisms and in metabolicstudies using stable isotopic tracers but the pre-cision would be inadequate for the study of nat-ural isotope fractionation effects. Later, in 1993,Peters and Beauchemin [161] measured isotoperatios in organotin compounds and found iso-tope ratios in reasonably good agreement withthe natural tin ratios. Precision was in the range0.3–3.6 % in peak height mode. Nowadays, itis considered that peak area offers more preciseand accurate isotope ratio measurements than peakheight [10, 173].

The optimisation of the measurement param-eters is critical in obtaining adequate isotoperatio measurements by GC-ICP-MS. When usingsequential instruments (quadrupole, single collectordouble focusing) the selection of the right inte-gration time will depend both on the ability tofollow adequately the peak profiles and on opti-mising counting statistics [10]. In general, the fasttransient signals generated by the GC are betterfollowed using short integration times per isotope.However, under those circumstances countingstatistics might be poor and isotope ratio precisionsinadequate. On the other hand, for longer integra-tion times counting statistics improve but ICP-MSis no longer able to follow the peak profiles accu-rately. Under those circumstances spectral skewin sequential instruments starts to be noticeableand isotope ratio precision is reduced [10]. These


Counts/sCounts/s Counts/s

5.0E5 2.0E5

1.0E5

5.0E5

2.5E52.5E5

235 236 237 238 239 240 241 242 243 235 236 237 238 239 240 241 242 243 235 236 237 238 239 240 241 242 243

Time (s)Time (s)Time (s)

208 : Pb

206 : Pb

208 : Pb

206 : Pb

208 : Pb

206 : Pb

40 ms10 ms 160 ms

Figure 4.2.4. Influence of integration time on the GC-ICP-MS peak profiles measured for MeEt3Pb, at masses 206 and 208, inleaded gasoline using a quadrupole instrument [10]. (Reproduced by permission of The Royal Society of Chemistry.)

two opposed effect are illustrated in Figure 4.2.4for the measurement of lead isotope ratios intetraalkyllead compounds in gasoline [10]. As canbe observed, for a very short integration time perisotope (10 ms) peak profiles are very well definedbut show high noise. For a long integration time(160 ms) peak profiles are poor and spectral skewis noticeable. A compromise between noise andspectral skew was obtained by keeping the totalintegration time between 150 and 200 ms [80, 188].Under those conditions from two to eight isotopescould be monitored simultaneously using integra-tion times from 25 to 100 ms [192].

In order to obtain accurate isotope ratio mea-surements using GC-ICP-MS we have to take intoaccount factors such as detector dead time, massbias, spectral interferences and chemical blanks.We will discuss briefly those factors and the wayin which they can be corrected.

(a) Detector dead time is the time needed for thedetection and electronic handling of single detec-tion events. This effect is the cause of nonlinearityof pulse-counting detectors at high counting ratesbut does not affect analogue detectors [193]. Whenan isotope ratio is measured, counting losses willaffect one isotope differently from the other and

so the measured isotope ratio will be inaccurateand will depend on the concentration level beingmeasured. There are different ways in which thedetector dead time for a given instrument can becomputed [194], but in most cases the measure-ment of isotope ratios by nebulisation of standardsat different concentration levels is necessary. Oncethe detector dead time is known, the measuredintensities on each point of the chromatogram haveto be corrected using the equation:

NT = N0

1 − N0τ

where N0 and NT are the observed and realcount rates (counts s−1) and τ is the detector deadtime (s). The use of analogue measurement modesis less sensitive than pulse counting but does notrequire dead time correction [195].

(b) Mass bias, or mass discrimination, is a well-known effect in all ICP-MS instruments. It consistin the preferential transmission of heavier ionsin the mass spectrometer which results in isotoperatio measurements which are biased by a constantfactor [196]. The value of this mass bias factordepends mainly on the mass difference between theisotopes and can change from element to element,


Table 4.2.11. Isotope ratios measured for natural standards of MBT, DBT and TBT using GC-ICP-MS. Uncertainty expressedas standard deviation for three injections of 800 pg as tin.

Isotope Natural Experimental ratios measured �M Averageratio ratio relative error

MBT (s) DBT (s) TBT (s)

116/120 0.4463 0.4350 (0.0034) 0.4357 (0.0025) 0.4334 (0.0065) 4 −0.02596117/120 0.2357 0.2296 (0.0020) 0.2287 (0.0027) 0.2313 (0.0026) 3 −0.02489118/120 0.7434 0.7333 (0.0054) 0.7348 (0.0022) 0.7373 (0.0067) 2 −0.01110119/120 0.2637 0.2599 (0.0001) 0.2610 (0.0030) 0.2586 (0.0040) 1 −0.01458122/120 0.1421 0.1424 (0.0008) 0.1434 (0.0006) 0.1431 (0.0017) −2 0.00624124/120 0.1777 0.1824 (0.0027) 0.1809 (0.0032) 0.1806 (0.0039) −4 0.02004

from instrument to instrument and from day today. In order to illustrate this effect, Table 4.2.11shows the experimental and theoretical tin isotoperatios measured by GC-ICP-MS for mono-, di-and tributyltin (MBT, DBT and TBT respectively)[188]. As can be observed, the relative errorin the measured ratio, (Rexp − Rtheo)/Rtheo, is afunction of the mass difference between the twoisotopes, �M . By plotting the relative error versusthe mass difference a linear function is usuallyobtained. The slope of this line is the mass biasfactor, K . Once K is determined for a given GC-ICP-MS system by injecting isotopic standards,all subsequent isotope ratio measurements canbe corrected [197]. For elements which do notshow natural variations, such as tin, the naturalelement showing natural isotope composition canbe used [198]. Following this approach, RuizEncinar et al. observed that the mass bias factormeasured for organotin compounds drifted slightlyafter repeated injections of a natural abundancemixed butyltin standard during a run of 4 h [10,175, 188]. This problem was solved by injectinga natural abundance standard every three samplesand correcting each group of three samples usingthe average of the K values measured beforeand after the samples. A different approach wasfollowed by Gallus and Heumann [173]. In thiscase, a system for the continuous introduction ofa volatile natural selenium compound was devisedand used for mass bias correction. Finally, Kruppet al. [176, 177] added natural thallium using on-line nebulisation for external mass bias correctionof lead isotopes with satisfactory results. It is worthnoting that mass bias is an instrumental effectand hence independent of the elemental species

being monitored (see Table 4.2.11). This meansthat mass bias correction could be performed inthe same chromatogram using a different elementalspecies added to the sample with well-knownisotope composition [173, 175].

(c) Chemical blanks affect the measurement ofisotope ratios in elemental speciation in a subtleway: only those species which show significantblank values will be affected. For example, highinorganic tin blanks will not affect the measure-ment of isotope ratios for butyltin compounds afterethylation [188]. To understand the effect of chem-ical blanks we need to consider two possibilities:isotope ratio measurements for source character-ization, such as Pb source studies, or for isotopedilution analysis. In the latter case, chemical blanksare treated as other samples and the concentrationin the blank computed by isotope dilution analy-sis and subtracted from that in the sample. Whenthe measurements are performed for source char-acterisation chemical blanks are more difficult totake into account as the isotope composition ofthe blank will be slightly different from that in thesample. In those cases blank subtraction of abso-lute intensities at different masses might be theonly alternative, but only when the blank valuesare much smaller that those in the sample. Thereduction of chemical blanks to negligible levelsby working with ultrapure reagents is advisable.Blank values affecting isotope ratio measurementshave been reported for the speciation of Hg [197]and Pb [10] by GC-ICP-MS. For the speciationof butyltin compounds in sediments blank valueshave been reported to be negligible [175, 188] buthigh blank levels have been found for the analysisof seawater samples when high preconcentration


factors are utilised [62, 187]. This may be due totheir use as stabilizers in different plastics or tocontamination of the laboratory atmosphere. Fromour experience, using only cleaned glass contain-ers and working under clean room conditions areessential to measure low levels of butyltin com-pounds in seawater samples [187].

(d) Spectral interferences by GC-ICP-MS are lesssevere than when using conventional nebulisation.Isobaric interferences (e.g. 204Hg and 204Pb) arenormally separated in the chromatographic columnas the elution time of different lead and mercurycompounds will be different and polyatomic inter-ferences (e.g. 38Ar40Ar on 78Se) will only increasethe chromatographic baseline by a constant value.The integration of the GC peak will normally com-pensate for a small baseline value. Only in thecases when the baseline value is high and noisyin comparison with the analyte peak (e.g. 16O2

on 32S) will we need to resort to high resolutionmass spectrometers [17] or collision cell instru-ments. No other serious spectral interferences havebeen reported for isotope ratio work using GC-ICP-MS.

4.4.1 Applications of isotope ratiomeasurements

There are many different fields in which isotoperatio measurements with GC-ICP-MS have proveduseful. For example, in the study of isotope frac-tionation, the determination of isotope ratios indifferent elemental species can offer informationwhich will be lost after total digestion of the sam-ple. Many environmental, biological and geolog-ical processes produce compounds which mightshow natural or man-made isotope fractionation.Examples of those elemental species which mightshow changes in isotope ratios and can be detectedby GC-ICP-MS are those of sulfur and lead [176].Of course, the better the level of precision reachedin the GC-ICP-MS coupling the larger this field ofapplication will be.

The differentiation between sources of leadorganometallic compounds was recently demon-strated by Ruiz Encinar et al. [10] using quad-rupole ICP-MS. The measurement of lead isotoperatios by GC-ICP-MS of organolead standardsallowed them to differentiate between lead sources.For example, Figure 4.2.5 shows some of the

0.46

0.45

0.44

0.43

0.42

0.41

0.4

0.39

0.83 0.85 0.87 0.89 0.91 0.93 0.95

207/206

207/

208

Organolead compoundsin leaded gasoline

TML

TEL

DEL

DML

EthylatedPb (Merck)

Figure 4.2.5. Lead isotope ratios measured for different organolead standards [10]. (Reproduced by permission of The RoyalSociety of Chemistry.)


results obtained: tetraalkyllead compounds presentin gasoline (Me4Pb, Me3EtPb, Me2Et2Pb, MeEt3Pband Et4Pb) could be differentiated from buty-lated Et3Pb+ (TEL) or Me3Pb+ (TML) standardsobtained from ABCR (Karlsruhe, Germany) whichthemselves showed a different lead signature. Thereaction of the trialkylated species with iodinemonochloride to form the corresponding dialky-lated species (DEL and DML), which were thenbutylated to form the volatile species, did not pro-duce drastic isotope changes. Finally, an inorganiclead ICP standard from Merck was ethylated usingtetraethylborate and also showed a different leadsignature by GC-ICP-MS (see Figure 4.2.5). Theexcellent precision results obtained by Krupp et al.using a multicollector ICP-MS [176, 177] for leadisotope ratio measurements by GC-ICP-MS arevery promising for these type of studies.

Another field of application of isotope ratiomeasurements by GC-ICP-MS is in the studyof artefacts during sample preparation. The useof certain sample preparation procedures mayalter the speciation of a given element. In thesecases, the use of isotopically labelled elementalspecies has allowed researchers to study these pro-cesses in depth. The isotopically labelled speciescan be followed throughout the whole analyti-cal procedure and any change in the isotopiccomposition of another species will be shown byisotope ratio measurements. In this way, Hintel-mann et al. [197] measured methylation veloci-ties of inorganic mercury in sediments by using199Hg-enriched (92.57 at%) mercury nitrate. Later,the same authors studied methylation of inorganicmercury under different sample preparation con-ditions [199] such as distillation and acid andbasic extractions. They concluded that methylationdepended both on the type of sample (speciallythe organic matter content) and on the samplepreparation procedure employed. In a similar studyusing inorganic 202Hg (97 at%), Garcıa Fernandezet al. [68] did not observe methylation of inorganicmercury during derivatisation processes involvingNaBEt4 or NaBPr4. Similar results were obtainedby Ruiz Encinar et al. [175] in the study of trans-butylation reactions for butyltin compounds duringextraction and derivatisation with NaBEt4. For this

purpose, a highly pure DBT standard enriched in118Sn (98.44 at%) was synthesized and added toa certified sediment containing all three butyltincompounds. The measurement of isotope ratiosby GC-ICP-MS demonstrated that only the DBTpeak showed an altered 120/118 isotope ratio. Thesame authors observed that lead isotope ratioswere unchanged during derivatisation with NaBEt4using enriched 204Pb [10].

The applications of isotope dilution analysis fortrace element speciation will be discussed later inthis chapter.

4.5 Comparison of different ICP-MSinstruments

Most of the GC-ICP-MS applications described todate have been performed on quadrupole ICP-MSinstruments. This is mainly due to their relativehigher abundance in analytical chemistry labo-ratories but we also have to take into accountthat their good sensitivity, speed and robustnessmake them suitable for most GC-ICP-MS appli-cations. However, there are certain applicationswhere other type of ICP-MS instruments wouldbe a better choice.

Double focusing instruments offer higher sen-sitivities at low resolution power but their mainadvantage over quadrupole systems is in the spe-ciation of spectrally interfered elements such assulfur, phosphorus and silicon where the highcontinuous background makes speciation withquadrupole systems impossible. For example,Rodriguez Fernandez et al. [17] measured volatilesulfur compounds in fermented saliva using a dou-ble focusing ICP-MS instrument as detector atmass 32 (95 at% abundance). Working at 3000 res-olution power they could separate the sulfur peakfrom the polyatomic 16O2

+. By using the mostabundant sulfur isotope they could reach detectionlimits between 7 and 33 µL L−1 (v/v) for 1 mLof sample. Figure 4.2.6 shows one of the chro-matograms obtained for headspace sampling ofsaliva incubated at 37 ◦C for 48 h. Most of thesesulfur compounds could be identified and they arethought to be related to bad breath problems [17].


0 100 200 300 400 500 600

?

?

80000

70000

60000

50000

40000

30000

20000

10000

0

cps

(m/z

= 3

2)

Time (s)

CH

3SH

(CH

3)2S

(C2H5)2S

(C2H5)2S2

(CH3)2S2

H2S

Figure 4.2.6. GC-ICP-MS chromatogram of volatile sulfur compounds in fermented saliva obtained with a double focusinginstrument at mass 32 and resolution power of 3000 [17]. (Reproduced by permission of The Royal Society of Chemistry.)

The speciation of organosilicon compounds wasalso performed using an analogous system at res-olution 3000 and mass 28 [200].

The high sensitivity capabilities of doublefocusing instruments was evaluated by Prohaskaet al. [37] for arsenic speciation at mass 75 inlow resolution. This instrument was evaluatedfor the detection of volatile arsenic compounds(arsine, mono-, di- and trimethylarsine) in amicrocosm experiment. Similarly, high sensitivityapplications for selenium speciation have beendescribed [201].

A current limitation of quadrupole andsingle collector double focusing instruments isthe sequential nature of their measurements.For certain applications, such as isotope ratiomeasurements or multielemental speciation, asimultaneous measurement of several isotopesor elements would be advantageous. This

simultaneous measurement is offered nowadays bytime of flight instruments and multicollector sectorfield instruments.

The expectations that arose with the introduc-tion of time of flight ICP-MS instruments forelemental speciation have not been completelyfulfilled. In principle, the measurement of manyisotopes in fast transient signals, such as thoseobtained by GC separations, would be better per-formed on simultaneous instruments avoiding thewell-known ‘spectral skew’ effect of sequentialinstruments. Also, for the precise measurement ofisotope ratios, simultaneous instruments would bebetter suited. The main problem of TOF-ICP-MSinstruments for elemental speciation is the needto acquire whole mass spectra continuously. Whenwe need to focus on a few masses only, we wouldlike to spend the maximum time integrating thosemasses and not to waste precious time on unwanted


portions of the spectrum. In this way, the sensitiv-ity of TOF instruments is ca. 5–10 times lowerthan quadrupole instruments and precision of iso-tope ratios for GC transient signals was between1.2 and 2.9 % RSD [202, 203] and limited bycounting statistics. However, when high concen-trations were used and the measurements wereperformed in analogue mode isotope ratio pre-cisions using GC-ICP-TOF-MS ranged between0.3–0.5 % RSD [204]. For quadrupole instruments,where the use of time is more efficient, isotoperatio precisions were clearly below 1 % in countingmode [10, 175, 188]. The accuracy of isotope ratiomeasurements using TOF instruments was ade-quate in analogue mode but reduced seriously incounting mode when peaks larger than 20 000 cpsreached saturation [204]. Of course, when trulymultielemental speciation is required and no a pri-ori information of the species present is given,such as the studies performed on landfill and othergases [3–6], TOF instruments could be the bet-ter choice.

The situation is completely different usingmulticollector instruments. In these cases weare continuously monitoring several isotopes, sotime usage is optimal. Minor fluctuations in theplasma are also compensated so isotope ratios,even for fast transient signals, should be veryprecise indeed. To date, very few applicationsof GC-ICP-MS using multicollector instrumentshave been published [176, 177, 205] but weshould expect much more in the near future.Isotope ratio precisions for major lead isotopesin tetraethyllead were in the range 0.02–0.07 %RSD. The accuracy of the measurements wasalso excellent for NIST 981 lead isotope stan-dard (between 0.02 and 0.15 % deviation fromthe certified values). This isotope ratio accuracywas accomplished by on-line mass bias correc-tion using thallium which was nebulised contin-uously and mixed with the GC eluent using a Tpiece [176, 177]. Detection limits using a collisioncell single focusing multicollector [176] rangedfrom 2.9 fg (208Pb) to 126 fg (204Pb) and ade-quate isotope ratio precisions were obtained atconcentrations as low as 0.5 ng mL−1 of Et4Pb,as lead.

5 APPLICATION OF THE GC-ICP-MSCOUPLING

5.1 Environmental applications

We consider it adequate to classify the environ-mental applications of GC-ICP-MS with respectto the nature of the analysed samples. We willconsider first gaseous samples and the liquid andsolid samples.

(a) Gaseous samples. The atmosphere can beconsidered the main route of dispersion oforganometallic compounds in the environment.These compounds are present in the atmosphereas both volatile compounds and aerosols and canbe formed by natural processes or generated byhuman activity. It is clearly established nowa-days that we need to know the concentration oforganometallic compounds in the atmosphere tounderstand the biogeochemical cycle of certain ele-ments [106]. In this way, studies on the specia-tion of Sn, Se, As, P, In, Ga, Hg and Pb [106]and Pb [72] in atmospheric samples have beencarried out by GC-ICP-MS. Also, characterisa-tion of the source of different lead organometal-lic compounds in atmospheric particulate sampleshas recently been carried out by measuring leadisotope ratios [10]. In this study the authors com-pared lead isotope ratios in different organoleadspecies measured for leaded gasoline, airborne par-ticulates from both an urban and a rural area and anurban dust reference material (CRM 605). In addi-tion, the samples were digested and the isotopiccomposition of total lead was measured by nebuli-sation using thallium as internal isotope standard.The results obtained are illustrated in Figure 4.2.7.As can be observed, the isotopic composition oforganolead species in the atmosphere of Oviedowas similar to that found in leaded fuel but dif-ferent from those in the reference material. On theother hand, total lead in Oviedo had an isotopic sig-nature in between that of the rural area (industriallead from a nearby aluminium smelter) and leadedfuel from gasoline showing the relative contribu-tion of both sources to lead pollution in Oviedo.

The atmospheric transport or organolead com-pounds from central Europe to the artic was studied

APPLICATION OF THE GC-ICP-MS COUPLING 191

0.47

0.4625

0.455

0.4475

0.44

0.4325

0.425

0.4175

0.41

0.4025

0.3950.8525 0.8685 0.8845 0.9005 0.9165 0.9325 0.9485 0.9645 0.9805 0.9965

Pb207/Pb206

Pb2

07/P

b208

Digested airborneparticulates (Oviedo)

Leaded gasoline

Urban Dust BCR 605

Organolead speciesin airborne particulates(Oviedo)

Digested airborne particulates (San Claudio)

Figure 4.2.7. Lead isotope ratios measured for different atmospheric particulate samples [10]. (Reproduced by permission of TheRoyal Society of Chemistry.)

by Adams et al. [19] by measuring organoleadcompounds in arctic snow and ice. They observedthat meteorological variations and solar light werethe main factors influencing the transport and fateof organolead compounds in the atmosphere.

Volatile species from many different elementshave been detected also in landfill gases [4, 5, 204].In these studies Feldmann et al. have reported thepresence of hydrides and methylated species of As,Se, Sn, Sb, Te, Hg, Pb and Bi. Even so, volatileMo and W hexacarboniles were detected.

(b) Liquid samples. The low concentrationsin which elemental species are found in watersrequire, in many cases, the use of preconcen-tration procedures. Solid–liquid extraction, solid-phase microextraction [62] and purge and cryo-genic trapping [13] are the more used procedures.Using cryogenic trapping, Amoroux et al. [13]observed the presence of tin hydrides in seawa-ter and concluded that chemical and biochemicalmethylation processes taking place in the sed-iments lead to the mobilization of tin speciesto the atmosphere. The same authors [11] stud-ied the presence of volatile Se, Sn, Hg and Pbspecies in estuarine waters. Heisterkamp et al. [80]studied the presence of organolead compounds in

rainwater, tap water and snow samples. Rainwa-ter showed the highest concentrations of thesecompounds, dimethyllead being the more abundant(concentrations in the range of ng l−1). Open-oceanseawater was analysed by Tao et al. [62] in thesearch for organotin compounds. Both butylatedand phenylated tin species were detected at thepg l−1 range in these samples. Recently, RodrıguezGonzalez et al. [187] determined the levels ofbutyltin compounds in coastal seawaters using iso-tope dilution GC-ICP-MS. Levels up to 80 ng kg−1

of TBT (as Sn) were found in marinas while lev-els below 1 ng kg−1 were found on beaches and inopen areas.

(c) Solid samples. Most environmental solidsamples studied in elemental speciation consistof sediments. Sediments act as the final sink fororganometallic compounds and many biologicaltransformations can occur there. The presence ofelemental species in sediments is normally due toanthropogenic origin and so these species are usu-ally adsorbed on the surface of sediment particles.That means that extraction methods used for ele-mental speciation in sediments rarely require totaldigestion of the sample [109]. Extraction meth-ods used involve acidic leaching with the help


of complexing agents or organic solvents such asmethanol. Mechanical shaking, ultrasonic extrac-tion, soxhlet extraction, distillation and microwaveassisted extraction have all been used for elemen-tal speciation in sediments. However, we have totake into account that extraction methods are theweakest link in the traceability chain for elementalspeciation. The use of isotopically enriched ele-mental species has shown to be a viable alternativemethod to study and optimise sample preparationprocedures [197, 206–208]. Tin, mercury and leadare the elements for which most speciation studiesin sediments have been carried out by GC-ICP-MS[30, 51, 97, 169, 209].

GC-ICP-MS studies on soils have been alsoperformed. Prohaska et al. [37] created a closedmicrocosm in which the biomobility of arseniccould be studied under different conditions.

5.2 Biological applications

Biological samples require total decomposition ofthe tissue as the elemental species are incorporatedin the matrix. This decomposition step has to beperformed with extreme care in order to keepthe nature of the species unchanged. Controlleddecomposition of biological tissues for elementalspeciation has been performed in differentways: (i) enzymatic hydrolysis using differentproteases and lipases, (ii) alkaline hydrolysis usingKOH [102] or tetramethylammonium hydroxide(TMAH) [117, 119] or (iii) acid hydrolysis usingHCl in a saline medium [210]. The use ofmicrowave assisted extraction [117, 119] orultrasonic extraction [210] has been proposedrecently to accelerate the hydrolysis processes.

The elements for which most applications havebeen published in biological samples are Hg, Snand Se. The high toxicity of methylmercury andits bioaccumulation in fish explains the interest forthe speciation of this element in biological sam-ples. In the case of tin, the toxicity of butyltincompounds for oysters and mussels is well knownand, in the case of selenium, its biological win-dow between essentiality and toxicity is very nar-row. The increasing use of selenium supplements,

in which selenium can be present as differentselenium species, has been addressed by differentauthors. In this sense, Perez Mendez et al. havedemonstrated that the coupling of a gas chromato-graph to a quadrupole or a double focusing ICP-MS instrument resulted in one of the most sensitiveand selective methods for the measurement ofselenoaminoacids in food supplements. Also, theconcept of chiral speciation was demonstrated bythe separation of D- and L-selenoaminoacids usinga chiral stationary phase [201, 211].

The detection of sulfur compounds in biologicalsamples is an area of activity where many GC-ICP-MS applications can be foreseen. For example,the detection of volatile sulfur compounds infermented saliva by double focusing ICP-MS afterGC separation has been published [17] and GC-ICP-MS methods for the detection of methionineand other sulfur-containing aminoacids have beenpresented [212].

5.3 Isotope dilution analysis withGC-ICP-MS

If we consider that obtaining quality data intrace analysis for total elemental concentrationsis a difficult task, we have to agree that thedifficulties will be much greater when we try todetermine different elemental species in which oneelement is distributed [213]. The complexity ofthe matrices, the different analytical steps whichneed to be performed (digestion, extraction, clean-up, preconcentration, derivatisation, separation anddetermination) and the possibility of speciesinterconversion make elemental speciation adifficult and error-prone task [21]. The applicationof ‘primary’ analytical methods, such as isotopedilution analysis in combination with massspectrometry [214] could be a powerful toolin the search for quality assurance in traceelement speciation [215]. All the well-knownadvantages of isotope dilution analysis [216],namely the correction for incomplete recoveries orlow derivatisation yields, correction for sensitivitydrifts, ideal internal standard and high precisionmeasurements, can be used advantageously in

APPLICATION OF THE GC-ICP-MS COUPLING 193

trace metal speciation. This concept was firstdescribed by Heumann for elemental speciation[213, 217]. In these papers, Heumann indicatedtwo possible ways in which isotope dilutionanalysis could be applied in elemental speciation:(i) using a species-unspecific spike or (ii) usingdifferent species-specific spikes. The first approachis useful when the structure or identity of thespecies is unknown and/or no standard is availablefor quantification. Practically, the spike is addedpost-column after the separation of the speciesand can be used for quantification by convertingthe original chromatogram (counts s−1 versus time)into a mass flow chromatogram (ng s−1 versustime) using the isotope dilution equation. Theintegration of the chromatographic peaks willprovide directly the amount of trace element in thiselemental species [218, 219]. This first approachhas been used only after HPLC or CE separationsand does not offer all the advantages of isotopedilution analysis listed previously (only correctsfor sensitivity drifts).

The full advantages of isotope dilution analysisfor elemental speciation are realized only withthe second approach: the use of one or moreisotopically enriched species of the element ofinterest. This implies that we need to knowthe chemical structure of the species of interestand that we can obtain or synthesize thosespecies isotopically enriched. For this approach theenriched spike(s) is(are) added at the beginningof the analytical procedure and that makes fulluse of all isotope dilution advantages. Thereare several examples of publications on isotopedilution analysis in combination with GC-ICP-MS(GC-ICP-IDMS) but these have focused only onthree elements: Hg [197, 199, 220, 221], Se [173]and Sn [175, 187, 188, 207, 208].

Hintelmann et al. [199] studied the formationof methylmercury during the extraction of inor-ganic mercury from different environmental sam-ples. They used both Hg(II) nitrate, enriched in199Hg, and Hg(II) chloride, enriched in 200Hg, tocompute methylation reactions, and methylmer-cury chloride, enriched in 201Hg, to calculate therecovery of the extraction technique employed.

The results obtained indicated important overes-timations of MeHg+ using distillation as a sam-ple preparation technique. The conclusions of thisstudy were that methylation of mercury dependedon the type of sample, on the relative concentra-tions of inorganic and methylmercury and on theextraction technique employed.

In a different study, Snell et al. [220] used198Hg-enriched (96 at%) Hg(II) and MeHg+ todevelop analytical methodologies with quantitativerecoveries when computation was performed usingthe isotope dilution equation. The use of standardaddition calibration provided results of less qual-ity. Recently, Demuth and Heumann [221] usedthe double spike technique to study methylationand demethylation reactions for mercury duringderivatisation. Their results indicated that certainextraction conditions favoured the formation ofHg(0) and decomposition of the original species.

Gallus and Heumann [173] developed backin 1996 a methodology for the determination ofSe(IV) and Se(VI) species by GC-ICP-IDMS.Selenite was derivatised to piazselenol (volatile)and this compound determined by GC-ICP-MS.Then, selenate was reduced to selenite andderivatised in the same way. The concentrationof selenate was calculated by difference fromthat of selenite. They used 82Se-enriched (62 at%)species, added to the sample prior to derivatisa-tion or reduction, and measured the ratios 77/82and 78/82 in the sample. The determinations usingboth isotope ratios led to comparable results. Thedetection limit of the method was estimated to be0.02 ng ml−1 for both selenium species.

Ruiz Encinar and coworkers [175, 188, 207,208] evaluated the possibilities of isotope dilu-tion analysis for tin speciation in sediments usingdifferent isotopically enriched spikes. In the firststudy, [175] the synthesis of enriched dibutyltin(DBT), using 118Sn (98.44 at%), was describedand applied to the determination of DBT in cer-tified reference materials with satisfactory results.Isotope equilibration was achieved by mechani-cal shaking of the sample and spike during 12 husing 4 ml of a 75/25 mixture of acetic acidand methanol. No interconversion of the different


butyltin species in the sediment was detected usingthis extraction procedure.

In a second study [188] the synthesis of a119Sn-enriched (82.4 at%) mixture of MBT, DBTand TBT was described. This mixed spike wascharacterized by reverse isotope dilution analysisand applied to the simultaneous determination ofthe three butyltin species in certified sediments. Inthis case, isotope equilibration and extraction wereperformed as for the 118Sn-enriched DBT and nointerconversion reactions were detected during thereverse isotope dilution experiments. Isotope ratioprecision on triplicate injections was, on average,0.8 %, depending on peak size, and the precisionof the whole procedure for independent triplicateanalyses was between 1 and 1.5 %. This indicatedthat satisfactory results could be obtained evenfor single injection, single extraction experimentswhich could give a considerable reduction in timein comparison with other calibration strategies.

The third study, published recently by thesame research group [207, 208], compared threeextraction techniques: mechanical shaking, ultra-sonic extraction and microwave assisted extrac-tion for the speciation of butyltin compounds in

sediments. The basic idea was to be able tocompare the extraction efficiency of the three tech-niques for MBT, DBT and TBT and the possi-bility of interconversion reactions during extrac-tion. For this purpose, a mixed spike was preparedusing the enriched species synthesized previously[175, 188]. The final spike mixture used containedMBT and TBT, both enriched in 119Sn, and DBTenriched in 118Sn. The isotopic composition of thisspike can be clearly seen in Figure 4.2.8. As canbe observed, the abundance of the 120Sn isotope,the main isotope in natural tin, is very low inall species.

By measuring the ratios 120/118 and 120/119 inspiked certified reference materials and assumingthat the decomposition of butyltin compoundswould follow a simple debutylation mechanisms,the authors could calculate the concentration ofall three butyltin species in the sample and thecorresponding decomposition factors.

The results showed that MBT was stronglybound to the sediment matrix, as previouslysuspected [222], and that mechanical shaking wasnot able to extract it quantitatively. On the otherhand, all three extraction techniques tested could

220 240 260 280 300 320 340 360 380 400

Time/s

2.0E5

1.0E5

Cou

nts

MBT

DBT

TBT

120118

119

Figure 4.2.8. GC-ICP-MS chromatogram of a 118Sn and 119Sn double isotope spike for butyltin speciation [207]. (Reproducedby permission of The Royal Society of Chemistry.)

CONCLUSIONS 195

extract DBT and TBT easily. One interestingconclusion of this study was that microwaveassisted extraction caused the decomposition ofbutyltin compounds with transformation factorsup to 16 % depending on the butyltin speciesconsidered, the microwave power applied and theextraction time used.

The use of isotope dilution analysis in elemen-tal speciation will continue to increase in the nearfuture. The ability to compensate for nonquantita-tive recoveries will be of tremendous interest instudying extraction techniques where the recover-ies are strongly influenced by the sample matrix,such as solid-phase microextraction. The analysisof butyltin compounds in natural and seawater,another area where high preconcentration factorsare required, would benefit from the use of isotopedilution methodologies [187]. Also, the applicationof double spike techniques, or even triple spiketechniques will help in establishing sound extrac-tion procedures for elemental speciation and inthe development of more reliable certified refer-ence materials. Finally, the application of isotopedilution methodologies for the GC-ICP-MS deter-mination of other elements and species, such asSe- and S-containing aminoacids, will be a realityin the near future.

5.4 Reference materialsand quality control

Elemental speciation, until now restricted more orless to research laboratories, is coming slowly intoenvironmental regulations in different countries.For example, tributyltin is regulated in seawater(10 ng l−1 as TBT) and drinking water (63 ng l−1

as TBT) in the USA. This trend is also fol-lowed by the EU where TBT has to be mea-sured in continental waters (directive 76/464/EEC)[223]. Other elemental species, such as Cr(VI),MeHg+, BrO−

3 and ‘inorganic As’, are alreadyincluded in various ‘black lists’ and have to bemeasured in different environmental and biologi-cal samples (e.g. inorganic As in fish products).This means that sound speciation procedures willhave to be implemented in routine or contract

laboratories and that new reference materials forelemental speciation will be needed in the nearfuture.

We all know that sample preparation proceduresare still the ‘Achilles heel’ of modern speciationmethodologies [224, 225]. If we need to developnew reference materials or reference methodolo-gies to be used in contract laboratories we willhave to pay much more attention to these samplepreparation steps. It has been observed in previouscertification campaigns [226] that certification ofcertain elemental species was impaired by the widerange of interlaboratory results or by the lack ofstability of the material itself. Isotope dilution anal-ysis with species-specific spikes will surely play apredominant role in this context.

There are several areas where isotope dilutionanalysis will help in the preparation of ‘speciated’reference materials. First, homogeneity studieswould be better performed using isotope dilution.The homogeneity of a given material cannot becertified to a value lower than the uncertainty inthe measurement procedure used. Hence, if wecould improve our measurement uncertainty by5–10 times by using isotope dilution methodolo-gies the better the homogeneity studies wouldbe. The same can be said for stability studieswhere enriched species could be used to under-stand why a certain elemental species is degrad-ing and what the degradation product is. Thiswill help in devising alternative stabilization pro-cedures which would not alter the speciation inthe sample. Finally, the certification campaignwould benefit from the use of speciated isotopedilution analysis as has been demonstrated previ-ously for the preparation of trace element referencematerials.

6 CONCLUSIONS

Trace metal speciation methodologies are very welldeveloped in research laboratories. The coupling ofa high efficient separation technique (HPLC, GC orCE) to a sensitive and selective elemental detectionmethod (MIP-AES or ICP-MS) seems to be idealfor elemental speciation. However, we have to take


into account that new elemental species are dis-covered continuously and many ‘unknown’ speciesare still found in the literature (see for exampleFigure 4.2.6). To identify these species we needto resort to detection techniques which are capa-ble of producing ‘molecular’ information, such aselectron impact MS, electrospray MS or new tun-able GD-MS, MIP-MS or even ICP-MS, which arecomplementary to the elemental detection method.

Having said that, the speciation of elementswhich form volatile compounds or which canbe derivatised to form volatile compounds isperfectly adapted for GC separation coupled toelemental detectors. Techniques such as MIP-AESor ICP-MS are the most sensitive and selectivewhich can be used in detectors for GC. It isclear that the speciation of Hg, Sn and Pb inenvironmental and biological samples is betterperformed using GC-ICP-MS. This is clear whenwe compare the analytical characteristics of thiscoupling with those of any other combinationpublished in the literature. Recently, GC-ICP-MShas started to be applied for the speciation of Seand S with excellent results. We have to expectmore developments in the near future for thespeciation of these elements.

The main characteristic of GC-ICP-MS, inwhich it clearly differs from other atomic detec-tion techniques such as MIP-AES, is its ability toprovide isotopic information. This extra informa-tion can be used in two ways: to investigate thesources of elemental species, as in the studies per-formed on lead and sulfur, or to use isotope dilu-tion methodologies to improve the quality of theanalytical information obtained. We should expectmore developments in this field as the combinationof GC with multicollector ICP-MS instruments hasonly started to be evaluated.

7 REFERENCES

1. Templeton, D. M., Ariese, F., Cornelis, R., Danielson,L. G., Muntau, H. van Leeuwen Lobinski, R. H. P.,Pure Appl. Chem., 72, 1453 (2000).

2. Quevauviller, P., Astruc, M., Morabito, R., Ariese, F.and Ebdon, L., Trends Anal. Chem., 19, 180 (2000).

3. Feldmann, J. and Hirner, A. V., Int. J. Environ. Anal.Chem., 60, 339 (1995).

4. Feldmann, J., Koch, I. and Cullen, W. R., Analyst., 123,815 (1998).

5. Feldmann, J. and Cullen, W. R., Environ. Sci. Technol.,31, 2125 (1997).

6. Feldmann, J., J. Anal. At. Spectrom., 12, 1069 (1997).7. Radziuk, B., Thomassen, Y., van Loon, J. C. and Chau,

Y. K., Anal. Chim. Acta, 105, 255 (1979).8. Kim, A. W., Foulkes, M. E., Ebdon, L., Hill, S.,

Patience, R. L., Barwise, A. G. and Rowland, S. J.,J. Anal. At. Spectrom., 7, 1147 (1992).

9. Montes Bayon, M., Gutierrez Camblor, M., GarcıaAlonso, J. I. and Sanz-Medel, A., J. Anal. At. Spectrom.,14, 1317 (1999).

10. Ruiz Encinar, J., Leal Granadillo, I. A., Garcıa Alonso,J. I. and Sanz-Medel, A., J. Anal. At. Spectrom., 16, 475(2001).

11. Amoroux, D., Tessier, E., Pecheyran, C. and Donard,O. F. X., Anal. Chim. Acta, 377, 241 (1998).

12. Amoroux, D. and Donard, O. F. X., Geophys. Res. Lett.,23, 1777 (1996).

13. Amoroux, D., Tessier, E. and Donard, O. F. X., Envi-ron. Sci. Technol., 34, 988 (2000).

14. Hewitt, C. N. and Harrison, R. M., Anal. Chim. Acta,167, 277 (1985).

15. Bloom, N. and Fitzgerald, W. F., Anal. Chim. Acta,208, 151 (1988).

16. Ballantine, D. S. and Zoller, W. H., Anal. Chem., 56,1288 (1984).

17. Rodriguez Fernandez, J., Montes Bayon, M., Pereiro, R.and Sanz-Medel, A., J. Anal. At. Spectrom., 16, 1051(2001).

18. Robinson, J. W., Kiesel, E. L., Goodbread, J. P.,Bliss, R. and Marshall, R., Anal. Chim. Acta, 92, 321(1977).

19. Adams, F. C., Heisterkamp, M., Candelone, J. P.,Laturnus, F., Van de Velde, K. and Boutron, C. F.,Analyst , 123, 7671 (1998).

20. Quevauviller, P., Donard, O. F. X., Maier, E. A. andGriepink, B., Mikrochim. Acta, 109, 169 (1992).

21. Morabito, R., Massanisso, P. and Quevauviller, P.,Trends Anal. Chem., 19, 113 (2000).

22. Kingston, H. M., Huo, D., Lu, Y. and Chalk, S., Spec-trochim. Acta , 53B, 299 (1998).

23. Campbell, A. D., Pure Appl. Chem., 64, 227 (1992).24. Dirkx, W. M. R., Lobinski, R. and Adams, F. C., in

Quality Assurance for Environmental Analysis , Que-vauviller, Ph., Maier, E. A. and Griepink, B. (Eds),Elsevier, Amsterdam, 1995, Chapter 15.

25. Donard, O. F. X. and Pinel, R., in Environmental Anal-ysis using Chromatography Interfaced with Atomic Spec-troscopy , Harrison, R. M. and Rapsomanikis, S. (Eds),Ellis Horwood, Chichester, 1989, Chapter 7.

26. Weber, J. H., Trends Anal. Chem., 16, 73 (1997).27. Le, X. C., Cullen, W. R. and Reimer, J. K., Anal. Chim.

Acta , 285, 277 (1994).28. Jin, K., Shibata, Y. and Morita, M., Anal. Chem., 63,

986 (1991).

REFERENCES 197

29. Donard, O. F. X., Rapsomanikis, S. and Weber, J. H.,Anal. Chem., 58, 772 (1986).

30. Segovia Garcıa, E., Garcıa Alonso, J. I. and Sanz-Medel, A., J. Mass Spectrom., 32, 542 (1997).

31. Clark, S. and Craig, P. J., Appl. Organomet. Chem., 2,33 (1988).

32. Sullivan, J. J., Torkelson, J. D., Wekell, M. M.,Hollingworth, T. A., Saxton, W. L., Miller, G. A.,Panaro, K. W. and Uhler, A. D., Anal. Chem., 60, 626(1986).

33. Martin, F. M. and Donard, O. F. X., J. Anal. At. Spec-trom., 9, 1143 (1994).

34. Martin, F. M., Tseng, C.-M., Belin, C., Quevauviller, P.and Donard, O. F. X., Anal. Chim. Acta, 286, 343(1994).

35. Pannier, F., Astruc, A., Astruc, M. and Morabito, R.,Appl. Organomet. Chem., 10, 471 (1996).

36. Dodd, M., Pergantis, S. A., Cullen, W. R., Li, H.,Eigendorf, G. K. and Reimer, K. J., Analyst , 121, 223(1996).

37. Prohaska, T., Pfeffer, M., Tulipan, M., Stingeder, G.,Mentler, A. and Wenzel, W. W., Fresenius’ J. Anal.Chem., 364, 467 (1999).

38. Ritsema, R. and Donard, O. F. X., Appl. Organomet.Chem., 8, 571 (1994).

39. Ishizaka, T., Nemoto, S., Sasaki, K., Suzuki, T. andSaito, Y., J. Agric. Food. Chem., 37, 1523 (1989).

40. Fent, K. and Muller, M. D., Environ. Sci. Technol., 25,489 (1991).

41. Honeycutt, J. B. J. and Riddle, J. M., J. Am. Chem.Soc., 83, 369 (1961).

42. de la Calle-Guntinas, M. B., Scerbo, R., Chiavarini, S.,Quevauviller, P. and Morabito, R., Appl. Organomet.Chem., 11, 693 (1997).

43. Maguire, R. J. and Huneault, H., J. Chromatogr., 209,458 (1981).

44. Harrison, R. M. and Radojevic, M., Environ. Technol.Lett., 6, 129 (1985).

45. Gomez-Ariza, J. L., Morales, E., Giraldez, I.,Beltran, R. and Escobar, J. A. P., Fresenius’ J. Anal.Chem., 357, 1007 (1997).

46. De Diego, A., Tseng, C. M., Stoichev, T., Amoroux, D.and Donard, O. F. X., J. Anal. At. Spectrom., 13, 623(1988).

47. Bloom, N. S., Can. J. Fish. Aquat. Sci., 46, 1131 (1989).48. Rapsomanikis, S., Donard, O. F. X. and Weber, J. H.,

Anal. Chem., 58, 35 (1986).49. Ashby, J. R. and Craig, P. J., Appl. Organomet. Chem.,

5, 393 (1991).50. Cai, Y., Rapsomanikis, S. and Andreae, M. O., J. Anal.

At. Spectrom., 8, 119 (1993).51. Ritsema, R., de Smaele, T., Moens, L., de Jong, A. S.

and Donard, O. F. X., Environ. Pollut., 99, 271 (1998).52. Rapsomanikis, S. and Craig, P. J., Anal. Chim. Acta,

248, 563 (1991).53. Jimenez, M. S. and Sturgeon, R. E., J. Anal. At. Spec-

trom., 12, 597 (1997).

54. Sturgeon, R. E., Wiilie, S. N. and Berman, S. S., Anal.Chem., 61, 1867 (1989).

55. Rapsomanikis, S., Analyst , 119, 1429 (1994).56. Chau, Y. K., Yang, F. and Brown, M., Anal. Chim.

Acta , 338, 51 (1997).57. Ashby, J., Clark, S. and Craig, P. J., J. Anal. At. Spec-

trom., 3, 735 (1988).58. Kuballa, J., Wilken, R.-D., Jantzen, E., Kwan, K. K.

and Chau, Y. K., Analyst , 120, 667 (1995).59. Rodriguez Pereiro, I., Schmitt, V. O., Szpunar, J.,

Donard, O. F. X. and Lobinski, R., Anal. Chem., 68,4135 (1996).

60. Cai, Y., Monsalud, S., Jaffe, R. and Jones, R. D.,J. Chromatogr. A, 876, 147 (2000).

61. Montigny, C., Lespes, G. and Potin-Gautier, M., J.Chromatogr. A, 819, 221 (1998).

62. Tao, H., Rajendran, R. B., Quetel, C. R., Nakazato, T.,Tominaga, M. and Miyazaki, A., Anal. Chem., 71, 4208(1999).

63. Szpunar, J., Schmitt, V. O., Lobinski, R. and Monod,J. L., J. Anal. At. Spectrom., 11, 193 (1996).

64. Lespes, G., Desauziers, V., Montigny, C. and Potin-Gautier, M., J. Chromatogr. A, 826, 67 (1998).

65. Lobinski, R., Boutron, C. F., Candelone, J. P., Hong, S.,Szpunar, J. and Adams, F. C., Anal. Chem., 65, 2510(1993).

66. Lobinski, R., Szpunar, J., Adams, F. C., Tesseidre, P. L.and Cabanis, J. C., J. Assoc. Off. Anal. Chem., 76, 1262(1993).

67. De Smaele, T., Moens, L., Dams, R., Sandra, P., Vander Eycken, J. and Vandyck, J., J. Chromatogr. A, 793,99 (1998).

68. Garcıa Fernandez, R., Montes Bayon, M., GarcıaAlonso, J. I. and Sanz-Medel, A., J. Mass Spectrom.,35, 639 (2000).

69. Heisterkamp, M. and Adams, F. C., J. Anal. At.Spectrom., 14, 1307 (1999).

70. Pons, B., Carrera, A. and Nerın, C., J. Chromatogr. B ,716, 139 (1998).

71. Bulska, E., Emteborg, H., Baxter, D. C., Frech, W.,Ellingsen, D. and Thomassen, Y., Analyst , 117, 657(1992).

72. Leal Granadillo, I. A., Garcıa Alonso, J. I. and Sanz-Medel, A., Anal. Chim. Acta, 423, 21 (2000).

73. Radojevic, M., Allen, A., Rapsomanikis, S. and Harri-son, R. M., Anal. Chem., 58, 658 (1986).

74. Chakraborti, D., De Jonghe, W. R. A., Van Mol, W. E.,Van Cleuvenbergen, R. J. A. and Adams, F. C., Anal.Chem., 56, 2692 (1984).

75. Gui-Bin, J. and Qun-Fang, Z., J. Chromatogr. A, 886,197 (2000).

76. Stab, J. A., Cofino, W. P., van Hattum, B. andBrinkman, U. A. T., Fresenius’ J. Anal. Chem., 347,247 (1993).

77. Liu, Y., Lopez-Avila, V., Alcaraz, M. and Beckert,W. F., J. High Resolut. Chromatogr., 17, 527 (1994).


78. Stab, J. A., Brinkman, U. A. T. and Cofino, W. P., Appl.Organomet. Chem., 8, 577 (1994).

79. Meinema, H. A., Burger-Wiersma, T., Versluis-deHaan, G. and Ch. Gevers, E., Environ. Sci. Technol., 12,288 (1978).

80. Heisterkamp, M., De Smaele, T., Candelone, J. P.,Moens, L., Dams, R. and Adams, F. C., J. Anal. At.Spectrom., 12, 1077 (1997).

81. Nagase, M., Toba, M., Kondo, H. and Hasebe, K., Ana-lyst , 123, 1091 (1998).

82. Cai, Y., Alzaga, R. and Bayona, J. M., Anal. Chem., 66,1161 (1994).

83. Quevauviller, P., Fortunati, G. U., Filippelli, M., Bor-toli, A. and Muntau, H., Appl. Organomet. Chem., 12,531 (1998).

84. Quevauviller, P., Ebdon, L., Harrison, R. M. andWang, Y., Appl. Organomet. Chem., 13, 1 (1999).

85. Wenclawiak, B. W. and Krah, M., Fresenius’ J. Anal.Chem., 351, 134 (1995).

86. Minganti, V., Capelli, R. and De Pellegrini, R., Frese-nius’ J. Anal. Chem., 351, 471 (1995).

87. Clausen, J. and Nielsen, S. A., Biol. Trace Elem. Res.,15, 125 (1988).

88. Lamkin, W. M. and Gehrke, C. W., Anal. Chem., 37,383 (1965).

89. Husek, P. and Sweeley, C., J. High Resolut. Chro-matogr., 14, 751 (1991).

90. Vazquez Pelaez, M., Montes Bayon, M., Garcıa Alonso,J. I. and Sanz-Medel, A., J. Anal. At. Spectrom., 15,1217 (2000).

91. Pawliszyn, J., Trends Anal. Chem., 14, 113 (1995).92. Alpendurada, M. F., J. Chromatogr. A, 889, 3 (2000).93. Arthur, C. L. and Pawliszyn, J., Anal. Chem., 62, 2145

(1990).94. Cai, Y. and Bayona, J. M., J. Chromatogr. A, 696, 113

(1995).95. Tutschku, S., Mothes, S. and Wennrich, R., Fresenius’

J. Anal. Chem., 354, 587 (1996).96. Gorecki, T. and Pawliszyn, J., Anal. Chem., 68, 3008

(1996).97. Moens, L., De Smaele, T., Dams, R., Van Den

Broeck, P. and Sandra, P., Anal. Chem., 69, 1604(1997).

98. Lespes, G., Desauziers, V., Montigny, C. and Potin-Gautier, M., J. Chromatogr. A, 826, 67 (1998).

99. Jiang, G. B., Liu, J. Y. and Yang, K. W., Anal. Chim.Acta , 421, 67 (2000).

100. Aguerre, S., Bancon-Montigny, C., Lespes, G. andPotin-Gautier, M., Analyst , 125, 263 (2000).

101. Millan, E. and Pawliszyn, J., J. Chromatogr. A., 873, 63(2000).

102. Vercauteren, J., De Meester, A., De Smaele, T., Van-haecke, F., Moens, L., Dams, R. and Sandra, P., J.Anal. At. Spectrom., 15, 651 (2000).

103. Lord, H. and Pawliszyn, J., J. Chromatogr. A, 885, 153(2000).

104. Baltussen, E., Sandra, P., David, F. and Cramers, C., J.Microcolumn Sep., 11, 737 (1999).

105. Vercauteren, J., Peres, C., Devos, C., Sandra, P., Van-haecke, F. and Moens, L., Anal. Chem., 73, 1509(2001).

106. Pecheyran, C., Quetel, C. R., Lecuyer, F. M. M. andDonard, O. F. X., Anal. Chem., 70, 2639 (1998).

107. Seligman, P. F., Valkirs, A. O. and Lee, R. F., Environ.Sci. Technol., 20, 1229 (1986).

108. Schebek, L. and Andreae, M. O., Environ. Sci. Technol.,25, 871 (1991).

109. Lobinski, R. and Adams, F. C., Spectrochim Acta PartB , 52, 1865 (1997).

110. Eiden, R., Scholer, H. F. and Gastner, M., J. Chro-matogr. A, 809, 151 (1998).

111. Munari, F., Colombo, P. A., Magni, P., Zilioli, G., Tres-tianu, S. and Grob, K., J. Microcolumn Sep., 7, 403(1995).

112. Ceulemans, M., Lobinski, R., Dirkx, W. M. R. andAdams, F. C., Fresenius’ J. Anal. Chem., 347, 256(1993).

113. Cooke, W. S., Today’s Chem. Work , 5, 16 (1996).114. Lobinski, R., Rodriguez Pereiro, I., Chassaigne, H.,

Wasik, A. and Szpunar, J., J. Anal. At. Spectrom., 13,859 (1998).

115. Rodriguez Pereiro, I., Schmitt, V. O. and Lobinski, R.,Anal. Chem., 69, 4799 (1997).

116. Wasik, A., Rodriguez Pereiro, I. and Lobinski, R., Spec-trochim. Acta Part B , 53, 867 (1998).

117. Slaets, S., Adams, F., Rodriguez Pereiro, I. and Lobin-ski, R., J. Anal. At. Spectrom., 14, 851 (1999).

118. Lobinski, R., Sidelnikov, V., Patrushev, Y., RodriguezPereiro I. and Wasik, A., Trends Anal. Chem., 18, 449(1999).

119. Rodriguez Pereiro, I., Monicou, S., Lobinski, R., Sidel-nikov, V., Patrushev, Y. and Yamanaka, M., Anal.Chem., 71, 4534 (1999).

120. Wasik, A., Rodriguez Pereiro, I., Dietz, C., Spuznar, J.and Lobinski, R., Anal. Commun., 35, 331 (1998).

121. Forsyth, D. S. and Marshall, W. D., Anal. Chem., 55,2132 (1983).

122. Soderquist, C. J. and Crosby, D. G., Anal. Chem., 50,1435 (1978).

123. Zarnegar, P. and Mushak, P., Anal. Chim. Acta, 69, 389(1974).

124. Uchida, H., Shimoishi, Y. and Toci, K., Environ. Sci.Technol., 14, 541 (1980).

125. Andreae, M. O., Anal. Chem., 49, 820 (1977).126. Ebdon, L., Hill, S. and Ward, R. W., Analyst , 111, 1113

(1986).127. Ebdon, L., An Introduction to Atomic Absorption Spec-

troscopy , Heyden, London, 1982.128. Kolb, B., Kemmer, G., Schlesser, F. H. and Wiedek-

ing, E., Fresenius’ J. Anal. Chem., 221, 166 (1966).129. Chau, Y. K., Wong, P. T. S. and Goulden, P. D., Anal.

Chim. Acta , 85, 421 (1976).130. Wolf, W. R., Anal. Chem., 48, 1717 (1976).

REFERENCES 199

131. Hahn, M. H., Mulligan, K. J., Jackson, M. E. andCaruso, J. A., Anal. Chim. Acta, 118, 115 (1980).

132. Ebdon, L., Ward, R. W. and Laethard, D. A., Analyst ,107, 129 (1982).

133. Foster, R. C. and Howard, A. G., Anal. Proc., 26, 34(1989).

134. Forsyth, D. S. and Cleroux, C., Talanta , 38, 951 (1991).135. Segar, D. A., Anal. Lett., 7, 89 (1974).136. Parris, G. E., Blair, W. R. and Brinckman, F. E., Anal.

Chem., 49, 378 (1977).137. Apte, S. C. and Gardner, M. J., Talanta , 35, 539 (1988).138. Chau, Y. K., Wong, P. T. S. and Goulden, P. D., Anal.

Chem., 47, 2279 (1975).139. Chau, Y. K., Wong, P. T. S., Bengert, G. A. and Dunn,

J. L., Anal. Chem., 56, 271 (1984).140. Forsyth, D. S. and Marshall, W. D., Anal. Chem., 57,

1299 (1985).141. Muller, M. D., Anal. Chem., 59, 617 (1987).142. Dachs, J., Alzaga, R., Bayona, J. M. and Quevauviller,

P., Anal. Chim. Acta, 286, 319 (1994).143. Gui-Bin, J., Qun-Fang, Z. and Bin, H., Bull. Environ.

Contam. Toxicol., 65, 277 (2000).144. McCormack, A. J., Tong, S. C. and Cooke, X. D., Anal.

Chem., 37, 1470 (1965).145. Rosenkranz, B. and Bettmer, J., Trends Anal. Chem.,

362, 489 (1998).146. Kato, T., Uehiro, T., Yasuhara, A. and Morita, M., J.

Anal. At. Spectrom., 7, 15 (1992).147. Witte, C., Szpunar, J., Lobinski, R. and Adams, F. C.,

Appl. Organomet. Chem., 8, 621 (1994).148. Ceulemans, M., Witte, C., Lobinski, R. and Adams,

F. C., Appl. Organomet. Chem., 8, 451 (1994).149. Mena, M. L., McLeod, C. W., Jones, P., Withers, A.,

Mimganti, V., Capelli, R. and Quevauviller, P., Frese-nius’ J. Anal. Chem., 355, 456 (1995).

150. de la Calle Guntinas, M. B., Ceulemans, M., Witte, C.,Lobinski, R. and Adams, F. C., Mikrochim. Acta, 120,73 (1995).

151. Estes, S. A., Uden, P. C. and Barnes, R. M., Anal.Chem., 53, 1829 (1981).

152. Creed, J. T., Mohamad, A. H., Davidson, T M., Ata-man, G. and Caruso, J. A., J. Anal. At. Spectrom., 3, 923(1988).

153. Windsor, D. L. and Denton, M. B., Appl. Spectrosc., 32,366 (1978).

154. Windsor, D. L. and Denton, M. B., J. Chromatogr. Sci.,17, 492 (1979).

155. Windsor, D. L. and Denton, M. B., Anal. Chem., 51,1116 (1979).

156. Brown, R. M., Northway, S. J. and Fry, R. C., Anal.Chem., 53, 934 (1981).

157. Brown, R. M. and Fry, R. C., Anal. Chem., 53, 532(1981).

158. Van Loon, J. C., Alcock, L. R., Pinchin, W. H. andFrench, J. B., Spectrosc. Lett., 19, 1125 (1986).

159. Chong, N. S. and Houk, R. S., Appl. Spectrosc., 41, 66(1987).

160. Peters, G. R. and Beauchemin, D., J. Anal. At. Spec-trom., 7, 965 (1992).

161. Peters, G. R. and Beauchemin, D., Anal. Chem., 65, 97(1993).

162. Murillo, M. and Mermet, J. M., Spectrochim. Acta ,42B, 1151 (1987).

163. Kim, A. W., Hill, S., Ebdon, L. and Rowland, S. J., J.High Resolut. Chromatogr., 15, 665 (1992).

164. Pretorius, W. G., Ebdon, L. and Rowland, S. J., J.Chromatogr., 646, 369 (1993).

165. Ebdon, L., Evans, H., Pretorius, W. G. and Rowland,S. J., J. Anal. At. Spectrom., 9, 939 (1994).

166. Story, W. C., Olson, L. K., Shen, W.-L., Creed, J. T.and Caruso, J. A., J. Anal. At. Spectrom., 5, 467 (1990).

167. Evans, E. H. and Caruso, J. A., J. Anal. At. Spectrom.,8, 427 (1993).

168. Evans, E. H., Pretorious, W., Ebdon, L. and Row-land, S., Anal. Chem., 66, 3400 (1994).

169. Prange, A. and Jantzen, E., J. Anal. At. Spectrom., 10,105 (1995).

170. De Smaele, T., Verrept, P., Moens, L. and Dams, R.,Spectrochim. Acta , 50B, 1409 (1995).

171. De Smaele, T., Moens, L., Dams, R. and Sandra, P.,Fresenius’ J. Anal. Chem., 355, 778 (1996).

172. De Smaele, T., Moens, L., Dams, R. and Sandra, P.,LC-GC Int., 14, 876 (1996).

173. Gallus, S. M. and Heumann, K. G., J. Anal. At. Spec-trom., 11, 887 (1996).

174. Pritzl, G., Stuer-Lauridsen, F., Carlsen, L., Jensen, A. K.and Thorsen, T. K., Int. J. Environ. Anal. Chem., 62, 147(1996).

175. Ruiz Encinar, J., Garcıa Alonso, J. I. and Sanz-Medel, A., J. Anal. At. Spectrom., 15, 1233 (2000).

176. Krupp, E. M., Pecheyran, C., Meffan-Main, S. andDonard, O. F. X., Fresenius’ J. Anal. Chem., 370, 573(2001).

177. Krupp, E. M., Pecheyran, C., Pinaly, H., Motelica-Heino, M., Koller, D., Young, S. M. M., Brenner, I. B.and Donard, O. F. X., Spectrochim. Acta , 56B, 1233(2001).

178. Vela, N. P., Olson, L. K. and Caruso, J. A., Anal.Chem., 65, 585A (1988).

179. Garcıa Alonso, J. I., Sanz-Medel, A. and Ebdon, L.,Anal. Chim. Acta, 283, 261 (1993).

180. Stab, J. A., Cofino, W. P., Van Hattum, B. and Th.Brinkman, U. A., Fresenius’ J. Anal. Chem., 347, 247(1993).

181. Ashby, J. R. and Craig, P. J., Sci. Total Environ., 78,219 (1989).

182. Arakawa, Y., Wada, O., Yu, T. H. and Iwai, H., J.Chromatogr., 216, 209 (1981).

183. Duebelbeis, D. O., Kapila, S., Yates, D. E. and Mana-han, S. E., J. Chromatogr., 351, 465 (1986).

184. Tolosa, I., Bayona, J. M., Albaiges, J., Alencrasto, L. F.and Tarradellas, J., Fresenius’ J. Anal. Chem., 339, 646(1991).


185. Jiang, G. B., Ceulemans, M. and Adams, F. C., J. Chro-matogr. A, 727, 119 (1996).

186. Sato, K., Khori, M. and Okochi, H., Bunseki Kagaku,45, 575 (1996).

187. Rodrıguez Gonzalez, P., Ruiz Encinar, J., GarcıaAlonso, J. I. and Sanz-Medel, A., J. Anal. At. Spectrom.,17, 824 (2002).

188. Ruiz Encinar, J., Monterde Villar, M. I., Gotor San-tamarıa, V., Garcıa Alonso, J. I. and Sanz-Medel, A.,Anal. Chem., 73, 3174 (2001).

189. Horlick, G. and Montaser, A., Analytical characteristicsof ICP-MS, in Inductively Coupled Plasma Mass Spec-trometry , Montaser, A. (Ed.), Willey-VCH, New York,1998, pp. 503–614.

190. Quevauviller, P., Astruc, M., Ebdon, L. and Griepink,B., Appl. Organomet. Chem., 8, 639 (1994).

191. Abalos, M., Bayona, J. M., Campano, R., Granados, M.,Leal, C. and Prat, M. D., J. Chromatogr. A, 788, 1(1997).

192. Garcıa Alonso, J. I., Ruiz Encinar, J., Leal Granadillo,I. A., Garcıa Fernandez, R., Montes Bayon, M. andSanz-Medel, A., Trace metal speciation in environmen-tal analysis using inductively coupled plasma mass spec-trometry, in Advances in Mass Spectrometry , Vol. 15,Gelpı, E. (Ed.), John Wiley and Sons, Ltd, Chichester,2001, pp. 265–282.

193. Vanhaecke, F., Moens, L. and Taylor, P., Use of ICP-MS for isotope ratio measurements, in InductivelyCoupled Plasma Spectrometry and its Applications,Hill, S. J. (Ed.), Sheffield Academic Press, Sheffield,1999, pp. 145–207.

194. Nelms, S. M., Quetel, C. R., Prohaska, T., Vogl, J. andTaylor, P. D. P., J. Anal. At. Spectrom., 16, 333 (2001).

195. Ruiz Encinar, J., Garcıa Alonso, J. I., Sanz-Medel, A.,Main, S. and Turner, P. J., J. Anal. At. Spectrom., 16,315 (2001).

196. Heumann, K. G., Gallus, S. M., Radlinger, G. andVogl, J., J. Anal. At. Spectrom., 13, 1001 (1998).

197. Hintelmann, H., Evans, R. D. and Villenueve, J. Y., J.Anal. At. Spectrom., 10, 619 (1995).

198. Rosman, K. J. R. and Taylor, P. D. P., J. Anal. At.Spectrom., 14, 5N (1999).

199. Hintelmann, H., Falter, R., Ilgen, G. and Evans, R. D.,Fresenius’ J. Anal. Chem., 358, 363 (1997).

200. Klemens, P. and Heumann, K. G., Determination oftotal silicon and silicon species by isotope dilution ICP-HRMS, Paper presented at the Second InternationalConference on Double Focusing Sector Field ICP-MS,Vienna, September 12–15, 2001.

201. Perez Mendez, S., Montes Bayon, M., Blanco Gonzalez,E. and Sanz-Medel, A., J. Anal. At. Spectrom., 14, 1333(1999).

202. Costa Fernandez, J. M., Bings, N. H., Leach, A. M. andHieftje, G. M., J. Anal. At. Spectrom., 15, 1063 (2000).

203. Bings, N. H., Costa Fernandez, J. M., Guzowski, J. P.Jr., Leach, A. M. and Hieftje, G. M., Spectrochim. Acta ,55B, 767 (2000).

204. Haas, K., Feldmann, J., Wennrich, R. and Stark, H-J.,Fresenius J. Anal. Chem., 370, 587 (2001).

205. Donard, O. F. X., Metal speciation and trace metalcycling in the environment. Paper presented at theFourth Euroconference on Environmental AnalyticalChemistry, Visegrad, Hungary, September 14–19, 2000.

206. Huo, D., Lu, Y. and Kingston, H. M., Environ. Sci.Technol., 32, 3418 (1998).

207. Ruiz Encinar, J., Rodrıguez Gonzalez, P., GarcıaAlonso, J. I. and Sanz-Medel, A., Anal. Chem., 74, 270(2002).

208. Garcıa Alonso, J. I., Ruiz Encinar, J., RodrıguezGonzalez, P. and Sanz-Medel, A., Anal. Bioanal. Chem.,373, 432 (2002).

209. Rajendran, R. B., Tao, H., Nakazato, T. andMiyazaki, A., J. Anal. At. Spectrom., 15, 1757 (2000).

210. Yu, Q., Quian, J. and Frech, W., J. Anal. At. Spectrom.,15, 1586 (2000).

211. Perez Mendez, S., Blanco Gonzalez, E. and Sanz-Medel, A., J. Anal. At. Spectrom., 15, 1109 (2000).

212. Montes Bayon, M., Garcıa Alonso, J. I., VazquezPelaez, M., Rodriguez de la Flor, R. and SanzMedel, A., Coupling GC to double focusing ICP-MS toperform speciation of sulphur and selenium containingaminoacids. Paper presented at the First InternationalConference on High resolution ICP-MS, Norfolk, Vir-ginia, May 25–28, 2000.

213. Heumann, K. G., in Metal Speciation in the Environ-ment , Broekaert, J. A. C., Gucer, S. and Adams, F.(Eds), NATO ASI Series , Vol. G 23, Springer, Berlın,1990, pp. 153–168.

214. De Bievre, P., Anal. Proc., 30, 328 (1993).215. Sanz-Medel, A., Spectrochim. Acta , 53B, 197 (1998).216. Fasset, J. D. and Paulsen, P. J., Anal. Chem., 61, 643A

(1989).217. Heumann, K. G., Int. J. Mass Spectrom. Ion Processes ,

118/119, 575 (1992).218. Sariego Muniz, C., Marchante Gayon, J. M., Garcıa

Alonso, J. I. and Sanz-Medel, A., J. Anal. At. Spectrom.,16, 587 (2001).

219. Rottmann, L. and Heumann, K. G., Anal. Chem., 66,3709 (1994).

220. Snell, J. P., Stewart, I. I., Sturgeon, R. E. andFrech, W., J. Anal. At. Spectrom., 15, 1540 (2000).

221. Demuth, N. and Heumann, K. G., Anal. Chem., 73,4020 (2001).

222. Ceulemans, M. and Adams, F. C., Anal. Chim. Acta,317, 161 (1995).

223. Quevauviller, P., Spectrochim. Acta , 53B, 1261 (1998).224. Quevauviller, P. and Morabito, R., Trends Anal. Chem.,

19, 86 (2000).225. Adams, F. C. and Slaets, S., Trends Anal. Chem., 19, 80

(2000).226. Quevauviller, P., Astruc, M., Morabito, R., Ariese, F.

and Ebdon, L., Trends Anal. Chem., 19, 180 (2000).

4.3 Capillary Electrophoresis in Speciation Analysis

Bernhard MichalkeInstitute for Ecological Chemistry, Neuherberg, Germany

Abbreviations . . . . . . . . . . . . . . . . . . . . . . 2011 Speciation – Introduction . . . . . . . . . . . . . 202

1.1 Importance of element speciationanalysis . . . . . . . . . . . . . . . . . . . . . . 202

2 Position of Capillary Electrophoresis inSpeciation . . . . . . . . . . . . . . . . . . . . . . . . 202

3 Principles of Metal Speciation by CapillaryElectrophoresis . . . . . . . . . . . . . . . . . . . . . 2043.1 Injection . . . . . . . . . . . . . . . . . . . . . 2043.2 Separation . . . . . . . . . . . . . . . . . . . . 2043.3 Different separation modes . . . . . . . . 2053.4 Problems in speciation, problems and

risk for species preservation,suitability of (complexing) buffers,metal release in IEF . . . . . . . . . . . . . 207

3.5 Separation of varying species types 2083.5.1 Element species in different

oxidation states . . . . . . . . . . . 2083.5.2 Analysis of organometallic

compounds . . . . . . . . . . . . . . 2083.5.3 Analysis of elements bound to

organic compounds such asproteins . . . . . . . . . . . . . . . . . 208

4 Detection Modes and Their Advantages andProblems . . . . . . . . . . . . . . . . . . . . . . . . . 2084.1 UV detection and indirect UV

detection (iUV) . . . . . . . . . . . . . . . . 2084.2.1 A few examples of speciation

using CE-UV/iUV . . . . . . . . . 209

4.2.2 A few examples of speciationusing CE-UV and withETV-ICP-MS for qualitycontrol . . . . . . . . . . . . . . . . . 210

4.3 Inductively coupled plasma massspectrometry detection . . . . . . . . . . . 212

4.4 The interfacing to ICP-MS . . . . . . . . 2134.4.1 Requirements of the interface 2134.4.2 Technical solutions . . . . . . . . 2144.4.3 Potential of CE-ICP-MS . . . . 2154.4.4 Limitations of CE-ICP-MS 2164.4.5 A few examples of speciation

using CE-ICP-MS . . . . . . . . . 2174.5 ESI-MS detection . . . . . . . . . . . . . . . 2184.6 Problems of ESI-MS in speciation 2194.7 CE-ESI-MS . . . . . . . . . . . . . . . . . . . 220

4.7.1 Requirements of the ESIinterface and solutions . . . . . . 220

4.7.2 Potential of CE-ESI-MS . . . . . 2204.7.3 Limitations of CE-ESI-MS . . . 2204.7.4 A few examples of speciation

using CE-ESI-MS . . . . . . . . . 2205 Combination of CE-ESI-MS and

CE-ICP-MS: Maximized SpeciesInformation . . . . . . . . . . . . . . . . . . . . . . . 221

6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . 2227 References . . . . . . . . . . . . . . . . . . . . . . . . 222

ABBREVIATIONS

CE capillary electrophoresisCZE capillary zone electrophoresisDIN direct injection nebulizer

EOF endoosmotic flowESI-MS electrospray ionization mass

spectrometer


ICP-MS inductively coupled plasmamass spectrometer

IEF isoelectric focusingITP isotachophoresisLIF laser induced fluorescenceLOD limit of determinationMCN microconcentric nebulizerMECC micellar electrokinetic

chromatography

1 SPECIATION – INTRODUCTION

1.1 Importance of element speciationanalysis

The quality and quantity of the various speciesof an element in a matrix are highly responsiblefor the mobility, bioavailability and finally theecotoxicological or toxicological impact of thatelement rather than its total concentration [1–5].Therefore, only a knowledge of the species enablesan assessment of whether it is toxic, without(known) impact at a specific concentration, orwhether it is essential. This stresses the necessityof speciation analysis to determine the speciationof an element in a specific matrix.

Definitions related to speciation have beengiven earlier in this book. However, it must benoted that the use of capillary electrophoresis (CE)in speciation analysis is closely linked to theanalytical activity of identifying and measuringspecies, including identification of the bindingpartners of elements as well as providing methodsfor quality control [6].

2 POSITION OF CAPILLARYELECTROPHORESIS IN SPECIATION

As yet there is no such instrument as an easy‘speciation analyzer’. Combination and hyphen-ation of separation technologies and element- ormolecule-selective detection systems are gener-ally the basis for speciation analyses. Separa-tion and detection methods already established inother fields have to be combined partly in new

ways and modified according to the particularspeciation problems. Therefore, element speciesmust be separated before being analyzed, eitherby nonselective detectors (e.g. UV) or by element-or molecule-selective detector such as ICP-MSor ESI-MS. One of the most powerful separa-tion techniques is CE. It provides a most effi-cient separation of species and is often superiorto liquid chromatography separation techniques. Inaddition, a single CE instrument even allows sev-eral different modes in separation: CZE, MECC,IEF, ITP and CEC. These separation mecha-nisms are commonly based on the application ofa high voltage. However, their separation prin-ciples are quite different, providing completelydifferent mechanisms of characterization and iden-tification for element species. The latter is ofparticular significance, as species identification israrely done by one single method but needs multi-dimensional strategies. In addition, this variabilityis most advantageous in finding separation solu-tions for nearly every specific separation prob-lem. While CE supplements conventional HPLCmethods, it shows unique promise for speciationpurposes because it causes only a minor distur-bance to the existing equilibrium between differentspecies. There is no stationary phase, which hasa huge surface area and gives various possibili-ties for undesired interactions. Therefore, speciesintegrity is thought to be less easily affected thanwith HPLC.

The use of CE in speciation is complex: it maybe used as a primary separation mechanism or asa secondary separation technique after e.g. HPLC,in a second dimension for identification. Often it iscombined with nonselective (direct or indirect) UVdetection. In whatever dimension it is employed,parallel runs in different separation modes on onesample are performed for a wide characteriza-tion. As an improvement fractionation may be per-formed at the outlet of the capillary to analyzefurther the separated element species by ICP-MSfor the respective metals. This approach mustbe recognized as a preliminary stage for finallyhyphenated techniques linked to selective detec-tors for either elements or whole molecules. Thevarious combinations of CE techniques in different

POSITION OF CAPILLARY ELECTROPHORESIS IN SPECIATION 203

sample

ICP-MS

ESI

CE

HPLC column

fractioncollector

UVdetector

UVdetector

fractioncollector

MS/MS

InterfaceCE-ICP-MS

plasma +torch

Figure 4.3.1. The position of capillary electrophoresis in element speciation analysis: CE can be used as a primary separationsystem with UV detection and fractionation for subsequent element determination. A more elegant way is its use in hyphenatedsystems, with either ICP-MS or ESI-MS. Then isotopic or molecular information can be obtained. CE can also be used as asecondary separation device, e.g. after HPLC separation, for quality control using the detection capabilities shown.

stages of multidimensional strategies and with dif-ferent detectors, which supply information aboutisotopes and elements (ICP-MS) or moleculesand structural compounds (ESI-MS/MS), makeCE immensely valuable in speciation analysis.Figure 4.3.1 schematically shows the position ofCE in speciation analysis.

1. In Figure 4.3.1 it is shown that CE may be usedas the primary separation technique. The sampleis introduced to a CE instrument and species areseparated by one of the different separation modes.Detection may be done by direct UV or indirectUV if analytes are UV transparent. However,conductivity detectors or LIF are also employed.

2. CE may be also used for quality controland species identification as a second dimension

separation device. Here, the first separation andcharacterization of species are performed byanother method, such as HPLC, combined withfraction collection and subsequent element detec-tion. In order to improve the reliability of speci-ation results, it is necessary to subject the HPLCfractions to other separation techniques based ondifferent properties of the molecules for clear iden-tification. Element concentrations are determinedfor specific HPLC peak fractions, which are furtherinvestigated by (different) CE methods. A nearlyoverall characterization is possible in this way, pro-vided that no species alteration has occurred dur-ing the primary (HPLC) separation. It is assumedthat the identified species and the elements deter-mined in the fractions are indeed associated withone another. However, this assumption does not


necessarily hold in all cases, even if they aredetected in the same fraction. To overcome thisuncertainty CE may be run additionally with frac-tion collection at the end of the CE capillary.The fractions are again introduced into an elementdetector. Elements are shown to be attached to aspecific species, both after HPLC and after CEseparation. The species identification, as well asthe knowledge about species–element associationhere, is based on orthogonal strategies, providinghigh testifying power.

3. A more elegant choice is to use a hyphen-ated system of CE with an element-selectivedetector such as ICP-MS. The high resolv-ing power of CE and its capability of differ-ent separation technologies are here combinedwith element and/or isotope information. Com-binations with ICP-OES have also been occa-sionally described, but cannot play an importantrole in speciation at physiologically or environ-mentally normal concentrations. For these appli-cations the detection sensitivity of ICP-OESis insufficient. Specific problems in interfac-ing the two systems arise from the interfaceitself.

4. Finally, even with CE-ICP-MS not all prob-lems in speciation are solved. Generally, com-binations of separation and element detectionsystems identify species by means of comparingretention/migration times of analytes and standardcompounds. When standards are missing iden-tification is rarely possible. Even a characteri-zation is difficult when employing hyphenatedsystems, as no fractions for subsequent inves-tigations are available. Therefore, CE-ICP-MSshould be complemented by a parallel set-up of CE-ESI-MS. The electrospray ionizationsource is capable of handling low flow ratestypical of CE and provides a smooth ion-ization, where the species is not altered (inmany cases). Therefore, with this hyphenationthe high resolving power of CE and its capacityfor different separation technologies is combinedwith molecular (MS) and/or structural informa-tion (MS/MS).

3 PRINCIPLES OF METALSPECIATION BY CAPILLARYELECTROPHORESIS

3.1 Injection

The first method of sample introduction is to usea positive pressure at the inlet or, vice versa, anegative pressure (suction) at the outlet. Given ahomogenous sample composition all componentsof the sample are injected at their respectiveconcentration without any preference. Typicalsampling volumes range between 5 and 50 nL.Hydrodynamical injection modes are based on asimilar principle: either the sample at inlet is sethigher than the outlet or the latter is positionedbelow the inlet. In both modes gravity is forcingthe sample into the capillary during the time theheight difference is maintained.

Another injection method is to apply a highvoltage to the sample for a short injection time,typically around 5–15 s. This injection modediscriminates between charged and nonchargedanalytes or anions and cations. Conversely, thereis the possibility of introducing selectively onlyanalytes of a desired charge (positive or negative)by the application of an injection voltage with theappropriate polarity. This is called electrokineticinjection. A high charge density of the analytehelps for a preferred injection by the systemand finally leads to a preconcentration duringinjection. However, it has been found that theinjection reproducibility of slow moving speciesis much worse. Thus, electrokinetic injectionis not considered to be a reliable method inelemental speciation [7]. It also lacks on a widedynamic range.

3.2 Separation

CE uses the separation principle of differencesin the electrically driven mobility of chargedanalytes, similar to conventional electrophoresis.Here, an electric field is applied along an open-tube column with a low inner diameter at highvoltage, typically between 20 and 30 kV. This

PRINCIPLES OF METAL SPECIATION BY CAPILLARY ELECTROPHORESIS 205

is the reason for its comparatively short analysistime and very high separation efficiency. Usually,200 000–700 000 theoretical plates are achieved.The lack of a stationary phase having a largesurface is considered as a further advantage ofCE in speciation [8]. The molecules move withdifferent velocities in the electric field. The bordersof the analyte bands do not show laminar profiles,but for the CE show typically sharp vertically ones.This again improves resolution and also the signalto noise ratio and thus affects detection sensitivity.The latter is high for total analyte masses but lowwhen looking at analyte concentrations. The reasonis the very small total sample volumes, being in therange of 5–50 nL.

In contrast to conventional flat bed elec-trophoresis this technique is easily adapted forautomation and quantitative analysis. The result-ing electropherograms can be similarly processedlike LC chromatograms. Usually a complete CEanalysis is very fast and completed earlier than anLC separation of the same analytical problem.

3.3 Different separation modes

Development of the method and changing ofanalytical parameters are readily done by justchanging the electrolyte system when replacinga few milliliters of one buffer system with otherbuffers. Thus, several different separation princi-ples are available, distinguishing and separating thespecies according to their different physicochemi-cal properties. Table 4.3.1 gives an overview of theseparation modes, separation principle and targetanalytes (according to ref. 9).

Detection may be performed on the column byUV and LIF, or at the end of the capillary by

conductivity or (preferably) element-selective andmolecule-selective detectors. Combinations usedfor element speciation are discussed below. Ascheme of a conventional CE system is given inFigure 4.3.2.

At pH values higher than 3 a laminar flow isbuilt up, the more the pH takes on basic values.This flow is called endoosmotic flow (EOF). Itis induced by the negatively charged inner cap-illary surface where the silanolic groups of thebare fused silica capillary attract positively chargedions. An ‘electrical double layer’ is set up wherehydrated cations are moved to the cathode by theelectric field and thus produce the EOF. Oftenthe EOF is faster than the current-driven move-ment of anions in the opposite direction. A veryfast EOF can result in an alteration in the sep-aration of species. On the other hand, the EOFhelps for a prolonged separation time and thusincreased resolution when the electrically drivenmovement and the EOF are opposed. Usually, theEOF is directed towards the cathode as long asthe capillary walls remain negatively charged. Thismeans that only cationic metal species are mov-ing by electrophoresis in the same direction as theEOF. However, without special measures to adjustthe differences in electrophoretic mobilities, thesimple co-electroosmotic mode offers efficient sep-arations for only a limited number of cations. Incontrast, many anionic metal complexes promisea good resolution due to inherent differences inelectrophoretic mobilities. However, anionic elec-trophoretic mobilities and the EOF have differentdirections. Such a counter-electroosmotic migra-tion substantially reduces the range of anionic ana-lytes that can be separated for detection at thecathode end.

Table 4.3.1. Overview of the separation modes, separation principle and target analytes [9].

Separation mode Abbreviation Principle Target analytes

Capillary zone electrophoresis CZE Charge densitycharge/mass ratio

Charged molecules, amino acids,proteins

Capillary isoelectricfocusing

cIEF Isoelectric point Proteins, peptides

Capillary isotachophoresis cITP Analyte specific conductivity Differently dissociated moleculesMicellar electrokinetic capillary

chromatographyMECC Hydrophobicity Neutral molecules with different ability to

enter charged hydrophobic micelles


+ −

electrode

capillary detection window

filter

high voltage

UV-source

inlet buffer vialcontaining electrolyte

PC for instrument control and data acquisition

detector, typically UV

outlet buffer vialcontaining electrolyte

electrode

20 µA

Figure 4.3.2. Principle scheme of capillary electrophoresis.

The EOF may be varied by applying coatingsto the capillary wall: it can be slowed down,totally avoided or may be even reversed. Thesecoatings may be of a dynamic nature by purgingthe capillary with suitable additives before eachanalysis, or bound permanently to the wall bycovalent bondings. Uncoated silanolic groups arelikely to adsorb basic molecules, preferentiallybasic proteins. Therefore, in order to prevent theprecipitation of proteins and to suppress the EOF,capillaries with a polymer-coated inner surface arerecommended for such separations.

The principle of CZE separation assumes thatthe difference in ionic mobilities has a decisiveeffect on the resolution. The ionic mobilities ofanalytes are related, in turn, to their charge den-sities, i.e. charge-to-size ratios. In MECC, whichoperates under the same conditions as CZE butwith micellar electrolytes, the principle mechanismof separation depends on whether the analytes arecharged or not. For charged analytes, both elec-trophoretic migration in the aqueous electrolyteand solubilization into micelles play a role. Theseparation of electrically neutral compounds, onthe other hand, is dominated by the distribution

between the aqueous electrophoretic and micellarphases only. Accordingly, the analyte’s hydropho-bicity governs the distribution ratio and determinesits migration behavior. The separation of smallmetal ions is an area where CE is being usedto an increasing extent [10]. The distinctive fea-ture of metal cations, metal-complexed ions andmetal oxoanions is their high charge-to-size ratioand hence high electrophoretic mobility. As astriking electrophoretic property, it should lead torapid separations with high efficiency. However formetal cations, many of which are of nearly iden-tical charge and hydrated ionic radius, the differ-ences in mobility are not sufficient to provide goodseparations. Further technologies are given by iso-tachophoresis. This method is mostly achieved bythe use of discontinuous buffer systems, distin-guished by differences in conductivity. When thesample is positioned between a ‘terminating buffer’at the inlet, having a very low conductivity and aleading electrolyte with a higher conductivity thanthe sample compounds, the analytes will be posi-tioned according to increasing conductivity as soonas the high voltage is turned on. The major advan-tage of isotachophoresis in speciation analysis is

PRINCIPLES OF METAL SPECIATION BY CAPILLARY ELECTROPHORESIS 207

its capability to be combined on line with CZE.This results in a prefocusing of the sample beforeit is separated by CZE. Therefore, the sample vol-ume can be increased partly up to 300 nL (in 1.5 mcapillaries) without loss in resolution and thus con-centration detection limits may be improved.

3.4 Problems in speciation, problems andrisk for species preservation, suitability of(complexing) buffers, metal release in IEF

One limitation is the very small sample volume tobe analyzed, typically only a few nanoliters. Thiscauses problems in ensuring a representative sam-ple and sets high demands on homogeneity of thesample. Further, the concentration detection limitsare generally about two orders of magnitude worsethan those for LC separations. The detection capa-bility of the UV detectors generally used principallyfollows the Lambert–Beer law, where the length ofthe light path through the sample governs the sen-sitivity. In CE this length is the inner diameter ofthe capillary, which is typically 25–75 µm. How-ever, since capillary electrophoresis is hyphenatedto powerful detectors such as ICP-MS, LODs of0.05–1 µg L−1 have been reported [11–13]. Thehigh voltage itself can alter the integrity of elemen-tal species. Similarly, the buffer/electrolyte compo-sition and nature of additives can have detrimentaleffects on species stability [14].

The mobility of analytes is dependent on theactual strength of the electric field and the (pH-dependent) EOF. Unfortunately, the EOF may beinfluenced by the sample itself, which may havea high buffering capacity, thus locally changingthe pH and EOF. Furthermore, the electric fieldis changing along the capillary due to differencesin analyte conductivity, again being partly afunction of sample composition. Therefore, themigration time of a specific compound maybe shifted from standard to sample and fromsample to sample. This is widely reported inthe literature and known as ‘migration timeshifts’. The reproducibility within one sample isgenerally high, provided the buffer reservoirs arenot depleted, but the intersample reproducibilityis sometimes worse. Species identification by

comparison with standard electropherograms isthus most questionable. The addition of an internalstandard helps to correct species migration timeaccording to standard electropherograms. Anotherpossibility is the standard addition procedure ofthe compounds investigated. In the case thatcomigration of other analytes is excluded thisclearly identifies the species.

The application of the high voltage itself mayalter the structure of metal species. Loosely boundmetals may be removed. Also, recomplexationis possible by several buffer components investi-gated, e.g. for borate and phosphate buffers. Thishas to be specifically considered when the netcharge of the species is changed and subsequentlyit is moving to the inlet. It will no longer bedetected anymore.

When using hyphenated techniques an addi-tional suction flow may be forced on the ‘open-tube’ capillary. Suction is likely to be induced bya necessary sheath flow in the nebulizer interfaceor by the nebulization gas itself at the end of thecapillary.

When cIEF is used an EOF must be strictlyavoided [15]. Therefore, only coated capillaries aresuitable for cIEF. This technique is considered toshow excellent advantages in separating peptidesand proteins (but is restricted to such molecules)concerning resolution and concentration detectionlimits. The reason is that the whole capillary isfilled with the sample. The total sample volumetherefore is around 500–1500 nL (depending oncapillary length) compared to CZE methods withca. 5–20 nL. The sample must be mixed with‘ampholytes’ (approximately 2 % in the sample),which determine the pH gradient inside the cap-illary and thus the final position of metal–proteinspecies in the capillary. Unfortunately, one mustconsider that the ampholytes may alter speciesand lead to recomplexation. This is especiallystressed as peptides and proteins show no netcharges at their respective isoelectric points andsome loosely bound metals may then be removedby the high voltage.

When applying MECC, additives for micellationare needed in the electrolyte. Metal contaminationsmay be occurring at the polar surface of micelles,


which mimics element species. Hydrophobicorganic–metal complexes with loosely bound met-als may be altered.

3.5 Separation of varying species types

3.5.1 Element species in differentoxidation states

Free metal ions usually show similar elec-trophoretic mobilities and insufficient stability inelectrolyte solutions. In addition, detectability isdifficult using conventional detectors, due to thelack of detectable properties for metal ions. Com-plexation presents a valuable approach for per-forming speciation of metals with different oxi-dation states [16]. Complete conversion of met-als into charged complexes, which takes placeupon addition of a complexing reagent to a sam-ple before introduction to the capillary, rendersthem different in charge density while alteration ofthe initial concentration ratio of different oxidationstates is prevented. For speciation involving metaloxidation forms of opposite charges, precapillarycomplexation is a straightforward strategy to getthe same charge and electrophoretic direction forthe species. The exchange kinetics must be slowenough (compared with the time required for theseparation) to obtain the individual peaks for eachspecies [14]. To preserve the stability or chemicalintegrity of complexed metal species the numberof electrolyte constituents should be kept to a min-imum. Just for improved detection simple carrierelectrolytes containing only protonated imidazolefor indirect UV detection should be used.

3.5.2 Analysis of organometallic compounds

Since different species of one metal with alkylor aryl substituents often have similar mobili-ties, the separation power of CE usually needsenhancement. Enhanced separation efficiency canbe gained by selecting a suitable electrolyteadditive, such as a weak complexing reagentor β-cyclodextrin, which is capable of formingcomplexes with organometallic compounds. Oth-erwise, organometallic–ligand complexes formed

before introduction into the capillary can be usedeffectively.

3.5.3 Analysis of elements bound to organiccompounds such as proteins

There are many metal-containing species of bio-logical significance that have been subjected to CEspeciation. In accordance with their charges, thesespecies can be separated while moving towardsthe cathode. For preventing precipitation of pro-teins and to suppress the EOF, capillaries witha polymer-coated inner surface are recommendedfor such separations. CE can also be used for theanalysis of the interaction between metals and pro-teins as assessed by mobility-shift assays. Impor-tant structural information about metalloproteinssuch as transferrin, concerning heterogeneity of theattached carbohydrate chains as well as the degreeof metal saturation, can be obtained by both CZEand cIEF techniques.

4 DETECTION MODES AND THEIRADVANTAGES AND PROBLEMS

4.1 UV detection and indirect UVdetection (iUV)

One major area of research in applying CE formetal speciation is the development of sensitivedetectors. The small diameter of separation capil-laries leads to high efficiencies in CE separationsbut is also responsible for the major limitation indetection sensitivity. Although the mass sensitivityis very high, the concentration sensitivity is gen-erally one or two orders of magnitude lower thanthat for HPLC. Since real-world samples containmetal species at the µg L−1 level or lower, sensi-tive detectors must be coupled to the capillary ina CE system.

Conventional detectors are predominantly UVdetectors. They are mostly considered as non-specific detectors, because in only a fewcases do elemental species show typical UVspectra. Species-selective electropherograms mustbe obtained by using a UV scanning detectionmode. However, UV detection is nonspecific for

DETECTION MODES: ADVANTAGES AND PROBLEMS 209

most types of speciation analysis. This generallycan lead to a ‘false’ detection of element species.Overlapping of peaks, resulting possibly in incor-rect quantification, may not be apparent. The detec-tion sensitivity is mostly too poor as many speciesdo not show sufficient UV response. Therefore,their detection generally requires derivatizationusing a suitable chromophore prior to separation.According to experiences of the Standard Measure-ment and Testing program of the EU ‘derivatiza-tion steps are often far from being controlled’. Therespective steps are ‘rarely fully understood’ [17].

In contrast, indirect UV absorbance detectionis presently the most versatile method to solve theproblem of universal detection for metal speciationby CE. The key to this approach is the displace-ment of a highly absorbing electrolyte co-ion bythe sample ions. When choosing a UV-active co-ion, a close mobility match to the analyte ionsis required; otherwise, asymmetrical peak shapesare generated. Various cationic and anionic co-ions such as imidazole, pyridine and chromate havebeen successfully utilized for indirect detection ofmetal species in CE.

Unfortunately, even with indirect UV detection,the sensitivity is rarely adequate for monitoringnaturally concentrated species in the real world.The addition of chromophores to the buffer mayagain affect species integrity in some cases.Another possibility is to link ‘invisible’ elementspecies to complexing UV-active complexes. Thishas been shown to provide detection sensitivity,but with the risk of species alteration. Originalspecies information is easily lost.

4.2.1 A few examples of speciation usingCE-UV/iUV

Several attempts have described to speciate sele-nium compounds usually by direct UV detec-tion. Unfortunately, the speciation methods mostlyhad to be limited to separate standards orstandard-added real samples, as detection sensi-tivity was inappropriate. Albert et al. [18] sepa-rated four standard selenium compounds within20 min at concentrations considerably higher thanthose found in e.g. biological samples. However,the high separation potential of the techniqueis demonstrated clearly (Figure 4.3.3). A similar

2

1

3

4

0 5 10 15 20 min

Figure 4.3.3. Electropherogram of Se compounds with the following CE conditions: fused silica capillary 52 cm × 50 µm ID;electrolyte, 80 mM phosphate buffer (pH 8.5); 2 mM TTAB; hydrodynamic injection 5 s; separation at −12 kV; direct UV detectionat 200 nm. Peak 1 = Se(VI), peak 2 = Se(IV), peak 3 = SeC, peak 4 = SeM. Reprinted from Analusis, Albert, M. M. et al.,Vol. 21, pp. 403–407, 1993, copyright notice of Springer-Verlag [18].


1

2

3 4

5

6

0.04

0.03

0.02

0.01

0

−0.01

4 6 8 10 12 14

minutes

AU

Figure 4.3.4. Electropherogram of a human milk sample for the following CE conditions: fused silica capillary50 cm × 50 µm ID; electrolyte, 10 mM Na2CO3; pH 11.6; pressure injection 25 psi s; separation at −18 kV; direct UV detec-tion at 200 nm. Peak 1 = Se-cystamine, peak 2 = Se(VI), peak 3 = SeC, peak 4 = Se(IV), peak 5 = SeM, peak 6 = Se-carryingglutathione (reprinted from [19]). Peak positions were determined by standard addition. Se(VI) and Se(IV) seem to comigratewith other compounds, as they were proven to be absent in the native sample.

approach was taken by Michalke [19], which isshown in Figure 4.3.4. Here human milk was ana-lyzed using an alkaline electrolyte and direct UVdetection. The arrows mark the positions where Sestandards are observed when added to the sam-ple. Here too, the impressive separation capabilityof CZE is shown. However, the nonspecific UVdetector shows various signals that are not due toSe compounds. It is also demonstrated that someof the Se species are not resolved from major com-pounds (e.g. peak 2) or are very close to detectionlimit and thus difficult to quantify. It should benoted that subsequent analysis of this sample byhyphenation with ICP-MS excluded the presenceof Se in compounds 2 and 4. Obviously, unknownbut UV-active non-Se compounds were migrat-ing at positions where added Se standards alsoappeared. The pitfalls of nonselective detectionare clearly demonstrated here. Further speciation

experiments have been performed, e.g. for As spe-ciation, organolead species and organotins. Thelatter were determined by indirect UV detectionat concentrations around 5 mg L−1.

4.2.2 A few examples of speciation usingCE-UV and with ETV-ICP-MS for qualitycontrol

The use of CE as a second separation sys-tem (two-dimensional) with conventional detectiontechnique in multidimensional concepts with otherseparation techniques can be an easy means ofquality control. It helps to increase the certainty ofidentification of element species. However, even inthis case comigration is not excluded completely.Furthermore, the attribution of element concentra-tions in consecutive factions (e.g. from HPLC) tospecies, determined in the same fractions by CE, is


still not always guaranteed. Complex schemes arenecessary to overcome the remaining problems. Asuitable method is to make a first identificationby retention times in the first separation dimen-sion (HPLC). Collected fractions are subsequentlysubjected to CE for further analysis to provide atwo-dimensional identification. It has been proven[20] that in rare cases of very complex matri-ces comigration still takes place. Thus, in parallelanother separation method by CE – either based oncompletely different electrolyte systems or usinganother separation mode, such as cIEF – was intro-duced. Coincidence of identification in each of thethree separations finally provides a high certaintyof identification. Figure 4.3.5 shows a flow chartof the two experiments performed in parallel.

To obtain additional information about thecorrect attribution of elements to identified speciesa fraction collection may be set at the outlet of

the capillary. The CE fractions then are analyzede.g. by ETV-ICP-MS, for the elements of interest.The element concentrations are compared to thecorresponding concentrations in HPLC fractionsof the ‘first dimension separation’. Electrothermalvaporization is suggested for sample introductioninto ICP-MS, because ETV is capable of handlingvery small sample volumes (few µL). This isnecessary to avoid an unnecessary increase indilution of analytes. In any case, the dilutionof off-migration analytes from CE into the vialis immense!

For quantitative working (which is recom-mended for quality control) it is necessaryto know the injection volume of CE exactly.Michalke and Schramel [21] introduced a suit-able but time-consuming method: a standard com-pound of known concentration was injected intothe capillary applying different injection times in

fractionation

CE method 1

MT1. standard compounds

MT2. analysing fractions

3. analysing fractions + standard addition

identification

CE method 2

MT1. standard compounds

MT2. analysing fractions

3. analysing fractions + standard addition

identification

sample

HPLC

UV-detection

standard

HPLC

UV-detectioncomparison of RTidentification (?)

comparison of concentrations from identified compounds in consecutive fractions (CE method 1 and 2); comparison with element concentrations in consecutive fractions

element detection

Figure 4.3.5. Flow chart of consecutive investigations when using CE for quality control (in analogy to [20]).


consecutive experiments. The capillary was thenpurged and the effluent collected into vials con-taining 100 µL of buffer. The capillary volume was0.7 µL and thus made a negligible contribution tothe final volume. The element was measured in thecollection vial and the exact injection volume cal-culated according to the standard concentration, themeasured concentration in vial and the vial volume(100 µL). Finally, a calibration curve was calcu-lated from consecutive experiments with varyinginjection times. When using this setup platinumcomplexation by methionine as well as aging anddegradation of the Pt–methionine complex wasdemonstrated. The whole procedure for qualitycontrol, species identification and element attribu-tion was performed for Se species in the sameexperiment [21]: after separating Se compoundsvia SEC, two different CZE methods were appliedfor identification. Finally, fractions were collectedat the end of the capillary and subjected to ETV-ICP-MS. Thus, Se was attributed to specific Sespecies after HPLC separation and in addition aftera subsequent CZE separation. For quality controlreasons fractions of each collected peak as wellas of the inlet and outlet vials were monitoredfor Se. Only the identified Se species showed ameasurable Se concentration. Figure 4.3.6 demon-strates the electropherogram of one CZE separa-tion, which was made after an SEC separation. Theidentified Se-carrying glutathione gives the onlyfraction with Se.

A more elegant way is to use ICP-MS directlyfor on-line detection in capillary electrophoresis.The following section gives some informationabout the potential and about some problemsarising from this detection system when used forCE detection in elemental speciation.

4.3 Inductively coupled plasma massspectrometry detection

The big advantages of this method are its multiele-ment capability and high sensitivity [22]. Isotopicand elemental information is gained. The ioniza-tion source is an inductively coupled Ar plasma.

The sample introduction is performed by aninterface, connecting the CE system with the

0 0 0 0 0

97

0

0.009

0.008

0.007

0.006

0.005

0.004

0.003

0.002

0.001

0

15 16 17 18 19 20

minutesA

U

peak fractionation and Se-determinationby ETV-ICP-MS

SEC fraction containing GSH and Se. Total Se = 0.14 ng (100%).grey: sample, black: sample + GSH-addition (identification)

reco

very

[%]

inle

t

peak

1

Pea

k 2

peak

3

peak

4

peak

5

outle

t

110

90

70

50

30

10

−10

1

2

3

4

5

Figure 4.3.6. Analysis of an SEC fraction from human milk,containing Se-carrying glutathione: identification by standardaddition and quantification of Se in CE fractions. Se-carryingglutathione was the only Se species in this SEC fraction.

ICP-MS instrument. Even in hyphenated systemsdetection limits for element concentrations inelement species have been reported to be in


the 10–100 ng L−1 range [11, 13, 23]. Thereare quadrupole systems available (qICP-MS) andhighly resolving sector field ICP-MS (sf-ICP-MS).

qICP-MS is considerably cheaper, but detectionlimits generally are worse than those for sf-ICP-MS. The quadrupole mass filter provides a resolu-tion of 300. Therefore, ‘polyatomic interferences’are likely, especially in the mass range 40–80. Aprominent interference is that one on 75As. The75As isotope is a mono-isotope, which is easilyinterfered with the 40Ar35Cl+ cluster [24]. The lat-ter is produced in the Ar plasma when chlorine isintroduced by buffers or as a sample component.The highly resolving sf-ICP-MS can distinguishbetween the interference and the element isotope,as the mass resolution is 7500–10 000 (instead of300 for qICP-MS). However, when using a highmass resolution the detection sensitivity is reduced.When employing (only) element-selective detec-tors such as an ICP mass spectrometer one hasto realize that only the element in the species isdetected, not the whole molecule. This gives theadvantage that the separation of molecules withdifferent elements does not need to be complete.The detector can distinguish them. However, as themolecule itself is not seen, identification is onlypossible by comparing retention times. In naturalsamples this cannot always be done with certainty.The hyphenation of CE to ESI-MS can help inthis matter. For recognition of polyatomic inter-ferences the monitoring of several isotopes of oneelement can be helpful. Only when the natural iso-tope ratio is determined in a peak, are interferencesunlikely. Unsatisfactory sensitivity is still a prob-lem in samples of the lowest concentration. It isrecommended to monitor the most abundant iso-topes of an element, except when these isotopesare major targets of interferences; as an example80Se is mentioned, being totally overlapped bythe 40Ar40Ar cluster. Even the second most abun-dant isotope 78Se suffers strong interference andthus the signal to noise is worse than for 77Se,although the abundance of the latter is a factor ofthree lower.

On the other hand, an ICP mass spectrometer isa sequential detector, monitoring the programmedisotopes for several ms. When programming too

much isotopes for determination in parallel thedetector gets too slow for highly resolved andfast appearing peaks on one specific isotope. Thiscauses a loss in chromatographic resolution of thehyphenated system.

4.4 The interfacing to ICP-MS

Much effort has been devoted to interfacingCE with inductively coupled plasma (ICP) massspectrometry (MS). It has been demonstrated thatsuch hyphenated CE techniques could provide notonly sub-µg L−1 detection limits for the analysisof many types of environmental samples, butalso the capability for multielement monitoring ofvarious metal functionalities [11]. At present anefficient interface is still a challenge. In addition,the tiny amounts of sample result in concentrationdetection limits that are usually higher than ofchromatographic methods

4.4.1 Requirements of the interface

The most critical point in hyphenating CE to ICP-MS is the interface itself. It has to fulfill severalrequirements. One is the closing of the electri-cal circuit from CE at the end of the capillaryalthough the outlet of the capillary is connectedto a nebulizer. Another problem is the low flowrate of CE, generally not matching the flow ratesfor an efficient nebulization. Thus, it must be sup-plemented either by an additional sheath flow or byincreasing the flow through the capillary itself. Thefirst solution results in an undesired, considerabledilution of analyte species. The second alters dra-matically the separation, unless the capillary flowis increased after a suitable pause (e.g. a 20 minpause without a flow) only for transportation ofthe species bands to the detector (two-step mode).There is another undesired flow occurring when thenebulization gas is turned on (‘suction flow’) or asheath flow builds up a backpressure to the cap-illary (reversed flow). As CE capillaries are prin-cipally ‘open-tube systems’ nebulizer suction orbackpressure can cause flow rates of up to 1 m cap-illary length per minute (approx. 2 µL min−1). This


is twice the volume of frequently used capillaries(50 cm × 50 µm ID). Effective separation is thenno longer achieved. The nebulization efficiency isoften a function of exact and reproducible position-ing of the end of the capillary at the nebulizationgas stream. Owing to the low sample volumes ana-lyzed this is a critical point in setting up a system.Finally, the interface must preserve the high res-olution and separation power from CE when thespecies is transferred to the detector. Peak broad-ening or memory effects must be avoided to ensurethat no signals are artifacts. For several commer-cial CE systems it may be a problem to maintaintemperature control of the whole capillary up tothe interface. Avoiding capillary heating, however,is a critical point for reproducibility, efficient sep-aration and preventing sample degradation duringseparation. This problem may be overcome by atemperature-controlled liquid flow around the cap-illary, ensuring that the capillary is not overheated.Finally, the composition of the sheath electrolytemust be checked and has been investigated by sev-eral groups [25, 26]. It turned out that nitric acidwas best suited to be used for this task, althoughthe sheath electrolyte also has the function of anoutlet electrolyte. However, as mostly +/− polar-ity was used H+ ions did not enter the capillaryand decrease the background electrolyte pH. Fur-ther, the use of an inorganic acid instead of asalt solution prevents the nebulizer from crustingand blocking.

4.4.2 Technical solutions

The closing of the electrical circuit of CE duringnebulization is a primary problem needing asolution. The first published attempt was based ona Meinhard design and used a silver-painted endto the capillary, which was grounded [11]. A verylow current of only few µA was measured. Mostresearchers have employed a coaxial electrolyteflow around the CE capillary. The grounded outletelectrode is in contact with this electrolyte flowin all cases. Usually a current of 10–30 µAwas determined. Furthermore, the sheath flowwas used to adapt the flow rate to a suitablenebulization efficiency. These interface models

were based on a (modified) Meinhard design, onmodified MCN nebulizers or a modified DIN.The optimization of nebulization efficiency wasperformed by an optimal adoption of the flow ratewhen using systems based on MCN or DIN or byan exact positioning of the CE the position of thecapillary at the point of nebulization, employing amicrometer screw.

The reduction of dead volume and preserva-tion of resolution were achieved using labora-tory made special spray chambers. Michalke andSchramel [27] set up a spray chamber with anadditional gas flow, which coats the inner sur-face and inhibits condensation. The mass transportinto the ICP mass spectrometer was accelerated.Schaumloffel and Prange [26] constructed a low-volume spray chamber similar to that used byPolec et al. [28] and the modified DIN [23] oper-ated without spray chamber.

Suction through the capillary is often notconsidered in papers. However, if dealt witheither it is quantified and alterations on separationestimated or attempts are made to avoid thisundesired flow. There are two solutions based oncapillary dimension: (a) using a long CE capillary(1.5 m) with standard inner diameter of 50 µm [12]or (b) a short (2 cm) but narrow interface capillary(25 µm ID) set at the end of the CE capillary [26].

Both are based on the law of Hagen andPoiseuille [29].

V

t= r4

8

π

η

p1 − p2

L(4.3.1)

V = volume, which flows in the capillary at atime interval t ,

r = radius of the capillaryp1 = pressure at capillary’s inlet,p2 = pressure at capillary’s outlet,

if there is a suction θ , then p2 < p1, andθ = p1 − p2

η = viscosity of the buffer,L = capillary length

The value of θ = p1 − p2 is the force (suction)which pulls the electrolyte volume ‘V ’ during thetime interval t through the capillary. θ decreaseswith increased length or decreased inner diameter


of the capillary. Another solution in the literatureis to apply a negative pressure at the inlet duringseparation. However, exactly meeting the point ofequilibrium between nebulizer suction and countersuction at the inlet is complicated to achieve. Otherexperiments aimed for a (fragile) equilibriumbetween nebulization gas flow (influencing thesuction), and the sheath flow (feeding the suctionflow). However, in this case the nebulization gasflow must fulfill two different tasks, which areefficient nebulization and controlling the dynamicflow. Logically, an optimized compromise betweenthe two tasks is difficult to achieve. Finally, self-aspiration of the sheath flow has been suggestedto overcome the suction flow but has not beenwidely used in the literature. If the suction flow isneglected separation usually is altered markedly.Before starting to analyze samples the suctionflow should be checked and quantified. Attemptshave been described in the literature to checkthe suction flow. Michalke et al. [12] checked itin three steps. (1) The capillary was filled withbuffer, the high voltage turned on and the nebulizergas off; this determined the CE current. (2) Thecapillary inlet was exposed to air and the nebulizergas turned on for ca. 60 min. (3) The capillaryinlet was again put into the buffer vial, the highvoltage was turned on and nebulizer gas off; theCE current was determined again. When suctionoccurred, air was drawn into the capillary andinterrupted the electrical circuit. In this case thecurrent must be zero in step (3). Typically, nodifference in current was seen between steps (1)and (3) for 1.5 m capillaries. Schaumloffel andPrange [26] analyzed an Rb-containing standardsolution by CZE and varying nebulization gasstreams in consecutive runs. Here, too, no suctionflow was seen, as the standard was monitored at80 s during all runs, independently on the nebulizergas flow rates.

Hyphenated systems are mostly operated in aconventional mode: ICP-MS is detecting the efflu-ent from the capillary in real time. There is onlyone group of researchers who have switched fromthis mode to a ‘two-step mode’: Separation wasperformed in a very long capillary over 15–20 min,during which species were not leaving the capillary

[12]. Then the separated analyte bands were movedto ICP mass spectrometer within 2 min by pressureat the inlet. Advantages were seen in reducing asuction flow using the long capillary and by keep-ing the sheath flow low (10 µL h−1), as it was notused for improvement of nebulization efficiency.It contributed only a little (nearly no species dilu-tion) to the total flow during the pressure drivendetection step (1.5 µL min−1 = 90 µL h−1). Whenperforming IEF hyphenated to ICP-MS, such atwo-step procedure is obligatory. In this case thelong capillary, being completely filled with sampleand ampholytes, helps to improve concentrationdetection limits. Unfortunately, in a few CZE sep-arations this pressure driven postseparation flowcompromised the resolution.

Up to now interfaces for hyphenating CEto ICP-MS were laboratory-made or laboratory-modified systems based on commercially availablenebulizer parts. Recently, at least two interfaces(from refs [12 and 26]) were made available bycompanies. In Figure 4.3.7 schemes are presentedof the first published interface by Olesik, and ofthe two widely used interfaces due to Michalke orSchaumloffel.

4.4.3 Potential of CE-ICP-MS

When the interface is working reliably no specific,‘coupling problems’ occur and investigations canconcentrate on the broad potential of this tech-nique. The inaccessible advantages and potentialof CE-ICP-MS are its high separation capability,the short analysis time and the high selectivityand sensitivity of detection. The ICP mass spec-trometer accepts all buffers and modifiers withoutany problems, as the respective volumes reachingthe plasma are in the nL to µL range only. Thisdoes not affect plasma stability. Therefore, on-line preconcentration methods, such as ITP com-bined with CZE, are easily possible, providingspecies separation that is still acceptable by usingmarkedly increased sample volumes (= improvedconcentration detection limits). Buffer sandwichesor discontinuous buffer systems easily achieveimprovements in separation. Here, the sample plugmay be positioned in the middle of the capillary


CE capilary

Ptelectrode

nebulizercapillary

spraychamber

Argon

make-upliquid

Interface used widely by Michalke et al. 1996 - now

Silver PaintTo Complete Electrical Connection

Ferrule Seal

Fused Silica Capilary100 µm ID Ar

0.4 − 1.2 L/min

Interface introduced by Olesik et al. 1995

Interface used by Schaumlöffel et al. 1999 - now

CEcapillary

fixed tothe nutby a screw(black)and asilicone seal(white) nut for exact

positioning ofCE capillary

outlet buffer(electrolyte)

10 µL/h

Ar gas streamca. 0.85 L/min

Ar/H2 gas stream0.2 L/min

“O”-ring“O”-ring

Pt - electrode(grounded)

screw (black) withsilicone seal (white)

adapter to ICP-MS

spray chamber

nebulizer

and the pH of electrolytes is chosen in a way thatsome species appear anionic, others cationic. Thus,the former move towards the cathode, and the lat-ter towards the anode. Before leaving the capillaryeither the buffer pH is changed to induce eachspecies now to move to the ICP-MS instrument ora pressure driven detection step is started. The dif-ferent separation modes allow a separation solutionfor nearly all elemental species and a wide char-acterization of the sample. The powerful ICP-MSdetector provides elemental and isotope informa-tion, as well as multi element capability and lowdetection limits. Results are reported in the range0.05–30 µg L−1 depending on the species [11, 12,13, 23, 26]. Species identification is possible viamigration time and comparison with standard solu-tions. Further, there are no stationary phases thatcan impair species stability [30]. Several authorshave already demonstrated applications to real-world samples from the real world with very differ-ent matrices and very low species concentrations.

4.4.4 Limitations of CE-ICP-MS

Many problems can be related to the attempt todecrease concentration detection limits to con-centrations in the real world when using (partlyinadequate) stacking and separation conditions.Difficulties were often related to chemical inter-actions of samples, electrolytes and the capillaryor to detector interferences [31]. This is not sur-prising as species stability can easily be impairedby ‘incorrect’ CE conditions, predominantly com-plexing electrolytes, inadequate pH etc. [7, 14,32]. A very serious problem is a total or par-tial sticking of a compound to the capillary. Inthis case quantifications are typically wrong and,most critical ‘pseudo-species’ are detected. Suchpeaks may suggest species within a sample, but

Figure 4.3.7. Schemes of some interfaces used for CE-ICP-MS.The first was published by Olesik et al. [11]. Reprinted withpermission from Olesik et al., Anal. Chem., 67/1, 1–12. Copy-right 1995 American Chemical Society. The second used byMichalke et al. [12]. Fresenius’ Journal of Analytical Chem-istry, B. Michalke and P. Schramel, Vol. 357, pp. 594–599,1997, copyright notice of Springer-Verlag. The third was pub-lished by Schaumloffel et al. [26] Reproduced by permissionof Wiley-VCH, STM-Copyright and Licenses.


are only artifacts. Since comparison of migrationtimes of standards and samples performs speciesidentification, such artifacts may appear at a spe-cific migration time and thus may be ‘identified’ asa certain species. The well-known migration timevariations according to differences in ionic strengthof buffers or samples are a further problem forspecies identification [33]. Standard additions helpto overcome this uncertainty [20]. But generationof new species during analysis is also recognizedas a serious source of error [31, 32]. In these papersalkaline conditions helped to focus Sb species, butconversely altered some of them, as shown forSb(III) tartrate standards. Three peaks appearedfor one compound, proving decomposition. Thereare also problems known for the detection part.These problems can be summarized as variationsin detection times according to sample/buffer vis-cosity when using a two-step procedure and inter-fered mass signals due to polyatomic or isobaricinterferences [34]. The latter obliges one to ana-lyze at least two isotopes per element to detecta possible violation of natural isotope ratio, thuspointing to interferences. However, as ICP-MS is asequential detection system, the monitoring of toomany isotopes in parallel may result in missing fastmigrating peaks of one isotope. Detection limits ofthe CE-ICP-MS system up to now are either justsuitable or still too high for several real-world sam-ples. Thus, coupling to more sensitive detectors,e.g. HR-ICP-MS, is recommended and has alreadybeen partly achieved in the literature [25, 26, 35].

4.4.5 A few examples of speciation usingCE-ICP-MS

Applications to real-world samples, e.g. eluatesof tunnel dust and soil (platinum speciation [14]),plant extracts or sewage and fouling sludges (anti-mony, arsenic [32, 36]), or serum and human milk(selenium and iodine [13, 27]) or other body flu-ids (human cytosol) and tissues (metallothioneins,Cu, Cd; Co in cobalamins) have already beendemonstrated. Lobinski [35] used CE-ICP-MS inparallel to CE-ESI-MS. Surprisingly, this authorfound worse LOD when using CE-ICP-MS thanfor CE-ESI-MS. An explanation was seen in the

fact that detecting only Co in the molecule (ICP-MS detection) needs a higher mass detection sen-sitivity than does detecting the whole molecule(ESI-MS detection). The same group obtained asimilar result for MT speciation by CE coupledto ICP-MS or ESI-MS [37]. However, these find-ings are contradictory to findings of other groups,where the superior detection sensitivity of ICP-MSovercompensates this effect.

As another example Figure 4.3.8 shows theanalysis of Se species in a human milk sample,

400

600

800

1000

1200

1400

50 100

25.3

SeC

32.7

SeM

0

1000

2000

3000

4000

5000

6000

7000

8000

0 10 20 30 40 50 60 70

seconds

77S

e/io

ns/s

SeM

GSSeH

?

SeC

SeCM?

0 150

seconds

ions

/s

7.6

SeC

M

55.7

GS

SeH

Figure 4.3.8. A comparison of two different CE separationmodes hyphenated to ICP-MS for Se speciation in humanmilk. An advanced characterization is possible by employingtwo different CE modes. For CZE analysis the sample waspreconcentrated by freeze-drying. The compounds are (CZE):SeCM, SeC, SeM and GSSeH.


performed by CZE and in parallel by cIEF, eachcoupled to ICP-MS. 78Se and 77Se were monitored(only 78Se is shown). The electropherogramsshow multidimensional characterization by theuse of different separation methods. The cIEFseparation is superior as regards detection limitsand resolution. Both methods detect the samemajor Se species, but cIEF is able to separateand detect at least three unknown species inaddition. As species were analyzed in urine byvan Holderbeke et al. [25] or in several othermaterials by Schlegel [36]. An ArCl+ cluster didnot interfere with detection of species in bothpapers. For monitoring all relevant As species in asingle run a fast EOF was applied which pumpedanionic, uncharged and a cationic species to thedetector. One of the most frequent applications isthe analysis of metallothioneins. Several groupshave investigated the isoforms of this proteinand its Cd and Cu bondings. As examples,Polec et al. [28], Mounicou et al. [37] and Prangeet al. [38] are mentioned. The investigations ofPolec et al. and Mounicou et al. used both ICP-MSand ESI-MS hyphenations for MT characterizationand will be discussed in the corresponding section.Prange et al. analyzed MT samples in a volumeof 22 µL for Cu, Zn, Cd and Pb, employingtheir laboratory developed interface and an sf-ICP-MS system. They compared human brain Cu-MTfrom healthy persons with that from patients withAlzheimer’s disease. The two electropherogramsare shown in Figure 4.3.9.

4.5 ESI-MS detection

Electrospray ionization is an ionization pro-cess that may preserve the whole species intactunder optimal circumstances. ESI is suitable forextremely low flow rates. It is based on ‘ion evap-oration’, where charged droplets of the analytes aretransferred into gas phase. A volatile buffer con-sisting of a considerable amount of e.g. methanolsupports this ion evaporation. In fact the highvolatilization capability of CE electrolytes is nec-essary. The droplets are formed at the end of theCE capillary by the application of a high voltage,typically around 5 kV. The success of this detec-tion method is based on the ability to producemulticharged ions from element species of highmolecular weight such as metalloproteins and thusmaking the analysis of these compounds available,up to MW = 150 000–200 000 Da. The possibil-ity of coupling this detector to LC or CE systemsmakes it additionally immensely valuable [39].The soft ionization of element species finally givesthe chance to preserve the whole molecule (ele-ment species) when it is transferred to the gasphase and is subsequently analyzed in the massspectrometer [40]. Structural changes (mostly) donot take place as long as covalent bondings arepresent. In special cases (e.g. with selenium) sta-ble element–organic molecules can be analyzed.As an example selenocystamine is discussed: themost prominent signal is the protonated species([SeCM × H]+). The specific molecular mass at249 is detected, but up to 19 peaks in the mass

MT 1

MT 2

MT 3MT 1

MT 2 MT 3

63Cu 63Cu

2000

0.2

0.4

0.6

0.8

1.0

1.2

1.4

0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

700 1200

time/s

200 700 1200

time/s

cps/

cps

Rb-

85

Figure 4.3.9. Analysis of metallothioneins in brain cytosol from healthy persons compared with those in brain cytosol frompatients with Alzheimer’s disease (reprinted from [38] by permission of John Libbey Eurotext).


range m/z = 237–255 should be monitored, whichoriginate from the two Se atoms in the SeCMmolecule and the specific isotope pattern of Se.Some of the Se isotopes, however, show very lowabundances and thus may not be detected [41].Here, the (structural) information of two Se atomswhich are part of the SeCM molecule is availablewithout any further analytical efforts.

When applying collision induced dissociation(CID) together with an MS/MS system furtherstructural information can be gained. CID ispossible when the parent ions (filtered by the firstquadrupole) are accelerated by an electric field andthen pass through a region of minimal gas pressure(≤10−3 bar). This results in collision reactionsthat are governed by the kinetic energy from theacceleration of parent ions. Undesired ion–solventclusters may be destroyed or the parent ionsare fragmented into (molecule-specific) daughterions, selected by a second quadrupole. The latterprovides structural information about the elementspecies (parent ion). No other detection technique

is able to provide such detailed information aboutthe molecular weight and even the structure ofcompounds analyzed.

4.6 Problems of ESI-MS in speciation

One problem comes from the ion–solvent clus-ters. During the transfer of gas-phase ions intothe high vacuum (10−9 bar) a condensation ofsolvent molecules (e.g. methanol, water) to thegas-phase ions probably happens. The productis called an ion–solvent cluster. Production ofion–solvent clusters results in a splitting of onespecies into multiple signals, worsening detec-tion limits and increasing spectral complexity.Electrolytic processes at the metallic ESI tipneedle have been observed, resulting in the gen-eration of new species or a transformation ofspecies (e.g. by metal exchange). When ana-lyzing free metal ions such as Cu(II), multiplesignals from ion–solvent clusters are monitored(Figure 4.3.10). Most important, however, is the

0

20

40

60

80

100

60 70 80 90 100 110 120 130 140 150 160

m/z

rela

tive

abun

danc

e (%

)

63

6581

83

95

97

105107

123

125

137

139

155

157

[Cu(H2O)]+

Cu+

[Cu(MeOH)]+

Figure 4.3.10. Pitfalls of electrospray ionization: an ESI-MS spectrum of CuCl2 · xH2O in H2O (500 µg Cu L−1) is presented.Multiple ion–solvent clusters are seen. Collision offset voltage = −25 V. Reprinted from J. Chromatogr. A, Vol. 819, O. Schramelet al., pp. 231–242, Copyright (1998), with permission from Elsevier Science [48].


fact that native counter ions of the metal ion arereplaced by H2O and/or methanol, independent onthe counter ion initially present (e.g. [Cu(MeOH)]+instead of Cu2+2Cl−). This causes a total lossof initial species information. Destruction of theion–solvent cluster by CID is rarely possible dueto a reduction of Cu(II) to Cu(I) and the generationof only a few Cu(II)–solvent clusters follows this.These Cu-based clusters are easily identified by theisotope pattern coming from 63Cu/65Cu. Finally acharge reduction at the cluster may occur as foundby Agnes and Horlick[42]:

Cu(MeOH)n2+ −−−→ Cu(MeO)(MeOH)n−2

+

+ H(MeOH)+ (4.3.2)

4.7 CE-ESI-MS

4.7.1 Requirements of the ESI interfaceand solutions

ESI interfaces for CE are commercially available.The requirements of the interface are similarto those discussed earlier. The closing of theelectrical circuit from CE during ion evaporationis provided by an electrolyte sheath flow. Effectiveion production is possible using a suitable sprayvoltage easily controlled by instrument software.For older instruments a laboratory made devicefor reproducibly optimized positioning of the CEcapillary to the ESI-tip is still necessary and hasbeen described by Schramel et al. [41].

4.7.2 Potential of CE-ESI-MS

The potential of this hyphenated technique isdefined by the advantages from CE discussed forspecies separation combined with the possibilitiesand species information coming from ESI-MS. Incontrast to the CE-ICP-MS coupling, direct speciesdetection is available here. When the element ofinterest imposes its specific isotope pattern on thetotal molecule mass, elemental information withinthe species is also possible: this means maximizedspecies information is gained in one analytical

effort. If the elemental pattern is not seen onthe total molecular mass, structural information isprovided when applying MS-MS mode. Speciesidentification is gained via migration time and viam/z of species.

4.7.3 Limitations of CE-ESI-MS

The problems of CE-ESI-MS result from the lim-itations of CE discussed for species separationcombined with the problems from ESI-MS. A bigproblem is that only volatile buffers are possi-ble for ESI detection. Therefore, a free choicefor separation electrolytes is not available. Mostlya compromise between separation and detectioncapability is necessary. Furthermore, Mounicouet al. [37] found detrimental effects on separationwhen the sheath buffer was different from the inletbuffer. This is a significant disadvantage comparedwith CE-ICP-MS. Elemental information is possi-ble only in specific cases where the isotope patternis imposed on the molecule mass (e.g. selenoaminoacids). Unfortunately, the ionization process itself,as mentioned above, causes several severe prob-lems in speciation. Summarizing, these problemsare a total loss of species information by gas-phase ligand replacement, gas-phase intramolec-ular charge transfer, thermal decomposition onheated detector parts, ion-solvent cluster genera-tion combined with worsening of detection limitsand increasing problems of species identificationby spectral complexity, generally worse detectionlimits, resulting in species of low concentration notbeing monitored. Application to real-world sam-ples is thus rare.

4.7.4 A few examples of speciation usingCE-ESI-MS

There are several examples employing CE-ESI-MS. Most of them have been in combination withICP-MS technology and part of multidimensionalconcepts. In ref. [41] the possibilities and limita-tions of this technology were demonstrated usingspecies of Sb, Cu and Se as examples. Nearlyall other studies have analyzed metallothioneins

COMBINATION OF CE-ESI-MS AND CE-ICP-MS 221

3

10 15 20

5

6

0

0.4

0.8

1.2

×104

×106

1

2 4

CdCu

Zn

migration time, min

time, min

0 5 10 15 200

1

2

3

4

5

1

2

3

1300 1500 1700 1900

1369.0

1378.0

1388.0

1397.0

1392.0

1383.0

1710.0

1722.1

1746.0

1739.0

1734.0

1727.0

m/z, u

Inte

nsity

, cps

Inte

nsity

, cps

Inte

nsity

, cps

Figure 4.3.11. (a) An electropherogram of MT 1, analyzed byCZE-ICP-MS. Spectra for Cu, Cd and Zn are seen. (b) Thesame sample analyzed by CZE-ESI-MS, where the total ioncurrent is monitored. The peaks were further investigatedand the specific masses could be attributed to Cd7-MT2(m/z = 1382), accompanied by a mixed complex of Cd7-MT2Cd6Zn-MT2, Cd5Zn-MT2 and Cd4Zn-MT2. Other subisoformswere also identified. Reprinted from K. Polec et al., CellMol. Biol., 46/2, 221–235 (2000), reprinted with permissionfrom CMB [28].

or selenocompounds, each of them stable duringthe ESI process. As mentioned above, impressiveexamples are published in refs [37] and [28], bothstudies from the same group. They both inves-tigated metallothioneins first by CE-ICP-MS andsubsequently by CE-ESI-MS for advanced identi-fication of the metal–MT compounds. Comparingthe two methods they found a poorer resolutionin CE-ESI-MS and problems arising from theirinstrumental setup, which allowed no temperaturecontrol of the capillary outside the instrument.Mt isoforms were determined and subisoformswere additionally found, attributed to Cd, Cu andZn. The Mt isoforms were then analyzed by CE-ESI-MS and subsequently their m/z values weredetermined. Figure 4.3.11 shows the respectiveelectropherograms and mass identifications.

5 COMBINATION OF CE-ESI-MS ANDCE-ICP-MS: MAXIMIZED SPECIESINFORMATION

Summarizing the advantages and limitations ofboth hyphenation techniques we can use the highresolution power of CE for the separation of metalspecies [43, 44] combined with molecular andstructural information from ESI-MS and elementaland isotope information from ICP-MS detection.

Electrospray using soft ionization provides thefollowing informations about element species:

• direct detection and quantification of species[45, 46];

• possibly elemental information by a characteris-tic isotopic pattern [47, 48];

• structural information by using the MS/MStechnique [46].

Thus, information is gained about the elementwithin the species, resulting in total species infor-mation and identification. Additionally, speciesidentification is possible according to migrationtimes and standard addition procedures.

Unfortunately, some disadvantages are known:

• Only volatile buffers are appropriate for theESI process. Therefore, a compromise between


optimal separation and sensitive detection isoften unavoidable [46].

• Elemental information is achieved only indi-rectly by a characteristic isotope pattern. If theelement does not show a very characteristic iso-tope pattern and/or the whole species is ratherlarge (isotope pattern superimposed) no elemen-tal information is gained.

• Possible losses of species information are knowndue to gas-phase ligand replacements or gas-phase intramolecular charge transfers [42, 48].Electrolytic processes at the stainless steelcapillary tip are not excluded.

• Detection limits are strongly compound depen-dent. For metal ions low values between 3 and300 µg L−1 are reported without CE coupling.However, for metal-containing amino acids andpeptides detection limits of around 500 µg L−1

are still typical with CE coupling. This is toopoor for most real samples.

The use of an ICP mass spectrometer aselement-specific detector for CE is a useful com-plementary technique to ESI-MS. The advantagesof ICP mass spectrometer as a detector are:

• the very low detection limits of this hyphenatedsystem (<1 µg L−1 [12, 23, 26]);

• direct element information and element quantifi-cation;

• All buffer systems and stacking procedures fitwell to the detector, so no compromises betweenseparation and detection are necessary [23, 27];

• Species identification via migration times.

On the other hand there are some limitations:

• Detection of only the element is possible. Thereis no detection of the species itself.

• Unknown species cannot be identified. Theycan only be characterized, e.g. via cIEF/ICP-MS[49].

• m/z signals can suffer interference from poly-atomic interferences, resulting in pseudo-element signals [50, 51]. Detection limitsworsen.

Considering all these factors, CE-ESI-MS pro-vides maximum information, i.e. direct determina-tion of the element in its specific form. CE-ICP-MS

directly yields elemental information. Structuralinformation can be obtained indirectly by meansof CE data.

The combination of CE-ICP-MS and CE-ESI-MS can provide maximum species information.

6 CONCLUSION

Capillary electrophoresis is proving a very impor-tant and suitable tool for speciation investigations.Generally it acts as a separation method for ele-mental species before they are detected eithernonselectively or in elemental- and/or molecule-selective ways. Its use in orthogonal multidi-mensional strategies helps to provide speciesinformation with increased certainty. Speciationanalysis can often be performed where othermethodical approaches cannot promise success.Especially in combination with detection tech-niques such as ICP-MS and ESI-MS its very highpotential is realized and used for improved spe-ciation of the elements. Nevertheless, the variouslimitations and problems – such as loss of speciesinformation or mimicking of unknown species bygenerating artifacts – must be carefully consideredwhen using this technology. Finally, as CE canonly play a (essential) part in speciation analysistogether with other technologies, it should be used(only) in those fields where separation is difficultand cannot be achieved more easily and cheaplywith other methods, and where species concentra-tions are high enough for detection without any(severe) problems. Thus the typical sample to bespeciated by CE contains element species that aredifficult to separate (uses the high resolution ofCE) but in high species concentrations (low sampleintake causes no detection problem).

7 REFERENCES

1. Mota, A. M. and Simaes Goncalves, M. L., Direct meth-ods of speciation of heavy metals in natural waters, in Ele-ment Speciation in Bioorganic Chemistry , Caroli, S. (Ed.),John Wiley & Sons, Inc., New York, 1996, Chapter 2,pp. 21–87.

2. Morrison, G. M. P., Trace element speciation and itsrelationship to bioavailability and toxicity in naturalwaters, in Trace Element Speciation: Analytical Methods

REFERENCES 223

and Problems , Batley, G. E. (Ed), CRC Press, BocaRaton, FL, 1989, Chapter 2, pp. 25–42.

3. Florence, T. M., Trends Anal. Chem., 2, 162 (1983).4. Turner, D. R., Met. Ions Biol. Syst., 18, 137 (1984).5. Templeton, D. M., Ariese, F., Cornelis, R., Danielsson,

L.-G., Muntau, H., van Leeuwen, H. P. and Lobinski, R.,Pure Appl. Chem., 72, 1453 (2000).

6. Michalke, B., Fresenius’ J. Anal. Chem , 350, 2 (1994).7. Majidi, V. and Miller-Ihli, N. J., Analyst , 123, 803

(1998).8. Dunemann, L. and Begerow, J., Kopplungstechniken zur

Elementspeziesanalytik , VCH, Weinheim, 1995.9. Kuhn, R. and Hofstetter-Kuhn, S., Capillary Electro-

phoresis: Principles and Practice, Springer, Berlin, 1993.10. Timerbaev, A. R., Electrophoresis , 18, 185 (1997).11. Olesik, J. W., Kinzer, J. A. and Olesik, S. V., Anal.

Chem., 67, 1 (1995).12. Michalke, B. and Schramel, P., Fresenius’ J. Anal. Chem.,

357, 594 (1997).13. Michalke, B. and Schramel, P., Electrophoresis , 20, 2547

(1999).14. Michalke, B., Lustig, S. and Schramel, P., Electrophore-

sis , 18, 196 (1997).15. Bulletin 1641, Coatings: The Key to Success in Capillary

Electrophoresis of Proteins, Bio-Rad, Hercules, CA, 1992.16. Jen, J. F., Wu, M. H. and Yang, T. C., Anal. Chim. Acta,

339, 251 (1997).17. Quevauviller, Ph., Maier, E. A. and Griepink, B., Quality

control of results of speciation analysis, in ElementSpeciation in Bioorganic Chemistry , Caroli, S. (Ed.),John Wiley & Sons Inc., New York, 1996, Chapter 6,pp. 195–222.

18. Albert, M. M., Demesmay, C. and Rocca, J. L., Analusis ,21, 403 (1993).

19. Michalke, B., Speziesanalytik in Umwelt und bio-medizinischen Proben , Habilitationsschrift, TechnischeUniversitat Graz, Austria, 1999.

20. Michalke, B., Fresenius’ J. Anal. Chem., 351, 670 (1995).21. Michalke, B. and Schramel, P., J. Chromatogr. A, 750, 51

(1996).22. Hill, S. J., Bloxham, M. J. and Worsfold, P. J. J., Anal.

At. Spectrom., 8, 499 (1993).23. Liu, Y., Lopez-Avila, V., Zhu, J. J., Wiederin, D. R. and

Beckert, W. F., Anal. Chem., 67, 2020 (1995).24. Hill, S. J., Ford, M. J. and Ebdon, L., J. Anal. At. Spec-

trom., 7, 719 (1992).25. van Holderbeke, M., Zhao, Y., Vanhaecke, F., Moens, L.,

Dams, R. and Sandra, P., J. Anal. At. Spectrom., 14, 229(1999).

26. Schaumloffel, D. and Prange, A., Fresenius’ J. Anal.Chem., 364, 452 (1999).

27. Michalke, B. and Schramel, P., Electrophoresis , 19, 270(1998).

28. Polec, K., Mounicou, S., Chassaigne, H., Lobinski, R.,Cell Mol. Biol., 46, 221 (2000).

29. Formeln Physik Chemie Mathematik , Buch und ZeitVerlag GmbH, Koln, 1980.

30. Harms, J. and Schwedt, G., Fresenius’ J. Anal. Chem.,350, 93 (1994).

31. Michalke, B. and Schramel, P., J. Anal. At. Spectrom., 14,1297 (1999).

32. Michalke, B. and Schramel, P., J. Chromatogr. A, 1999,834, 341 (1999).

33. Bondoux, G., Jandik, P. and Jones, R. W., J. Chro-matogr. A, 602, 79 (1992).

34. ICP-MS Interferenz Tabelle, Finnigan MAT, Bremen,1995.

35. Lobinski, R., Appl. Spectrosc., 51, 260A (1997).36. Schlegel, D., Arsen-Speziationsanalytik mit Ionenchro-

matographie und Kapillarelektrophorese in Kopplung mitelementspezifischer Detektion, UfZ-Bericht 5/1999, ISSN0948–9452, Leipzig, 1998.

37. Mounicou, S., Polec, K., Chassaigne, H., Potin-Gautier, M.,Lobinski, R., J. Anal. At. Spectrom., 15, 635 (2000).

38. Prange, A, Schaumloffel, D., Richarz, A. and Bratter, P.,Speciation of metallothioneins in animal and humansamples from nanoliter volumes, in Metal Ions inBiology and Medicine, Vol. 6, Centeno, J. A., Colley, Ph.,Vernet, G., Finkelman, R. B., Gibb, H. and Etienne, J. C.(Eds), John Libbey Eurotext, Paris, 2000, pp. 430–432.

39. Smith, R. D., Loo, J. A., Barinaga, C. J., Edmonds, C. G.and Udseth, H. R., J. Chromatogr., 480, 211 (1989).

40. Cole, R. B., Electrospray Ionization Mass Spectrome-try – Fundamentals, Instrumentation and Applications .John Wiley & Sons, Inc., New York, 1997.

41. Schramel, O., Michalke, B. and Kettrup, A., Fresenius’J. Anal. Chem., 363, 452 (1999).

42. Agnes, G. R. and Horlick, G., Appl. Spectrosc., 48, 655(1994).

43. Jackson, P. E. and Haddad, P. R., Trends Anal. Chem.,12, 231 (1993).

44. Michalke, B., Fresenius’ J. Anal. Chem., 354, 557 (1996).45. Chassaigne, H. and Lobinski, R., Anal. Chem. Acta , 359,

227 (1998).46. Chassaigne, H. and Lobinski, R., Fresenius’ J. Anal.

Chem., 361, 267 (1998).47. Corr, J. J. and Anacleto, J. F., Anal. Chem., 68, 2155

(1996).48. Schramel, O. Michalke, B. and Kettrup, A., J Chro-

matogr. A, 819, 231 (1998).49. Michalke, B. and Schramel, P., J. Chromatogr. A, 807, 71

(1998).50. Seubert, A. and Meinke, R., Fresenius’ J. Anal. Chem.,

348, 510 (1994).51. Michalke, B. and Schramel, P., Analusis , 26, M51 (1998).

4.4 Gel Electrophoresis for Speciation Purposes

Cyrille C. CheryLaboratory for Analytical Chemistry, Ghent University, Belgium

Abbreviations . . . . . . . . . . . . . . . . . . . . . . 2241 Introduction and Definition . . . . . . . . . . . . 225

1.1 Introduction; gel electrophoresis andspeciation . . . . . . . . . . . . . . . . . . . . . 225

1.2 Speciation . . . . . . . . . . . . . . . . . . . . 2251.2.1 Applicability . . . . . . . . . . . . . 2251.2.2 Limitations . . . . . . . . . . . . . . 226

1.3 Apparatus . . . . . . . . . . . . . . . . . . . . 2261.4 Definitions . . . . . . . . . . . . . . . . . . . . 2271.5 Typical applications . . . . . . . . . . . . . 227

2 Techniques and Procedures . . . . . . . . . . . . 2282.1 Basics . . . . . . . . . . . . . . . . . . . . . . . 2282.2 Native/denaturing electrophoresis . . . 2282.3 Restricting medium: gradient

or linear gel . . . . . . . . . . . . . . . . . . . 2282.4 Stacking or sample concentration:

discontinuous buffers . . . . . . . . . . . . 2292.5 Application . . . . . . . . . . . . . . . . . . . 229

2.5.1 Nondenaturing electrophoresis 2312.5.2 Two-dimensional gel

electrophoresis (2DE) . . . . . . 2322.5.2.1 Isoelectric focusing

(IEF) . . . . . . . . . . . . 2332.5.2.2 Sodium dodecyl sul-

fate – polyacrylamidegel electrophoresis(SDS-PAGE) . . . . . . 234

2.5.2.3 Example: 2DE ofselenised yeast . . . . . 234

3 Detection of Trace Elements . . . . . . . . . . . 2353.1 General . . . . . . . . . . . . . . . . . . . . . . 2353.2 Detection of trace elements in

subsamples of the gel . . . . . . . . . . . . 2353.2.1 Liquid introduction system:

inductively coupledplasma – mass spectrometry(ICP-MS), AAS and AES . . . 235

3.2.2 Solid sample analysis:electrothermal vaporisation(ETV) – ICP; graphitefurnace – atomic absorptionspectrometry (GF-AAS) . . . . . 236

3.2.3 Nuclear analytical chemistry:scintillation counting, neutronactivation analysis (NAA) . . . 236

3.3 Detection of trace elements in awhole gel . . . . . . . . . . . . . . . . . . . . . 2363.3.1 Autoradiography . . . . . . . . . . 2363.3.2 Laser ablation – inductively

coupled plasma – massspectrometry (LA-ICP-MS) 237

3.3.3 Particle induced X-rayemission (PIXE) . . . . . . . . . . 238

3.3.4 Mass spectrometry (MS) . . . . 2384 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . 2385 Acknowledgements . . . . . . . . . . . . . . . . . . 2386 References . . . . . . . . . . . . . . . . . . . . . . . . 238

ABBREVIATIONS

1D one-dimensional2DE two-dimensional gel electrophoresis

AAS atomic absorption spectroscopyAES atomic emission spectroscopyC degree of cross-linkingCE capillary electrophoresis

INTRODUCTION AND DEFINITION 225

CHAPS 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate

DTT DithiothreitolETV electrothermal vaporisationGE gel electrophoresisGF graphite furnaceICP-MS inductively coupled plasma – mass

spectrometryIEF isoelectric focusingIPG immobilized pH gradientLA-ICP-MS laser ablation – inductively coupled

plasma – mass spectrometryMALDI matrix-assisted laser

desorption/ionisationMS mass spectrometryNAA neutron activation analysisPAGE polyacrylamide gel electrophoresispI isoelectric pointPIXE particle induced X-ray

emissionSDS sodium dodecyl sulphateSGE slab gel electrophoresisT total acrylamide concentrationTris tris(hydroxymethyl)aminomethaneV h volt hour

1 INTRODUCTION AND DEFINITION

1.1 Introduction; gel electrophoresisand speciation

Although far less used for speciation purposes thanits parent method, capillary electrophoresis (CE),gel electrophoresis (GE), or more precisely slabgel electrophoresis (SGE), is a very promisingmethod. It not only allows a rapid separation of acomplex mixture, even if detection can be tedious,but it also allows various separation mechanismsaccording to the need. Some of the terms andprinciples of the two methods are comparable andit is thus advisable to take a look at the previouschapter on capillary electrophoresis in parallel tothis one. Furthermore, although this chapter ismore dedicated to gel slabs, the one-dimensionalmethods can be transposed to gel rods or columns,as is the case for isoelectric focusing.

One may wonder why gel electrophoresis isstill interesting when capillary electrophoresis isso powerful. Let us give a striking image: gelelectrophoresis would indeed belong to the pastif this method was comparable in its figure ofmerit with thin layer chromatography and capillaryelectrophoresis with capillary chromatography; inother words, gel electrophoresis would be abasic quality control method and CE a methodwith unmatchable capacities, because of its highresolution and speed of analysis. However, gelelectrophoresis cannot be considered as a cheapalternative to CE, if we pursue the comparisonwith chromatography, but as a method with othergoals and a different output. It is an efficient andrapid separation method for complex mixtures; itallows detection with radiotracers; it can be two-dimensional; the amount of material is larger andallows an off-line identification by the means oftrypsin cleavage and mass spectrometry; and, lastbut not least, the material is always available forfurther studies: gels stored for years can be usedfor identification since the compounds have notbeen degraded.

Although the application of gel electrophoresisfor speciation purposes is just beginning, it isdefinitely worth applying on a large scale.

Some excellent books are recommended [1–3]to get a deeper and practical insight into gelelectrophoresis and its applications. In this chapter,only the use of gel electrophoresis for elementalspeciation purposes will be examined in detail.

1.2 Speciation

1.2.1 Applicability

The area covered by gel electrophoresis forspeciation purposes is charged macromolecules towhich any metal or metalloid is bound, covalentlyor not. Even if this chapter deals mainly withproteins, other macromolecules can be separatedby this method, such as DNA or humic acids.

This method has been used in combination withnumerous metals. The applicability is not limitedto certain elements, but by practical considerations


Table 4.4.1. Applications of SGE to elemental speciation: a survey.

Element Matrix SGE separation Detection Remark Reference

Co Serum Crossedimmunoelectro-

LA-ICP-MS Stability of species? [32]

phoresisFe Apotransferrin Native Autoradiography [30]

Bacterium SDS-PAGE PIXE [34]P Milk SDS-PAGE NAA [20, 22]Pb Humic acids SDS-PAGE LA-ICP-MS [33]Pt Grass SDS-PAGE Voltammetry Contamination from

electrodes?[15]

Serum Native 2DE Autoradiography Very promisingmethod

[11]

Se Soft tissues 2DE Autoradiography [25, 26]Yeast 2DE Autoradiography Stability checked [14]

ETV-ICP-MS Stability checked [36]LA-ICP-MS [5]

Glutathione peroxidase SDS-PAGE Mineralisation:HPLC-fluorescence

[16]

Glutathione peroxidase SDS-PAGE GF-AAS [18]ETV-ICP-MS Stability checked [36]

Soft tissues SDS-PAGE Liquid scintillation [19]LA-ICP-MS [37]HG-AFS [38]

V Serum Native Autoradiography Stability checked [39]Multi-element Kidney IEF NAA Stability of species? [23]

Liver IEF X-ray fluorescence [40]

such as the limits of the detection method and theamount of material that can be brought on the gel.

The most representative applications of gelelectrophoresis to speciation are summarized inTable 4.4.1.

1.2.2 Limitations

Although the detection of the metal is possible(see Section 3.2.), quantitation is still difficult.Up to now, gel electrophoresis has remained asemiquantitative method and this is also the casewhen it is used for speciation.

It should always be kept in mind that, althoughgel electrophoresis is a versatile separation method,speciation implies that the compound must be keptintact. The integrity of the analyte is fundamentaland even dictates the separation process; thereforethe choice of buffers, pH and electrodes, tomention the most important parameters, is crucial.

Some standard methods, which use detergentsand denaturing agents, cannot be considered forspeciation of protein–metal complexes becausethe basic structure of the complex is lost andthe protein is stripped of its metal. Ideally, even if

the metal is combined with the protein, the stabilityof the protein during the separation process has tobe checked; indeed, artefacts can occur, due amongother things to oxidation of residues of the protein,as in the case of selenoproteins.

Additional considerations have to be taken intoaccount, such as possible contamination of thegel through the electrodes [4]. This is exempli-fied by the use of platinum electrodes. The largecontact area between the electrode and the gel pro-motes the release of oxidized platinum into thegel, which, when platinum–protein complexes arestudied, leads to the detection of an artefact. Con-tamination from the gels has never been reportedand mineralisation of commercial gels followedby detection by ICP-MS has demonstrated thatno high blank value has to be feared [5]. Lackof purity of chemicals can still be a problemand therefore home-made gels should be preparedwith extreme caution and with the purest chemi-cals available.

1.3 Apparatus

Basically, an electrophoresis experiment requiresa high voltage generator, typically up to 2000 V,

INTRODUCTION AND DEFINITION 227

a set of electrodes and a temperature-controlledseparation surface or chamber. The last point iscritical since heat is produced through the Jouleeffect, which can disturb the separation. For smallgels, less than 10 cm2, a Pelletier cooling is veryefficient; for larger ones water cooling is the bestsolution. Several variations are available from thisstarting set, including submarine electrophoresis,vertical or horizontal electrophoresis. Other acces-sories may be necessary if the method is usedintensively, such as staining trays. Detection forelemental speciation will be discussed separately.

1.4 Definitions

In an electric field, charged molecules or com-plexes will migrate to the electrode bearingthe charge opposite to their global charge. Thisglobal charge is not necessarily the charge of themolecule, but rather the charge of the particlethat is formed during electrophoresis, for examplethe charge of the protein in an SDS micelle orthe charge of a metallo-complex. If a voltageV is applied between two electrodes separatedby a distance L, a field E appears according toequation (4.4.1). The migration velocity (ν) of aparticle in this field is proportional to the mobil-ity (µ) of the particle and the field strength (E)(equation 4.4.2), where µ is an intrinsic parameterof the particle.

E = V/L (4.4.1)

ν = µE (4.4.2)

The unit mainly used in electrophoresis isthe volt hour (V h), since this value is propor-tional to the displacement d of the particle.Indeed, it can be proved that the velocity,ν, becomes rapidly constant and by combiningequations (4.4.1) and (4.4.2):

d = νt = (µ/L)V t (4.4.3)

Since µ is a characteristic of the particle and L

of the system, the distance d to the electrode isrelated to a unit that has the dimension of a voltagemultiplied with a time, which is traditionallyexpressed in V h.

This unit is particularly interesting if a sepa-ration has to be translated from one gel size toanother. If a certain gel size is used to create andoptimise a method and half the number of lanesare used afterwards, the number of V h is invari-ant, the parameters that have to be adjusted aresimply current and power, both divided by two.In contrast, if the migration distance is divided bytwo, the number of V h is divided by two, althoughthis time it may be a rough approximation.

Migration occurs in a liquid medium, namelythe buffer, which is one of the key elements ofthe separation and especially of the stability of themetal bound to the macromolecule. This buffer,which is not necessarily a pH buffer but rather agood solvent for the particles, may be a single solu-tion or a combination of two or three solutions ifthe separation process requires it (see Section 2.4).

The gel is the second key parameter for agood separation of the particles. It determines toa first approximation which separation mechanismoccurs. Various gels are available, agarose andpolyacrylamide being the most common ones.Generally, the choice is made between the twoaccording to the size of the particles to beanalysed. For larger particles, typically over 10 nmin diameter, agarose gel is preferred, especially forthe analysis of DNA or RNA. Polyacrylamide gelis the polymer of choice for most proteins.

This polymer is obtained by copolymerisationof acrylamide and a cross-linking agent, usuallyN ,N ′-methylenebisacrylamide, which confers itsthree-dimensional structure on the gel. The poresize is defined by two parameters, C and T ,both expressed in per cent. They are related tothe polymerisation process and the quantities ofmonomer and cross-linking agent used, T beingthe mass of acrylamide per gel volume and C thepercentage of cross-linking agent in the gel. Forour purpose, it is only necessary to know that,if T increases, the pore size decreases (the morepolymer per volume, the less free volume).

1.5 Typical applications

Whereas applications of gel electrophoresis arenumerous, they are just emerging in the field


of elemental speciation. They range from sepa-ration of DNA/RNA, humic acids, proteins, todyes. A particular branch of gel electrophoresisis worth mentioning: two-dimensional gel elec-trophoresis (2DE). It is the most advanced andsuccessful method, this success being due to theincreasing importance of proteomics: the com-bination of PROTEin analysis and genOMICSaims at unravelling the mysteries of the expres-sion of the genome into proteins. Thus proteomicsaims to characterise all proteins present in aparticular sample, e.g., Saccharomyces cerevisiae,human serum, Eschericia coli to quote just afew. Two-dimensional electrophoresis has becomea widespread and very reliable analytical methodfor this purpose. To give a single example, 2DEenables the separation of more than 10 000 proteinson a single gel [6].

One further point which has to be mentionedabout the application of gel electrophoresis isthe widespread availability of ready-to-use gelsand buffers, making it less and less necessary tomake them oneself, which used to be most timeconsuming.

In the field of speciation by means of gelelectrophoresis, most work has been done onproteins. That is why this chapter will concentrateon this point. But the examples that are quotedhere can always be transposed to other types ofsamples, as long as the mechanism of separationis relevant (e.g., isoelectric focusing is of no usefor a molecule without pI).

2 TECHNIQUES AND PROCEDURES

2.1 Basics

Even before choosing the procedures, the follow-ing question has to be addressed: is the metalcovalently bound to the protein? If so, as is thecase for selenium in some proteins, the species arerelatively stable during the separation, and dena-turing conditions can be used; this means condi-tions where only the primary structure is preserved.Otherwise, nondenaturing electrophoresis must beapplied, even if this implies a loss of separationefficiency; nondenaturing is equivalent to native.

The first decision is therefore whether a native ora denaturing procedure is to be used. The otherquestions to be answered are related to the kind ofsample or the mixture of proteins to be separated:gradient or linear gel (Section 2.3), with or withoutstacking (Section 2.4). Any paired combination isfeasible, giving eight theoretical associations.

2.2 Native/denaturing electrophoresis

The first and most straightforward method is nativeelectrophoresis. This means that the proteins,without any modification to their secondary andtertiary structures, are submitted to electrophoresis.The buffer is chosen so that the protein is notdenatured. Biochemists use this separation methodwhen they are interested in the activity of theenzymes that are isolated, activity that would belost if the proteins were denatured. However, thistype of separation is subject to a major drawback,as no buffer system is suited to all separations.Firstly, no universal buffer exists for the separationof all proteins. In a buffer with a pH below10, proteins with a pI of 11, thus positivelycharged, migrate to the cathode, and proteins witha pI below 10, negatively charged, migrate to theanode; in other words, they migrate in oppositedirections and cannot be seen on a single gel. Asolution would be to use extreme pHs but thoseare prohibited since they denature the proteins.Secondly, no buffer exists that allows the correctpreservation of all metal–protein complexes, e.g.,some buffers may affect a vanadium–proteincomplex without affecting a platinum complex.

For speciation purposes, however, the nativemethod is compulsory when the metal is notcovalently bound to the protein. Should the proteinbe denatured, the complexing site would bedestroyed and the metal would be set free.

Various nondenaturing buffers have been pro-posed [7] and only one will be given as examplein Section 2.5.1.

2.3 Restricting medium: gradient orlinear gel

Basically, separation takes place in either a restrict-ing medium or a free medium. Restricting means

TECHNIQUES AND PROCEDURES 229

that the particles interact with the gel, either phys-ically, e.g., because of their size, or, with an evenbroader definition, chemically, because of inter-actions of the proteins with a pH gradient orwith antibodies.

For the separation according to size, exceptduring stacking, a restricting medium is preferred.There is still a question to be answered, namelywhether the gel must be used with a constantdensity or a gradient. A gradient is a continuouschange in density of the gel in the direction ofmigration, or in other words, a continuous gradientof the pore size. The gel begins with large poresizes and ends with restricted sizes, so that thefriction force constantly increases in the gel, up toa point where the velocity of the protein is verylow. A gradient also ensures the separation of abroad mass range, typically from 10 to 200 kDa, incomparison with smaller ranges for homogeneousgels. Thus, a gradient gel allows sharper andsometimes easier separations, especially when littleis known about the range of molecular masses ofinterest. Homogeneous gels still have advantages,especially when one is interested in a particularmass or family of proteins, since they offer akind of zoom process. Further, a more precisemass determination is possible, especially whenFerguson plots are used.

2.4 Stacking or sample concentration:discontinuous buffers

By stacking, one understands a process capableof concentrating the sample in the gel beforethe real separation occurs. This is particularlyinteresting if the analytes are present at lowconcentration. Indeed, stacking allows the proteinsto be concentrated in sharp bands, a prerequisitefor Rf measurements, i.e. migration distance,or speciation, when the metal concentration islow. The principle is isotachophoresis, or moreprecisely moving-boundary electrophoresis. Themethod is also widely known as discontinuous ordisc electrophoresis because of the discontinuity ofproperties between stacking and separating gels inbuffer and pH.

For clarity, the basics of the method will besummarized here, for a system migrating from thecathode to the anode. The gel is physically madeof two zones, the first one, where the sample isapplied, being the stacking gel, where the gel isnot restricting; the second is made of the resolvinggel, where separation occurs. A set of three buffersmust be chosen, but even though this task isvery tedious it is one of the best documented [7].The goal is to obtain three ions with increasingmobilities from the cathode to the anode:

• a terminating ion, with a low mobility, at thecathode in our case;

• a leading ion, with the highest mobility of theions, present in the gel and the anode;

• a common counter ion.

The proteins are applied at the cathode and, in amanner of speaking, are sandwiched between thetwo ions. When a voltage is applied, moleculesrange according to their mobilities, from theleading ion, the proteins, to the terminating ion.This occurs with a most interesting characteristic,a constant concentration in one band, dependenton the concentration of the leading ion. Throughthis effect the proteins are pre-separated in sharpbands (see Figure 4.4.1).

At the border between stacking and separatinggel, a new force, the friction with the gel material,affects the macromolecules. The terminating iondoes not interact with the gel, since its size isnegligible in comparison with the pore size, andmigrates further. At this stage, the proteins aresurrounded by the terminating ion and migratefarther, but this time following the principles ofzone electrophoresis.

2.5 Application

In order to give a practical idea of a separation bygel electrophoresis, two methods have been cho-sen: one-dimensional native electrophoresis andtwo-dimensional high resolution electrophoresis.Those separations illustrate the most extreme casesin trace element speciation with gel electrophore-sis: 2DE is used when high separation capac-ity is needed but when the species can resist


3. Sample in the separating gel.

migration

leading ion

migration

Gel description.

stacking gellarge pores

separation gelsmall pores

1. Sample application.

anode bufferterminating ionlow mobility

samplegel buffer in both gelsleading ionhigh mobility

2. Sample in the stacking gel.

Separation starts.

The proteins are separated according to the principles of isotachophoresis.Sharp bands are obtained.

terminating ion

proteins

Separation occurs according to friction with the gel.

In the separating gel, the terminating buffer goes farther but the proteins, whichencounter now a high friction, are separated according to mass to charge ratio.

proteins

+

Figure 4.4.1. Principle of stacking.

denaturation, whereas nondenaturing electrophore-sis is used when the complex is fragile andwould not resist to a denaturing separation. Theseexamples are largely inspired by dedicated andextended chapters in the books aforementioned;they also take into account the availability of com-mercial sets of gels and buffers.

Common features can be recognized in bothexamples. First of all, the stability of the speciesis a point that cannot be stressed enough and must

be checked by an independent method. Since thegels may trap oxygen during the polymerisation,if the species are sensitive to oxidation, reducingagents can be used such as thioglycolate [8].

Secondly, the proteins must be brought intosolution, in a sample buffer compatible withthe separation method. The quantity of proteinnecessary for an optimum detection of the formeris given by rules of thumb: in fact, this quantityprimarily depends upon the detection method,


silver-staining, rubies-staining, coomassie blue or14C. For example, a sensitive method such assilver-staining requires around 10 mg mL−1 as totalconcentration of proteins; for the visualisation of aspecific protein, circa 10 ng protein are necessary,a quantity that is dependent not only on the proteinitself, but also on the separation method. Indeed,in a one-dimensional experiment, the protein isspread over a whole band in comparison with aspot from 2DE, concentrated on 1 mm2, thereforerequiring less protein. One must also take intoaccount the quantity of trace element in thesample and the detection of the metal. However,a limiting factor is that gel electrophoresis hasalso a maximum loading capacity above which noseparation occurs. A compromise has to be foundbetween the two parameters. If a high amount ofprotein has to be used to detect the trace element,the use of coomassie blue for the proteins can beconsidered, which is about 50 times less sensitivethan silver-staining; should the latter be used, therisk is high that the whole gel would be darkenedby the proteins.

2.5.1 Nondenaturing electrophoresis

Although one-dimensional electrophoresis doesnot imply that the method be nondenaturing, sucha combination has been chosen for simplicity. Theconverse is true, nondenaturing electrophoresisbeing mostly one-dimensional up to this date.

The choice of the buffer system is the first step.A set of buffers optimum for the separation ofproteins with acidic pI (below circa pI 8), andwidely used, is a slight modification of the setdescribed by Laemmli [9]:

• electrode buffer, glycine/Tris base, pH 8.3;• gel buffer, Tris base/HCl, pH 6.4;• polarity, separation towards anode, the sample

is applied at the cathode;• 100 mL of each solution is sufficient. Buffers

strips and gel are rehydrated with ade-quate solution.

This set belongs to the class of discontinuousbuffers, with Tris as the common ion, chloride theleading ion and glycine the trailing ion.

For other separations, such as separations ofbasic proteins (pI above 8), other sets havebeen optimised and can be found in authoritativereviews [7] or books [3].

The second step is the preparation of the sample.The sample, proteins in our example, must bedissolved in a buffer compatible with the method,they must be kept in solution and, of course, thespecies must be stable. In order to be compatiblewith the electrophoresis procedure, the sampleshould not contain too much salt. In fact, a samplewith too high an ionic strength is one of the maincauses of failure of a separation, since a highcharge concentration causes a drop in the resistanceof the solution. Indeed small ions migrate moreeasily and in such a case the proteins stay at theirpoint of application. A further requirement is agood solvent that allows a smooth penetration ofthe sample in the gel. This step may be critical,especially for large or hydrophobic molecules, forwhich it is difficult to go from a free solution to asolution in a gel. Therefore, an optimal sample isthe combination of the following:

• the desalted original sample;• diluted in the gel buffer, which is the first

solution with which the proteins will be incontact in the gel, with 10 % glycerol, to mimicthe gel concentration and therefore facilitate thepenetration in the gel;

• mild detergents, useful to keep the proteins insolution and prevent aggregation, but there is areal danger that the proteins could be denaturedor the species degraded.

Parallel to the choice of the buffer, the problemof the stability of the compounds has to be tackled.As already mentioned, the choice of a buffersystem is the key not only to a good separationof proteins but also to speciation. In order to applythe separation method to speciation, it is advisableto first test the stability of the compounds to beseparated in the buffers. For example, if the speciesvanadium–protein has to be separated in a givenbuffer, experiments ought to be performed withthe species in the buffer to check whether freemetal is produced, i.e., whether the equilibriumbetween free vanadium and complexed vanadium


Gel

Cathode

Anode+

−

mig

ratio

n

sample troughscirca 10 µL

Anodebuffer

Cathodebuffer

Figure 4.4.2. Typical set for flatbed separation.

is disturbed. Such experiments can be ultrafiltrationor size-exclusion chromatography, any method thatis able to separate the metal–complexes from thefree element [39].

An example of incompatibility of a buffer setand a trace element is vanadium. Indeed, glycinecan complex vanadium and strips proteins fromthe trace element to which it was bound [10].Therefore, the use of buffers other than glycineor tricine has to be explored.

• Equipment: no equipment is standard for 1Delectrophoresis. There is a large choice betweenflatbed, submarine or vertical systems, each ofwhich has advantages over the other, but noneinfluencing speciation.

• Gel: various gel sizes can be used, 5 cm ×5 cm or 25 cm × 10 cm gels, most of whichare commercially available. If a specific gel isneeded, preparation is possible in the laboratory,at low cost; once again, one should refer to theappropriate literature. Another possibility is touse a rehydrated gel, wash it thoroughly, dryand rehydrate it in the necessary buffer.

The buffers are laid directly under the respec-tive electrode. The sample is brought on thegel either:

• directly, if the sample volume is small (1 µL);• onto a sample strip (1–10 µL);• in a sample trough (up to 15 µL) if these were

foreseen during polymerisation.

See Figure 4.4.2 for a typical set with aflatbed system.

The separation programme is also dependenton the apparatus but common features are alwaysidentifiable. A low voltage (ca. 25 V cm−1) isapplied in the first step to let the proteins enterthe gel material, i.e. the stacking gel. Afterwardsthe real separation begins.

For example, for a homogeneous gel, 25 cm(approximately 25 lanes) on 11 cm (separationlength), stacking gel T = 5 % (thus wide pores forthe stacking effect), C = 3 % (33 mm), resolvinggel T = 10 %, C = 2 % (77 mm), the programmecan be written as [2]:

Voltage Current Power Duration(V) (mA) (W) (min)

1st step 500 10 10 102nd step 1200 28 28 50

The separation is stopped when the dye, indicatingthe front line, is at the anode. The subsequent stepsare the detection of the trace elements and thevisualisation of the proteins.

2.5.2 Two-dimensional gelelectrophoresis (2DE)

Two-dimensional electrophoresis is the latest deve-lopment of gel electrophoresis, as evident from the


exponentially increasing number of articles aboutthis method. As already mentioned, the reason forthis success is due to proteomics, since a gel canmap nearly all proteins present in a sample. Forspeciation analysis, 2DE is only applicable if themetal under study is covalently bound, since themethod is denaturing in its most refined form, i.e.high resolution. Nondenaturing two-dimensionalelectrophoresis has been described [11], but isbased on different separation principles from thosementioned here and little has been publisheduntil now. That is why this part concentrates ondenaturing electrophoresis.

2DE unites two separation mechanisms whichexist independently from one another: isoelectricfocusing (IEF) and sodium dodecyl sulfate – poly-acrylamide gel electrophoresis (SDS-PAGE). It isthus a separation according to pI in the first dimen-sion and size in the second. In most of the con-temporary publications, IEF is used as the firstdimension and SDS-PAGE as the second dimen-sion. Both methods can be used independentlyand what is mentioned here is true for bothSDS-PAGE one-dimensional electrophoresis andIEF one-dimensional electrophoresis, with slightmodifications for the latter. One should refer tobooks [2, 3, 12] or articles [6, 13] for a practicaland state-of-the-art description of the method.

2.5.2.1 Isoelectric focusing (IEF)

This separation procedure relies on one of themajor characteristics of proteins, their isoelectricpoint. Indeed, the charge of a protein is pHdependent, and at a characteristic pH this netcharge is zero. It is possible to polymerise a gelwith a pH gradient, termed an immobilized pHgradient (IPG), or to use a chemically created pHgradient, formed by free carrier ampholytes. Theyare both commercially available. The choice ofthe form of gradient is sometimes important forthe quality of the separation. To begin with IEFseparations, especially as the first dimension of2DE, it is generally more secure to use the IPGtechnology, where the IPG strips, usually about5 mm wide and 5–20 cm long, are stored in adehydrated form.

If the acidic part of the gel is pointed towardsthe anode (+), a protein initially at the anode,which is positively charged below its pI, migratestowards (−) and thus towards its pI. At the pI,the charge is zero and the field does not influencethe particle any more. This reasoning is converselyvalid for a protein initially present at the cathode.Should the protein diffuse, below its pI, the chargeis positive and it is repelled by the anode (+), whileabove it the charge is negative and it is repelledby the cathode (−). That is why the separation isusually named focusing, stressing the fact that theprotein comes to a definite spatial point in the gelby a ping-pong mechanism.

Although not a requirement for IEF, it is betterto denature the proteins at this stage, firstly tobe compatible with the second dimension andsecondly to bring hydrophobic proteins in solution.The quantity of protein is also a determiningfactor for detection and must be adjusted tothe detection method, as already mentioned, butshould not exceed 200 µg in a narrow strip, about20 cm long. This solubilisation takes place in amixture, hereafter referred to as sample solution,containing [13]:

• urea, a chaotropic agent, used to break thehydrogen bonds in and between proteins and tounfold them;

• a nonionic detergent, such as CHAPS, to bringthe proteins in solutions without contributing tothe ionic strength of the solution;

• a reducing agent, DTT, to break the disulfidebonds in proteins;

• possibly a protease inhibitor, depending on thesample, in order to prevent any proteolysis.

Once again, it must be checked whether the sampleis stable in this solution. The same strategy asthat mentioned in Section 2.5.1 is recommended.In particular, since the separation occurs in a pHgradient, control of the stability of the species isnecessary at the extreme pHs, usually 3 and 11.

For our example, IPG, the strip must berehydrated for at least 10 h, either in the samplesolution or in a solution containing the samechemicals, except the analytes, the sample beingadded at the end of rehydration. After rehydration,


a low voltage (in our example 300 V) is applied for1 h to force the proteins into the gel and afterwardsa high voltage is used to achieve the focusing (upto 8000 V, if it can be reached) for at least 6 h.

Between the two dimensions an equilibrationstep is necessary to change the solution in whichthe proteins will be separated in the seconddimension, namely SDS.

2.5.2.2 Sodium dodecyl sulfate – polyacrylamidegel electrophoresis (SDS-PAGE)

The separation principle is based on the constantaffinity of SDS for proteins, since 1.4 g SDS binds1 g protein. From then on, the intrinsic charge ofthe protein is masked by the charge of SDS; thusthere is a constant charge per gram of protein.Instead of a separation according to mass andcharge, charge is here related to mass. This methodhas other advantages, when it can be applied tospeciation, such as an enhanced solubility of theproteins or a real random coil of the chain, whichallow an easier mass estimation of the particle.

For SDS-PAGE, the choice is once again leftopen between gradient or homogeneous; a gradientgel may be easier to begin with, since the massrange is broader, especially when little is knownabout the sample. Most of the time a stacking effectis used to enhance the resolution in the seconddimension. The last point can be recognised in thesetup of the buffers [9]:

• cathode buffer, glycine/Tris, 10 g L−1 SDS;• gel buffer, Tris/HCl, 10 g L−1 SDS;• anode buffer, Tris/HCl;• glycine is the trailing and chlorine the lead-

ing ion.

The IPG strip is laid parallel and next to thecathode (−). Since the micelles are negativelycharged, they migrate towards the anode.

The voltage programme is comparable withthe others already mentioned, for a typical gel(25 cm × 11 cm):

(i) 200 V (50 mA and 30 W maximum). Thevoltage is low for an optimum sample entry,until the sample front is 5 mm away from theIPG strip.

(ii) The strip must be then removed and thecathode buffer put in on its place to avoiddehydration of the PA gel.

(iii) 600 V (50 mA and 30 W maximum). Theseparation is stopped when bromophenol bluehas reached the anode.

Afterwards, the gel is ready for the detection ofthe trace element and the proteins.

2.5.2.3 Example: 2DE of selenised yeast

The aforementioned method was applied to theseparation of an extract of yeast, enriched inselenium [14]. The radiotracer 75Se was used toallow detection by means of the phosphor screentechnology. A key point of the separation isthe protection of the selenoamino acids againstoxidation, by a chemical derivatisation. Withoutthis precaution, the species are not stable duringelectrophoresis.

After separation, the proteins were fixed as men-tioned later and the selenium-containing proteinsdetected with a phosphor screen for 1 week. Afterdetection of the trace element, the gel was silverstained and the two pictures, autoradiogram andsilver staining, can be compared (see Figure 4.4.3and Figure 4.4.4 respectively).

Two practical points are worth mentioning. Firstof all, all gels and material are commerciallyavailable, which reduces the amount of work.However, if necessary, all these gels can be readily

cathode

anode

a b

Figure 4.4.3. Autoradiogram of a 2DE gel of yeast enrichedwith 75Se. a and b are two different spots, materialisedto ease the comparison between silver staining and autora-diography. Reprinted from Fresenius’ Journal of Analyti-cal Chemistry, Two-dimensional gel electrophoresis, C. C.Chery et al., Vol. 371, pp. 775–781, 2001, copyright noticeof Springer-Verlag [14].

DETECTION OF TRACE ELEMENTS 235

Figure 4.4.4. Silver staining of the gel used in Figure 4.4.3. Spots a and b are materialised for comparison with the autoradiogram.Note that the spots c do not appear on the autoradiogram, which proves that some proteins do not contain selenium. Reprintedfrom Fresenius’ Journal of Analytical Chemistry, Two-dimensional gel electrophoresis, C. C. Chery et al., Vol. 371, pp. 775–781,2001, copyright notice of Springer-Verlag [14].

made in the laboratory at low cost. Secondly, thewhole separation requires less than 2 h of work,although it lasts 20 h in total. The first dimensionis not labour intensive. Once the IPG is prepared,the separation occurs without external intervention(15 h); this can be done overnight. Preparation forthe second dimension requires about 1 h and therun, although best supervised from time to time,does not require further intervention.

3 DETECTION OF TRACEELEMENTS

3.1 General

Once again, the crucial question arises whether aspecies is stable during separation or detection.Indeed, a method widely used prior to the detectionof proteins is precipitation. By soaking the gelwith an acidic solution, the proteins precipitateand diffusion of the bands or spots is no longer aproblem. The gel is soaked in a solution of ethanol,acetic acid and water and left to dry. It is in factthe first step of detection by staining. If the gelmust be stored for a long period before analysis,glycerol can be added to this solution, otherwisethe gel may shrink. However, such a treatment isprohibited if the species are labile and the onlymethods left are either drying or freezing the gel.

A second choice to be made is to use eithera method that allows the detection of the trace

elements in the entire gel or a method that requiresthat the gel be cut in subsamples. The former ismore convenient but the latter allows quantitativeanalysis. Thus, the choice between the two setsof methods is dictated by strategy and need. Onehas to know which question is to be addressed:where is the trace element, or in which quantity isit present?

3.2 Detection of trace elementsin subsamples of the gel

Although this procedure is more time consuming,since an additional manipulation is required beforedetection, analysis of subsamples is more reliablefor quantitative analysis and more sensitive. Fur-thermore, they are sufficient when only a roughimage of the distribution of trace elements is neces-sary. From the analytical point of view, all methodsused for the detection of trace elements can infact be considered; however, for simplicity, onlythe methods whose applications have already beenpublished are presented here.

3.2.1 Liquid introduction system: inductivelycoupled plasma – mass spectrometry (ICP-MS),AAS and AES

In order to get a first image of the distributionof trace elements in gels, ICP-MS with a liquid


introduction system is a method of choice. Thegel must be excised and the pieces mineralisedby microwave assisted digestion. Each fractionis then quantified by ICP-MS. This method isparticularly tedious, but gives detection limits thatare excellent for most elements. Any other methodthat relies on dissolved subsamples is, however,applicable, such as AAS, AES, voltammetry [15]or fluorescence [16].

3.2.2 Solid sample analysis: electrothermalvaporisation (ETV) – ICP; graphitefurnace – atomic absorption spectrometry(GF-AAS)

These methods can be seen as improvements ofthe introduction system of the sample. Indeed,as solid sampling analysis, they do not requiredestruction of the pieces of gel and still offerexcellent detection limits. The sample is broughtin the oven, the organic material ashed and thetrace elements brought to the plasma in the caseof ETV [17, 36] or detected in the furnace forGF-AAS [18]. The disadvantages of the methodare that a precise optimisation of the temperatureprogramme and of the use of chemical modifiersare necessary and that the optimisation is valid foronly one element. Still, the detection is simplerthan with a liquid introduction system and can bequantitative if the system is adequately calibrated.

3.2.3 Nuclear analytical chemistry:scintillation counting, neutron activationanalysis (NAA)

Scintillation counting can be used with samplesradioactively labelled. According to the nuclearcharacteristics of the radiolabel, the method ofchoice will be either liquid scintillation counting orwell-type NaI(Tl) detection. The former requiresthat the subsample be solubilised, whereas thelatter is directly suitable for the solid piece of gel.Liquid scintillation has already been successfullyapplied to the detection of 75Se after PAGE [19].The advantage of the method is a higher sensitivitythan the phosphor screen technology, but at theexpense of a good resolution of the gel.

NAA [20–23] has been one of the first detec-tion methods used for trace elements in gels. Evenif this method implies a large investment anda heavy infrastructure, the method is interestingbecause it relies on a totally different principlefrom the spectrometric methods mentioned pre-viously. After the gel is sliced, the elements itcontains are activated in a nuclear reactor throughneutron bombardment. After a cooling time, theelements are detected by recording the wholegamma spectrum with a Ge(Li) detector and byassigning the peaks specific to each element. Quan-titation is once again possible. Theoretically, thismethod can be applied to a whole gel and anautoradiogram recorded (see later). However, thehigh neutron fluxes produce a high temperatureand a high radiation which damage the gel; fur-thermore, a high background is obtained since anyimpurity and traces of the reagents in the gel arealso activated.

3.3 Detection of trace elementsin a whole gel

The methods presented here rely on conceptsalready presented in Section 3.2, with the excep-tion of autoradiography, and are refinements of thesample introduction system.

3.3.1 Autoradiography

Autoradiography is the method of choice when thematerial submitted to electrophoresis is labelledwith a radiotracer. A whole and precise image ofthe distribution of radioactivity is obtained whenthe autoradiography screen is laid on the gel,recording a kind of photographic picture of theradioactive material. Furthermore, it is not limitedto gels, but can be applied to all thin materials,such as tissues.

Previously, X-ray films were used, but the con-temporary technology relies on phosphor screens[24], which allow a more rapid detection withhigher resolution. Further, a phosphor screen istheoretically infinitely reusable, as long as it hasnot been contaminated. After the signal has been

DETECTION OF TRACE ELEMENTS 237

read, the screen is erased by exposing it to visiblelight for a couple of minutes and is ready foruse again.

The phosphorescent material is a suspension ofBaFBr : Eu2+ crystals in a polymer. This crystal,when excited by radiation, sends an electron to theconduction bands, resulting in a different chemicalstructure, BaFBr− and Eu3+. When the excitedcrystal is exposed to light, typically from anHe–Ne laser (633 nm), this energy is enough todestabilise the excited electron, which falls backto its ground state, emitting a photon at 390 nm. Alaser is used because it allows an excellent spatialresolution. The luminescence is recorded with theposition of the laser and finally, when the wholescreen has been scanned, a precise image of theposition and intensity of the original radioactivityis obtained, in the form of a digitised picture withthe optical density as a function of the position.

Although both X-ray and phosphor screensprimarily allow the detection of 14C, 32P and35S, pure beta-emitters, they also may be appliedto other isotopes (75Se [14, 25–27], 63Ni [28],109Cd [29], 59Fe [30], 65Zn and 45Ca [31]). Theonly drawback is that the detection efficiency islower for some of these isotopes, depending on thenature of the radioactivity they emit. This meansthat a higher specific activity has to be used toget the same optical density as with the sameactivity of, e.g., 35S, or that a longer exposure timeis required.

From the last remark it is obvious that bothparameters for detection by autoradiography areof importance, i.e. specific activity and exposuretime. To get an idea of whether a sample can bedetected by this method, a simple test is required: adry gel can be rehydrated with a solution of knownactivity. By a simple weighing before and afterrehydration, the amount of activity is known inthe gel. The gel is left to dry, packed in plasticand exposed to the phosphor screen. Scanning ofthe gel will reveal if the original activity washigh enough.

Let us describe the practical use of a phos-phor screen. Just before use, the screen has tobe erased since a background builds up after awhile, even if the screen remains unused. After

separation, on the condition that the speciesremains stable with treatment, the proteins arefixed in the gel by an acidic solution and the gel isleft to dry. The staining should stop at this point,since a metal-based staining (e.g. silver staining)can quench the radiation emitted from the gel.In order to avoid contamination of the phosphorscreen by the tracers in the gel, the gel is wrappedin a plastic foil, for example Mylar, the thinnestpossible and only carbon-based to avoid loss ofradiation through absorption of the material. Aphosphor screen may also be stored at low temper-ature (less than −20 ◦C) while exposed to a gel,without apparent detriment to the quality of thepicture. Thus, radiography also allows the detec-tion of species that cannot be fixed in acid.

The phosphor screen is left for a perioddetermined as mentioned beforehand, typically1 day for 35S and 2 days for 75Se at similar specificactivities. For longer periods, laying the screen ina lead coffer can improve the signal to backgroundratio by lowering the natural surrounding activity.The screen is then read by laser densitometry,erased and stored. Standard softwares are availablefor the image treatment of the gel.

3.3.2 Laser ablation – inductively coupledplasma – mass spectrometry (LA-ICP-MS)

This method is a further step toward a reliabledirect introduction of the sample into the spec-trometer. Other combinations are plausible, suchas LA-AES, but the most powerful is presentedhere. LA-ICP-MS is a very promising techniquefor the detection of metals in gels after separation.

The detection method is itself an on-linehyphenation, between classical ICP-MS and alaser. The sample, the gel in other words, lies inan ablation cell; a spot is ablated by the laser andthe fumes from the ablation are brought from thecell to the plasma in a tube by a continuous gasflow, generally argon. ICP-MS then gives the ele-mental composition of the protein present at theablation site.

Various laser types are commercially available,from ArF to Nd : YAG lasers, with wavelengthsrespectively from 193 to 1064 nm. The ablation


crater ranges from the µm range up to 200 µm,and is thus ideal to achieve a high resolution inthe case of gel electrophoresis. Furthermore, thesample can be moved in two dimensions, as thelatest equipment is computer controlled, allowinga precise screening of the gel. This last point isvery advantageous for 2DE.

For a material such as polyacrylamide and theminimum protein spot size expected (about 200 nmin diameter), a low wavelength and very energeticbeam are not necessary; fractionation, whichmeans a different composition of the ablationcomposition from the original sample, is probablyno problem, allowing the use of widespread andaffordable lasers.

Up to now, this method has been used foronly few elements in gels (Co in serum [32], Pbin humic acids [33]) but can be extended to anyelement detected by ICP-MS [5].

3.3.3 Particle induced X-ray emission (PIXE)

This technique relies on the excitation of theelectron of the inner shell by a collimated beamof energetic charged particles, protons. Afterexcitation, the atoms emit characteristic X-rayspectra that allow their detection, mostly by anenergy dispersive analysis. However, this methodrequires a heavy investment since, e.g., a cyclotronis necessary to produce protons of a few MeV.

With an appropriate apparatus to translate thegel, the former can be scanned; the method hasalready been applied to one-dimensional gels,where a scanning in the direction of migration isthe easiest [34].

3.3.4 Mass spectrometry (MS)

The applicability of MS to gel electrophoresisis well known and widespread, especially inthe field of proteomics with matrix-assisted laserdesorption/ionisation (MALDI). The combinationof the two for speciation purposes has not yetbeen reported but no hindrance really exists, exceptfor the low specificity of the method for metals.MS may be thought of as a detection method

dedicated to organic molecules, in our case proteinidentification, but it has already been successfullyapplied to elemental speciation, for example to thespeciation of arsenic or selenium [35]. The samecan be applied with other elements, provided theygive a typical loss, for example as for Se, or arecognisable isotopic envelope. Two examples toexplain respectively the notions of typical loss andisotopic envelope: the loss of m/z 386 can onlybe attributed to Glu–Cys–(80Se)–Gly in ref. [35],and not to its sulfur analogue; if subpeaks can berecognised in a major peak and if these peaks areseparated by the same m/z as between isotopesof an element and with the same height ratio asin the natural abundance, the major peak can beattributed to a species containing this element.

4 CONCLUSION

Gel electrophoresis may seem tedious for elemen-tal speciation purposes but its figures of merit makeit worth giving it a try. To be applicable, one mustbe sure that the analytical data are representativeof the elemental species in the original sample.Thus, attention must be paid to the stability of thecompounds under investigation, even if this meansadditional tests.

The figures of merit for gel electrophoresis qual-ify this method as an integral part of speciation:high resolution, various separation mechanisms,the quantity of material after separation it yields,hyphenation with powerful tools such as MALDIand laser ablation – ICP-MS.

5 ACKNOWLEDGEMENTS

CCC is Research Assistant of the Fund for Scien-tific Research – Flanders (Belgium) (F.W.O.)

6 REFERENCES

1. Andrews, A. T., Electrophoresis. Theory, Techniques, andBiochemical and Clinical Applications , Oxford UniversityPress, New York, 1988.

2. Westermeier, R., Electrophoresis in Practice, VCH, Wein-heim, 1993.

REFERENCES 239

3. Hames, B. D. (Ed.), Gel Electrophoresis of Proteins ,Oxford University Press, Oxford, 1998.

4. Lustig, S., De Kimpe, J., Cornelis, R., Schramel, P. andMichalke, B., Electrophoresis , 20, 1627 (1999).

5. Chery, C. C., Gunther, D., Cornelis, R., Vanhaecke, F.and Moens, L., Electrophoresis, submitted (2003).

6. O’Farrell, P. H., J. Biol. Chem., 250, 4007 (1975).7. Chrambach, A. and Jovin, T. M., Electrophoresis , 4, 190

(1983).8. Bansal, M. P., Ip, C. and Medina, D., J. Soc. Exp. Biol.

Med., 196, 147 (1991).9. Laemmli, U. K., Nature, 227, 680 (1970).

10. Lustig, S., Lampaert, D., De Cremer, K., De Kimpe, J.,Cornelis, R. and Schramel, P., J. Anal. At. Spectrom., 14,1357 (1999).

11. Lustig, S., De Kimpe, J., Cornelis, R. and Schramel, P.,Fresenius’ J. Anal. Chem., 363, 484 (1999).

12. Rabilloud, T. (Ed.), Proteome Research: Two-Dimensio-nal Gel Electrophoresis and Identification Methods ,Springer, Berlin, 2000.

13. Gorg, A., Obermaier, C., Boguth, G., Harder, A., Sch-eibe, B., Wildgruber, R. and Weiss, W., Electrophoresis ,21, 1037 (2000).

14. Chery, C. C., Dumont, E., Cornelis, R. and Moens, L.,Fresenius’ J. Anal. Chem., 371, 775 (2001).

15. Messerschmidt, J., Alt, F. and Tolg, G., Electrophoresis ,16, 800 (1995).

16. Vezina, D., Belanger, R. and Bleau, G., Biol. Trace Elem.Res., 24, 153 (1990).

17. Wrobel, K., Gonzalez, E. B., Wrobel, K. and Sanz-Medel, A., Analyst , 120, 809 (1995).

18. Sidenius, U. and Gammelgaard, B., Fresenius’ J. Anal.Chem., 367, 96 (2000).

19. Qu, X. H., Huang, K. X., Wu, Z. X., Zhong, S., Deng,L. Q. and Xu, H. B., Biol. Trace Elem. Res., 77, 287(2000).

20. Stone, S. F., Zeisler, R., Gordon, G. E., Viscidi, R. P.and Cerny, E. H., ACS Symp. Ser., 445, 265 (1991).

21. Stone, S. F., Hancock, D. and Zeisler, R., J. Radioanal.Nucl. Chem., 112, 95 (1987).

22. Stone, S. F., Zeisler, R. and Gordon, G. E., Biol. TraceElem. Res., 26, 85 (1990).

23. Jayawickreme, C. K. and Chatt, A., J. Radioanal. Nucl.Ch., 124, 257 (1988).

24. Johnson, R. F., Pickett, S. C. and Barker, D. L., Elec-trophoresis , 11, 355 (1990).

25. Behne, D., Kyriakopoeulos, A., Weiss-Novak, C., Kalck-loesch, M., Westphal, C. and Gessner, H., Biol. TraceElem. Res., 55, 99 (1996).

26. Jamba, L., Nehru, B., Medina, D., Bansal, M. P. andSinha, R., Anticancer Res., 16, 1651 (1996).

27. Lecocq, R. E., Hepburn, A., Lamy, F., Anal. Biochem.,127, 293 (1982).

28. Nielsen, J. L., Poulsen, O. M. and Abieildtrup, A., Elec-trophoresis , 15, 666 (1994).

29. Scott, B. J. and Bradwell, A. R., Clin. Chim. Acta , 127,115 (1983).

30. Vyoral, D. and Petrak, J., Biochim. Biophys. Acta, 1403,179 (1998).

31. Scott, B. J. and Bradwell, A. R., Clin. Chem., 29, 629(1983).

32. Neilsen, J. L., Abildtrup, A., Christensen, J., Watson, P.,Cox, A. and McLeod, C. W., Spectrochim. Acta B , 53,339 (1998).

33. Evans, R. D. and Villeneuve, J., J. Anal. At. Spectrom.,15, 157 (2000).

34. Szokefalvi-Nagy, Z., Bagyinka, C., Demeter, I., Kovacs,K. L. and Quynh, L. H., Biol. Trace Elem. Res., 26, 93(1990).

35. McSheehy, S., Pohl, P., Szpunar, J., Potin-Gautier, M.and Łobinski, R., J. Anal. At. Spectrom., 16, 68 (2001).

36. Chery, C. C., Chassaigne, H., Verbeeck, L., Cornelis, R.,Vanhaecke, F. and Moens, L., J. Anal. At. Spectrom., 17,576 (2002).

37. Fan, T. W.-M., Pruszkowski, E. and Shuttleworth, S., J.Anal. At. Spectrom., 17, 1621 (2002).

38. Chen, C. Y., Zhao, J. J., Zhang, P. Q. and Chai, Z. F.,Anal. Bioanal. Chem., 372, 426 (2002).

39. Chery, C. C., De Cremer, K., Dumont, E., Cornelis, R.and Moens, L., Electrophoresis , 23, 3284 (2002).

40. Gao, Y. X., Chen, C. Y., Chai, Z. F., Zhao, J. J., Liu, J.,Zhang, P. Q., He, W. and Huang, Y. Y., Analyst , 127,1700 (2002).

CHAPTER 5

Detection

5.1 Atomic Absorption and Atomic EmissionSpectrometry

Xinrong Zhang and Chao ZhangTsinghua University, Beijing, China

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 2412 Flame and Hydride Generation AAS . . . . . 243

2.1 Technical developments in sampleintroduction . . . . . . . . . . . . . . . . . . . 243

2.2 Separation and preconcentration . . . . 2442.3 Chromatography coupled to flame

AAS . . . . . . . . . . . . . . . . . . . . . . . . 2462.4 Chromatography coupled to hydride

generation AAS . . . . . . . . . . . . . . . . 2473 Electrothermal AAS . . . . . . . . . . . . . . . . . 2514 Plasma AES . . . . . . . . . . . . . . . . . . . . . . . 253

4.1 ICP source . . . . . . . . . . . . . . . . . . . . 2534.1.1 Interface based on the

concentric and cross-flowpneumatic nebulizers . . . . . . . 253

4.1.2 Interface based on ultrasonicnebulizers . . . . . . . . . . . . . . . 255

4.1.3 Interface based onthermospray nebulizer . . . . . . 255

4.2 Other plasma sources . . . . . . . . . . . . 2555 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . 2576 References . . . . . . . . . . . . . . . . . . . . . . . . 257

1 INTRODUCTION

Speciation measurements of trace elements havebecome important because many studies haveshown that the determination of total amountsin a sample, without distinguishing between itschemical species, is no longer adequate [1–4].The main analytical challenges for trace elementspeciation are the very low concentration atwhich they often occur and the requirement foridentification of the chemical forms in whichthe element is present. Both requirements are farfrom satisfactorily met by most of the existing

commercial instrumentation available for inorganicand organic analysis [5, 6].

As indicated in ref. 6, elemental analysis hastraditionally aimed at complete analyte recoveryand high sensitivity, in order to measure the totalamount of a specific element contained in a sam-ple. Atomic absorption and emission spectrometrictechniques (AAS, AES and ICP-MS) have beendeveloped to achieve these goals. The method-ology developed for element determination aimsat complete matrix dissolution in order to opti-mize atomization and quantitation. With these tech-niques, however, little information can be gained


242 DETECTION

with respect to the chemical forms and structuresof the compounds in which the element is present.

Organic structure analysis, in contrast, hasalways been directed at molecular identificationof the analyte. The analytical efforts were there-fore focused on structural identification rather thanon recovery and sensitivity issues. The techniquesthat have been developed for the identificationof chemical structures include mass spectrome-try (MS), nuclear magnetic resonance spectroscopy(NMR), infrared/ultraviolet spectrometry (IR/UV),etc. Separation techniques, such as gas or liq-uid chromatography, could be coupled for pre-separation of the compounds before identification.As relatively high concentrations are usually avail-able for structure identification, sensitivity wouldnot be an important issue in organic analysis.

Speciation studies, however, require the detec-tion of trace level quantities of the elements inthe samples. At the same time, the measurementsshould provide information about the chemicalforms or structures in which the element is present.From this point of view, we can immediately seethe technical difficulties related to element specia-tion. Because of their low sensitivity, most populartechniques, (such as IR/UV, NMR and MS), whichhave been extensively and successfully used for theidentification of the forms and structures in organicanalysis, are usually no longer adequate for ele-ment speciation. Atomic spectrometric detectors,the powerful tools with high sensitivities for mostmetals, fail to provide information about the chem-ical forms in which an element is present. At thismoment, there is hardly any commercial instru-ment available yet for element speciation.

In laboratories this problem has been solvedby combining several analytical techniques. Mostsuccessful combinations result from the couplingof a separation technique and an element-selectivedetector [7–9]. In general, the detection limits ofthese hyphenated systems are strongly dependenton the selected detectors, although the detectionlimits of hyphenated techniques are inferior tothose of atomic spectrometric detectors alone. Thismay be due to the relatively small sample volumesthat can be introduced into the chromatographicsystem and the peak broadening that occurred

during separation, and also because of low sampleflow rates needed for compatibility with LC flows.

The element-selective detection techniques gen-erally used for speciation purposes include atomicabsorption spectrometry (AAS), atomic emissionspectrometry (AES), atomic fluorescence spec-trometry (AFS) and inductively coupled plasma –mass spectrometry (ICP-MS). In comparison withAES and ICP-MS, AAS is the more popular detec-tion method. It has been successfully coupled togas and liquid chromatography (GC/LC) in manylaboratories [5, 7]. As the eluent from GC/LC canbe easily introduced into flame AAS (FAAS), theinterface between chromatography and FAAS isvery simple. However, FAAS could not providethe necessary sensitivity for element speciation inmost cases. Hydride generation atomic absorptionspectrometry (HGAAS) coupled to chromatogra-phy is becoming a common technique for the spe-ciation of those elements that can form hydrides.The major merit of this hyphenated technique is itshigh sensitivity. In addition, the matrix effect canbe removed effectively before entering the AASdetector. Unfortunately, the technique is suitableonly for a limited number of elements, such asarsenic, mercury, selenium, lead, etc. Elements thatdo not produce hydrides cannot be determined byHGAAS. Electrothermal atomic absorption spec-trometry (ETAAS) combines high sensitivity withan extensive range of elements suitable for detec-tion. Unfortunately, the sequential nature of thedrying and ashing steps prior to atomization makesit difficult to couple onto the continuous flow of theHPLC effluent. Up to now, it remains a problem todesign an interface for on-line coupling of ETAASto a chromatographic system.

In comparison with AAS, plasma source AESoffers advantages of multielement operation, easycoupling to chromatography and acceptance ofthe continuous flow of LC eluent. The importantdisadvantages are associated with the sensitivityof the plasma to organic solvents and the overallinefficiency of the nebulizer system. The poortolerance of the plasma source for the commonmobile phase, such as ion-pair reagents, limitsthe application of this technique to the speciationmeasurements of the trace level of analytes.

FLAME AND HYDRIDE GENERATION AAS 243

The present chapter aims to review the develop-ment of AAS and AES as element-selective detec-tion methods in hyphenated techniques for elementspeciation. The important techniques that will bereviewed include flame AAS, HGAAS, ETAASand ICP-AES. The important coupling techniquescovered in the chapter will include chromatog-raphy and capillary electrophoresis. Consideringthat flow injection analysis (FIA) is easily cou-pled with AAS and AES, and that it could beused to differentiate between redox species such asCr(III)/Cr(VI), As(III)/As(V) and to perform on-line preconcentrations and separations, this tech-nique is concisely described in the present chapter.

2 FLAME AND HYDRIDEGENERATION AAS

Flame AAS is one of the most simple, cheap andreliable detectors that has been extensively cou-pled to chromatographic and other separating tech-niques for element speciation. The poor detectionlimit for trace elements remains the main problemwith this method. This can be partly ascribed tothe analyte transport inefficiency of the pneumaticnebulizer for sample introduction into the flameatomization system. The flame AAS nebulizer hasless than 5 % efficiency for sample introduction.Many efforts have therefore been undertaken toimprove its efficiency. Another way of improv-ing the detection limit in elemental speciation isto develop preconcentration techniques, includingcolumn absorption and atomic-trapping, that haveinterfaced well with flame AAS for elemental spe-ciation in recent years [10–12].

2.1 Technical developmentsin sample introduction

In order to increase the sample introduction effi-ciency, the aerosol chamber from a glass con-centric nebulizer system originally developed forICP, has been adapted for a nebulizer interfacefor FAAS [13]. This resulted in an almost 100 %analyte transport efficiency due to the improved

flow characteristics and efficient desolvation [14,15]. With this design the detector signal was tentimes better than that obtained using a conven-tional nebulizer.

The use of thermospray (TS) in FAAS is anothermethod of increasing sensitivity, because it greatlyimproves the efficiency of the sample introduction.A common design of TS apparatus includes a cap-illary through which the solution is pumped byan HPLC pump. The capillary is heated to a tem-perature of 100–200 ◦C either by thermal contactwith a heated block or by passing an electric cur-rent through the capillary itself [16]. Maintaining aconstant temperature is necessary to minimize dif-ferences in nebulization conditions with changes insample flow rate or composition. As the solutionto be nebulized progresses through the capillary,it first boils and then forms droplets. The advan-tage of TS is that the droplets are smaller thanthose produced by a pneumatic nebulizer, givingincreased signals and lower detection limits. Theuse of TS has greatly improved the sensitivityof hyphenated techniques for element speciation.Chang and Robinson have developed a TS appa-ratus for interfacing HPLC and FAAS [17]. Theinstrument was used for cadmium speciation stud-ies in urine. Their experimental data showed that a75 µm orifice and 0.05 cm i.d. capillary producedsensitivity much higher than that using a commer-cial flame atomizer. The desolvation mechanismof TS was also studied by the same authors [18].They believe that, with further modification of theTS design, even higher analytical sensitivity forultratrace metal analysis and better compatibilityfor interfacing HPLC with FAAS and ICP-AEScan be expected. The main limitation of couplingTS with FAAS is that the system tends to clogbecause of the capillary commonly used in a TSapparatus. As a result, TS-FAAS is poorly tolerantof the introduction of a high salt solution.

Berndt and Yanez developed a hydraulic highpressure nebulization (HHPN) technique with hightemperature (300 ◦C) superheated liquids for sam-ple introduction [19]. The liquid sample was neb-ulized providing aerosol yields of up to 90 % inflame AAS. This new nebulization method com-bines the advantages of HHPN and thermospray

244 DETECTION

techniques (very small aerosol droplets, highaerosol yield, nebulization of saturated salt solu-tions) and could be easily interfaced between chro-matographic and spectrometric setups.

In addition to the design improvements, variousquartz adapters have also been tested to increasethe sensitivity of FAAS detection [8]. Improve-ment factors of 2.0 to 3.5 have been observedfor the different elements using the conventionaladapter with an exit slot at 180◦ to the entranceslot. Using the adapter with slots at 120◦ to eachother, yielded improvement factors of 4 to 5. Ithas been found that the sensitivity can be addition-ally enhanced by shifting the conventional quartzadapter with slots at 180◦ by 10–15◦. Thus a 4–6times lower characteristic concentration for arsenichas been obtained. Obviously, the S-type convec-tion flow through the slots ensures more stabletemperature conditions in the quartz tube. How-ever, it should be noted that slotted quartz tubes arevulnerable to vitrification in the presence of alkalimetal ions in buffers; therefore they are pretreatedwith La(III) for improving long-term performance.Following a suitable separation scheme the con-centration of different species has been determined,e.g. As(III) and As(V) in wastewater and Tl(I) andTl(III) in soil extracts [20, 21].

2.2 Separation and preconcentration

The chromatographic separation of species is veryoften associated with their preconcentration. Forinstance, after extraction and chromatographic sep-aration of As(III) and As(V) in a water sample, thearsenic concentration in the eluent is 20 to 25 timeshigher than initially in the samples. After such a

separation/preconcentration step researchers oftenuse simple, cheap and reliable flame AAS meth-ods for the detection of the concentrated species[20, 22–25]. The on-line scheme has been exten-sively studied by combination of flow-injectionseparation/preconcentration and flame AAS detec-tion. Fe(II) and Fe(III) could be well separated byusing a C18-modified silica column combined withFIA flame AAS. Fe(III) was passed straight to theAAS detector whereas Fe(II) was trapped as Fe(II)-ferrozine and then eluted with methanol [23]. Asimilar procedure was used for the speciation ofCr(III)/Cr(VI) by inserting a microcolumn packedwith acidic alumina [24] or Se(IV)/Se(VI) by ananionic exchange column [25]. The detection lim-its are 0.6 ng for Se(VI) and 1 ng for selenite,respectively, using flame AAS detection.

Speciation of inorganic and methylmercurycould also be carried out by coupling FIA precon-centration with flame AAS, although atomic fluo-rescence spectrometry offers higher sensitivity [5].Figure 5.1.1 shows a typical manifold for the spe-ciation of mercury species using FIA preconcentra-tion. A mixed solution of inorganic mercury andmethylmercury was flushed onto a microcolumnpacked with sulphydryl cotton. Inorganic mercurywas not retained and was reduced to elemen-tal Hg◦ with SnCl2 solution, whereas methylmer-cury was absorbed on the column. Although themethylmercury is present at very low trace lev-els in the original solution, the absorption pro-cedure allows a high degree of preconcentration,proportional to the volume of sample processedthrough the column. After recording the inorganicmercury peak, the elution of methylmercury occursupon acidification of the column using 0.1 mol L−1

Peristaltic pump

Detector

Sample

Column

Mixing coil

Waste

Gas/liquidseparator

Valve

0.1mol/L HCl

Br−-BrO3−

SnCl2

Figure 5.1.1. Schematic diagram of inorganic and methylmercury speciation by coupling FIA and AFS with an on-linepreseparation technique using a sulphydryl cotton microcolumn [5].


HCl solution. On-line bromination and reaction ofmethylmercury with SnCl2 generated Hg◦, whichwas measured by the detector.

Preconcentration based on atom trapping hasbeen developed in recent years to increase thesensitivity of certain elements in flame AASfor element speciation. In this technique, atomsare trapped on a cooled tube mounted abovea conventional spray chamber–burner assembly.After a fixed sample collection time (conventionalnebulization), the tube is rapidly heated to atemperature that causes atomization of the trappedanalyte. Sensitivity enhancement depends uponthe collection time and the analyte element.The technique works reasonably well for volatileelements such as arsenic, lead, cadmium and zinc.Matrix occlusion interferences limit this method tosamples with a low dissolved solid content.

The combination of on-line hydride genera-tion, atomic trapping and flame AAS has receivedconsiderable interest during the last decades. Thishyphenated technique was first used in 1975 for

the determination of methylated forms of sele-nium in freshwater environments [26]. Since then,it has been used for the speciation of arsenic, anti-mony, selenium, germanium, mercury and organictin species. Table 5.1.1 lists the derivatizing con-ditions for the speciation of these elements usingatomic trapping. The on-line integration of the dif-ferent steps allows this approach to be used for awide range of applications for samples such as air,water, sediments, and biological tissues, etc.

The introduction of cold trapping and chro-matographic separation steps to hydride genera-tion provides both high sensitivity and selectivityfor real samples. This on-line hyphenated systemprovides derivatization by hydride generation orethylation, preconcentration by cryotrapping, sep-aration by packed column gas chromatographyand detection by quartz furnace atomic absorptionspectrometry. In addition to the direct gain in sen-sitivity achieved by hydride formation and coldtrapping for elements such as As, Bi, Sb, Se, Sn,Ge and Te, the system provides the possibility of

Table 5.1.1. Derivatizing conditions for speciation using hyphenated techniques.

Species Reagent Derivatizingconditions

Samplepretreatment

Reference

As(III) NaBH4 2 mL 2 % aqueous NaBH4 1–3 mL of 5 % potassiumhydrogen phthalate pH 3.5–4

27

As(V) MMA DMATrimethylarsine

NaBH4 4 × 2 mL 2 % aqueous NaBH4 pH 1–1.5 with 5 mL of saturatedsolution (10 % w/v) of oxalicacid in water

27

Sb(III) Sb(V) NaBH4 1 % NaBH4 and 5.0 mol L−1

HCl solutionpH 4.0 with 50 mmol L−1 citrate

solution28

Se(IV) Se(VI) NaBH4 1.8870 g NaBH4 in 0.5 % (w/v)NaOH and diluting with250 mL of the same solution

with 25, 10 and 5 % (v/v)HCl : HBr solutions

29

Selenomethionine NaBH4 0.3 % NaBH4 in 0.2 % NaOH 30Ge MexGe(4−x)+ NaBH4 6 mL of 20 % NaBH4 in 0.06 M

NaOH per 100 mL of sample5 mL of 1.9 M Tris-HCl + 10 mL

of 300 gL−1 NaCl + 1 mL of0.2 M EDTA per 100 mL ofsample

31

MexPb(4−x)+ NaBEt4 3 mL of 0.43 % NaBEt4 inwater

pH 4.1 32

Hg2+ NaBH4 1 mL of 0.4 % aqueous NaBH4 pH 4 33MeHg+ LiB(C2H5)3H 0.1 % solution of LiB(C2H5)3H

in THFpH 4 33

TRISna MexSn(4−x)+ NaBH4 2 × 1 mL of 1 % aqueousNaBH4

pH 6.5 with 4 mL of 2 M Tris-HCl 34

TRISn MexSn(4−x)+ NaBH4 1 mL of 4 % NaBH4 in 0.02 MNaOH

pH 2 with 0.2 mL of 5 M HNO3 35

n-BuxSn(4−x)+Et3Sn+

NaBH4 2 × 2.5 mL of 6 % aqueousNaBH4

pH 1.6 with 2 mL of 5 M HNO3 36

aTRISn, total recoverable inorganic tin; THF, tetrahydrofuran.

246 DETECTION

redox speciation of inorganic species of As (IIIand V), Sb(III and V) and Se(IV and VI). It isalso very efficient for most low boiling alkylatedspecies such as the mono-, di-, and trimethylatedforms, monomethylarsonate (MMA) and dimethy-larsinate (DMA); dimethylselenide, dimethyldis-elenide and diethylselenide; methyl-, ethyl-, andespecially butyltin (including tetrabutyltin); inor-ganic, methyl-, dimethyl-, and diethylmercury; aswell as inorganic, methyl- and ethyllead species,etc. However, it should be noted that speciessuch as arsenobetaine, arsenocholine, arsenosugarsand selenomethionine or selenocysteine require awet digestion procedure because these compoundshave higher organometallic forms that are notdirectly amenable to gaseous derivatizing methods.

2.3 Chromatography coupledto flame AAS

Combining gas chromatography (GC) with atomicdetectors produces powerful performance instru-ments for speciation analysis. This hyphenatedtechnique has been extensively studied in recent

years. Its main advantage over liquid chromatog-raphy is that the gas stream emerging from GCcan be readily introduced to flame AAS so as toovercome the problems associated with the over-all insensitivity of the flame system. The nebu-lizer of a flame AAS apparatus is only around5–10 % efficient for liquid samples. If the gasemerging from the GC can be introduced bypass-ing the nebulizer, an increase in sensitivity can beachieved. The main disadvantages of GC coupledto flame AAS for element speciation are that manyof the organometallic compounds are nonvolatileand therefore have to be derivatized prior to gaschromatographic determination. The samples mustalso be thermally stable and not break down at theoven temperature used in GC. The transfer lineslinking the two instruments must also be heatedin order to prevent condensation of the analyte.There must be no dead volume or cold areas in thelines. It was therefore applicable only for a limitednumber of determinations. Typical examples canbe found in Table 5.1.2.

In comparison with GC coupled to flame AAS,coupling liquid chromatography to flame AAS is

Table 5.1.2. Element speciation using GC coupled to flame AAS.

Analyte Chromatography Detection limits (DL) Reference

TRISn MexSn(4−x)+ Chromosorb GAW-DMCS 45–60,3 % SP2100

DL 20–25 pg for Sn. Reagent: NaBH4 32

Organotin Cryogenic trapping GC Interferences were found to be a problemand the reasons why are discussed

37

Organotin Ethylation GC DL 2–4 ng/g for methyltin and butyltin 38Organotin Ethylation CT/GC DL MeSn3+ 135 pg 39Tributyltin Hydride generation GC DL 0.1–1 ng L−1 40MexPb(4−x)+ Chromosorb WAW-DMCS

80–100, 10 % SP2100DL 9–10 pg as Pb Reagent: NaBEt4 32

Me4Pb GC DL 13 pg as Me4Pb 41Me4Pb Me3EtPb Me2Et2Pb

MeEt3Pb Et4PbPacked column, 3 % OV-101 on

gas chromQDL 12–25 pg as Pb 42

As(III) As(V) MMA DMA Glass beads (40 mesh) DL 19–61 pg as As. Reagent: NaBH4 43Arsenite Arsenate MMAA

DMAA TMAOHydride generation, trapping, GC Absolute detection limits in the range of

0.1–0.5 ng for all compounds44

As(III) As(V) DMA MMATMAO

Hydride generation, GC DL 200–400 and 2–10 ng L−1 forinorganic (As(III), As(V)) andmethylated (DMA, MMA, TMAO)arsenic species, respectively

45

Hg2+ MeHg+ Ethylation, cryogenic trapping, GCChromosorb W-HP (60/80 meshsize, 10 % SP2100)

Potential artifacts and interferingcompounds were studied during thespeciation analysis

46

Methyl-, ethyl- andphenylmercury

Hydride generation, GC with aSupelco SPB-1 capillary column

DL 16 ng, 12 ng and 7 ng for methyl-,ethyl- and phenylmercury, respectively

47

MMA, monomethylarsonate; DMA, dimethylarsinate; TMAO, trimethylarsine oxide; TRISn, total recoverable inorganic tin; Me, methly; Et, ethyl.


more popular for element speciation because itoffers several advantages over GC. One of themajor advantages is that the analyte can be sep-arated at ambient temperature with no need forderivatizing. This not only shortens the samplethroughput time but it also reduces possible lossesduring the process. Also, there are more variableoperational parameters; both the stationary andthe mobile phase can be varied simultaneouslyto achieve better separation. A large variety ofstationary phases are available and ion-exchange,normal and reverse phases, as well as gel per-meation chromatography allow the separation ofions, organometallics of low and high molecularmass as well as metal–protein compounds. Theeluent from HPLC can easily be introduced intothe nebulizer for aerosol formation in order toevaporate in the flame for atomization. A typicalexample of HPLC coupled to FAAS for the speci-ation of methyl- and ethyltin compounds has beendescribed by the ref. 48: An ODS Spherisorb S5W column (250 mm × 3.0 mm i.d.) is used forthe separation of the species. The mobile phaseswere acetone/pentane (3 + 2) at 1.0 ml min−1 formethyltin compounds and acetone/pentane (7 + 3)at 1.2 ml min−1 for ethyltin compounds. The AASdetection was carried out with an N2O–C2H2

flame. The detection limits were 11–19 ng Sn for50 µL sample injection. Applications using liquidchromatography coupled to flame AAS are sum-marized in Table 5.1.3.

The difficulty in coupling HPLC to FAAS isthe balance of optimal flow rates between HPLCseparation and AAS detection. The common flowrate for HPLC is around 1–1.5 mL min−1 butthe flow rate for FAAS is much higher, causingstarvation. Although an additional solvent canbe introduced into the nebulizer at the end ofthe HPLC column [67], this leads to undesirablesample dilution. Another possibility is to attach aTeflon funnel [68] to the nebulizer or to introducea small T-piece [69] into the transfer line.

An important disadvantage in interfacing HPLCwith flame AAS is the efficiency of sampleintroduction; only around 5 % of the sample couldbe introduced into the flame for detection. Anotherdisadvantage of this interface is the dispersion,

not only in the HPLC column but also in theinterface tube and FAAS detector, which decreasesthe sensitivity and resolution of the HPLC. Thesensitivity could be greatly improved by usinghydride generation (HG) for sample introduction,although this is limited only to the elements thatcan form hydrides, such as As, Sb, Se, Sn and Pb.

2.4 Chromatography coupled to hydridegeneration AAS

Figure 5.1.2 [59] shows a diagram of the chemicalhydride generator. After HPLC separation, theeluent is introduced into the hydride generatorand mixed first with hydrochloric acid, then with1–5 % NaBH4 solution. The gaseous hydridesformed in the reaction coil are separated ina gas–liquid separator, introduced by inert gasflow into the heated quartz absorption cell anddetected by AAS.

The most popular use of HPLC-HGAAS hasbeen for the speciation of reducible arsenicspecies [59, 70]. Four reducible arsenic species,e.g. As(III), As(V), monomethylarsonate (MMA)and dimethylarsinate (DMA) could be separatedby using an anionic exchange column or a C18column with ion-pair reagent of tetrabutylammo-nium ion as the mobile phase. The gas–liquidmixture upon derivatization can be introduced intothe gas–liquid separator and the hydride can bedetected by AAS. With this system, improveddetection limits can be achieved through theremoval of matrix interferences. The system is suit-able for a wide range of samples, including urineand serum [71–73]. A similar set-up can be usedfor the speciation of antimony, tin, lead, selenium,etc. It could also be used for mercury, although Hgdoes not form a hydride. The eluting mercury com-pounds are converted into Hg◦ vapour by reductionwith SnCl2 and sweeping the vapour through aquartz-windowed cell aligned in the light path ofAAS. A summary of several applications based onHPLC-HGAAS is given in Table 5.1.3.

A thermochemical hydride generation inter-face has been developed for HPLC-AAS byBlais and coworkers. This on-line interface is

248 DETECTION

Table 5.1.3. Application of hyphenated systems using liquid chromatography coupled to atomic absorption spectrometry.

Analyte Chromatography Detector Comments Reference

As Dionex anion exchange500 mm × 3 mm

HG-FAAS Detection limit (DL)3–16 ng mL−1

49

Seven As species IC-HPLC FAAS Hydrogen–argon entrainedflame, with a slotted tubeatom trap for signalenhancement. Analysis ofaqueous extracts of soilsamples from a pollutedland site

50

Cd-metallothionein HPLC, elution with linear gradientof Tris buffer

AAS DL 5 µg g−1 51

Zn2+ and Cd-metallothionein

HPLC FAAS New design of thermospraynebulizer was assessed

52

Cr3+, Cr6+ FI, microcolumn packed withactivated alumina (acidic form)

FAAS Recovery of 90–106 % fornatural water

24

Fe2+, Fe3+ Flow injection (FI), C18-modifiedsilica column

FAAS Fe3+ passes straight todetector, whereas Fe2+ istrapped as Fe2+-ferrozineand then eluted withmethanol

23

Hg Develosil-ODS (30 µm)precolumn, and STR-ODS-H(5 µm) column

FAAS DL 0.1 ng 53

Pb µBondapak C18 column FAAS usingair–C2H2 flame

DL 10 ng 54

Pb species Chelex 100. Spheron oxin.Amberlite XAC-2, C18 andcellulose sorbents modified withphosphoric acid andcarboxymethyl groups

FAAS DL 0.17 µg L−1 Cellulosesorbents were found tohave the best retentioncharacteristics

55

Sn ODS Spherisorb S5W,250 mm × 3.0 mm i.d.

FAAS usingN2O–C2H2 flame

DL 11–19 ng 48

Arsenic species An anion exchange and a CAS1ion exchange column connectedin series

HGAAS DL 1.6 ng mL−1 for As(III),As(V), MMA andp-APAa and 1.9 ng mL−1

for DMA, AsB and AsC

56

Arsenic species inserum of uraemicpatients

Cation exchange liquidchromatography

UV photo/oxidationand HGAAS

DL 1.0, 1.3, 1.5 and1.4 µg L−1 of arsenic forMMA, DMA, AsB andAsC

57

AsB, AsC, DMA Cation exchange chromatographyusing a new solid-phase typebased on the CBC technology

UV/MW reactor tothe HGAAS

DL about 1 µg L−1 for eacharsenic species

58

Arsenic species inhuman serum

Reversed phase ion-pairchromatography, polymer-basedanion exchange chromatographyand silica-based anion exchangechromatography

HGAAS DL 0.49, 0.44, 0.92 and0.40 µg L−1 for As(III),As(V), MMA and DMAin serum, respectively.

59

Inorganic arsenic andselenium

Separation of ions by HPLC usinga phosphate buffer

HGAAS Contaminated ground watersamples

60

(a) Traceconcentrations ofAs, Cd, Pb and Se

(b) Various arsenicand seleniumcompounds

A reversed phase (C-18), ion-pair(tetrabutylammonium) HPLCprocedure for the separation offour arsenic species

HGAAS or ETAAS DL 0.004 µg L−1 for bothAs and Se

61

Inorganic and totalmercury

HPLC CVAAS Mussel tissue 62


Table 5.1.3. (continued)

Analyte Chromatography Detector Comments Reference

Methylmercury andinorganic mercury

A reversed phase C-18 column CVAAS Quantitative recovery forboth inorganic mercuryand methylmercury froma spiked natural watersample

63

Sb(III) and Sb(V) HPLC, Hamilton PRP-X100 HGAAS-ICP-MS The detection limits were 5and 0.6 ng per 100 µLsample for Sb(III) andSb(V)

64

Sb(III) and Sb(V) inwastewaters

Two HPLC columns of differentlengths (PRP-X 100,250 mm × 4.1 mm i.d. andPRP-X 100, 100 mm × 4.1 mmi.d.)

HGAAS DL 1.0 and 0.8 µg L−1 forSb(V) and Sb(III),respectively

28

Inorganic andorganic antimonycompounds

Dionex AS14 for the separation ofSb(V) and Sb(III); ION-120column TMSbCl2 and Sb(V)

FI-HGAAS Detection limits of 0.4, 0.7,and 1.0 µg L−1 forTMSbCl2, Sb(III), andSb(V)

65

Organic andinorganic seleniumspecies in urine

A vesicle mediated HPLC Microwave-HGAASor ICP-MS

DL ranged between 1.0 and5.3 µg L−1

66

ap-APA, p-aminophenylarsenate; TMSbCl2, trimethylantimony dichloride.

AAS

HPLC pump

Pump

Coil

Separator

HPLC Column

Valve

Waste

HCl

NaBH4

Ar

Sample

Recorder

Mobilephase

Figure 5.1.2. Diagram of HPLC-HGAAS system for As speciation.

based on thermospray nebulization of the HPLCmethanolic eluent, pyrolysis of the analyte in amethanol–oxygen flame, gas-phase thermochemi-cal hydride generation using excess hydrogen, andcool diffusion flame atomization of the product ina quartz cell mounted in the AAS optical beam.It has been used for the determination of arsenicspecies and selenonium compounds [74, 75]. Thelow cost, high reproducibility and relatively lowdetection limits of this system make it suitable forspeciation study.

Mineralization may sometimes be necessary fororganometallic compounds because some of thesespecies cannot produce hydrides under reducingconditions. On-line digestion, based on microwavedigestion or UV photolysis has been extensivelystudied in recent years. Examples include thespeciation of organic arsenic, tin and seleniumspecies [76]. These approaches allow highly sen-sitive determinations of analyte species that donot form volatile hydrides by hydride genera-tion. Detection limits down to 1.5–2.0 ng mL−1

250 DETECTION

can be achieved for organoarsenic compounds,such as arsenobetaine (AsB), arsenocholine (AsC),trimethylarsine oxide (TMAO), and tetramethylar-sonium ion (TMAs+) [77].

Although microwave-assisted digestion has gai-ned wide acceptance as a rapid method for sampledecomposition in speciation analysis [61, 78], itis not without problems. In-stream treatment witha mixture of oxidants at temperatures reaching200–300 ◦C to achieve complete dissolution, mayproduce a high pressure in the tubing with theoverheated solution. This is even more seriouswhen NaBH4 solution (to produce hydride) hasbeen mixed in. A cooling system is thereforenecessary after the microwave digestion, whichcauses post-column broadening in the coupledHPLC-AAS.

A simpler on-line digestion method for organo-metallic species has been developed by using alow power UV lamp as a source of reactionenergy [57, 79, 80]. Organometallic compounds,such as arsenobetaine (AsB) and arsenocholine(AsC), which cannot produce hydrides directly byreaction with NaBH4 in acidic medium, can be

well digested in 1.5 % K2S2O8 solution with theuse of a 6 W UV lamp. This procedure allows theHPLC-HGAAS determination of AsB and AsC atlevels of 1.5 and 1.4 ng mL−1 in serum of uraemicpatients [57].

The main problem with UV on-line diges-tion is that a longer coil and a lower flow-ratehave to be used to improve the efficiency. Bothcause a broadening of the chromatographic peaksand consequently poorer separation of the species.To improve the separation, an argon segment-flow technique was developed for arsenic spe-ciation using HPLC-HGAAS with UV-assisteddigestion [57]. After separation of the arsenicspecies, argon was injected into the moving car-rier stream of the mobile phase immediatelyafter the column. This prevents physical dis-persion of the analytes and controls the peakbroadening, independently of mobile phase flow-rate, manifold geometry, coil length and diame-ter. Figure 5.1.3 compares the chromatograms ofthe unsegmented and segment-flow technique. Thesegment-flow method displays a marked improve-ment in resolution.

0 4

1

2

3

4

5

6

8

(a)

12 0

1 263 − 5

4 8

Time/min

(b)

12

Figure 5.1.3. Segmented and unsegmented flow techniques (column: Dionex Ionpac CS 10, 4 mm × 250 mm; mobile phase:100 mmol L−1 HCl–50 mmol L−1 NaH2PO4). (a) segmented flow; (b) unsegmented flow. peak 1, MMA; peak 2, DMA; peak3, AsB; peak 4, TMAO; peak 5, AsC; peak 6, TMAs [57]. MMA, monomethylarsonate; DMA, dimethylarsinate; TMAO,trimethylarsine oxide; AsB, arsenobetaine; TMAs, tetramethylarsonium ion.

ELECTROTHERMAL AAS 251

3 ELECTROTHERMAL AAS

Electrothermal AAS, which first appeared on themarket about 1970, provides enhanced sensitivitybecause the entire sample is atomized in a shortperiod and the average residence time of the atomsin the optical path is 1 s or more. A few microlitresof sample are first evaporated at low temperatureand then ashed at a somewhat higher tempera-ture in an electrically heated graphite tube. Afterashing, the current is rapidly increased to severalhundred amperes, which causes the temperatureto soar to 2000–3000 ◦C. The atomization of thesample occurs in a few milliseconds to seconds.The absorption is measured in the region immedi-ately above the heated surface. These proceduresoffer the advantage of high sensitivity for smallamounts of sample. Unfortunately, the sequentialnature of drying and ashing prior to atomizationmake it difficult to couple to the continuous flowof a chromatographic eluent. Another obstacle isthe matrix effect of samples on ETAAS detection,which makes chemical modification of the matrixand background correction necessary. Until nowthe on-line coupling of chromatography to ETAASdetection has been rather limited compared to thevarious types of AAS. The need for sampling theeffluent before injection of discrete aliquots hasled to the development of two basic proceduresof coupling: (1) collection of the effluent in frac-tions and analysis of each fraction by ETAAS;(2) periodic sampling of the effluent and injectioninto the furnace.

Procedures based on fraction collection implytheir off-line analysis, as there is no physicalconnection between the chromatographic systemand the detector. Thus the chromatograms are notobtained in real time. In spite of this disadvantage,the procedures based on ETAAS detection aresimple and sensitive and have been extensivelyapplied to the speciation of Al [81–83], As [20, 84,85], Cd [86], Cr [87–89], Fe [90, 91], Se [92–94],Sn [95, 96], Sb [97–99], Zn [100], etc. in naturalwater, urine, blood, soil, sediment, and airborneparticles [101].

Some researchers have designed an automateddevice for effluent sampling in real time. These

designs allow a portion of the flowing eluent tobe introduced periodically into the furnace formeasurement. Two types of autosampler have beendesigned for this purpose: one is using a PTFEflow-through cell that is fixed into a cup of theautosampler of the AAS. Eluent from the columnflows from the bottom of the cell and is drainedby one channel of the peristaltic pump. Portionsof the flowing effluent are taken periodically bythe autosampler arm. The interface between thegraphite furnace and the chromatographic systemis shown in Figure 5.1.4(a) [102].

In a second approach, the effluent is stored asdiscrete fractions in the fraction collector. Aftercomplete elution of the fractions, they are injectedoff-line into the graphite atomizer. Figure 5.1.4(b)shows the interface between the graphite furnaceand the chromatographic system [103]. The latter

(a)

10−50 µl

LC

UVl

(b)

10−50 µl

100−500 µl

LC

UV

Splitl

Figure 5.1.4. Two views of the interface between the graphitefurnace and the chromatographic system: (a) flow-throughcell [102]; (b) off-line injection [103].

252 DETECTION

method yields more data points per chromato-graphic peak, resulting in a better signal to noiseratio. Both methods were demonstrated to be suit-able for the speciation of different elements. Chro-matographic peaks are depicted by several bars,each of them representing the integrated signalfor a certain collection period. These systems donot provide a continuous, on-line, real-time anal-ysis. The resolution depends on the number ofpulses detected.

An interface similar to thermospray (TS) neb-ulization in AAS has been proposed for HPLC-ETAAS by Nygren et al. [104]. They used a fusedsilica capillary heated through stainless steel tub-ing (200 ◦C) where the effluent is volatilized asaerosols. The aerosols are directly introduced intoa glass carbon graphite furnace for detection. Thisglassy carbon material proved sufficiently temper-ature resistant up to at least 2000 ◦C. When thegraphite furnace is operated at 2000 ◦C, a tem-perature gradient from 2000 ◦C at the furnace tothe temperature below 1400 ◦C at the narrow partof the glassy carbon tube is formed to protect thefused silica capillary. The method offers the advan-tage of on-line and real-time analysis, but seriousinterference is caused by the matrix of eluent andsamples since no drying and ashing occur prior toatomization. This technique has been used to deter-mine di- and tributyltin species with a detectionlimit of 0.5 ng.

A technique for element speciation with anETAAS detector based on ‘permanent modifica-tion’ has been developed in recent years [105].This is, in fact, a very promising development inchemical matrix modification in view of increas-ing sample throughput with ‘fast’ programmes,reducing reagent blanks, preliminary elimination ofunwanted modifier components, compatibility withon-line and in situ enrichment, etc. The techniquewas first studied by applying a single, manualinjection of 50 µg Pd and 50 µg Ir on the inte-grated L’vov platform of the transversely heatedgraphite atomizer THGA tube, allowing up to 300complete cycles of hydride trapping and atomiza-tion in hydride generation ETAAS determinationof As, Bi and Se [106]. The vapour or hydridewas introduced by the tip of the quartz capil-lary tube that was inserted automatically fromthe outlet of the gas–liquid separator at the cen-tre of the graphite tube. The temperature of thegraphite furnace during atomization was increasedto 2000 ◦C over seconds using maximum power.This technique allows one to apply the ‘fast’ pro-grammes, which could therefore be suitable forcoupling the separation technique to ETAAS withincreasing sample throughput. Table 5.1.4 showsexamples of permanent modification in ETAAS forelement speciation.

Despite the better overall sensitivity of ETAAS,coupling with chromatographic separation is still

Table 5.1.4. Examples of permanent modification in ETAAS for element speciation.

Analyte Matrix Modifier Comments Reference

Si Serum W-treated pyrocoated GT Anion exchange HPLC-ETAAS for speciation;600 µL fractions collected; 10 µL injections

107

Sn Fruits, vegetables Mg; Pd; NH4H2PO4; Ti- orZr-treated GT

Speciation of tricyclohexyltin hydroxide after CHCl3extraction; Zeeman STPF or Ti-treated tubes

108

As Aquatic plant,biological tissues,urine, water

Ir-treated GT On-line UV photo-oxidation or MWD withFI-HG-ETAAS for ‘first-order’ speciation; LOD0.14 ng

109

Se Soil Pd-treated GT GC–trapping–ETAAS for speciation of (CH3)2 Se,(C2H5)2 Se and (CH3)2Se2

110

Se Aq. Solution BN-coated GT vs. Pd-Mg(NO)2 modifier addition

Se(IV), Se(VI) and selenomethionine studied 111

Se Seawater Ir-treated platform Speciation protocol: Se(IV) determined withoutpretreatment; total Se after pre-reduction in5 mol L−1 HCl

112

Sn Water, sediments 110 µg of Zr or 240 µg ofW and 2 µg of Irpermanent modifier

FI-HG-ETAAS for Sn(IV), monomethyltin,dimethyltin, trimethyltin, diethyltin andmonobutyltin

113

PLASMA AES 253

troublesome. The interface is difficult, mainlyowing to the discontinuous nature of the ETAAS.So far, ETAAS has been applied to the speciationstudies for about 20 elements in various environ-mental matrices. Among them, Cr has been exten-sively studied in different environment samples.As, Se, Sb, Pb, Sn, Hg and Al are the elementsthat have been more or less widely analysed. Cu,Cd, Zn, Fe and Ni have been studied somewhatand little work has been done on Co, Si, V, Mg,Mo, and Tl.

4 PLASMA AES

Element speciation based on the plasma sourcehas the advantage of accepting the continuousflow of the HPLC eluent and is therefore eas-ier to hyphenate. Another advantage is its mul-tielement capacity: Plasma AES can detect bothmetals and nonmetals. In the literature, differ-ent plasma sources have been proposed for ele-ment speciation coupled with chromatographicseparation, e.g. microwave-induced plasma (MIP),direct current plasma (DCP) and inductively cou-pled plasma (ICP). The best one is still ICPbecause the continuous HPLC flow quenches thedischarge of the MIP source, causing difficul-ties on the interface. DCP appears to be bet-ter than MIP since it provides a more stableplasma, especially with the introduction of mixedorganic–aqueous eluents. However, with DCPdirect interfacing, the detection was limited to lev-els above 100 ng mL−1 [114], several times higherthan with an ICP source.

4.1 ICP source

The ICP source was first developed in the1960s and is becoming the most important sourcefor atomic spectrometry [115]. The argon-basedplasma is compatible with aqueous aerosols andoffers high energy for drying, dissociation, atom-ization and ionization of the analytes. The temper-atures reached by an argon ICP vary from 4500 to10 000 K, depending on the definition of ‘temper-ature’ (kinetic temperature, electron temperature,

atomization temperature, ionization temperature)and the location inside the plasma. An ICP is there-fore called a nonthermal equilibrium plasma. Thetemperature in the inner channel, used for analyti-cal purposes, is about 5500–6500 K, high enoughto destroy all molecular bonds and even to ionizemany elements.

The sample introduction system in the ICPsource is usually designed for liquids, but solids orgases are also possible. The nebulizer and the spraychamber constitute the most critical units of anICP-chromatography set-up for element speciation.The standard configuration of an ICP includes apneumatic nebulizer for the formation of aerosolsand a spray chamber for the separation of thedroplets by size. Only small droplets should beable to enter the plasma, otherwise it becomesunstable or extinguishes. Therefore, only 1–5 %of the sample reaches the plasma torch with thepneumatic nebulizer. The first requirement fora nebulizer is its compatibility with flow rateand eluent composition. Water-based eluents areoften deleterious to cones because of their saltcontent. Eluents containing organic modifiers ororganic eluents tend to affect plasma stabilitybecause of the increased solvent vapour pressure.Therefore, a frequent observation has been thatthere is poor tolerance of the ICP for the mobilephases commonly used in HPLC, particularlywith ion-pairing or size exclusion LC separation.Ways to overcome these problems are directlyrelated to improve rates of liquid consumptionand efficiency of nebulization of the analyteinto the plasma. Typical examples of nebulizersused in interfacing HPLC-ICP include cross-flowand concentric pneumatic nebulizers, ultrasonicnebulizers and thermospray nebulizers.

4.1.1 Interface based on the concentricand cross-flow pneumatic nebulizers

Both concentric and cross-flow nebulizers arewidely used in ICP-AES. For the concentricnebulizer, the analyte solution is fed through acapillary surrounded by a second capillary, whilethe nebulizer gas flows through the space betweenthem and produces the aerosol. The concentric

254 DETECTION

nebulizers are robust and allow high operationstability because of their monolithic construction,but they tend to clog, especially with low aspirationgas flow rates. Another kind of nebulizer, cross-flow nebulizer is designed by feeding the solutionthrough a vertically mounted capillary, nebulizedby the gas flow from a horizontal capillary whichends close to the tip of the former one [116].Although the cross-flow design is more tolerantof solutions with high salt contents, it is subjectto periodic blockage by salt deposits and to tipblockage as a result of salting out, caused by areduction in temperature.

Concentric and cross-flow nebulizers have beenextensively utilized for element speciation bycoupling HPLC to ICP-AES. For instance, amethod was developed for arsenic speciation inmarine organisms by coupling an anion exchangecolumn with ICP-AES. The eluate was passeddirectly to a concentric nebulizer and detected byICP-AES [117]. The difference between axial andradial viewing for the speciation of polar siliconcompounds using HPLC-ICP-AES with a cross-flow nebulizer has also been documented [118]. Itappeared that no improvement in detection limitwas achieved using the axial system. However,since the flow rate is around 1 mL min−1, thecommon design is not ideal for the purposebecause of the dead volume or the liquid and gasinjection areas. Therefore, coupling microcolumnHPLC and CE to ICP-AES by using a concentricnebulizer was developed by several authors. ACE-ICP interface using a modified concentricnebulizer has been described and applied to theseparation and correlation of metal species inmetallothioneins of rabbit liver [119]. By replacingthe central tube of the concentric nebulizer, amodified concentric nebulizer was developed withthe CE capillary as the interface to couple to theICP-AES. The detection limit of Cr, based on peakarea, is approximately 10 µg L−1 [120].

Another type of concentric nebulizer for thesimultaneous determination of both hydride form-ing and nonhydride forming elements by ICP-AEShas been described [121]. The large droplets froma concentric nebulizer are trapped and react with aborohydride solution pumped into a small hydride

generator fitted in the spray chamber. The hydridesthat are formed and the sample aerosol enter theplasma simultaneously, while the nonvolatile reac-tion products and surplus reductant flow to thedrain. This system provides a more than 20-foldimprovement in detection limits for the hydrideforming elements without degrading the perfor-mance for other elements.

Microconcentric nebulizers have attracted muchattention as alternative sample introduction devicesfor ICP-AES. The main difference with respect toconventional concentric nebulizers is a conspic-uous reduction in their critical dimensions. Thisreduction allows a more efficient gas–liquid inter-action at low liquid flow rate, thus improving thenebulizer performance in terms of better efficiencyof nebulization and transportation of sample. Themain advantage of microconcentric nebulizers isthat they can give rise to detection limits similarto or better than those obtained with conventionalnebulizers, with liquid flow rates 10–40 timeslower that result in increased transport efficiencyand/or enhanced signal stability [122]. A study offive nebulizers associated to three spray cham-bers in conjunction with HPLC-ICP-AES foundthat best signal-to-noise ratios were obtained byusing a microconcentric nebulizer and a cyclonespray chamber without affecting the chromato-graphic resolution. Response of the ICP to eachspecies of As(III), As(V), DMA and MMA wasstrongly affected by the selection of the nebulizerand spray chamber [123].

The direct injection nebulizer is a microcon-centric pneumatic nebulizer that allows directnebulization of typically 100 µL min−1 of ana-lyte solution into the plasma. The direct injec-tion nebulizer offers a 100 % analyte transportefficiency, reduced memory effect, rapid responsetime and good precision. The speciation of organicand inorganic selenium in a biological certifiedreference material, using microbore ion-exchangechromatography coupled to ICP-AES via a directinjection nebulizer has been reported. Separa-tion for selenomethionine, selenite, selenate andselenocystine can be obtained within 5 min [124].The speciation of chromium using HPLC-ICP-MSvia a direct injection nebulizer with an absolute

PLASMA AES 255

detection limit of 3 pg for both Cr(III) and Cr(VI)has been reported. However, the direct injectionnebulizer requires a much more expensive setupthan a common concentric pneumatic nebulizer.

4.1.2 Interface based on ultrasonic nebulizers

Sample introduction by utilizing ultrasonic nebu-lizers has become popular in recent years. In anultrasonic nebulizer, a longitudinal acoustic waveis produced by an oscillator coupled to a trans-ducer. The latter is oriented in such a way thatthe direction of wave propagation is perpendicularto the interface between the gas and the sampleliquid to be nebulized. Aerosol formation occursby the action of ‘geysers’ when the amplitude ofthe waves becomes sufficiently large [116]. Ultra-sonic nebulizers are not subject to clogging, andyield detection limits that are currently 10–50times better than those of pneumatic nebulizers,due to their high aerosol transport efficiencies(up to 30 %). Moreover, a system conjugating anultrasonic nebulizer with flow-injection ICP-AESyielded a sensitivity enhancement factor of 225 forthe speciation of vanadium(IV) and vanadium(V)in river water samples [125].

Applicability of an ultrasonic nebulizer cou-pled to different chromatographic techniques hasbeen explored by several authors. Ion exchangechromatography coupled to ICP-AES by using anultrasonic nebulizer was used for the speciation ofarsenic(V) and monomethylarsonate in the ng L−1

range [126]. Ultrasonic nebulization ICP-AES inconjunction with size exclusion chromatographywas also studied. It found that speciation of Ca,Cu, Fe, Mg, Mn and Zn in human milk couldbe applied to assess the concentration range andbinding pattern of the elements in the milk of 60lactating mothers [127].

4.1.3 Interface based on thermospray nebulizer

The primary reason to employ alternative methodsof nebulization such as thermospray as an interfacebetween liquid chromatography and ICP-AES isto achieve better sensitivity and better limits

of detection, since ICP-AES is not sensitiveenough for trace element speciation in manysamples, when compared to ICP-MS. Thermospraynebulization is accomplished by pumping a liquidsample at moderately high pressure through anelectrothermally heated capillary. The aerosoldroplet size produced by the thermospray is muchsmaller than those of pneumatic nebulization,which is favourable for desolvation, volatilizationand atomization, as mentioned in Section 2.1.

In a review of new developments in thermo-spray sample introduction for atomic spectrometryfrom 1992 to 1997 [128], HPLC thermospray ICP-AES was summarized for speciation of tin [129],chromium [130], Se [131] etc. The general princi-ples and operational characteristics of thermosprayhave also been discussed, together with a reviewof the applications of thermospray sample intro-duction with atomic spectrometry detection [132].Speciation of Cr(III) and Cr(VI) was carried out byusing thermospray sample introduction with ICP-AES and parameters such as control temperature,pH, and pump rate were also studied [133]. Thedirect speciation of selenite and selenate with ther-mospray sample introduction coupled to ICP-AEShas also been developed [134].

4.2 Other plasma sources

Although the majority of sample introductionsystems described in the literature are concernedwith ICP-AES detection, other plasma sourceshave also been exploited for element speciation.Particularly, low power MIP has been of interestas an excitation source, because it provides goodsensitivity for a number of elements and isinexpensive and easy to operate. Some limitationsof the MIP discharge have been established, i.e.low tolerance of the introduction of even a limitedamount of sample and instability when operatedat low power. Hence, applications on this topicare limited to a few papers, which describeseveral designs of interfaces for efficient liquidintroduction or special modifications of cavities inorder to increase the plasma stability [125–137].

To overcome the problems of low tolerancefor sample introduction and instability for plasma

256 DETECTION

caused by the direct introduction of liquid samples,cold vapour and hydride generation techniqueswere applied to the speciation in LC-MIP-AES.These techniques offer the possibility for volatileanalytes to enter the plasma without the inter-fering mobile phase. This was achieved by on-line coupling of liquid chromatography to lowpower argon microwave-induced plasma for thespeciation of mercury and arsenic compoundswith continuous cold vapour or hydride generationtechniques [138]. Signal enhancements of around100 % were found in micelles of cetyltrimethy-lammmonium bromide. The MIP generated atreduced pressure might have more resistanceagainst molecular gases such as hydrogen pro-duced in the hydride generator than atmosphericpressure plasmas [139]. For mercury, detectionlimits were found to be between 0.15 µg L−1 forinorganic Hg and 0.35 µg L−1 for methylmercury.For arsenic, the authors cite values between 1 and6 µg L−1.

In comparison with the hyphenation difficultiesof liquid samples to MIP-AES, GC is advantageousfor the coupling to MIP-AES. First, the analytescan be introduced quantitatively into the plasmain the gaseous form and no nebulization anddrying is necessary. Secondly, GC separations canuse helium as the carrier gas, which is ideally

suited to helium MIPs. The interface can be builteasily with a simple heated transfer line with lowdead volume. On the other hand, GC requires thederivatizing of analytes prior to analysis, as thenative species normally occur in the ionic state andlack the necessary volatility and thermal stability.Table 5.1.5 summarizes the typical examples ofGC-MIP-AES for element speciation.

Direct current plasma (DCP) appears to offercertain advantages with regard to chromatogra-phy coupling. This source involves a low voltage(10–50 V), high current (1–35 A) discharge, sta-bilized by the flowing inert gas, usually argon.The interface includes a quartz jet tube to con-vey the GC effluent directly into the DCP plume.The DCP remains stable in a high solvent back-ground and its design does not require a vent-ing valve, especially with the introduction of themixed organic–aqueous eluents. GC coupled toDCP emission spectroscopy has been used forthe determination of organotins in fish and shell-fish [147]. A similar device was employed for thedetermination of methylmercury [148]. However,with DCP direct interfacing, there is the prob-lem of high detection limits, which is the reasonfor the scarcity of publications on element specia-tion using DCP. Speciation analysis with differentplasma sources has been reviewed [149].

Table 5.1.5. GC-MIP-AES for element speciation.

Species Samples Comments Detection limit Reference

Organolead compounds Snow samples In situ propylation withsodium tetrapropylboratederivatization

0.15–0.21 ng kg−1 (As, Pb) 139

Organolead compounds Tapwater and peat In situ butylation withtetrabutylammoniumtetra-butylborate forderivatization

sub-ng L−1 range 140

Mercury Certified referencematerials

Purge-and-trapmulticapillary GC

0.01 pg mL−1 formethylmercury

141

Butyltin organomercurytetraalkyllead

Sediment and gasoline Multicapillary GC sub-pg L−1 range 142

Organolead compounds Gasoline Multicapillary GC <1 ng mL−1 143Mercury species Natural water Packed column GC for

large volume injections8 pg L−1 144

Mercury species Canal waters Sulphydryl cottonmicrocolumn forpreconcentration

10 ng L−1 for methyl- andethylmercury; 16 ng L−1

for inorganic mercury

145

Organotin compounds Sediment samples Microwave-assistedleaching

2 ng g−1 146

REFERENCES 257

5 CONCLUSION

AAS and AES have been coupled to different sep-aration techniques for elemental speciation. Thepotential of these hyphenated systems is stronglydependent on the selection of the detectors, theinterfaces and the design for sample introduction.It remains a problem to develop an interface foron-line coupling ETAAS to chromatographic sys-tems, although this type of detector offers highsensitivity with an extensive range of elements fordetection when compared with other detectors suchas flame AAS, HGAAS and ICP-AES. Commer-cial instruments for element speciation should alsobe developed, since home-made hyphenated instru-ments have severely restricted the application ofthese methods in routine environmental and clini-cal laboratories for element speciation.

6 REFERENCES

1. Cornelis, R., Ann. Clin. Lab. Sci., 26, 252 (1996).2. Cornelis, R., Crews, H., Donard, O., Ebdon, L., Pitts, L.

and Quevauviller, P., J. Environ. Monit., 3, 97 (2001).3. Bernhard, M., Brinckman, F. E. and Sadler, P. J., The

Importance of Chemical Speciation in EnvironmentalProcesses , Springer, Berlin, 1986.

4. Caroli, S., Element Speciation in Bioinorganic Chem-istry , John Wiley & Sons Inc., New York, 1996.

5. Donard, O. F. X. Hyphenated techniques applied tothe speciation of organometallic compounds in theenvironment, in Environmental Analysis, Applicationsand Quality Assurance, Barcelo, D. (Ed.), Elsevier, NewYork, Chapter 16, 1993.

6. Donard, Trace metal speciation: finally, correctlyaddressing trace metal issues, in the Report on the con-sultants’ meeting to: Outline of a Co-ordinated ResearchProject on Trace Element Speciation to Enhance Tech-nology Transfer in This Field by the Development ofSuitable Methodologies and/or the Production of Suit-ably Characterized Reference Materials , InternationalAtomic Energy Agency, Vienna, November 23–26,1998.

7. Ellis, L. A. and Roberts, D. J., J. Chromatogr. A, 774,3 (1997).

8. Havezov, I., Fresenius’ J. Anal. Chem., 355, 452 (1996).9. Cornelis, R., De Kimpe, J. and Zhang, X. R., Spec-

trochim. Acta, Part B , 53, 187 (1998).10. Matusiewicz, H., Spectrochim. Acta B , 52, 1711 (1997).11. Pasullean, B., Davidson, C. M. and Littlejohn, D., J.

Anal. At. Spectrom., 10, 241 (1995).

12. Garbos, S., Bulska, E. and Hulanicki, A., At. Spectrosc.,21, 128 (2000).

13. Gustavsson, A., Spectrochim. Acta, Part B , 42, 118(1997).

14. Nygren, O., Nilsson, C. A. and Gustavsson, A., Spec-trochim. Acta, Part B , 44, 589 (1989).

15. Gustavsson, A. and Nygren, O., Spectrochim. Acta, PartB , 42, 883 (1987).

16. Saverwijns, S., Zhang, X. R., Vanhaecke, F., Cor-nelis, R., Moens, L. and Dams, R., J. Anal. At. Spec-trom , 12, 1047 (1997).

17. Chang, P. P. and Robinson, J. W., Spectrosc. lett., 30,193 (1997).

18. Chang, P. P. and Robinson, J. W., Spectrosc. lett., 29,1469 (1996).

19. Berndt, H. and Yanez, J., Fresenius’ J. Anal. Chem.,355, 555 (1996).

20. Russeva, E., Havezov, I. and Detcheva, A., Fresenius’J. Anal. Chem., 347, 320 (1993).

21. Tsakovsky, S., Ivanova, E. and Havezov, I., Talanta ,41, 721 (1994).

22. Naghmush, A. M., Pyrzynska, K. and Trojanowicz, M.,Talanta , 42, 851 (1995).

23. Krekler, S., Frenzel, W. and Schulze, G., Anal. Chim.Acta , 296, 115 (1994).

24. Sperling, M., Xu, S. K. and Welz, B., Anal. Chem., 64,3101 (1992).

25. Kolbl, G., Kalcher, K. and Irgolic, K. J., Anal. Chim.Acta , 284, 301 (1993).

26. Chau, Y. K., Wong, P. T. S. and Goulden, P. D., Anal.Chem., 47, 2279 (1975).

27. Braman, R. S., Johnson, D. L., Craig, C., Foreback,C. C., Ammons, J. M. and Bricker, J. L., Anal. Chem.,49, 621 (1977).

28. Satiroglu, N., Bektas, S., Genc, O. and Hazer, H., Turk.J. Chem., 24, 371 (2000).

29. Gallignani, M., Valero, M., Brunetto, M. R., Burguera,J. L., Burguera, M. and Petit de Pena, Y., Talanta , 52,1015 (2000).

30. Chatterjee, A., Shibata, Y. and Morita, M., Microchem.J., 69, 179 (2001).

31. Hambrick, G. A., Freolich, P. N., Andreae, M. O. andLewis, B. L., Anal. Chem., 56, 421 (1984).


33. Craig, P. J., Mennie, D., Ostah, N., Donard, O. F. X.and Martin, F., Analyst , 117, 823 (1992).

34. Braman, R. S. and Tompkins, M. A., Anal. Chem., 51,12 (1979).

35. Andreae, M. O. and Byrd, J. T., Anal. Chim. Acta, 156,147 (1984).

36. Randall, L., Donard, O. F. X. and Weber, J. H., Anal.Chim. Acta , 184, 197 (1986).

37. Martin, F. M., Tseng, C. M., Belin, C., Quevauviller, P.and Donard, O. F. X., Anal. Chim. Acta, 286, 343(1994).

258 DETECTION

38. Shawky, S., Emons, H. and Durbeck, H. W., Anal.Commun., 33, 107 (1996).

39. Martin, F. M. and Donard, O. F. X., Fresenius’ J. Anal.Chem., 351, 230 (1995).

40. Ritsema, R., Mikrochim. Acta, 109, 61 (1992).41. Bergmann, K. and Neidhart, B., Fresenius’ J. Anal.

Chem., 356, 57 (1996).42. Radziuk, B., Thomassen, Y., Van Loon, J. C. and Chau,

Y. K., Anal. Chim. Acta, 105, 255 (1979).43. Howard, A. G. and Comber, S. D. W., Mikrochim. Acta,

109, 27 (1992).44. Molenat, N., Astruc, A., Holeman, M., Maury, G. and

Pinel, R., Analusis , 27, 795 (1999).45. Guerin, T., Molenat, N., Astruc, A. and Pinel, R., Appl,

Organomet. Chem., 14, 401 (2000).46. Tseng, C. M., De Diego, A., Wasserman, J. C., Amour-

oux, D. and Donard, O. F. X., Chemosphere, 39, 1119(1999).

47. He, B. and Jiang, G. B., Fresenius’ J. Anal. Chem., 365,615 (1999).

48. Burns, D. T., Glockling, F. and Harriot, M., Analyst ,106, 22 (1981).

49. Ricci, G. R., Shepard, L. S., Colovos, G. and Hes-ter, N. E., Anal. Chem., 53, 610 (1981).

50. Hansen, S. H., Larsen, E. H., Pritzl, G. and Cornett, C,J. Anal. At. Spectrom., 7, 629 (1992).

51. Pan, A. H., Wang, Z. and Ru, B., Biomed. Chromatogr.,5, 193 (1991).

52. Chang, P. P. and Robinson, T. W., J. Environ. Sci.Health, Part A, 28, 1147 (1993).

53. Munaf, E., Haraguchi, H., Ishii, D., Takeuchi, T. andGota, M., Anal. Chim. Acta, 235, 309 (1990).

54. Messman, J. D. and Rains, T. C., Anal. Chem., 53, 1632(1983).

55. Naghmush, A. M., Pyrzynska, L. and Trojanowicz, M.,Talanta , 42, 851 (1995).

56. Han, H. B., Liu, Y. B., Mou, S. F. and Ni, Z. M., J.Anal. At. Spectrom., 8, 1085 (1993).

57. Zhang, X. R., Cornelis, R., De Kimpe, J. and Mees, L.,Anal. Chim. Acta, 319, 177 (1996).

58. Sur, R., Begerow, J. and Dunemann, L., Fresenius’ J.Anal. Chem., 363, 526 (1999).

59. Zhang, X. R., Cornelis, R., De Kimpe, J. and Mees, L.,J. Anal. At. Spectrom., 11, 1075 (1996).

60. Rassler, M., Michalke, B., Schramel, P., Schulte-Host-ede, S. and Kettrup, A., Int. J. Environ. Anal. Chem., 72,195 (1998).

61. Tyson, J. F., J. Anal. At. Spectrom., 14, 169 (1999).62. Rio-Segade, S. and Bendicho, C., Ecotox. Environ.

Safe., 42, 245 (1999).63. Rio-Segade, S. and Bendicho, C., Talanta , 48, 477

(1999).64. Smichowski, P., Madrid, Y., Guntinas, M. B. D. and

Camara, C., J. Anal. At. Spectrom., 10, 815 (1995).

65. Krachler, M. and Emons, H., J. Anal. At. Spectrom., 15,281 (2000).

66. LaFuente, J. M. G., Dlaska, M., Sanchez, M. L. F. andSanz-Medel, A., J. Anal. At. Spectrom., 13, 423 (1998).

67. Yoza, N. and Ohashi, S., Anal. Lett., 6, 595 (1973).68. Slavin, W. and Schmidt, G. J., J. Chromatogr. Sci., 17,

610 (1979).69. Ebdon, L., Hill, S. J. and Jones, P., Analyst , 110, 515

(1985).70. Cornelis, R., Zhang, X. R., Mees, L., Christense, J. M.,

Byrialsen, K. and Dyrschel, C., Analyst , 123, 2883(1998).

71. Zhang, X. R., Cornelis, R., De Kimpe, J., Mees, L. andLameire, N., Clin. Chem., 44, 141 (1998).

72. Zhang, X. R., Cornelis, R., De Kimpe, J., Mees, L. andLameire, N., Clin. Chem., 43, 406 (1997).

73. Zhang, X. R., Cornelis, R. and Mees, L., J. Anal. At.Spectrom., 13, 205 (1998).

74. Blais, J. S., Momplaisir, G. M. and Marshall, W. D.,Anal. Chem., 62, 1161 (1990).

75. Blais, J. S., Huyghues-Despointes, A., Momplaisir,G. M. and Marshall, W. D., J. Anal. At. Spectrom., 6,225 (1991).

76. Howard, A. G., J. Anal. At. Spectrom., 12, 267 (1997).77. Ouyang, J., Shi, Y. and Zhang, X. R., Chin. J. Anal.

Chem., 27, 1151 (1999).78. Le, X. C., Cullen, W. R. and Reimer, K., Talanta , 41,

495 (1994).79. Howard, A. G. and Hunt, L. E., Anal. Chem., 65, 2995

(1993).80. Tsalev, D. L., Sperling, M. and Welz, B., Spectrochim.

Acta, Part B , 55, 339 (2000).81. Bantan, T., Milacic, R., Mitrovic, B. and Pihlar, B.,

Fresenius’ J. Anal. Chem., 365, 545 (1999).82. Bantan, T., Milacic, R., Mitrovic, B. and Pihlar, B., J.

Anal. At. Spectrom., 14, 1743 (1999).83. Wrobel, K., Gonzalez, E. B. and Sanz-Medel, A., Trace

Elem. Med., 10, 97 (1993).84. Van Cleuvenbergen, R. J. A., Van Mol, W. E. and

Adams, F. C., J. Anal. At. Spectrom., 3, 169 (1988).85. Burguera, M. and Burguera, J. L., J. Anal. At. Spec-

trom., 8, 229 (1993).86. Fernandez, F. M., Tudino, M. B. and Troccoli, O. E., J.

Anal. At. Spectrom., 15, 687 (2000).87. Kelko-Levai, A., Varga, I., Zih-Perenyi, K. and Lasz-

tity, A., Spectrochim. Acta, Part B , 54, 827 (1999).88. Vassileva, E., Hadjiivanov, K., Stoychev, T. and

Daiev, C., Analyst , 125, 693 (2000).89. Horvath, Z., Lasztity, A., Varga, I., Meszaros, E. and

Molnar, A., Talanta , 41, 1165 (1994).90. Bermejo, P., Pena, E., Dominguez, R., Bermejo, A.,

Fraga, J. M. and Cocho, J. A., Talanta , 50, 1211 (2000).91. Vanlandeghem, G. F., Dhaese, P. C., Lamberts, L. V.

and Debroe, M. E., Anal. Chem., 66, 216 (1994).

REFERENCES 259

92. Yan, X. P., Sperling, M. and Welz, B., Anal. Chem., 71,4353 (1999).

93. Robles, L. C., Feo, J. C., de Celis, B., Lumbreras, J. M.,Garcia-Olalla, C. and Aller, A. J., Talanta , 50, 307(1999).

94. Gammelgaard, B. and Larsen, E. H., Talanta , 47, 503(1998).

95. Astruc, A., Dauchy, X., Pannier, F., Potin-Gautier, M.and Astruc, M., Analusis , 22, 257 (1994).

96. Bermejo-Barrera, P., Soto-Ferreiro, R. M., Dominguez-Gonzalez, R. and Bermejo-Barrera, A., Ann. Chim., 86,495 (1996).

97. de la Calle Guntinas, M. B., Madrid, Y. and Camara, C.,J. Anal. At. Spectrom., 8, 745 (1993).

98. Garbos, S., Rzepecka, M., Bulska, E. and Hulanicki, A.,Spectrochim. Acta Part B , 54, 873 (1999).

99. Farkasovska, I., Zavadska, M. and Zemberyova, M.,Chem. Listy , 93, 173 (1999).

100. Guenther, K. and Waldner, H., Anal. Chim. Acta, 259,165 (1992).

101. Das, A. K. and Chakraborty, R., Fresenius’ J. Anal.Chem., 357, 1 (1997).

102. Laborda, F., Vicente, M., Mir, J. M. and Castillo, J. R.,Fresenius’ J. Anal. Chem., 357, 837 (1997).

103. Vickrey, T. M., Howell, H. E. and Paradise, M. T., Anal.Chem., 51, 1880 (1979).

104. Nygren, O., Nilsson, C. A. and Frech, W., Anal. Chem.,60, 2204 (1988).

105. Tsalev, D. L., Slaveykova, V. I., Lampugnani, L., D’Ul-ivo, A. and Georgieva, R., Spectrochim. Acta, Part B ,55, 473 (2000).

106. Shuttler, I., Feuerstein, M. and Schlemmer, G., J. Anal.At. Spectrom., 7, 1299 (1992).

107. Wrobel, K., Blanco Gonzalez, E., Wrobel, K. and Sanz-Medel, A., Analyst , 120, 809 (1995).

108. Giordano, R., Ciaralli, L., Gattorta, G., Ciprotti, S. andCostantini, S., Microchem. J., 49, 69 (1994).

109. Willie, S. N., Spectrochim. Acta Part B , 51, 1781(1996).

110. Jiang, G.-b., Ni, Z.-m., Zhang, L., Li, A., Han, H.-b.and Shan, X.-q., J. Anal. At. Spectrom., 7, 447 (1992).

111. Krivan, V. and Kueckenwaitz, M., Fresenius’ J. Anal.Chem., 342, 692 (1992).

112. Cabon, J. Y. and Erler, W., Analyst , 123, 1565 (1998).113. Tsalev, D. L., D’Ulivo, A., Lampugnani, L., Di Mar-

co, M. and Zamboni, R., J. Anal. At. Spectrom., 11, 979(1996).

114. Morita, M., Uehiro, T. and Fuwa, K., Anal. Chem., 53,1806 (1981).

115. Montaser, A., Inductively Coupled Plasma Mass Spec-trometry , Wiley-VCH, New York, 1998.

116. Boumans, P. W. J. M., Inductively Coupled PlasmaEmission Spectrometry , John Wiley & Sons, Inc., NewYork, 1987.

117. Amran, M. B., Lagarde, F. and Leroy, M. J. F., Mikro-chim. Acta , 127, 195 (1997).

118. Ebdon, L., Foulkes, M., Fredeen, K., Hanna, C. andSutton, K., Spectrochim. Acta, Part B , 53, 859 (1998).

119. Deng, B. Y. and Chan, W. T., Electrophoresis , 22,2186 (2001).

120. Chan, Y. Y. and Chan, W. T., J. Chromatogr. A 853,141 (1999).

121. Huang, B. L., Can. J. Appl. Spectrosc., 39, 117 (1994).122. Hettipathirana, T. D. and Davey, D. E., J. Anal. At.

Spectrum., 13, 483 (1998).123. Chausseau, M., Roussel, C., Gilon, N. and Mer-

met, J. M., Fresenius’ J. Anal. Chem., 366, 476 (2000).124. Emteborg, H., Bordin, G. and Rodriguez, A. R., Ana-

lyst , 123, 245 (1998).125. Wuilloud, R. G., Wuilloud, J. C., Olsina, R. A. and

Martinez, L. D., Analyst , 126, 715 (2001).126. Ochsenkuhn-Petropulu, M. and Schramel, P., Anal.

Chim. Acta , 313, 243 (1995).127. Bocca, B., Alimonti, A., Coni, E., Di Pasquale, M.,

Giglio, L., Bocca, A. P. and Caroli, S., Talanta , 53, 295(2000).

128. Conver, T. S., Yang, J. F. and Koropchak, J. A., Spec-trochim. Acta, Part B , 52, 1087 (1997).

129. Van Berkel, W. W., Balke, J. and Maessen, F. J. M. J.,Spectrochim. Acta, Part B , 45, 1265 (1990).

130. Roychowdhury, S. B. and Koropchak, J. A., Anal.Chem., 62, 484 (1990).

131. Laborda, F., de Loos-Vollebregt, M. T. C. and deGalan, L., Spectrochim. Acta, Part B , 46, 1089 (1991).

132. Zhang, X. H., Chen, D., Marquardt, R., John, A. andKoropchak, J. A., Microchem. J., 66, 17 (2000).

133. Zhang, X. H. and Koropchak, J. A., Anal. Chem., 71,3046 (1999).

134. Yang, J. F., Conver, T. S. and Koropchak, J. A., Anal.Chem., 68, 4064 (1996).

135. Hansen, G. W., Huf, F. A. and DeJong, H. J., Spec-trochim. Acta, Part B , 40, 307 (1985).

136. Kollotzek, D., Oechsle, D., Kaiser, G., Tschoepel, P.and Toelg, G., Fresenius’ J. Anal. Chem., 318, 485(1984).

137. Rosenkranz, B. and Bettmer, J., Trends Anal. Chem., 19,138 (2000).

138. Costa-Fernandez, J. M., Lunzer, F., Pereiro-Garcia, R.and Sanz-Medel, A., J. Anal. At. Spectrom., 10, 1019(1995).

139. Heisterkamp, M. and Adams, F. C., J. Anal. At. Spec-trom., 14, 1037 (1999).

140. Heisterkamp, M. and Adams, F. C., Fresenius’ J. Anal.Chem., 362, 489 (1998).

141. Pereiro, I. R., Wasik, A. and Lobinski, R., Anal. Chem.,70, 4063 (1998).

142. Pereiro, I. R., Wasik, A. and Lobinski, R., J. Chro-matogr. A, 795, 359 (1998).

260 DETECTION

143. Pereiro, I. R. and Lobinski, R., J. Anal. At. Spectrom.,12, 1381 (1997).

144. Hanstrom, S., Briche, C., Emteborg, H. and Bax-ter, D. C., Analyst 121, 1657 (1996).

145. Mena, M. L., McLeod, C. W., Jones, P., Withers, A.,Minganti, V., Capelli, R. and P. Quevauviller, P., Fre-senius’ J. Anal. Chem., 351, 456 (1995).

146. Pereiro, I. R., Wasik, A. and Lobinski, R., Fresenius’ J.Anal. Chem., 363, 460 (1999).

147. Krull, I. S., Panaro, K. W., Noonan, J. and Erickson, D.,Appl. Organomet. Chem., 3, 295 (1989).

148. Panaro, K. W., Erickson, D. and Krull, I. S., Analyst ,112, 1097 (1987).

149. Lobinski, R. and Adams, F. C., Spectrochim. Acta, PartB , 53, 1865 (1997).

5.2 Flow Injection Atomic Spectrometryfor Speciation

Julian F. TysonUniversity of Massachusetts Amherst, MA, USA

1 Flow Injection Analysis . . . . . . . . . . . . . . 2612 Flow Injection and Atomic Spectrometry 2623 Speciation of the FIAS Literature . . . . . . . 2634 Overview of Chemical Reactions for

Speciation . . . . . . . . . . . . . . . . . . . . . . . . 2645 Hydride Generation . . . . . . . . . . . . . . . . . 269

5.1 Antimony . . . . . . . . . . . . . . . . . . . . . 2705.2 Arsenic . . . . . . . . . . . . . . . . . . . . . . 2705.3 Selenium . . . . . . . . . . . . . . . . . . . . . 2715.4 Mercury . . . . . . . . . . . . . . . . . . . . . . 2725.5 Other elements . . . . . . . . . . . . . . . . . 272

5.5.1 Bismuth . . . . . . . . . . . . . . . . 272

5.5.2 Cadmium . . . . . . . . . . . . . . . 2725.5.3 Lead . . . . . . . . . . . . . . . . . . . 2735.5.4 Tin . . . . . . . . . . . . . . . . . . . . 273

6 Solid-phase Extraction . . . . . . . . . . . . . . . 2736.1 Arsenic . . . . . . . . . . . . . . . . . . . . . . 2736.2 Chromium . . . . . . . . . . . . . . . . . . . . 2746.3 Selenium . . . . . . . . . . . . . . . . . . . . . 2746.4 Other elements . . . . . . . . . . . . . . . . . 275

7 Other Procedures . . . . . . . . . . . . . . . . . . . 2778 Comparison with Other Speciation

Procedures . . . . . . . . . . . . . . . . . . . . . . . . 2789 References . . . . . . . . . . . . . . . . . . . . . . . . 278

1 FLOW INJECTION ANALYSIS

The need for greater automation of chemicalanalysis procedures is a major driving force inanalytical chemistry research and development.Automation brings a number of advantages in termsof the analytical figures of merit, such as improvedlong-term precision, as well as in the fiscal figuresof merit such as reduced cost per analysis. One wayof viewing flow injection (FI) is as a procedure forthe automation of serial determinations. Indeed, theearly publications in the FI literature concentrateon this aspect of the technique. Explanations ofthe basic principles were often made in terms ofFI being ‘segmented flow without the bubbles inslightly narrower, nonwettable tubing’.

Since the term ‘flow injection analysis’ was firstused, in 1975, to describe a procedure for chemicalanalysis, controlled fluid flow as an integral part

of an analytical measurement procedure is now sowidespread in current analytical methodology thatit is difficult to provide a concise definition of FI.Definitions which are based on concepts such as‘the gathering of information from a concentra-tion gradient formed when a well-defined zone offluid, dispersed in a continuous unsegmented streamof a carrier, flows through a suitable detector’, toparaphrase a definition provided by Ruzicka andHansen [1], who are widely acknowledged to be twoof the technique’s inventors, would nowadays beconsidered to include high performance liquid chro-matography (HPLC), capillary zone electrophoresis(CZE), and – as gases are fluids – gas chromatog-raphy (GC). The scope of what is considered to beflow injection can be ascertained from the materialcollected at the flow analysis database [2], in whichthere are now over 10 000 citations.

262 DETECTION

Flow injection was also at the forefront of whathas turned out to be another major theme of currentanalytical chemistry research and development, thedrive for miniaturization of chemical measurementsystems. It was not long before the theory ofthe characteristics of fluid flow and mixing inclosed conduits was being examined, and thebenefits of even further decreases in working scalesbecame apparent. If mass transfer processes inliquids are to be driven by diffusion, then rapidevents are only achieved by keeping the relevantdistances small.

FI is a microscale operation. That is, thevolumes of sample solution involved are in themicroliter range, the flow rates involved are inthe µL s−1 range, masses of solid reagents are inthe mg range, the volumes of reactors are in theµL to mL range. As FI is also a low pressureoperation, the tubing used for conduits is typically0.5–0.8 mm in diameter and not more than a fewmeters in length, hence fluids can be propelled byperistaltic (or syringe) pumps working at pressuresof a few hundred kPa.

Dispersion is controlled by the hydrodynamicsof such flow systems which are dominated bylaminar flow. In a closed circular pipe, the flowis characterized by a parabolic velocity gradientranging from zero at the walls to twice theaverage linear velocity at the center. For mostsystems, some dispersion is desirable as this is themechanism by which sample solution and reagentsolution are mixed. Typically, rapid radial mixingis required with minimum longitudinal mixing.These features can be promoted by inducing theappropriate secondary flow patterns which arefunctions of coiling, confluence point geometryand the insertion of packing materials into theflow lines. For a given manifold design, samplevolume, and flow rate(s), the dispersion processesare highly reproducible as are residence times.Thus all samples and standards are subjected toexactly the same chemistry for exactly the sametime, and thus it is not necessary for any processto be independent of residence time; as long asprocesses which affect the magnitude of the signalultimately measured by the spectrometer progressto an extent which gives a signal above that

equivalent to the detection limit, then the manifoldmay be used for quantitative analysis.

The key features of FI are often summarized assample injection, controlled dispersion and repro-ducible timing. Many of the flow-based proce-dures for the preconcentration of trace elements fordetermination by atomic spectrometry (AS) wouldnowadays be considered under the general head-ing of FIAS, even though the procedure did notinvolve sample ‘injection’ as such, rather the pro-cedure would be based on the introduction of arelatively large volume of sample by the continu-ous introduction of solution for a controlled time.Once separation from the matrix components hadoccurred, the procedure might well be based onthe handling of a limited zone of analyte withinthe flow manifold.

Twenty five years or so after the appearance ofthe first papers, it is easily seen that FI as a meansof automating chemical analysis procedures hasunderpinned the commercial success of the tech-nique. However, the concept of handling fluids bypressure-driven flow in narrow, nonwettable con-duits proved to be extremely versatile – capable of‘automating’ many procedures in addition to mix-ing of homogeneous aqueous solutions – and thusthe original invention gave rise to a considerablenumber of research themes in a considerable num-ber of research laboratories around the world.

2 FLOW INJECTION AND ATOMICSPECTROMETRY

The scale of flow injection is compatible withnormal operation of the various atomic spectrom-eters as detectors for the flow systems: flameatomic absorption spectrometry (FAAS) requiresintroduction at 5 mL min−1, inductively coupledplasma (ICP) optical emission spectrometry (OES)or mass spectrometry (MS) requires introductionat 1 mL min−1, and graphite furnace (GF) AASrequires about 20 µL of sample solution.

Although it got off to a relatively slow start,the concept of using FI as a sample handling sys-tem for atomic spectrometry has proved to be aparticularly fruitful research topic, and the flow

SPECIATION OF THE FIAS LITERATURE 263

injection atomic spectrometry (FIAS) combinationcould easily be considered a major category ofanalytical chemistry methodology in the same wayas the acronym HPLC represents a major collec-tion of current analytical chemistry methodology.In 1998, Sturgeon [3] wrote, ‘without doubt, thegreatest impact on sample processing and introduc-tion for atomic spectrometry has derived from thefields of FI technology and microwave radiation.’The combination of flow analysis techniques withatomic spectrometry detection (FIAS) now forms asignificant subdiscipline of the flow analysis field.Several textbooks have appeared, the most recentof which [4] included a chapter on cryofocusing(for ‘metal and metalloid speciation in the envi-ronment’), and a chapter on chromatographic sep-arations (for ‘trace element speciation in biologicalsystems’); thus the boundaries between ‘flow injec-tion’ and other flow-based analytical methods arebecoming blurred. The transient signals producedby FI sample introduction can model the responseof an atomic spectrometer to the transient concen-tration profiles in HPLC eluents [5, 6].

The literature is regularly reviewed as AtomicSpectrometry Updates in the Journal of AnalyticalSpectrometry; the Updates are entitled Advancesin Atomic Emission, Absorption and FluorescenceSpectrometry, and Related Techniques and AtomicMass Spectrometry, published in the June andAugust issues, respectively, cover the develop-ments relating to FIAS. The relevant literature islarge. Each year several hundred papers appearin the original primary literature concerned withsome aspect of the combination of FI with atomicspectrometry. Many of these are what might betermed ‘application’ papers, i.e. the contents arethe description of a new method for the deter-mination of X (some analyte or analytes) in Y(some matrix). There are relatively few publica-tions in which a genuinely new FI technique isdescribed. Many of the application papers are con-cerned with overcoming some limitation of theatomic spectrometry technique. Such limitationsare often related to detection capability, or inac-curacies due to the presence of some potentiallyinterfering matrix component. Thus a considerableamount of research effort is being devoted to the

development of FI procedures for the preconcen-tration and/or separation of the analyte(s) from thematrix. Categorizing the current research activityin this fashion does not relate immediately to theissues of speciation.

3 SPECIATION OF THE FIASLITERATURE

However, one possible categorization of the litera-ture relating to FIAS topics is to divide publishedwork into two categories: those papers in whichsome aspect of speciation is featured, and thosepapers in which some aspect of speciation is notfeatured. In the first category would be work inwhich the principal use of FI was (a) to transportthe sample solution to the spectrometer, (b) to pre-concentrate all of the analyte, or (c) to separate allof the analyte from the matrix. Solutions contain-ing high quantities of dissolved solids or suspendedparticles can be handled by FI. There are a sur-prising variety of ways, not all of which wouldappear to have been discovered yet, for the dilutionof sample solutions en route to the spectrometer,so that off-range samples can be handled withoutremoval from the autosampler tray, dilution andreanalysis. It is also possible to add reagents, suchas ionization buffers or releasing agents, by themixing of sample and reagent solutions in a suit-ably designed manifold.

In the second category, would be work in whichsome selectivity had been created via selectiveintroduction to the instrument: most likely byselective chemical reaction in the FI manifold. Theselectivity induced would be based on a separation,and thus the use of FIAS for speciation purposesis founded on the implementation of reactionsfor separation in a flow injection manifold wherethe key processes are controlled by fluid flowin closed, fixed-geometry conduits, reactors andother vessels.

The separations that are typically implementedin a FI system might be termed ‘nonchromato-graphic’. They are usually binary in character,i.e. the analytes are separated into two groups onthe basis of the process. These processes include,

264 DETECTION

for example, precipitation, liquid–liquid extrac-tion, solid-phase extraction (retention on a solidreagent), chemical vapor generation (conversion toa gaseous derivative), and dialysis. Continuous-flow sample delivery with such techniques wasthe subject of a book [7], which covers manyof the techniques that are used for speciationanalysis, though the speciation was not a majortopic of the book, nor indeed was flow injec-tion. A slightly more recent book [8] approachedthe topic of separation from the FI perspective.The book also covered the closely related topic ofpreconcentration. Again, speciation did not figurevery prominently in the table of contents or inthe index. Although atomic spectrometry featuredprominently among the instrumental techniquesused for quantification, other spectroscopic andelectrochemical techniques were also covered. InWelz’s ‘attempt at a forecast’ of where speciationanalysis is going [9], FI-based procedures are fea-tured prominently. The extent to which variousinstrumental techniques have been used with FIis apparent from a relevant review article [10] inwhich the literature up to 1995 was reviewed underthe title Chemical speciation by flow-injectionanalysis. As the authors of this review point out,FI has the ability to help overcome a general prob-lem encountered in speciation analyzes involvingseparations, namely the redistribution of speciesfollowing the disturbing of the equilibrium bythe separation, as the time between reaction andmeasurement can be kept short (maybe only afew seconds).

There are several reports of procedures inwhich FIAS is used as part of a method whichresults in the production of information aboutthe speciation of the element, or elements, ofinterest, but in which the diagnostic chemistry isnot performed in the FI manifold. For example,As(III) was selectively determined in fish tissue bydistillation of AsCl3 followed by FI-HG-AAS [11].Such procedures have been considered outside thescope of this chapter, which deals with speciationmethodology in which the chemistry which formsthe basis of the speciation has been performed online in the FI manifold.

4 OVERVIEW OF CHEMICALREACTIONS FOR SPECIATION

To a first approximation, the response of anyatomic spectrometer is independent of the chemicalform of the element(s) introduced, and thusthere is little scope for using the response ofthe spectrometer as the basis of a measurementof chemical speciation. Selectivity towards thedifferent chemical forms of the element of interestare therefore produced by reaction chemistrydesigned to physically separate fractions of theanalyte prior to measurement. In principle, anychemical reaction or process that produces suchseparation has the potential for application as aspeciation procedure; however, to be useful inthe context of helping to solve problems by theprovision of reliable information about chemicalcomposition, it is preferable that the separationbe easily related to well-defined chemical species.There are, though, schemes of speciation whichare quite widely used, but which are basedon separations for which the species separatedare not well defined. For example there areschemes for the estimation of various formsof elements in soils based on leaching withsolutions of increasing complexing or solubilizingability [12, 13]. In 1985, Van Loon wrote [14],‘I treat such procedures with some degree ofskepticism because past experience suggests thatit is not possible in most cases to restrict theextraction to a particular soil fraction. However,in spite of these problems agricultural scientistsfind these procedures helpful’. It is possible [15]to implement such procedures in a FI format.

This separation basis for speciation means thatthe analysis, typically, proceeds in several stages,each of which requires a fresh portion of thesample material: following suitable pretreatement,the total element will be determined, then a freshsample portion will be obtained and a measurementmade of the element that is separated underthe conditions selected. This might be repeatedwith a different set of conditions selective foranother analyte species or a different separationscheme might be used. Finally, a fraction will bequantified by calculation as the difference between

OVERVIEW OF CHEMICAL REACTIONS FOR SPECIATION 265

the total and the sum of the various speciatedfractions measured.

As with other analytical methods in which reac-tion chemistry precedes instrumental measurement,the species quantified are only defined in terms ofthe chemistry which forms the basis of the proce-dure. For example, it is common to describe thereaction in which acidified potassium antimonyltartrate, ammonium molybdate and ascorbic acidare added to the sample in the appropriate orderfollowed, possibly, by heating and preceded, pos-sibly, by the addition of masking agents as ‘thedetermination of phosphate’. What, in fact, is mea-sured are all species that will react to give a bluecolor under the conditions used; such species mightinclude other phosphorus-containing oxo-anions,and oxo-anions of other elements. Most speciationprocedures are based on the goal of being ableto related what is measured to the concentrationof a defined chemical entity as this, in general,leads to information which aids in understandingthe problem being studied. However, this is notthe only goal of speciation analysis. An importantquestion about materials that might be ingested(either deliberately as in the case of food anddrink) or accidentally (as in the case of soil nearpressure-treated decks) concerns the bioavailabil-ity of potentially harmful elements. In assessing the‘arsenic status’ of a soil, it may be more important

to know how much arsenic is bioavailable than itis to know how much arsenic is present as arsen-ite, arsenate, monomethyl arsonic acid and so on.Devising simple chemical tests that are reliableindicators of bioavailability is not a trivial prob-lem, and is an active area of research at present.

All of the separation procedures commonlyused in analysis can be adapted to the FI format.Even distillation can be performed in an FIsystem [11, 16] though there appears to be littleinterest in such systems at present. The onlywell-established chemical separation procedurethat has not yet been adapted to the FI formatis fire assay. There is some variation in thedesign of manifold for the various separationprocedures and the way such manifolds areinterfaced with the spectrometer (not always atrivial exercise in the case of graphite furnaceatomizer instruments); however, it is possible toidentify a prototypical FI manifold for separatinganalyte species either from each other or fromother sample components. This is shown inFigure 5.2.1. The sample solution is either injectedinto a carrier (C), or is pumped continuously, andmerged with a reagent R. It might be necessaryto add more than one reagent (for examplethe addition of a complexing agent followed,after a suitable reaction time, by the addition ofan extracting solvent). The sample components

Inject

C

R1

R2

R3

DSep Mod

stream switchingmodule

W

Pump

Figure 5.2.1. Generalized FI manifold for speciation studies. C, carrier stream; R1, R2, R3, reagents; Sep Mod, separationmodule; D, detector; W, waste. Typically R1 and R2 are selective for some component of the sample which is injected via asuitable valve at the location shown by the ‘Inject’ arrow. The reagent R3 would be involved in the release of sample componentsretained in the separation module.

266 DETECTION

pass into the separation module (whose exactnature depends on the nature of the separation).There are several possible next stages: (1) selectedsample species can pass to the detector with othermatrix components of the sample, (2) selectedsample species can pass to the detector whileother matrix components of the sample pass towaste, (3) sample species are retained, matrixcomponents pass to the detector or to waste, thensample species are released from the separationmodule, (4) sample species are retained, matrixcomponents pass to the detector or to waste, thensample species are sequentially released from theseparation module, or (5) a combination of (1) or(2) with (3) or (4). Although the figure is drawnas though the separation module was mounted inone particular orientation, it may well be that it islocated in a stream switching module that allowsit to be moved from one flow line to another andfor the direction of flow to be reversed.

The most widely used separation modules, forspeciation and other applications, are (a) the strip-ping coil and gas–liquid separator and (b) thesolid-phase extraction cartridge (often referred toas a ‘minicolumn’) or coil. Less widely useddevices are (a) the liquid–liquid extraction coiland phase separator, (b) the particulate filter, and(c) the dialyzer. In one or two cases there issome ambiguity as to the nature of the separa-tion process. For example, when ammonium pyrro-lidine dithiocarbamate is added to an aqueoussolution of trace metals, complexes are formedwith can be retained on a C-18 cartridge oron the walls of an open tubular reactor. Asthese complexes are generally thought to beinsoluble, it is not entirely clear whether theretention is by solid phase extraction or filtra-tion. More detailed illustrations of the imple-mentation these separation schemes are given inFigure 5.2.2.

Most of the speciation schemes implementedwith the aid of FI technology are based ondifferences in thermodynamic properties of theparticipants in the reactions which form the basisof the separation. For example, hydride generation(HG) is quite widely used to discriminate betweendifferent oxidation states of inorganic selenium.

For most samples, the soluble inorganic seleniumexists as either selenite (SeO3

2−) or selenate(SeO4

2−) in which selenium is in the 4+ state orthe 6+ state, respectively. Extensive experimentalevidence supports the contention that, when asolution of selenate in acid is mixed with analkaline solution of borohydride (BH4

−) so thatthe resulting solution is acid, no volatile seleniumspecies is formed; whereas if the experiment isrepeated with selenite a substantial percentage(approaching 100) of this species is convertedto hydrogen selenide (H2Se). Thus HG providesthe basis for the sequential determination of thesetwo species. In the case of solid-phase extraction,the basis of separation is the formation of aproduct with the immobilized reagent that ismore stable than (a) the solution-phase reactantform of the target species and (b) all otherpossible products with other species in the solutionphase. Thus when the solution and solid areseparated, only one species is retained by thesolid reagent. For example, an anion exchangecolumn will retain negatively charged forms ofchromium from aqueous solution (provided theconditions are appropriate) and positively chargedforms will remain in solution. This proceduretherefore provides one possible basis for theseparation of chromium in the +6 oxidation state(as these species are negatively charged) fromchromium in the +3 state (species which arepositively charged).

Features of chemical reactions such as ionicstrength effects, pH effects and buffer capacitymust also be borne in mind. The position of theequilibrium of reactions involving ionic speciescan be significantly affected by ionic strength.Samples which have substantial inorganic matri-ces (such as seawaters) may have ionic strengthsorders of magnitude greater than those of the cal-ibration standards. Some samples may have highacid concentrations as a result of the pretreatementprocedure which overwhelm the buffer capacity ofthe solution used for pH adjustment. Some samplesmay have a buffer capacity high enough to resistthat of the buffer solution.

There are some schemes which are basedon kinetic considerations. For example, in the

OVERVIEW OF CHEMICAL REACTIONS FOR SPECIATION 267

solid phase extractant

(a)

sample waste

eluent

(c)

stripping coil

to atomic spectrometervia dryer

pumped to waste

stripping gas

gas-liquid separator

(b)

solid phase extractant

spectrometer eluent

(d)

extraction coil

organicextractant

aqueoussample

organicphase

phaseseparator

aqueouswaste

Figure 5.2.2. FI separation devices. (a) Solid-phase extractant mounted in column in six-port rotary valve in ‘load’ position;(b) valve switched to elute position; (c) chemical vapor generation manifold showing gas–liquid separator filled with glassbeads; (d) liquid–liquid extraction manifold showing conical phase separator in which less dense organic solvent is removedat the upper outlet; (e) dialysis (or supported liquid membrane) separating donor and acceptor solutions; (f) in-line filter forcollection of precipitate [can be mounted in loop of injection valve as shown for (a) and eluted by back-flushing as shownin (b)].

268 DETECTION

(e)

donor solution

acceptor solutiondialysis membrane

(f)

filter

sample

precipitant

Figure 5.2.2. (continued)

speciation of aluminum in natural waters, thefraction identified as being potentially the mosttoxic to aquatic organisms contains monomeric,inorganic hydrated ions. These species are oftenquantified on the basis of the speed of the reactionof aluminum species with chelating reagents,such as 8-hydroxyquinoline (8-HQ). This ideawas proposed prior to the availability of FItechniques, but may be readily implemented inan FI manifold as the combination of controlledinjection volume and flow rate allows the contacttime between analyte species and reagent tobe precisely controlled. Thus if the 8-HQ isimmobilized in a short column the so-called ‘fastreacting’ or ‘labile’ aluminum can be determinedby measurement of the aluminum that is retainedon the passage of the injected slug though sucha column.

As residence times in FI systems are on theorder of seconds, reactions which are exploited asthe basis of speciation measurements are usuallyfast. Some reactions which are quite widely usedare not fast enough at room temperature, and toimplement these in a flow injection system requiressome additional features to be incorporated intothe manifold or its operation. To some extent,residence times can be increased by slowing theflow or increasing the length of the manifoldtubing or both. There are some practical limitationsto these approaches arising from the limitationsof peristaltic pumps. Firstly it is difficult topump reliably (i.e. with good precision and nopulsations) at very low flow rates (µL min−1), andsecondly increasing the tube length increases the

back pressure and it can become impossible topump the solutions though the manifold. Residencetimes can be increased by stopping the flow fora desired period. This requires computer controlover the pumping device, but is probably thebest approach if this stopped-flow mode is used.Another approach to this problem is to increasethe rate of the desired reaction by raising thetemperature. This can be done by immersingthe reaction coil in a conventional heating bathor by irradiation with microwaves. This latterapproach provides an additional possibility forspeciation as the energy is controllable. Thusit is possible to make a measurement of thespecies introduced to the spectrometer arisingfrom the chemical reactions occurring with themicrowaves on and a measurement of the speciesproduced with the microwaves off. For example,(selenate + selenite) can be determined by HG ifthe sample contains hydrochloric or hydrobromicacids and the microwaves are ‘on’ (as selenate isreduced to selenite). When the microwaves areoff, only selenite is detected as selenate is not‘borohydride active’.

The kinetics of mass transfer processes shouldalso be borne in mind. If a solid-phase extractionprocedure is to be used as the basis of separationin an FI system, then the amount of materialretained on a small column in an FI systemwill depend on the extent to which the speciesin solution get close to the immobilized reagentspecies on the surface of the solid support.If the support material creates large interstitialspaces and the analyte species are dilute and a

HYDRIDE GENERATION 269

high volumetric flow rate is used, the retentionefficiency may be very poor (only a few percent ofthe target material is retained). Thus a considerableinaccuracy is incurred if the concentration of thenonretained element is equated with species A andthe element subsequently eluted is equated withspecies B. Diffusion in liquids is slow; to use itas a mechanism of mass transfer means that thedistances involved must be small (or residencetimes must be long). One of the reasons thatHPLC is HP is that mass transfer by diffusionin the interstitial spaces is effective because theinterstitial spaces are small. This is achieved atthe expense of flow rate as the back pressurefor a column of useful dimensions is high. Highpressure pumping (with HPLC pumps) is notnormally considered a feature of FI. Residencetimes can be manipulated as discussed in theprevious paragraph.

Some speciation schemes depend on the kinet-ics of heat transfer to effect temporal separa-tion of volatile species after cryotrapping. Thisprocedure is normally applied following gener-ation of volatile derivatives of several speciesin the sample. For example, inorganic arsenic,monomethylarsonic acid and dimethylarsinic acidcan be derivatized with borohydride to arsine,monomethyl arsine and dimethyl arsine, all threeof which can be trapped at liquid nitrogen temper-atures. When the trap is warmed, the three com-pounds elute in order of lowest boiling (arsine)first, highest boiling (dimethylarsine) last. In aninteresting variation on this experiment, Bur-guera et al. trapped the arsines in a coil insidea microwave oven [17] and remobilized themby microwave heating when the arsines werevolatilized in the reverse order, i.e. the compoundwith highest boiling point (but highest dipolemoment) was volatilized first.

5 HYDRIDE GENERATION

The basic principles of hydride generation (HG)were comprehensively covered in a recent bookby Dedina and Tsalev [18], which despite thetitle does cover HG for atomic fluorescence and

emission as well. There has been a substantialbody of work published since this book waspublished (maybe as many as a thousand papers)whose contents have been reviewed in the regularAtomic Spectrometry Update articles publishedbimonthly in the Journal of Analytical AtomicSpectrometry. The salient features of FI-HG werecovered by Dedina in a chapter [19] in a recent FIbook [4]. In addition to the general advantageousfeatures of FI compared with batch procedures,there is an additional feature of FI-HG that isbeneficial: kinetic discrimination over interferencescaused by the reaction of borohydride with matrixcomponents. In general, the primary HG reactionis fast compared with the competing interferingreactions, and improved tolerance to interferencesis obtained in an FI system as the residence timeof the hydride in the liquid phase is very short.Much of the work published recently concerning thedetermination of the ‘hydride-forming’ elements isconcerned with some aspect of speciation, thougha substantial number of papers describe HPLCseparation with element specific detection. By farthe greatest interest at present is in the determinationof arsenic and selenium species. Other elementsof interest include tin, lead, and antimony. FI-HGhas been used for the determination of bismuth,cadmium, and germanium. Mercury species mayalso be determined via HG, though the determinationof inorganic mercury is normally considered not toinvolve the formation of a hydride. A commonlyused reagent is tin(II) in hydrochloric acid solutionwhich reduces Hg(II) species to elemental Hg, whichcan be blown out of solution with a stream of argonor nitrogen. Inorganic mercury also reacts withborohydride, BH4

−, to release elemental mercuryvapor. If mercury hydride is formed it is consideredto be unstable with respect to elemental mercury andhydrogen and to have a very short lifetime.

Hydride generation may be used as the basisof a speciation scheme by virtue of selectivederivatization. The most common reagent isborohydride, BH4

−. Not all forms of the so-called ‘hydride-forming’ elements are ‘borohy-dride active’ (i.e. give a volatile derivative, whichmay be easily atomized), when mixed with aborohydride solution followed by acidification.

270 DETECTION

Howard [20] has reviewed speciation based onreaction with borohydride, covering discrimina-tion based on (a) pH control, (b) cryotrapping,and (c) HPLC. Tsalev, in a recent comprehen-sive review [21] surveyed HG procedures for spe-ciation. HG may also be used in conjunctionwith other separation schemes, such as solid-phaseextraction (in the case of FI-based speciation) orHPLC. In both cases, consideration has to be givento the borohydride activity of the separated speciesand further chemical transformation may be nec-essary in the case of species which are inactive.Most procedures of this sort are based on the con-version of the borohydride-inactive species to thesimple inorganic anion of lowest oxidation state.This may involve reduction, by agents such aschloride, bromide, iodide, or ascorbic acid in thecase of inorganic ion oxidation state adjustment,or combinations of oxidation and reduction whenorganocompounds containing bonds between car-bon and the element of interest are to be deter-mined. Oxidants such as alkaline persulfate, aciddichromate or permanganate have been used. Allof these types of reactions have been promotedby microwave heating [22, 23], and there is evi-dence that the desired conversions can be obtainedunder less aggressive chemical conditions than areneeded when conventional heating is applied. It isalso possible to promote some of these reactionswith UV light [20], an attractive proposition fora FI procedure as flow-through UV photoreactorsare commercially available.

5.1 Antimony

An HG-inductively coupled plasma atomic emis-sion spectrometry (ICP-AES) procedure for thedetermination of antimony(III) and antimony(V),based on the kinetics of the reduction reactionof antimony(V) with L-cysteine, was devised byFeng et al. [24]. They found that the degree ofconversion was a linear function of time for aperiod of up to 10 min. Their method was basedon measurements made 2 and 8 min after theaddition of the reducing agent. Potential interfer-ences from transition metals were removed witha column of Chelex-100. Ulrich [25] devised an

ICP-AES method in which the reactivities ofvarious antimony species [inorganic Sb(III), Sb(V)and trimethylstilboxide (TMeSbO)] towards fluo-ride formed the basis of the separation. In thepresence of fluoride and iodide, only trimethyl-stilboxide was detected; in the presence of flu-oride both Sb(III) and TMeSbO were detected.In the presence of iodide, Sb(V) was reduced toSb(III) which was detected along with TMeSbO.The limits of detection were around 1 µg L−1

for each species. Deng et al. showed [26] thatthe partial contribution of the signal from Sb(V)obtained in their FI-HG-AFS procedure could beeliminated by the addition of 8-hydroxyquinolineadded originally to suppress the interferences bytransition metals in the analysis of lake watersand sediments.

5.2 Arsenic

Nielsen and Hansen [27] devised an AAS proce-dure for the determination of As(III) and As(V) inwater samples. Total As was measured followingon-line reduction with ascorbic acid and potassiumiodide in a knotted reactor immersed in an oil bathat 140 ◦C. Without the heating bath and reduc-ing agents, and with mild acid conditions, onlyAs(III) was determined. The procedure was appliedto the analysis of a certified drinking water mate-rial. A similar procedure was developed by Kren-zelok and Rychlovsky [28] who reduced As(V)to As(III) with potassium iodide (in 6 M HCl)alone at 60 ◦C. A membrane gas–liquid separa-tor was used. Willie [29] devised what he called a‘first-order’ speciation procedure with detection byETAAS following trapping of the volatile arsenic-containing species on the interior of the atomizer.Conditions were used under which volatile deriva-tives were formed from As(III), As(V), MMA, andDMA, but not from arsenobetaine, arsenocholineor tetramethylarsonium. On-line UV oxidation ormicrowave heating in the presence of alkalinepersulfate converted all species to borohydride-active forms. The procedure was applied to theanalysis of several water and urine reference mate-rials. A somewhat similar method was proposedby Cabon and Cabon [30] for the analysis of

HYDRIDE GENERATION 271

seawater. They showed that As(III) was not stablewith respect to oxidation to As(V) in acidi-fied seawater and slowly converted to As(V) incoastal waters at the natural pH. The methodwas extended [31] by cryogenically trapping thevolatile hydrides from As(III) and As(V), gener-ated under differing conditions, and MMA andDMA on a chromatographic phase followed bysequential release. It was claimed that ‘evidencefor the biotransformation of arsenic in seawaterwas clearly shown’. Burguera et al. [17] remo-bilized the arsines trapped in a liquid-nitrogen-cooled coil by irradiation with microwaves. Thesefour species can, apparently, be differentiated onthe basis of reaction conditions [32]; whereas theaddition of L-cysteine produced equal sensitivi-ties for each species [33]. This was thought to bedue to the reduction of the As(V) species (arse-nate, MMA and DMA) to organosulfur–As(III)species, which would be identical with or respondidentically to the species formed when arseniteand L-cysteine react. This reagent has also beenstudied by Tsalev et al. [34] who showed that thespecies produced on reaction with L-cysteine couldbe separated by HPLC. They also showed [35]that UV-assisted oxidation with persulfate wouldtransform arsenite, MMA, DMA, arsenobetaine,arsenocholine, trimethylarsine oxide and tetram-ethylarsonium to arsenate. The reaction was usedfor the determination of total As by FI-HG-AAS, and for post-column derivatization followingHPLC separation.

Electrochemical HG has the same sort ofselectivity as borohydride and so can be applied forspeciation. Pyell et al. [36] devised a system thatwas selective for As(III) and, with the addition of acryotrap, was selective for arsine over the variousmethylated arsine species.

There are several reports of procedures in whichthe quantification was by FI-HG-atomic spectrom-etry following various pretreatments in the batchmode. Perhaps the most important of these isthe procedure developed by Chaterjee et al. [37]in support of the studies of the contaminationof well-water in West Bengal, described by theresearchers as ‘the biggest arsenic calamity inthe world’.

5.3 Selenium

An AAS procedure for the determination ofSe(IV) and Se(VI) in seawater was developed byFernandez et al. [38] in which the Se(VI) wasreduced in the FI manifold to Se(IV) on addition ofconcentrated hydrochloric acid and passing thougha coil immersed in a heating bath at 140 ◦C.Hydride generation was performed at 0 ◦C byplacing the relevant part of the manifold in anice bath. Stripeikis et al. [39] measured relevantreaction kinetic parameters for such a system andclaimed to have the best throughput, 60 h−1, ofany FI system described so far. Microwave heatingis also suitable for the conversion of Se(VI) toSe(IV) by hydrochloric acid, and such a procedurewas developed by Burguera and coworkers [40] forthe determination of these species in citrus fruitjuices and geothermal waters. They later modifiedthe procedure [41] so that reduction was effectedby a mixture of hydrochloric and hydrobromicacids; in comparison to the procedure which usesonly hydrochloric acid, lower concentrations ofthe acids (10 %) could be tolerated. He et al. [42]determined these selenium species in seawater bya similar procedure, but in which quantificationwas by AFS. The AFS detection limits of 5 ng L−1,were about 50 times better than those of the AASmethods. Moreno et al. [43] compared on- and off-line procedures for the extraction (and reduction)of selenium species from wastewaters and sewagesludges, and concluded that accurate analyzes bythe on-line method required calibration by themethod of standard additions. Varade and Luque deCastro [44] showed that subcritical water (250 ◦Cand 200 bar) extracted both inorganic and organiccompounds from sludges, but that the organiccompounds did not survive the extraction. The on-line reduction to Se(IV) was helped by microwaveirradiation.

It has been reported [45, 46] that UV irradiationconverts both Se(VI) and various organoseleniumcompounds to Se(IV). This has been used forthe speciation of selenium species in seawaterby various off-line pretreatments [45] and forHPLC post-column derivatization [46]. This post-column derivatization has also been achieved by

272 DETECTION

microwave-assisted reaction with a mixture ofpotassium bromate and hydrobromic acid [47].

Clearly most researchers thought that organose-lenium compounds are not borohydride active;however, recently published evidence [48] sug-gests that selenomethionine, selenoethionine andtrimethylselenonium do react with borohydrideto produce methyl selenol and dimethyldise-lenide, ethyl selenol and diethlyldiselenide, anddimethylselenide, respectively. It has also beenshown that trimethylselenonium can be determinedby HG-AAS with the same sensitivity as for selen-ite [49]. The reasons for these differences in find-ings over the reactivity of these organoseleniumcompounds are not yet clear, but may have some-thing to do with (a) the acidity of the solution or(b) the presence of matrix components capable ofreacting with borohydride.

5.4 Mercury

Gallignani et al. [50] showed that inorganic mer-cury could be determined in the presence ofmethylmercury by the use of stannous chlorideas the reductant to generate mercury vapor. Totalmercury was determined after on-line microwave-assisted oxidation with persulfate. For deter-mination by AAS, the limit of detection was0.1 µg L−1. Rio-Segade and Bendicho devisedsomewhat similar procedures for distinguishingbetween inorganic mercury and methylmercury.In extracts containing both species only inor-ganic mercury was determined when stannouschloride was used as the reductant [51]. Whenon-line oxidation with potassium peroxodisulfate(persulfate) in sulfuric acid was triggered by heat-ing [52], total mercury was measured; with theheating off, only inorganic mercury was deter-mined. In the former procedure, mercury specieswere extracted from fish tissue into hydrochloricacid with the help of ultrasound. However, it hasbeen reported [53] that such a procedure can con-vert methylmercury to inorganic mercury, but asthe results obtained for the analysis of three fishreference materials were in agreement with the cer-tified values there may be conditions under which

no conversion occurs. Tao et al. [54] obtainedaccurate results for marine tissue reference mate-rials following dissolution of mercury species intetramethylammonium hydroxide solution. Inor-ganic mercury was selectively determined withstannous chloride after the addition of L-cysteine.Burguera et al. [55] determined mercury speciesin fish-egg oil. A surfactant was added to gen-erate an emulsion which was much easier topump in the FI system than the original vis-cous oil. Inorganic mercury was determined onthe addition of borohydride (when presumablymethylmercury was not detected) and total mer-cury after oxidation with persulfate. However, forthe 22 samples examined no inorganic mercurywas detected whereas the organic mercury con-tent ranged from 2 to 3 µg L−1. There is con-flicting evidence in the literature concerning theproduct of the reaction of methylmercury withborohydride. Some researchers [e.g. 49] find thatmercury vapor is produced, whereas others [e.g.56, 57] find that methylmercuryhydride is pro-duced. Recent work [58] suggests that the con-centration of both the borohydride and the sodiumhydroxide, used to stabilize the borohydride solu-tion, are important factors.

5.5 Other elements

5.5.1 Bismuth

Bismuth species in urine were determined by AASfollowing ethylation with tetraethylborate [59].The procedure was free from interferences fromother components in urine and was applied aftera simple 1 + 1 dilution. The limit of detectionwas 2 µg L−1 and the method was used to followthe urinary clearance of therapeutic doses ofbismuth subcitrate

5.5.2 Cadmium

Post-column HG with ICP-MS detection allowedthe detection of cadmium-containing metalloth-ioneins in eel liver and kidney separated byvesicle-mediated, reversed-phase HPLC [60]. The

SOLID-PHASE EXTRACTION 273

procedure was validated by the examination of rab-bit liver cadmium metallothioneins.

5.5.3 Lead

Organolead compounds in the concentration range0.25–8 mg L−1 were selectively determined in thepresence of inorganic lead by FI-HG-ETAAS [61].A carrier stream containing hydrochloric acidand ethylenediaminetetraacetic acid suppressed theinorganic lead signal for concentrations up to10 mg L−1.

5.5.4 Tin

Post-column UV photooxidation in the presenceof persulfate at 95–100 ◦C converted a variety ofalkyltin compounds to forms that could be detectedby HG-AAS [62] with better than 80 % recovery.However, tetrabutyltin was only 15 % recovered.

6 SOLID-PHASE EXTRACTION

More papers are published about the implemen-tation of a solid-phase extraction procedure in anFI mode with direct coupling to an atomic spec-trometer than about any other FIAS topic. Mostof these publications are concerned not with spe-ciation analysis, but with either separation of ana-lyte(s) from a potentially interfering matrix or withpreconcentration, or both. Many of the methodsdeveloped are designed to extend the capabilitiesof flame AAS (FAAS).

The possible applications for speciation analy-sis are based on the selective retention of speciesby a column packed with a suitable solid-phaseextractant. Unretained species may either be dis-carded to waste or be directed to the spectrometerfor determination. Species retained on the columnmay then be eluted with an appropriate reagentand measured by the spectrometer. In some cases,it is possible to sequentially elute more than oneretained species.

Terminology can be a little confusing in thisarea as some authors refer to the procedure

described in the previous paragraph as ‘extractionchromatography’. In chromatographic terms, spe-cies which are not retained have a capacity factorof 0. Capacity factor (often given the symbol k′)is the number of column volumes, over and abovethe void volume, needed to elute a component fromthe column. Species which are retained have verylarge capacity factors, maybe approaching infinity.The term ‘chromatography’ will not be used inrelation to these solid-phase extraction proceduresunless it is clear that in the procedure described thespecies were being eluted under conditions whichproduced k′ > 0 and � infinity.

6.1 Arsenic

A procedure in which the complex between As(III)and ammonium diethyldithiophosphate was selec-tively retained on a C-18 column was devisedby Pozebon et al. [63]. The retained species waseluted with 120 µL of ethanol into an autosam-pler cup for transferal (30 µL plus 10 µL of 0.1 %palladium nitrate solution) into a graphite furnaceatomizer. Total arsenic was determined after reduc-tion with potassium iodide and ascorbic acid andhence As(V) by difference. The working rangewas 0.3–3 µg L−1. Tyson [64] developed a pro-cedure in which the hydride was generated fromthe surface of an anion-exchange resin by thesequential retention of analyte and borohydridefollowed by the passage of a discrete volume ofacid through the column. A speciation procedurefor As(III), As(V) and methylated arsenicals wasdevised based on control of pH of the carrierstream and some off-line oxidations. At pH 2.3only arsenate is not fully protonated and hencethis was the only species retained by the col-umn. Reaction with hydrogen peroxide in nitricacid oxidized As(III) to As(V) and thus As(III)could be found by difference. Further oxidationwith alkaline persulfate in a sealed vessel in amicrowave oven converted all species to As(V)and thus the methylated arsenic species werefound, again, by difference. The limit of detectionwas 4 ng L−1 for a 10 mL sample volume. Burgueraet al. [65] trapped As(V), MMA and DMA on acombined cation–anion exchange column and then

274 DETECTION

sequentially removed them by eluting first with0.006 M trichloroacetic acid [to remove As(V)],then with 0.2 M trichloroacetic acid (to removeMMA) and finally with 5 M ammonium hydrox-ide solution (to remove DMA). As(III) was notretained and was determined by difference aftera separate determination of total As after an off-line, microwave-assisted, acid decomposition. Thespecies in the various eluents were determined byHG-AAS after ‘reduction’ with a solution con-taining 15 % (m/V) potassium iodide and 5 %(m/V) ascorbic acid. The limits of detection werearound 0.1–0.3 µg L−1 in a 2 mL urine sample.A similar approach has been used by Yalcin andLe [66] who retained (a) DMA on a resin-based,cation exchange resin, followed by elution with1 M HCl, (b) MMA and As(V) on a silica-based,anion exchanger, followed by elution of the MMAwith 0.06 M acetic acid and of the As(V) with 1 MHCl, and (c) all four species on alumina. As(III)was not retained on the cation or anion exchangematerials. The procedure was applied to the deter-mination of the species in drinking water down toconcentrations around 0.05 µg L−1.

6.2 Chromium

All of the published FIAS work is concerned withmeasurement of Cr(III) and Cr(VI). Sperling andcoworkers [67] developed an ETAAS procedurein which Cr(VI) was selectively retained as thecomplex with diethyldithiocarbamate (DDC) onC-18 silica and then eluted with ethanol. TotalCr was determined after oxidation of Cr(III) toCr(VI) with persulfate, and hence Cr(III) wasdetermined by difference. The limits of detectionwere 20 ng L−1. They also developed a procedurefor FAAS [68] in which an alumina column wasrendered selective to each of the species in turnby varying the pH of the carrier solution. AtpH 2 Cr(VI) was retained, and at pH 7, Cr(III)was retained. The eluents were ammonia solution(0.5 M) and nitric acid (1 M), respectively. Thelimits of detection were 1 µg L−1. More recently,titanium dioxide has been used in a similarfashion. Vassileva [69] developed a procedure forthe retention of each species on two separate

columns followed by determination by ICP-AES,and Yu et al. [70] selectively retained Cr(VI) atpH 2 with detection of the unretained Cr(III) byICP-MS. However, the Cr(VI) was not elutedfor subsequent determination, rather a total Crvalue was obtained by direct introduction so thatCr(VI) was determined by difference. The limitsof detection were 70–80 ng L−1.

Several other solid-phase extraction materialshave been used for one or other of the species.Naghmush et al. [71] investigated the performanceof various functionalized cellulose absorbents, achelating resin and some ion exchange resins. ForFAAS the detection limits were around 1 µg L−1.Jiminez et al. [72] collected Cr(III) on Amber-lite IR-20 cation exchange resin and Cr(VI) onAmberlite IRA-400 anion exchange resin. TheFAAS detection limits were 10 and 1 µg L−1

for Cr(III) and Cr(VI), respectively. Kelko-Levaiet al. retained Cr(III) on ‘IDAEC’ and Cr(VI)on the anion exchanger diethylaminoethyl (DE)-cellulose [73]. After elution the elements werequantified either by ETAAS or by total reflectionX-ray spectrometry.

The method of Sperling et al. [67], based onretention of the DDC complexes on C-18, wasrecently modified by Rao et al. [74] so that bothspecies were retained: Cr(VI) at pH 1–2 and Cr(III)at pH 4–9 [in the presence of Mn(II) which wasreported to enhance the FAAS signal for Cr(III)by a factor of 10]. The eluent was methanol andthe overall enrichment factor was 500 for a 5 minpreconcentration, allowing determination of bothspecies in the concentration range 0.2–200 µg L−1.Cespon-Romero et al. [75] devised a procedurein which Cr(III) was selectively retained on apoly(aminophosphonic acid) chelating resin fol-lowed by elution with hydrochloric acid (0.5 M).Total Cr was determined after off-line reduction ofCr(VI) with ascorbic acid. The limit of detectionwas 0.2 µg L−1 for a 6.6 mL sample volume.

6.3 Selenium

Selenium(IV) and (VI) were preconcentrated on analumina column (activated by the carrier streamof 0.01 M nitric acid) by workers in Camara’s

SOLID-PHASE EXTRACTION 275

group [76]. Following elution with 2 M ammoniasolution the Se(IV) was determined by on-lineHG-AAS. Total selenium was determined byreduction prior to introduction to the column, andthus Se(VI) was determined by difference. For25 mL of sample, a preconcentration factor of50 was obtained with detection down to around6 ng L−1. Pyrzynska et al. [77] devised an off-lineprocedure based on the same preconcentration, butwith successive elution of the Se(IV) with 1 Mammonia and the Se(VI) with 4 M ammonia, anddetermination by ETAAS. The limit of detection ofdetection for Se(IV) was 50 ng L−1. A sequential-elution, on-line procedure was devised by Bryceand coworkers [78]. The two selenium specieswere retained on a strong anion-exchange materialfollowed by elution of Se(IV) with 2 M formic acidand of Se(VI) with 6 M HCl. On-line reduction toSe(IV) was aided by passage through a microwaveoven. Determination was by HG-AFS and for asample volume of 600 µL and an elution volumeof 350 µL, the detection limit was 40 ng L−1.Yan et al. [79] selectively retained Se(IV) as thepyrrolidine dithiocarbamate complex on C-18,from 4.2 mL of sample, followed by elution with26 µL of ethanol directly into a graphite furnaceatomizer pretreated with iridium. The detectionlimit was 4 ng L−1.

Carrero and Tyson [80] retained Se(IV) togetherwith borohydride on an anion exchange resinfollowed by HG directly from the solid phaseby the passage of a slug of acid. For a samplevolume of 9 mL the detection limit was 100 ng L−1.The work has been further adapted [81] for thedetermination by HG-ETAAS with in-atomizertrapping, achieving a detection limit of 4 ng L−1

for a sample volume of 20 mL. In this revisedprocedure, the selenium was loaded first onto thecolumn followed by an appropriate amount ofborohydride. No signal was obtained from anySe(VI) retained in the column.

6.4 Other elements

The so-called ‘fast reacting’ aluminum specieswere retained on an Amberlite XAD-2 (nonionic)column following a reaction time of 3 s with

8-hydroxyquinoline in a procedure devised byFairman and Sanz-Medel [82]. The retained Alwas eluted with 1 M HCl and determined byICP-AES. The procedure also incorporated apreconcentration of up to 18-fold giving a limitof detection of 2 µg L−1. It was shown that thenontoxic AlF2+ species was not included in the‘fast reacting’ retained aluminum. The methodwas also adapted for field sampling and fordetermination by ICP-MS [83].

Hulanicki and coworkers have developed pro-cedures for the selective retention of anti-mony(III) prior to determination by GFAAS [84]or FAAS [85]. In the first procedure, the Sb(III)was retained on C-18 as the chelate with ammo-nium pyrrolidine dithiocarbamate followed byelution with ethanol directly into the graphite fur-nace atomizer. Total antimony was determinedafter reduction with L-cysteine. The limit of detec-tion was 7 ng L−1. The stability of Sb(III) andSb(V) spiked into several natural matrices wasinvestigated, and it was found that, while bothspecies were stable in urine, Sb(III) was not sta-ble in tapwater. In the second method [85], Sb(III)was retained on a ‘DETA sorbent with grafteddiethylenetriamine groups’ then eluted with nitricacid directly into a flame atomizer to give adetection limit of 0.9 µg L−1. Total antimony wasdetermined by GF-AAS and thus Sb(V) was deter-mined by difference. Antimony species added to awell water which reached rocks formed during theoligocene period was determined.

An on-line UV decomposition procedure wasdevised by Comber et al. [86] to investigate thespeciation of copper in natural waters. Complex-ation with three model ligands, glycine, NTA andETDA, was investigated. With the lamp off, only84 %, 45 % and 2 % respectively of the copperwas collected by a column of Chelex-100 chelat-ing resin, but when the lamp was turned on, over90 % of the copper was collected. A procedure forthe speciation of copper and manganese in cow’smilk has been devised by Abollino et al. [87]in which four operationally defined ‘species’were measured. The separations were based on(a) precipitation with casein (species associatedwith proteins), (b) species retained on an anion

276 DETECTION

exchange column, (c) species retained on a chelat-ing cation exchange column and (d) species notassociated with proteins not retained by eithercolumn. The initial casein precipitation was per-formed off line, but the two ion exchange resinswere incorporated into an FI manifold, in whichthe sample could be directed to one or other ofthe columns followed by back-flushing with 2 MHCl and determination by ICP-AES. Total copperand casein-precipitated copper were determined byGF-AAS.

Several procedures for the differentiation ofiron(II) and iron(III) have been devised, and theearlier ones are summarized in ref. 10. Morerecently Bagheri et al. [88] retained Fe(III) selec-tively on 2-mercaptobenzimidazole loaded ontosilica gel; Fe(II) was not retained. The Fe(III)was removed with thiocyanate solution. Total ironwas determined after oxidation of Fe(II) withhydrogen peroxide. The limit of detection wasaround 1 µg L−1.

A procedure for separation of tetralkyllead andthe sum of inorganic lead and organolead specieshaving a smaller number of alkyl groups wasdeveloped by Naghmush et al. [89]. All of thespecies were retained on Cellex P, a functional-ized cellulose sorbent. The two groups of retainedspecies were eluted with ethanol and nitric acid,and determined by FAAS. The limit of detectionfor inorganic lead was 0.2 µg L−1 for a sample of50 mL loaded at 7 mL min−1. A somewhat morecomplicated procedure which provided greater res-olution among a similar set of analytes was devisedby Valcarcel and coworkers [90]. Inorganic leadwas precipitated as the chromate which was con-tinuously collected then dissolved in nitric acidand the lead determined by FAAS. The trialkylcations trimethyllead and triethyllead, in the fil-trate, were retained on a column of the fullereneC-60 following derivatization with diethyldithio-carbamate. This retention was selective dependingon the conditioning of the column: either n-hexaneor isobutyl methyl ketone was used.

Despite the considerable interest in the deter-mination of mercury species, there has been verylittle work published in which an FI solid-phaseextraction procedure has formed the basis of

the distinction between species. As long ago as1992, Wei and McLeod [91] devised a method inwhich methylmercury was selectively retained onsulfhydryl cotton. However, descriptions of furtherusage of this method are somewhat ambiguous.The original research group applied the proce-dure to field sampling of the Manchester ShipCanal [92], but also suggested that the materialcould be loaded with both inorganic Hg as wellas organomercury species to form a possible ref-erence material [93]. Cai et al. [94] described aprocedure in which methyl- and ethylmercury wereretained on sulfhydryl cotton followed by elu-tion and conversion to the bromides for subse-quent determination by GC with AFS detection.Apparently inorganic mercury was not retained.However, Yu and Yu recently described a pro-cedure [95] in which both inorganic mercury andlead species were retained on the material, whereasKwokal and Branica devised a procedure [96] forthe preconcentration of methylmercury. Both ofthese methods were applied to the analysis ofwaters. Frech and coworkers preconcentrated bothinorganic and organomercury as the dithiocarba-mate complexes prior to elution and determination(after butylation) by GC [97], or by HPLC withpost-column hydride generation [98].

Organotin compounds were retained on a silica-based C-18 column as the first step in a GCprocedure for the determination of these com-pounds in water developed by Szpunarlobinskaet al. [99]. The species were ethylated on the col-umn by the passage of a solution of tetraethylbo-rate, eluted with methanol and injected onto a GCcolumn for separation followed by detection bymicrowave-induced plasma atomic emission spec-trometry (MIP-AES). Grotti et al. [100] interfacedthe HPLC separation of butyltin compounds withgraphite furnace AAS via a HG manifold.

Vanadium(IV) and (V) were determined [101]in river water by a procedure in which V(IV)was selectively retained on a column of AmberliteXAD-7 as the complex with 2-(5-bromo-2-pyridylazo)-5-diethylaminophenol while V(V) wasmasked with 1,2-cyclohexanediaminetetraaceticacid. The retained vanadium was eluted withnitric acid and determined by ICP-OES. In the

OTHER PROCEDURES 277

absence of the masking agent both oxidation stateswere retained and so V(V) was determined bydifference. The limit of detection for a 10 mLsample was 20 ng L−1 for an instrument equippedwith an ultrasonic nebulizer.

7 OTHER PROCEDURES

Adsorption of insoluble derivatives on the innerwall of open tubular reactors forms the basisof preconcentration procedures for a number ofelements. When coupled with chemistry that isselective for a particular oxidation state of an ele-ment, such a procedure has the potential to providespeciation information. Nielsen and Hansen [102]devised such a procedure for the determinationof Cr(VI) by ETA-AAS. The Cr(VI) complexwith ammonium pyrrolidine dithiocarbamate wasretained on a knotted PTFE tubular reactor andthen eluted with 55 µL of ethanol directly intothe graphite furnace atomizer. A preconcentrationfor 60 s at 5 mL min−1 gave a signal enhancementof 20-fold and a detection limit of 4 ng L−1. Themethod has also been adapted for implementationin the sequential injection mode [103]. Gaspar andPosta showed [104] that PEEK tubing also col-lected this complex. They obtained a detectionlimit by FAAS of 2 µg L−1 for a sample vol-ume of 5 mL. In addition to the usual water sam-ples, they applied the method to the analysis ofcigarette ash. Yan et al. [105] retained the Fe(III)pyrrolidinedithiocarbamate (PDC) complex on theinterior of a PTFE reactor from a solution con-taining 0.07–0.4 M HCl, prior to elution with 1 Mnitric acid and detection by ICP-MS. When theacidity of the sample solution was decreased to0.001–0.004 M HCl, both Fe(II) and Fe(III) wereretained and thus Fe(II) could be determined bydifference. For loading at 5 mL min−1 for 30 s,a detection limit of 80 ng L−1 was obtained. Theenhancement factor was 12, the retention effi-ciency was 80 %, and the throughput was 21 h−1.This procedure was an adaptation of an earliermethod [106] for the determination of inorganicarsenic species by ICP-MS. The As(III)–PDCcomplex was retained on PTFE tubing, from solu-tions whose acidity ranged from 0.01 to 0.7 M with

respect to nitric acid, prior to elution with 1 Mnitric acid. Neither As(V) nor mono-nor dimethyl-arsenic species were retained. After reduction ofAs(V) to As(III) with L-cysteine, total inorganicarsenic was determined (again, the methylatedforms were not retained) and hence As(V) wasdetermined by difference. The detection limit was20–30 ng L−1.

In addition to the retention of organic com-plexes on the interior of such reactors, it has beenshown possible to collect metal hydroxide pre-cipitates and to exploit co-precipitation both asa means of preconcentration and as the basis ofa speciation scheme. Zou et al. [107] selectivelyretained Cr(III) by coprecipitation with lanthanumhydroxide, prior to dissolution in 0.5 M HCl anddetermination by FAAS. For a loading period of110 s the limit of detection was 0.8 µg L−1 and themethod was applied to the analysis of water andhuman hair. Nielsen et al. [108] collected Se(IV)with the same chemistry; but, after elution withHCl, the selenium was determined by HG-AAS.The detection limit was an impressive 5 ng L−1.

FI liquid–liquid extractions for speciation havenot been developed to any extent, probably becauseof the difficulty of achieving reliable phase separa-tion. Nielsen et al. [109] extracted the PDC com-plex of Cr(VI) into MIBK, 55 µL of which wasthen delivered to a graphite furnace for determina-tion by AAS to give a detection limit of 3 ng L−1.Phase separation was achieved in a small coni-cal PTFE vessel with a stainless steel base. Theprocedure was applied to various water samplesincluding the wastes from incineration and desul-furization plants.

In principle, it is possible to connect flowinjection detectors in series (as long as the onlydestructive detector is the last in the sequence),though there do not appear to be many proce-dures developed based on this concept. Girardand Hubert [110] described a procedure for thedetermination of chromium species in which amolecular absorption detector was connected inseries with a flame atomic absorption spectrom-eter. Cr(VI) was detected by visible absorptionspectrometry of the 1,5-diphenylcarbazide com-plex and total chromium was detected by AAS.

278 DETECTION

The method was applied to the analysis of stain-less steel welding dust. It was found that for amaterial with 30 000 mg L−1 of Cr in the solid, ofwhich 25 000 mg L−1 was ‘extractable’, the Cr(VI)content was 22 500 mg L−1. The method was basedclosely on a procedure first described by Lynchet al. [111] in 1984.

FI chemical vapor generation with cryotrappingcan be used as the first stage in a procedure inwhich the derivatives are subsequently separatedby GC and detected with an atomic spectromet-ric detector. A recent example of this methodol-ogy [112] demonstrates that considerable temporalseparation of the two stages is possible as thederivatives of arsenic, germanium, mercury, andtin with borohydride were formed in an FI systemand cryotrapped on board ship with subsequentanalysis in the laboratory by GC with ICP-MSdetection. Alkylarsenic compounds other than themethylated species were tentatively identified insamples taken from the Rhine estuary.

8 COMPARISON WITH OTHERSPECIATION PROCEDURES

The capability of FI-based speciation proceduresis somewhat limited when compared with the per-formance of procedures based on high perfor-mance liquid or gas chromatography coupled withelement-specific detection, in the sense that FI pro-cedures are only capable of providing informationabout a limited number of analyte species. FI pro-cedures become rather cumbersome when the goalof the analysis is to quantify four (or more) compo-nents. However, if the goal is to provide informa-tion about a limited number of components basedon, for example, the classifications ‘inorganic’,‘organic’, ‘toxic’, ‘nontoxic’, ‘fast reacting’, andso on, then FI procedures offer the advantages ofspeed, simplicity and relatively low cost. Whilea fully automated, computer-controlled FI systemwill be considerably more expensive than a manu-ally operated system, the capital and running costswill be considerably less than those of a highperformance liquid chromatograph. As other ana-lytical performance parameters, such as detection

limit are a function of the detection technique; toa first approximation these are the same for thetwo approaches. However, it should be borne inmind that HPLC procedures cause a substantialon-line dilution (maybe by as much as a factorof 100) and may thus be inferior to procedures inwhich there is preconcentration step (such as bytrapping a generated hydride on the interior of agraphite furnace). It should not be assumed thatuseful speciation information can only be obtainedby the combination of some high resolution sepa-ration technique coupled with plasma-source massspectrometric detection.

9 REFERENCES

1. Ruzicka, J. and Hansen, E. H., Flow Injection Analysis ,2nd edn, John Wiley & Sons, Inc., New York, 1988,p. 380.

2. Chalk, S. J., The Flow Analysis Database, http://www.fia.unf.edu/fad.lasso (accessed June 2001).

3. Sturgeon, R. E., J. Anal. At. Spectrom., 13, 351 (1988).4. Sanz-Medel, A. (Ed.), Flow Analysis with Atomic Spec-

trometric Detectors , Elsevier, Amsterdam, 1999.5. Beauchemin, D., J. Anal. At. Spectrom., 13, 1 (1988).6. Dempster, M. A. and Marcus, R. K., J. Anal. At. Spec-

trom., 14, 43 (1999).7. Valcarcel, M. and Luque de Castro, M. D., Non-

chromatographic Continuous Separation Techniques ,Royal Society of Chemistry, Cambridge, 1991.

8. Fang, Z., Flow Injection Separation and Preconcentra-tion , VCH, New York, 1993.

9. Welz, B., Spectrochim. Acta, Part B , 53, 169 (1998).10. Campanella, L., Pyrzynska, K. and Trojanowicz, M.,

Talanta , 43, 825 (1996).11. Oygard, J. K., Lundebye, A. K. and Julshamn, K., J.

AOAC Int., 82, 1217 (1999).12. Woolson, E. A., Axley, J. H. and Kearney, P. C., Soil

Sci. Soc. Am. Proc., 37, 254 (1973).13. Iu, K. L., Pulford, I. D. and Duncan, H. J., Anal. Chim.

Acta , 106, 319 (1979).14. Van Loon, J. C., Selected Methods of Trace Metal

Analysis , Wiley-Interscience, New York, 1985, p. 285.15. Lin, S. L., Man, H. and Shu, Z., Lab. Robotics

Automat., 12, 86 (2000).16. Pilhar, B. and Costa, L., Anal. Chim. Acta, 114, 275

(1980).17. Burguera, J. L., Burguera, M., Rivas, C. and Carrero,

P., Talanta , 45, 531 (1998).18. Dedina, J. and Tsalev, D., Hydride Generation Atomic

Absorption Spectrometry , John Wiley and Sons, Ltd,Chichester, 1995.

REFERENCES 279

19. Dedina, J., Flow methods in gas–liquid separations,in Flow Analysis with Atomic Spectrometric Detectors ,Sanz-Medel, A. (Ed.), Elsevier, Amsterdam, 1999.

20. Howard, A. G., J. Anal. At. Spectrom., 12, 267 (1997).21. Tsalev, D. L., J. Anal. At. Spectrom., 14, 147 (1999).22. Jin, Q., Liang, F., Zhang, H., Zhao, L., Huan, Y. and

Song, D., Trends Anal. Chem., 18, 479 (1999).23. Burguera, M. and Burguera, J. L., Quim. Anal., 15, 112

(1996).24. Feng, Y. L., Narasaki, H., Chen, H. Y. and Tian, L. C.,

Anal. Chim. Acta, 386, 297 (1999).25. Ulrich, N., Anal. Chim. Acta, 417, 201 (2000)26. Deng, T. L., Chen, Y. W. and Belzile, N., Anal. Chim.

Acta , 432, 293 (2001).27. Nielsen, S. and Hansen, E. H., Anal. Chim. Acta, 343,

5 (1997).28. Krenzelok, M. and Rychlovsky, P., Coll. Czech. Chem.

Commun., 63, 2027 (1998).29. Willie, S. N., Spectrochim. Acta, Part B , 51, 1781

(1996).30. Cabon, J. Y. and Cabon, N., Anal. Chim. Acta, 418, 19

(2000).31. Cabon, J. Y. and Cabon, N., Fresenius’ J. Anal. Chem.,

368, 484 (2000)32. Rude, T. R. and Puchelt, H., Fresenius’ J. Anal. Chem.,

350, 44 (1994).33. Le, X. C., Cullen, W. R. and Riemer, K. J., Anal. Chim.

Acta , 285, 277 (1994).34. Tsalev, D. L., Sperling, M. and Welz, B., Talanta , 51,

1059 (2000).35. Tsalev, D. L., Sperling, M. and Welz., B., Spectrochim.

Acta, Part B , 55, 339 (2000).36. Pyell, U., Dworkschak, A., Nitcshke, E. and Neid-

hart, B., Fresenius’ J. Anal. Chem., 363, 495 (1999).37. Chatterjee, A., Das, D., Mandal, B. K., Chowdhury,

T. R., Samanta, G. and Chakraborti, D., Analyst , 120,643 (1995).

38. Fernandez, M. G. C., Palacios, M. A. and Camara, C.,Anal. Chim. Acta, 283, 386 (1993).

39. Stripeikis, J., Costa, P., Tudino, M. and Troccoli, O.,Anal. Chim. Acta, 408, 191 (2000).

40. Burguera, J. L., Carrero, P., Burguera, M., Rondon, C.,Brunetto, M. R. and Gallignani, M., Spectrochim. Acta,Part B , 51, 1837 (1996).

41. Gallignani, M., Valero, M., Brunetto, M. R., Burguera,J. L., Burguera, M. and de Pena, Y. P., Talanta , 52,1051 (2000).

42. He, Y. Z., Azouzi, H., Cervera, M. L. and de laGuardia, M., J. Anal. At. Spectrom., 13, 1291 (1988).

43. Moreno, A. E., Perez-Conde, C. and Camara, C., J.Anal. At. Spectrom., 15, 681 (2000).

44. Varade, C. M. R. and Luque de Castro, M. D. L., J.Anal. At. Spectrom., 13, 787 (1998).

45. Cabon, J. Y. and Erler, W., Analyst , 123, 1565 (1998).46. Vilano, M., Padro, A., Rubio, R. and Rauret, G., J.

Chromatogr. A, 819, 211 (1998).

47. La Fuente, J. M. G., Dlaska, M., Sanchez, M. L. F. andSanz-Medel, A., J. Anal. At. Spectrom., 13, 423 (1998).

48. Chatterjee, A., Shibata, Y., Yoneda, M., Banerjee, R.,Uchida, M., Kon, H. and Morita, M., Anal. Chem., 73,3181 (2001).

49. Chatterjee, A. and Shibata, Y., Anal. Chim. Acta, 398,273 (1999).

50. Gallignani, M., Bahsas, H., Brunetto, M. R., Burgu-era, M., Burguera, J. L. and de Pena, Y. P., Anal.Chim. Acta , 369, 57 (1988).

51. Rio-Segade, S. and Bendicho, C., J. Anal. At. Spectrom.,14, 263 (1999).

52. Rio-Segade, S. and Bendicho, C., Spectrochim. Acta,Part B , 54, 1129 (1999).

53. Capelo, J. L., Lavilla, I. and Bendicho, C., Anal. Chem.,72, 4979 (2000).

54. Tao, G. H., Willie, S. N. and Sturgeon, R. E., J. Anal.At. Spectrom., 14, 1929 (1999).

55. Burguera, J. L., Quintana, I. A., Salager, J. L., Bur-guera, M., Rondon, C., Carrero, P., de Salager, R. A.and de Pena, Y. P., Analyst , 124, 593 (1999).

56. Puk, R. and Weber, J. H., Anal. Chim. Acta, 292, 175(1994).

57. Palmer, C. D., Ph.D. Dissertation, University of Mas-sachusetts, Amherst, 2001.

58. Palmer, C. D., Rio-Segade, S. and Tyson, J. F., work inprogress.

59. Mota, J. P. V., de la Campa, M. R. F. and Sanz-Medel, A., J. Anal. At. Spectrom., 13, 431 (1998).

60. Infante, H. G., Sanchez, M. L. F. and Sanz-Medel, A.,J. Anal. At. Spectrom., 15, 519 (2000).

61. Bettmer, J. and Cammann, K., Appl. Organomet. Chem.,8, 615 (1994).

62. Tsalev, D. L., Sperling, M. and Welz, B., Spectrochim.Acta, Part B , 55, 339 (2000).

63. Pozebon, D., Dressler, V. L., Neto, J. A. G. and Cur-tius, A. J., Talanta , 45, 1167 (1998).

64. Tyson, J. F., J. Anal. At. Spectrom., 14, 169 (1999).65. Burguera, J. L., Burguera, M. and Rivas, C., Quim.

Anal., 16, 165 (1997).66. Yalcin, S. and Le, X. C., J. Environ. Monitor., 3, 81

(2001).67. Sperling, M., Xin, X. F. and Welz, B., Analyst , 117,

629 (1992).68. Sperling, M., Xu, S. K. and Welz, B., Anal. Chem., 64,

3101 (1992).69. Vassileva, E., Analysis , 28, 878 (2000).70. Yu, J. C., Wu, X. J. and Chen, Z. L., Anal. Chim. Acta,

436, 59 (2001).71. Naghmush, A. M., Pyrzynska, K. and Trojanowicz, M.,

Anal. Chim. Acta, 288, 247 (1994).72. Jiminez, M. S., Martin, L., Mir, J. M. and Castillo,

J. R., At. Spectrosc., 17, 201 (1996).73. Kelko-Levai, A., Varga, I., Zih-Perenyi, K. and Lasztity,

A., Specrochim. Acta, Part B , 54, 827 (1999).74. Rao, T. P., Karthikeyan, S., Vijayalekshmy, B. and

Iyer, C. S. P., Anal. Chim. Acta, 369, 69 (1998).

280 DETECTION

75. Cespon-Romero, R. M., Yebra-Biurrun, M. C. andBermejo-Barrero, M. P., Anal. Chim. Acta, 327, 37(1996).

76. Larraya, A., Cobofernandez, M. G., Palacios, M. A. andCamara, C., Fresenius’ J. Anal. Chem., 350, 667 (1994).

77. Pyrzynska, K., Drzewicz, P. and Trojanowicz, M., Anal.Chim. Acta , 363, 141 (1998).

78. Bryce, D. W., Izquierdo, A. and Luque De Castro,M. D., J. Anal. At. Spectrom., 10, 1059 (1995).

79. Yan, X. P., Sperling, M. and Welz, B., Anal. Chem., 71,4353 (1999).

80. Carrero, P. E. and Tyson, J. F., Analyst , 122, 915(1997).

81. Carrero, P. E. and Tyson, J. F., Spectrochim. Acta, PartB , 53, 1931 (1998).

82. Fairman, B. and Sanz-Medel, A., Fresenius’ J. Anal.Chem., 355, 757 (1996).

83. Fairman, B. and Sanz-Medel, A., J. Anal. At. Spectrom.,10, 281 (1995).

84. Garbos, S., Rzepecka, M., Bulska, E. and Hulanicki, A.,Spectrochim. Acta, Part B , 54, 873 (1999).

85. Garbos, S., Bulska, E. and Hulanicki, A., At. Spectrosc.,21, 128 (2000).

86. Comber, M. H. I., Eales, G. J. and Nicholson, P. J. D.,J. Automat. Chem., 14, 5 (1992).

87. Abollino, O., Aceto, M., Bruzzoniti, M. C., Mentasti, E.and Sarzanini, A., Anal. Chim. Acta, 375, 299 (1998).

88. Bagheri, H., Gholami, A. and Najafi, A., Anal. Chim.Acta , 424, 233 (2000).

89. Naghmush, A. M., Pyrzynska, K. and Trojanowicz, M.,Talanta , 42, 851 (1995).

90. Baena, J. R., Gallego, M. and Valcarcel, M., Spec-trochim. Acta , 54, 1869 (1999).

91. Wei, J. and McLeod, C. W., Talanta , 39, 1537 (1992).92. Jian, W., Mena, M. L., McLeod, C. W. and Rollins, J.,

Int. J. Environ. Anal. Chem., 57, 99 (1994).93. Mena, M. L. and McLeod., C. W. Mikrochim. Acta,

123, 103 (1996).

94. Cai, Y., Jaffe, R., Alli, A. and Jones, R. D., Anal. Chim.Acta , 334, 251 (1996).

95. Yu, F. and Yu, T. L., Spectrosc. Spectr. Anal., 20, 898(2000).

96. Kwokal, Z. and Branica, M., Croat. Chem. Acta, 73, 97(2000).

97. Emteborg, H., Baxter, D. C. and Frech, W., Analyst ,118, 1007 (1993).

98. Xin, X. F., Frech, W., Hoffmann, E., Ludke, C.and Skole, J., Fresenius’ J. Anal. Chem., 361, 761(1998).

99. Szpunarlobinska, J., Ceulemans, M., Lobinski, R. andAdams, F. C., Anal. Chim. Acta, 278, 99 (1993).

100. Grotti, M., Rivaro, P. and Franche, R., J. Anal. At.Spectrom., 16, 270 (2001).

101. Wuilloud, R. G., Wuilloud, J. C., Olsina, R. A. andMartinez, L. D., Analyst , 126, 715 (2001).

102. Nielsen, S. and Hansen, E. H., Anal. Chim. Acta, 366,163 (1998).

103. Nielsen, S. and Hansen, E. H., Anal. Chim. Acta, 422,47 (2000).

104. Gaspar, A. and Posta, J., Anal. Chim. Acta, 354, 151(1997).

105. Yan, X. P., Hendry, N. J. and Kerrich, R., Anal. Chem.,72, 1879 (2000).

106. Yan, X. P., Kerrich, R. and Hendry, N. J., Anal. Chem.,70, 4736 (1998).

107. Zou, H. F., Xu, S. K. and Fang, Z. L., At. Spectrosc.,17, 112 (1996).

108. Nielsen, S., Sloth, J. J. and Hansen, E. H., Analyst , 121,31 (1996).

109. Nielsen, S. C., Sturup, S., Spliid, H. and Hansen, E. H.,Talanta , 49, 1027 (1999).

110. Girard, L. and Hubert, J., Talanta , 43, 1965 (1996).111. Lynch, T. P., Kernoghan, N. J. and Wilson, J. N., Ana-

lyst , 109, 839 (1984).112. Tseng, C. M., Amouroux, D., Brindle, I. D. and Donard,

O. X. F., J. Environ. Monitor., 2, 603 (2000).

5.3 Detection by ICP-Mass Spectrometry

Frank VanhaeckeLaboratory for Analytical Chemistry, Ghent University, Belgium

Gunda KollenspergerAnalytical Chemistry Research Group, BOKU Universitat fur Bodenkultur Wien, Vienna, Austria

1 Inductively Coupled Plasma-MassSpectrometry . . . . . . . . . . . . . . . . . . . . . . 2811.1 Introduction . . . . . . . . . . . . . . . . . . . 2811.2 Operating principle . . . . . . . . . . . . . . 2811.3 Figures of merit . . . . . . . . . . . . . . . . 2831.4 Spectral interferences . . . . . . . . . . . . 284

1.4.1 Introduction . . . . . . . . . . . . . . 2841.4.2 Cool plasma conditions . . . . . 2851.4.3 Aerosol desolvation . . . . . . . . 2861.4.4 Multipole collision cell . . . . . 2861.4.5 Dynamic reaction cell . . . . . . 2861.4.6 High mass resolution . . . . . . . 287

1.5 Nonspectral interferences . . . . . . . . . 2872 Calibration . . . . . . . . . . . . . . . . . . . . . . . . 289

2.1 Traditional approaches . . . . . . . . . . . 2892.2 Isotope dilution . . . . . . . . . . . . . . . . 290

3 Use of the ICP as a Soft Ionization Source 293

4 High Performance Liquid Chromatography(HPLC)-ICPMS . . . . . . . . . . . . . . . . . . . . 2944.1 Coupling . . . . . . . . . . . . . . . . . . . . . 2944.2 Illustrative applications . . . . . . . . . . . 296

5 Gas Chromatography (GC)-ICPMS . . . . . . 2995.1 Coupling . . . . . . . . . . . . . . . . . . . . . 2995.2 Illustrative applications . . . . . . . . . . . 300

6 Capillary Electrophoresis (CE)-ICPMS 3046.1 Coupling . . . . . . . . . . . . . . . . . . . . . 3046.2 Illustrative applications . . . . . . . . . . . 306

7 Supercritical Fluid Chromatography(SFC)-ICPMS . . . . . . . . . . . . . . . . . . . . . . 3087.1 Coupling . . . . . . . . . . . . . . . . . . . . . 3087.2 Illustrative applications . . . . . . . . . . . 309

8 Alternative Approaches . . . . . . . . . . . . . . . 3099 References . . . . . . . . . . . . . . . . . . . . . . . . 310

1 INDUCTIVELY COUPLEDPLASMA-MASS SPECTROMETRY

1.1 Introduction

Inductively coupled plasma-mass spectrometry(ICPMS) is a remarkably powerful and versatiletechnique for (ultra)trace element determination,characterized by extremely low limits of detection(LODs), a wide linear dynamic range, multiele-ment capabilities, surveyable spectra and a highsample throughput. As such, ICPMS cannot beused for elemental speciation, as all molecules

introduced into the high temperature ion source(an Ar ICP) are broken down into atoms, whichare subsequently ionized. However, if the differ-ent species of interest can be separated from oneanother before their introduction into the plasma,ICPMS can be used as a highly sensitive andelement-specific on-line multielement detector.

1.2 Operating principle

Although the use of laser ablation or – to a lesserextent – electrothermal vaporization permits the

282 DETECTION

direct analysis of solid samples, ICPMS is mainlyintended for the analysis of (aqueous) sample solu-tions. Traditionally, such a sample solution is con-verted into an aerosol by means of a pneumaticnebulizer. In order to ensure a stable plasma andan efficient atomization and ionization in the ICP,the larger droplets (d > 10 µm) are removed fromthe sample aerosol by means of a spray chamber.Subsequently, the aerosol is swept by the Ar neb-ulizer or carrier gas into the ICP, which can beconsidered as an extremely hot (ionization tem-perature approximately 7500 K) electrical flame,generated at the end of a quartz torch. During theirresidence in this ICP, the droplets are desolvatedand the sample molecules are broken down intoatoms, which are subsequently excited and ionized.For the majority of elements, the efficiency of ion-ization in an Ar ICP exceeds 90 % (Figure 5.3.1),and in spite of their high first ionization potential,important metalloids and nonmetals, such as As,

Se, S or Cl, are still sufficiently ionized to allowsensitive determination.

Since the ICP is operated under atmosphericpressure, whereas in the mass spectrometer a highvacuum is required, an interface between bothcomponents is necessary (Figure 5.3.2). This inter-face consists of two successive, coaxial and water-cooled cones, with a small central aperture – thesampling cone and the skimmer. As a result ofthe difference in pressure between the expansionchamber – the region between the sampling coneand the skimmer – and the ion source, a fractionof the ICP is extracted into the interface regionand undergoes supersonic expansion [3]. Becauseof the sudden drop in particle density, ion–electronrecombination or other reactions are avoided, suchthat the composition of the extracted gas is ‘frozen’and hence, is representative of that of the ICP. Themajority of the extracted gas is subsequently evac-uated by means of a vacuum pump, but a central

1.1

1.0

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

degr

ee o

f ion

izat

ion

ionization energy (eV)

3 5 7 9 11 13 15

Figure 5.3.1. Ionization efficiency in the ICP as a function of the analyte’s first ionization energy [1]. Reprinted from AnIntroduction to Analytical Atomic Chemistry, L. Ebdon, E. H. Evans (Ed.), A. Fisher and S. J. Hill, 1998, John Wiley & SonsLimited. Reproduced with permission.

INDUCTIVELY COUPLED PLASMA-MASS SPECTROMETRY 283

ELANCPU

CEM

Quadrupolelon Lenses

TurboPump

TurboPump

MechPump

MechPump

System Computer

ICP Torch Spray Chamber

Sample

Argon

RF Supply

Figure 5.3.2. Schematic diagram of a quadrupole-based ICPMS instrument [2]. Reproduced by permission of PerkinElmer Sciex.

beam enters a higher vacuum region by passingthrough the skimmer aperture. This beam consistsof ions, electrons and neutral particles. A nega-tively charged extraction lens selectively attractsthe positive ions, which are subsequently trans-ported and introduced as efficiently as possibleinto the mass analyzer. The latter is accomplishedby the ion lens system, the construction and com-plexity of which differ markedly from one type ofinstrument to another.

Originally, all ICP-mass spectrometers availablewere equipped with a quadrupole filter for massanalysis. Although nowadays instruments equippedwith a double-focusing sector field mass spec-trometer, a time-of-flight analyzer or even an iontrap are also commercially available, the majority(approximately 90 %) of instruments used world-wide are still equipped with a quadrupole fil-ter. Such a quadrupole filter acts as a bandpassmass filter and transmits only those ions with amass-to-charge ratio within a narrow mass window(approximately 1 u). Of course, by changing thevoltages applied to the quadrupole rods, the posi-tion of the mass window can be selected and theoperator has the choice between scanning of theentire mass spectrum or a smaller mass region onthe one hand or monitoring the intensity of a lim-ited number of pre-selected analyte signals (peakjumping or hopping) on the other.

Finally, the ions transmitted by the massanalyzer have to be detected. The majority ofinstruments use an electron multiplier for thispurpose, although some manufacturers equip theirinstrumentation with a Daly-type detector. Anelectron multiplier is an extremely sensitive detec-tion device, permitting each individual ion to becounted. The intensity of each output signal iscompared with that of a threshold to suppress thebackground noise. Photon noise on the other handis reduced by using a photon or shadow stop and/orby mounting the detector or the entire mass ana-lyzer off axis. In order to cope with higher countrates, electron multipliers can also be operated inanalog mode, where ions are no longer counted,but an output signal, the intensity of which is pro-portional to the intensity of the ion beam strikingthe detector surface, is measured instead.

1.3 Figures of merit

LODs of course vary among instruments (governedby background noise and sensitivity) and from oneelement to the other (governed by mass number,ionization potential and the isotopic abundance ofthe isotope monitored). When not limited by avery high ionization potential (e.g., Cl, Br) or anincreased background due to blank contamination

284 DETECTION

and/or interfering ions (see later), low- to sub-ng L−1 LODs can be obtained. Owing to theextremely low background observed with sectorfield ICPMS (<0.2 counts s−1 vs. <1–10 countss−1 for quadrupole-based instruments) and its highion transmission efficiency, the instrumental LODwith this type of instrumentation is ≤1 pg L−1. Inmany instances, it is no longer the instrumentaldetection power, but the purity of the reagents andrecipients used that limits the LODs attainable.Additionally, modern ICPMS instruments offer alinear dynamic range of 8 to 9 orders of magnitude.Short-term RSDs on signal intensities of ≤2 %are typical, while the isotope ratio precision is≥0.05 %. Although the robustness of ICPMSinstruments has been improved during the pastdecade, it is not advisable to introduce solutionswith a total content of dissolved solids exceeding1–2 g L−1.

While quadrupole-based ICPMS instrumentsshow only unit mass resolution (peak widthapproximately 0.5 u over the entire mass spec-trum), sector field ICPMS also permits measure-ments at a higher mass resolution (Rmaximum ≥10 000). Hence, in many instances, spectral over-lap of signals from ions showing the same nominalmass can be avoided, permitting lower LODs tobe obtained for traditionally problematic elements.This topic will be discussed into more detail in alater section.

With quadrupole-based ICPMS instrumentation,the complete mass range – from Be to U – canbe scanned in approximately 0.1 s. Owing tohysteresis of the magnetic sector, the scanningspeed with a sector field mass spectrometer islower and scanning the full mass range takes about0.5 s. With a recently introduced newly developedmagnetic field regulator however, the scanningspeed can be enhanced to a value almost similarto that of quadrupole-based instrumentation. WithTOF-ICPMS on the other hand, up to 30 000full mass spectra can be recorded every second.The sensitivity of TOF-ICPMS instrumentation,however, is inferior to that of quadrupole-basedICPMS by 1–2 orders of magnitude. Additionally,a recent investigation showed that, even whenworking with transient signals only lasting a few

seconds, the ‘simultaneous’ monitoring of up to20 mass-to-charge ratios – a number hardly everexceeded in speciation work – is feasible withquadrupole-based instrumentation [4].

Finally, the purchase price (2002) of a quadru-pole-based or TOF-based ICPMS instrument isin the order of 200 000 ¤; that of a sector fieldinstrument is about twice as high.

1.4 Spectral interferences

1.4.1 Introduction

Very soon after the introduction of ICPMS, itbecame clear that nonspectral (matrix-induced sig-nal suppression or enhancement) and spectral(overlap of the signals of ions showing a differencein mass <0.5 u) interferences were its most promi-nent disadvantages. While, in most cases, non-spectral interferences could be fairly easily copedwith – e.g., by means of sample dilution, the useof (a) carefully selected internal reference(s) orapplication of standard additions or isotope dilu-tion instead of (an) external standard(s) for cali-bration – the occurrence of spectral interferencesproved to be more troublesome. Despite the hightemperature in the ICP, molecular ions, originat-ing from the plasma gas (Ar), entrained air, thesolvent and/or the matrix, occur in ICPMS andtheir signal may complicate the mass spectrumand analyte quantification to a large extent. Espe-cially for complex matrices and in the lower massrange (≤80 u), obtaining accurate results or suf-ficiently low LODs is therefore not self-evident.As an illustration, Table 5.3.1 presents a numberof elements – the majority of which are of interestfrom a biomedical and/or environmental point ofview – and the molecular ions potentially givingorigin to spectral overlap with the signal from themost abundant analyte isotope.

Self-evidently, throughout the years, effortshave been made to avoid, reduce or correct for theeffects of spectral interferences. In the first place,whenever possible, problems should be avoidedby appropriate selection of the nuclide(s) moni-tored, although selecting an isotope with a lower


Table 5.3.1. Examples of spectral interferences potentiallyencountered using ICPMS. For each element, only themajor isotope (highest isotopic abundance) is given. Theactual occurrence of spectral overlap depends on the matrixcomposition.

Element Major isotopeand isotopic

abundance (%)

Molecular ions (potentially)causing spectral overlapat low mass resolution

Al 27Al, 100 % 12C14NH+, 13C14N+Si 28Si, 92.2 % 14N2

+, 12C16O+P 31P, 100 % 14N16OH+S 32S, 95.0 % 16O2

+K 39K, 100 % 38ArH+Ca 40Ca, 96.9 % 40Ar+Sc 45Sc, 100 % 12C16O2H+, 28Si16OH+, 29SiO+Ti 48Ti, 73.8 % 32S16O+, 31P16OH+V 51V, 99.8 % 35Cl16O+, 37Cl14N+Cr 52Cr, 83.8 % 40Ar12C+, 35Cl16OH+Fe 56Fe, 91.7 % 40Ar16O+, 40Ca16O+Cu 63Cu, 69.2 % 40Ar23Na+, 31P16O2

+Zn 64Zn, 48.6 % 32S16O2

+, 31P16O2H+As 75As, 100 % 40Ar35Cl+Se 80Se, 49.6 % 40Ar2

+

isotopic abundance obviously leads to a deteriora-tion in sensitivity and detection power. Especiallyin elemental speciation work, where total elementconcentrations that are already low are furtherdistributed over different species, such a reduc-tion in sensitivity is often not acceptable. Alsothe conditions of sample preparation and speciesseparation should be selected such that the occur-rence of spectral interferences is avoided to thelargest possible extent. In relatively simple cases,mathematical correction has also been used suc-cessfully, although very rapidly one may run intorather complex calculation schemes, requiring theuse of dedicated computer programs. Sometimes,even blank correction, preferably using a matrix-matched blank, may be sufficient. However, veryoften, more effective, technically more complexand unfortunately more expensive measures arerequired for coping with spectral overlap.

1.4.2 Cool plasma conditions

With all modern ICPMS instruments, the plasmacan be operated under cool plasma conditions [3,5] – obtained by using a low RF power and anincreased nebulizer gas flow rate. In order to also

enable operation of the ICP under cool plasmaconditions with instruments equipped with a loadcoil that is not electrically balanced, insertion ofa grounded metal plate in between the coil andthe ICP torch is necessary to capacitively decoupleboth components [5]. Otherwise, the occurrenceof intense secondary discharges counteracts thedesired effect. The use of cool plasma conditionsleads to a substantial reduction in the intensityof Ar+ and Ar-based ions and hence makes thedetermination of ultratrace amounts of traditionallydifficult elements, such as K, Ca and Fe, possi-ble. Unfortunately, the use of these cool plasmaconditions also brings about important disadvan-tages: elements characterized by a high ionizationpotential are no longer efficiently ionized, the sig-nal intensity of other types of molecular ions (e.g.,oxide ions) increases and matrix effects becomemore pronounced. Hence, cool plasma conditionsare only useful for the determination of relativelylight elements in fairly clean and simple matricesand are predominantly used in the semiconductorindustry for the analysis of high purity chemicalreagents. Their use in speciation work has onlyseldom been reported.

Vanhaecke et al. [6] evaluated the merits ofcool plasma conditions to cope with the spec-tral overlap of the signals of ArC+ and ClO(H)+with 52Cr+ and 53Cr+ analyte signals. This over-lap hampered accurate quantification of Cr(III) andCr(VI) species, separated from one another usinga microbore anion exchange HPLC column, inindustrial process solutions. They came to the con-clusion that, while the intensity of ArC+ couldbe sufficiently suppressed, the ratio ClO(H)+/Cr+deteriorated on switching to cool plasma condi-tions. Therefore, these authors preferred to use ahigher mass resolution (see later) instead. Despitethe important loss in signal intensity inherent toan increase in mass resolution, sub-µg L−1 LODswere obtained. The increased tendency of oxideformation observed under cool plasma conditionswas used to advantage by Divjak and Goessler [7]in sulfur speciation. Sulfide, sulfite, sulfate andthiosulfate were separated from one another usingan anion exchange HPLC column, while detectionwas accomplished using ICPMS. Since the signal

286 DETECTION

of 32S+ suffers from a major spectral overlap by theO2

+ signal, the instrument was tuned for a maxi-mum SO+/S+ ratio and the 32SO+ ion signal wasmonitored for analytical purposes instead. In thecase of highly saline samples, the determinationof sulfide was reported to be hindered by severesignal suppression.

1.4.3 Aerosol desolvation

Aerosol desolvation reduces the occurrence ofoxide- and Cl-based ions. Since the introductionof ICPMS, cooled spray chambers – originally bymeans of a thermostated water jacket, later on alsoby means of Peltier elements – have been used tolimit the introduction of water (vapor) into theplasma. More efficient ways of sample introduc-tion, e.g., thermospray or ultrasonic nebulization,require the use of more effective desolvation sys-tems [8]. Basically, these units consist of a heatedpart, in which the solvent is vaporized and a cooledpart, in which the solvent vapor is condensed andremoved. Membrane desolvation is a more novelapproach. In a membrane desolvator, the Ar carriergas loaded with sample aerosol is directed througha cylinder, the walls of which are made out of amicroporous material. Heated Ar (the sweep gas)flows around this central cylinder in the oppo-site direction to the carrier gas flow. Because ofthe elevated temperature, the solvent is vaporizedand the gaseous solvent molecules can penetratethe semipermeable membrane and are carried offby the aforementioned sweep gas. In addition toreducing the level of oxide- and Cl-based molecu-lar ions, application of a membrane desolvator alsopermits the direct analysis of (volatile) organic sol-vents [9]. In the latter case, addition of O2 (at a lowgas flow rate) to the plasma is advisable to avoidcarbon deposition on the torch and the interface.

1.4.4 Multipole collision cell

Several manufacturers produce ICPMS instrumen-tation equipped with a multipole collision cell.Such a collision cell consists of six (hexapole) oreight (octopole) rods to which an RF-only voltage

is applied. In order to cope with the occurrence ofAr-based molecular ions, which give rise to spec-tral overlap, H2 can be added to the cell. As aresult of the occurrence of selective ion–moleculereactions, e.g., charge, atom and proton transfer,the signal intensities of Ar+ and of Ar-containingmolecular ions such as ArC+, ArO+, ArCl+ andAr2

+ are suppressed by three or more orders ofmagnitude, such that trace levels of Ca, Cr, Fe,As and Se can be accurately determined [10, 11].Application of He as a thermalization gas resultsin both an improvement in the ion transmissionefficiency (collisional focusing) and an enhancedreaction efficiency.

Marchante-Gayon et al. [12] used reversedphase and ion-pairing HPLC to separate the Sespecies present in human urine from one another.By introducing H2 as a reaction gas and He asa thermalization gas in the hexapole collision cellof the ICPMS detector, the signal intensity of theargon dimer (Ar2

+) ion, which normally precludesthe use of the most abundant Se isotope 80Se(isotopic abundance: 49.6 %), could be suppressedby orders of magnitude. This experimental set-up was used for urinary Se speciation before andafter intake of commercially available nutritionalsupplements.

1.4.5 Dynamic reaction cell

The dynamic reaction cell is a similar approach,but owing to the use of a quadrupole unit instead ofa hexapole or octopole configuration the cell can besimultaneously used as a bandpass mass filter. As aresult, the lifetime of newly created and unwantedspecies can be limited, such that they neither giverise to a signal in the mass spectrum nor takepart in further reactions [13]. As a result, a largerselection of gases, e.g., NH3, CH4, or NO2, canalso be used, such that the application range is notlimited to Ar+ and Ar-containing molecular ions.Spectral overlap due to the occurrence of an oxideion can, e.g., be avoided by converting the latterinto a dioxide or higher oxide ion [14] and similaratom transfer reactions have also been shown to besuccessful for overcoming isobaric overlap [15].Since hexapole or octopole arrangements do not


show mass filtering capabilities, an extension of theapplication range of collision cells is aimed at viadiscrimination between analyte ions and newly (in-cell) created ions on the basis of their energy [16].

By using CH4 as a reaction gas in a dynamiccell, Sloth and Larsen [17] were able to suppressthe signal intensity of the 40Ar2

+ ion by approxi-mately five orders of magnitude. Hence, monitor-ing of the major isotope of Se (80Se) was enabledand sensitive detection of selenoamino acids, sep-arated by cation exchange HPLC, could be accom-plished (absolute limit of detection: approximately5 pg as Se).

1.4.6 High mass resolution

Finally, the most universal strategy to cope withspectral overlap is created by using a double-focusing sector field mass spectrometer instead ofthe more traditional quadrupole filter in ICPMSinstrumentation. Sector field mass spectrome-ters offer a far higher mass resolution than doquadrupole filters, such that ions differing inmass by only a fraction of a mass unit canstill be separated from one another (Figure 5.3.3)and straightforward quantification becomes self-evident, despite the presence of molecular ordoubly charged ions of the same nominal mass-to-charge ratio. The maximum mass resolution (≥10 000) offered by present-day sector field ICPMSinstruments is sufficient to overcome the largemajority of spectral interferences known, the onlylimitation of this approach being the substantialloss in ion transmission efficiency observed onincreasing the mass resolution.

Rottmann and Heumann [19] were the first toreport on the use of a sector field ICPMS instru-ment operated at a higher mass resolution as adetector in a speciation study. This study aimedat obtaining insight into the association of metalswith different fractions of the dissolved organicmatter in natural waters and applied a combina-tion of an HPLC system, equipped with a sizeexclusion column, and an ICPMS instrument forthis purpose. Interference-free determination ofFe, whose accurate quantification is hampered atlow mass resolution due to spectral overlap of

the signals of 56Fe+ and 40Ar16O+, was accom-plished at a resolution setting of 3000. Cabezueloet al. [20] used anion exchange fast protein liq-uid chromatography coupled to sector field ICPMSfor studying Al species in blood serum of bothureamic patients and healthy subjects. Accuratedetermination of low Al concentrations was hin-dered by the occurrence of 13C14N+, 12C15N+ and12C14NH+. These molecular ions mainly originatedfrom the ammonium acetate mobile phase used.At a higher mass resolution setting (R = 3000)however, Al could be measured interference free.Hence, despite the fact that its basal level in nor-mal serum is below 5 µg L−1 (total element con-centration), speciation of Al was accomplished.Vanhaecke et al. [6] used a higher mass resolu-tion to ensure accurate results in Cr speciationwork. Cr(III) and Cr(VI) were established to co-elute from the anion exchange column with Cl−and HCO3

−, respectively, such that the signalsof both 52Cr+ and 53Cr+ suffered from spectraloverlap, due to the occurrence of 40Ar12,13C+,35Cl16OH+ and 37Cl16O+ ions. At a resolution set-ting of 3000, both Cr signals could be measuredinterference free.

1.5 Nonspectral interferences

While nonspectral interferences – matrix-inducedsignal suppression or enhancement – can often befairly easily coped with when total element con-centrations have to be determined, the situationis more complicated in the context of elemen-tal speciation.

Often, the elements to be speciated are presentat a low level only and as the total element contentis further ‘distributed’ over a number of species,dilution is in many cases unacceptable. For com-plex samples, also matrix matching – imitation ofthe sample matrix by adding high purity chemi-cals to the standard solutions – is not self-evident.When aiming at the determination of total elementconcentrations, both nonspectral interferences andsignal instability and/or drift can often be correctedfor by using a carefully selected internal reference.To all blank, sample and standard solutions, an

288 DETECTION

56

Intensity56Fe+ + 40Ar 16O+

55 57 m/z

m/z

Intensity

40Ar 16O+

56Fe+

55.935 55.957

Figure 5.3.3. Schematical representation of the mass spectrum at m/z = 56, as obtained using (a) a quadrupole-based ICPMSinstrument and (b) a sector field ICPMS instrument operated at a higher mass resolution [18]. Reproduced by permission ofInternational Scientific Communications.

equal amount of an internal reference element isadded and it is assumed that this element undergoesthe same suppression or enhancement as the ana-lyte element(s). All calculations are subsequentlycarried out using the ratio of the signal intensity ofthe analyte element to that of the internal reference.Usually, accurate correction is obtained by using

an internal reference element with a mass num-ber close to that of the analyte elements, althoughsome matrices (the example of C-containing matri-ces being notorious) selectively enhance the signalintensity of specific elements (e.g., As, Sb and Se)by affecting the ionization process [21]. As a resultof the use of a separation technique, adding an

CALIBRATION 289

internal reference element to the sample solutiondoes not offer a solution for nonspectral inter-ferences, as the prerequisite of a ‘common fate’of this internal reference element and the samplecomponents is not fulfilled. Saverwyns et al. [22]solved this problem to a large extent by mixingthe effluent of their HPLC column – used to sep-arate Cr(III) and Cr(VI) from one another – witha continuous flow of a Co standard solution. ForGC-ICPMS, using an internal reference is even lessobvious. De Smaele et al. [23] used H2 with 1 %Xe as a carrier gas and continuously monitored the126Xe+ signal as an internal reference. Chen andHouk [24] pointed out that the signals of somepolyatomic ions can also be used as an internalreference for analytes at a nearby mass-to-chargeratio. Especially in speciation work, this approachmay be useful.

Finally, when using the method of standardadditions or isotope dilution as a means ofcalibration (see later), nonspectral interferences areautomatically corrected for as sample and standardare affected to exactly the same extent.

2 CALIBRATION

2.1 Traditional approaches

Self-evidently, when using ICPMS for detectionin speciation work, the ‘traditional’ calibrationapproaches of internal standardization, externalstandardization and the method of standard addi-tions can be used.

Internal standardization is the simplest approach,in which the concentration of the species involvedis estimated from the signal intensity of a singlecompound with known concentration. With thisapproach, the results obtained should not beconsidered as being more than semiquantitative,because in a chromatographic separation process,not only the content of the trace element(s) whichis (are) the subject of the speciation study, but alsothat of the matrix (which may also contain othersubstances present at much higher concentrationlevels), will vary as a function of time. This maylead to a time-dependent variation in the degree

of nonspectral interference. In addition, as anentire separation process may take a considerabletime (e.g., several minutes), signal drift mayalso further compromise the results obtained. Anadvantage of internal standardization, however, isthat not for every species detected a correspondingstandard is required. In many instances, acquiringthese standards is not obvious and in the caseof complex molecules from biochemical (e.g.,proteins) or environmental (e.g., humic acids)origin often impossible.

When the identity and structure of all of thespecies of interest are known to the analyst and thecorresponding standards are commercially avail-able or can be synthesized, external calibrationbecomes possible. As a result of the wide linearrange exhibited by ICPMS, single point calibra-tion is often used instead of a calibration line.It should be realized that since sample and stan-dard are measured at a different moment, whilethe standard will normally show a different matrixcomposition, the accuracy of quantitative results isoften still not guaranteed. Application of an inter-nal reference signal permits one to correct for thevarying sensitivity of the detector – as a result ofvariations in the composition of the column efflu-ent and/or as a function of time – and can henceimprove the reliability of the data produced. Theuse of such an internal reference was discussed inSection 1.5. Feldmann [25] even described the useof such a Rh+ internal reference signal (obtainedby continuous nebulization of a Rh standard solu-tion into the ICP) for semi-quantitative determina-tion of elemental species, introduced into the ICPafter separation by GC and for which no standardswere (commercially) available.

Adding the species-specific standards to thesample (standard additions) brings about an auto-matic correction for nonspectral interferences assample and standard undergo the same matrix-induced suppression or enhancement. Signal driftor instrument instability between measurement ofthe sample and of the sample to which standardhas been added however, is still not corrected for.Simultaneous application of an internal referencesignal can hence further ameliorate the quality ofthe results obtained.

290 DETECTION

2.2 Isotope dilution

As a mass spectrometric technique, ICPMS alsoprovides the analyst with isotopic information,such that also isotope dilution (ID) can be usedfor calibration, provided that the element studiedshows at least two isotopes, the signal of whichis free from spectral overlap. In this approach,a known amount of tracer – i.e. the chemicalcompound to be determined, in which the elementstudied is characterized by an isotopic compositionsufficiently different from the natural one – isadded to a known amount of sample. Fromthe change induced in the isotopic compositionof the analyte element by this spiking process(Figure 5.3.4), the concentration of the targetspecies can be accurately calculated by using thefollowing equations:

Rsample =1nsample

2nsample,

Rtracer =1ntracer2ntracer

, Rblend =1nblend2nblend

nsample = θ1tracer

θ1sample

Rsample

Rtracer

[Rtracer − Rblend

Rblend − Rsamp

]ntracer

(5.3.1)

nsample = θ2tracer

θ2sample

[Rtracer − Rblend

Rblend − Rsamp

]ntracer (5.3.2)

where 1n is the signal intensity for thelighter isotope, 2n is the signal intensity for theheavier isotope, θ1 is the isotopic abundance of thelighter isotope and θ2 is the isotopic abundance ofthe heavier isotope.

Equation (5.3.1) is used if the tracer is enrichedin the lighter isotope, whereas equation (5.3.2) ispreferred when the tracer is enriched in the heav-ier isotope. If all ratios (Rsample, Rtracer and Rblend)are measured experimentally, the ratios obtainedcan be introduced into the appropriate equation assuch. When tabulated values are used for Rsamp

and/or Rtracer instead, the experimental results haveto be corrected for mass discrimination [26]. It isuseful to mention in this context that, althoughmost elements show an isotopic composition that

sample + tracer = blend

Rblend = 0.688

Rtracer = 0.208

signal isotope 2

signal isotope 1

Rsample = 2.125

Figure 5.3.4. Schematic representation of the principle of IDMS.

CALIBRATION 291

is constant, some show natural (e.g., Sr, Pb) orman-made (e.g., Li and U) variations. For theseelements, experimental determination of the iso-topic composition in the sample and in commer-cial standards, respectively, is required. Finally,the information required for the tracer – ntracer andθ (enriched isotope) or enrichedntracer = θ (enrichedisotope) ntracer – can be obtained from the corre-sponding certificate or, if necessary, via reverse ID.The latter is an ID procedure, whereby the enrichedtracer is considered as the sample and a standardof natural isotopic composition as the tracer.

ID shows distinct advantages over other cali-bration techniques. Once isotopic equilibration hasbeen established, analyte losses cause no furtherdeterioration in the final analysis result becausethey do not affect the isotope ratio. Additionally,isotope ratios are practically unaltered by nonspec-tral interferences, instrument instability or signaldrift either, such that these phenomena are alsoautomatically corrected for. Actually, variationsin the matrix composition and/or measurementparameters may have a small effect on the massdiscrimination, but in the context of ID this effectis of no significance. Hence, ID has the capabilityof providing the analyst with more accurate andmore precise analysis results.

A limitation of ID as a calibration approach typ-ical of elemental speciation work is the lack ofcommercially available isotopically enriched stan-dards. Therefore, several groups have been obligedto synthesize species-specific tracers in house.

Heumann et al. [27] have used ID for thedetermination of iodide and iodate in mineralwaters. These species were separated from oneanother using ion exchange HPLC. As I is mono-isotopic (127I), a long-lived radionuclide (129I)had to be used in the production of species-specific tracers. If no organoiodine species werepresent, the sum of the iodide and iodate resultswas seen to be in excellent agreement withthe total concentration of I. Ebdon et al. [28]used ID for the determination of trimethylleadTML – a degradation product of the correspond-ing tetraalkyllead, added as an anti-knocking agentto petrol – in artificial rainwater. For this pur-pose, 206Pb-enriched TML was synthesized from

‘a radiogenic lead’ reference material (NIST 983),by converting the latter into PbCl2, which wassubsequently derivatized using a MeMgI Grig-nard reagent. TML was separated from inor-ganic Pb and triethyllead (TEL) by means ofreversed phase ion-pairing HPLC. The resultsobtained agreed well with the corresponding refer-ence values.

The use of species-specific tracers, enablingthe use of ID in elemental speciation is notlimited to HPLC-ICPMS and several authors havereported on the use of this approach in GC-ICPMSapplications as well.

Encinar et al. [29] synthesized a mixture ofisotopically enriched mono- (MBT), di- (DBT)and tributyltin (TBT) tracers by butylation ofelemental Sn, enriched in 119Sn, using BuCl andEt3N and I2 as catalysts. The concentrations ofthe three target species in the mixed tracer weredetermined by reverse ID. Commercial MBT,DBT and TBT standards (with Sn of naturalisotopic composition) were used for this purpose.The entire process included derivatization of thetarget species using NaBEt4, separation usingcapillary gas chromatography and detection byICPMS. Subsequently, this mixed tracer solutioncould be used for the simultaneous determinationof TBT (mainly originating from the use ofantifouling paints) and its degradation productsDBT and MBT in sediment samples. Providedthat complete extraction of the target speciesfrom the sediment samples is accomplished, suchthat complete isotopic equilibration is guaranteed,ID permits more reliable results to be obtained.Once the above-mentioned prerequisite is fulfilled,other sources of error – incomplete derivatization,analyte losses during extraction of the derivatizedtarget species in an organic solvent and/or themeasurement itself – no longer affect the finalresult. The accurate results obtained for twocertified reference materials illustrated the utilityof this approach.

Snell et al. [30] produced isotopically enriched(CH3)2Hg, CH3HgCl and HgCl2 and used theseas species-specific tracers for ID purposes in theanalysis of natural gas condensates by means of

292 DETECTION

GC-ICPMS. Even in the case of species inter-conversion, which is normally detrimental in spe-ciation work, species-specific ID offers a solu-tion. Hintelmann et al. [31], e.g., established thatthe formation of additional CH3Hg+ from inor-ganic Hg during sample preparation may lead toan overestimation of the ‘original’ CH3Hg+ con-tent in environmental samples. This alteration ofthe distribution of Hg over its various chemicalforms during the pre-treatment could be unequiv-ocally demonstrated by adding an isotopicallyenriched stable tracer of Hg2+ to reference mate-rials that were subjected to sample preparationand subsequent analysis by means of HPLC-ICPMS or GC-ICPMS. Species-specific ID wassuggested to cope with this problem, becauseisotopically enriched standards are expected toundergo the same changes as the species to bedetermined [32].

Species-specific ID, however, is only possibleif the identity of the species is known and theircomposition and structure is sufficiently simple topermit synthesis of a corresponding tracer. In order

to preserve the advantages offered by ID to thelargest possible extent when the aforementionedconditions are not fulfilled, Heumann and cowork-ers introduced the use of species-unspecific ID inHPLC applications [19, 33, 34]. In this approach,the species of interest are first separated from oneanother using an appropriate form of HPLC and theHPLC effluent is subsequently mixed with a con-tinuous flow of enriched tracer (in a Y junction)prior to its introduction into the ICP (Figure 5.3.5).Hence, in this approach, isotopic equilibration isonly obtained in the ICP, where due to the hightemperature all compounds are broken down intoatoms, irrespective of their original chemical form.Finally, the experimental set-up also contains aflow injection valve, permitting injection of a stan-dard solution of natural isotopic composition forcalibration of the tracer flow (reverse ID).

As is shown in Figure 5.3.6 for Cu, with thisapproach, the isotope ratio of interest is monitoredas a function of time. When no copper-containingspecies are eluted from the column, the measuredratio is that of the enriched spike (approximately

Unspiked sample

Pump for the continuousaddition of a species-

unspecific spike

Guard column

Separation column

Sample injection valve

Standard injectionfor calibration of

spike flow

UV detector ICP-MS

HPLC pump

Figure 5.3.5. Schematic diagram of the instrumental set-up for HPLC-ICPMS using species-unspecific ID for calibration [19].Reprinted with permission from L. Rottmann and K. G. Heumann, Analytical Chemistry, 66, 3709. Copyright (1994) AmericanChemical Society.

USE OF THE ICP AS A SOFT IONIZATION SOURCE 293

Spike isotope ratio

Natural isotope ratio

Retention time/min

65C

u:63

Cu

25

15

10

20

5

0 10 15 205

Figure 5.3.6. 65Cu/63Cu isotope ratio as a function of theretention time during analysis of a river water sample using acombination of an HPLC system, equipped with a SEC columnand ICPMS, with species-unspecific ID for calibration [27].Reproduced by permission of the Royal Society of Chemistry.

20 for the 65Cu-enriched Cu(NO3)2 spike usedin this example). Cu-containing fractions of theeluent, however, shift the isotope ratio of interestin the direction of the natural isotope ratio, ascan be seen from the two ‘negative’ peaks in thechromatogram. Finally, the original chromatogram,showing the variation of the Cu isotope ratioas a function of time, can be converted into achromatogram, which directly displays the amountof Cu present in the effluent as a function of time.This approach is, of major interest, for e.g., thequantification of metal complexes with proteins inbody fluids or with humic substances in watersof different origin, the complex composition andstructure of which do not permit synthesis ofa species-specific spike. In the case of species-unspecific ID, nonspectral interferences, signaldrift and instrument instability are automaticallycorrected for; losses during the sample pre-treatment and/or species interconversion on theother hand are not corrected for since isotopicequilibration is only accomplished in the ICP.

This approach was, e.g., used to study metalinteractions with dissolved organic materials innatural aquatic systems. For this study, an HPLCsystem, equipped with a size exclusion column(SEC) was coupled on line with a UV detec-tor and an ICP-mass spectrometer. Simultane-ous registration of the presence of UV-absorbing

organic matter and the metals in the column efflu-ent permitted conclusions to be drawn about theinteractions between the metals under investigationand humic substances, which form the major partof dissolved organic matter in waters of differentorigin. In addition, the ID approach allows accu-rate quantitative information to be obtained. Lateron, simultaneous ICPMS detection of both the met-als of interest and the content of dissolved organiccarbon (DOC) was made possible by also adding13C-enriched benzoic acid as a species-unspecifictracer [35].

3 USE OF THE ICP AS A SOFTIONIZATION SOURCE

With commercially available ICPMS instrumenta-tion, the ICP is operated at atmospheric pressureand at RF powers exceeding 1000 W (except incase of operation under cool plasma conditions,which requires a reduction of the RF power bysome hundreds of watts). Molecules introducedinto such a plasma are broken down into atoms,which are subsequently ionized, such that all struc-tural information is lost. During the past coupleof years however, efforts have also been made tocouple low power low pressure ICP ion sourcesto an MS detection system [36, 37]. With suchinstrumentation, both molecular and atomic spec-tra can be obtained, depending on the RF powerused (5–90 W). The molecular spectra obtained atlow RF power resemble electron impact spectra,such that application of existing library spectracan facilitate species identification on the basisof the structural information provided. In additionto the RF power, the plasma pressure and com-position (pure He and mixed Ar/He plasmas areused) and the introduction of reaction gases havebeen demonstrated to influence the molecular frag-mentation. These ICPMS devices with low pres-sure low power plasmas have been combined withGC, to separate organo-Br, -Hg -Pb and -Sn com-pounds. Both qualitative and quantitative informa-tion could be obtained with this set-up. In spite ofthe promising results, no low power low pressureICPMS instrumentation is commercially available.

294 DETECTION

4 HIGH PERFORMANCE LIQUIDCHROMATOGRAPHY (HPLC)-ICPMS

4.1 Coupling

HPLC-ICPMS offers an unmatched performancefor the detection and determination of nonvolatilemetallospecies [38]. As a result of the differentseparation strategies that can be used – (1) normalphase and reversed phase partition, (2) ionexchange (3) size exclusion, (4) affinity and(5) adsorption chromatography – HPLC is extre-mely versatile. The interaction of the mobilephase with the sample constituents has a directeffect on the distribution coefficients (except inthe case of size exclusion chromatography, SEC)and contributes to this versatility. The choice ofan HPLC technique primarily depends on theresearch objective. Currently, reversed phase (RP)partition HPLC is one of the most frequently usedapproaches. It implies distribution of the targetcompounds between a nonpolar stationary phaseand a relatively polar mobile phase.

An important advantage in the coupling ofHPLC with ICPMS is the compatibility of thechromatographic effluent flow rate and the liq-uid flow rate required for stable pneumatic neb-ulization. Moreover, HPLC is operated at roomtemperature. Hence, interfacing of the two tech-niques is straightforward. However, problems mayarise upon introduction of HPLC effluents into theICP [38, 39]. LC mobile phases generally con-sist of some combination of organic solvents, saltsin buffer solutions and/or ion-pairing reagents.Generally, ICPMS requires more dilute buffersto be used and only tolerates lower concentra-tions of organic solvents than ICPOES. In ionexchange chromatography, buffer concentrationsoften exceed 0.1 M. Such concentrations are likelyto cause short-term signal suppression or enhance-ment and can cause blockage of the nebulizerand/or the sampling cone as well as erosion of thesampling cone and the skimmer [39]. These irre-versible changes in the interface aperture configu-ration lead to unwanted sensitivity losses. In RPC,the organic solvent must be coped with, whereasthe buffer concentration is seldom a problem. High

loads of organic solvents negatively influence theperformance because of the decreased plasma sta-bility – even plasma extinction can be observed inextreme cases – and deposition of carbon on thetorch and sampling cone [39]. Water-cooled spraychambers and introduction systems equipped witha desolvation unit are therefore used to reduce theamount of solvent introduced into the plasma. Anincrease in forward RF power can improve theplasma stability, but is accompanied by an increaseof the reflected power, which is harmful to the RFgenerator in the long run [40]. The deposition ofcarbon on the sampling cone, causing an elevatednoise level and excessive signal drifts, can in mostcases be minimized by addition of oxygen (ca. 3 %V/V) to the nebulizer gas flow, although this maylead to a reduced sampling cone lifetime [39, 40].Finally, it has to be considered that mobile phasesmay give rise to a more complex mass spectrum,leading to more spectral interferences, particularlyat a mass-to-charge ratio <80.

Figure 5.3.7 shows the experimental set-up ofa versatile HPLC-ICPMS system, based on a con-ventional pneumatic nebulizer interface [41]. Theinterface itself is simple: a piece of narrow boretubing connects the outlet of the LC column withthe liquid flow inlet of a pneumatic nebulizer(Meinhard nebulizer or PFA low flow nebulizer).Post-column addition of an internal reference ele-ment allows correction for changes in the plasmaconditions and/or other fluctuations in sensitivity.Post-column effluent split, accomplished by a vari-able micro splitter valve, is optional as a remedyto reduce the amount of mobile phase introducedinto the ICP, although at the expense of a lossin sensitivity. The dead volume is reduced to thelargest possible extent by using capillaries withsmall diameters, micro flow-splitters and miniatureT-pieces. To enhance the overall performance ofthe system, a software-controlled four-port valve,which is installed after the separation column,allows switching from chromatographic effluent to1 % HNO3 in-between measurements to rinse thesample introduction system.

A major limitation of the hyphenation of HPLCand ICPMS, is the low analyte transport efficiencyto the plasma (usually 5 % or less), inherent to

HIGH PERFORMANCE LIQUID CHROMATOGRAPHY (HPLC)-ICPMS 295

Separation Column(Dionex CS5A)

Peristalic Pump(1% HNO3)

HPLCmetal free(DX500, Dionex)

ICP-SMSElement 1, Finnigan MAT

Micro HPLC Pumpmetal free(st3001psw, Sanwa Tsusho)

1, 2 mL min−1

Micro Flow Splitter

1:10

1% HNO3 + 0,1 ng g−1 ln

1 mL min−1MeinhardNebulizer

Scott-typeSpraychamber

Figure 5.3.7. Experimental set-up of an HPLC-ICPMS system: flow splitting and make-up flow addition are used for reductionof nonspectral interferences [41]. Reproduced by permission of the Royal Society of Chemistry.

DIN sample introduction capillary

sample 5 µl

column 1 mm i.d.

60 µl min−1

40 µl min−1

gradientHPLCpump

isocraticHPLCpump

Eluent

Makeup liquid

nebulizing gas

control unit

O2

Ar

Figure 5.3.8. Schematic representation of a DIN-based interface for the coupling of microbore HPLC to ICPMS [42]. Reproducedby permission of the Royal Society of Chemistry.

pneumatic nebulization [40]. Hence, an increase intransfer efficiency would result in a correspondingimprovement in LODs. The use of other typesof nebulizers (e.g., an ultrasonic nebulizer) canincrease the analyte transport efficiency, but canalso give rise to an additional extra-column deadvolume and is therefore not recommendable. One

option that shows promise for certain applicationsis the use of a direct injection nebulizer (DIN) [42](Figure 5.3.8). This DIN is a microconcentricpneumatic nebulizer, which is positioned insidethe central tube of the ICP torch. Also the morerecently developed direct insertion high efficiencynebulizer (DIHEN) allows the direct and 100 %

296 DETECTION

efficient introduction of ∼ 100 µL min−1 flow ratesinto the ICP. The DIN interface offers a lowdead volume, short wash-out times and virtuallyno memory effects. By using this introductiontechnique, peak broadening can be minimizedand transport efficiency can approach 100 % withmobile phase flow rates up to 0.1 mL min−1 formicrobore columns. Lobinski et al. [42] reportedthat carbon deposits were less important for thistype of interface than for conventional pneumaticnebulizer interfaces. Moreover, the system showedbetter salt compatibility. As a drawback, the effectof organic solvents on the plasma stability wassubstantial and necessitated addition of oxygen,while the degree of plasma robustness may bereduced. Oxide ion formation levels in the plasmawere not reported in this study, but are also knownto be considerably higher.

4.2 Illustrative applications

HPLC-ICPMS has been used for the speciation ofa variety of biochemically and/or environmentallyinteresting elements. Most studies have focusedon the assessment of the environmental and/orhealth risks by the determination of potentiallydangerous species of a single target element [38,39] in different environmental compartments (air,natural water, soils, sediments and biota) or inhuman body fluids.

As a result of the large difference in toxicitybetween its different chemical forms, many pub-lications have reported on the speciation of Asusing HPLC-ICPMS. Already in 1993, Larsenet al. [43] used HPLC-ICPMS for the determina-tion of eight arsenic compounds in human urine.In a first phase (Figure 5.3.9.(a)), dimethylarsi-nate (DMA), As(III), monomethylarsonate (MMA)and As(V) were separated from one another andfrom the positively charged As species (eluting inthe void volume) using an anion exchange resin.In a second phase (Figure 5.3.9.(b)), the otherAs compounds – aresenobetaine (AsB), trimethy-larsine oxide (TMAO), arsenocholine (AsC) andthe tetramethylarsonium ion (TMAs) – were sepa-rated from one another and from the negatively

charged species using a cation exchange resin.More recently, it was demonstrated that, whenusing an anion exchange resin that also showssome nonpolar activity (nonpolar sites in additionto ion exchange sites), AsB – which is unchargedat the pH value of the mobile phase used – couldbe determined within the same chromatographicrun as DMA, As(III), MMA and As(V), while onlythe cationic arsenic compounds elute in the voidvolume [44]. Under these conditions, one mea-surement of a urine sample is sufficient to drawmeaningful conclusions concerning (professional)exposure to inorganic As, without risk of misin-terpretation due to the intake of food of marineorigin (a major source of the nontoxic AsB).Also the chromatographic resolving power and theLODs (≤0.05 µg L−1) of anion exchange chro-matography – ICPMS for As speciation have beenimproved considerably [45, 46].

Recently, other more explanatory HPLC-ICPMS studies have aimed at the elucidation ofthe mechanisms of biotransformation of inorganicmetal ions and simple inorganic species [38].A trend in the field of speciation by HPLC-ICPMS is the detection and identification ofligand – metal complexes in biological samples.Metalloproteins, trace metal complexation in bloodand blood plasma, selenoproteins in human andanimal body fluids and tissues, metallodrugs andtheir interaction with proteins are subjects ofinvestigation in these studies [47].

An interesting HPLC-ICPMS study on the inter-action of cis-[Pt(NH3)2Cl2] (cisplatin) with 5′-guanosine monophosphate (5′-GMP) was reportedby Hann et al. [48]. The anti-tumoral activity ofplatinum drugs is based on their coordinative bind-ing with lone pairs of electrons of DNA bases. As aconsequence, the structure as well as the function-ality of the DNA is modified and cell replicationis inhibited. Hence, investigation of the speciesformed by the interaction of cisplatin with DNAbases is a key to understanding the activity ofthis drug. The combination of high performanceion chromatography with sector field ICPMS pro-vided unambiguous stoichiometrical informationon the major GMP adduct. Cisplatin was incu-bated with 5′-GMP under physiological conditions

HIGH PERFORMANCE LIQUID CHROMATOGRAPHY (HPLC)-ICPMS 297

5−8

9

12

3

4

Time/s

5

4

3

2

1

0 100 200 300

(a)

1

6

57 8

0 100 150 20050

15

10

5

Time/s

(b)

Figure 5.3.9. (a) Chromatogram obtained for a fourfold diluted urine sample, spiked with 1, DMA; 2, As(III); 3, MMA; 4,As(V); 5, AsB; 6, TMAO; 7, AsC; 8, TMAs and 9, Sb(OH)6

− (used as an internal reference) using an HPLC system (equippedwith an anion exchange resin) coupled on-line to an ICP-mass spectrometer. (b) Chromatogram obtained for a fourfold dilutedurine sample, spiked with 1, DMA; 5, AsB; 6, TMAO; 7, AsC and 8, TMAs using an HPLC system (equipped with a cationexchange column) coupled on-line to an ICP-mass spectrometer [43]. Reproduced by permission of the Royal Society ofChemistry.

for 180 h. The simultaneous detection of P and Ptby sector field ICPMS and the determination ofthe instrumental response (sensitivity) for both ele-ments resulted in an P/Pt elemental ratio of 2/1,corresponding to the molar ratio in the bisadductcis-[Pt(NH3)2(GMP)2]2−. Higher mass resolution

proved to be mandatory for the accurate determina-tion of 31P. Additionally, the time-dependent reac-tion course of the cisplatin-5′ –GMP system couldbe followed as can be observed in Figure 5.3.10.The concentration decrease of 5′-GMP and the for-mation of adducts was monitored on the basis

298 DETECTION

1.0 × 106

8.0 × 105

6.0 × 105

4.0 × 105

2.0 × 105

5 10 20

0 h

3 h

5 h

18 h

180 h

26 h

53 h

0.0

5 10 20 2515

2.0 × 106

0

inte

nsity

/cps

@ 1

94,9

648

mu

inte

nsity

/cps

@ 1

94,9

648

mu

11 12 13 14

1 × 105

2 × 105

3 × 105

4.0 × 106

6.0 × 106

8.0 × 106

1.0 × 107

0 h3 h

5 h

18 h

180 h

26 h

53 h

0 h

3 h5 h

18 h

180 h

26 h

53 h

15

time/min

(a)

(b)

(c)

time/min

time/min

inte

nsity

/cps

@ 3

0,97

38 m

u

Figure 5.3.10. Time-dependent monitoring of reaction of cis-[Pt(NH3)2Cl2] (cisplatin) with 5′-guanosine monophosphate(5′-GMP): P (a) and Pt (b) chromatograms obtained after different incubation times (bisadduct observed at a retention time of15.5 min). (c) shows the intermediate monoadduct (observed at a retention time of 12.0 min) [48]. Reproduced from Fresenius’Journal of Analytical Chemistry, 370, 581, 2001, copyright notice of Springer-Verlag.

GAS CHROMATOGRAPHY (GC)-ICPMS 299

of UV and ICPMS detection. An intermediatemonoadduct was observed together with the majorproduct, the bisadduct cis-[Pt(NH3)2(GMP)2]2−.

Other interesting HPLC-ICPMS applicationsexploit the multielement capability of ICPMS.Wang et al. [49] used size exclusion chromatog-raphy coupled to sector field ICPMS to studymetal–protein binding. They investigated the asso-ciation of various elements with proteins of humanand bovine serum. V, Co, Cd and Mo werefound in two distinct protein fractions. Mn wasobserved in three protein fractions and in thelow molecular weight fraction. The latter fractionalso contained some lanthanides. Most lanthanides,however, were bound to proteins in the mass rangeof 70–90 kDa. Alkali metals and Tl were foundas free metals not bound to proteins. Jakubowskiet al. [50] studied the metabolism of platinum inbiological systems. Pt, S and C (13C) were moni-tored simultaneously. By means of this approach,90 % of the Pt found in a grass sample couldbe assigned to inorganic platinum. The remaining10 % occurred in four different organic fractions.

A review by Szpunar and Lobinski [47] reportedon the potential of HPLC-ICPMS for the studyof biomacromolecular complexes. BidimensionalHPLC with ICPMS detection, i.e. separation ofmetallopolypeptides by size-exclusion HPLC andsubsequent signal identification by anion exchangeHPLC-ICPMS, was discussed. Species identifi-cation was accomplished on the basis of thematching of their retention times with those ofstandards. Species of interest are biomacromolecu-lar metal complexes found in plants (e.g., polysac-charides, phytochelatins), biological fluids (e.g.,proteins, porphyrins) or animal tissues (e.g., met-allothioneins). The authors emphasized that, at thispoint of research, limitations in terms of separationselectivity and signal identification are becomingincreasingly conspicuous because of the unavail-ability of standards. The parallel use of electro-spray MS (ESMS) was proposed as a promisingalternative. The gap between the LODs offered byICPMS and ESMS, respectively, and the difficultyof unambiguous attribution of a peak in the massspectrum to a species are considered as currentlimitations.

Up to now, HPLC-ICPMS analysis is almostexclusively research, and most of the publishedmethods have not, or have not sufficiently, con-sidered some of the fundamental requirements ofroutine analysis [51]. However, for selected appli-cations, such as the separation of Cr(III) andCr(VI) and the simultaneous separation of differentoxidation states of As and Se, so-called speciationkits are commercially available. These kits, espe-cially developed for use with ICPMS, consist ofa microbore HPLC column and a guard disk forcolumn protection [52].

5 GAS CHROMATOGRAPHY(GC)-ICPMS

5.1 Coupling

Compared with HPLC-ICPMS, GC-ICPMS offersa higher resolving power and 100 % introductionefficiency, allows a more stable plasma andgives rise to fewer spectral interferences as aresult of the plasma being dry and finally leadsto less sampling cone and skimmer wear. Ofcourse, GC-ICPMS can only be used for theseparation and detection of sufficiently volatileand thermally stable compounds or compoundsthat can be derivatized into a volatile form.Also the coupling of a gas chromatograph withan ICP-mass spectrometer is somewhat morecomplicated as a heated transfer line is required,such that condensation of the species that havebeen separated from one another in the gaschromatograph and hence peak broadening can beavoided. Additionally, typical effluent flow rateswith GC are in the order of 1 mL min−1, whilefor ICPMS, a carrier gas flow rate in the order of1 L min−1 is required to obtain an annular plasma.As a result, addition of a make-up gas is required.

Recently, a transfer line enabling coupling ofGC to ICPMS was commercially introduced [53].Many research groups, however, have developedtheir own metallic or quartz heated transfer line.De Smaele et al. [54] reported the developmentof a transfer line permitting rapid coupling anddecoupling of a capillary gas chromatograph

300 DETECTION

transfer line

T-piece

separately heated part

Ar make-up gas

GC

ICPMS

Figure 5.3.11. Heated transfer line enabling hyphenation of CGC with ICPMS without significant band broadening [54].Reprinted from Spectrochimica Acta, Vol. 50, De Smaele et al., A flexible interface for . . ., p. 1409, 1995 with permissionfrom Elsevier Science.

to an ICPMS instrument. Their transfer line(Figure 5.3.11) basically consists of two resistivelyheated stainless steel tubes. The first stainless steeltube contains the deactivated fused silica capillarythat transports the GC effluent into the center of theICP. The Ar make-up gas is resistively heated inthe second stainless steel tube and is subsequentlyintroduced into the first stainless steel tube, whereit flows around the fused silica capillary. Thisset-up ensures an equable temperature all overthe transfer line. Finally, the whole transfer lineis thermally isolated. To avoid peak broadening,even under the most demanding circumstances, thelast part of the transfer line was resistively heatedseparately. Therefore, this part was lengthwise cutin two halves, which were electrically isolatedfrom one another over their entire length withpolyimide tape, except for an electrical contact atthe very end.

The transfer line developed by Montes Bayonet al. [55] is interesting, because its design ismarkedly different from that of the majority oftransfer lines described in the literature. In this

design, the last part of the GC capillary isinserted into a short piece of heated metallictube. This concentric assembly is inserted intoa metallic T-piece (Figure 5.3.12). Unheated Armake-up gas is introduced into the T-piece viathe perpendicular side arm. Since this make-upgas is forced to flow through the narrow openingbetween the metallic tube and the T-piece, itacquires a higher velocity and is hence able toprevent condensation of the separated species onthe walls of the T-piece or within the flexiblenonheated PTFE tubing used for connection to theICPMS instrument. This interface was reported toperform well in GC-ICPMS analysis, aiming at thedetermination of organometallic compounds of Hg,Pb, S, Se and Sn.


The major application field of GC-ICPMS involvesthe determination of organometallic compounds ofSn, Hg and Pb in environmental matrices.


Chromatographiccolumn

Ar make-up

Stainless steeltubing

1/4” Swagelok Tee

Coppertubing

ICP-MS

Metallic block

Column

GC Oven

Heater

Temp. sensor

Figure 5.3.12. Unheated transfer line for hyphenation of CGC and ICPMS [55]. Reproduced by permission of the Royal Societyof Chemistry.

Ritsema et al. [56], e.g., reported on the useof GC-ICPMS for the determination of butyltincompounds in harbor sediments, sampled in twomarinas along the Dutch coast. For many years,TBT was added as the active component to anti-fouling paints, intended to prevent the growth ofalgae and mussels on the hulls of ships and indocks. As a result of the growing awareness of

the outspoken toxicity of TBT, and to a lesserextent of its degradation products DBT and MBT,its use was regulated more strictly and prohibitedfor pleasure boats in many countries, including theNetherlands. To investigate the effects of this ban,the content of butyltin compounds in sedimentssampled between 1992 and 1995 was determined.After being leached from the sediments using a

302 DETECTION

MeOH–acetic acid mixture, the species of interestwere ethylated using NaBEt4 and the derivatizedspecies were extracted into hexane. These extractswere subsequently subjected to GC-ICPMS anal-ysis. As no decreasing trend was observed as afunction of the sampling time between 1992 and1995, while high TBT/DBT ratios were found inall sediment extracts, it was concluded that sed-iments can act as a reservoir of TBT. TBT ishence slowly released into the harbor water, suchthat intolerably high levels are still expected for along time. Vercauteren et al. [57] used GC-ICPMSfor the determination of traces of the pesticidefentin (triphenyltin) in environmental samples. Theanalyte of interest was extracted from the matri-ces investigated (potatoes and mussels) by meansof headspace solid-phase microextraction (SPME)after digestion using TMAH or a mixture ofKOH and EtOH and in situ derivatization usingNaBEt4. Leal-Granadillo et al. [58] investigatedthe possibility of determination of organolead com-pounds in airborne particulate matter using GC-ICPMS after their extraction into an organic sol-vent and derivatization by means of a Grignardreaction. Femtogram LODs (as Pb) were accom-plished for trimethyl- (TML), dimethyl- (DML),triethyl- (TEL) and diethyl-Pb (DEL). Analysis ofairborne particulate matter sampled in the Span-ish city Oviedo revealed the presence of sev-eral organo-Pb compounds. Armstrong et al. [59]demonstrated a LOD <1 pg (as Hg) for methylmer-cury (MM) using GC-ICPMS. The accuracy of theresults obtained was demonstrated by the agree-ment between experimental results and the cor-responding certified contents of MM in marinetissue certified reference materials. Finally, theapproach developed was used for the determina-tion of MM in ‘real-life’ samples of ringed sealand beluga whale.

Multielement approaches have also beendescribed in the literature. Jantzen and Prange [60]carried out an extensive study on the occurrenceof organometallic species of Sn, Hg and Pbin sediments along the River Elbe. The speciesof interest were extracted from the matrix in

an organic solvent after in situ derivatizationwith NaBEt4 and were subsequently subjectedto GC-ICPMS analysis. Peycheran et al. [61]collected air samples with a cryogenic trap andsubsequently analyzed these using GC-ICPMS fortheir content of volatile metal species. InorganicHg and tetraalkyl-Pb compounds were found tobe the major species in samples originating fromthe Bordeaux urban environment (France). Thequantitative results obtained for the organometalliccompounds of Pb illustrated the beneficial effect ofthe decreased use of leaded petrol. The efficiencyand high sample throughput of the methodpermits the influence of meteorological factors andautomotive traffic parameters to be studied intomore detail. By using a combination of in situpurging and cryogenic trapping, this method couldalso be used for analyzing natural waters [62].Low to sub-pg L−1 LODs were reported fororganometallic compounds of Hg, Pb, Se andSn (Figure 5.3.13). The method was used toinvestigate the occurrence of these species inestuaries in France, Belgium and the Netherlands.

Finally, in a number of instances, organometal-lic compounds of other elements have also beenstudied. Gallus and Heumann [63] determinedthe selenite (directly) and selenate (= total Se −selenite) content in water samples by means of GC-ICPMS. As selenite is a nonvolatile species, it wasconverted into a volatile piazoselenol by reactionwith 1,2-diamino-4-trifluoromethylbenzene priorto its determination. Feldmann et al. [64] providedevidence for the biomethylation of Bi – an ele-ment widely used in alloys, cosmetics and pharma-ceutical products – to trimethyl-Bi (TMB) usingGC-ICPMS for the analysis of gases of sewagesludge digesters and landfill gases. Peycheranet al. [65] demonstrated the utility of GC-ICPMSfor monitoring the workspace air in the semicon-ductor industry in view of professional exposureto volatile species of As, In and P. The sameapproach can also be used to check the purityof volatile reagents used in the production ofsemiconductors. Finally, Gruter et al. [66] recentlyreported on a GC-ICPMS-based method, capable


1000

900

800

700

600

500

400

300

200

100

0

Inte

nsity

(ar

bitr

ary

units

)

0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5

Retention Time (min)

208Pb

202Hg

120Sn

82Se

Et4Sn

Me2Hg

Me4Pb

Me4Sn

Me2Se

Me2Se2

Et4Pb

Et2Hg

Figure 5.3.13. Multielement chromatogram for an aqueous standard solution, obtained after purging and cryogenic trappingfollowed by GC-ICPMS [62]. Reprinted from Analytica Chimica Acta, Vol. 337, Amouroux et al., Sampling and probing volatile. . ., p. 241, 1998 with permission from Elsevier Science.

1 2 3 4 5 6 7 8 9 10 11 12 13

200

160

120

80

40

0

−40

−80

−120

retention time [min]

boili

ng p

oint

[°C

]

(CH3)GeH3

(CH3)SbH2

(C4H9)SnH3

(C2H5)4Pb

(C4H9)2SnH2

SnH4 SbH3

BiH3

(CH3)

(CH3)2Se

(CH3)2SbH

(CH3)3Sb

(CH3)2Hg

(CH3)3Bi

(CH3)2Te

(CH3)4Sn

(CH3)SnH3

(CH3)3As(CH3)2AsH

(CH3)AsH2

(CH3)SnH3

(CH3)3GeH

(CH3)2SnH2

(CH3)2GeH2

AsH3GeH4

Figure 5.3.14. Relationship between boiling point and elution time of volatile organometal(loid) compounds for a purge-and-trapGC-ICPMS set-up [66]. Reprinted from Journal of Analytical Chemistry, A new HG/LT-GC/ICP-MS multielement speciationtechnique for real samples in different matrices, U. M. Gruter, J. Kresimon and A. V. Hirner, Vol. 368, p. 67, 2000, copyrightnotice of Springer-Verlag.

304 DETECTION

of determination of the metal(loid)–organic com-pounds of twelve elements – As, Bi, Ge, Hg, I,Mo, Pb, Sb, Se, Sn, Te and W – with LODsbelow 1 pg. Derivatization was accomplished bymeans of hydride generation, and the volatilespecies were subsequently collected in a cryo-genic trap. By heating the trap slowly from – 196to 150 ◦C, and introducing the sequentially re-volatilized compounds into a GC, complete separa-tion could be ensured. A clear correlation betweenboiling point and retention time could be estab-lished (Figure 5.3.14). Soil samples from munici-pal waste deposits were subjected to analysis withthis instrumental set-up, in view of the possibil-ity of methylation caused by the microbiologicalactivity and the reducing conditions often typi-cal for these matrices. For nine elements, organicspecies were found with concentrations in theupper ng (as element) per kg level. In a wastedeposit that had been closed for a longer time(15 years), a lower extent of biomethylation wasestablished. This can probably be explained by analmost complete decomposition of organic mate-rial, leading to reduced bioactivity.

6 CAPILLARY ELECTROPHORESIS(CE)-ICPMS

6.1 Coupling

CE-ICPMS combines a high separation efficiencywith the sensitivity of ICPMS and can be con-sidered as the most important recent trend in thefield of hyphenated techniques [39]. CE techniqueshave important advantages over more conventionalseparation approaches. CE is able to separatecationic, anionic and neutral species, it showspotential to handle labile complexes (e.g., nonco-valently bound) and colloidal systems, it is rapidand the columns are relatively simple and cheap.Hence, a variety of compounds with metal func-tionality are amenable to speciation by CE-ICPMS.The method represents an interesting alternative toHPLC-ICPMS, when ‘gentle’ separation schemesare required to preserve the true chemical infor-mation in a real-life sample. Moreover, since the

technique is characterized by an extremely lowsample consumption, it is ideally suited when theamount of sample available is limited, a situationfrequently encountered in biological, biomedical ornuclear research.

As only very small sample volumes (typically1–100 nL) are introduced in CE, it is gener-ally difficult to obtain satisfactory LODs in termsof concentration for most species. Accordingly,development of highly sensitive and selectivedetectors has been very important and challengingsince CE came into existence. Olesik et al. [67]published the first paper on CE-ICPMS in 1995.The first hyphenation was accomplished by usinga conventional concentric nebulizer in combinationwith a conical spray chamber. It was evident fromthis first publication that the key to the analyticalsuccess of CE-ICPMS is the design of the inter-face. Establishing efficient sample transport with-out degradation of separation resolution proved tobe a major challenge. The basic requirements ofthe hyphenation can be summarized as follows.(1) The interface must include an electrical con-nection at the CE capillary exit end, enabling appli-cation of an electrical field gradient along the CEcapillary as a driving force for species separation.A stable electric current is crucial for reproducibleseparations. (2) The interface must adapt the flowrate of the electro-osmotic flow (EOF) inside theCE capillary (nL min−1 range) to the liquid flowrate required by the nebulizer. (3) The interfacemust prevent nebulizer suction from causing lami-nar flow inside the capillary. Nebulizer suction wasidentified [67] as the principal factor jeopardizingseparation resolution in CE-ICPMS. (4) BecauseICPMS is a post-column detection method, somedead volume is inevitable. The challenge is tomake the impact of this dead volume as smallas possible.

During the last few years, several interfaces forcoupling of CE with ICPMS have been devel-oped with various degrees of success. Theseconstructions are mostly based on Meinhard [68,69], microconcentric [70–72] or ultrasonic neb-ulizers [73], but also interfaces based on cross-flow [70], high efficiency [74] or direct injectionnebulizers [75] have been developed. Almost all

CAPILLARY ELECTROPHORESIS (CE)-ICPMS 305

spray chamber

argon

make-up liquid

CE capillary

Pt electrode

nebulizer capillary

Figure 5.3.15. Schematic representation of a commerciallyavailable CE-ICPMS interface based on a modified microcon-centric nebulizer [76]. Reprinted from Journal of AnalyticalChemistry, D. Schaumloffel and A. Prange, Vol. 364, p. 452,1999, copyright notice of Springer-Verlag.

interfaces developed to date use (either home-builtor commercially available) spray chambers with alow dead volume, such as the cyclonic or the coni-cal type to reduce memory effects and peak broad-ening. Schaumloffel and Prange [76] developedan interface based on a modified microconcen-tric nebulizer, which is now commercially avail-able (Figure 5.3.15). Flow rates between 2 and12 µL min−1 were obtained in the self-aspirationmode for this modified nebulizer. A conductingmake-up buffer flows concentrically around the CEcapillary exit to provide the necessary electrical

connection as well as to supplement the low EOFin order to achieve stable nebulizer operation. Across-shaped piece connects the nebulizer with theCE capillary (vertical fittings). The two horizontalfittings are intended for the platinum electrode andfor the make-up liquid, respectively. The make-up liquid is transported by self-aspiration of thenebulizer and is in electrical contact with the plat-inum electrode. Mixing occurs at the end of theCE capillary. The nebulizer is plugged into a lowvolume spray chamber of about 5 mL for minimiz-ing band broadening of the CE signals. A Teflontube of 60 cm length connects this interface to theICPMS torch. The possibility to adjust each param-eter (e.g., the capillary position) in an exact andreproducible way is one of the key advantages ofthis interface. The authors report a dilution factorof only 5–20 for the CE analytes, depending onCE capillary dimensions and voltage.

A cheap, stable and dependable interfacecan also be obtained by combining a commer-cially available microconcentric nebulizer with alow volume spray chamber [70–72] such as thecyclonic spray chamber (Figure 5.3.16). Laminarflow in the CE capillary can be counterbalancedby using a liquid sheath flow. Either the sheathflow is pumped at matched flow rates or the nebu-lizer is operated in the self-aspiration mode, with aleveled sheath liquid reservoir. As a drawback, thisuse of a sheath flow compromises sensitivity to ahigher degree compared to the commercially avail-able design, since the nebulizer accommodatesflow rates in the 100 µL min−1 range, such that

microconcentricnebulizer micro mist

stainless steel tee stainless steel tubing

tygon tubing

nebulizing gas(0,45−1, 1I Ar min−1)

sheath liquid(10−150 µl min−1)

CE capillary

Figure 5.3.16. Schematic representation of a CE-ICPMS interface based on a commercially available microconcentric nebulizer(Micromist, Glass Expansion, Switzerland), used in combination with a cyclonic spray chamber [71]. Reproduced by permissionof the Royal Society of Chemistry.

306 DETECTION

the sample is significantly diluted (up to factor of1000). Lower flow rates compromise nebulizationefficiency and result in memory effects. Majidi andMiller-Ihli [70] reported the occurrence of spikesand peak tailing in the electropherogram due topoor transport and mixing efficiency of the CEeffluent with sheath buffer prior to nebulization.With higher flow rates again spikes were observed.This was explained by release of analytes from thewall of the low volume spray chamber. Applicationof a negative pressure to the inlet vial is an alter-native approach for counterbalancing the nebulizersuction [72].

Other interesting interfaces based on direct sam-ple injection have been developed. The aim was toimprove sample transfer efficiencies up to 100 %,without adverse effects derived from the spraychamber, such as analyte–chamber interactions,band broadening and/or sample losses. A DINinterface was first implemented by Liu et al. [75].The CE capillary was placed concentrically insidethe DIN sample introduction capillary so that theliquid sample was directly nebulized into the cen-tral channel of the plasma torch. The DIN’s lowdead volume offered faster sample introduction andshorter wash-out times. Majidi et al. [77] reportedthe use of a DIHEN for CE-ICPMS. With a sam-ple uptake rate of 85 µL min−1, the LODs couldbe improved by a factor of 2 compared with across-flow nebulizer interface. Recently, Bendahlet al. [78] developed a demountable DIHEN oper-ating at low sample uptake rates (10 µL min−1)for CE-ICPMS hyphenation (Figure 5.3.17). Asa drawback, both DINs and DIHENs are knownto be difficult to install and optimize on ICPMSinstruments. Moreover, the DIN requires a gas dis-placement pump for sample delivery and a coaxial

carrier gas in addition to the nebulizer gas andits cost is also a consideration. Tangen et al. [79]found severe problems due to the interference ofthe high voltage used for the CE separation withthe RF power supply of the ICPMS plasma with aDIN interface. Finally, it has to mentioned that theabsence of a spray chamber may also have negativeeffects, such as an increased oxide ion formation.


To date, the application of CE-ICPMS to real sam-ples has been reported in only a few papers. Mostresearch has focused on hyphenation-related prob-lems and aimed at the construction of a robustinterface and improvement of the LODs. The ana-lytical figures of merit have been evaluated onthe basis of different applications. Van Holderbekeet al. [71] showed the potential of CE-ICPMS forarsenic speciation. Polec et al. [80] evaluated aninterface based on a self-aspirating micronebulizerfor a study of metal binding by recombinant andnative metallothioneins. Ackley et al. [81] testedthe method for analysis of metalloporphyrins stan-dards. Michalke and Schramel [82] investigatedthe capability of CE-ICPMS for Sb speciation.On the basis of iodine speciation, Michalke [83]addressed problems related to the conflicting situ-ation of improving LODs by increased sample vol-umes and preconcentration by stacking procedureson the one hand and preserving separation effi-ciency and species stability on the other. Detectionof pseudospecies artefacts as a result of chemicalinteraction of samples, electrolytes and the capil-lary and spectral interferences was stated as majorproblem in CE-ICPMS.

argon platinum tubeCE-capillarysample capillaryshield

sheath liquid

Figure 5.3.17. Demountable DIHEN for interfacing CE to ICPMS [78]. Reproduced by permission of the Royal Society ofChemistry.

CAPILLARY ELECTROPHORESIS (CE)-ICPMS 307

Michalke and Schramel [84] applied CE-ICPMSto Se speciation in human serum and breast milkof early lactation. Se speciation is of growinginterest, due to the toxic and essential propertiesexhibited by this element. CE separation of sixSe species – including both inorganic and organicspecies – was accomplished. In parallel, CE sepa-rations based on isoelectrical focusing have beendeveloped to discover the isoelectric points of theorganic Se species. LODs of the CE-ICPMS sys-tem were found to be just acceptable or still toohigh for the samples investigated. Therefore, pre-concentration was necessary to detect Se species inhuman milk. In accordance with other studies, noinorganic Se was observed in this type of samples.Ten different Se species were detected in humanserum. Unfortunately, the signals were close to thedetection limit, while identification of the specieswas not included in this study.

In a recent paper, Prange et al. [85] investigatedmetallothionein isoforms in human brain cytosolby CE-ICPMS. For the first time, CE-ICPMS wasused in comparative studies on the distribution

of isoforms of metallothioneins in brain samplestaken from subjects with Alzheimer’s disease andfrom a control group. The isoforms were sepa-rated by CE and the elements Cu, Zn, Cd and Swere detected by sector field ICPMS. For accu-rate determination of some of these elements, theuse of a higher mass resolution was a prerequi-site. Defatting cytosol with subsequent acetoni-trile precipitation for protein elimination was foundto be the optimum sample preparation procedure.Figure 5.3.18 shows the electropherogram of acerebellum cytosol for the elements Cu, Zn and Cdafter acetonitrile precipitation. The authors suggestthat the peaks observed correspond to the speciesknown as MT-1, MT-2 and MT-3 on the basisof migration time comparison with reference MTfrom rabbit liver. Similar patterns of MT isoformswere found for all brain regions of the subjectsinvestigated. For subjects with Alzheimer’s dis-ease, the levels of MT-1 and MT-3 were reportedto be reduced in all temporal and occipital sam-ples, while the Cu pattern was observed to beaffected. Furthermore, the detection of sulfur by

200.000

150.000

100.000

50.000

300 400 500 600 700 800 900 1.000 1.100 1.200

MT 3

MT 1

MT 2

63Cu

64Zn114Cd

time/s

inte

nsity

/cps

250.000

0200

Figure 5.3.18. Electropherogram of a cerebellum cytosol showing the signals for the elements Cu, Zn and Cd after acetonitrileprecipitation. The authors suggest that the peaks observed correspond to MT-1, MT-2 and MT-3 on the basis of migration timecomparison with that for reference MT from rabbit liver [85]. Reprinted from Journal of Analytical Chemistry, A. Prange et al.,Vol. 371. p. 764, 2001, copyright notice of Springer-Verlag.

308 DETECTION

ICPMS and quantification by ID have been inves-tigated as a new method for the quantification ofMT isoforms. After metal quantification, the molarratios of sulfur to metals can be used to character-ize the composition of the metalloprotein complexof the MT isoforms. Thus, a suggestion for theirstoichiometric formulae can be given.

7 SUPERCRITICAL FLUIDCHROMATOGRAPHY (SFC)-ICPMS

7.1 Coupling

SFC can be regarded as a hybrid techniquebetween gas and liquid chromatography, combin-ing some of the best features of each, i.e. thehigh diffusion coefficient of GC with the solubil-ity properties of LC [38]. Compounds which aretraditionally difficult to separate by GC, such asthermally labile, nonvolatile and high molecularmass compounds can be separated by SFC withrelative ease. Compared to LC, SFC is faster dueto the lower viscosity of the mobile phase andhigh diffusion coefficients of the analyte. Whenusing SFC, the use of organic solvents and the rel-atively long on-column residence times observedwith many LC techniques can be avoided [40].Critical temperatures and critical pressures of themobile phases used in SFC lie within the usualHPLC chromatographic conditions, such that littleinstrument modification is required. Pressure con-trol is the primary variable for the chromatographicseparation. Carbon dioxide is so far the most com-monly used mobile phase.

In SFC-ICPMS, the mobile phase changes fromthe supercritical fluid to the gaseous state beforeentering the plasma. Technically, the decompres-sion is accomplished in the interface by imple-menting a restrictor connected to the end of theanalytical column [39]. This process of expansionis subject to the Joule–Thomson effect and resultsin a net cooling. Hence, the restrictor zone must besufficiently heated. Generally, coupling SFC to ICPtorches requires similar conditions to those for GC,where a heated transfer line connects the oven withthe torch. Nebulizer and spray chamber typicallyused for liquid sample introduction into the plasmacan be eliminated. In summary, the main factorsto be considered for coupling SFC and ICPMS areas follows. (1) Sufficient heat should be providedto the restrictor (temperature > 150 ◦C) in orderto avoid cluster formation and wall condensation.(2) Plasma response to the supercritical fluid iscritical. (3) As with any other coupled technique,a sufficiently high analyte transport efficiency iscrucial for successful hyphenation.

There is a marked preference in selecting eitherpacked microcolumns or capillary columns overpacked SFC columns for SFC-ICPMS, since thehigh mobile phase flow rates with the latter(compared to that with capillary systems) maycause severe perturbations in the plasma [86].Interfacing capillary SFC and ICPMS instrumentswas first accomplished by Shen et al. [87]. Mostof the SFC-ICPMS studies reported to date usethis functional design (Figure 5.3.19). It providesa heated transfer line connecting the SFC ovenwith the ICPMS torch. The restrictor is mountedin a heated copper tube (200 ◦C). This assembly

heated gas

transfer line1/8" - 1/16" swagelok unionfrit restrictor

1/8" copper tube

Figure 5.3.19. Schematic representation of an SFC-ICPMS interface [87]. Reprinted with permission from W. Shen et al.,Analytical Chemistry, 63, 1491. Copyright (1991) American Chemical Society.

ALTERNATIVE APPROACHES 309

is inserted into the central tube of a regular ICPtorch. The tip of the restrictor is positioned flushwith the end of the central tube of the ICPtorch. A three-way Swagelock union was usedfor introduction of a heated Ar make-up gas totransport the analyte into the plasma. The make-up gas flow rate approximates the normal nebulizergas flow rate used in ICPMS. Plasma and auxiliarygas flows are not altered. The combination ofheating the transfer line and restrictor zone andproviding a heated make-up gas flow ensures aproper restrictor temperature. Shen et al. evaluatedthe performance of this instrumental set-up on thebasis of the separation of tetraalkyltin compounds.The authors critically discussed the backgroundspectral features of Ar ICPMS with CO2 asan SFC eluent. Major background interferenceswere identified as 12C+, 12C16O2

+ and 40Ar12C+.Also, the effect of the mobile phase CO2 onthe sensitivity – due to plasma quenching – wasinvestigated. At CO2 flow rates typical for capillarySFC (<1 mL min−1), the effects of CO2 on theplasma were found to be minimal and thus normalICP operating conditions could be used. At higherflow rates, prolonged CO2 introduction resulted incarbon deposition on the sampler and skimmer.


Speciation studies using SFC-ICPMS have recentlybeen reviewed by Vela and Caruso [86]. The num-ber of publications and applications has been lim-ited, in part due to the popularity of and the wideavailability of detailed information on LC and GCseparations [40]. So far, most SFC-ICPMS stud-ies have focused on a better understanding of theprocesses occurring in the transfer line or the inter-face. Another factor to consider is the fact that themost common mobile phase for SFC, CO2, is notideal for most organometallic compounds. Elutionproblems and strong interaction with the station-ary phase are problems related to the mismatchingin terms of polarities of mobile phase and analytecompounds. The use of organic modifiers or for-mation of nonpolar metal complexes from polarorganometallic compounds have been studied assolutions for these problems.

The potential of ICPMS as an element-selectivedetector for SFC was initially demonstrated withorganotin compounds [87]. Tetrabutyltin, tributyltinchloride, triphenyltin chloride, and tetraphenyltinhave been separated in a single chromatographic runwith LODs in the subpicogram range. The analyticalfigures of merit of SFC-ICPMS and SFC-FIDwere compared systematically [88]. The resolutionobtainable with SFC-FID was not always observedin SFC-ICPMS. Temperature fluctuations in thetransfer line and the restrictor zone – affecting themobile phase density and velocity – were identifiedas the main source of peak broadening. As anadditional consequence, variations in retention timeswere observed. However with SFC-ICPMS, LODswere improved by one order of magnitude comparedto SFC-FID.

Carey et al. [89] evaluated the performance ofSFC-ICPMS for the determination of organochro-mium compounds. Owing to the spectral interfer-ences (ArC+ at m/z 52), nitrous oxide replacedcarbon dioxide as the mobile phase. For the deter-mination of ionic chromium and different Cr oxi-dation states, it was necessary to complex thechromium prior to injection into the SFC system.β-Ketonate compounds were chosen as complex-ing agents. Problems with the detection of ther-mally labile chromium complexes were encoun-tered. This was related to the manner in which therestrictor was heated.

8 ALTERNATIVE APPROACHES

The large majority of elemental speciation stud-ies using ICPMS are accomplished by means ofon-line coupling of a chromatographic or elec-trophoretic column. However, some alternative andoften creative methods have been described thatdeserve proper attention.

Jin et al. [90] accomplished speciation of Ge bytrapping the hydrides of inorganic, monomethyl-(MMGe) and dimethyl-Ge (DMGe) – formed byreaction with NaBH4 and removed from the reac-tion vessel by a flow of He – in a cooled trap.Subsequently, re-volatilization of the compoundsas a function of their boiling point could be accom-plished, such that fractionated introduction into the

310 DETECTION

ICP enabled quantitative species-specific resultsto be obtained. Sub-pg LODs were achieved. Allcited Ge compounds were detected in natural andwaste waters, while in the latter sample, other,unknown Ge species were also observed.

Willie et al. [91] used graphite furnace – ICPMSto determine the contents of inorganic mercury andmethylmercury (MM) in fish tissue. This approachis also based on the different volatilities of thetarget compounds. In a first measurement, Hg isvaporized from the samples of interest (solubilizedusing TMAH) and introduced into the ICP, suchthat the total content of this element could be deter-mined. For the second measurement, iodoaceticacid, sodium thiosulfate and acetic acid are addedto the sample. During the drying step of the tem-perature program, MMI is already removed fromthe furnace, such that during the vaporization step,only the inorganic fraction of the Hg is introducedinto the ICP and measured. An even more elegantapproach was developed by Gelaude et al. [92]. Inthis approach, a carefully optimized temperatureprogram enabled the vaporization of MM and inor-ganic Hg to be separated in time, such that bothcompounds could be determined within the samemeasurement. Quantification was accomplished bymeans of species-unspecific ID using a permeationtube, containing elemental Hg enriched in 200Hg.Analysis of certified reference materials demon-strated the suitability of this approach.

Feng et al. [93] reported on the simultaneousdetermination of inorganic As(III) and As(V) usinghydride generation – ICPMS. Since the reactionconditions can be selected such that As(III) isreadily converted into the corresponding hydride,while As(V) is not converted at all, it is possibleto either determine the As(III) content, or the totalAs content, the latter after complete pre-reductionof As(V) to As(III). Hence, determination of bothAs(III) and As(V) (by subtraction) is possible, butrequires two measurements of the same sample.Feng et al. modified this approach by addingL-cysteine as a reducing agent, which slowlyconverts As(V) into As(III). By measuring theAs+ signal intensity at two moments in time afterthe start of the reaction, the contents of bothAs(III) and As (V) can be calculated, such that one

measurement suffices for each sample. A similarapproach was developed for the simultaneousdetermination of Sb(III) and Sb(V) [94].

Finally, a very promising approach was describedby Nielsen et al. [95], who used a combinationof gel electrophoresis and laser ablation ICPMSfor studying the binding of Co to blood serumproteins. For this purpose, human serum (enrichedwith Co) was subjected to immunoelectrophoresis.By rastering the agarose gel obtained in thisseparation process with an IR laser and analyzingthe material ablated from the gel by ICPMS, a Codistribution map was obtained. By comparing thisto the protein distribution map, obtained on stainingthe proteins with Coomassie Blue, the main Co-binding serum proteins could be identified. Evansand Villeneuve [96] used a similar approach tostudy the association of Pb to humic and fulvicacids. For this purpose, isotopically enriched Pb wasadded to the sample and the aforementioned organiccompounds were subsequently separated from oneanother as a function of their molecular size ona polyacrylamide gel. Finally, the dried gel plateswere subjected to LA-ICPMS analysis to find outwhich organic compounds Pb was bound to.

9 REFERENCES

1. Fisher, A. and Hill, S. J., Inductively coupled plasmamass spectrometry, in An Introduction to AnalyticalAtomic Spectrometry , Ebdon, L. and Evans, E. H. (Eds),John Wiley & Sons, Ltd, Chichester, 1998, Chapter 5,pp. 115–136.

2. Elan 5000 Users Manual, 0993–8429, May 1992, Rev. B.3. Niu, H. and Houk, R. S., Spectrochim. Acta, Part B , 51,

779 (1996).4. Resano, M., Verstraete, M., Vanhaecke, F. and Moens, L.,

J. Anal. At. Spectrom., 16, 1018 (2001).5. Sakata, K. and Kawabata, K., Spectrochim. Acta, Part B ,

49, 1027 (1994).6. Vanhaecke, F., Saverwyns, S., De Wannemacker, G.,

Moens, L. and Dams, R., Anal. Chim. Acta, 419, 55(2000).

7. Divjak, B. and Goessler, W., J. Chromatogr. A., 844, 161(1999).

8. Minnich, M. G. and Houk, R. S., J. Anal. At. Spectrom.,13, 167 (1998).

9. Brenner, I. B., Zander, A., Plantz, M. and Zhu, J., J.Anal. At. Spectrom., 12, 273 (1997).

10. Feldmann, I., Jakubowski, N. and Stuewer, D., Frese-nius’ J. Anal. Chem., 365, 415 (1999).

REFERENCES 311

11. Feldmann, I., Jakubowski, N., Thomas, C. andStuewer, D., Fresenius’ J. Anal. Chem., 365, 422 (1999).

12. Marchante-Gayon, J. M., Feldmann, I., Thomas, C. andJakubowski, N., J. Anal. At. Spectrom., 16, 45 (2001).

13. Tanner, S. D. and Baranov, V. I., At. Spectrosc., 20, 45(1999).

14. Simpson, L. A., Thomsen, M., Alloway, B. J. andParker, A., J. Anal. At. Spectrom., 16, 1375 (2001).

15. Bandura, D. R., Baranov, V. I. and Tanner, S. D., Frese-nius’ J. Anal. Chem., 370, 454 (2001).

16. Shaw, P., Spence, B., Lee, K. and Nelms, S., Speed andversatility with CCTED ICP-MS. How fast switchingand energy discrimination can get the highest qualitydata in the shortest possible time. Paper presented atthe 2002 Winter Conference on Plasma Spectrochemistry,Scottsdale, AZ, USA, January 6–12, 2002.

17. Sloth, J. J. and Larsen, E. H., J. Anal. At. Spectrom., 15,669 (2000).

18. Lapitajs, G., Greb, U., Dunemann, L., Begerow, J.,Moens, L. and Verrept, P., Int. Lab., May, 21 (1995).

19. Rottmann, L. and Heumann, K. G., Anal. Chem., 66,3709 (1994).

20. Cabezuelo, A. B. S., Bayon, M. M., Gonzalez, E. B.,Garcia Alonso, J. I. and Sanz-Medel, A., Analyst , 123,865 (1998).

21. Vanhaecke, F., Riondato, J., Moens, L. and Dams, R.,Fresenius’ J. Anal. Chem., 355, 397 (1996).

22. Saverwyns, S., Van Hecke, K., Vanhaecke, F., Moens, L.and Dams, R., Fresenius’ J. Anal. Chem., 363, 490(1999).

23. De Smaele, T., Moens, L., Dams, R. and Sandra, P.,Fresenius’ J. Anal. Chem., 355, 778 (1996).

24. Chen, X. S. and Houk, R. S., J. Anal. At. Spectrom., 10,837 (1995).

25. Feldmann, J., J. Anal. At. Spectrom., 12, 1069 (1997).26. Heumann, K. G., Gallus, S. M., Radlinger, G. and Vogl, J.,

J. Anal. At. Spectrom., 13, 1001 (1998).27. Heumann, K. G., Rottmann, L. and Vogl, J., J. Anal. At.

Spectrom., 9, 1351 (1994).28. Ebdon, L., Hill, S. J. and Rivas, C., Spectrochim. Acta,

Part B , 53, 289 (1998).29. Encinar, J. R., Monterde Villar, M. I., Santamaria, V. G.,

Garcia Alonso, J. I. and Sanz-Medel, A., Anal. Chem.,73, 3174 (2001).

30. Snell, J. P., Stewart, I. I., Sturgeon, R. E. and Frech, W.,J. Anal. At. Spectrom., 15, 1540 (2000).


32. Hintelmann, H. and Evans, R. D., Fresenius’ J. Anal.Chem., 358, 378 (1997).

33. Rottmann, L. and Heumann, K. G., Fresenius’ J. Anal.Chem., 350, 221 (1994).

34. Heumann, K. G., Gallus, S. M., Radlinger, G. and Vogl, J.,Spectrochim. Acta, Part B , 53, 273 (1998).

35. Vogl, J. and Heumann, K. G., Anal. Chem., 70, 2038(1998).

36. O’Connor, G., Ebdon, L., Evans, E. H., Ding, H., Olson,L. K. and Caruso, J. A., J. Anal. At. Spectrom., 11, 1151(1996).

37. Waggoner, J. W., Milstein, L. S., Belkin, M., Sutton,K. L., Caruso, J. A. and Fannin, H. B., J. Anal. At.Spectrom., 15, 13 (1999).

38. Szpunar, J., Witte, C., Lobinski, R. and Adams, F., Fre-senius’ J. Anal. Chem , 351, 351 (1995).

39. Sutton, K., Sutton, R. C. and Caruso, J. A., J. Chro-matogr. A., 789, 85 (1997).

40. Vela, N. P., Olson, L. K. and Caruso, J. A., Anal. Chem.,65, 585 A (1993).

41. Hann, S., Prohaska, T., Kollensperger, G. andStingeder, G., J. Anal. At. Spectrom., 15, 721 (2000).

42. Lobinski, R., Pereiro, I. R., Chassaigne, H., Wasik, A.and Szpunar, J., J. Anal. At. Spectrom., 13, 859 (1998).

43. Larsen, E. H., Pritzl, G. and Hansen, S. H., J. Anal. At.Spectrom., 8, 557 (1993).

44. Saverwyns, S., Zhang, X., Vanhaecke, F., Cornelis, R.,Moens, L. and Dams, R., J. Anal. At. Spectrom., 12, 1047(1997).

45. Kavanagh, P., Farago, M. E., Thornton, I., Goessler, W.,Kuehnelt, D., Schlagenhaufen, C. and Irgolic, K. J., Ana-lyst , 123, 27 (1998).

46. Lintschinger, J., Schramel, P., Halatak-Rauscher, A.,Wendler, I. and Michalke, B., Fresenius’ J. Anal. Chem.,362, 313 (1998).

47. Szpunar, J. and Lobinski, R., Pure Appl. Chem., 71, 899(1999).

48. Hann, S., Zenker, A., Galanski, M., Bereuter, T. L.,Stingeder, G. and Keppler, B. K., Fresenius’ J. Anal.Chem , 370, 581 (2001).

49. Wang, J., Houk, R. S., Dreessen, D. and Wiederin, D. R.,J. Biol. Inorg. Chem., 364, 546 (1999).

50. Jakubowski, N., Thomas, C., Klueppel, D. andStuewer, D., Analysis , 26, M37 (1998).

51. Welz, B., Spectrochim. Acta, Part B , 53, 169 (1998).52. Vanhaecke, F. and Moens, L., Fresenius’ J. Anal. Chem ,

364, 440 (1999).53. Agilent GC-ICP-MS Interface Technology, Agilent Tech-

nologies application note, 5988-3071EN (2001).54. De Smaele, T., Verrept, P., Moens, L. and Dams, R.,

Spectrochim. Acta, Part B , 50, 1409 (1995).55. Montes Bayon, M., Gutierrez Camblor, M., Garcia

Alonso, J. I. and Sanz-Medel, A., J. Anal. At. Spectrom.,14, 1317 (1999).

56. Ritsema, R., De Smaele, T., Moens, L., de Jong, A. S.and Donard, O. F. X., Environ. Pollut., 99, 271 (1998).

57. Vercauteren, J., De Meester, A., De Smaele, T., Van-haecke, F., Moens, L., Dams, R. and Sandra, P., J. Anal.At. Spectrom., 15, 651 (2000).

58. Leal-Granadillo, I. A., Garcia-Alonso, J. I. and Sanz-Medel, A., Anal. Chim. Acta, 423, 21 (2000).

59. Armstrong, H. E. L., Corns, W. T., Stockwell, P. B.,O’Connor, G., Ebdon, L. and Evans, E. H., Anal. Chim.Acta , 390, 245 (1999).

312 DETECTION

60. Jantzen, E. and Prange, A., Fresenius’ J. Anal. Chem.,353, 28 (1995).

61. Peycheran, C., Lalere, B. and Donard, O. F. X., Environ.Sci. Technol., 34, 27 (2000).

62. Amouroux, D., Tessier, E., Peycheran, C. and Donard,O. F. X., Anal. Chim. Acta, 377, 241 (1998).


64. Feldmann, J., Krupp, E. M., Glindemann, D., Hirner,A. V. and Cullen, W. R., Appl. Organomet. Chem., 13,739 (1999).

65. Peycheran, C., Quetel, C. R., Lecuyer, F. M. M. andDonard, O. F. X., Anal. Chem., 70, 2639 (1998).

66. Gruter, U. M., Kresimon, J. and Hirner, A. V., Frese-nius’ J. Anal. Chem., 368, 67 (2000).

67. Olesik, J. W., Kinzer, J. A. and Olesik, S. V., Anal.Chem., 67, 1 (1995).

68. Lu, Q., Bird, S. M. and Barnes, R. M., Anal. Chem., 67,2949 (1995).

69. Michalke, B. and Schramel, P., J. Chromatogr. A, 750, 51(1996).

70. Majidi, V. and Miller-Ihli, N. J., Analyst , 123, 803(1998).

71. Van Holderbeke, M., Zhao, Y., Vanhaecke, F.,Moens, L., Dams, R. and Sandra, P., J. Anal. Atom. Spec-trom., 14, 229 (1999).

72. Taylor, K. A., Sharp, B. L., Lewis, D. J. and Crews,H. M., J. Anal. Atom. Spectrom., 13, 1095 (1998).

73. Kirlew, P. W., Caruso, J. A. and Castillano, M. T. M.,Spectrochim. Acta B , 53, 221 (1998).

74. Kinzer, J. A., Olesik, J. W. and Olesik, S. V., Anal.Chem., 68, 3250 (1996).

75. Liu, Y., Lopez-Avila, V., Zhu, J. J., Wiederin, D. R. andBeckert, W. F., Anal. Chem., 67, 2020 (1995).

76. Schaumloffel, D. and Prange, A., Fresenius’ J. Anal.Chem., 364, 452 (1999).

77. Majidi, V., Qvarnstrom, J., Tu, Q., Frech, W. andThomassen, Y., J. Anal. At. Spectrom., 14, 1993 (1999).

78. Bendahl, L., Gammelgaard, B., Jons, O., Farver, O. andHansen, S. H., J. Anal. At. Spectrom., 16, 38 (2001).

79. Tangen, A., Lund, W., Josefson, B. and Borg, H., J.Chromatogr. A., 826, 87 (1998).

80. Polec, K., Szpunar, J., Palacios, O., Gonzalez-Duarte, P.,Atrian, S. and Lobinski, R., J. Anal. At. Spectrom., 16,567 (2001).

81. Ackley, K. L., Day, J. A. and Caruso, J. A., J. Chro-matogr. A., 888, 293 (2000).

82. Michalke, B. and Schramel, P., J. Chromatogr. A., 834,341 (1999).

83. Michalke, B., J. Anal. At. Spectrom., 14, 567 (1999).84. Michalke, B. and Schramel, P., J. Chrom. A., 807, 71

(1998).85. Prange, A., Schaumloffel, D., Bratter, P., Richarz, A. N.

and Wolf, C., Fresenius’ J. Anal. Chem., 371, 764 (2001).86. Vela, N. P. and Caruso, J. A., J. Biochem. Biophys. M.,

43, 45 (2000).87. Shen, W., Vela, N. P., Sheppard, B. S. and Caruso, J. A.,

Anal. Chem., 63, 1491 (1991).88. Vela, N. P. and Caruso, J. A., J. Chromatogr., 641, 337

(1993).89. Carey, J. M., Vela, N. P. and Caruso, J. A., J. Chro-

matogr. A., 622, 329 (1994).90. Jin, K., Shibata, Y. and Morita, M., Anal. Chem., 63, 986

(1991).91. Willie, S. N., Gregoire, D. C. and Sturgeon, R. E., Ana-

lyst , 122, 751 (1997).92. Gelaude, I., Dams, R., Resano, M., Vanhaecke, F. and

Dams, R., Anal. Chem., 74, 3833 (2002).93. Feng, Y.-L., Chen, H.-Y., Tian, L.-C. and Narasaki, H.,

Anal. Chim. Acta, 375, 167 (1998).94. Feng, Y.-L., Narasaki, H., Tian, L.-C. and Chen, H.-Y.,

At. Spectrosc., 21, 30 (2000).95. Nielsen, J. L., Abildtrup, A., Christensen, J., Watson, P.,

Cox, A. and McLeod, C. W., Spectrochim. Acta, Part B ,53, 339 (1998).

96. Evans, R. D. and Villeneuve, J. Y., J. Anal. At. Spectrom.,15, 157 (2000).

5.4 Plasma Source Time-of-flight MassSpectrometry: A Powerful Toolfor Elemental Speciation

Andrew M. Leach, Denise M. McClenathan and Gary M. HieftjeIndiana University, Bloomington, IN, USA

1 Introduction . . . . . . . . . . . . . . . . . . . . . . 3132 Time-of-flight Mass Spectrometry . . . . . . 3153 Mass Resolution of TOFMS . . . . . . . . . . 315

3.1 Compensation for an initial spatialdistribution . . . . . . . . . . . . . . . . . . . 316

3.2 Compensation for initial energydistribution . . . . . . . . . . . . . . . . . . . 317

4 TOFMS Instrumentation . . . . . . . . . . . . . 3184.1 Orthogonal acceleration

time-of-flight mass spectrometry . . . 3184.2 Axial acceleration time-of-flight

mass spectrometry . . . . . . . . . . . . . 3194.3 Other considerations for TOFMS

with continuous ionization sources 3195 Performance of ICP-TOFMS Systems . . . 322

5.1 Sensitivity, noise and limits ofdetection . . . . . . . . . . . . . . . . . . . . 322

5.2 Mass resolving power andabundance sensitivity . . . . . . . . . . . 322

5.3 Isotope ratio measurement . . . . . . . . 3235.4 Mass analyzer comparison . . . . . . . 323

6 Hyphenated TOFMS Speciation Analysis 3246.1 Gas chromatography . . . . . . . . . . . . 3246.2 Liquid chromatography . . . . . . . . . . 3266.3 Capillary electrophoresis . . . . . . . . . 3266.4 Electrothermal vaporization . . . . . . . 327

7 Modulated Ionization Source TOFMSSpeciation Analysis . . . . . . . . . . . . . . . . . 3287.1 Two-state modulated systems . . . . . 3287.2 Single-state modulated systems . . . . 330

8 Conclusions . . . . . . . . . . . . . . . . . . . . . . 3309 Acknowledgements . . . . . . . . . . . . . . . . . 332

10 References . . . . . . . . . . . . . . . . . . . . . . . 332

1 INTRODUCTION

To generate speciation information, two types ofdata must be obtained about a sample: which ele-ments are present and what the state or immedi-ate environment of those elements is. The mostcommon methods used to obtain this informationinvolve a combination of separation techniques andelemental detectors. These systems differentiatespecies on the basis of selective retention (chro-

matography), electrophoretic mobility (electropho-resis), or volatility (electrothermal vaporization)prior to optical or mass spectral elemental detec-tion. A more novel method for the genera-tion of speciation information that is unique tomass spectrometry is the use of a modulatedor pulsed ionization source that provides bothatomic and molecular or fragmentation informationabout the sample in a sequential and repeti-tive fashion.

314 PLASMA SOURCE TIME-OF-FLIGHT MS

Restricted measurement time is a commonfeature of virtually all speciation techniques. Sep-aration methods produce analyte peaks with lim-ited widths that arrive sequentially at the detector.Similarly, modulated ionization sources producerapidly alternating windows of analyte informa-tion. The measurement of such transient signalscan be extremely difficult with conventional mass-spectral systems.

Historically, the most common mass spectrom-eters used in elemental analysis are sequentiallyscanned systems such as the quadrupole mass fil-ter or the scanned sector-field mass spectrometer.With scan-based systems, elemental information isgenerated by selectively allowing ions of a sin-gle mass-to-charge ratio (m/z) to traverse the massspectrometer and be detected. To produce a spec-trum, the mass-selection device is scanned in asequential manner. In most laboratory situationsin which analysis time is not limited, scanned sys-tems provide excellent sensitivity and good spec-tral resolution. However, when measurement timeis restricted, the fact that only one m/z is measuredat a time produces an inverse relationship betweenthe best achievable sensitivity (and thus precision)and the maximum number of elements and isotopesthat can be monitored. Equation (5.4.1) describesthe relationship between sensitivity and mass cov-erage range for scanned systems when analysistime is limited:

t =∑

n

c

r(5.4.1)

In this equation, t is the available analysis time,which is related, for example, to the length of achromatographic peak, c is the number of signal(ion) counts needed for a given m/z to achievea desired level of precision and sensitivity, n isthe number of m/z to be monitored, and r is thecount rate (ions s−1) for a given m/z. If the analysistime t is held constant, the only way to enhancethe sensitivity or precision of a measurement is toincrease the number of counts (c) collected for anindividual m/z; consequently, fewer elements (n)can be monitored.

An additional problem associated with themeasurement of transient signals by means ofscanned detection systems is the quantitation error

known as spectral skew [1]. Transient signals, suchas those produced by most separation techniques,exhibit a temporally changing concentration pro-file. Spectral skew is the relative enhancement orsuppression of signals from adjacent m/z that isproduced by time-dependent changes in analyteconcentration that are on the same time scale asthe scan time of the detection system. The combi-nation of the ever-increasing speed of modern sep-aration systems and the instrumentally restrictedmaximum scan rate of quadrupoles and sector-fieldinstruments will cause spectral skew to become anincreasingly troublesome difficulty.

Trapping mass analyzers such as the quadrupoleion trap and Fourier transform (FT) ion cyclotronresonance mass spectrometer eliminate not onlyspectral skew but also the tradeoff among sensitiv-ity, precision and mass coverage range, since theysimultaneously extract all m/z from the ionizationsource [2]. Additionally, the ability to selectivelymodify or remove trapped m/z through ion chem-istry makes trap-based analyzers highly attractive.However, limited ion capacity (104 to 106 ions)has restricted the sensitivity and dynamic rangeof these systems. Furthermore, when operated ineither a scanned or simultaneous (FT or image-current) detection mode, trapping spectrometersoffer relatively low duty cycles. Their restrictedspectral generation rate limits the applicabilityof trapping systems to the measurement of tran-sient signals.

Sector-field instruments equipped with multiplediscrete detectors or detector arrays are anotherclass of mass spectrometer that simultaneouslyextracts selected m/z and avoids the disadvan-tages of scanned systems for the analysis of tran-sient signals. Both detector geometries benefitfrom a 100 % duty cycle and practically unlim-ited spectral generation rates. However, multide-tector systems are typically limited to monitoringless than ten elements and isotopes per analy-sis [3]. Detector-array systems are able to moni-tor large sections or a complete elemental massspectrum simultaneously, but to date have beenlimited by the performance of available detectortechnology [4].

MASS RESOLUTION OF TOFMS 315

2 TIME-OF-FLIGHT MASSSPECTROMETRY

Time-of-flight mass spectrometry (TOFMS) is oneof the oldest and perhaps simplest forms of massanalysis. First proposed in 1948 by Cameronand Eggers [5], TOFMS separates ions basedon the fundamental relationship between kineticenergy (KE ), mass (m) and velocity (v) shown in(equation 5.4.2):

KE = 12 mv2 (5.4.2)

The acceleration of a population of ions througha potential field so that all ions receive the samekinetic energy will result in a mass-dependentvelocity dispersion, with light ions achievinghigher velocities than heavier ions. Following thisacceleration process, ions are allowed to separatebased on their velocity difference within a field-free drift region commonly known as a flighttube. Detection of the ion arrival times (relatedto velocity) at a detector positioned at the end ofthe flight tube produces a mass spectrum.

Time-of-flight mass spectrometry has recentlyseen a dramatic increase in application, fueledby bioanalytical research that exploits the theo-retically infinite mass range afforded by TOFMSfor the measurement of large peptides and DNAfragments. Although the ability to measure high-mass ions is relatively unimportant in elemen-tal mass spectrometry (commonly limited to

m/z 1–238), other characteristics have motivatedatomic spectroscopists to utilize TOFMS [6–8].Because ions of different mass are detected as afunction of time, TOFMS is not, strictly speak-ing, a simultaneous detection technique. However,since all of the ions that produce a given time-of-flight mass spectrum are extracted simultaneouslyfrom the ionization source, the tradeoff betweensensitivity and mass coverage range inherent toscanned systems is eliminated, and spectral skewis avoided. Additionally, the limited atomic massrange, moderate acceleration potentials (−2000 V),and short flight lengths (1 m) can produce com-plete elemental mass spectra in less than 50 µs.With spectral generation rates greater than 20 kHz,TOFMS is a nearly ideal detection system for tran-sient signals.

3 MASS RESOLUTION OF TOFMS

Resolving power (m/�m) in TOFMS is pro-portional to ion flight time and calculated asR = T/2�t , where T is the total flight time fromthe extraction region to the detector and �t isthe width of the analyte peak. The factor of 2 inthe denominator is a result of the squared rela-tionship between mass and velocity. The ultimateresolving power of TOFMS systems is limited byseveral factors related to the spatial and velocitydistributions of the ion packet extracted for massanalysis. Figure 5.4.1 shows a diagram of a simpletwo-stage acceleration TOFMS instrument. Ions

R G0 G1Detector

E1

s a D

E2 Field-Free Region

Figure 5.4.1. Diagram of a two-stage acceleration time-of-flight mass spectrometer. The extraction region is defined by therepeller plate R and gridded electrode G0, having a width of s and a potential difference of E1. The acceleration region isdefined by gridded electrodes G0 and G1, having a width of a and a potential difference of E2. The field-free drift region has alength of D that ends at the ion detector.


enter the mass spectrometer and fill the extractionregion defined by the repeller plate (R) and agridded electrode (G0). While ions are filling theextraction region, R and G0 are held at the samevoltage, ordinarily ground potential. At a set delaytime, the repeller is pulsed to a high positive volt-age (for the analysis of positive ions). This eventproduces a linear potential field between R andG0 and causes ions within the extraction region tomove toward G0. Following the extraction regionis a second, constant-acceleration stage defined bytwo gridded electrodes, G0 and G1. Acceleratedions then enter a field-free drift region where massseparation occurs, ending at the detector.

3.1 Compensation for an initial spatialdistribution

If a packet of isomass ions is accelerated throughthe same potential difference, the ions will achievean equal velocity and arrive at the detectorwith a time spread corresponding to their initialspatial distribution in the direction of the flighttube. This situation will obviously compromiseresolution. To compensate for the length of theextraction region, a technique known as spacefocusing was conceived and developed by Wileyand McLaren [9]. Application of a voltage gradientacross the extraction region produces a relationshipbetween the initial position of an ion and thepotential through which that ion is accelerated.With a positive voltage applied to the repeller plate,positive ions of a given m/z that are positionedclose to the repeller are accelerated to a greaterfinal velocity than ions closer to the G0 electrode.The difference in velocity of isomass ions causesthe ions with elevated velocities to overtake theslower ions within the flight tube. The positionwhere the spatial distribution of the extracted ionpacket is minimized is known as the space-focusplane. Conveniently, the position of the space-focus plane is the same for ions of all m/z.

Optimal mass resolving power requires com-pensation for the initial spatial distribution ofions within the extraction region. Equation (5.4.3)describes the basic premise of space focusing,

where two ions, one positioned in the middle ofthe extraction region (s1/2) and one positioned adistance �s from s1/2, produce a distribution offlight times.

t (s1/2 ± �s) − t (s1/2) = minimum (5.4.3)

To achieve the highest possible mass resolutionthe temporal spread of ion arrival times at thedetector must be minimized. Equation (5.4.3) canbe analyzed with a Taylor series expansion toproduce equation (5.4.4):

∞∑n=1

tn(s1/2)

n!(±�s)n − t (s1/2)

= t ′(s1/2)(±�s) + 1

2t ′′(s1/2)(±�s)2 + · · ·

(5.4.4)In this power series, t ′ and t ′′ denote first- andsecond-order coefficients. To achieve enhancedresolving power, successively higher-order coeffi-cients must be included in calculating the locationof the optimal space-focus plane. First-order spacefocusing occurs when �t = 0 for the t ′ term, whilesecond-order space focusing is achieved by simul-taneous solution of the t ′ and t ′′ terms.

Analysis of the equations of motion for ionswithin the TOF system shows that the ratioof the field strengths within the extraction (E1)and acceleration (E2) regions can be used tocontrol the quality and position of the space focusplane [10]. With the two-stage acceleration systemshown in Figure 5.4.1, a solution for first-orderspace focusing finds the position of lowest spatialdispersion (D) to be:

D = s

[(1 − g

f

)(√1 + 2f

)3 + g

(2 + 1

f

)]

(5.4.5)

where s and a are the widths of the extractionand acceleration regions, respectively, g = a/s andf = (E2/a)/(E1/s). Higher mass resolution canbe accomplished with second-order space focusingif specific operational conditions are observed.Second-order space focusing is achieved whenE1 = E2, where a space focus plane is producedat D = s + 2a.

MASS RESOLUTION OF TOFMS 317

At this point, it should be noted that thequadratic relationship between ion velocity andinitial position within the extraction region pro-hibits the complete solution of the Taylor seriesexpansion with the previously described two-stagelinear field system. However, use of a quadraticextraction field would produce a linear relation-ship between ion velocity and initial position andthus allow ideal space focusing [11].

Additionally, second-order space focusing typi-cally produces a focal plane close to the extractionregion, by which location little mass dispersion hasbeen achieved. Since resolution is directly relatedto total flight time, only low resolving powers areobtained at this primary space-focus plane. How-ever, as will be discussed in the following section,the space focus plane is ideally suited to act as avirtual source for an ion mirror that re-images theplane at a greater flight length.

A full mathematical description of space focus-ing is beyond the scope of the present text. For amore in-depth review of space-focusing theory, thereader is referred to the literature [9, 10, 12–14].

3.2 Compensation for initial energydistribution

The equations of motion for space focusing arecommonly solved by assuming that the extractedions have no initial energy and thus an energydistribution of zero. However, in reality, most ion-ization sources used for elemental analysis areoperated at atmospheric pressure, which requiresthe use of a vacuum interface to couple the sourcewith the mass spectrometer. Transmission from theionization source to the mass spectrometer gener-ates an isokinetic (equal-velocity) ion beam andthus a mass-dependent energy [15]. The velocityof the ion beam is dictated by the most abundantspecies, argon for an inductively coupled plasma(ICP), but will exist in a Maxwellian distributioncorresponding to the temperature of the ionizationsource. Additionally, potentials imparted by theionization source, such as the offset potential of anICP, will augment the distribution of ion velocitieswithin the TOFMS extraction region. The exis-tence of this energy distribution will produce a

spread of final post-acceleration isomass ion veloc-ities and thus will result in degraded resolution.

The most common method used to compen-sate for an initial ion energy distribution is anelectrostatic ion mirror, also known as a reflec-tron. Developed by Mamyrin and coworkers [16,17], a properly designed reflectron can significantlyreduce timing errors caused by initial energy dis-tributions of up to 10 % of the total accelerationpotential. Ion mirrors are potential ramps consist-ing of a series of resistively separated ring elec-trodes (Figure 5.4.2). High-energy ions penetratefarther into the reflectron than ions of lower energyand thus follow a longer path to the detector. Thisenergy-dependent flight length causes the fasterions to travel a longer distance and to arrive at thedetector coincident with slower ions that traverseda shorter path. A second function of the ion mir-ror is to re-image the original space focus plane.The bent ion path of a reflectron-equipped TOFinstrument can effectively double the flight lengthof the system without an increase in the instrumentsize. This enhanced flight length will result in anadditional gain in mass resolving power.

A special resolution consideration arises fromthe possibility that ions might be moving towardthe repeller (away from the detector) prior to theirextraction for mass analysis. These ions would

V

R1R2

FT

Figure 5.4.2. Energy compensation with a two-stage reflectionion mirror. The reflectron is a voltage (V ) ramp composedof a retardation region (R1) and a reflection region (R2). Ionswith different energies disperse within the flight tube (FT ).High-energy ions (ž) penetrate the reflectron to a greaterextent than lower energy ions (Ž) and thus travel a greaterdistance within the ion mirror. High-energy ions catch up withlower-energy ions in the second passage though the flight tube,to arrive simultaneously at the detector.


possess a negative initial velocity with respect totheir final velocity imparted by the TOF extractionand acceleration processes. Upon initiation of therepeller pulse these ions would first decelerateprior to being re-accelerated toward the detector.This turnaround time error is difficult to overcomeand can result in degraded mass resolution.

4 TOFMS INSTRUMENTATION

Although several time-of-flight geometries havebeen explored, two systems, orthogonal accelera-tion (oa) and axial acceleration (aa) have enjoyedthe widest acceptance. These systems differ in therelative angle between the axis of the primary ionbeam from the ionization source and the axis ofthe TOF flight tube. Each system possesses advan-tages relative to the other and will be discussedfrom a historical standpoint here.

4.1 Orthogonal acceleration time-of-flightmass spectrometry

Hieftje and coworkers [18–21] performed the initialexperiments that coupled an ICP source to a time-of-flight mass spectrometer with an instrument basedon orthogonal acceleration. Figure 5.4.3 shows aschematic diagram of the orthogonal accelerationprocess. In this system, the primary ion beam entersthe extraction region between the repeller plate andthe G0 electrode. At a set time, the repeller is pulsedto a positive voltage and the ions are acceleratedinto the flight tube, positioned at a 90◦ angle tothe original axis of ion propagation. Because themost significant initial ion energy distribution isoriented perpendicular to the TOF axis, orthogonalextraction systems can achieve relatively high massresolving power (>2000 FWHM). Additionally, thewidth of the primary ion beam in the direction ofthe flight tube can be restricted by optics positionedprior to the extraction region. As a result, lessstringent demands are placed upon space focusing,so resolution is improved [20].

Although the initial velocity component ofthe primary ion beam, oriented perpendicular to

ExtractionR

G0

G1

Non-extraction

Figure 5.4.3. Diagram of an orthogonal acceleration TOFMSsystem. The extraction region is defined by the repeller plate(R) and a gridded electrode (G0), commonly held at groundpotential. The acceleration region extends from G0 to a secondgridded electrode (G1), held at flight-tube potential. While ionsare entering the extraction region the repeller is held at groundpotential. Extraction of ions for mass analysis is initiated whenthe repeller is pulsed to a positive potential, injecting ions intothe flight tube at a 90◦ angle with respect to their originalaxis of propagation. Arrows denote the relative direction andmagnitude of ion velocity vectors, but are not drawn to scale.

the field-free region, does not significantly limitmass resolving power, this energy produces manyof the instrumental difficulties experienced withoa-TOFMS systems. The isokinetic ion beamextracted from the ionization source results ina mass-dependent energy perpendicular to theTOFMS axis. Light ions have low initial energyand are easily redirected by the repeller pulse.However, heavier ions possess more energy andthus follow a more curved path to the detector.Differences in the flight path of ions result in beambroadening that can require the use of an extendedextraction zone and larger, more expensive detec-tors that have elevated susceptibility to noise. Ini-tial experiments with ICP-oa-TOFMS attemptedto correct for this mass-dependent ion trajectoryby means of steering plates positioned within theflight tube [19]. Adequate mass dispersion hasoccurred at the steering plates to allow the applica-tion of a voltage ramp to the plates that redirects allm/z along roughly the same flight path. Althoughthis compensation technique reduced beam diver-gence, some loss in mass resolving power wasexperienced.

Guilhaus [22] has demonstrated a simpler tech-nique, known as spontaneous drift, that allows

TOFMS INSTRUMENTATION 319

ions to follow their natural, mass-dependent path.Although this method results in a broadened ionbeam and can require the use of enlarged reflec-trons and detectors, spontaneous drift avoids theslight degradation of mass resolving power causedby steering plates. A smaller detector can be usedwith the spontaneous drift technique if a longerextraction region is employed [23]. By taking lightions from the leading edge of the ion packet whiledetecting heavy ions from the trailing edge, thespatial distribution within the extraction region canbe used to compensate for the mass-dependention trajectory.

4.2 Axial acceleration time-of-flight massspectrometry

To address several of the limitations of oa-TOFMS, a second geometry has been developedthat accelerates ions coaxially to their initial axis ofpropagation from the ionization source. A diagramof the axial acceleration process is shown inFigure 5.4.4. Similar to oa-TOFMS, a two-stageacceleration system is employed. However, in anaxial system, the repeller plate is replaced with a

ExtractionR

G0

G1

Non-extraction

Figure 5.4.4. Schematic diagram of an axial accelerationTOFMS system. The extraction region is defined by the griddedrepeller electrode (R) and a second gridded electrode (G0),commonly held at ground potential. The acceleration regionextends from G0 to a third gridded electrode (G1), held atflight-tube potential. Ions enter the extraction region throughthe repeller (held at ground potential). Extraction of ions formass analysis is initiated when the repeller is pulsed to apositive potential, injecting ions into the flight tube at a 0◦angle with respect to their original axis of propagation. Arrowsdenote the relative direction and magnitude of ion velocityvectors, but are not drawn to scale.

gridded electrode. Ions pass through the repellerelectrode to enter the TOF extraction region.This simple alteration results in several importantchanges to the operational characteristics of an aa-TOFMS system.

With an aa-TOFMS system, the axis with thelargest initial ion energy spread is positioned col-inear with the TOFMS axis. This coaxial geome-try provides both benefits and disadvantages whencompared to an orthogonal acceleration system.Since the isokinetic energy of the primary ionbeam is directed along the TOFMS axis, themass-dependent trajectory to the detector expe-rienced with orthogonal systems is significantlyreduced [7]. The lower energy distribution per-pendicular to the TOFMS axis also lessens beamdivergence within the field-free region and resultsin greater ion transmission efficiency to the detec-tor. Conversely, the increased energy distribu-tion directed along the TOF axis places greaterdemands on the energy-compensation ability of thereflectron ion mirror. Additionally, to improve theefficiency of ion utilization in a TOFMS, it is desir-able to make the length of the extraction zone aslong as possible. Because this length is in the direc-tion of the flight tube in an aa-TOFMS, greatercare is needed to achieve the best space focusing.

4.3 Other considerations for TOFMSwith continuous ionization sources

The open geometry of TOF systems requires thatunwanted ions be strictly controlled to reducenoise and interferences. Ions that reside in theextraction region during a repeller pulse areaccelerated into the flight tube for analysis. Thearrival of these ions at the detector is temporallyreferenced to the extraction pulse and proportionalto mass. Ions that pass through the extractionregion when the repeller is at ground potentialwill also be accelerated into the flight tubeand can result in detector events. However, theflight time of a nonextracted ion cannot becalculated since there is no referenced start time.Because nonextracted ions will regularly enter theacceleration region, these ions will be seen as acontinuous background signal.


Adventitiously, extracted and background ionscan be differentiated on the basis of theirenergy [24]. Extracted ions will attain an energyproportional to the sum of the repeller and acceler-ation potentials. In contrast, background ions willreceive energy only from the acceleration regionsince they enter that region at a time when arepeller pulse is not being applied. Placement of anenergy barrier with a potential slightly higher thanthe acceleration voltage immediately prior to thedetector will stop background ions while passingextracted ions. As shown in Figure 5.4.5, a sim-ple energy barrier can be constructed from a seriesof three gridded electrodes, of which the exteriorelectrodes are held at flight-tube potential whilethe center electrode provides the energy discrimi-nation.

Use of an energy discrimination system cansignificantly reduce continuum ion background.However, although ions can be stopped by apotential field, neutral species will be unaffected.

DEDR G1G0 FT

(a)

(b)

Figure 5.4.5. Effect of energy discrimination system on con-tinuum background ions. Ion optics consists of: repeller (R),grounded electrode (G0), acceleration electrode (G1), flighttube (FT ), energy discrimination electrodes (ED), and detector(D). The middle ED electrode is held at a positive potential toreject continuum background ions (Ž). (a) Repeller potentialat ground. None of the ions entering the flight tube has suf-ficient energy to overcome ED barrier. (b) Repeller electrodeis pulsed to a positive potential to start time-of-flight anal-ysis sequence. Ions (ž) present within the extraction region(defined by R and G0) during this pulse gain additional energyand are subsequently able to pass the ED barrier to result indetector events.

Ions that are neutralized after acceleration, mostlikely through charge exchange with residual gasin the flight tube, will retain a significant portionof their initial energy. If charge exchange occursbetween the reflectron and the detector, the high-energy neutrals that are produced will traverse theenergy discrimination barrier and result in detectorevents and elevated background.

To limit the production of high-energy neutrals,the pressure inside the flight tube must be as lowas possible and the number of ions (especiallyAr+) admitted to the flight tube must be restricted.Time-of-flight mass spectrometry is an inherentlypulsed technique. However, most of the ionizationmethods used for elemental analysis produce ionsin a continuous fashion. To reduce the number ofunwanted ions presented to the mass spectrometer,the primary ion beam can be modulated.

The position of the flight tube at a 90◦ anglerelative to the axis of the primary ion beammakes the orthogonal extraction process a nearlyideal modulation approach. In theory, all ionsenter the extraction region from one side and exitthrough the other. Only when the repeller plate ispulsed to a positive potential are ions redirectedinto the flight tube for mass analysis. However,ion scattering occurs; also, the radial velocity ofsome ions in the primary ion beam is sufficientto cause unwanted ions to gain entrance to theflight tube [24]. These undesired ions can thenundergo charge exchange and result in elevatedbackground noise.

To further restrict the number of unwantedions within the TOFMS system, an ion gatecan be positioned ahead of the extraction region(Figure 5.4.6). In a fashion similar to that usedin the energy discrimination system, a modulationgate can be constructed of three gridded electrodes.The two exterior electrodes (M1 and R) aremaintained at ground potential while the centerelectrode (M2), held at a positive potential, stopsthe passage of ions. During a defined temporalwindow, the gate electrode is brought to a lowpotential to transmit ions. Returning the gateelectrode to its original elevated potential thenproduces an ion packet. Other modulation gatesincorporate steering plates or lenses to deflect the

TOFMS INSTRUMENTATION 321

M1 M2

M2

G0R M1 M2 G0R M1 M2 G0R

R

0+

0+

Time

Potential

Figure 5.4.6. Temporal potential sequence of a three-electrode pre-acceleration modulation system. Modulation electrode M1remains at ground potential throughout analysis. With modulation grid M2 at a positive potential all ions (Ž) are prevented fromentering the TOF extraction region (defined by repeller R and ground electrode G0). When M2 is dropped to ground potential, ionspass through the modulation system to fill the extraction region. After a set delay, M2 is returned to its original potential, impedingthe transmission of additional ions and generating an ion packet (ž). A subsequent delay allows ions within the modulationsystem to enter the TOF extraction region. Finally, R is pulsed to a positive potential to start the TOF analysis process.

primary ion beam [20]. Although beam modulationcan be beneficial in an orthogonal extractionsystem, modulation is critical to the performanceof axial systems, in which the primary ion beam isoriented directly into the flight tube.

While the simultaneous extraction of all m/z

has proven to be one of the keys to the successof TOFMS for elemental analysis, it imparts somedifficulties. The primary ion beam of most conven-tional plasma sources is composed predominantlyof ionized support gas and solvent vapor. Argonin an ICP can be greater than 200-fold more abun-dant than analytes present at the mg kg−1 level.Additionally, ambient gas and solvent ions includ-ing N+, O+, OH+, N2

+, NO+, O2+ and several

argon-containing polyatomics can be present athigh concentrations. For ultratrace measurements,the disparity between matrix and analyte concen-trations is great. Therefore, matrix ions must beremoved from the extracted ion packet to avoiddetector saturation while maintaining sufficientdetector gain for analyte measurements.

The most common technique used to removeunwanted ions from the extracted ion packet isselective ion deflection. Although ion deflectioncan be accomplished via a number of instrumentalconfigurations, the basic operation is the same.

Ions are extracted into the TOF drift region andallowed to undergo partial mass separation. Adeflection system as simple as two parallel steeringplates is positioned at the first space focus plane.Application of a potential pulse to one platewill selectively alter the trajectory of ions withinthe deflection region, causing them to miss thedetector. A sequence of pulses can be used toremove more than one m/z window.

The performance of an ion deflection systemcan be evaluated on the basis of several crite-ria, most notably deflection efficiency (the ratio ofions in a selected mass window with and with-out the application of a deflection pulse) anddeflection-window width (important for minimiz-ing the effects of ion deflection on adjacent m/z).The temporal width of a deflection window isdefined by the combination of the physical dimen-sions of the deflection system in the time-of-flightaxis and the width of the voltage pulse. Deflec-tion efficiency is dictated by the applied potentialused to steer the unwanted ions, and the lengthof time the selected ions experience the deflectionfield (proportional to deflection window width). Asshould be evident from the previous statements,the efficiency and temporal width of deflectionsystems are usually inversely related. Traditional


parallel-plate systems provide good deflection effi-ciency, ranging from 103 to 104, but are limitedin temporal narrowness by the physical dimen-sions of the plates. As a result, they often resultin the simultaneous deflection of several m/z evenat relatively low masses (<40 m/z). An alternativedeflection system described by Vlasak et al. [25]consists of a series of closely spaced parallelwires; pulses of opposite potential are simultane-ously applied to adjacent wires. This system iscommonly referred to as a comb deflector dueto its physical appearance. The temporal fidelityachieved with a comb deflector is significantly bet-ter than that of parallel-plate systems due to thenarrow region (in the time-of-flight axis) in whichions experience the applied deflection pulse. How-ever, as expected, the deflection efficiency of combdeflectors is typically worse than parallel-plate sys-tems, ranging from 102 to 103.

5 PERFORMANCE OF ICP-TOFMSSYSTEMS

Although time-of-flight mass spectrometers havebeen used with a number of plasma ioniza-tion sources including glow discharges (GD) andmicrowave plasma torches (MPT), this section willfocus on ICP-based systems due to their recent suc-cess. Two commercial ICP-TOFMS instrumentsare already available, the orthogonal accelera-tion Optimass 8000 (GBC Scientific EquipmentPty. Ltd., Dandenong, VIC, Australia) and theaxial acceleration Renaissance (LECO, St. Joseph,MI, USA). This section compares the charac-teristics of several instruments, both academicand commercial.

5.1 Sensitivity, noise and limits ofdetection

Theoretically, it is expected that the in-line geom-etry of the axial acceleration system should exhibitlower beam divergence and better ion-transmissionefficiency than the orthogonal system. In turn,increased transmission efficiency would result inimproved sensitivity. However, the sensitivity of

both commercial instruments is found to be inthe range of 500 to 10 000 cps µg L−1 per m/z

(range caused by heavy-mass bias in both instru-ments) [26, 27]. The similar sensitivities of thetwo acceleration geometries can be explained bythe relative size of each instrument’s extractionregion. To gain sensitivity in an orthogonal accel-eration system the length of the extraction regioncan be increased to improve the instrument dutycycle (percentage of ions utilized). However, alonger extraction region in an axial accelera-tion instrument will place greater demands onthe space-focusing abilities of the typical two-stage acceleration system and might result indegraded resolution.

The folded flight path produced by the pres-ence of a reflectron minimizes photon-related noisein both acceleration geometries. The orthogonalacceleration system demonstrates very low detec-tor background with less than one count per secondper m/z for most elements [26]. As would beexpected, the axial acceleration instrument exhibitsslightly higher background (1–10 cps per m/z)due to the greater likelihood of high-energy neu-trals within the flight tube [27]. The combinationof comparable sensitivity and reduced noise pro-vided the orthogonal acceleration instrument withsomewhat better limits of detection than the axialsystem. Depending upon operating conditions, theaxial instrument exhibited detection limits in thesingle ng L−1 range, while the orthogonal systemwas typically in the sub-ng L−1 regime. The limitsof detection for both instruments should improveas the instruments mature.

5.2 Mass resolving power and abundancesensitivity

In general, the mass resolving power of ICP-TOFMS instruments improves with increased m/z.Although the temporal separation between adjacentm/z peaks is greater for low masses, higherresolution is achieved for heavy ions due to longerflight times. The orthogonal geometry producedresolving powers (FWHM) of 500 for 6Li and 2200for 238U [26]. The axial system specifies a lower

PERFORMANCE OF ICP-TOFMS SYSTEMS 323

resolving power (but at the more stringent criterionof 10 % peak height) of 615 for 209Bi due to theincreased demands on space focusing and elevatedinitial energy spread [27]. Work performed ona similar axial system in this laboratory hasresulted in resolving powers (FWHM) rangingfrom 410 for 7Li to 1140 for 209Bi [28]. Althoughboth instruments achieved resolution significantlygreater than convention quadrupole-based systems(operated in the first Mathieu stability region),resolution was inferior to sector-field instrumentsand insufficient for separation of important ICP-MS isobars such as 52Cr with 40Ar12C (R = 2400)and 56Fe with 40Ar16O (R = 2500).

Abundance sensitivity is the effect of adjacentm/z on one another, calculated in terms of equiv-alent concentrations. High abundance sensitivitymeans that trace measurements can be made inthe presence of high concentrations of an ele-ment at an adjacent m/z. The orthogonal accel-eration TOFMS produced abundance sensitivitiesof 2.8 × 106 on the low-mass side and 1.4 × 104

on the high-mass side [26]. A noncommercialaxial system provided abundance sensitivities of6.7 × 105 on the low-mass side and 1.7 × 104 onthe high-mass side [29]. Both geometries achievedlow-mass side abundance sensitivities similar toquadrupole instruments, but were worse on thehigh-mass side due to peak tailing. It is worth not-ing that the abundance sensitivity measured withquadrupole systems is typically better for the high-mass side than for the low-mass side (opposite tothe trend seen with TOF-MS instruments).

5.3 Isotope ratio measurement

Multiplicative noise in the ionization source isa major cause of degraded isotope-ratio preci-sion, especially when the source fluctuations areon the same time scale as the scan rate of themass spectrometer [30]. Although faster scan ratesimprove ratio precision, all m/z must be sam-pled from the plasma simultaneously to com-pensate fully for ionization-source multiplicativenoise. Mahoney et al. [6] demonstrated that thesimultaneous extraction capability of ICP-TOFMS

could provide isotope ratio precision on the orderof 0.056 % RSD. Good agreement with valuespredicted by counting statistics suggests that fur-ther improvements in precision are possible withlonger integration times. The commercial orthogo-nal acceleration instrument provided isotope-ratioprecision that was as good as 0.2 % RSD (depend-ing upon concentration and integration time, asexpected) [26]. At higher concentrations, Van-haecke and coworkers [31] reported precision bet-ter than 0.05 % RSD with a commercial axialacceleration system. The measurement precisionof both commercial instruments was shown toimprove based on counting statistics up to inte-gration times between 30 and 60s [26, 27]. Beyondthese integration lengths other noise sources appearto be dominant in the determination of isotope-ratio precision.

Instrumental mass bias and, even more impor-tant, its stability, are important factors that oftenlimit isotope ratio accuracy. In general, mass biasfor the two commercial ICP-TOFMS instrumentsis found to drop with increased m/z [26, 32]and to be similar in magnitude to that reportedfor quadrupole- and magnetic sector-based instru-ments [33]. Mass-bias fluctuation and drift havebeen reported with both commercial ICP-TOFMSsystems. Sturgeon et al. [26] attributed minorshort-term variations in mass bias experienced withthe orthogonal acceleration instrument to changesin ion-optic voltages. With the axial accelerationsystem, Emteborg and coworkers [32] reportedsudden, unexplained shifts in mass bias thatdegraded the long-term stability of the instrument.However, the authors noted that the rapid dataacquisition speed of TOF-MS instruments limitedthe need for long-duration measurements. As withother types of ICP-MS instruments, proper calibra-tion will allow accurate isotope ratio measurementsto be performed with ICP-TOFMS systems.

5.4 Mass analyzer comparison

Table 5.4.1 displays a comparison of the figuresof merit typically achieved by the three types ofmass analyzers currently competing in the com-mercial ICP-MS market. Instruments based on


Table 5.4.1. Comparison of mass analyzer figures of merit.

ICP-QMSa ICP-SFMSa,b ICP-TOFMS

Sensitivity(cps µg L−1)

105 106 103 –104 per m/z

Background (cps) 1–10 <0.1 1–10 per m/zResolving powerc Unit Mass Up to 10 000 Up to 1500Abundance

sensitivityd106/107 106/106 106/104

Isotope ratioprecisione (%RSD)

0.1 0.05 0.02

Ion sampling mode Sequential Sequential SimultaneousSpectral generation

rate2000–3000 m/z s−1

10 full spectra s−1350 m/z s−1

1.7 fullspectra s−1

6 000 000 m/z s−1 f

20 000 fullspectra s−1

Cost $ $$ $

aRef. 34.bRef. 35.c10 % valley definition.d Low-mass side/high-mass side.e10 min integration time.f Based on 300 m/z measured at 20 000 Hz.

quadrupole mass filters have historically domi-nated research in the field of ICP-MS. Quadrupolesystems are in general robust and of relativelylow cost while offering good sensitivity. Lowresolving power, typically unit-mass resolution,and relatively high noise have been the main dis-advantages of ICP quadrupole mass spectrome-ters (QMS) [34]. Sector-field mass spectrometers(SFMS) combine excellent sensitivity and lownoise to produce the best limits of detection cur-rently achievable with ICP-MS, often less than0.1 pg g−1 [34, 35]. Additionally, most ICP-SFMSsystems can be operated in a high-resolution mode(R > 7500) that eliminates the majority of com-mon elemental isobaric interferences. Criticismof sector-field instruments typically stems fromtheir high cost and slow scan speeds. Fast spec-tral generation rates and the potential of high-resolution measurements have resulted in therecent interest in the use of time-of-flight massspectrometers for elemental analysis. However,the widespread acceptance of ICP-TOFMS is cur-rently hindered by somewhat higher detection lim-its. It is important to note that ICP-TOFMS isthe least mature of the three classes of mass ana-lyzer compared here, so it can be expected thatthe performance of time-of flight instruments willrapidly improve.

6 HYPHENATED TOFMSSPECIATION ANALYSIS

The most obvious method for the generationof speciation information is to couple a sepa-ration system with an element-selective detec-tor. Several outstanding reviews on the topic ofhyphenated speciation analysis have recently beenpublished [36–39]. The following section high-lights hyphenated speciation systems that employplasma source time-of-flight mass spectrometry(PS-TOFMS).

6.1 Gas chromatography

Capillary gas chromatography (CGC) is a conve-nient method for the separation of volatile speciesbased on their selective partitioning between sta-tionary and mobile phases. Separations typicallyrequire several minutes with analyte peak widthsas narrow as hundreds of milliseconds. To provideadequate temporal resolution, single m/z samplingrates of at least 10 Hz are typically desired. Onthis time scale, multielemental analysis of com-plex mixtures with conventional scan-based massanalyzers can be experimentally difficult.

Leach and coworkers [40] demonstrated the useof an ICP-aa-TOFMS instrument for the speciation

HYPHENATED TOFMS SPECIATION ANALYSIS 325

analysis of a simple mixture of organometalliccompounds. Although this study did not analyzecomplex mixtures, the figures of merit for the useof ICP-TOFMS with gas chromatography wereexplored. The high spectral generation rate ofTOFMS provided excellent temporal resolution offast transient signals. Simultaneous extraction ofall m/z provides the capability to perform iso-topic analysis on GC peaks. Figure 5.4.7 illus-trates the detection of all ten tin isotopes duringthe separation of tetramethyl- and tetraethyltin.Calculation of the 118Sn/120Sn isotopic distribu-tion for ten injections containing 5 pg of tin percompound produced an isotope ratio accuracy of0.28 % with a precision of 2.88 % RSD. This pre-cision was found to be in good agreement withcounting statistics. Detection limits in the tens offemtograms (calculated as the metal) per injectionwith a dynamic range of six orders of magni-tude were demonstrated for three organotin andorganolead compounds.

A helium microwave plasma torch (MPT) wascoupled to an oa-TOFMS system for the detec-tion of halogenated hydrocarbons separated bygas chromatography [41]. Figure 5.4.8 shows thedetection of several chlorine-containing hydrocar-bons at m/z 12 for carbon and m/z 35 for chlo-rine. The simultaneous extraction of all m/z by

112 m/z Ratio120

116

0100

200

50150

0

500

1000

1500

2000

2500

3000

Cou

nts

(100

mse

c in

t.)

Chromatographic Time (sec)

Et4Sn

Me4Sn

Figure 5.4.7. Multi-isotopic detection during the separation oftetramethyltin (Me4Sn) and tetraethyltin (Et4Sn). Injection con-tained 50 pg of tin per compound. Reproduced by permissionof the Royal Society of Chemistry.

0.0

0.100 20 40 60 80 100 120 140

0.20

0.30

0.40

0.50

0.60

0.70

0 20 40 60 80 100 120 140

0.50

1.0

1.5

2.0

Time (sec)

(a)

Time (sec)

(b)

Sig

nal (

V)

Sig

nal (

V)

chlorohexane

chloropentane

chlorobutane

Methanol

Figure 5.4.8. Isotope-specific chromatograms of halogenatedhydrocarbons (chlorobutane to chlorohexane) in methanol.Twin boxcar averagers used for data collection. (a) Signal from12C. (b) Signal from 35Cl. Reprinted with permission fromB. W. Pack, J. A. C. Broekaert, J. P. Guzowski, J. Poehlman,and G. M. Hieftje, Anal. Chem., 70, 3957. Copyright (1998)American Chemical Society.

TOFMS allowed the direct identification of ana-lyte peaks by their distinctive empirical formulas(Figure 5.4.9). Detection limits for the GC-MPT-TOFMS were found to be in the low femtomolerange for all analytes.

In a similar study, Guzowski and Hieftje [42]demonstrated the use of a simple gas-samplingglow discharge (GSGD) coupled with an oa-TOFMS for the detection of halogenated GC elu-ents. This low-pressure ionization source provideda cost-effective alternative to bulkier atmospheric-pressure sources such as the ICP. Additionally,


0.00.0 1.0 2.0 3.0 4.0 5.0

0.10

0.15

0.20

0.25

0.30

0.35

0.50

1.0

1.5

2.0

2.5

3.0

3.5H

e-M

PT

Cl/C

Rat

ios

He-

MP

T C

l/C R

atio

s

Theoretical Cl/C Ratio

(b)

Theoretical Cl/C Ratio

(a)

0.10 0.15 0.20 0.300.25 0.35

Figure 5.4.9. Comparison of Cl/C ratios obtained from flow-cell MPT-TOFMS measurements to values derived fromempirical formulas. (a) Ratios obtained from both aromaticand aliphatic species. Compounds used and correspond-ing Cl/C ratios: chlorotoluene (Cl : C 0.143), chlorobenzene(Cl : C 0.167), dichlorobenzene (Cl : C 0.333), chlorohep-tane (Cl : C 0.143), chlorohexane (Cl : C 0.167), chloropen-tane (Cl : C 0.200), chlorobutane (Cl : C 0.250), chloropropane(Cl : C 0.333), methylene chloride (Cl : C 2.0), chloroform(Cl : C 3.0), and carbon tetrachloride (Cl : C 4.0). Correlationcoefficient, r = 0.994. (b) An expanded view of the low Cl : Cratio region in (a) which consists of a homologous series ofaliphatic halogenated hydrocarbons (chloropropane to chloro-heptane). Reprinted with permission from B. W. Pack, J. A. C.Broekaert, J. P. Guzowski, J. Poehlman, and G. M. Hieftje,Anal. Chem., 70, 3957. Copyright (1998), American ChemicalSociety.

reduced-pressure operation lessened the likelihoodof air entrainment, resulting in simplified massspectra. Speciation analysis was performed on arange of halogenated hydrocarbons with atomicdetection limits in the low- to mid-femtogram persecond regime. A dynamic range greater than threeorders of magnitude was reported. As will be

discussed in the next section, this GSGD could beoperated in both atomic and molecular ionizationmodes to provide further speciation information.

The recent development of multicapillarycolumns [43, 44] and ultrafast GC [45] willlikely foster the application of PS-TOFMS to thespeciation analysis of volatile compounds. Withthese methods producing peak widths as shortas 10 ms, the ability of scan-based analyzers tomonitor multielemental separations will quickly beoverwhelmed. For example, a minimum of fivedata points should be collected per analyte peak toprovide acceptable temporal fidelity [46]. Althoughcommercial quadrupole systems offer scan ratesas high as 3000 amu s−1 [34], measurementswill be limited to fewer than six elementsor isotopes for a 10 ms transient. Additionally,spectral skew will then severely compromise theaccuracy of the quantitative information generatedby scanned systems.

6.2 Liquid chromatography

Conventional liquid chromatography is widelyused in the field of speciation analysis for theseparation of nonvolatile species. However, PS-TOFMS has experienced limited application tothe element-selective detection of LC eluents dueto the relatively slow speed of such separations.Ferrarello et al. [47] demonstrated the separationof several metallothionein-like proteins with anICP-aa-TOFMS instrument. This study showed thepotential of PS-TOFMS for fast protein speciation.The authors noted that the simultaneous extractioncapability of ICP-TOFMS would be attractivefor isotope ratio or isotope dilution experimentswith LC separations. The recent development ofultrahigh pressure LC (pressures between 20 and40 kpsi), which generates peak widths on theorder of 100 to 1000 ms in duration, should resultin the further application of PS-TOFMS to LCspeciation [48].

6.3 Capillary electrophoresis

High separation efficiency and minimal requiredsample volume has led to the increased popularity

HYPHENATED TOFMS SPECIATION ANALYSIS 327

of capillary electrophoresis for elemental specia-tion analysis. Costa-Fernandez et al. [49] demon-strated the use of CE for the rapid separationof a mixture of cations in a cyanide buffer.In addition to V(V), Cr(VI), Ni(II) and Cu(II),speciation of Co(II) and Co(III), and As(III),As(V), and dimethylarsinic acid was performed(see Figure 5.4.10). The entire analysis was com-pleted in less than 70 s, with peak widths on theorder of 1–3 s FWHM. Although the separation didnot fully resolve several of the analytes, element-selective detection provided the ability for iden-tification and quantitation. This study examinedseveral experimental variables including bufferconcentration, flow rates and separation voltages.Additionally, suction flow associated with the useof pneumatic nebulization was addressed.

Like other separation techniques, capillary elec-trophoresis is constantly being driven to faster timescales. Microfluidic technology that allows thecomplete manipulation and separation of analyteson a single microchip-sized substrate providesfast, efficient, low-volume measurements [50].Although microfluidic CE has yet to be used for

elemental speciation with PS-MS, it seems likelyin the near future.

6.4 Electrothermal vaporization

In electrothermal vaporization (ETV), a smallvolume of sample solution, typically ranging from1 to 20 µL, is heated within a graphite furnace.Application of a temperature ramp to the furnacecauses analytes to be released as a function ofvolatility. Mahoney and coworkers [51] used thiselement-selective volatility to provide an orthogo-nal means of enhancing the effective spectral reso-lution of ICP-TOFMS. With an oa-TOFMS systemthat exhibited a nominal resolution of 2000, sev-eral isobars (112Cd/112Sn, 113Cd/113In, 114Cd/114Sn)that would require resolution as high as 300 000could easily be separated.

Although ETV is not commonly associatedwith speciation analysis, differences in compoundvolatility can be used to provide analyte separation.Guzowski et al. [52] demonstrated the speciationof two iron compounds with ETV-GSGD-TOFMS(see Figure 5.4.11). A furnace temperature of

V(V)

Cr(VI)

Co(II)

Co(III ) Cu(II)

As(III) As(V)

DMAs

020

4060

80100

120

0

200

400

600

800

1000

1200

50 55 60 65 70 75

Migration Time (s)

m/z Ratio

Cou

nts

(250

ms

int.)

Ni (II)

Figure 5.4.10. CE-ICP-TOFMS separation and multielemental detection of V(V), Cr(VI), Co(II), Co(III), Ni(II), Cu(II), As(V),As(III) and DMAs cyanide complexes. Injection contained 100 pg of each metal analyte per compound (400 pg of metal perarsenic species). Reproduced by permission of the Royal Society of Chemistry.


0

500

1000

1500

0 50 100 150

2000

56F

e+ sig

nal (

mV

)

ferrocene

time (seconds)

ferrouschloride

(a)

00 50 100 150

10

20

30

40

50

60

FeC

p+ sig

nal (

mV

)

ferrocene

time (seconds)

(b)

Figure 5.4.11. Chemical speciation of two iron compounds,ferrocene (5 µg) and ferrous chloride (250 ng), achieved withETV and with a GSGD operated in the modulated mode(10 Hz). Helium gas sampling glow discharge: atomic current40 mA, molecular current 15 mA, and discharge pressure5.10 Torr helium. ETV parameters: ash 400 ◦C (120 s); ramp300 ◦C s−1; atomize, 2200 ◦C (4 s); and transfer line flow rate90 ml min−1 helium, heated at 130 ◦C. The inorganic salt wasinitially added to the cell, and the solvent evaporated at300 ◦C, then the organic matrix was added after the cell hadcooled. (a) Atomic mode; signal collection at m/z 56, 56Fe+.Response observed for both organometallic and inorganiccomponents. (b) Molecular ionization mode; fragmentationsignal collected at m/z 120, FeCp+. The ferrocene sublimesat a temperature significantly lower than the appearancetemperature of the inorganic salt. Reprinted from Spectrochim.Acta, Part B, Vol. 55B, J. P. Guzowski, J. A. C. Broekaert,and G. M. Hieftje, pp. 1295–1314, Copyright (2000), withpermission from Elsevier Science.

400 ◦C readily volatilized ferrocene. Elevation ofthe furnace temperature to 2200 ◦C caused therelease of iron chloride from the furnace into theplasma. Measurement of atomic and molecular

information provided the unique identification ofboth analytes without the need for standards.

7 MODULATED IONIZATIONSOURCE TOFMS SPECIATIONANALYSIS

As was mentioned previously, ionization sourcescan be operated under different conditions to pro-vide distinct analyte mass spectra. ‘Hard’ ioniza-tion conditions generate atomic information, while‘softer’ conditions result in molecule fragmenta-tion or formation of the molecular ion. Operatedindependently or in combination with a hyphenatedseparation technique, the combination of atomicand molecular information allows the unique iden-tification of analytes.

7.1 Two-state modulated systems

Guzowski et al. [53] have demonstrated that agas sampling glow discharge can be operated inat least two distinct ionization modes. Use ofhelium as a discharge gas at pressures between1 and 10 Torr, currents between 10 and 130 mA,and potentials between 200 and 4000 V producedrelatively clean atomic mass spectra. With a lowerpressure (0.1–3.0 Torr) argon discharge operated atcurrents of 3–20 mA and potentials of 30–300 V,a more diffuse plasma was formed. This reduced-pressure discharge acts as a softer ionization sourceto produce molecular fragmentation similar tothat in conventional 70 eV electron impact (EI)systems. In the atomic mode, the GSGD providedexcellent sensitivity with limits of detection forseveral halogenated hydrocarbons in the tens ofpicograms per second range. Operation in themolecular mode made possible the use of EIlibraries for the unique identification of analytesbased on their distinctive fragmentation patterns.

In later studies by the same group [54] it wasdiscovered that the different ionization modes ofthe GSGD could be exploited with a single dis-charge gas at a constant pressure. This simpleexperimental alteration greatly enhanced the attrac-tiveness of the GSGD. With a single discharge gas(helium) at a pressure of 5.5 Torr, a change in the

MODULATED IONIZATION SOURCE TOFMS SPECIATION ANALYSIS 329

current–voltage characteristics of the plasma couldgreatly affect the ionization properties. Operationat a current of 25 mA and a potential of −350 Vgenerated atomic ions. With a current of 15 mAand a potential of +210 V, the surface area of thedischarge was increased by greater than a factor of10, to produce a diffuse plasma with soft ioniza-tion capabilities. Because current and voltage werethe only two experimental factors that dictatedthe GSGD ionization properties, simple electron-ics could be used to rapidly switch the source fromatomic to molecular mode at rates up to 100 Hz.Figure 5.4.12 shows the atomic and molecular

80 100 120 140 160 180 200

m/z

(a)

2000

1500

1000

500

0

sign

al (

mV

)

0

100

200

300

400

sign

al (

mV

)

200180160140

m/z

(b)

12010080

Figure 5.4.12. Atomic and molecular mass spectra for bro-moform vapor swept into the GSGD while the plasma wasmodulated at 10 Hz. Data were collected on a digital oscil-loscope (1000 transients averaged). (a) Atomic ionizationmode; discharge conditions, 5.50 Torr helium, 30 mA, 350 V.(b) Molecular fragmentation mass spectrum; discharge condi-tions, 5.50 Torr helium, 20 mA, 250 V. Reproduced by permis-sion of the Royal Society of Chemistry.

mass spectra of bromoform generated by theswitched GSGD. In the atomic spectrum the twoisotopes of bromine are clearly visible. The molec-ular spectrum exhibits good agreement with thespectrum from a 70 eV EI source. Notably absentfrom the GSGD fragmentation spectrum is themolecular ion, suggesting that the GD is slightlymore energetic than a conventional EI source.

Combined with gas chromatography, theswitched GSGD provides high sensitivity andstructural identification of transient analyte peaks.Figure 5.4.13 demonstrates how the molecular

0

0.5 1.0

time (min)

(a)

1.5 2.0

10

20

30

40

C2H

3+ sig

nal (

mV

)

chloropentane

chlorohexane

chloroheptane

o-dichlorobenzene

solvent

0

2

4

6

8

10

12

14

0.5 1.0 1.5 2.0

time (min)

(b)

C6H

3+ sig

nal (

mV

)

o-dichlorobenzene

solvent

Figure 5.4.13. Chromatogram of chlorinated hydrocarbonswith the ionization source operated in the static molecu-lar-fragmentation mode with boxcar averagers used for datacollection. Plasma conditions: pressure, 5.50 Torr; current,20 mA 250 V. (a) m/z 27 (C2H3

+) signal; a common fragmentfor these analytes. (b) m/z 75 (C6H3

+) signal illustrating theselectivity of this method. Reproduced with permission fromJ. P. Guzowski, Jr., and G. M. Hieftje, Anal. Chem., 72, 3812.Copyright (2000) American Chemical Society.


mode of the GSGD complements the elementaldetection of the atomic ionization conditions bymonitoring fragments unique to specific analytessuch as C6H3

+ from o-dichlorobenzene [42]. Todetect the modulated signals generated by theGSGD, particularly when coupled with GC, thefast spectral generation rate (15 000 complete massspectra per second) of TOFMS was essential.

7.2 Single-state modulated systems

The pulsed glow discharge has been suggestedas a source nearly ideally suited for use withTOFMS [55–58]. Higher instantaneous powerprovides greater sensitivity than in DC dischargeswithout overheating the source cathode. Addition-ally, the ionized discharge gas that is extractedfrom the source can be temporally discriminatedagainst to improve signal to noise.

Steiner et al. [59] exploited the time-dependentenergy properties of a millisecond pulsed glowdischarge to produce a source capable of pro-viding both elemental and molecular information(Figure 5.4.14). Immediately upon initiation of theGD pulse, the source potential accelerates elec-trons toward the cathode. These fast electronsstrike gas atoms (argon) to produce a high-energydischarge capable of electron-impact like ioniza-tion and molecular fragmentation. After approxi-mately 0.5 ms the discharge stabilizes and assumes

−5 0 5 10 15

PrepeakElectron Ionization

0.0−0.5 ms

AfterpeakPenning Ionization

5.5−7.0 ms

Glow DischargePower Pulse

Time (ms)

Figure 5.4.14. Depiction of glow discharge pulse sequenceillustrating the location and duration of each analytical temporalregion (prepeak and afterpeak). Reproduced by permission ofthe Royal Society of Chemistry.

pseudo-steady-state conditions during which bothdischarge gas and cathode vapor are ionized.After termination of the discharge pulse, elec-trons recombine with ionized argon to producelong-lived metastable atoms. These metastablesgenerate significantly less fragmentation and pro-duce molecular ions through Penning ioniza-tion pathways.

Majidi and coworkers [60] later extended theuse of modulated sources for the generation ofconcurrent elemental and molecular informationto microsecond pulsed discharges. Comparisonof Figures 5.4.15 (a) and (b) shows that 80 µsafter plasma ignition, the fragmentation patternproduced for p-xylene is in fairly good agreementwith that from a 75 eV EI spectrum. The massspectrum produced 470 µs after plasma ignition isdominated by the molecular ion. The authors notedthat although identical fragmentation patterns areexpected for other isomers of xylene, use ofa separation technique prior to the GD wouldprovide additional information needed for theunique identification of each analyte.

8 CONCLUSIONS

Plasma source time-of-flight mass spectrometryis a relatively new addition to the arsenal oftechniques available to analytical chemists. Atpresent, several figures of merit of PS-TOFMS,most notably sensitivity, lag behind those of moreconventional mass-analyzer configurations. Con-tinuing improvement in vacuum interface geom-etry and space-focusing capability should causethis gap to decrease. Although the mass resolvingpower of most PS-TOFMS systems is currently notsufficient to resolve many of the common isobarsassociated with elemental analysis, recent devel-opment in the field of ion guide collision cellspromises to push resolving power above 3000in the near future by reducing the energy distri-bution of sampled ions through collisions withbuffer gas molecules [28, 61]. Additionally, ionchemistry in collision cells with reactive gasesprovides a means to selectively eliminate specificisobar interferences.

CONCLUSIONS 331

0

2

4

6

8

10

20 30 40 50 60 70 80 90 100 110 120

Mass (amu)

p-Xylene reference

(a)

0.010

0.008

0.006

0.004

0.002

0.000

Inte

nsity

(V

)

80 µs

10 20 30 40 50 60 70 80 90 100 110 1200

Mass (amu)

(b)

0 10 20 30 40 50 60

(c)

70 80 90 100

470 µs

110 120

Inte

nsity

(V

)

0.010

0.008

0.006

0.004

0.002

0.000

Mass (amu)

Figure 5.4.15. Mass spectra of p-xylene: (a) NIST EI reference spectrum obtained at 75 eV; (b) µs-pulsed GD profile obtained80 µs after the plasma ignition; (c) µs-pulsed GD profile obtained 470 µs after the plasma ignition. Reproduced by permissionof the Royal Society of Chemistry.

The greatest strengths of TOFMS that makeit extremely attractive for the elemental anal-ysis of complex systems are the simultaneousextraction of all m/z and high spectral gener-ation rate. The combination of these two traits

allows PS-TOFMS to monitor fast transient signalswithout loss of temporal fidelity or corruptionof quantitative data by spectral skew. Coupledwith a variety of ionization sources including theICP, MPT and GD, TOFMS has proven to be


a capable system for the detection of hyphen-ated separations. As the speed of chromatographicand electrophoretic methods increases, the use ofTOFMS in speciation analysis will likely experi-ence a dramatic rise. Additionally, the develop-ment of rapidly modulated ionization sources willdictate the use of detectors such as TOFMS thatcan monitor the large quantity of spectroscopicinformation generated.

9 ACKNOWLEDGEMENTS

This research was supported in part by the USDepartment of Energy through grant DOE DE-FG02-98ER14 890, by the LECO Corporation, andby ICI Technologies.

10 REFERENCES

1. Holland, J. F., Enke, C. G., Allison, J., Stults, J. T., Pick-ston, J. D., Newcome, B. and Watson, J. T., Anal. Chem.,55, 997 (1983).

2. Duckworth, D. C. and Barshick, C. M., Anal. Chem., 70,709 (1998).

3. Halliday, A. N., Christensen, J. N., Lee, D.-C., Hall,C. M., Luo, X. and Rehkamper, M., Multiple-collectorinductively coupled plasma mass spectrometry, in Inor-ganic Mass Spectrometry , Barshick, C. M., Duckworth,D. C. and Smith, D. H. (Eds), Marcel Dekker, New York,2000, pp. 291.

4. Solyom, D. A., Burgoyne, T. W. and Hieftje, G. M., J.Anal. At. Spectrom., 14, 1101 (1999).

5. Cameron, A. E. and Eggers, D. F., Rev. Sci. Instrum., 19,605 (1948).

6. Mahoney, P. P., Ray, S. J. and Hieftje, G. M., Appl.Spectrosc., 51, 16 (1997).

7. Myers, D. P., Ray, S. J. and Hieftje, G. M., Inorganictime-of-flight mass spectrometry, in Inorganic MassSpectrometry , Barshick, C. M., Duckworth, D. C. andSmith, D. H. (Eds), Marcel Dekker, New York, 2000,pp. 447.

8. Guilhaus, M., Spectrochim. Acta, Part B , 55, 1511 (2000).9. Wiley, W. C. and McLaren, I. H., Rev. Sci. Instrum., 26,

1150 (1955).10. Piseri, P., Iannotta, S. and Milani, P., Int. J. Mass Spec-

trom. Ion Processes , 153, 23 (1996).11. Flory, C. A., Taber, R. C. and Yefchak, G. E., Int. J.

Mass Spectrom. and Ion Proc., 152, 169 (1996).12. Cotter, R. J., Time-of-Flight Mass Spectrometry. Instru-

mentation and Applications in Biological Research , ACSProfessional Reference Books, Washington DC, 1997.

13. Boesl, U., Weinkauf, R. and Schlag, E. W., Int. J. MassSpectrom. Ion Processes , 112, 121 (1992).

14. Sanzone, G., Rev. Sci. Instrum., 41, 741 (1970).15. Niu, H. and Houk, R. S., Spectrochim. Acta, Part B , 51,

779 (1996).16. Karataev, V. I., Mamyrin, B. A. and Shmikk, D. V., Sov.

Phys.-Tech. Phys., 16, 1177 (1972).17. Mamyrin, B. A., Karataev, V. I., Shmikk, V. and Zag-

ulin, V. A., Sov. Phys. JETP , 37, 45 (1973).18. Myers, D. P. and Hieftje, G. M., Microchem. J., 48, 259

(1993).19. Myers, D. P., Li, G., Yang, P. and Hieftje, G. M., J. Am.

Soc. Mass Spectrom., 5, 1008 (1994).20. Myers, D. P., Li, G., Mahoney, P. P. and Hieftje, G. M.,

J. Am. Soc. Mass Spectrom., 6, 400 (1995).21. Myers, D. P., Li, G., Mahoney, P. P. and Hieftje, G. M.,

J. Am. Soc. Mass Spectrom., 6, 411 (1995).22. Guilhaus, M., J. Am. Soc. Mass Spectrom., 5, 588 (1994).23. Guilhaus, M., Selby, D. and Mlynski, V., Mass Spectrom.

Rev., 19, 65 (2000).24. Mahoney, P. P., Ray, S. J., Hieftje, G. M. and Li, G., J.

Am. Soc. Mass Spectrom., 8, 125 (1997).25. Vlasak, P. R., Beussman, D. J., Davenport, M. R. and

Enke, C. G., Rev. Sci. Instrum., 67, 68 (1996).26. Sturgeon, R. E., Lam, J. W. H. and Saint, A., J. Anal. At.

Spectrom., 15, 607 (2000).27. Tian, X., Emteborg, H. and Adams, F. C., J. Anal. At.

Spectrom., 14, 1807 (1999).28. Leach, A. M. and Hieftje, G. M., Int. J. Mass Spectrom ,

212, 49 (2001).29. Ray, S. J. and Hieftje, G. M., J. Am. Soc. Mass Spectrom ,

submitted (2003).30. Furuta, N., J. Anal. At. Spectrom., 6, 199 (1991).31. Vanhaecke, F., Moens, L., Dams, R., Allen, L. and Geor-

gitis, S., Anal. Chem., 71, 3297 (1999).32. Emteborg, H., Tian, X., Ostermann, M., Berglund, M.

and Adams, F. C., J. Anal. At. Spectrom., 15, 239 (2000).33. Heumann, K. G., Gallus, S. M. and Vogl, J., J. Anal. At.

Spectrom., 13, 1001 (1998).34. Montaser, A., Inductively Coupled Plasma Mass Spec-

trometry , Wiley-VCH, New York, 1998.35. Jakubowski, N., Moens, L. and Vanhaecke, F., Spec-

trochim. Acta, Part B , 53B, 1739 (1998).36. Sutton, K., Sutton, R. M. C. and Caruso, J. A., J. Chro-

matogr. A, 789, 85 (1997).37. Lobinski, R., Appl. Spectrosc., 51, 260 (1997).38. Lobinski, R., Pereiro, I. R., Chassaigne, H., Wasik, A.

and Szpunar, J., J. Anal. At. Spectrom., 13 (1998).39. Lobinski, R. and Szpunar, J., Anal. Chim. Acta, 400, 321

(1999).40. Leach, A. M., Heisterkamp, M., Adams, F. C. and

Hieftje, G. M., J. Anal. At. Spectrom., 15, 151 (2000).41. Pack, B. W., Broekaert, J. A. C., Guzowski, J. P.,

Poehlman, J. and Hieftje, G. M., Anal. Chem., 70, 3957(1998).

42. Guzowski, J. P., Jr. and Hieftje, G. M., Anal. Chem., 72,3812 (2000).

REFERENCES 333

43. Pereiro, I. R., Schmitt, V. O. and Lobinski, R., Anal.Chem., 69, 4799 (1997).

44. Pereiro, I. R., Wasik, A. and Lobinski, R., J. Chro-matogr., 795, 359 (1998).

45. van Deursen, M. M., Beens, J., Leclercq, P. A. andCramers, C. A., J. Chromatogr. A, 878, 205 (2000).

46. Leclercq, P. A. and Cramers, C. A., Mass Spectrom. Rev.,17, 37 (1998).

47. Ferrarello, C. N., Bayon, M. M., de la Campa, R. F. andSanz-Medel, A., J. Anal. At. Spectrom., 15, 1558 (2000).

48. Wu, N., Collins, D., Lippert, J. A., Xiang, Y. and Lee,M. L., J. Microcolumn Separations, 12, 462 (2000).

49. Costa-Fernandez, J. M., Bings, N. H., Leach, A. M. andHieftje, G. M., J. Anal. At. Spectrom., 15, 1063 (2000).

50. Kutter, J. P., Trends Anal. Chem., 19, 352 (2000).51. Mahoney, P. P., Ray, S. J., Li, G. and Hieftje, G. M.,

Anal. Chem., 71, 1378 (1999).52. Guzowski, J. P., Broekaert, J. A. C. and Hieftje, G. M.,

Spectrochim. Acta, Part B , 55B, 1295 (2000).

53. Guzowski, J. P., Broekaert, J. A. C., Ray, S. J. andHieftje, G. M., J. Anal. At. Spectrom., 14, 1121 (1999).

54. Guzowski, J. P., Jr. and Hieftje, G. M., J. Anal. At.Spectrom., 15, 27 (2000).

55. Hang, W., Yang, P., Wang, X., Yang, C., Su, Y. andHuang, B., Rapid Commun. Mass Spectrom., 8, 590(1994).

56. Harrison, W. W. and Hang, W., J. Anal. At. Spectrom.,11, 835 (1996).

57. Hang, W., Baker, C., Smith, B. W., Winefordner, J. D.and Harrison, W. W., J. Anal. At. Spectrom., 12, 143(1997).

58. Harrison, W. W., Hang, W., Yan, X., Ingeneri, K. andSchilling, C., J. Anal. At. Spectrom., 12, 891 (1997).

59. Steiner, R. E., Lewis, C. L. and Majidi, V., J. Anal. At.Spectrom., 14, 1537 (1999).

60. Majidi, V., Moser, M., Lewis, C., Hang, W. and King,F. L., J. Anal. At. Spectrom., 15, 19 (2000).

61. Guzowski, J. P. and Hieftje, G. M., J. Anal. At. Spec-trom., 16, 781 (2001).

5.5 Glow Discharge Plasmas as Tunable Sourcesfor Elemental Speciation

R. Kenneth MarcusClemson University, Clemson, SC, USA

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 3342 Glow Discharges as Speciation Detectors 3353 Gas Sampling Glow Discharges . . . . . . . . 336

3.1 Optical emission detection . . . . . . . . 3363.2 Mass spectrometric detection . . . . . . 340

4 Liquid Sampling Glow Discharges . . . . . . 3484.1 Optical emission detection . . . . . . . . 3494.2 Mass spectrometric detection . . . . . . 352

5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . 3536 References . . . . . . . . . . . . . . . . . . . . . . . . 354

1 INTRODUCTION

The very essence of the elemental speciationexperiment is the identification and quantificationof the various chemical entities present in a givenanalytical specimen. In the vast majority of thesedeterminations, the first step in the process is someform of chemical separation used to isolate thecomponents in chemical space. This separationcan occur as a selected complexation, immobiliza-tion, precipitation, or some form of liquid, gas, orsupercritical fluid chromatography. More recently,capillary electrophoresis and field flow fraction-ation have been employed for separation. Oncethe chemical species have been separated, the sec-ond step of the speciation experiment involves thedetection and quantification of the target centralmetal ion. Most often, either inductively coupledplasma optical (atomic) emission and mass spec-trometries (ICP-OES/MS) or flame atomic absorp-tion (FAA) is used as the detector element. Notethat in the latter case, the identity of the analytemust be known a priori.

The typical speciation experiment involves liq-uid or gas chromatography followed by atomic

spectroscopy. Chromatography alone cannot iden-tify specific chemical species. Identities can onlybe inferred by correlating analytical retention timesto standard mixtures, if direct overlaps in compo-sition exist. While ICP-OES and MS are excellentmeans of making trace metal determinations, theessence of the methods requires that any analyte-containing entities be in the free atom state. Itmust be kept in mind that there are cases wherespeciation does not involve metal ions explicitly,so capabilities in elemental detection must extendto nonmetals including ‘gaseous’ elements and thehalides. In any case, if only elemental analysis ispossible, then there is no direct evidence of theinitial analyte’s identity generated in either the sep-aration or the detection process. Thus, the typicalspeciation experiment involves qualitative analysisby inference, based on relative chemical reactivityor chromatographic retention times for chemicalentities containing the monitored element. Clearly,the most effective way to perform elemental spe-ciation is to employ detection means that providedirect evidence of the chemical form of each con-stituent. It is possible that chemical separation willnever be circumvented as an integral part of the

GLOW DISCHARGES AS SPECIATION DETECTORS 335

experiment, but there will definitely always be aneed for chemically specific means of detection.

Beyond the ability for the aforementionedatomic spectroscopic methods to provide element-specific information, chemical speciation in theareas of biological or environmental analysisrequires far greater amounts of information. Thereare two more relevant levels of specificity thatcan be related to the chemical identity of chro-matographically separated species. The first levelof complexity is the empirical formulae of com-pounds, basically the relative number of atomsof each element in the eluting molecule. Forexample, the empirical formula for tetraethyl lead(Pb(CH2CH3)4 is PbC8H20. A determination of theempirical formula for this compound is not specificto this particular molecule though.

The greatest level of specificity comes frommethods that allow the unambiguous identificationof a complete analyte entity. For complete determi-nations, there are few spectroscopic methods thatallow unambiguous identification of any molecule.Mass spectrometry is one such method, but onlywhen aspects of the complete molecular entity(i.e., molecular weight) and selective informa-tion related to molecular structure (i.e., fragmenta-tion patterns) exist. When both of these aspectsare used in gaining speciation information viamass spectrometry such methods outweigh con-ventional atomic spectrometry sources that, by def-inition, decompose samples down to their atomicform to report elemental information. Scientificallydesirable methods are those that provide elementspecific information, which is complemented byrelative atomic ratios as well as complete molecu-lar weight signatures. Thus the detector elementmust inherently be capable of delivering differ-ent amounts of energy to the analyte to produceatomic, fragment, and molecular signals. Conven-tional spectrochemical sources operating at atmo-spheric pressure are not capable of providing thisrange of information for a number of reasons.Alternatively, low pressure glow discharge (GD)sources do indeed have such capabilities. GDsources, operating at pressures of ∼1 Torr in raregas environments and at powers of less than 60 Whave proven the basic capabilities of producing

optical emission spectra which are reflective of ananalyte’s empirical formula and mass spectra thatare composed of the desired information.

Described here are the design aspects andanalytical characteristics of GD sources employedin elemental speciation studies. Sources designedfor receiving either gaseous or liquid samples willbe presented. In addition, sampling in either theoptical emission or mass spectrometry modes willbe demonstrated for both sample forms. Whilethere are no commercially available instrumentsin which glow discharge sources are employed toexplicitly monitor liquid or gaseous streams, it isbelieved that work of the sort described here willgo far toward bringing these capabilities to wideracceptance [1].

2 GLOW DISCHARGESAS SPECIATION DETECTORS

GD sources offer a number of interesting pos-sibilities as speciation detectors for gaseous andliquid sample analysis. Many of the key points aredescribed below with reference to how they are dif-ferent from spectrochemical sources that are morewidely employed in this area [2–4].

• Inert gas environment – Because they operate atreduced pressures and in inert environments, theplasma is not inherently contaminated by atmo-spheric species such as water, oxygen, hydro-carbons, and fragments thereof. As a result, GDsources can be employed for elemental anal-ysis across the entire periodic table (save thedischarge gas), which is paramount in the abil-ity to perform empirical formula determinationsof organometallics by optical emission or massspectrometry. In terms of mass spectrometricanalysis, the absence of background gases elim-inates possibilities of deleterious side reactions.

• Low temperature environment – Specifically forthe case where mass spectrometry is the desiredsampling mode, the gas-phase temperatures anddensities must be sufficiently low so as notto induce dissociation in molecular species.Thermal excitation of vibrational modes inanalyte molecules can serve either to dissociate

336 GLOW DISCHARGE PLASMAS AS TUNABLE SOURCES

species directly or to make them more prone tofragment in the course of subsequent ionizationevents. The kinetic temperatures of most GDsources lie in the range of 100–500 K, asopposed to ∼2500 K in combustion flames and∼10 000 K in ICP sources. Of course, hightemperatures are desirable in cases where onlyelemental analysis is required and are the defacto goal in all optical emission experiments.

• Low power operation – While it is a naturalassumption that low operation powers will yieldlow gas phase temperatures, the desire for lowpower operation has another role in terms ofdesigning speciation detectors. In all forms ofchromatography there is a desire to match thesize, cost, and experimental complexity of thedetector element with that of the separationstage. GD devices operating at average powersof less than 60 W (i.e., less than a light bulb)can be constructed to fit in the palm of yourhand. Therefore, these are sources that canbe coupled to chromatographs rather than achromatograph coupled to the source. This maybe a matter of semantics, but is very importantwhen one wishes to have technology adapted bychromatographers, environmental chemists, andnutritionists.

• Analytical versatility – Given that there aremany forms of chromatographic separation andsample introduction, one must consider therange of sample forms that a potential speciationdetector can receive. In regards to GD sources,this is an issue of sample introduction asopposed to the spectrochemical source per se.Glow discharges readily accept vapor-phasesample forms; thus gas chromatography andvapor generation are viable means of speciesseparation. On the other hand, direct liquidinjection is not a viable option. As will bediscussed in subsequent sections of this chapter,liquid introduction can be accomplished in astraightforward manner. In the end, a single GDsource can be designed to accept eluents fromeither gaseous or liquid streams.

In the following sections, a number of analyt-ical applications of glow discharge devices in thefield of elemental speciation will be described. The

discussion will be restricted to GD experimentsthat yield higher levels of chemical informationthan simple determinations of metal ion content inflowing streams. In the case of optical emissionsampling, the nonmetal elements provide infor-mation toward elucidating empirical formulae. Inmass spectrometric analysis, the ability to pro-duce information on both molecular and elementalconstituents must be realized. Based on examplesappearing in the literature to date, GD plasmasshould play a very prominent role in the rapidlyexpanding area of chemical speciation.

3 GAS SAMPLING GLOWDISCHARGES

All GD sources have in their very nature anaspect of operation with a flowing gas stream.Commercial GD-OES and GDMS systems relyon a stream of argon (typically) passing throughthe source as a means of providing the dischargegas. The argon flow also serves to flush debrisfrom the system and, in the case of MS analysis,to carry analyte ions from the source volumeinto the mass analyzer stage. In the case ofsolid specimen analysis, argon is the dischargegas of choice because of its high sputteringefficiency combined with high-lying metastablelevels. Metastable levels of approximately 11.5 eVplay an important role in the excitation andionization characteristics in the discharge negativeglow region. In the case of gaseous sampleintroduction, sputtering is not required; on thecontrary it is detrimental. Therefore, the dischargegas can be chosen solely on the basis of achievingthe desired excitation/ionization characteristics,with helium providing the most energetic of GDplasmas. In general, gaseous sample introduction isachieved by simply transporting the analyte vaporto the source in a flow of He gas, as is the case inmost gas chromatography separations.

3.1 Optical emission detection

The underlying principles for the use of GD-OESin chemical speciation were presented before the

GAS SAMPLING GLOW DISCHARGES 337

term had actually become common. The potentialcapabilities of the use of GD sources as detectorsfor elemental speciation were first demonstratedin 1989 by Puig and Sacks who used a heatedinjector to introduce hydrocarbon vapors into ahollow cathode discharge (HCD) and monitoredthe optical emission of the halides F, Cl, andBr [5]. The discharge source operated at pressuresof 20–40 Torr He and at currents of up to 100 mA.Transient signals resulting from the introductionof 250 µL samples suggested that chromatographicintegrity could be retained while yielding detectionlimits in the range of 0.6–17 ng level for thehalides introduced as small halocarbon molecules.Similarly, Winefordner and coworkers extendedthe analyte set to include I and S [6]. Thoseresearchers suggested future applications in gaschromatography as well.

The initial studies explicitly pointing to the useof GD-OES as a speciation tool (i.e., the abil-ity to generate empirical formulae) were presentedin a series of papers by Hieftje and cowork-ers [7–10]. That early work followed the conceptof McLuckey and coworkers who developed anatmospheric pressure sampling glow discharge fororganic mass spectrometry [11]. The gas samplingglow discharge (GSGD) incorporated a metal-lic capillary to deliver volatilized organic com-pounds through the cathode region of a planarGD source [7]. Those studies showed that smallmolecules could be effectively dissociated suchthat empirical formulas of halogen-versus-carboncontent could be derived in a straightforward fash-ion. Detection limits on the 5 ng s−1 were achievedfor chlorine, with subsequent studies also illustrat-ing the use of hydride generation for determina-tions of As in solution [8].

A detailed study of the effects that organic vaporintroduction had on the discharge operating char-acteristics for a Grimm-type source geometry waspresented by Hieftje et al. [10]. It was found thatthe discharge voltage increased upon injection ofthe organic compounds, which were manifest inchanges in the spectral background. This points tothe importance of obtaining background-correctedsignals on the chromatographic time scale. Threedifferent introduction geometries were explored

regarding the exit of the silica capillary: (1) flushwith the face of the plane cathode, (2) a conicaldepression in the cathode surface, and (3) a 2 mmdeep, 1 mm diameter hollow cathode. Spatially-resolved optical measurements clearly indicate thatthe plasma energetics are localized within the cav-ities of geometries (2) and (3), with a very diffusedistribution observed in the plane cathode case.These studies also considered the temporal stabil-ity and noise structure of the plasma. The Cl(II)emission from dichloromethane introduced contin-uously over a 30 min period showed a variationof only 0.9 % RSD. The associated analyte noisepower spectrum was ‘white’ up to a frequencyof 800 Hz. The detection limit for chlorine intro-duced as dichloromethane was determined to be20 ng s−1, higher than the previous studies whichused more open geometries [7]. The degradation indetection limit was attributed to the higher spectralbackground present in the restricted Grimm-sourcegeometry. Linear response curves were demon-strated for the C/Cl content for a range of halo-carbons, with acceptable responses noted as wellfor the cases of C/H and H/Cl ratios.

In the area of source development, Sanz-Medel and coworkers have published an excellentseries of papers on the use of glow dischargedevices as optical emission detectors for gaseoussample introduction [12–17]. Their work differedfrom that of the previously cited authors in thatradio frequency (RF) powered plasma sourceswere employed. In the case of optical emis-sion detection, RF powering affords a more ener-getic plasma in terms of electron energies anddensities [18]. They have described two basicdesigns for gaseous sample introduction, which aredepicted in Figure 5.5.1. Early studies involveda modification of the so-called ‘Marcus’ geom-etry normally used for solids analysis [12]. Inthis approach (Figure 5.5.1(a)), the gaseous sam-ple was introduced through the limiting (anode)orifice (shown in greater detail on the right-handside of the figure), tangential to the cathode sur-face. In this way, the eluent was delivered directlyinto the plasma negative glow region. Differentintroduction points for the plasma make-up gashave also been investigated. The make-up gas is


4.0 f

32.5 f

47.5 f

1.0 f8.0

20.04.0

Gas entry

Groove for the o-ring

(a)

(b)

“i.d.”

Gas entry

Circulatingwater

toPressure

gauge

Vacuumexit

Limiting disc“Cathode”

Macor

Glass

Coaxial Cable

Brass torquebolt

Cathode cavity (1mm i.d.)

40 mm

He inlet

Pipes for water 5−10 ºCcirculation

Capillary from GC

Negativeglow

Figure 5.5.1. Basic discharge geometries employed by Sanz-Medel and coworkers for RF-GD-OES analysis of gaseous samples.(a) Introduction through limiting orifice of Marcus-type source geometry. Reproduced from reference 12 with permission fromthe American Chemical Society. (b) Hollow cathode geometry. Reproduced from reference 15 with permission from the RoyalSociety of Chemistry.


required because the He flow rate from the sampleintroduction is not sufficiently high to maintaina stable plasma. Even the earliest studies pro-duced detection limits for nonmetals (C, Cl, Br,and S), introduced in the form of organic vaporsthrough an exponential dilution flask, that weresuperior to those of more widely used meth-ods. Further demonstration of the methodologyinvolved more thorough evaluations of the plasmaoperation parameters as well as alternative meth-ods of sample introduction.

Figure 5.5.2 illustrates the versatility of the RF-GD-OES sampling method through the three basicforms of sample introduction employed by theOviedo group; exponential dilution [12], volatilevapor generation (either cold vapor, hydride, oroxidative) [13, 14], and gas chromatography [15,16]. In one application, volatile chlorine (Cl2) was

oxidatively generated from chloride in solutionand detected via Cl(I) emission in the NIR regionof the spectrum (837.6 nm) [13]. Both RF anddirect current (DC) modes of plasma operationwere investigated. Detection limits for the RFmode (0.1 ng mL−1) were found to be superior tothose for the DC, and better than any previouslyreported values, without sacrifice in other figuresof merit. In addition, empirical formulas fora range of chloro-hydrocarbons and a pair ofsulfur compounds were accurately determined. Gaschromatography separations were employed formercury determinations in fish tissue extracts [15].Detection limits for methyl-, ethyl-, and inorganicmercury ranged from 1–3 ng mL−1 when the Hg(I)253.6 nm transition was monitored. Unfortunately,only chromatographic retention time was used toidentify species, rather than the use of nonmetals to

He

GD

ClampKMnO4

(H2SO4 conc.)

SampleH2SO4 (40%)

WasteGas-liquidseparator

H2SO4

H2O

electric heating tapemagneticstirrerexponential

dilutor

restrictorto hood

He

Rf GD

Rf-GD-OES

He

He

HeFLOW

METER

Liquid waste

SnCl2 solution

HCI 1 MOxidantsolution

HPLC SystemFocused microwave

digestorH2SO4

H2SO4

(a) (b)

(c)

Figure 5.5.2. Examples of different modes of gaseous sample generation for introduction to an RF-GD-OES source. (a) Oxidationof chloride to chlorine [13]. Reproduced with permission from the Royal Society of Chemistry. (b) Exponential dilution of organicvapors [12]. (c) HPLC separation followed by microwave oxidation to generate Hg vapor [16]. Reproduced with permission fromMaik Nauka, Russia.


determine empirical formula. A different approachto introducing the same materials involved the useof liquid chromatography to separate the organicand inorganic mercury species, followed by in-line microwave digestion, and finally cold vaporgeneration of Hg0 that was introduced into theplasma [17]. Liquid chromatography separated thevarious Hg species in time, prior to more or lessgeneric vapor generation. Both NaBH4 and SnCl2were evaluated as the reducing agents to producethe mercury vapor, with realized detection limitsof 1.8 and 0.2 ng mL−1, respectively. The samplepreparation and separation steps were viewed bythe authors to be more useful and rapid than theaforementioned GC methodology. In comparisonwith more common ICP methods, the GD-OEStechnique ‘. . . has favorable detection limits, lowerinstrumentation costs, and very low plasma gasconsumption’ [16].

The group of Sanz-Medel has also looked atdifferent source geometries for the use of GD-OES for gas specimen analysis. It has long beenknown that hollow cathode geometries yield plas-mas that are of higher density and greater energythan those employing planar (flat) cathode geome-tries [19]. Figure 5.5.1(b) depicts the RF-HC-GD-OES source developed in that laboratory [15].Different from the planar source arrangement, thegas eluent and the entirety of the He dischargeare introduced in the base of the plasma. Arsenicand antimony hydride generation was used as thesample introduction platform for the comparisonstudies [14]. While the DC powered flat cathodegeometry produced lower limits of detection thanthe other possible modes and geometries, it wasdetermined that the RF-HC combination was moreanalytically useful based on low detection limits,greater sensitivity for low sample volumes, andgreater temporal fidelity of signal transients. TheRF-HC-GD-OES source was used as a GC detectorfor organic and inorganic mercury species [15]. Inthis instance, the species in solution were isolatedby first performing a Grignard reaction to formthe mercury chlorides, followed by extraction in amixture of diethyldithiocarbamate in toluene. Sim-ilar to many plasma sources used for GC detection,the authors point out that the eluting solvent vapor

can extinguish the plasma, and so the dischargewas not initiated until the solvent front had passed.Very high quality chromatograms were realizedin the separation of fish tissue reference mate-rials. Here again, the HC geometry yielded farlower detection limits (5–10×) than the flat cath-ode geometry, which in turn were lower than thosefor GC with MIP-OES detection. A recent reportfrom this group has illustrated the use of solid-phase microextraction (SPME) as a very powerfulmeans of sample preconcentration, while also elim-inating solvent effects in performing tin and leadspeciation experiment by RF-(HC)GD-OES [20].

3.2 Mass spectrometric detection

Similar to the case of GD-OES analysis ofgaseous samples, the use of GDMS for elementalspeciation applications has its roots very muchin early works by McLuckey and coworkers inthe development of the ASGD source for organicmass spectrometry [11]. Those studies illustratedwell the fact that a reduced pressure GD sourcecan produce fragmentation patterns that are verysimilar in nature to those of electron impact (EI)sources. This is an important feature as onlyEI sources are reproducible to the point of thegeneration of ‘universal’ spectral libraries. Ofcourse, in the generic case of elemental speciation,one desires the ability to observe molecular ionsas well as chemically significant fragmentationpatterns. This is the strength of EI and is in factrealized for the most part with the low pressureGD sources applied to speciation.

The first dedicated effort in the use of GDMSfor elemental speciation was described by Carusoand coworkers at the University of Cincinnati [21].That work involved designing a cubic dischargecell that was mounted in the torch position ofa commercial ICP-MS instrument. Figure 5.5.3(a)depicts the design of the RF-GDMS sourceused for elemental speciation based on a directinsertion probe (DIP) as described by Duckworthand Marcus [25]. A high-purity aluminum disk(∼6 mm diameter) served as the cathode. Thefused silica GC capillary was passed through aheated steel capillary mounted on a flange of


Sampling Cone

Al Sample GD Probe

1/16 Tube

1.4 mm Orifice

Capillary

Flange

(a)

(c)

(b)

rf

skimmer

front plate

1 mm samplingorifice

capillary

1/16” tube

cathode

1/16" tube

capillary

cathode

front plate

skimmer

Figure 5.5.3. Three configurations of RF-GDMS sources used for gas chromatography detection. (a) Open cell geometry usinga six-way cube [21]. (b) Reduced-volume source with movable probe [22]. (c) Coaxial introduction of gas eluent throughcathode [23]. Reproduced with permission from the Royal Society of Chemistry.

the cube, such that it terminated within 1 mmof the ion exit orifice. Placement of the cathodeapproximately 1 cm from the orifice ensured thatthe eluting species entered directly into the plasmanegative glow for ionization and extraction tothe mass analyzer. Those early studies clearlyshowed that the extent of fragmentation foralkyltin and lead species was dependent on thedischarge conditions of power and pressure, withgenerally very good overlap observed with EIspectral libraries. Detection limits for Sn of 1 pgwere shown to be very competitive with moreestablished methods.

Two subsequent design modifications weremade to the RF-GD source used for elemen-tal speciation. The second design by Caruso andcoworkers involved a dramatic reduction in theplasma volume as depicted in Figure 5.5.3(b) [23].Based on the structure of the RF plasma, higher

discharge pressures (∼25 Torr He) were used withthis source than the initial design (<1 Torr). Coolon-column injections could be employed read-ily with this geometry, with no need for solventventing in many instances. Comparison betweenchromatograms obtained with a flame ionizationdetector and that from the GDMS detection for amixture of alkyltin compounds showed very goodcorrelation in terms of peak shape and resolu-tion. The roles of carrier gas flow rate, dischargepressure, and cathode–anode separation distancewere evaluated for three organotin compounds,which showed that each compound responded inthe same way. This would indicate that thereshould be minimal matrix effects. The RF-GDmass spectra were shown to be very similar againto EI libraries, as illustrated in Figure 5.5.4(a)for tetramethyltin. By the same token, the massspectra of related alkyltin compounds produce


170 180 190

120

135 165

150180

12

8

4

0

Inte

nsity

/105 c

ount

s rf-GD

100

50

0

120135 150

165

110 130 150 170 190 210 230 250

Rel

ativ

e ab

unda

nce

m/z

NIST

120

135 145165

197

211

227

242

10

8

6

4

2

0

TMPT

120

149

179207

236135

10

8

6

4

2

0

TET

Inte

nsity

/105 c

ount

s

120

177

135

235

TBT3

2

1

0

m/z

(b)(a)

110 130 150 170 190 210 230 250

Figure 5.5.4. Mass spectra of TMT. (a) RF-GD ionization source (RF power = 30 W, pressure = 26 Torr He, samplingdistance = 2.5 mm) and NIST reference library [21]. (b) RF-GD mass spectra of eluting TET, TMT, and TBT [23]. Reproducedwith permission from the Royal Society of Chemistry.

very distinct, yet easily interpreted, fragmentationspectra (Figure 5.5.4(b)). A very important set ofexperiments was performed to elucidate how dif-ferences in discharge conditions (i.e., power andpressure) and sampling distance are manifest inthe degree of fragmentation for the tetramethyltincompound. Interestingly, while discharge condi-tions effect the absolute signal intensities, thereare no appreciable differences in degree of dis-sociation. On the other hand, changes in samplingdistance do yield different spectral structures. Sim-ply put, while the bare metal ion signal (120Sn+)definitely decreases as a function of distance, thereis lesser degrees of fragmentation at short distances(i.e., a greater fraction of molecular species).Table 5.5.1 lists the determined figures of meritfor the RF-GDMS analysis of alkyltin compounds.As can be seen, the figures are quite excellent,with detection limits on the subpicogram level.

The responses across the family of compounds arereally quite uniform, with the only deviation is fortetrabutyltin, the least volatile of the molecules.

One final variation in the source designs inves-tigated by the Caruso group involved somethingakin to a hollow cathode geometry, with the GCcapillary actually passing through the center ofthe DIP and terminating at the cathode surface,as shown in Figure 5.5.3(c) [24]. With this geom-etry, the eluting compounds pass coaxially throughthe RF-GD plasma. In this way, variations of thesampling distance also effect the time each ana-lyte is present in the discharge. The analyticalfigures of merit for this geometry are included aswell for tetramethyltin in Table 5.5.1. While thereare few substantive differences, one very interest-ing improvement is seen in the >3X larger slope(i.e., sensitivity) realized with the latter sourcegeometry. Different as well, increases in discharge


Table 5.5.1. Figures of merit of RF-GDMS analysis of alkyltin compounds introduced by gas chromatogra-phy [22, 23].

Parameter TMT[21] TET[21] TMPT[21] TBT[21] TMT[22]

Linear range studied (decades) 2.5 2.5 2.5 2.5 2Slope/counts s−1 pg−1 681 615 379 262 2237Correlation coefficient 0.9950 0.9874 0.9713 0.9991 0.997Log–log slope 1.011 1.126 1.187 0.9196 0.973MDAa pg−1 1.2 2.0 2.6 87 1.0Detection limit pg−1 0.6 0.6 1.2 13 0.6RSD (%) <5 % <5 % <5 % <5 % 3.5

a Minimum detectable amount (MDA) based on peak height, all others based on peak area. Reproduced with permission fromthe Royal Society of Chemistry.

pressure did indeed produce greater amounts offragmentation, enhancing the relative amounts ofatomic ions. Thus, there is indeed an amountof tunability that can be achieved in the use ofRF-GDMS for elemental speciation. In a subse-quent study principally using low pressure induc-tively coupled plasmas, the mixing of Ar and Hedischarge gases was also shown in the case ofthe RF-GD to allow a greater amount of con-trol of discharge fragmentation of organometalliccompounds [25]. This tunability arises from thefact that the two gases have appreciably differentmetastable energies.

A very enlightening series of papers has illus-trated the potential of using different regimes ofglow discharge powering to effect the productionof mass spectra that are either ‘atomic’ or ‘molecu-lar’ in nature. Continuing along the lines describedpreviously for their work with the gas samplingglow discharge (GSGD) employed for opticalemission detection [7–10], Hieftje and coworkershave coupled that source to a time-of-flight (TOF)mass spectrometer to allow greater ion through-put and simultaneous analysis across the desiredmass regions [26–30]. A 1.6 mm o.d. stainlesssteel capillary was used to introduce analyte vaporsinto the discharge, where the planar cathode wasmachined to effect a small hollow in which theplasma was struck [26]. The authors showed thatthe extent of analyte fragmentation could be con-trolled via the use of either He or Ar dischargegases, with the former yielding ‘atomic’ spectrapermitting assignments of empirical relationshipsand the latter producing ‘molecular’ spectra. Themajority of the studies involved the evaluation of

how discharge parameters affected ‘atomic’ ionsignals. Unfortunately, the responses of the signalsfor the different fragment ion species were notthe same, such that measured empirical formulaewere very dependent on discharge current and pres-sure. Detection limits for a range of chloro- andbromohydrocarbons were in the general range of40–90 pg s−1 for the molecular ions, and perhaps∼2 × lower in the case of halogen detection.

Given the experimental complications of chang-ing discharge gases during the course of a sin-gle gas chromatogram, Hiefje and coworkers havedeveloped a GSGD that produces either atomicor molecular species on-the-fly through the useof a DC, ‘switched’ discharge source [27, 28].The switching between the ‘atomic’ and ‘molec-ular’ modes was affected by operation at high(30 mA) and low (20 mA) discharge currents at a50 % duty cycle at frequencies of up to 150 Hz.As illustrated in Figure 5.5.5, the pulsing of the

Discharge Modulation(0−100 Hz)

Repeller Pulse(15 kHz)

Atomic Trigger

Molecular Trigger

Steering PlateModulation(0−100 Hz)

atomic

molecular

−2360 V

−2060 V

Figure 5.5.5. Pulse sequences employed in a ‘switched’GDMS source for the acquisition of atomic and molecular massspectra on a TOF mass analyzer [28]. Reproduced with permis-sion from the Royal Society of Chemistry.


discharge occurs at such a rate that many TOFmass spectra can be accumulated during eachhalf-cycle [28]. Different discharge configurationsand powering schemes were investigated usingbromoform introduced via an exponential dilu-tion flask at the model compound. Ultimately, aslight depression formed in the cathode used toeffect a hollow cathode geometry. Figure 5.5.6illustrates the versatility of the switched sourceoperation in the case of the GC separation of aseries of n-chlorohydrocarbons [27]. As seen inthe top two chromatograms, the respective sig-nals for the 12C and 35Cl ions correlate very well,with the Cl+ signal showing much better signal tonoise characteristics, particularly when the muchlower mole fraction of chlorine is considered. Thechromatogram extracted from the propyl group(C3H5

+) signals also showed excellent termpo-ral agreement with the atomic chromatograms,

with very good signal-to-noise as well. Calibrationcurves were shown to be linear on a log–log scaleranging from the picogram to nanogram levelswith detection limits ranging from 1 to 18 pg s−1

for halocarbons in both the atomic and molecu-lar modes, with the former generally being on thesingle pg s−1 level. Mass spectra acquired undermolecular conditions were shown to correlate withconventional EI source.

In addition to the use of direct vapor and gaschromatographic sample delivery, the Indiana Uni-versity group illustrated the use of electrothermalvaporization (ETV) as a means of introductionof inorganic and organic compounds [29]. In thisapproach, the appearance temperature of a givenanalyte provides insight into the chemical identity.For example, metals of different volatility appearin temporal order of their melting points (e.g., Zn→ Sn → Cd). Here too, the GSGD was operated

120

100

80

60

40

20

01 2 3 4 5 6 7 8

CI-C5

CI-C6CI-C7

CI-C8 CI-C9 CI-C10

12C

+ sig

nal (

mV

)

time (min)

(a)

2 3 4 5 6 7

25

20

15

10

5

1 8

CI-C5

CI-C6 CI-C7 CI-C8

CI-C9 CI-C10

35C

I+ sig

nal (

mV

)

time (min)

(b)

2 3 4 5 6 71 8

C3H

5+ sig

nal (

mV

)

time (min)

CI-C5

CI-C6 CI-C7 CI-C8CI-C9

CI-C10

20

15

10

5

25

0

(c)

Figure 5.5.6. Gas chromatography separation of a number of n-chloroalkanes employing the atomic (12C+ and 35Cl+ in A andB respectively) and molecular (C3H5

+ in C) modes of GSGD operation [27]. Reproduced with permission from the AmericanChemical Society.


50 1000 150

2000

1500

1000

500

0

56F

e+ sig

nal (

mV

)

ferrouschloride

ferrocene

(a)

time (seconds)

0 50 100 150

60

50

40

30

20

10

0

FeC

p+ sig

nal (

mV

) ferrocene

(b)

time (seconds)

Figure 5.5.7. Speciation of organic (ferrocene) and inorganic(ferrous chloride) iron species by selective volatilization anduse of a switched-source GSGD and TOF-MS to monitor(a) atomic iron at 56Fe+ and (b) the molecular analyte atFeCp+ (m/z = 120 Da) [28]. Reproduced with permissionfrom Elsevier Science.

in the switching mode to optimize ‘atomic’ signalsand also identify ‘molecular’ species. Figure 5.5.7illustrates the differentiation of organic iron (fer-rocene) from inorganic iron (ferrous chloride). Inthe top figure, the mass spectra produced in theatomic portion of the waveform are monitored for56Fe+. The transient signal plot shows two Fespecies of different volatility. On the other hand,monitoring of m/z = 120 (FeCp+) clearly indi-cates that the first peak corresponds to an ironatom with a cyclopentadiene ligand (i.e., ferroceneafter loss of one cyclopentadiene unit). The GDexperiment here clearly provides greater speciesinformation than would have been obtained in themore conventional coupling of ETV with ICP-MS.

Majidi and coworkers have taken the conceptof a switched glow discharge into a very differentrealm by the use of microsecond time scale pulsingof the plasma [31]. These experiments build upon

a number of papers in the use of pulsed-sourceGDMS in elemental analysis of solids. Harrisonand coworkers have shown that on–off pulsingof the GD produces mass spectra that representtwo different modes of ionization [32, 33]. Inthe discharge-on period, the mass spectra areindicative of the situation where electron ionizationis the dominant mechanism. In the post-pulseregime (within a few microseconds of plasmacessation), the mass spectra clearly indicate thationization occurs via a Penning-type collision withremnant metastable discharge gas atoms. It mustbe kept in mind that the EI process is energeticenough to ionize sputtered atoms and other gaseousmolecules, while the Penning process involves afixed energy dictated by the discharge gas species(i.e., ∼11.5 eV for Ar). The use of microsecondpulse lengths provides great temporal selectivityin ion sampling as the post-pulse region can besampled selectively in terms of pulsing of therepeller gate to the orthogonal TOF mass analyzer.In essence, the temporal evolution of plasmaenergetics can be selectively used to yield massspectra of different amounts of fragmentation.

Figure 5.5.8 illustrates the concept of tempo-ral sampling to achieve different levels of speciesinformation [31]. The mass spectrum of ethylben-zene acquired by pulsing the repeller 45 µs afterthe ignition of the discharge pulse (which has a20 µs width) displays very much more fragmen-tation than the reference EI spectrum. While theflight time from the plasma source to the repellergate is not provided, it is assumed that theseions reflect more the ionization conditions of theplasma-on regime. On the other hand, the spec-trum acquired from the ion population passingthe repeller 305 µs after plasma ignition indicates‘softer’ ionization conditions than standard 70 eVEI. A number of alkylated aromatic compoundsshow similar temporal qualities. Use of the tem-poral characteristics of these compounds and thecommon inorganic thermometric species tungstenhexacarbonyl (W(CO)6) provides a picture of theionization conditions as a function of delay timethat is presented in Figure 5.5.9. As can be seen,the energy begins very high, ultimately decreas-ing to the metastable level of the Ar discharge.


0

10

20

30

40

50

60

70

80

90

100

0 10 20 30 40 50 60 70 80 90 100 110 120

Mass (amu)

(a)

Rel

ativ

e A

bund

ance

(%

)Ethylbenzene reference

45 µs

0.035

0.030

0.025

0.020

0.015

0.010

0.005

0.000

0 10 20 30 40 50 60 70 80 90 100 110 120

Mass (amu)

(b)

Inte

nsity

(V

)

0.004

0.003

0.002

0.001

0.000

Mass (amu)

(c)

Inte

nsity

(V

)

0 10 20 30 40 50 60 70 80 90 100 110 120

305 µs

Figure 5.5.8. Mass spectra of ethylbenzene. (a) NIST EI reference spectrum. (b) µs-pulsed sampled 45 µs after plasma ignition(25 µs after termination). (c) µs-pulsed sampled 305 µs after pulse plasma ignition (285 µs after termination) [31]. Reproducedwith permission from the Royal Society of Chemistry.


0

5

10

15

20

25

30

0 100 200 300 400 500

Plasma Delay Time (µs)

App

aren

t Ion

izat

ion

Ene

rgy

(eV

)

Figure 5.5.9. Measured plasma ionization energy as a function of time for µs-pulsed GD based on introduction of hydrocarbonand W(CO)6 vapors [30]. Reproduced with permission from the Royal Society of Chemistry.

79Br+ 81Br+

93{79BrN}+ 97{81BrO}+

800

700

600

500

400

300

200

100

0

60 80 100 120 220200180140 160

95{79BrO, 81BrN}+

Sig

nal (

mV

)

m/z

(a)

60 80 100 120 220200180140 160

Sig

nal (

mV

)

m/z

200

150

100

50

0

91{79BrCH}+

93{81BrCH}+

171{79Br2CH}+

173{79Br81BrCH}+

175{81Br2CH}+

(b)

Figure 5.5.10. Atomic and molecular modes of the MPT-ASGD tandem source shown for bromoform introduction. (a) Atomicmode with both MPT and ASGD powered. (b) Molecular mode with ASGD powered (no MPT) [30]. Reproduced with permissionfrom Elsevier Science.


This approach to speciation offers a great deal ofexperimental flexibility, though perhaps too muchwith regards to practical implementation. Futurestudies were suggested involving the use of otherdischarge gases and the addition of chemical ion-ization agents to the discharge.

One other approach has been used in theconcept of switched sources for elemental spe-ciation of gaseous samples. Ray and Hieftjehave recently described a tandem microwaveplasma torch–atmospheric sampling glow dis-charge (MPT-ASGD) apparatus mounted on aquadrupole mass analyzer [30]. In this approach,the gaseous sample is subject to the colli-sional/ionization environment of the MPT and thenthat of the ASGD. As such, three modes of oper-ation are possible: MPT alone, ASGD alone, orthe MPT-ASGD in series. In this implementation,the GD was maintained by a positive potentialon the sampling aperture, with the ions transmit-ted to the MS effectively through the cathode thatwas actually the second-stage skimmer cone of theinstrument. This was termed reverse-biased oper-ation. In practice, use of the MPT alone providedvery little direct ionization. Figure 5.5.10 depictsthe mass spectra obtained in the two other opera-tional modes. As expected, the combined sourcesprovided exclusively atomic spectra (relative to theparent halocarbon analyte) as analyte moleculeswere subjected to the two collisionally active plas-mas. Finally, use of the ASGD by itself yieldedmass spectra that were similar in structure to EIlibraries, as would be expected given its rela-tively low current operation. The authors noted thatreverse-biased ASGD operation produced ions ofvery high kinetic energies, which can be detrimen-tal to throughput and mass resolution. This is thesame situation noted by Harrison and coworkers inearly hollow cathode GDMS, where sampling ionsthrough the cathode fall lead to many mass spectro-metric difficulties [34]. Other difficulties presentedincluded an electrical ‘communication’ betweenthe plasmas and the need to separately ignite theMPT with a Tesla coil. The latter point makesswitching on chromatographic time scales impos-sible in the current method. The authors do suggestthat there may be a variety of other ‘tandem

source’ combinations which may be effective in thearea of elemental speciation by mass spectrometry.

4 LIQUID SAMPLING GLOWDISCHARGES

As alluded to early in this chapter, a typi-cal approach to chemical speciation is the on-line detection of species separated by liquidchromatography followed by ICP-MS. This ofcourse alleviates the possibility of direct speciesidentification. Based on the work described in theprevious sections, the coupling of gaseous sampleintroduction method to low pressure GD sourcesis a relatively straightforward line of reasoning.On the other hand, the introduction of solutionphase samples into glow discharge sources is sim-ilar to the introduction of liquid samples into thelow pressure ion sources used in organic massspectrometry in terms of the detrimental effectsof solvent vapors to the plasma operation. Elec-tron impact and chemical ionization sources areincapable of affecting desolvation and the presenceof residual solvent vapors causes severe depres-sion of analyte ion signals and leads to high levelsof spectral interference [35]. By analogy, it is notsurprising that liquid chromatography/mass spec-trometry (LC/MS) interfaces, the moving belt andthe particle beam [36–39], have also been imple-mented for sample introduction into GD devices.The moving belt and particle beam devices fall intothe category of ‘transport-type’ LC/MS interfaces.While achieved by different means, these inter-faces include aspects of on-line sampling, desolva-tion, solvent vapor removal, and analyte deliveryinto low pressure environments. In LC/MS appli-cations, analytes are delivered solvent free to ionsources operating at pressures <10−4 Torr, at solu-tion flow rates in the case of conventional LCseparations in the range 0.2–2.0 mL min−1. Thus,use of these interfaces with GD sources would pro-vide an environment wherein the analyte shouldhave no memory of its solution phase heritage.While the moving belt has been used in a simpleelemental analysis, it has not been used to performspeciation-types of experiments.

LIQUID SAMPLING GLOW DISCHARGES 349

Nebulizer

Desolvationchamber

MS ionvolume

Momentumseparator

Vacuum

M+ Heatedtarget

He

LC

Figure 5.5.11. Basic components of a particle beam LC/MS interface.

A more widely used transport-type LC/MS inter-face is the particle beam (PB), shown schematicallyin Figure 5.5.11. PB interfaces, as first described byWilloughby and Browner [38], include some formof a nebulizer housed in a heated chamber wheredesolvation of the aerosol takes place, followed bya multistage momentum separator. The separatoraffects particle enrichment by removal of nebu-lizer gases and solvent vapors through the vacuum.In simple terms, mixtures of gas and particulatematter passing through small (≤1 mm diameter)orifices across pressure differentials produce anexpansion based on the relative momenta of thestream components. Lightweight species (gases)tend to expand radially and are skimmed from thebeam, while more massive analyte particles con-tinue to pass through successive orifices. Use of twoor more stages of differential pumping yields a truebeam of particles, free from residual gases. Com-plete removal of the solvent vapors allows particledelivery to ionization volumes operating at pres-sures of ∼10−4 Torr where they are flash vaporizedinto the gas phase. The PB interface provides ameans of sampling liquids at flow rates of up to∼1.5 mL min−1, while allowing ionization by elec-tron impact, chemical ionization and many othermethods. Marcus and coworkers have used thisapproach to introduce particles into GD sources foranalysis by optical emission and mass spectrome-tries as depicted in Figure 5.5.12.

Similar to the observations with transport-type interfaces for organic LC/MS, the noted

difficulties with the MB interface eventually ledthe exploration of new interface approaches forsolution introduction, specifically the PB inter-face [37, 38]. The ability to work across a widerange of solution phase compositions and flowrates, while still delivering dry analyte particles tothe mass spectrometer ionization volume are theadvantages of the PB approach. Strange and Mar-cus first described the use of a PB interface for theintroduction of liquids into GD plasmas [40]. Inthat work, Cu and Al disks were employed as thecathode in the discharge and the target at whichthe particle beam impinged. The observed opticalemission spectra indicated that there was little orno water solvent carried over from the momentumseparator at solution flow rates of 0.1–1.0 mL min.Scanning electron micrographs of particles col-lected at the cathode surface indicated that particlesranging from 0.5–10 µm in diameter were deliv-ered to the source.

4.1 Optical emission detection

A series of reports by You and Marcus describedthe design aspects, the sample introduction char-acteristics, and the analytical performance ofthe particle beam-hollow cathode-optical emis-sion spectrometry (PB-HC-OES) system [41–44].The HC geometry was adopted to obtain a moreenergetic plasma as well as to provide a heated sur-face (the wall) to vaporize the introduced particle.


He Gas

ParticleBeam

MomentumSeparator

VacuumGauge

TC

HeaterCartridges

LC

ThermalNebulizer

DesolvationChamber

He Gas

Vacuum Gauge

Thermocouple

Hollowcathodehν

(a)

Glow Discharge

LC

ThermalNebulizer

DesolvationChamber

PumpStage 1

PumpStage 2

ParticleBeam

MomentumSeparator

Ar Gas

Vacuum Gauge

Heater Cartridge

Direct InsertionProbe

M+

(b)

Figure 5.5.12. PB sample introduction coupled to (a) HCGD atomic emission and (b) mass spectrometry sources.

As expected, the response in optical emissionof the introduced analyte species depended onthe discharge operating conditions of dischargegas pressure and identity and current [41]. Theresults were in general agreement with thoseobtained with other HC sources, e.g., He wasfound to be a better excitation environmentthan Ar, and the optimum discharge gas pres-sure was in the 2–4 Torr range. Furthermore,the analyte emission intensity was found to beroughly proportional to the discharge current (upto 100 mA). A cursory evaluation of the roles of

the nebulizer tip temperature, the solution flowrate, and the cathode block (vaporization) tem-perature was also presented. Analyte responseprofiles for flow injection introduction showed lit-tle or no dispersion or tailing, suggesting thatthe memory effects were low and that the devicecould be effectively used as a chromatographicdetector. For the case of analytical calibrationcurves obtained for Na and Cs (nitrates) with neataqueous solutions, detection limits of 0.05 and0.1 µg mL−1 for 200 µL injections were achieved,which are about two orders of magnitude lower

LIQUID SAMPLING GLOW DISCHARGES 351

than those in the case of the planar cathode PB-GD geometry.

Detailed studies have been performed to under-stand the nebulization and particle transport char-acteristics in PB-HC-OES [42, 43]. The respectiveroles of the nebulizer tip temperature, the He neb-ulizer gas flow rate, and the use of a supplementalHe gas flow in the desolvation chamber were stud-ied as a function of the analyte solution flow rate.Each of these parameters was found to directlycontribute to the nebulization and vaporization pro-cesses in a more or less straightforward manner. Itwas important for the use of the device for LCapplications to find that the Cu(I) analyte emis-sion was insensitive to the water:methanol solventcomposition, ranging from 100 % water to 100 %methanol. As with other pneumatic nebulizer sys-tems, they obtained the highest analyte responseswith the use of small (∼50 µm) inner diameter sil-ica capillaries and high He nebulizer gas flow rates.Calibration curves for aqueous Cu, Pb, Fe andMg (including the aforementioned HCl addition)exhibited very good linearity, and the detectionlimits were found to range from 12 to 25 ng mL−1

for 200 µL injections.One of the key advantages of the use of low

pressure, inert gas plasmas lies in the fact thatthere is no continuous emission background fromatmospheric species such as C, N, O, H, etc. Of

course, these are also components of most liquidsample matrices and LC mobile phases. Therefore,the determination of such elements depends on theuse of highly efficient solvent removal methods,such as the PB interface. The use of standard flameand furnace atomic absorption and ICP sources forsuch determinations is less attractive due to lowdesolvation efficiencies and constant atmosphericbackground. The access to the above ‘gaseous’ ele-ments enables some level of molecular informationto be obtained in elemental speciation experiments.Namely, it is possible to determine atomic ratiosof these elements and to elucidate of the empiricalformula of a compound eluting from an LC col-umn. The application of PB-GD-OES to empiricalformula determinations, and by extension to ele-mental speciation was demonstrated. Basically, aplot of the ratio of the H(I) optical emission signalto that of C(I) as a function of the molar ratios ofthose elements for a range of aliphatic amino acidswas quite linear [44]. A similar response curve forthe H(I)/N(I) ratio is shown in Figure 5.5.13, illus-trating the ability to distinguish between a numberof aromatic amino acids as well [45]. It can beseen that the agreement is excellent. Table 5.5.2depicts some of the analytical figures of merit forthe application of LC/PB-GD-OES for the analy-sis of a pair of organomercury compounds as wellas a number of amino acids. Elemental detection

H/N Ratio in Amino Acids

H/N

Em

issi

on In

tens

ity R

atio

s

0

5

10

15

20

25

30

35

40

2 3 4 5 6 7 8 9 10 11 12

y = 2.6179x − 4.0443R2 = 0.9632

Histidine

Tryptophan

Proline

b−3, 4-Dihydroxy-Phenylalanine

Figure 5.5.13. Comparison of experimentally obtained H(I) and N(I) optical emission intensity ratios with the empirical formulavalues (H/N) for a range of aromatic amino acids determined by PB-HC-OES [45]. Reproduced with permission from the RoyalSociety of Chemistry.


Table 5.5.2. Analytical response characteristics for organomer-cury compounds by PB-HC-AES [45]. Reproduced with per-mission from the Royal Society of Chemistry.

Analyte Wavelength(nm)

r2 Detection limit(in 200 µL injections)

Hg inthimerosal

435.8 0.9979 2.9 × 10−11 molthimerosal(0.03 ppm Hg)

Hg inmerbromin

435.8 0.9995 4.1 × 10−11 molmerbromin(0.04 ppm Hg)

C inthimerosal

538.0 0.9834 3.5 × 10−8 molthimerosal(19 ppm C)

C inmerbromin

538.0 0.9934 5.2 × 10−8 molmerbromin(63 ppm C)

Na inthimerosal

589.0 0.9943 1.4 × 10−10 molthimerosal(0.02 ppm Na)

Br inmerbromin

614.9 0.9997 1.3 × 10−10 molmerbromin(0.1 ppm Br)

H inthimerosal

656.3 0.9877 2.2 × 10−9 molthimerosal(0.1 ppm H)

limits for H, C, and N in amino acid specimenscan be performed in the range from ∼0.1 to 3 ppmfor 200 µL injection volumes. This corresponds tomolecular detection limits of 10−9 M! It is envi-sioned that this approach has great potential forapplications in elemental speciation.

4.2 Mass spectrometric detection

In order to obtain the most comprehensive infor-mation in elemental speciation, one would wantunambiguous molecular weight and structural infor-mation of compounds eluting from the separationcolumn. Thus, ideally, one would have a mass spec-trometry ion source capable of accepting LC flowsand still produce meaningful molecular and elemen-tal mass spectra. Traditional ‘organic’ ionizationsources such as electron impact (EI) and chemicalionization (CI), while excellent at providing infor-mation on the molecular level, are virtually uselessfor the ionization of free metal atoms and smallinorganic complexes. By the same token, ICP-MSis a very powerful means of performing elemen-tal analysis of liquid specimens, but by its nature

all molecules are decomposed down to the atomiclevel. One accordingly needs an ionization sourceof modest energy such that organic compounds canstill be kept intact, while some atomic species yieldsimple ‘elemental’ mass spectra.

Gibeau and Marcus have used the sameapproach as described above, namely a PBinterface and a GD ionization source, to realizea versatile LC detector [46]. Different from theOES application, the beam of particles impingesdirectly on a disk-type cathode where they arevaporized/sputtered into the gas phase for subse-quent ionization. While not nearly so efficient asan HC source (to be pursued in the future), workwith this geometry has demonstrated that the basicapproach provides the sorts of information desiredin elemental speciation. As in the case of OES,experiments with amino acids have been done tocharacterize this technique. In Figure 5.5.14, thePB-GDMS spectra obtained for 200 µL injectionsof 150 ppm solutions for a set of selenoamino acidsare shown [47]. These spectra are very straightfor-ward to interpret and provide the exact type ofinformation required in comprehensive speciationexperiments. All other ’small’ molecules exam-ined to date have yielded mass spectra that arevery similar to those generated with traditionalEI sources, and in fact identification with the aidof standard databases is possible. One observessignals representative of the molecular weight ofthe molecule, and the successive loss of organicfunctional groups. The mass spectra obtained with‘elemental’ solutions are very simple in structure,with little or no evidence of oxide species or thelike. The sensitivity of the method at an early stageof development is comparable with that obtainedwhen coupling PB interfaces to EI sources, andthe detection limits for both organic molecules andelemental species are at the single nanogram level.Surprisingly, this value is in line with commercialICP-MS instrumentation. A separation and specia-tion of mixtures of inorganic lead and organoleadcompounds show the flexibility of the LC/PB-GDMS technique, as with the GD ionization sourcespectra are produced which accurately representboth types of species [47]. Also, separation and

CONCLUSIONS 353

Se

NH2

O

OH

Inte

nsity

0

1000

2000

3000

4000

80Se+

(SeCH3)+

(SeCH2CH3)+

(CH3SeCH2CH2)+(M-CH4)+

(M-COOH)+

(M-H)+

40 80 120 200160

Inte

nsity

0

500

1000

1500

2000

168+

80Se+

(SeCH3)+

(SeCH2CH3)+

(SeCH2CH2NH2)+

40 80 120

(a)

(b)

(c)

200160

SeNH2

O OH

Se

168

NH2

HO

O

Se

NH2

O

OH

m/z

Inte

nsity

0

200

400

600

800

1000

80Se+

(CH3CH2SeCH2)+

(M-CH4)+

(M-COOH)+

(M-H)+

(M-CH2CH3)+

(SeCH3)+

(SeCH2CH3)+

40 80 120 200160

Figure 5.5.14. PB-GD mass spectra of (a) seleno-DL-cystine, (b) seleno-DL-methionine, and (c) seleno-DL-ethionine [47].Reproduced with permission from Elsevier Science.

identification of mixtures of polyaromatic hydro-carbons (PAHs), steroids, and selenoaminoacidshas been carried out. It is believed that this method-ology holds a great promise in providing morecomprehensive speciation information than anyother MS source.

5 CONCLUSIONS

The preceding and following chapters of thisvolume set out very clearly the challenges andopportunities that exist in the very importantarea of elemental speciation. Issues of sampling,


preservation, sample preparation, separation anddetection are integrally linked. Unfortunately, thepart of the experiment that allows the actual identi-fication of species is probably the least evolved ofthem all. Existing, commercially available meth-ods have been adapted from their initial forms tobe very selective and sensitive detectors for ele-mental (predominately metal) species. To deliverthe target information though, detectors designedexplicitly for these jobs provide the best opportu-nity to assemble the required systems. Low pres-sure, low power GD devices offer the possibility toobtain the desired information in the biochemicaland environmental analysis communities. Small,low cost platforms that enable the ready couplingof a variety of sample introduction forms have afar greater chance of acceptance by the nonplasmacommunity. Ultimately, though, it is the informa-tion content provided by GD-OES and GDMS thatmust answer the call of the speciation community.Based on studies presented to date from a hand-ful of laboratories, the promise is there and awaitsthe support of the instrumentation community torealize its suggested potential.

6 REFERENCES

1. Marcus, R. K., Evans, E. H. and Caruso, J. A., J. Anal.At. Spectrom., 15, 1 (2000).

2. Chapman, B. N., Glow Discharge Processes , Wiley-Interscience, New York, 1980.

3. Marcus, R. K., (Ed.) Glow Discharge Spectroscopies ,Plenum, New York, 1993.

4. Baude, S., Broekaert, J. A. C., Delfosse, D., Jakubowski,N., Fuechtjohann, L., Orellana-Velado, N. G., Pereiro, R.and Sanz-Medel, A., J. Anal. At. Spectrom., 15, 1516(2000).

5. Puig, L. and Sacks, R., Appl. Spectrosc., 43, 801 (1989).6. Ng, K. C., Ali, A. H. and Winefordner, J. D., Spec-

trochim. Acta , 46B, 309 (1991).7. Starn, T. K., Periero, R. and Hieftje, G. M., Appl. Spec-

trosc., 47, 1555 (1993).8. Periero, R., Starn, T. K. and Hieftje, G. M., Appl. Spec-

trosc., 49, 615 (1995).9. Broekaert, J. A. C., Starn, T. K., Wright, L. J. and Hieftje,

G. M., Spectrochim. Acta , 48B, 1207 (1993).10. Broekaert, J. A. C., Starn, T. K., Wright, L. J. and Hieftje,

G. M., Spectrochim. Acta , 53B, 1723 (1998).11. McLuckey, S. A., Glish, G. L., Asano, K. G. and Grant,

B. C., Anal. Chem., 60, 2220 (1988).

12. Centineo, C., Fernandez, M., Pereiro, R. and Sanz-Medel,A., Anal. Chem., 69, 3702 (1997).

13. Rodriguez, J., Pereiro, R. and Sanz-Medel, A., J. Anal.At. Spectrom., 13, 911 (1998).

14. Orellana-Velado, N. G., Pereiro, R. and Sanz-Medel, A.,J. Anal. At. Spectrom., 13, 905 (1998).

15. Orellana-Velado, N. G., Pereiro, R. and Sanz-Medel, A.,J. Anal. At. Spectrom., 15, 49 (2000).

16. Martinez, R., Pereiro, R., Sanz-Medel, A. and Bordel, N.,Fresenius’ J. Anal. Chem., 371, 746 (2001).

17. Orellana-Velado, N. G., Fernandez, M., Pereiro, R. andSanz-Medel, A., Spectrochim. Acta , 56B, 113 (2001).

18. Pan, X., Hu, B., Ye, Y. and Marcus, R. K., J. Anal. At.Spectrom., 13, 1159 (1998).

19. Slevin, P. J. and Harrison, W. W., Appl. Spectrosc. Rev.,10, 201 (1975).

20. Orellano-Velado, N. G., Pereiro, R. and Sanz-Medel, A.,J. Anal. At. Spectrom., 16, 376 (2001).

21. Olson, L. K., Belkin, M. and Caruso, J. A., J. Anal. At.Spectrom., 11, 491 (1996).

22. Olson, L. K., Belkin, M. and Caruso, J. A., J. Anal. At.Spectrom., 12, 1255 (1997).

23. Belkin, M., Waggoner, J. W. and Caruso, J. A., Anal.Commun., 35, 281 (1998).

24. Milstein, L. S., Waggoner, J. W., Sutton, K. L. and Caruso,J. A., Appl. Spectrosc., 54, 1286 (2000).

25. Duckworth, D. C. and Marcus, R. K., J. Anal. At. Spec-trom., 7, 711 (1992).

26. Guzowski, J. P., Jr., Broekaert, J. A. C., Ray, S. J. andHieftje, G. M., J. Anal. At. Spectrom., 14, 1121 (1999).

27. Guzowski, J. P., Jr. and Hieftje, G. M., Anal. Chem., 72,3812 (2000).

28. Guzowski, J. P., Jr. and Hieftje, G. M., J. Anal. At.Spectrom., 15, 27 (2000).

29. Guzowski, J. P., Jr., Broekaert, J. A. C. and Hieftje,G. M., Spectrochim. Acta , 55B, 1295 (2000).

30. Ray, S. J. and Hieftje, G. M., Anal. Chim. Acta, 445, 35(2001).

31. Majidi, V., Moser, M., Lewis, C., Hang, W. and King,F. L., J. Anal. At. Spectrom., 15, 19 (2000).

32. Klingler, J. A., Savickas, P. J. and Harrison, W. W., J.Am. Soc. Mass Spectrom., 1, 138 (1990).

33. Harrison, W. W. and Hang, W., Fresenius J. Anal. Chem.,355, 803 (1996).

34. Bruhn, C. G., Bentz, B L. and Harrison, W. W., Anal.Chem., 51, 673 (1979).

35. Chapman, J. R., Practical Organic Mass Spectrometry:A Guide for Chemical and Biochemical Analysis , JohnWiley & Sons, Ltd, Chichester, 1993.

36. Scott, R. P. W., Scott, C. G., Munroe, C. G. M. andHess, J., Jr., J. Chromatogr., 99, 395 (1974).

37. Games, M. B., Adv. Mass Spectrom., 10B, 323 (1986).38. Willoughby, R. C. and Browner, R. F., Anal. Chem., 56,

2626 (1984).39. Creaser, C. S. and Stygall, J. W., Analyst , 118, 1467

(1993).

REFERENCES 355

40. Strange, C. M. and Marcus, R. K., Spectrochim. Acta ,46B, 517 (1991).

41. You, J., Fanning, J. C. and Marcus, R. K., Anal. Chem.,66, 3916 (1994).

42. You, J., Depalma, P. A., Jr. and Marcus, R. K., J. Anal.At. Spectrom., 11, 483 (1996).

43. You, J., Dempster, M. A. and Marcus, R. K., J. Anal. At.Spectrom., 12, 807 (1997).

44. You, J., Dempster, M. A. and Marcus, R. K., Anal.Chem., 69, 3419 (1997).

45. Dempster, M. A. and Marcus, R. K., J. Anal. At. Spec-trom., 15, 43 (2000).

46. Gibeau, T. E. and Marcus, R. K., Anal. Chem., 72, 3833(2000).

47. Gibeau, T. E. and Marcus, R. K., J. Chromatogr., A 915,117 (2001).

5.6 Electrospray Methods for Elemental Speciation

Hubert ChassaigneEuropean Commission – Joint Research Centre, Geel, Belgium

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 3562 Electrospray and Related Ionization

Techniques . . . . . . . . . . . . . . . . . . . . . . . . 3572.1 Electrospray . . . . . . . . . . . . . . . . . . . 3572.2 Pneumatically assisted electrospray 3582.3 Microelectrospray and

nanoelectrospray . . . . . . . . . . . . . . . . 3593 Sample Introduction and Analysis in

Tandem Mass Spectrometry . . . . . . . . . . . 3593.1 Sample introduction in electrospray 3593.2 Tandem quadrupole MS/MS system 360

3.2.1 MS mode 3613.2.2 MS/MS mode . . . . . . . . . . . . 3623.2.3 In-source collision-induced

dissociation mode . . . . . . . . . 364

3.2.4 Quantitative analysis . . . . . . . 3653.3 Quadrupole time-of-flight MS/MS

system . . . . . . . . . . . . . . . . . . . . . . . 3654 Applications in Elemental Speciation

Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 3664.1 Selenium speciation in yeast . . . . . . . 3674.2 Arsenosugars in seaweeds . . . . . . . . . 3704.3 Cadmium-induced phytochelatins in

plants . . . . . . . . . . . . . . . . . . . . . . . . 3704.4 Cadmium complexes with

metallothioneins in animaltissues . . . . . . . . . . . . . . . . . . . . . . . 374

5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . 3766 Acknowledgements . . . . . . . . . . . . . . . . . . 3767 References . . . . . . . . . . . . . . . . . . . . . . . . 376

1 INTRODUCTION

In terms of analytical developments, the demon-stration of analytical craft and skills of an analystto determine a particular elemental distribution ina sample are aimed at the detection of unknownelemental species, their identification and/or struc-tural elucidation [1, 2]. Exciting potential oppor-tunities are offered by electrospray (ES andrelated techniques) for a soft ionization of metal-containing species and by tandem mass spec-trometry (MS/MS) for a precise determination ofmolecular weight and structural characterization ofmolecules at trace levels in complex matrices.

The evolution in speciation analysis is partiallydue to the advent and spread of electrospray tandemquadrupole and quadrupole time-of-flight (Q-TOF)mass spectrometers in analytical laboratories and

the coupling of ES MS/MS with high resolutionseparation techniques, such as high performanceliquid chromatography (HPLC – MS/MS) [1, 3, 4].Microbore and capillary LC are compatible withthe majority of electrospray MS interfaces [5] andprovide high sensitivity for elemental trace analysis.

Speciation affects the bioavailability and toxic-ity of elements and so is important in toxicologyand nutrition [6, 7]. Both essential and nonessen-tial elements are taken into the body with food-stuffs and some elements may be biologicallyincorporated in food itself. Classic speciation anal-yses put strong emphasis on the evaluation of riskinduced by the contamination of foodstuffs [2].The greatest area of interest in the biologicalfield concerning trace element speciation probablyrelates to its influence on the bioavailability of

ELECTROSPRAY AND RELATED IONIZATION TECHNIQUES 357

essential elements and the availability and toxicityof toxic metals [8]. Organometallic species andmetal complexes arising in plants and animalsfrom biotransformation of metals and metalloidsare important in understanding the mechanisms oftheir biological effects.

Electrospray ionization is the method of choicefor organometallic species, proteins, oligopeptidesand their metal complexes. This chapter is illus-trated with a number of examples related to traceelement speciation analysis in biochemistry. Thecompounds of interest are endogenous and exoge-nous metal and metalloid species in foodstuffs (Sespecies in yeast and As in seaweeds), metal com-plexes with peptides and oligopeptides in plants(Cd-induced phytochelatins) and metal–proteincomplexes (Cd and Zn complexes with metalloth-ioneins) in animal tissues.

2 ELECTROSPRAY AND RELATEDIONIZATION TECHNIQUES

2.1 Electrospray

Droplets are generated when a high voltage isapplied to a liquid stream; this technique isknown as electrospray (ES) [9, 10]. The part of

the ionization source in which the ES processtakes place is operated at atmospheric pressure(Figure 5.6.1). The liquid sample is delivered via afused silica capillary inserted into a metal needle.The needle is held at a potential difference (typi-cally 3–5 kV) relative to the mass spectrometer’sentrance orifice. The typical flow range operatedwith this technique is 1–100 µL min−1. The volt-age on the needle causes the spray to be charged asit is nebulized. The droplets evaporate in a regionmaintained at a vacuum of about 1 mbar caus-ing the charge to increase on the droplets. Largerdroplets explode into smaller droplets and so onuntil protonated analyte molecules are released intothe gas phase [11]. Ions in the partial vacuum ofthe ion block are extracted and focused electro-statically (by ion lenses) into a quadrupole or ahexapole system which efficiently transports ionsinto the mass analyser. Pure ES in the context ofMS is accomplished without a nebulizing gas.

A new development in ES source geome-try is the orthogonal sampling technique. Thespray is directed perpendicularly to the spectrom-eter’s entrance hole (not shown). By modifyingthe geometry, the source design is claimed toresult in increased ruggedness and sensitivity com-pared with the early on-axis sampling systems,

Sprayer

High voltage3–5 Kv

Dissolvedsample

Vacuum pump

Ionlenses

Chargeddroplets

Atmosphere

≈ 1 mbar10−4−10−5 mbar

Entrance to ms

Plate

Extractioncone

Figure 5.6.1. Principle and geometries of an ES ion source (in-axis sampling system).

358 DETECTION

because fewer large droplets, neutral componentsand particles may enter the vacuum chamber ofthe mass spectrometer and contaminate the massanalyser. Ions are extracted orthogonally fromthe spray into the sampling cone aperture leav-ing large droplets, involatile materials, particu-lates and unwanted components to collect in thevent port.

2.2 Pneumatically assisted electrospray

An alternative ES technique uses a gas to assistnebulization and desolvation of liquids in ES.This method is called pneumatically assistedES (Figure 5.6.2). The source is an ambienttemperature ion source (without the application ofany heat) that typically accepts flow rates from1 to 200 µL min−1 without flow splitting. In thisway quasimolecular ions can be generated fromvery labile and high molecular weight compounds,without any degradation.

New pneumatically assisted ES exploits a dualorthogonal sampling technique to deliver twostages of sampling, one for ruggedness (contam-ination avoidance) and a second for sensitivity inmass spectrometry (Figure 5.6.2):

• In the first stage the spray is directed perpendic-ularly past the sampling cone as in the simpleorthogonal source geometry. Ions in solution areemitted into the gas phase.

• The second orthogonal step enables the volumeof gas (and ions) sampled from atmosphere tobe increased by a factor of 2–4 compared withconventional sources. The jet passes orthogo-nally to the second aperture so the flow intoit is significantly decreased. Ions in the partialvacuum of the ion block are extracted elec-trostatically into the quadrupole or hexapole,which efficiently transports ions to the massanalyser.

Background noise is significantly reduced bydual orthogonal sampling, contributing to the

High voltage3–5 Kv

Nebulizer gas

Sprayer

Dissolvedsample

Atmosphere

Extractioncone

Ionlenses

10−5 mbar

≈ 1 mbar

Samplingcone

Cleanable plate

Entrance to ms

X 2–4 Ion

block

Figure 5.6.2. Principle of pneumatically assisted ES for labile and high molecular mass compounds (interface with a dualorthogonal sampling technique).

SAMPLE INTRODUCTION AND ANALYSIS IN TANDEM MS 359

very good limits of detection and the highsensitivity obtained.

Additional modifications of the ion sourceare proposed to increase the sample utilizationefficiency and thus the sensitivity obtained in MS.Heated pneumatically assisted ES sources achievehigh sensitivity for quantification at high flowrates (up to 1000 µL min−1) (not shown). Theadvantages are:

• superior ionization efficiencies without thermaldegradation of labile compounds for improvedsensitivity;

• enhanced stability at high flow rates and highaqueous mobile phases for improved results andreliability with gradient HPLC methods;

• compatible with a wide range of flow rates,separation techniques and solvent compositionsfor increased flexibility.

2.3 Microelectrosprayand nanoelectrospray

Other developments resulting from the miniatur-ization of the ES source due to demands forlow analyte consumption in the direct introduc-tion mode and hence low flow rates in chromatog-raphy. They became known as microelectrospray(flow rate ca 100–200 nL min−1) and nanoelectro-spray (since the flow rate is ca 10–20 nL min−1)(not shown). Other terms have been coined suchas microspray and so on. The nanoelectrospraytechnique was introduced in 1994 [12, 13]. Thisin turn can provide more concentrated solutionsfrom small amounts of sample and/or long analysistimes because of the low flow rates. The samplesare loaded into pulled capillaries that have beencoated with metal. The flow typically starts whena voltage is applied (although a little pressure issometimes needed) and then continues until thesample is depleted or the metal sputters away.

The advantages of the miniaturization of thesprayer are:

• near 100 % sample utilization;• minimized contamination of the instrument;• elimination of cross contamination due to dis-

posable spray capillaries;

• ability to spray from purely aqueous as well aspurely organic solvent;

• flexibility for a variety of on-line electrospraycouplings (e.g. capillary LC).

An example of application in speciation analysiswas the identification of arsenic-containing com-pounds (arsenosugars) at the picogram level usingnanoelectrospray MS [14]. The application of themethod to real samples reveals the potential of thetechnique for trace analysis.

3 SAMPLE INTRODUCTIONAND ANALYSIS IN TANDEM MASSSPECTROMETRY

3.1 Sample introduction in electrospray

With recent design improvements to provideflexibility, ease-of-use, and fast installation, thesources are now ideal for any other ES approachsuch as coupling to HPLC (nanoscale to analytical)techniques. However, typical LC eluents do notreadily lend themselves to direct coupling withMS. The coupling of reversed-phase HPLC withES is the most compatible and has been widelydescribed in the literature [15, 16]. It is alsogenerally agreed that the greatest sensitivity in ESand pneumatically assisted ES is achieved withthe lowest liquid flow and the smallest diameterHPLC column [17].

Before beginning an LC/MS analysis, the massspectrometry should be optimized for the specificapplication in speciation analysis. The best methodof tuning is either by infusion or by direct loopinjection of the sample solution. In this way, theeffect of different mobile phase compositions onsample ionization can be tested in positive and/ornegative ionization mode, and suitable solvents canthus be selected.

For direct analysis of samples in ES,50/50 H2O–AcN is typically used as the mobileflow phase. Although 100 % water can be usedin ES, better sensitivity is obtained with someorganic solvent present in the water. Even 5–10 %of MeOH or AcN increases the stability of the

360 DETECTION

nebulization process and the desolvation step.When using high percentages of water it is usuallynecessary to increase the nebulization gas and/orthe source temperature to aid desolvation.

If the pH of the mobile phase needs to bereduced to enhance LC separations, then aceticand formic acids are suitable. Formic acid is moreacidic and less is required to reach a desired pH.Normally, less than 1 % of acid would be added,but often with LC more needs to be added toreach the required pH. Care should be taken duringnegative ion analysis, as the addition of acid tothe mobile phase can suppress ionization. Weaklyacidic compounds will not form deprotonated ionsin acidic conditions.

Inorganic acids (e.g. H2SO4, H3PO4) shouldnot be used with LC-MS even though there areexamples in the literature of very low levels ofHCl (<0.005 %) being utilized. Care should betaken if employed, as HCl is corrosive, and evenat low concentrations, can cause corrosion. Thisassumes that the ionization process is not degradedby the addition of such acids. The ion source wouldrequire more regular cleaning, however, if theseadditives were used.

Buffers such as phosphate, Tris, and Hepes can-not be used in ES. Even trace levels of these inter-fere with the ES process. Only volatile bufferssuch as ammonium acetate (CH3COONH4) orammonium formate (HCOOHN4) can be used.To improve chromatography without degradingthe MS performance it is best to use as little

CH3COONH4 as possible up to a maximum con-centration of 0.1 M. In many cases CH3COONH4

can be used to replace phosphate buffers which arehighly incompatible with LC-MS systems.

Excess Na+, K+, and detergents (such as sodiumdodecyl sulfate) are very bad for ES and frequentlyresult in no data. Detergents, by their nature, areconcentrated at the surface of a liquid. This causesa problem in ES as the ionization relies on theevaporation of ions from the surface of a droplet.The detergent congregates at the surface of thedroplet thus suppressing other ions.

3.2 Tandem quadrupole MS/MS system

The ES source operates at atmospheric pres-sure, so a quadrupole analyzer, which does notemploy high voltages, is easier to interface to theES source. Tandem quadrupole mass spectrom-etry makes use of a mass analyser (quadrupoleMS 1 in Figure 5.6.3) to select a particular m/zvalue (usually the molecular ion) for a CID(collision-induced dissociation) in a collision cell(quadrupole or hexapole). The fragment ions,called product ions, that form in the collision cellare mass analysed (by the quadrupole MS 2) torecord a product ion spectrum. This spectrum con-tains structural information about the ion selectedin the tandem MS experiment.

A tandem quadrupole MS/MS system hasa mass-to-charge ratio (upper) limit typicallyaround m/z 3000, in both MS and MS/MS

Quadrupolems 1

Quadrupolems 2

Hexapolecollision cell

Hexapoleion bridge

Ionsource

Detector

Pre-filter Post-filter

Figure 5.6.3. Tandem quadrupole mass spectrometer (MS/MS system).


modes. The mass accuracy is ca 0.01 % over theentire mass range. The typical resolution powerof such an instrument is 3000–4000 (FWHM).This system brings unparalleled performance forquantification and identification in the speciationfield. The tenfold increase in sensitivity, enhancedrobustness, and ease of use of the innovativenew ion sources (that accept higher flows andthe presence of salt contents in the samples)combined with new collision cells (e.g. linearaccelerator for a better fragmentation efficiency)deliver higher productivity with greater confidencein the results.

3.2.1 MS mode

In the MS operating mode, the first quadrupoleMS 1 (see Figure 5.6.3) is swept over a givenmass range and MS 2 is operated in RF-onlymode. The sensitivity (signal-to-noise ratio) can besignificantly increased by scanning a narrow massrange or by using the single ion monitoring modeto focus on particular masses of interest.

The ES mass spectra are characterized by quasimolecular ions, e.g. [M + H]+, [M + Na]+ etc. Inthe simplest case, one or more protons can beattached to the analyte molecule leading to theformation of a singly charged ion (generally formetal- and metalloid-containing species with amass up to 1000 Da) or multicharged ions (e.g.in the case of polypeptides and proteins andtheir complexes with metals). In the MS mode,no identification can be performed. The onlyinformation obtained is the molecular mass andthe isotopic pattern, allowing the confirmation ofknown compounds and their complexes.

The MS spectrum obtained in the case of lowmolecular mass compounds (up to 1000 Da) is verysimple and contains only singly charged molecu-lar ions (singly protonated molecular ions). How-ever, for species containing an element havingseveral isotopes, such as organoselenium com-pounds, an envelope corresponding to protonatedmolecules with the most abundant isotopes canbe observed. Figure 5.6.4(a) shows the spectrumobtained for a common selenoamino acid encoun-tered in biological samples, selenomethionine [18].

Its shows an m/z envelope within which theisotopic pattern of selenium can be identified(protonated molecular ion [M + H]+ at m/z 198containing the most abundant 80Se isotope). A par-tial fragmentation can be obtained despite the rela-tively low extraction energy used (potential of theorifice or the extraction cone). Thus a fragment M-17 corresponding to the loss of an OH group fromthe carboxylic group is observed in Figure 5.6.4(a).

ES is particularly useful for obtaining themolecular weight of analytes that would normallyexceed the normal mass range of sector andquadrupole instruments. The most obvious featureof an ES spectrum is that the molecular ionsof large molecules carry multiple charges, whichreduce their mass-to-charge ratio compared with asingly charged species. This allows mass spectrato be obtained for large molecules.

For polypeptides and proteins and their com-plexes with metals, the formation of multiplyprotonated molecular ions occurs [10, 19, 20].Figure 5.6.5(a) shows an MS spectrum obtainedfor a class of metalloproteins, Cd-induced metal-lothioneins isolated from animal tissues (isoformMT-2 from rabbit liver) [21, 22]. Two differentenvelopes of multicharged peaks can be seen inthe spectrum. Assuming that adjacent peaks in theion envelope differ by only one charge and that thecharging is due to protonation, the relation betweena multiply charged peak at m/z and the relativemolecular mass is shown in Figure 5.6.5(a) [23].The higher the number of m/z signals that can beseen for a given compound in the mass spectrum,the more precise is the result of the molecular massdetermination. Three sub-isoforms of MT-2 (whichdiffer by a few amino acids in their composition)with very close molecular masses can be identifiedin the preparation. They were identified as MT-2a,MT-2b and MT-2c (Figure 5.6.5(a)) with molec-ular masses of 6125.5, 6146.0, and 6155.5 Da,respectively. Apart from the metallothioneins, thesample also contains another protein (multichargedpeaks +15, +14 and +13) with the molecular mass15 570.0 Da, identified as superoxide dismutase.

362 DETECTION

80 100 120 140 160 180 200

1.0

0.8

0.6

0.4

0.2

Inte

nsity

, cps

m/z, uma

747677788082

177.0

178.0

179.0

196.0

195.0

194.0

183.0

200.0

198.0

181.0

CH3Se(CH2)2 CH

NH2

COH+

[Se−Met+H]+

0.88%

8.95%7.65%

23.51%

49.62%

9.39%

(a)

×106

80 100 120 140 160 180 200

1.6

1.2

0.8

0.4

Inte

nsity

, cps

m/z, uma

78.0

76.082.0

198.0196.0

194.0

[Se−Met+H]+

Se+

80.0

(b)

×105

(c)

80 100 120 140 160 180 200

6.0

5.0

4.0

3.0

2.0

4.0

3.0

2.0

1.0

1.0 74.0

74.0

198.0

102.0

102.0

109.0107.0

135.0

133.0

152.0

150.0

181.0

179.0

196.0

×105

×105

+H+

198(Se80)/196(Se78)

181(Se80)/179(Se78)

−OH

COOH

NH2

152(SE80)/150(Se78)

109(Se80)/107(Se78)

135(Se80)/133(Se78)

102

CH3 Se CH2 CH2 CH

74

m/z, uma

Inte

nsity

, cps

Figure 5.6.4. Illustration of the potential of ES MS/MS in speciation analysis using different data acquisition modes.Characterization of a selenium-containing compound (selenomethionine): (a) MS spectrum, (b) SCID mode for elemental analysis,(c) MS/MS spectrum. Reprinted from Trends in Analytical Chemistry , Vol. 19, H. Chassaigne, V. Vacchina and R. Lobinski,p. 300, Copyright (2000), with permission from Elsevier Science.

3.2.2 MS/MS mode

In the MS/MS mode, mass spectrometer 1 isused as a filter to transmit the quasi molecularion [M + H]+ (see Figure 5.6.3) to the collisionchamber where it is fragmented by collision withneutral gas molecules in a process referred to ascollision-induced dissociation (CID). The collisiongas is typically nitrogen and the collision energiesused can be in the range 10–50 eV depending onthe mass of the compounds and to obtain optimumfragmentation. The mass of mass spectrometer 1

is fixed and mass spectrometer 2 is swept over agiven mass range to determine the ions that resultfrom the fragmentation of the precursor molecularion. The resulting fragment ions are analysed bythe second mass analyser (see quadrupole massspectrometer 2 in Figure 5.6.3) allowing their massdetermination and information on the molecule tobe obtained.

In particular, for species containing an ele-ment having more than one stable isotope, suchas selenomethionine, valuable information canbe obtained by fragmenting the two protonated


molecules containing the adjacent most abun-dant isotopes (m/z 196 and 198 corresponding to78Se and 80Se, respectively) (Figure 5.6.4(c)) [18].Selenium-containing fragments are separated bythe distance of two mass units whereas fragmentsthat do not contain selenium will remain at thesame m/z value, thus facilitating the interpreta-tion of the mass spectra. However, peaks obtainedby fragmentation of the molecular signals due todifferent isotopes may vary by 0.1–0.5 Da witha quadrupole instrument, which makes it diffi-cult to identify the elemental species in the CIDMS spectra [24].

Some studies clearly show the limitationsof a triple quadrupole instrument for speciationstudies and especially in the case of monoisotopic

elements (e.g. arsenic). Since arsenic has only oneisotope, the molecular ions observed in the MSmode may suffer from a severe risk of overlaps inthe case of real matrices which would mean thatthe positive identification of arsenic-containingcompounds exclusively on that basis would beunreliable [25]. However, a deeper insight intothe structure of the compounds can be gained byanalysing product ions resulting from the CID ofthe protonated molecular ions.

ES MS/MS is also widely used to determine thesequence of polypeptides and proteins. Peptidesof limited molecular mass (up to 2000 Da) canbe subject to sequence analysis by MS/MS [26].Peptides fragment primarily at the amine bonds toproduce a ‘ladder’ of sequence ions. The charge

1040 1080 1120 1160 1200 1240 1280 1320

m/z

(a)

Inte

nsity

, cps

6+

apo-MT-21022.0

5+

apo-MT-21226.0

15+

1039.0

a

0.7

2.1

3.5

4.9

6.3

7.7

9.1

×105

14+

1113.0

13+

1199.0

b

a

b

c

c

MMMMT-2a MW = 6125.5 ± 0.5MT-2b MW = 6146.0 ± 0.5MT-2c MW = 6155.5 ± 0.5

Superoxide dismutaseMW = 15570.0 ± 0.4MW = (m/z × z) − nH

(m/z)Hi

(m/z)Hi − (m/z)Low

ZLow =

Figure 5.6.5. Potential of ES MS/MS for the analysis of proteins and peptides. (a) Determination of the molecular mass ofCd-induced metallothionein MT-2 (rabbit liver) using the multiply charged ion envelope. Three sub-isoforms are identifiedas MT-2a, MT-2b, and MT-2c. Reprinted from the Journal of Analytical Chemistry , Vol. 361, H. Chassaigne, R. Lobinski,pp. 267–273, Figure 4, 1998, copyright notice of Springer-Verlag. (b) CID mass spectra of a Cd-induced peptide isolated fromplants (phytochelatin PC4) Reprinted from Trends in Analytical Chemistry , Vol. 19, H. Chassaigne, V. Vacchina and R. Lobinski,p. 300, Copyright (2000), with permission from Elsevier Science.

364 DETECTION

gGlu Cys gGlu Cys Gly

b2b1 b3 b4 b5 b6 b7 b8 MH+

y8 y7 y6 y5 y4 y3 y2 y1

gGlu Cys gGlu Cys

Predicted sequence

Symbol Mass ions y

E, Glu

C, Cys

E, Glu

C, Cys

E, Glu

C, Cys

G, Gly

129.04

103.01

129.04

103.01

129.04

103.01

57.02

772.20

643.15

540.14

411.10

308.09

179.05

76.04

ECECECECG

ions b

130.05

233.06

362.10

465.11

594.15

697.16

826.21E, Glu

C, Cys

129.04

103.01 929.21

986.24

875.20

1004.25

b2233

y5540 b6

697

500 1000

m/z

(b)

2.0

4.0

6.0

8.0

Inte

nsity

, cps

×103 [M + H]+

1004

b8929

y8875

y7772y6

643

y3308

MS/MS spectrum

Phytochelatin PC4


can be retained on the amino terminus (type b ion)or on the carboxy terminus (type y ion). Thus acomplete series of ions from both types allowsthe determination of the amino acid sequence. Thepotential of this mode is illustrated in the case ofpeptides such as cadmium-induced phytochelatinsin plants (phytochelatin PC4 with MW 1004 Da)(Figure 5.6.5(b)) [18].

3.2.3 In-source collision-induceddissociation mode

The ES interface can be tuned to provide fragmentions that unlock the structure of eluting moleculesand this leads to the formation of elemental ions ofinterest in speciation analysis. In-source fragmen-tation can be controllably induced by the voltage

applied to the extraction cone or orifice in the inter-face (see Figure 5.6.2). Source collision-induceddissociation (SCID) is promoted by increasing thisvoltage, resulting in a CID mass spectrum which isa unique fingerprint of the analyte. In the exampleof a selenoamino acid (selenomethionine) (seeFigure 5.6.4(b)), the protonated ion [M + H]+ atm/z 198 can be broken down by increasing theionization energy. At a sufficiently high ioniza-tion energy, elemental Se cations can be obtainedfrom the sample [18, 27]. The detection limits forthe elements, however, are 2–3 orders of magni-tudes higher than in the case of inductively coupledplasma mass spectrometry (ICP MS) [18].

The type of information obtained in this oper-ational mode of ES mass spectrometry is ofparamount importance in speciation analysis. In


the case of complex samples when a number ofconcomitant ions can interfere with the identifi-cation of the element isotopic pattern (e.g. forselenium) and especially for a monoisotopic ele-mental species (e.g. for arsenic), the SCID may bean alternative.

3.2.4 Quantitative analysis

For quantification, the tandem quadrupole massspectrometer can be operated in the differentacquisition modes as previously described:

• Full scan in MS (scanning over a definedmass range).

• SIM (selected ion monitoring) mode. The massspectrometer sensitivity (signal-to-noise ratio)can be significantly increased, by narrow massscanning or by selected ion monitoring.

• Full scan in MS/MS.• MRM (multiple reaction monitoring) mode.

MRM allows MS/MS to be used for multipleanalytes in direct analysis and coeluting analytesin HPLC. MRM allows a few parent ions to beisolated and dissociated under optimum condi-tions with monitoring of all the correspondingproduct ions.

An example of quantification using a quadrupoleinstrument was given by Madsen et al. [28]. Theidentity of arsenic-containing species (arsenosug-ars) was confirmed by LC-ES MS with a variablefragmenter voltage which provided simultaneouselemental and molecular detection. LC-ES MS wasfurther used to quantify four arsenosugars, pro-ducing values within 5–14 % (depending on thecompound) of the ICP MS data.

3.3 Quadrupole time-of-flightMS/MS system

A quadrupole time-of-flight (Q-TOF) instrumentis an MS/MS system combining the simplic-ity of a quadrupole (mass spectrometer 1) withthe ultrahigh efficiency of a TOF mass analyser(mass spectrometer 2) (Figure 5.6.6). The system

exploits a TOF mass analyser to achieve simul-taneous detection of ions across the full massrange. This is in contrast to conventional instru-ments (tandem quadrupole system) that must scanover one mass at a time. A Q-TOF instrumentoffers up to 100 times more sensitivity than tan-dem quadrupole instruments when acquiring fullproduct ion (MS/MS) mass spectra.

A Q-TOF system has a mass-to-charge ratio(upper) limit typically in excess of m/z 20 000, inboth MS and MS/MS modes, enabling the analy-sis of very large molecules as multiply chargedions. The high resolving power (5000–10 000FWHM) enables improved mass measurementaccuracy for small molecules, charge state iden-tification of multiply charged ions and greaterdifferentiation of isobaric species. The inherentstability of the reflectron TOF analyser routinelydelivers excellent mass measurement accuracy(0.0002–0.0005 %) for molecules of low molec-ular mass (up to 1000 Da). Exact mass mea-surements enable the masses of molecular and/orproduct ions (MS/MS) to be confirmed for knowncompounds. For unknowns the number of plausi-ble structures may be restricted to a small numberwith the aid of additional chemical information(metallic or nonmetallic element of interest inthe structure).

For identification, the Q-TOF hybrid MS/MSsystem is a powerful qualitative tool providinghigh full-scan sensitivity and excellent resolutionof intact compounds as well as product ion spectra.The system can easily determine the elementalcomposition of various compounds.

For a Q-TOF instrument the acquisition modesfor quantification are:

• Full scan in MS with the reflectron TOF analyser(mass spectrometer 2 in Figure 5.6.6).

• Full scan in MS/MS. CID mode of molecularions selected by mass spectrometer 1 andsimultaneous detection of fragment ions acrossthe full mass range by mass spectrometer 2(Figure 5.6.6).

The performances of a Q-TOF mass spec-trometer has been evaluated in terms of accu-racy and precision for the identification of low

366 DETECTION

Pre-filter

Quadrupolems 1

Hexapoleion bridge

Ionsource

Pusher

Collision cell

TOFms 2

Detector

Post-filter

Reflectronsystem

Figure 5.6.6. Geometry of a quadrupole TOF mass spectrometer system (MS/MS system).

molecular weight selenium compounds [24, 29].The accuracy of mass measurements, evaluatedwith standards (selenomethionine, selenoethionineand selenocystine) was found to be between 0.005and 0.01 % (0.01–0.02 Da) [24]. Previous work inthis field was done with a triple quadrupole massanalyser [30, 31]. A triple quadrupole instrumentsoffers 0.1–0.5 mass resolution which makes therecognition of the selenium pattern in a mass spec-trum often ambiguous, especially when a foreigncompound gives a peak at one of the molecularmasses within the isotopic cluster. The precisionof the molecular mass measurement in the MS/MSmode was evaluated as 0.004 % for selenomethio-nine and selenocystine [24].

Another study shows the performance of aQ-TOF instrument for the characterization ofarsenic compounds in biological samples [32].The mass accuracy obtained with this instru-ment enables the characterization of arsenosug-ars in a complex sample. These results offernew perspectives for the rapid recognition, and

subsequent identification, of unknown arsenicspecies in crude extracts, without the need forextensive purification or previously character-ized standards.

4 APPLICATIONS IN ELEMENTALSPECIATION ANALYSIS

The technique was shown to be capable ofproducing gas-phase ions of highly labile andnonvolatile compounds, such as peptides, proteinsand oligosaccharides [18]. This ionization processis so gentle that noncovalently bound metalcomplexes, such as complexes of Cd and As [27],were also shown to be desolvated and were studiedby MS. However, the sample must be soluble insolvents (AcN, MeOH, water, . . .) and the speciesstable at very low concentrations. Examples arediscussed to illustrate the potential of ES inspeciation analysis.

APPLICATIONS IN ELEMENTAL SPECIATION ANALYSIS 367

4.1 Selenium speciation in yeast

The fact that selenium (Se) is essential for humanbeings was demonstrated only recently in someregions where the intake of selenium is very low.Certain forms of cancers [33, 34] and cardiovascu-lar diseases [35] have in some studies been asso-ciated with a low intake of selenium by people.Hence, supplements have been used for improvingthe Se status in the human body and to prevent dis-ease to progress. However, the chemical form andconcentration in which Se is introduced in to thebody are of primary importance. Further investiga-tions have suggested that selenized yeast produceselenomethionine along with other organoseleniumcompounds. Selenized yeast makes the seleniumboth bioavailable and provides it in a form whichthe body may use beneficially. Since the bioavail-ability and the toxicity of Se are closely corre-lated with its chemical form and concentration,the information on Se speciation is vital [36, 37].For these reasons the demand for accurate andsensitive methods for Se speciation in nutritionalsupplements, e.g. Se-enriched yeast, has rapidlyincreased.

Selenocompounds cannot be analysed by ESMS in yeast extracts directly because of the pres-ence of a matrix composed of high molecularweight compounds and salts suppressing the sig-nal [24]. A sequential extraction and a fraction-ation of the extract by preparative size exclusionchromatography is often required to isolate the lowmolecular weight fractions [38]. Further purifica-tion of low molecular weight compounds by HPLC(e.g. anion exchange) may be necessary [39, 40].An approach was proposed by Casiot et al. [41]who isolated the major selenocompounds in yeastextracts and identified them on the basis of theCID pattern of the protonated molecular ions cor-responding to the adjacent Se isotopes by EStandem quadrupole MS/MS. Thus, selenomethio-nine and Se-adenosylhomocysteine were success-fully identified [36, 41]. In contrast, previous workwas carried out with a triple quadrupole systemwhich offers poor resolution and often makes therecognition of the Se pattern in a mass spectrumambiguous [24].

Figure 5.6.7(a) shows the mass spectrum obtain-ed for a chromatographic fraction [18, 30]. Thespectrum reveals the presence of an unresolved ioncluster, matching the characteristic abundance ofthe Se isotope pattern, centred at m/z 433. The MSmode gives valuable information on the molecularmass of the species and the presence of one Sein the molecule. Information on the identity of thecompound can be obtained by fragmenting the quasimolecular ion by CID. The MS/MS spectra obtainedfor the fragmentation of the ions at m/z 431 and433 (corresponding to the two most abundant Seisotopes 78Se and 80Se, respectively) are shown inFigure 5.6.7(b). The comparison of the two spectraallows the differentiation between fragments that donot contain Se (m/z signal in both spectra at the samevalue) and fragments that contain Se (m/z signalin both spectra appear with a difference of 2 Da).The structure proposed for this compound was Se-adenosyl-homocysteine [30, 31].

In recent work, the purification of seleniumcompounds from the yeast extract has been simpli-fied for further characterization of low molecularweight Se compounds by Q-TOF mass spectrome-try [24]. In quadrupole MS of this fraction onlythe most abundant cluster centred at m/z 433could be seen [30] whereas a number of otherclusters, e.g. m/z 182, 196, 298, are observed inthe MS spectrum of low molecular weight com-pounds (Figure 5.6.8(a)). When performing ele-mental analysis calculations with the instrument,the compound at m/z 433.0350 can be assigned theformula C14H20N6O5Se [24]. In Figure 5.6.8(b)the observed and the calculated isotopic patternof the molecule are compared. An insight into thestructure of the compound can be obtained by CID(data not shown). The data are similar to thoseobtained with a triple quadrupole MS/MS system(see Figure 5.6.7(b)). The advantage of the Q-TOFanalyser compared with the triple quadrupole sys-tem is the rapidity of spectra combined with highresolution, which enables the rapid fragmentationof all the molecular ions within the cluster usinga small amount of sample. ES Q-TOF, especiallywhen in tandem mode, provides novel informa-tion regarding the identity of selenocompounds inselenized yeast [24].

368 DETECTION

385.0

433.0

350 400 450 500 550 600 650

m/z

(a)

1.0

2.0

3.0

4.0

5.0

Inte

nsity

, cps

Inte

nsity

, cps

431.0

[M + H]+

×106

415 420 425 430 435 440 445

m/z, uma

1.0

2.0

3.0

4.0

5.0 Se78

431

Se82

435

Se80

433

Se76

429

Se77

430

×106

100 150 200 250 300 350 400

m/z

(b)

136. 0

182. 0

250. 0

298. 0

433. 0

0.5

1.0

1.5

2.0

1.2

1.0

0.8

0.6

0.4

0.2Inte

nsity

, cps

x105

x105136. 0

180. 0

250. 0

296. 0

431.0

[M + H]+

N

N

NH2

N

NHOOCCH−(CH2)2−Se−

NH2

CH2

HO OH

O

+2H+ 136

296 (78Se) / 298 (80Se)

250

180 (78Se) / 182(80Se)

+H+ 431 (78Se) / +H+ 433 (80Se)

Se-adenosyl-homocysteine

Figure 5.6.7. Identification of a selenocompound in a selenized yeast extract by ES quadrupole MS/MS. (a) ES MS spectrum ofa fraction collected in the RP HPLC – ICP MS chromatogram of a water yeast extract. Reproduced by permission of the RoyalSociety of Chemistry. (b) ES MS/MS spectra of the selenium-containing ions at m/z 431 and 433. The fragmentation patternof the identified compound, Se-adenosylhomocysteine, is shown in the inset. Reprinted from Trends in Analytical Chemistry ,Vol. 19, H. Chassaigne, V. Vacchina and R. Lobinski, p. 300, Copyright (2000), with permission from Elsevier Science.


Inte

nsity

, cou

nts

100.

0611

114.

0764

136.

0407

149.

0092

179.

9535

181.

9551

196.

1514

428.

0486

431.

0441

433.

0350

494.

9636

100 150 200 250 300 350 400 450

I

II

500

×105

0.0

1.0

2.0

(a)

426 436434432430

m/z

Inte

nsity

, %

Inte

nsity

, cou

nts

427.

0577

429.

0486

430.

0550

432.

0529

434.

0461

435.

0398

431.

0441

433.

0350

436.

0439

428426 428 430 432

m/z

434 436 438

00

0.2

0.4

0.6

0.8

1.0

1.2

1.4×105

20

40

60

80

100

HOOC SeO

NH2N

N

N

N

NH2

HO OH

(b)

Figure 5.6.8. Confirmation of the elemental composition of a selenium compound in yeast extract by ES TOF MS(Se-adenosylhomocysteine). (a) ES MS spectrum of a yeast fraction collected after preparative anion exchange chromatogra-phy – ICP MS. (b) Zoom of the m/z 433.0350. Se-containing species compared with a theoretical pattern for the species shown.Reproduced by permission of the Royal Society of Chemistry.

370 DETECTION

4.2 Arsenosugars in seaweeds

Arsenic biotransformations by marine life areknown to give rise to a wide range of organoarseniccompounds. Arsenobetaine (AsB) is the mostabundant species in marine animals whereasshellfish and algae may contain a class of arsenic-containing ribofuranosides (so-called arsenosug-ars) [14, 42, 43]. This class of compounds hasbeen raising growing concern since recent reportsindicating the possibility of metabolizing arseno-sugars to the carcinogenic dimethylarsinic acid(DMAA) by the human body [44]. As arsenosugarstandards are not available, the characterization ofsuch compounds in seaweeds requires their isola-tion [45] and purification before their analysis bymass spectrometry.

The poor potential of tandem quadrupole massspectrometry in the case of arsenic-containingspecies is hampered by the fact that this element ismonoisotopic so it is difficult to attribute a peak inthe mass spectrum to an arsenic species. However,a confirmation of the identity of the arsenosugarspresent can be obtained by the fragmentation ofthe pseudomolecular ions and mass spectrometryof the resulting fragments.

The matrix suppression of ES prevents itsapplication for crude extracts but the techniquemay be used for signal identification in fractionshighly purified by multidimensional chromato-graphic techniques, e.g. size exclusion HPLC andion exchange HPLC [46–48]. Figure 5.6.9 showsthe CID mass spectra of the protonated moleculesof four arsenosugars (A–D) extracted from algaeand purified by complementary chromatographictechniques [25]. As indicated in the earlier stud-ies [46], the common feature of the MS/MS spectraof arsenosugars is the presence of an ion signal atm/z 237 corresponding to the oxonium ion of thedimethylarsinoylpentose moiety. Another charac-teristic ion is that at m/z 195, which results fromthe break-up of the furane ring and indicates theattachment of the dimethylarsinoyl moiety to the5′ position of the furane ring [25]. ArsenosugarsA and C lose the SO3 moiety readily; the result isthe presence of an M+−80 fragment in the caseof sugar C or of an M+−98 fragment (m/z 295)

in the case of arsenosugar A. The fragments atm/z 97 and 80 have been assigned to OSO3H andSO3, respectively. The fragmentation never leadsto the bare As+ and or to the AsO+ ion common inthe mass spectra of simpler organoarsenic species(such as arsenobetaine), even when high extractionenergy is used [25].

In recent studies, the use of accurate, highresolution MS has been shown to be a power-ful analytical technique of great promise for theidentification of unknown arsenic compounds incrude algal extracts [32]. Q-TOF mass spectra ofa fractionated extract (by cation exchange chro-matography) were obtained over the m/z range0–400 on an instrument calibrated to a mass accu-racy of 0.0002 %. Figure 5.6.10(a) shows the pres-ence of a peak at m/z 329.1. However, since themass spectrum contains a number of peaks andarsenic has only one stable isotope (in contrastto selenium), the selection of this particular m/zvalue has required an assumption that arsenosug-ars were present in the mixture. When performingelemental analysis calculations with such an instru-ment, increasing the mass accuracy of the mea-surement reduces the number of possible elementalcombinations for the ion being investigated. Thispeak may correspond to the protonated moleculeof arsenosugar B (see Figure 5.6.9(b)). CID tan-dem MS was performed on the m/z 329.1 ion(Figure 5.6.10(b)). The resultant mass spectrum ofthe fragment ions was also subject to a search usingthe elemental composition tool, which confirmedthe composition of arsenosugar B. ES Q-TOFMS with high mass accuracy allows the numberof peaks that could correspond to arsenic to beconsiderably reduced, facilitating recognition ofunknown arsenic-containing compounds and rep-resenting a useful application on a more widelyavailable instrument.

4.3 Cadmium-induced phytochelatinsin plants

Like all organisms, plants present a dilemma inthat metals such as Cu and Zn are essential tracemetal nutrients taking part in redox reactions,electron transfers, a multitude of enzyme-catalyzed


100

96.7

148.

6

149

195

237

281

295

375

393

+H+

−H2O

195.

123

7.1

280.

629

5.1

374.

8392.

8

150

Intensity, cps

200

250

300

350

m/z

, am

u

0.5

1.0

1.5

x104

H3C

AsO CH

3

O

HO

OH

O

OH

SO

3H

(a)

329

+H+

Intensity, cps

100

12345

x104

150

96.9

164.

7

195.

023

7.0

269.

432

8.8

200

250

165

195

237 30

035

0m

/z, a

mu

H3C

AsO CH

3

O

HO

OH

O

OH

OH

(b)

100

80.2

96.7

195.

0237.

1

328.

9

237

391

483

329

−H2O

391.

0

483.

1

+H+

150

200

250

300

350

400

m/z

, am

u

Intensity, cps

0.2

0.4

0.6

0.8

1.0

x104

H3C

AsO CH

3

O

HO

OH

OO

HO

POO

HO

OH

OH

(d)

100

80.0

97.0

195.

0

195

409

+H+

237

237.

0

329

329.

2

150

200

250

300

350

m/z

, am

u

12345x104

Intensity, cps

H3C

AsO CH

3

O

HO

OH

O

OH

OS

O3H

(c)

Fig

ure

5.6.

9.Id

entifi

catio

nof

arse

nic

com

poun

dsin

alga

lex

trac

tsby

ES

quad

rupo

leM

S/M

S.

MS/

MS

spec

tra

ofth

epr

oton

ated

mol

ecul

eio

nsof

arse

nic-

cont

aini

ngri

bofu

rano

side

s(a

rsen

osug

ars)

extr

acte

dfr

omal

gae

and

puri

fied

byco

mpl

emen

tary

chro

mat

ogra

phic

tech

niqu

es.

(a)

Ars

enos

ugar

A;

(b)

arse

nosu

gar

B;

(c)

arse

nosu

gar

C;

(d)

arse

nosu

gar

D.

Rep

rint

edfr

omA

naly

tica

Chi

mic

aA

cta

,V

ol.

410,

S.M

cShe

ehy,

M.

Mar

cine

ck,

H.

Cha

ssai

gne,

and

J.Sz

puna

r,p.

71,

Cop

yrig

ht(2

000)

,w

ithpe

rmis

sion

from

Els

evie

rSc

ienc

e.

372 DETECTION

(a)

100

%

60 80

60.1 83.1

87.0101.0

118.1120.1

144.1147.1

116.1 160.1

169.0

189.2

241.1

212.3275.1 311.3

329.1313.3 336.2

m/z

m/z

236.1

100 120 140 160 180 200 220 240 260 280 300 320

60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360

340 360 380 400

0

100

%

0

(b)

97.0

195.0

237.0

329.1

330.1

Figure 5.6.10. Identification of arsenic compounds in algae extracts by ES quadrupole TOF MS/MS. (a) MS spectrum of anextract roughly purified by cation exchange preparative chromatography. (b) ES MS/MS spectrum of the protonated moleculeion of the arsenic-containing ion at m/z 329.1. The fragmentation pattern allows the confirmation of the identity of the compoundas arsenosugar B. Reproduced by permission of The Royal Society of Chemistry.

reactions, and structural function in nucleic acidmetabolism. However the same metals present athigh concentrations, and even low concentrationsof the more potent ions of Cd, Hg, etc., arestrongly poisonous, resulting in growth inhibitionand death of the organism [49]. In order to surviveplants must have developed efficient and specificmechanisms by which heavy metals are taken upand transformed into a physiologically tolerableform, providing the essential elements for theplant’s metabolic function [49]. Some studies haverevealed that the majority of higher plant speciesdetoxify the metals by chelating them to peptidesof the family of phytochelatins [50].

Phytochelatins (PC) are metal-binding peptideswhich are enzymatically synthesized from glu-tathione (GSH). These peptides have been shownto be induced in plants by various metals suchas Cd, Cu and several other metals [49, 51].In addition to PCs which possess the typical

(γ Glu-Cys)n-Gly (n = 2–11) sequence, severalvariant structures have also been detected whichdiffer in the C-terminal amino acid and are calledisophytochelatins (iso-PC) [52]. Homologues ofglutathione with C-terminal linked Glu insteadof Gly and isophytochelatins (Glu) (iso-PC(Glu),(γ Glu-Cys)n-Glu, n = 2–3) were recently isolatedfrom maize plants exposed to cadmium [53, 54].In addition, desglycine phytochelatins (desGly-PC,(γ Glu-Cys)n), which lack the C-terminal aminoacid residue, were first discovered in maize [52].

The classical approach used for the determina-tion of PC extract from plant is based on reversed-phase HPLC with on-line derivatization of thesulfhydryl groups with Ellman’s reagent (DTNB)and spectrophotometric detection. Reversed-phasechromatography of Cd-exposed and control maizeroot extract was used for isolation and purifica-tion of phytochelatins and phytochelatin-relatedpeptides (chromatograms not shown) [55]. In the


(a)

(b)

715

772

844947

1004

600 700 800 900 1000

m/z

1.0

2.0In

tens

ity, c

ps

iso-PC3(Glu)

desGly-PC3

desGly-PC3

PC3

desGly-PC4

desGly-PC4

PC4

1004

PC4

586643

875

X 106

x 106

567

715

772

844947

600 700 800 900 1000

m/z

1

2

Inte

nsity

, cps

551729

875643586

iso-PC3(Glu)

PC3

891

Figure 5.6.11. Identification of Cd-induced phytochelatins in plant root extracts (maize) by ES quadrupole MS. (a) MS spectrumof a fraction collected in reversed-phase HPLC – UV of Cd-exposed plant root extracts. (b) MS spectrum after substraction of thecontrol (blank). Reprinted from Phytochemistry , Vol. 56, H. Chassaigne, V. Vacchina, T. M. Kutchan, and M. H. Zenk, p. 657,Copyright (2001), with permission from Elsevier Science.

ES MS spectrum of the chromatographic fractionshown in Figure 5.6.11, several minor peaks areobserved that can be attributed to peptides whichmay be naturally present in plants or to Cd-inducedpeptides. The peaks at m/z 715, 772, 947 and 1004

may correspond to the molecular ions of desGly-PC3, PC3, desGly-PC4 and PC4, respectively. Thespectrum also shows the presence of the peak atm/z 844 that is attributed to the molecular ionof iso-PC3(Glu). The sequence of this last form

374 DETECTION

C C C E

233 362 465 844

380612 483

γE

697594

251 148

130

715

163

251

362

465

715

100 200 300 400 500 600 700 800

m/z

0.4

0.8

1.2

1.6

2.0

2.4

2.8

Inte

nsity

, cps

844

612

483

380233

x 103

γEγE

[M + H]+

Figure 5.6.12. Identification of Cd-induced phytochelatins in plant root extracts (maize) by ES quadrupole MS/MS. MS/MSspectrum of a peak corresponding to a PC-related peptide detected in the MS spectrum at m/z 844 (cf. Figure 5.6.11).The fragmentation pattern allows the identification of the compound, as iso-phytochelatin, iso-PC3(Glu). Reprinted fromPhytochemistry, Vol. 56, H. Chassaigne, V. Vacchina, T. M. Kutchan, and M. H. Zenk, p. 657, Copyright (2001), with permissionof Elsevier Science [55].

was confirmed by fragmenting the molecular ionin the CID mode (Figure 5.6.12). The completeseries of y-type ions was observed and a goodsensitivity was obtained with the triple quadrupoleinstrument [55]. However, ES Q-TOF with highermass accuracy and resolution should represent thefuture in peptide and oligopeptides research andespecially in the speciation field, facilitating recog-nition of unknown compounds in crude or partiallypurified extracts.

4.4 Cadmium complexes withmetallothioneins in animal tissues

Characterization of macromolecules involved inthe sequestration of heavy metals as well as inthe metalloregulation in animals has been attract-ing considerable interest. Metallothioneins (MTs)are cysteine-rich, metal-binding proteins involvedin the detoxification of metals and metabolism of

essential metals, and are found in various organsof animals [56]. Certain MTs exist as different iso-forms coded from multiple genes whose individualexpression can only be determined by fastidioustechniques (RNA assay using specific DNA). Themethodology developed for analysis of translatedproteins is also a challenge since sequence variationbetween MT isoforms encountered in mammalianspecies can vary from one to a few amino acids (in60–62 amino acids).

Figures 5.6.13(a) and (b) show a typicalreversed-phase chromatogram of the rabbit liverMT-2 isoform sample (commercially available)obtained with on-line inductively coupled plasmaMS and ES MS detection, respectively. Theidentification on the basis of the retention time inHPLC is impossible because of the unavailabilityof standards of particular protein isoforms withsufficient purity. The major peak is split in two,


4.0 × 104

×103

3.0

2.0

1.0

0

3

6600 6800

Molecular mass

7000

6000 6200

Molecular mass, u

6400 6000 6200

Molecular mass, u

6400

2

1

×103

3

2

1

1 2

0

×103

3.6

2.4

1.2

0

2

1

1

2

1

3.5

1.4

0.7

0 10 20 30

Retention time, min

(b)

(c)

apo-MT-2a apo-MT-2a

apo-MT-2c

apo-MT-2b

apo-MT-2c

Abs

orba

nce,

254

nm

Inte

nsity

, cps

40 50

2.8

2.1

20 25 30

Time, min

(a)

114Cd63Zn

35 40

1.25 × 105

Cd6Zn-MT-2a

Cd7-MT-2a

Cd7-MT-2b

Cd7-MT-2c

Cd5Zn2-MT-2a

×103

6600 6800

Molecular mass

7000

2

1

2 Cd7-MT-2a

Cd7-MT-2b

Cd7-MT-2cCd6Zn-MT-2a

Cd5Zn2-MT-2aCd4Zn3-MT-2a

1.00

40%CH3OH

20%CH3OH

0.75

0.50

0.25

0

Figure 5.6.13. Characterization of Cd, Zn-metallothionein complexes (MT-2) by reversed-phase HPLC with ES quadrupole MSdetection. (a) HPLC – ICP MS chromatogram with Cd and Zn specific detection of the MT-2 sample. (b) HPLC – ES MSchromatogram under the same chromatographic conditions. Reconstructed mass spectra taken at the maxima of the peaks 1 and2 are shown in the insets. (c) Reconstructed mass spectra at the maxima of the peaks 1 and 2 after acidification of the columneffluent. Reprinted from Trends in Analytical Chemistry, Vol. 19, H. Chassaigne, V. Vacchina, and R. Lobinski, p. 300, copyright(2000), with permission of Elsevier Science [18].

376 DETECTION

which might suggest the existence of two sub-isoforms (which differ by one or a few aminoacids in their composition) of the MT-2 isoform.The mass spectra taken at the maximum of peaks1 and 2 (spectra in the insets to Figure 5.6.13(b))give more information on the identity of the elutedcompounds. Since the separation is carried outat pH 7.0 to ensure the stability of metal–MTcomplexes, several artefacts due to the formationof mixed metal complexes (Cd, Cd–Zn) coexistand make the interpretation of the spectra difficult.

This problem can be solved by a post-columnacidification whose purpose is to remove themetals from the metallothionein complexes andto detect the apo-MT [57]. Figure 5.6.13(c) (afterpost-column acidification) confirms the complexityof the mass spectra of peaks 1 and 2 causedby the presence of different metallo species thateclipse the occurrence of species with differentamino acid composition. It is clearly seen thatthe two major peaks in the spectra belong to theMT-2 and MT-2c sub-isoforms, with respectivemasses of 6126.0 and 6150.0 Da. The differencebetween the mass spectra of peaks of metallo andepo-metallo proteins allows the determination ofthe stoichiometry of the separated compounds (seespectra in the insets to Figure 5.6.13b).

5 CONCLUSION

The main advantages of ES and related ionizationtechniques in elemental speciation analysis arethe following:

• soft ionization method;• suitable for analyzing large biomolecules;• suitable for analyzing polar and even ionic

compounds (e.g. metal complexes);• enables coupling of MS/MS and HPLC.

The latest developments of the technology con-cern the better sensitivity obtained and the lowervulnerability of the ionization source to the matrixcomposition or the eluent composition in HPLC.New pneumatically assisted ES combined with adual orthogonal sampling technique allow con-tamination avoidance and improve the sensitiv-ity obtained in MS. Additionally, the ionization

and desolvatation processes can be supported by aheated probe for analysis at high flow rates andto achieve a high sensitivity. The microelectro-spray and nanoelectrospray techniques extend thecapacities of the ionization technique for the anal-ysis of very small quantities of compounds and fornanoscale LC separations (capillary HPLC).

ES MS/MS allows a precise determination ofthe molecular mass and the elucidation of themolecular structure of chemical species, and forthis reason is turning out to be a necessary tool forthe identification and characterization of unknownmetal- and metalloid-containing compounds. Inthis chapter we have highlighted the performancesand the limitations in terms of sensitivity, massmeasurement precision and resolution of a triplequadrupole mass spectrometer preventing the iden-tification of new species. This study also evaluatesthe potential of an ES Q-TOF mass spectrometerfor the investigation of new compounds in specia-tion analysis. ES TOF MS with higher resolutionand mass accuracy allows the detection of newspecies in a complex mixture without the needfor fastidious purification. When it is used in tan-dem Q-TOF mode, novel information regarding theidentity of compounds is obtained.

6 ACKNOWLEDGEMENTS

H. Chassaigne acknowledges a postdoctoral fel-lowship from the European Commission–JointResearch Centre.

7 REFERENCES

1. Cornelis, R., De Kimpe, J. and Zhang, X., Spectrochim.Acta , 53, 187 (1998).

2. Lobinski, R., Fresenius’ J. Anal. Chem., 369, 113 (2001).3. Bettmer, J., Anal Bioanal Chem , 372, 33 (2002).4. Sanz Medel, A., Spectrochim. Acta , 53, 197 (1998).5. Haskins, W. E., Wang, Z., Watson, C. J., Rostand, R. R.,

Witowski, S. R., Powell, D. H. and Kennedy, R. T.,Anal. Chem., 73, 5005 (2001).

6. Taylor, A., Branch, S., Halls, D. J., Owen, L. M. W. andWhite, M., J. Anal. At. Spectrom., 15, 451 (2000).

7. Taylor, A., Branch, S., Fisher, A., Halls, D. J. andWhite, M., J. Anal. At. Spectrom., 16, 421 (2001).

8. Templeton, D. M., Fresenius’ J. Anal. Chem., 363, 505(2001).

REFERENCES 377

9. Yamashita, M. and Fenn, J. B., J. Phys. Chem., 88, 4451(1984).

10. Fenn, J. B., Mann, M., Meng, C. K., Wong, S. F. andWhitehouse, C. M., Mass Spectrom. Rev., 9, 37 (1990).

11. Kebarle, P. and Tang, L., Anal. Chem., 65, 972A (1993).12. Wilm, M. S. and Mann, M., Int. J. of Mass Spectrom. Ion

Process., 136, 167 (1994).13. Emmett, M. R. and Caprioli, R. M., J. Am. Soc. Mass

Spectrom., 5, 605 (1994).14. Pergantis, S. A., Wangkarn, S., Francesconi, K. A. and

Thomas-Oates, J. E., Anal. Chem., 72(2), 357 (2000).15. Wachs, T., Conboy, J. C., Garcia, F. and Henion, J. D.,

J. Chromatogr. A, 29, 357 (1991).16. Niessen, W. M. A. and Tinke, A. P., J. Chromatogr. A,

703, 357 (1995).17. Whitehouse, C. M., Dreyer, R. N., Yamashita, M. and

Fenn, J. B., Anal. Chem., 57, 675 (1985).18. Chassaigne, H., Vacchina, V. and Lobinski, R., Trends

Anal. Chem., 19, 300 (2000).19. Fenn, J. B., Mann, M., Meng, C. K., Wong, S. F. and

Whitehouse, C. M., Science, 246, 64 (1989).20. Smith, R. D., Loo, J. A., Ogorzalek-Loo, R. R., Bus-

man, M. and Udseth, H. R., Mass Spectrom. Rev., 10, 359(1991).

21. Chassaigne, H. and Lobinski, R., Fresenius’ J. Anal.Chem., 361, 267 (1998).

22. Chassaigne, H. and Lobinski, R., Anal. Chem., 70, 2536(1998).

23. Kellner, R., Mermet, J. M., Otto, M. and Widmer, H. M.,Analytical Chemistry , Wiley-VCH, Weinheim, 1998.

24. McSheehy, S., Szpunar, J., Haldys, V. and Tortajada, J.,J. Anal. At. Spectrom., 17, 507 (2002).

25. McSheehy, S., Marcinek, M., Chassaigne, H. and Szpu-nar, J., Anal. Chim. Acta, 410, 71 (2000).

26. Yates, J. R. III, McCormack, A. L., Link, A. J., Schi-eltz, D., Eng, J. and Hays, L., Analyst , 121, 65R (1996).

27. LeBlanc, J. C. Y., J. Anal. At. Spectrom., 12, 525 (1997).28. Madsen, A. D., Goessler, W., Pedersen, S. N. and Fran-

cesconi, K. A., J. Anal. At. Spectrom., 15, 657 (2000).29. Lindemann, T. and Hintelmann, H., Anal. Bioanal.

Chem., 372, 486 (2002).30. Casiot, C., Vacchina, V., Chassaigne, H., Szpunar, J.,

Potin-Gautier, M. and Lobinski, R., Anal. Commun., 36,77 (1999).

31. McSheehy, S., Pohl, P., Szpunar, J., Potin-Gautier, M.and Lobinski, R., J. Anal. At. Spectrom., 16, 68 (2001).

32. Pickford, R., Miguens-Rodriguez, M., Afzaal, S., Speir, P.,Pergantis, S. A. and Thomas-Oates, J. E., J. Anal. At. Spec-trom., 17, 173 (2002).

33. Clark, L. C., Combs, G. F., Turnbull, S. W., Slate, E. H.,Chalker, D. K., Chow, J., Davis, L. S., Glover, R. A.,

Graham, G. F. and Gross, E. G., J. Am. Med. Ass., 276,1957 (1996).

34. Ip, C., Birringer, M., Block, E., Kotrebai, M., Tyson,J. F., Uden, P. C. and Lisk, D. J., J. Agric. Food Chem.,48, 2062 (2000).

35. Neve, J., J. Cardiovasc. Risk , 3, 42 (1996).36. Kotrebai, M., Birringer, M., Tyson, J. F., Block, E. and

Uden, P. C., Anal. Commun., 36, 249 (1999).37. Templeton, D. M., Ariese, F., Cornelis, R., Danielsson,

L. G., Muntau, H., Van Leeuwen, H. P. and Lobinski, R.,Pure Appl. Chem., 72, 1453 (2000).

38. Chassaigne, H., Chery, C. C., Bordin, G. and Rodriguez,A. R., J. Chromatogr. A, 976, 409 (2002).

39. McSheehy, S., Pannier, F., Szpunar, J., Potin-Gautier, M.and Lobinski, R., Analyst , 127, 223 (2002).

40. Larsen, E. H., Hansen, M., Fan, T. and Vahl, M., J. Anal.At. Spectrom., 16, 1403 (2001).

41. Casiot, C., Szpunar, J., Lobinski, R. and Potin-Gautier, M.,J. Anal. At. Spectrom., 14, 645 (1999).

42. Shibata, Y., Morita, M. and Fuwa, K., Adv. Biophys., 28,31 (1992).

43. Francesconi, K. A. and Edmonds, J. S., Oceanogr. Mar.Biol. Annu. Rev., 31, 111 (1993).

44. Le, X. C., Cullen, W. R. and Reimer, K. J., Clin. Chem.,40, 617 (1994).

45. Francesconi, K. A., Edmonds, J. S. and Stick, R. V., J.Chem. Soc. Perkin Trans., 1, 1349 (1992).

46. Corr, J. J. and Larsen, E., J. Anal. At. Spectrom., 11, 1215(1996).

47. McSheehy, S. and Szpunar, J., J. Anal. At. Spectrom., 15,79 (2000).

48. McSheehy, S., Pohl, P., Lobinski, R. and Szpunar, J.,Analyst , 126, 1055 (2001).

49. Zenk, M. H., Gene, 179, 21 (1996).50. Gekeler, W., Grill, E., Winnacker, E. L. and Zenk, M.

H., Z. Naturforsch., 44, 361 (1989).51. Rauser, W. E., Plant Physiol., 109, 1141 (1995).52. Grill, E., Winnacker, E. L. and Zenk, M. H., FEBS Lett.,

197, 115 (1986).53. Meuwly, P., Thibault, P. and Rauser, W. E., FEBS Lett.,

336, 472 (1993).54. Meuwly, P., Thibault, P., Schwan, A. L. and Rauser,

W. E., Plant J., 7, 391 (1995).55. Chassaigne, H., Vacchina, V., Kutchan, T. M. and Zenk,

M. H., Phytochemistry , 56, 657 (2001).56. Stillman, M. J., Shaw, C. F. and Suzuki, K. T., Metal-

lothionein Synthesis, Structure and Properties of Metal-lothioneins, Phytochelatins and Metalthiolate Complexes ,VCH, New York, 1992.

57. Chassaigne, H. and Lobinski, R., J. Chromatogr. A, 829,127 (1998).

5.7 Elemental Speciation by Inductively CoupledPlasma-Mass Spectrometry with High ResolutionInstruments

R. S. HoukIowa State University, Ames, IA, USA

1 Introduction . . . . . . . . . . . . . . . . . . . . . . 3792 Mass Analysis with Magnetic and

Electrostatic Sectors . . . . . . . . . . . . . . . . 3792.1 Mass analysis in radial magnetic

field . . . . . . . . . . . . . . . . . . . . . . . . 3792.2 Kinetic energy analysis in radial

electric field . . . . . . . . . . . . . . . . . . 3802.3 Focusing properties of electrostatic

and magnetic fields . . . . . . . . . . . . . 3812.4 Double focusing principle . . . . . . . . 3822.5 Slit widths and resolution . . . . . . . . 383

3 Use of Magnetic Sectors with Ions froman ICP . . . . . . . . . . . . . . . . . . . . . . . . . . 3873.1 Acceleration of ions . . . . . . . . . . . . 3873.2 Effect of load coil configuration . . . 3883.3 ‘Cool’ plasma . . . . . . . . . . . . . . . . . 3893.4 Space charge effects . . . . . . . . . . . . 3903.5 Beam shaping with quadrupole or

hexapole lenses . . . . . . . . . . . . . . . 3923.6 Collision cells . . . . . . . . . . . . . . . . 392

4 Examples of Scanning, High ResolutionSector Instruments . . . . . . . . . . . . . . . . . 3934.1 Finnigan Element . . . . . . . . . . . . . . 3934.2 VG Axiom . . . . . . . . . . . . . . . . . . . 393

5 Examples of Multicollector Instruments 3965.1 VG Axiom with multicollector array 3965.2 Micromass Isoprobe . . . . . . . . . . . . 3965.3 NU Plasma . . . . . . . . . . . . . . . . . . . 3965.4 Finnigan Neptune . . . . . . . . . . . . . . 3965.5 Mattauch–Herzog instrument 399

6 Speciation Measurements with SectorInstruments . . . . . . . . . . . . . . . . . . . . . . . 3996.1 Dry sample introduction . . . . . . . . . 3996.2 Micronebulizers . . . . . . . . . . . . . . . 4026.3 Gradient elution . . . . . . . . . . . . . . . 4026.4 Scan speed and data acquisition

issues . . . . . . . . . . . . . . . . . . . . . . . 4036.5 Blanks . . . . . . . . . . . . . . . . . . . . . . 405

7 Capabilities and RepresentativeApplications with LC Separations . . . . . . 4067.1 High resolution and accurate mass

measurements . . . . . . . . . . . . . . . . . 4067.2 High sensitivity . . . . . . . . . . . . . . . 4067.3 Selected applications . . . . . . . . . . . . 407

8 Other Separation Techniques . . . . . . . . . . 4118.1 CE separations . . . . . . . . . . . . . . . . 4118.2 Gel electrophoresis . . . . . . . . . . . . . 4118.3 GC separations . . . . . . . . . . . . . . . . 412

9 High Resolution with Mass AnalyzersOther Than Magnetic Sectors . . . . . . . . . 4139.1 High resolution quadrupoles . . . . . . 4139.2 Fourier transform ion cyclotron

resonance . . . . . . . . . . . . . . . . . . . . 4139.3 Time-of-flight MS . . . . . . . . . . . . . . 413

10 High Resolution Measurements withPlasmas Other Than the ICP . . . . . . . . . . 413

11 Conclusion . . . . . . . . . . . . . . . . . . . . . . . 41412 Acknowledgements . . . . . . . . . . . . . . . . . 41413 References . . . . . . . . . . . . . . . . . . . . . . . 414

MASS ANALYSIS WITH MAGNETIC AND ELECTROSTATIC SECTORS 379

1 INTRODUCTION

This chapter provides information pertinent to useof high resolution mass spectrometers with an ICPsource. The use of these instruments for analysis ofbiological materials in general has been reviewedrecently [1]. The magnetic sector is by far the mostwidely used instrument for high resolution mea-surements with the ICP, and most of the chapterdeals with these devices. The terms ‘high reso-lution,’ ‘sector field,’ and ‘magnetic sector’ havebecome synonymous in the ICP lexicon, althoughthere are other ways to achieve high resolution, andsectors are used at low resolution where possible.Compared to quadrupole instruments, scanningmagnetic sector instruments offer the followingadvantages: (a) higher sensitivity1 at unit mass res-olution, (b) sufficient resolution to separate manychemically different ions at the same nominal m/z

value, e.g., 56Fe+ (m/z = 55.9349) from 40Ar16O+(m/z = 55.9567)2, (c) the ability to identify an ionconclusively by accurate m/z measurements3, and(d) lower instrument background (1 count s−1 orless). Whether the higher sensitivity and lowerbackground translate into better detection limitsdepends on the blank signals for the element(s)of interest. In the author’s opinion, the ability toconclusively identify the atomic analyte ion basedon high resolution and accurate m/z measurementsmakes sector instruments particularly attractive forspeciation measurements. The potential for betterdetection limits is also invaluable, because the ana-lyte compounds are typically diluted during thechemical separation.

The other general type of magnetic sectorinstrument, the multicollector, is also discussedin this chapter. These instruments are capableof highly precise isotope ratio measurements.

1 Sensitivity refers to analyte signal per unit concentration, i.e., theslope of a calibration curve. The minimum detectable amount of analyteis more properly referred to as the detection limit.

2 Throughout this chapter, the m/z values cited for a singly chargedion have had the mass of one electron removed from the sum of themasses of the various atoms.

3 The differences in the binding energies of the nuclei 56Fe, 40Arand 16O and the numbers of protons, neutrons and electrons in eachatom lead to slightly different m/z values for 56Fe+ and 40Ar16O+.Tables of accurate mass values are available from various handbooksand manufacturers.

Usually, the magnetic field is fixed, and ionsof different m/z ratios are measured in separatedetectors. Noise from the ICP is cancelled whenthe ratio of signals is measured, so these devicesshould provide high ratio precision for analytes inthe transient chromatographic peaks observed inspeciation studies [2, 3].

Thus, both general types of magnetic sec-tor instrument have potential uses in speciation.A concise description of the operating princi-ples of magnetic sector instruments is given,followed by a summary of their value andlimitations when used with on-line chemical sep-arations, the usual mode of providing specia-tion information with ICP-MS. Particular ways toenhance the performance of sector instruments,such as solvent removal and reduction of theplasma potential, and representative applicationsare also summarized.

2 MASS ANALYSIS WITH MAGNETICAND ELECTROSTATIC SECTORS

Much of the discussion in this section follows thatin Roboz [4]. More advanced treatments can befound in other references [5, 6].

2.1 Mass analysis in radial magnetic field

Figure 5.7.1 shows a positive ion of mass m andcharge z moving with velocity v in a vacuumthrough a uniform magnetic field of strength B.The magnetic field is oriented perpendicular to theplane of the paper; one magnet pole piece is abovethe plane of the figure, the other pole piece isbelow. Movement of the ion charge through themagnetic field generates a magnetic force Fm thatacts in a direction perpendicular to the directionof v. The magnetic force thus deflects the ion ina curved path. As the ion moves further throughthe magnetic field, the orientation of v changes,and Fm always acts perpendicular to the directionof v at each position in the field. Thus, the ioncontinues to move in a curved path as long as itis in the magnetic field. The magnitude of the ionvelocity remains the same, but the direction of v

380 ELEMENTAL SPECIATION BY ICP-MS

+

+

+

V

V

VFm

Fm

Fm

B

Figure 5.7.1. Deflection of positive ion moving at velocity vthrough a region of uniform magnetic field, strength B. Themagnetic field lines are oriented into the plane of the page. Themagnetic force Fm acts perpendicular to the direction of v.

changes, so the magnetic interaction does induceacceleration and represents a force in that sense.

The ion undergoes uniform circular motionsuch that the radius of curvature rm balances themagnetic force at the applied field strength

Fm = Bzev = mv2/rm (5.7.1)

To induce the ion to travel from the sourceinto the magnetic field, it is usually acceleratedthrough a potential difference V , which gives it akinetic energy

KE = 12mv2 = zev (5.7.2)

where e = electron charge. Note the importantdifference between capital V for acceleratingvoltage and lower case v for velocity.

These two equations can be combined to givethe following general form for ion motion in amagnetic field

m/z = B2r2m/2V (5.7.3)

where the constant e has been combined with z

and will be accounted for shortly.Some comment about units is also in order. The

standard unit for B is Tesla (T); 1 T = 104 G =108 lines of force m−2 = 1 weber m−2, where G =gauss. The symbol B is sometimes also referredto as magnetic flux density. Some older literatureuses the symbol H instead of B.

A convenient reduced form of equation (5.7.3)is as follows

m/z = 4.83 × 10−5B2r2m/V (5.7.4)

where m/z is measured relative to 12C = 12.00000. . ., z = +1, +2, −1, −2 . . ., rm is in cm, B isin gauss, and V is in volts. A singly chargedion (z = 1) at m = 100 in a magnetic field B =103 gauss with V = 2000 V moves with a radiusof curvature of 64 cm. To separate and detect suchan ion, the flight tube through the magnetic fieldis constructed with this value of rm.

Equations (5.7.2) and (5.7.3) also show thatthere are several ways to obtain a spectrum:

(1) Scan B while holding rm and V constant.(2) Scan V while holding rm and B constant.(3) Keep B and V constant and detect ions of

various m/z values at different values of rm

with a position-sensitive detector.

Generally, a particular instrument designed tochange m/z value by either methods (1) or (2) isused for scanning applications, while one meant tomeasure different ions at different radial positionsis referred to as a multicollector. Equation (5.7.3)also shows that the m/z scale is not linear in eitherB, V , or rm. Peaks for heavier ions are spacedmore closely. The instrument software usuallydisplays the spectrum as if the m/z scale werelinear, however.

2.2 Kinetic energy analysis in radialelectric field

Equations (5.7.2) and (5.7.3) show that a spreadof ion kinetic energy (reflected in a spread ofV values) causes a variety of m/z values to beobserved at particular values of B and rm. Toimprove mass resolution, a method for selectingions with a narrow spread of kinetic energy isoften used. A simplified schematic diagram of suchan electrostatic analyzer is given in Figure 5.7.2.Again, an ion is accelerated through potentialdifference V into the gap between two curvedmetal plates. The electric field between the platesis V ′/d , where d is the plate spacing. The ionexperiences a radial electrostatic force Fe that isoriented toward the center of curvature of theplates. This force deflects the ion in a curved path


FeFe

vv

re

fe

+V ′/2

−V ′/2

+

++

Figure 5.7.2. Deflection of ion moving inside a radial electro-static condenser, gap thickness d, radius of curvature re, angleof deflection φe, radial electric field E = V ′/d. The electricforce Fe acts perpendicular to the direction of ion velocity v.

between the sector plates. Again, the condition foruniform circular motion is

Fe = mv2/re = zeE (5.7.5)

where E = radial electric field = V ′/d . Thisexpression can be combined with equation (5.7.2)for the ion kinetic energy to derive the followinggeneral expression

re = 2V/E (5.7.6)

Thus, ions of a given kinetic energy V (actuallyzeV ) can be selected by adjusting E (i.e., theapplied voltages V ′). Mass does not appear inthis expression, so an electrostatic analyzer suchas that in Figure 5.7.2 does not provide a massspectrum. It is typically used in combination witha magnetic analyzer to select ions of the desiredrange of kinetic energies to improve the resolution.Alternatively, a multipole collision cell can beused to reduce the spread of kinetic energy bycollisional energy transfer, as discussed in moredetail below.

2.3 Focusing properties of electrostaticand magnetic fields

Electrostatic and magnetic analyzers can tolerateion beams of a finite width or angular divergenceand still provide good performance. These devices

rm

fm

B

lm′ lm″

2a

(a)

Object Image

(−)Imagele′

le″Radius reAngle fe

(+)

Object

(b)

Figure 5.7.3. Focusing effect of magnetic sector (a, subscriptm) and electrostatic sector (b, subscript e). The image distanceis l′ and the focused image is formed at l′′. See alsoequations (5.7.7)–(5.7.12).

have focusing properties illustrated in Figure 5.7.3.For either the electrostatic analyzer (subscript e,Figure 5.7.3(b)) or the magnetic analyzer (sub-script m, Figure 5.7.3(a)), ions injected directlyon center travel along the path dictated by theequations described above. An ion injected abovethis central path travels a longer distance in thefield and is deflected slightly more, whereas anion injected below the central path is deflected lessextensively. Such off-center ions can be focused toa point outside the field at positions l′′e or l′′m. Thesefocal positions are the usual positions for entranceor exit slits. In the simple cases (perpendicular ionentry, source, center of curvature and image allon same plane for magnetic analyzer) shown inFigure 5.7.3, their locations can be predicted fromthe following expressions:

f 2e = (l′e − ge)(l

′′e − ge) (5.7.7)

fe = re√2 sin(

√2φe)

(5.7.8)

ge = re√2 tan(

√2φe)

(5.7.9)


f 2m = (l′m − gm)(l′′m − gm) (5.7.10)

fm = rm/ sin φm (5.7.11)

gm = rm/ tan φm (5.7.12)

The focal properties of more complex geometriesand corrections for the fringe fields (especially thatof the magnet) can be determined [7].

These considerations govern the selection ofparameters used in construction of the instrument,such as angle of deflection φ, radius of curvaturer and slit positions l′ and l′′. For example, theintermediate slit between the electrostatic andmagnetic analyzer is located at a position thatprovides a focused image for both, i.e., at theobject position l′′ for the first analyzer and theimage position l′ for the second.

2.4 Double focusing principle

The focusing properties of both the electrostaticand magnetic field can be used in combinationto provide very narrow ion beams (i.e., highspectral resolution) without a disastrous loss oftransmission. Velocity focusing occurs when ionsof the same m/z value but different kineticenergy recombine at the same point. Curve g ofFigure 5.7.4 represents the set of such velocityfocusing points plotted together into a velocityfocusing curve for ions of various m/z values. Atthe same time the broadening caused by the initialangular divergence in the ion beam is minimum atcertain points on the direction focusing curve r.

The condition for double focusing is that thevelocity dispersion suffered by the ions in the

A ′

A′′e1, A′1

A′′e, A′

B

dN1

M1

M2

A′′1

A′′2

g r

Figure 5.7.4. Double focusing mass analyzer. For ions of various m/z values, the direction focusing curve is r, where the angularaberration due to the finite angular width of the beam is minimum. The velocity focusing curve is g, where ions of the same m/zvalue but different kinetic energy dispersed by the electrostatic analyzer recombine. For the particular settings of acceleratingvoltage and magnetic field strength, ions of m/z value M1 are separated from others with both velocity and direction focusingat point A′′

1. Reproduced from ref. 4 with permission from Springer-Verlag.


electrostatic analyzer is exactly compensated bythe magnetic field, while the direction-focusing ofboth fields and the mass dispersion of the magneticfield are maintained. For the instrument shown inFigure 5.7.4, this double focusing condition is metfor ions of m/z = M1 at point A′′

1, where the twofocusing curves intersect.

At the value of B that brings the ion beamfor M1 to the double focusing point, the lighterion M2 has a smaller radius of curvature rm andleaves the magnetic sector at a position wherethe direction focusing curve diverges from thevelocity focusing curve. The ion M2 thus producesa broader image because it is not observed at thesingle position where the direction and velocityfocusing compensate precisely. The values of V

and/or B could be adjusted to bring the paths forion M2 to the double focusing point A′′

1, which isone way to scan the m/z value transmitted by suchinstruments.

2.5 Slit widths and resolution

Resolution R is defined in the usual way as R =m/�m, the m/z value measured divided by theseparable mass difference. It can be defined withvarious values for the allowable valley betweenpeaks, relative to the peak heights, i.e., 5 % valley,10 % valley, 50 % valley, etc. The ratio m/�m issometimes also called resolving power.

The resolution necessary to separate a givenpair of ions can be estimated as follows. Supposewe wish to resolve 75As+ (m/z = 74.9210) from40Ar35Cl+ (m/z = 74.9307), a common problemion. The resolution required is

R = m/�m ∼ 75/(74.9307 − 74.9210) = 7800

If the peaks were perfect triangles of equalheight, they would be separated to baseline at thisresolution setting. If ArCl+ is much more abundantthan As+, substantially better resolution would beneeded to account for tailing of the peaks. Notealso that the atomic ion is the lightest one atm/z = 75, which is usually the case at m/z valuesbelow m/z ∼ 100 [8].

The resolution is affected by factors such asthe widths of the slits and ion beam, kineticenergy spread of the ion beam, and aberrations, i.e.,imperfections due to fringe fields and other causes.Roboz [4] gives several general equations thatillustrate the relation between these parameters:

R = 1

(S1 + S2)/rm + �V/V= rm

S1 + S2 + βrm

(5.7.13)

where �V/V is the kinetic energy spread �V

relative to the accelerating voltage V and β is anumber that describes the magnitude of aberration.This equation shows that the resolution increases atsmaller slit widths, smaller kinetic energy spread,and larger accelerating voltage.

For a double focusing instrument, it can beshown that the resolution is roughly related to theradius of curvature of the electrostatic analyzer [4]:

R ∼ are/S (5.7.14)

where the entrance and exit slits are of equalwidth S and the proportionality constant a isapproximately 1–2. Thus, an instrument with anelectrostatic analyzer of radius 10 cm with 10 µmslits would be expected to have a resolution ofthe order of 10 000, roughly what is achieved withICP-MS instruments.

The relation between slit width and peak shapeis also important. If the slits are wide open(Figure 5.7.5 top), the beam is narrower than theslit. The resulting peak has an extended flat section,corresponding to transmission of the entire beamthrough the last analyzer and exit slit to thedetector. Such a peak shape provides the bestprecision, as the signal at the top of the peakis not strongly affected by small variations inaccelerating voltage or magnetic field strength. It isalso straightforward to hop back and forth betweenthe centers of such flat-topped peaks, in the samefashion that the voltages applied to a quadrupolecan be varied to hop between peaks.

To achieve higher resolution, the slits aremade narrower than the inherent beam width(Figure 5.7.5 bottom). Ideally, the peaks would betriangular, but there is typically some curvature on


Low Resolution

High Resolution

Ion Image

Collector Slit

Collected BeamIntensity

Ion Image

Collector Slit

Collected BeamIntensity

Figure 5.7.5. Relation between beam size, slit width and peak shape at low and high resolution settings. Courtesy of ThermoElemental.

the sides. Naturally, both peak height and area aresacrificed to achieve higher resolution. It is alsomore difficult to control the accelerating voltageand magnetic field strength with sufficient accuracyto hop directly onto the sharp tip of a peak inhigh resolution. Traditionally, these instrumentsare scanned when operated in high resolution.Recent improvements in magnet stability allow thewidth of the scanned region to be only a narrow

strip in the center of the peak, so that little time isspent elsewhere except near the center of the peak.

Examples of peaks actually measured on aFinnigan Element are shown in Figure 5.7.6. Notethe different sensitivity scales. For a given elementand isotope, the peak height decreases roughlytenfold when the resolution is changed from 300to 4000 (slit widths ∼20 µm), followed by anothertenfold loss from 4000 to 8000. The slits are only


(a)

(b)

64Zn+

66Zn+

67Zn+

68Zn+

70Zn+

67Zn+

Figure 5.7.6. Actual peaks from Finnigan Element. (a) Zn+, low resolution, R ∼ 300, 10 µg L−1 Zn. (b) Expanded view of67Zn+ showing flat top of peak. (c) Separation of V+ from ClO+, medium resolution, R ∼ 4000, 10 µg L−1V. (d) Separation ofAs+ from ArCl+, high resolution, R ∼ 10 000, 10 ppb As in 1 % aqueous HCl. The accurate m/z value for 75As+ is 74.9216,very close to the peak centroid. (e) Separation of 78Se+ from 38Ar40Ar+, high resolution, R ∼ 10 000, 10 µg L−1 Se.


R = 10,000

75As+

40Ar35Cl+

R = 4,000

35Cl16O+

51V+

(c)

(d)


USE OF MAGNETIC SECTORS WITH IONS FROM AN ICP 387

(e)

R = 10,000

38Ar40Ar+

78Se+


10 µm wide at the high resolution setting, whichplaces stringent requirements on their manufacture,mounting and physical condition and on thefocusing conditions needed to put the ion beamonto the slit.

3 USE OF MAGNETIC SECTORSWITH IONS FROM AN ICP

3.1 Acceleration of ions

With a quadrupole instrument, the sampler andskimmer are typically grounded. The ions leavethe skimmer with the kinetic energy gainedfrom entrainment in the flow of argon, plus acontribution from the plasma potential. The kineticenergy possessed by an ion is given roughlyby the plasma potential plus the difference involtage between the region where it was formedand the region of interest [9]. For quadrupoles,

the ion kinetic energy need be only a fewelectronvolts, whereas magnetic sectors performbest when the ions are accelerated to severalthousand electronvolts. The method by which theions are accelerated is therefore important.

In the initial ICP magnetic sector experiment byBradshaw et al. [10], the ions were accelerated byapplication of a high positive voltage to both thesampler and skimmer (Figure 5.7.7(a)). The flighttube through the analyzers is then kept grounded.Naturally, the sampler and skimmer must be iso-lated electrically from the rest of the vacuum sys-tem. Some care is also necessary to prevent arcingfrom the high voltage interface through the vacuumtubing to the interface pump, which is typicallyoperated at or near the interface voltage using iso-lation transformers. Several present manufacturersuse this approach.

The reader may wonder why positive ions arenot simply repelled by the high positive voltage(+4 to +8 kV) applied to the sampler and skimmer.


+5 kV

Sampler

ICP

Skimmer Ionoptics

TO MS

Floatingflight tube

(a)

−5 kV

−5 kV

Sampler

ICP

Skimmer Ionoptics

TO MS

Floatingflight tube

(b)

Figure 5.7.7. Voltage arrangements for extracting ions into magnetic sector mass analyzer. (a) Sampler and skimmer float atpositive accelerating voltage, while flight tube is grounded. (b) Sampler and skimmer are grounded while flight tube floats belowground. Reprinted from Spectrochimica Acta, Part B, Vol. 51, Niu and Houk, pp. 779–815, Copyright (1996), with permissionfrom Elsevier Science.

There are several reasons. A space charge sheathforms around the inner edge of the sampler andskimmer, which shields the plasma from the poten-tial applied to the cones. Some say the plasmapotential also floats up near the accelerating volt-age. Ions pass through the sampler and skimmermostly by virtue of the gas flow rather than theapplied potentials.

An alternate way to accelerate the ions isto leave the sampler and skimmer groundedand apply a negative voltage to the flight tube(Figure 5.7.7(b)). This is the approach takenin the Finnigan Element and Neptune instru-ments [11, 12]. Finnigan claims that the high volt-age can be switched more rapidly by applying it to

the flight tube than to the interface, which permitsfaster peak switching with the grounded interface.

3.2 Effect of load coil configuration

Present magnetic sector ICP-MS instruments use aload coil that is grounded at the downstream end(Figure 5.7.8). On some instruments, a C-shapedmetal shield (also called a guard electrode orplasma screen) can be inserted between the coiland the outer tube of the torch (Figure 5.7.8). Theshield removes the capacitive coupling between theload coil and the plasma and reduces the plasmapotential [9, 13]. It is usually grounded.


Metal shieldinserted betweencoil and torch

YO &Y (I) Emission

Y (II)Emission

± HV

Figure 5.7.8. Shielded ICP torch. The usual initial radiationzone from red YO and Y(I) emission and blue analytical zonefrom Y(II) emission are also shown.

There are various reports on the effects ofthe shield with sector instruments. On FinniganElement devices, which use the grounded interface(Figure 5.7.7(b)), grounding the shield improvesthe sensitivity roughly tenfold. Appelblad et al.[14] report that operating with the shield inplace but floating allows selection of a value forthe aerosol gas flow rate that minimizes matrixeffects by mimicking the ‘cross-over point’ seenin ICP emission spectrometry. Some magneticsector instruments that use the high voltageinterface (Figure 5.7.7(a)) do not report sensitivityimprovements using the shield [15].

3.3 ‘Cool’ plasma

On any of the sector instruments, grounding theshield allows use of so-called ‘cool’ plasma con-ditions. The usual adjustable plasma conditionsare power, aerosol gas flow rate, and samplingposition. At a particular sampling position andpower, a plot of M+ ion signal versus aerosolgas flow rate generally gives a pyramidal shape(Figure 5.7.9) [16]. A similar shaped plot (normal-ized to the same maximum) displaced to higheraerosol gas flow rate is seen for the generally unde-sirable MO+ ions.

The variation of signal with these operatingparameters is interrelated. At higher power, theaerosol gas flow rate that generates maximum M+sensitivity is displaced to lower aerosol gas flowrate. If the plasma is retracted further away from

Nor

m. i

on s

igna

l

M+

Aer. gas flow rate

MO+

Figure 5.7.9. Plots of ion signal versus aerosol gas flow rateshowing different zones of abundance for M+ and MO+. Bothplots have been normalized to the same height; the maximumsignal for MO+ is usually much lower than that for M+,depending on the element and the solvent loading. Adaptedfrom ref. 16 with permission.

the sampler, lower power and/or higher aerosol gasflow rate are required to maximize sensitivity.

The ‘zone model’ [16, 17] has been describedto explain these effects. Figure 5.7.8 also showsthe zones characteristic of the various speciesformed as the sample travels through the ICP.If a concentrated solution containing yttrium (orsome other element whose MO species emit inthe visible) is introduced into the plasma, theplume comprising emission from MO and neutralM atoms is often called the initial radiation zone(IRZ). The IRZ protrudes further downstreamas power decreases and aerosol gas flow rateincreases. This can be thought of as ‘pushing’the MO and M species further downstream in theplasma until they are heated sufficiently to becomeatomized and ionized.

At the power, aerosol gas flow rate, and sam-pling position where M+ sensitivity is maximum,the sampler is typically just 1 to 2 mm downstreamfrom the tip of the IRZ. Physically, the plume ofM+ ions expands rapidly, and only ions from asmall cross-section of the plasma just in front ofthe sampler traverse both sampler and skimmer [9].These conditions yield best M+ sensitivity for allinstruments. The background spectrum under suchconditions is dominated by atomic ions O+ andAr+ and ions with a few protons like H2O+ and


ArH+. However, if the power is reduced and/oraerosol gas flow rate is increased, with the shieldgrounded, the background spectrum changes to onedominated by molecular ions such as H3O+, NO+and O2

+. This is the ‘cool’ plasma condition [18].Atomic background ions are much less prevalent,and so are their reaction products ArH+, ArN+,ArO+ and Ar2

+. The changes to plasma conditionsthat induce the ‘cool’ plasma would also causean increase in plasma potential and secondary dis-charge, so the ‘cool’ plasma can be generated onlyif the shield is in place and grounded or the voltageapplied to the load coil is inherently balanced.

The primary use of the ‘cool’ plasma is toremove the worst of the polyatomic ions thatplague measurement of K+, Ca+, and Fe+. Thereare also reports that operation under ‘cool’ con-ditions decreases atomic ion background, espe-cially for volatile elements such as Li, Na, andPb. Although the traditional use of the ‘cool’plasma has been to allow measurement of suchelements with quadrupole instruments, the highresolution capability of scanning magnetic sectorsis of further value. For example, measurementson a Finnigan Element show low levels of ‘new’background ions such as (H17

3 O+)(H162 O)2 and

(H2D16O+)(H162 O)2 at m/z = 56 (Figure 5.7.10).

The protons (m = 1.0078) make these ions easyto resolve from 56Fe+ (m/z = 55.9349, slightlyunder 56). These background ions are probablyalso present in quadrupole instruments and areoften ascribed to ‘Fe blank’. The ability to sep-arate the interferences conclusively, identify ionsdefinitively by accurate m/z measurements, andmeasure the true blank due to atomic ions are bigadvantages of magnetic sector instruments.

There are several drawbacks to the ‘cool’plasma. Elements such as Ce and W that formstrongly bound oxides are seen only as MO+ orMO2

+ ions, and elements with first ionizationenergies much above that of NO (9.25 eV) are notionized efficiently in ‘cool’ mode, unless they areseen as oxides. The latter category includes Asand Se, two key elements in speciation. Matrixeffects are generally more severe in ‘cool’ mode.There have been relatively few publications on the

‘cool’ plasma for speciation with magnetic sectorinstruments [19].

3.4 Space charge effects

In the plasma, the charge on the positive ionsis balanced by an equal number of electrons.The extracted ion beam remains quasineutral untilit leaves the skimmer. Preferential loss of thelight, highly mobile electrons results in a beam ofpositive ions leaving the skimmer with total currentof the order of 1 mA. The maximum current Imax

of ions at mass-to-charge ratio m/z that can befocused through a lens of length L and diameterD is given by

Imax ∼ 0.9(z/m)1/2(D/L)2V 3/2 (5.7.15)

where V is the ion kinetic energy inside the lens(volts), m/z is given relative to 12C = 12, and Imax

is in µA [20]. It is important to note that m/z isthe value for the major ion comprising the bulkof the ion beam, usually considered to be 40 forAr+ in the ICP. For example, it is not correct tosay that equation (5.7.15) shows that space chargeeffects are less severe for heavy analytes. Spacecharge effects (i.e., defocusing and loss of ions)can still be significant at current values less thanthis calculated maximum.

For a quadrupole instrument, V ∼ 200 V or less,for which Imax is 100 µA or less. Since this valueis well under the expected ion current of ∼1 mA,low voltage extraction optics are susceptible tothe space charge effect. For a sector instrumentwith accelerating voltage of 4000 and 8000 V,Imax is 9 mA and 25 mA, respectively, higher thanthe expected ion current through the skimmer.Thus, space charge effects are likely to be lesssevere in sector instruments. Whether they aretotally absent or unimportant is another matter.Space charge defocusing can still be significantat currents well below the limit calculated fromequation (5.7.14). The effects are exacerbatedif the beam is focused to a fine image, andsharp focusing is much more important in sectorinstruments than in quadrupoles. Certainly massbias and matrix effects are still present with sector


39K+

H3O+•H218O

600 W, 1.8 L/min14 mm sample depth

(a)

(b)

600 W

R = 4000

(H318O)+(H2

16O)(H216O)

(H317O)+(H2

16O)(H216O),

(H2D16O)+(H216O)(H2

16O)

56Fe 57Fe

Figure 5.7.10. Background spectra at m/z = 39, 56 and 57 in cool plasma mode, Finnigan Element, R = 4000. Note watercluster ions and their separation from 39K+, 56Fe+ and 57Fe+. Data provided by D. R. Wiederin, Elemental Scientific Inc.


instruments, although evidence has been presentedthat they have rather different fundamental causesthan the effects seen with low voltage, quadrupoleinstruments [21]. The entire matter of preciselyhow a beam is formed from the plasma emanatingout of the skimmer is still an issue for basicstudy [22].

3.5 Beam shaping with quadrupoleor hexapole lenses

The beam leaving the skimmer has a circularcross-section, while the entrance and exit slitsare rectangular. Ions in the outer sections of thecircular beam will be lost unless the shape of thebeam is changed. This can be done with the set ofquadrupole lenses shown in Figure 5.7.11. Thereare two sets of four metal rods with DC voltagesapplied as shown. The positive voltages on thehorizontal poles squeeze the ions in that direction,while the negative voltages allow the beam toexpand in the vertical direction. A combination oftwo or more of such lenses can produce the desiredfocusing properties in both dimensions [23].

Hexapole lenses operate on the same principle,except that there are six poles, of course. Note

that these quadrupole or hexapole lenses areDC devices; do not confuse them with the RFmultipole collision cells described below.

3.6 Collision cells

In recent years, collision cells4 have been suppliedwith quadrupole ICP-MS instruments for eitherremoval of interfering ions or conversion of M+analyte ions to more easily measured species. Oxi-dation of As+ to AsO+ to avoid ArCl+ interferenceis an important example of the latter process [24].The operating principles for these devices havebeen described [25, 26]. Ions pass through an RFquadrupole, hexapole, or octopole that contains acollision gas. The pressure is sufficient that the ionsundergo several collisions while inside the cell. Achemically unreactive gas (e.g., He) can be usedto thermalize the ions, and/or a reactive one (suchas H2 or NH3) can be used to induce a desiredchemical reaction. The term ‘chemical resolution’

4 Some workers distinguish between ‘collision cells’ that are not massanalyzers and use relatively lower pressure, fewer collisions and havemore likelihood of endothermic reactions from ‘reaction cells’ that arequadrupole mass analyzers and use higher pressure and exothermicion–molecule reactions. The present discussion does not distinguishbetween these various devices.

+ +

−

−

ENT.slit

Skimmer

Figure 5.7.11. DC quadrupole lens used between skimmer and electrostatic or magnetic sector. The beam leaving the skimmeris round, while the applied voltages ‘squeeze’ the beam into a rectangular shape that matches the slit.

EXAMPLES OF SCANNING, HIGH RESOLUTION SECTOR INSTRUMENTS 393

is sometimes used to describe the higher selectivityimparted by the collision cell. Some means of dis-tinguishing desired analyte ions from unwantedproduct ions generated in the ion–molecule reac-tion is usually necessary.

These same principles can be used with mag-netic sector analyzers. Micromass also uses an RFhexapole collision cell to reduce the kinetic energyspread of ions, which eliminates the need for anelectrostatic analyzer, at least for their low res-olution multicollector device (Figure 5.7.12). Foreach elastic, nonreactive collision with a stationaryneutral of mass m2, an ion of mass m1 with initialkinetic energy E1 loses energy to E′

1 according toequation (5.7.16) [27–29]:

E′1 = E1

[m2

1 + m22

(m1 + m2)2

](5.7.16)

These collisions also scatter ions, but the RFfield applied to the hexapole rods helps containthem, so transmission and sensitivity do not suffertoo much. Polyatomic ions have a larger cross-section for kinetic energy loss than atomic ions atthe same nominal m/z value. Yamada et al. [30]exploit this property to show that subsequentkinetic energy analysis can be used to discriminateagainst polyatomic ions. Bandura et al. [31] andVanhaecke et al. [32] show that such collisions canalso be used to reduce high-frequency noise in theion signal from fluctuations in the plasma. Thisimproves the precision in isotope ratios measuredwith quadrupole instruments, and should also helpimprove precision in scanning sector devices. Ofcourse, the same collisional processes used withquadrupole instruments to remove interferences,convert the analyte M+ ion to one more readilymeasured, or reduce noise from the ICP could alsobe implemented with sector devices.

4 EXAMPLES OF SCANNING, HIGHRESOLUTION SECTORINSTRUMENTS

It is not feasible to describe all the presentlyavailable devices in detail. Several of the more

widely used models are discussed to illustrate theworkings of these instruments.

4.1 Finnigan Element

This instrument is based on the work of Giessman,Jakubowski and associates [11, 12]. Of the variousmagnetic sector instruments, this is the only onethe author has used personally. As mentionedpreviously, this system uses the grounded interface(Figure 5.7.7). After the skimmer, the ions areaccelerated through a graphite electrode at highnegative potential. The flight tube is kept belowground potential by whatever voltage is neededto transmit the ions of interest, up to 8 kV.There are quadrupole lenses for beam shapingand focusing. The sectors are arranged in areverse Nier–Johnson geometry (Figure 5.7.13).The 60◦ magnetic analyzer is followed by the90◦ electrostatic analyzer, exit slit, and detector.Thus, the ions are mass analyzed in the magneticsector first, then ions of a selected band ofkinetic energies are transmitted through the exitslit to the detector. There are three fixed slitsettings for low resolution (R = 300), mediumresolution (R = 4000), and high resolution (R =8000 or better). The slit settings can be changedquickly during an acquisition, so medium or highresolution can be used only at m/z values wherethey are needed.

It is faster to scan m/z by changing acceleratingvoltage (V in equation 5.7.3), so the magneticfield is usually kept at the desired value whileions in a certain range are measured. This canbe done by electrostatic peak switching at lowresolution or electrostatic scanning at mediumor high resolution. The magnetic field is thenchanged to a value in the next desired m/z

range, and these ions are measured by changingaccelerating voltage.

4.2 VG Axiom

This instrument is descended from the original ICPsector MS work of Bradshaw et al. [10], which


ICP Torch

Hexapole

Interface Close-offValve

Beam Focus& Accelerator

Extended Multicollectorwith Motorised Detector

Positioning

FlightTube

Turbomolecularpumping system

Fast, LaminatedMagnet

Schematic of the IsoProbe

(a)

Hexapole

Collision ChamberPump

AccelerationChamber Pump

To magneticmass spectrometer

IntermediatePump

ExpansionVolume Pump

Zone ofsilence

ICPtorch

Shock waveArgonplasma

Sampling cone(sampler)

Skimmer cone

Hexapole collisioncell

Flight tubeLOS valve

AccelerationLenses

(b)

Figure 5.7.12. Micromass Isoprobe multicollector instrument with hexapole collision cell used to reduce kinetic energy spreadon ion beam. (a) overall instrument; (b) detail of interface and collision cell. Courtesy of Micromass.

began interest in the concept. Marriott et al. [33]have described the Axiom. The high voltageinterface (Figure 5.7.7(a)) is used. The ion opticalsystem is shown in Figure 5.7.14. The electrostaticanalyzer is first, which is called forward geometry.

Scanning measurements are done with a singleelectron multiplier at the double focusing positionafter the magnet. The slit widths can be selectedbetween a wide variety of values, so the resolutionis continuously selectable between 300 and 12 000.

EXAMPLES OF MULTICOLLECTOR INSTRUMENTS 395

ICP

ELECTROSTATICANALYZER

MAGNETICSECTOR

Figure 5.7.13. Thermo Finnigan Element instrument, reverse Nier-Johnson geometry. Courtesy of Thermo Elemental.

Figure 5.7.14. Ion optics of Axiom instrument from Thermo Elemental. The ions pass from the source at upper right throughan electrostatic analyzer first, followed by the magnetic sector. The exit optics are such that either a single electron multiplier oran array of Faraday cups can be used at the detector plane with good focusing. Reproduced from ref. 33 with permission of theRoyal Society of Chemistry.


5 EXAMPLES OFMULTICOLLECTOR INSTRUMENTS

For high precision isotope ratio measurements ofmetallic elements for geochemical studies, the bestpossible precision is desired, so thermal ionizationMS is the norm. Sample throughput would behigher, multielement studies would be possible,and certain problem elements could be measuredmore readily if highly precise isotope ratios couldbe obtained by ICP-MS.

Scanning instruments measure the various iso-topes of interest sequentially. The ion beam fromthe ICP fluctuates due to effects such as passageof large droplets and oscillation of the plasma ataudio frequencies [34, 35]. Fast scanning or peakhopping helps compensate for the effects of thisnoise when ratios are measured, and isotope ratioscan be measured with precision of ∼0.02 % rel-ative standard deviation (RSD) on scanning sec-tor instruments. However, nearly all of this noiseis eliminated if ion signals at different m/z val-ues can be measured truly simultaneously. Walderet al. [36] did the first multicollector measure-ments on ions from an ICP, using an ICP extractioninterface and beam shaping optics added onto aspectrometer designed primarily for thermal ion-ization. These studies demonstrated that ICP-MScould attain similar precision to thermal ionizationand led to the development of present ICP multi-collector devices.

5.1 VG Axiom with multicollector array

The electrostatic analyzer must come first witha multicollector. The ion optical system of theAxiom is designed such that ions of a substantialrange remain in reasonable focus outside the m/z

value corresponding to perfect double focusing(mass M1 in Figure 5.7.4). The two focal curves gand r lie nearly together on either side of point A′′

1,so ions of a range of masses of about 10 % of M1

are reasonably well focused at the same values ofB and V . Thus, this instrument can also be usedin multicollector mode. The electron multiplier isreplaced by a set of separate Faraday cup detectors(like those shown in Figure 5.7.12(a)). Ions at

up to eight adjacent m/z values are measuredtruly simultaneously with this arrangement. Thereis a central slit for an electron multiplier formeasurement of low abundance isotopes.

Such simultaneous measurements yield isotoperatios whose precisions are, in the short term atleast, limited only by counting statistics and thestability of the mass bias corrections. The responseof a Faraday cup is very stable, but it providesno amplification like that of an electron multiplier,so measurements with multicollectors require moreanalyte than do scanning instruments.

5.2 Micromass Isoprobe

As shown in Figure 5.7.12, this instrument doesnot have an electrostatic analyzer. A collision cellis used to ‘thermalize’ ions, i.e., to reduce theirkinetic energies and kinetic energy spread, beforethey enter the magnetic sector. Instead of isolatingions of a range of kinetic energies and rejectingthe others, the operating concept of the collisioncell is to utilize a larger fraction of the ions thanis the case with an electrostatic sector.

A spectrum for the various isotopes of Tland Pb is shown in Figure 5.7.15. Note the flattopped, highly stable peaks, one of the desirablecharacteristics of these multicollector instruments.This device is also capable of resolution of ∼4000.

5.3 NU Plasma

This company provides a relatively compact devicecapable of resolution of ∼500 (Figure 5.7.16).The zoom lens adjusts the separation between ionbeams so that the spacing between collectors neednot be changed mechanically when the magnetmass is changed. In addition to the usual Faradaycup detectors, ions at up to three m/z values canbe monitored using electron multipliers with anion deflector assembly. A larger version capableof high mass resolution is also available.

5.4 Finnigan Neptune

This instrument (Figure 5.7.17) has the ICP, inter-face and ion lenses from an Element added onto an

EXAMPLES OF MULTICOLLECTOR INSTRUMENTS 397

Figure 5.7.15. Scan of mass spectral peaks for Tl+ (m/z = 203 and 205) and Pb+ (m/z = 204, 206, 207, 208) from MicromassIsoprobe. Reproduced courtesy of Micromass.

100 10−4 10−8 10−9 Pressure in mBar

ROTARYPUMP

ICP SOURCEnu plasma

SCHEMATIC

MonitorPlate

ESA

MAGNET

VACUUM SYSTEM

ELECTRONICS

SOFTWARE

ZOOM LENS

VARIABLEDISPERSION

COLLECTOR ARRAY

Lens 1 Lens 2 Lens 3

TURBO PUMPS

Figure 5.7.16. NU Plasma multicollector instrument with zoom lens and multiple electron multiplier detectors. Figure providedcourtesy of NU Instruments.


ESA

NEPTUNE

Figure 5.7.17. Neptune multicollector instrument. Note the magnification provided by having the image distance roughly twicethe object distance, which facilitates operation of the multicollector array. Figure provided courtesy of Thermo Elemental.

Flat top peaks: high mass resolution

55.974

∆M/M = 100 ppm 56Fe

40Ar16O

0

0.2

0.4

0.6

55.984 55.994 56.004 56.014

mass (u)

Inte

nsity

(V

)

56.024 56.034 56.044

High resolution entrance slit

High resolution exit slit

Resolution definition:R(10% valley) = 8000

(triangular peak shape)

Figure 5.7.18. High resolution separation of 56Fe+ from 40Ar16O+ in multicollector mode. Figure provided by Thermo Elemental.

electrostatic and magnetic analyzer from the Tritoninstrument for thermal ionization. The large massanalyzer has a dispersion of 81 cm, and a zoomlens is used. This device can provide flat topped

peaks at a resolution 3000 for simultaneous multi-collection at up to three m/z values, as shown for56Fe+ in Figure 5.7.18. Resolution up to 8000 canbe achieved with sharply pointed peaks.

SPECIATION MEASUREMENTS WITH SECTOR INSTRUMENTS 399

Magnetic shunt

To monitoring coulomb meter

To monitoring ammeter

Monitor slit cover

Suppressor

b-Slit

Main slit

rerm (min.)

rm (max.)fe

fm

Ground

−

+

Earth slit

Accelerating slit

Ionization chamber

Ion source Electrostatic field Magnetic field

Energy selectorEntrance slitBeam aperture (a-Slit)

Beam height aperture

Detector system

Slit andcollector unit

Ion beam (II)

Ion beam (I)

Figure 5.7.19. Mattauch–Herzog geometry. The ion beam leaves the electrostatic analyzer as a parallel beam. All m/z valueswithin the upper m/z limit of the device come back into focus along a single plane near the boundary of the magnetic field.Reproduced from ref. 4 with permission from Jeol.

5.5 Mattauch–Herzog instrument

The multicollector instruments described previ-ously are meant to monitor ions at a set of eightto ten adjacent m/z values. The Mattauch–Herzoggeometry can measure all m/z values at once. Inthis geometry (Figure 5.7.19), φe is 31.83◦. Thefocusing equations (5.7.7)–(5.7.9) can be solvedfor this angle to show that the ions leave the elec-trostatic analyzer as a parallel beam (l′′e →∼). Themagnetic sector satisfies the equation

sin φm = √2 sin

√2φe (5.7.15)

The ion path is an S shape; the direction ofdeflection in the electrostatic analyzer is oppositeto that in the magnetic sector [4].

Under these conditions, the direction and veloc-ity focusing curves are straight lines that coincide.Thus, ions of the entire m/z range achieve dou-ble focusing along the same plane, as shown inFigure 5.7.19. This arrangement is well suited toa planar, array detector and was used with photo-graphic plates in the old spark source MS. An ICP-MS instrument of this sort with an electrical arraydetector would be capable of truly simultaneousmeasurement of the entire m/z range, provided the

detector is large enough to observe the whole rangewith sufficient dispersion. Work along these lineshas been described by Cromwell et al. [37, 38] andis proceeding under the direction of Hieftje andDenton [39–41].

6 SPECIATION MEASUREMENTSWITH SECTOR INSTRUMENTS

6.1 Dry sample introduction

The composition of the solvent that feeds thenebulizers is a matter of concern in ICP-MS. Avariety of schemes for solvent removal can beimplemented with any ICP-MS instrument:

(1) cooled spray chamber (Figure 5.7.20).(2) traditional desolvation [42] (Figure 5.7.21(a)),

i.e., heating the aerosol followed by cooling ator near the freezing point of the solvent.

(3) additional solvent removal by membrane orcryogenic desolvation (Figures 21(b) and (c)).

There are obvious advantages to removing sol-vent for ICP-MS, many of which are of particularadvantage in high resolution measurements. Otherfactors such as temperature remaining the same,


(b)

(a)

CoolantDrain

Aerosol out

Figure 5.7.20. Cooled spray chambers for solvent removal: (a) cooled double pass Scott chamber; (b) Cyclone chamber, sideand top views. In both chambers, most of the large droplets are deposited at the bends, while fine droplets pass out to the plasma.

lower levels of polyatomic ions containing oxygenatoms would be expected from simple mass bal-ance considerations,

MO+ −−−⇀↽−−− M+ + O

assuming most of the O atoms come from the sol-vent. The magnitude of the atomic mass defectvaries with mass [8], and some important poly-atomic ion interferences are just within the resolu-tion capabilities of the instruments. For example,separation of 140Ce16O+ from 156Gd+ requires

nominal resolution of at least 7200, with theaccompanying large sacrifice in sensitivity. Thus,using plasma conditions that reduce the abundanceof MO+ is useful. Removal of organic solventsminimizes carbon deposition and helps maintain aproperly hot plasma, and solvent removal obviatesnoise from passage of wet droplets through theplasma. In certain cases, removing solvent leadsto enhancements in sensitivity with sector instru-ments, which is welcome even if the basic reasonsare not clear. In general, the plasma potential is


To ICP Conventional desolvator

Condenser

Drain

Heating tube

USN

(a)

(b)

Heating/coolingcoils

(Cu loops)

Secondcryocondenser

(glass)

Cryogenic desolvator

140 °C

−80 °C(absoluteethanol)

To ICPDryaerosol

fromconventionaldesolvator

(c)

Membrane desolvatorTo ICP

Ar

Dry aerosol fromconventional desolvator

Ar

Figure 5.7.21. Additional solvent removal options: (a) conventional desolvation with heated chamber and condenser at or near thefreezing point of the solvent; (b) cryogenic desolvation. The aerosol undergoes repeated heating and cooling steps. (c) Membranedesolvation. Reproduced from ref. 42 with permission of the Royal Society of Chemistry.


lower when the plasma is dry, so ions from a dryplasma would be expected to have lower spreads ofkinetic energy, which could improve transmissionand sensitivity.

Generic disadvantages of solvent removal areband broadening, longer rinse out time, and worsememory effects, especially for volatile elementssuch as Hg, B, Os and I. Volatile species can alsobe lost during desolvation. Such memory or lossproblems are generally less severe if the dropletsare not heated to dryness.

First we discuss solvent removal options withconventional nebulizers that operate at liquidflow rates of 0.5 mL min−1 or more and havedroplet transport rates to the plasma of ∼1–2 %.These devices normally require spray chambers toremove the very large droplets and to keep thetotal solvent load below the maximum the plasmacan tolerate. The Scott-type double pass chamber(Figure 5.7.20(a)) [43] remains common, althoughthe cyclone chamber (Figure 5.7.20(b)) [44] showsmoderately faster rinse out behavior and is alsoseeing extensive use recently.

Either of these spray chambers can be cooledto temperatures near the freezing point of thesolvent. The aerosol leaving the spray chambercan also be heated and then either cooled againor passed through a membrane desolvator. Thesample constituents are injected into the plasma asdry particles with some solvent vapor remaining. Ifthe aerosol is to be heated anyway, a conical spraychamber that allows higher initial droplet transportcan be tolerated. The higher droplet productionrate and transport from the spray chamber are theprimary reasons behind the sensitivity advantageof ultrasonic nebulizers [45].

It must be admitted that solvent removal beyondsimply cooling the spray chamber is not common-place in speciation measurements, with either sec-tor or quadrupole instruments, largely because ofthe perceived problem of dead volume after theaerosol is produced. In the author’s experience,it is much more important to minimize dead vol-ume in the liquid connection between column andnebulizer. For well-behaved species, i.e., those notoverly volatile or prone to memory effects, whetherthe additional dead volume and rinse out time is a

problem depends on the time duration of the peaksprovided by the chemical separation. Species thatare volatile can be readily lost or suffer long mem-ory effects when their aerosols are heated.

On-line hydride generation is another option forsolvent removal for elements that form volatilehydrides (e.g., As, Se, Pb, Sn) or for Hg. It isimportant to verify that the various chemical formsare indeed converted to hydrides with reproducible,hopefully complete, recovery. On-line microwavedigestion just before the hydride generation cellhas been used for this purpose [46].

6.2 Micronebulizers

These devices operate at low liquid flow rates,100 µL min−1 or less for this discussion. Aerosol istransported more efficiently out of the spray cham-ber with micronebulizers, which compensates forthe lower liquid uptake. Thus, sensitivity can beas good as or better than that obtained from con-ventional nebulizers. It is also easier to removesolvent; indeed, for water at liquid flow rates of40 µL min−1 or less, the solvent evaporates nat-urally anyway [47]. At lower uptake rates, theaerosol can be fed directly into a membrane des-olvator, without the intermediate condenser. Sincethe total solvent load can be lower, the plasma cantolerate higher concentrations of organic solvent,and there should be less variation of sensitivity assolvent composition changes during gradient elu-tion. The trend in LC separations is toward lowerflow columns anyway, the outlets of which can befed to micronebulizers with less extensive split-ting. Those micronebulizers with very low liquiduptake rates (e.g., 20 µL min−1 for the PFA-20,Elemental Scientific, Inc., or even 5 µL min−1 forthe cross-flow device recently described by Tsun-oda et al. [48]) are readily adapted to capillaryelectrophoresis (CE) separations since it is easy tocompensate for their low natural suction. For thesereasons, use of micronebulizers for speciation isexpected to increase in the near future.

6.3 Gradient elution

As with other ICP-MS devices, the sensitivitygenerally varies with the solvent composition,


mainly because the characteristics of the aerosolchange with solvent composition. Changing thefraction of organic solvent present also results information of polyatomic ions from the solvent;the sector instrument has the desirable featurethat many of these polyatomic ions can beseparated from M+ ions using high resolution.Nevertheless, these effects complicate the use ofsolvent gradients in LC.

Browner and coworkers [49] use an oscillatingcapillary nebulizer (OCN) for LC measurementswith solvent gradients. The droplets are producedby vibration of a capillary tip, not by the usualpneumatic process. The droplets are producedmechanically, and the droplet characteristics arenot greatly affected by the solvent composition.

Other measures that facilitate use of gradientelution are as follows:

(1) Restrict the range of solvent compositions used.(2) Measure the change in sensitivity during the

gradient by adding analyte to the solvents,then apply correction factors at the appropriateretention times when analyzing samples.

(3) Add another solvent stream post-column thatis the inverse of the stream through the col-umn [50]. For example, if a gradient from10 % to 30 % methanol–water goes throughthe column, mix in a stream whose composi-tion goes from 90 % to 70 % methanol–water,then nebulize the mixed stream. Of course,a second gradient pump is required. Dilu-tion and post-column band broadening arealso expected.

(4) Continuous introduction of an internal stan-dard post-column, followed by measurementof the ratio of analyte signal to that forthe internal standard. As demonstrated byHeumann et al. [51] and Garcia Alonso et al.[52], the internal standard can be an enrichedminor isotope of the analyte, the signal fromwhich will be affected by changes in solventcomposition to the same degree as that for theanalyte. The m/z windows measured for bothanalyte and internal standard must either befree of polyatomic ions, or the polyatomic ionbackground must remain constant. Again, thehigh resolution capability helps separate these

interferences and facilitates accurate measure-ments using organic solvents or other addi-tives, as demonstrated below.

(5) Remove solvent (Figures 5.7.20 and 5.7.21)and/or add oxygen to the aerosol gas flow tominimize carbon deposition on the samplerand skimmer. Of course, extra oxygen willenhance the abundance of oxide ions.

6.4 Scan speed and dataacquisition issues

As with any scanning instrument, the rate of scan-ning or hopping from peak to peak is importantwhen monitoring a transient signal. First we con-sider single collector instruments that operate witha fixed value of rm. As shown in equation (5.7.3),the m/z value can be changed by changing eitherthe magnetic field B or the accelerating voltage V .The magnet takes longer to stabilize upon changethan does the accelerating voltage, and the accel-erating voltage must be at least a certain value toensure proper extraction and transmission of ions.Thus, measurements are commonly made in oneparticular m/z range of interest by changing V

while keeping B constant. The magnet setting isthen changed to a different value for the next set ofanalytes, with more measurements made by chang-ing V , and so on.

For the flat-topped peaks seen in low resolution(Figures 5.7.6(a) and (b)), the beam is muchnarrower than the slit, so it is relatively easyto change B and/or V and still have the beampass through the slit. Thus, the B and V valuescan be adjusted to hop directly onto the m/z

value of interest with little dead time, evenwhen the magnetic field is changed. The ‘deadtime’ involved in such hops is of the orderof a few milliseconds, comparable to that of aquadrupole MS.

With the sharply pointed peaks seen in mediumor high resolution (Figure 5.7.6(b) and (c)), it ismuch harder to ‘find’ the peak top reliably, so them/z value is usually scanned, even when the intentis to monitor only one m/z value. It then becomesnecessary to select experimental parameters tooptimize data collection. The use of such scans


M+ ANALYTE PEAK20 samples/peak

Interference

m/zMasswindow

Figure 5.7.22. Schematic diagram of medium or high resolution scan showing data acquisition parameters. Each vertical linerepresents one m/z position, i.e., a slightly different setting of accelerating voltage at a given magnetic field setting.

in high resolution does limit the number of m/z

values that can be monitored with a desired signal-to-noise ratio during a transient signal.

A description of a typical set of scan parametersis in order here. The terminology is taken fromthat used for the Finnigan Element; other deviceshave different jargon but follow the same concepts.Consider the schematic high resolution mass scanshown in Figure 5.7.22. The mass window is thepercentage of the nominal peak width chosenin which to acquire data; a value of 100 %could be used to encompass only 56Fe+ and not40Ar16O+. The samples per peak value gives thenumber of settings of accelerating voltage V perpeak, represented by the 20 lines for M+ inFigure 5.7.22. Sample time is the amount of timespent at each value of V , i.e., the time the massanalyzer spends on each line shown. This value is

5 ms or longer, with small values necessary forfast transients or to monitor more elements perchromatographic peak. There is also a settling timeneeded for the m/z value to stabilize, typically1 ms if only V is changed, 300 ms if the magnetsetting is changed. Thus, each line shown inFigure 5.7.22 would be measured in 6 ms, for atotal time of 120 ms spent measuring just the Fe+peak each time through the sequence of peaksselected. On new Element instruments, the masswindow can be selected to be as little as 10 % ofthe peak width reproducibly in medium or highresolution. All the signals for each line, or that fora chosen fraction of the lines, within the specifiedmass window can be added together when thechromatogram is plotted.

In the author’s experience, there are twomeasures that can be taken to speed up data


acquisition: avoid changing the slit width duringeach cycle (this imposes a 1 s delay), and minimizethe number of magnetic field settings wherepossible. On the Element, at least six m/z valuesat one or two B settings can be monitored onchromatographic peaks lasting 10 s with little lossof peak definition.

Multicollector instruments are generally justkept at a fixed magnetic field setting, as the usualobjective is the best possible precision. It has beendemonstrated recently that these devices are capa-ble of high precision during GC peaks [2, 3]. Withthe recent advent of zoom optics for the collec-tors, as described above, ions from various m/z

ranges can be imaged onto the detectors withoutlarge-scale mechanical adjustments of the spacing.Mattauch–Herzog instruments with array detectorsshould be able to monitor as many m/z values asdesired, if the dispersion is enough to separate thepeaks sufficiently onto a detector of practical size.

6.5 Blanks

The high sensitivity of magnetic sector ICP-MSdevices does not automatically translate into betterdetection limits. The blank level, i.e., the signalfrom the element of interest in the solvent, oftenrestricts the detection limit. In most experiments,small peaks are observable for all the elementsin the blank. The organic solvents and organicbuffer additives are particularly prone to suchcontamination.

The usual clean sample handling proceduresneed to be followed in speciation measurementsas well. One key point is whether the contaminantelement binds to the molecules in the sample.If not, the baseline in the chromatogram merelybecomes elevated, as shown for Cr in humanserum in Figure 5.7.23 [53]. The Cr+ in thisbaseline can be subtracted from that which is inthe chromatographic peaks, so the Cr bound to

140 kDa

Time (s)

Cou

nt r

ate

(c/s

)

0

0

200

400

600

800

1000

1200

1400

1600

1800

100 200 300 400 500

10 kDa

Figure 5.7.23. Chromatogram for Cr in human serum eluted from size exclusion column. The numbers give the estimatedmolecular weights of the proteins based on calibration with protein standards. Reproduced from ref. 53 with permission fromthe American Chemical Society.


proteins can still be estimated. Of course, a highbackground level also carries more noise with it,and much of the Cr apparently bound to proteinsmay actually come from contamination, which ishard to rule out completely.

In LC experiments, some measures that helpreduce trace element contamination are as follows:

(1) Conventional columns and pumps containstainless steel exposed to the eluent, so usemetal-free pumps and columns lined with glassor some other non-metallic material.

(2) Sanz-Medel and coworkers [1, 54, 55] use asmall scavenger column between the pump andinjector. The scavenger column is packed withChelex 100 or ion exchange resin and removesmetals from the eluent stream just before it isused for the separation.

(3) In the author’s experience, Teflon tubing iscleaner than PEEK.

(4) Clean containers are particularly importantin many biological applications. One wayto minimize contamination is to use largesample containers so that the ratio (samplevolume/wall area of container) is relativelylarge, which minimizes the contribution ofthe element coming from or going into thecontainer wall. This is often not an optionfor the small volumes available for enzyme orprotein solutions.

(5) Some instruments are installed completely intoclean room facilities. Others find it useful tokeep the sample handling area enclosed with apositive pressure of filtered air to exclude dust.

7 CAPABILITIESAND REPRESENTATIVEAPPLICATIONS WITH LCSEPARATIONS

This discussion is meant to illustrate the capabili-ties of sector instruments for speciation measure-ments. It is not a full review of all such studies.

7.1 High resolution and accuratemass measurements

As stated previously, the ability to separate theatomic ion of interest from polyatomic ions at

the same nominal m/z value provides very highconfidence that the ions measured are the onesdesired. An example is shown in Figure 5.7.24.The objective is to measure chromium speciationin serum by size exclusion chromatography withICP-MS. Although there is a substantial Cr+background, additional Cr from the sample elutesin two chromatographic peaks. The spectral scan inFigure 5.7.25 shows that only a small percentageof the total signal at m/z = 52 is from 52Cr+.The measured m/z value for 52Cr+ agrees wellwith the expected value (m/z = 51.9405). Thisidentification is simplified by the fact that, in thism/z range, the atomic ion is the lightest one likelyin each particular m/z window [8].

Most of the remaining ions in Figure 5.7.24 are40Ar12C+ (hereafter just referred to as ArC+), andthere is a hint of yet a third ion on the low masstail of ArC+. A low resolution measurement wouldascribe much too high a value for chromium con-centration in this chromatographic peak. Applica-tion of a correction for ArC+ based on that comingfrom the solvent would improve the situation butstill ignores the possibility that the carbon from theeluting protein could increase the amount of C+ inthe plasma and the corresponding level of ArC+seen in the spectrum. Use of a solvent gradientwith an organic modifier is another reason why theArC+ signal can change during the chromatogram.In the author’s opinion, spectral separation of theM+ analyte ion from the polyatomic ions is themost general solution to such overlap problems,especially when there are two or more polyatomicions at the nominal m/z value of interest.

7.2 High sensitivity

In low resolution mode the sensitivity from a sectorinstrument is approximately 100 times higherthan that obtained using quadrupole instruments.Of course, the sensitivity for the element ofinterest in the blank is also higher, so thedetection limits are not automatically 100 timesbetter. This sensitivity enhancement is of particularvalue in cases where the blank is low and/orwhere the sample is diluted extensively duringthe chromatographic separation. An example is

CAPABILITIES AND APPLICATIONS WITH LC SEPARATIONS 407

0

20000

40000

60000

80000

100000

120000

140000

160000

180000

200000

220000

240000

260000

SIG

NA

L

51.91 51.92 51.93 51.94 51.95 51.96 51.97 51.98 51.99

m /z

0

10

20

30

40

50

60

70

80

90

100

52Cr+

40Ar 12C+

Cr MASS SCAN1ST CHROM. PEAK

Figure 5.7.24. Mass scan during first chromatographic peak from Figure 5.7.23 showing spectral resolution of 52Cr+ from40Ar12C+. Reproduced from ref. 53 with permission from the American Chemical Society.

shown in Figure 5.7.25. Uranium and thoriumcan be readily observed at ambient levels inserum without prior chemical preconcentration,even though the separation dilutes the analyte by afactor of ∼30. The retention times show that theseelements are bound to proteins. Even if the samplehad been contaminated, the additional uranium andthorium are still attached to proteins [53].

7.3 Selected applications

The data shown in Figures 5.7.23–5.7.25 illus-trate results obtained in the author’s group withsize exclusion separations and a magnetic sectorICP-MS device. Other such studies include mul-tielement speciation in serum [56] and liver [57].Harrington et al. [58] report the use of this sep-aration method for quantification of Fe speciesin meat. Size exclusion chromatography has sev-eral desirable features: separations can be done atphysiological pH, organic solvents that would be

200 300 400 500Time (s)

0

500

1000

1500

2000

2500

3000

3500

Cou

nt r

ate

(c/s

)

630 kDa

80 kDa50 kDa

20 kDa

Th+

U+

Figure 5.7.25. Selected ion chromatogram for U and Th inhuman serum, low spectral resolution. The concentrations areestimated to be approximately 1 ng L−1 U and 3 ng L−1 Th.Reproduced from ref. 53 with permission from the AmericanChemical Society.

expected to unfold or denature proteins are usu-ally not used, and the relation between retentiontime and molecular weight can be determined bycalibration with known proteins. Thus, at least the


molecular weights of unknown proteins can beestimated without isolated standards of the samecompounds. Compared to other chromatographicmethods, size exclusion generally provides poorerchromatographic resolution, however.

Sanz-Medel and coworkers have used fast pro-tein LC separations with a magnetic sector instru-ment to measure speciation of several elements,especially Al in serum [54, 55]. A resolution of4000 is sufficient to separate 27Al+, the only Alisotope, from polyatomic ions at m/z = 27. Alu-minum in transferrin measured by LC-ICP-MSyields the same twin chromatographic peaks asa transferrin standard, which shows that the bulkof the Al in serum is bound to transferrin. Thisgroup also studied other elements using the samechromatographic method [59] and measured metaldistribution patterns in mussels using size exclu-sion separations [60].

Jakubowski et al. [61, 62] use a reversed phase,isocratic separation with a hydraulic high pres-sure nebulizer [63] and solvent removal to measureselenium-containing amino acids, selenomethion-ine, selenoethionine, selenocystamine, and seleno-cystine. Performance at either low or high massresolution was compared. As expected, low reso-lution gave better Se+ sensitivity, but backgroundions from the organic solvent were found at highresolution. Examples were 12C6H5

+ at m/z = 77and 12C4H16

2 O2+ and 12C4H16

6 O+ at m/z = 82. Thelatter m/z position is often thought to be ‘clean’for Se. Fortunately, the H atoms (m = 1.0078) dis-place the masses of these organic ions well abovethose of the atomic Se+ isotopes, and these poly-atomic ions were readily separated from the cor-responding Se+ isotopes at R = 1400. Detectionlimits were 2–4 pg at low resolution and 0.4–2 pgat R = 1400.

Jakubowski’s group [64] then applied thisLC-ICP-MS method to the measurement of Secompounds in herring gull eggs. A Se-selectivechromatogram obtained after injection of suchan extract is shown in Figure 5.7.26. The topframe shows the original chromatogram from a1 : 5 diluted extract. At least six separate Se com-pounds are apparent. There is little problem mon-itoring Se in these samples; the extracts contain

a total of about 57 µg L−1 Se. The chromatogramin the bottom frame was obtained from a dilutedextract spiked with four standard compounds at2 µg L−1 Se each. These spikes elute with fourof the peaks seen in the original chromatogram,but compounds 3, 4, 5, and 6 remain unidenti-fied. The problem of identification of the speciesresponsible for chromatographic peaks that do notelute with standard compounds remains endemicto ICP-MS. In its usual mode of operation, theICP destroys structural information about the com-pounds. Jakubowski et al. [65] have also investi-gated Pt speciation in plants using size exclusionchromatography. Sulfur was also measured to indi-cate when general proteins elute.

Measurement of phosphorylation of proteinsand peptides is another important potential applica-tion of LC-ICP-MS. The sensitivity of molecularmass spectrometric methods differs for differentcompounds, and the sensitivity generally decreasesas the extent of phosphorylation increases. Thesemolecular methods can be used to determinemolecular weight and sequence, and this informa-tion can then be combined with measurement of31P+ and 32S+ to both quantify the protein anddetermine the extent of phosphorylation. Interfer-ences such as NOH+ at m/z = 31 can be separatedat R = 4000.

Wind et al. [66, 67] recently reported such mea-surements. Capillary LC (flow rate ∼4 µL min−1)was interfaced to a microflow nebulizer (PFA100) and a Finnigan Element. The low liquidflow rate reduces changes in sensitivity with elu-ent composition, but these effects still occur. Touse gradient elution, the 31P+ and 32S+ sen-sitivity was measured while continuously intro-ducing standard phosphate and cysteine in theeluent during a gradient (Figures 5.7.27(a) and(b)). The ratio of P+/S+ sensitivity measured forthe standard mixture at each retention time wasused to correct the signals for each element dur-ing subsequent separation of proteins or peptidesusing the same solvent gradient (Figure 5.7.27(c)).The measured P/S ratio could then be comparedwith the known or assumed ratio for caseins andfor peptides with known phosphorylation, withgood agreement. Detection limits were ∼0.1 pmol.

CAPABILITIES AND APPLICATIONS WITH LC SEPARATIONS 409

160000

140000

120000

100000

80000

60000

40000

20000

0

0 60 120 180 240

(a)

300 360 420 480

160000

140000

120000

100000

80000

60000

40000

20000

0

0 60 120 180 240

Time/s

(b)

Inte

nsity

/cou

nts

s−1

300 360 420 480

#1

#1

#2

#3

#4

#5 #6

#2

#3

#4

#5 #6S

elen

ocys

tine

Sel

enoc

ysta

min

e

Sel

enom

ethi

onin

e

Sel

enoe

thio

nine

Figure 5.7.26. Selected ion chromatogram measured at R = 400 for 82Se in (a) extracts from herring gull eggs diluted 1 : 5, and(b) diluted extract spiked with standard organoselenium compounds at 2 µg L−1. Reproduced from ref. 64 with permission of theRoyal Society of Chemistry.

This methodology will prove very useful in stud-ies of post-translational modification of proteinsand peptides.

LC-ICP-MS measurements are also valuable foranalysis of DNA. Wang et al. [53] found that var-ious metal cations bound to DNA fragments, eventoxic metals such as Pb and Cd (Figure 5.7.28).Chromium(III) and chromate also bound to DNA,which suggested that the CrO4

2− anion was first

reduced to a Cr cation, probably Cr3+. The car-cinogenic role of chromate is related to its abilityto oxidize DNA, either directly or through interme-diates. Jakubowski’s group [68] quantified DNAadducts with styrene oxide using LC-ICP-MS incombination with electrospray ionization. The ICP-MS monitored 31P+ at R = 3000. The P+ sensitiv-ity was the same for P from DNA as for inorganicphosphate, so standards of the actual adducts were


60000

50000

40000

30000

20000

10000

0

0 5 10 15 20 25

Time, min

(a)

Inte

nsity

, cps

30 35 40 45

2.5

2.0

1.5

1.0

0.5

0.0

0 5 10 15 20 25

Time, min

(b)

P/S

rat

io

30 35 40 45

100

80

60

40

20

0

0 5 10 15 20 25

Time, min

(c)

% In

tens

ity

30 35 40 45

32S

31P

32S

31P

Figure 5.7.27. Determination of P/S ratio during LC separation. (a) Variation of sensitivity with eluent composition duringgradient elution. (b) P/S signal ratio measured from (a). (c) Elution of α-casein with correction derived from (b). Reproducedfrom ref. 67 with permission from the American Chemical Society.

OTHER SEPARATION TECHNIQUES 411

200 300 400 500

Time (s)

15000

10000

5000

0 0

1000

2000

3000

4000

5000

Cd+ C

OU

NT

RA

TE

(c

/s)

Pb+ C

OU

NT

RA

TE

(c

/s)

53 kDa

6 kDa

38 kDa

21 kDa

Pb+

Cd+

(a)

Figure 5.7.28. Selective ion chromatograms for 208Pb and114Cd in mixture of DNA restriction fragments separated bysize exclusion chromatography. The baseline at m/z = 208represents Pb eluting from the column, tubing, etc. AdditionalPb and Cd elutes with the DNA fragments of the approximatemolecular weights shown. Reproduced from ref. 53 withpermission from the American Chemical Society.

not necessary, which is a general advantage ofatomic spectroscopic methods. The detection limitwas 20 pg, or 14 modifications in 108 bases, whichis comparable with that of other methods for detec-tion of DNA damage.

Finally, Hassellov et al. [69] used field flowfractionation with ICP-MS to study trace metaladsorption onto colloidal particulates. Suspendedparticles of various sizes were separated andnebulized into a Finnigan Element. The majorelements Si, Al, minor elements Fe and Mn, andadsorbed trace elements Cs, La and Pb could befollowed into various particle size fractions.

8 OTHER SEPARATIONTECHNIQUES

8.1 CE separations

Another potential niche for the high sensitivityprovided by sector instruments is for CE, where theabsolute amount of sample injected is very low. An

ICP-MS device is a mass flow sensitive detector,so it struggles in very low flow applications, andit needs all the sensitivity it can get for CE.The process of nebulizing the CE effluent mustnot disturb the flow in the capillary, which isa challenge because of the suction generated bymost nebulizers. Various workers have describedmethods to compensate for suction [70]. Using thehigh sensitivity of sector instruments, Prange andSchaumloffel [71] and Tsunoda et al. [72, 73] haverecently described special nebulizer arrangementsfor CE separations that generate little or nosuction and do not disturb the quality of the CEseparation. These developments exploit the factthat, at very low liquid flow rate, the nebulizeddroplets evaporate quickly into solid particles,which can be transported to the ICP with veryhigh efficiency.

An example is shown in Figure 5.7.29 fromPrange and Schaumloffel [71]. An Element isused to monitor four elements (As, Se, Sb andTe) in a single injection. Two magnet settingsare necessary. Despite the sequential nature ofdetection and the need to wait for the magnet tostabilize, these four elements can be monitoredreadily in electrophoretic peaks that are very sharp,10 s peak widths in some cases. A proper interfacebetween the CE capillary and the nebulizer isessential in this measurement. Recent studies havedemonstrated separation of various isoforms of themetal storage protein metallothionein, for whichthe high efficiency and soft nature of CE arecritical [74]. In another use of sector instrumentswith CE for speciation, Sanz-Medel and coworkers[75] report measurement of Hg2+, MeHg+ andEtHg+ using a simple T-interface.

8.2 Gel electrophoresis

Electrophoresis in a flat configuration is, of course,also possible and offers the advantage of separationin two dimensions, typically isoelectric point (pI)and molecular weight. Several workers have usedthis format for metal speciation. The bands canbe cut out of the gel, then dissolved and analyzedas solutions [76]. Laser ablation has been used to


10

123456789

101112

ArsenocholineArsenobetaineArseniteDimethylarsinic acidPhenylarsonic acidMonomethylarsonic acidArsenateSeleniteSelenateAntimonateTelluriteTellurate

1 34

115

12

62

7

8

9

600000

400000

200000

0100 200 300

time/s

inte

nsity

/cou

nts

s−1

400 500

Figure 5.7.29. CE-ICP-MS separation of 12 species from four elements in one sample injection. Two magnet settings were used.The concentrations (as element) are As, Sb, Te 100 µg L−1 each, and Se 1 mg L−1. Reproduced from ref. 71 with permission ofthe Royal Society of Chemistry.

locate bands that contain the elements of interestwithout removing them from the gel [77].

8.3 GC separations

Rather quietly, GC-ICP-MS has become widelyused in some important real applications, partic-ularly for monitoring of Pb and Sn species inthe marine environment. Such species are gener-ally extracted from the samples, after derivatiza-tion if necessary. This sample preparation step alsoremoves most interferences.

Compared with liquid phase separations, capil-lary GC has several general advantages:

(1) There is no nebulizer, so all the analyte can beintroduced into the plasma.

(2) There is no solvent, with the accompanyingadvantages of a dry plasma.

(3) The chromatographic resolution and ability toseparate many components are often superiorfor GC.

(4) The separation is tuned by changing thecolumn temperature, which does not induce

a large change in sensitivity of the ICP-MS device.

Naturally, GC is only useful for compounds thatare volatile and thermally stable or can be con-verted into derivatives that have these properties.

There have been several publications on GC-ICP-MS with sector instruments. Krupp et al. [2, 3]converted Pb to Et4Pb and introduced this com-pound through a capillary GC interface into a mul-ticollector instrument. Good precision was obtaineddespite the short duration of the GC peaks (∼3 s)because no scanning was necessary. The GC wasconnected to the ICP by a heated interface. Theyalso simultaneously introduced a nebulized stan-dard solution containing Tl for instrument opti-mization and mass bias corrections on-the-fly, i.e.,as the Pb signals were also being measured.

Although not strictly speciation, Evans et al. [78]also report measurements of Hg isotope ratios usingtransient Hg signals from Hg vapor trapped fromcoal samples. Aerosols containing Tl were mixedwith the Hg vapor for mass bias corrections. Fourdifferent multicollectors and three other ICP-MSinstruments were tested for this application.

HIGH RESOLUTION MEASUREMENTS WITH PLASMAS OTHER THAN THE ICP 413

GC separations have also been used withscanning sector instruments. Sanz-Medel andcoworkers [79] used an Element to measure 32Sin compounds that cause bad breath. This peakwas readily resolved from 16O2

+ at the mediumresolution setting of the instrument. Of the manypossible compounds in breath samples, this is asensitive, fast and unambiguous way to find thosethat contain sulfur. They use a simple interfacewith flexible tubing so that the GC can be movedindependently of the ICP-MS. Prohaska et al. [80]measured arsenic species from soils by both LCand GC separations. The GC capillary was simplyfed into the usual concentric nebulizer, which wasconnected directly to the injector tube of the ICP.

9 HIGH RESOLUTION WITH MASSANALYZERS OTHER THANMAGNETIC SECTORS

Although magnetic sectors are by far the mostcommon devices for high resolution measurementswith an ICP, other types of mass analyzer havebeen or could be used for such measurements.These are mentioned briefly below, in the eventthat they develop further into widespread usage.

9.1 High resolution quadrupoles

The standard quadrupole operates in the firststability region and is normally limited to unitmass resolution. Douglas and coworkers [81] havedescribed the use of quadrupoles in alternatestability regions for either high resolution orunit mass resolution of high kinetic energy ions.Compared with work in the first stability region,the quadrupole control must provide higher powerto reach these new regions, which limits the massrange, but this is only a minor limitation for atomicions such as those from an ICP. Resolution up to8800, sufficient to separate 56Fe+ from 40Ar16O+to baseline with 10 % of the original ion signalremaining, has been demonstrated.

Amad and Houk [82] achieved resolution of22 000 with a multiple pass quadrupole operatedin the first stability region, although this concepthas not yet been applied to ions from an ICP.

9.2 Fourier transform ioncyclotron resonance

Eyler and coworkers [83] have published one paperon use of FT-ICR-MS with ions from an ICP. Thistype of instrument has the highest resolution ofany mass analyzer and is capable of separatingdifferent atomic ions at the same nominal mass.In more recent work, mass resolution up to600 000, sufficient to separate 40Ca+ from 40Ar+ tobaseline, was achieved [84]. As yet, the sensitivityis much lower than that obtained from beam-typeinstruments.

9.3 Time-of-flight MS

One of the most important recent developmentsin organic MS has been the capability to dohigh resolution and accurate mass measurementswith TOF mass analyzers. These instruments arethe basis of the very successful GC-TOF andquadrupole-TOF instruments. With orthogonal ionextraction [85], these devices now have a highduty cycle for ions produced from continuoussources such as electrospray, and they are also veryeffective at measuring product ions produced intandem MS experiments.

There are presently two commercially availableTOF-MS instruments with an ICP source, fromLECO and GBC. Although the present ICP-TOF-MS devices were probably not designed with thisin mind, resolution sufficient to separate manyof the worst cases of polyatomic ion overlapought to be possible, with the potential duty cycleadvantages of TOF for transient samples. In arecent paper, Adams and coworkers [86] describethe use of TOF-MS for multielement speciationin cytosols.

10 HIGH RESOLUTIONMEASUREMENTS WITH PLASMASOTHER THAN THE ICP

At first glance, it should be possible to use anyof the alternate plasmas with a magnetic sectorinstrument, although few such measurements have


been reported as yet. Nam et al. [33] used aHe ICP with a home-made Mattauch–Herzoginstrument, partly in the hope that the arraydetector would perform better with He+ at one endof it than with Ar+ in the middle. As usual with aHe plasma, Ar+, NO+, or other species persistedas major ions anyway.

11 CONCLUSION

Magnetic sector instruments provide both highsensitivity and selectivity for speciation measure-ments. If ions in the same m/z range are mon-itored so that the magnetic field setting is notchanged, the scan speed of a sector is compara-ble to that of quadrupoles. A longer settling time(∼300 ms) is needed if the magnetic field settingis changed. Their spectra acquisition rate is muchslower than that of TOF instruments, but they pro-vide much higher basic signal-to-noise ratios thanpresent TOF devices. It is expected that usage ofsector instruments in speciation measurements willincrease with the newer devices that are depend-able and simple to use. Multicollector instrumentsalso have a role in speciation measurements, asisotope ratios can be measured on transient peakswith high precision without scanning.

12 ACKNOWLEDGEMENTS

The Ames Laboratory is operated for the USDepartment of Energy by Iowa State Universityunder Contract W-7405-Eng-82. This work wassupported in the Chemical and Biological SciencesProgram by the Director of Science, Office ofBasic Energy Sciences, Division of ChemicalSciences. The author thanks Daniel R. Wiederinfor providing Figure 5.7.10, David B. Aeschlimanfor measuring the spectra in Figure 5.7.6, the latterworker and Jill Ferguson for critical commentson the manuscript, and Jin Wang, Dawn Dreessenand Regine Schoenherr for making many of thespeciation measurements reported. Early speciationwork by our group was performed on a magneticsector MS provided by CETAC Technologies,Omaha, NE. Chuck Douthitt provided valuableliterature surveys.

13 REFERENCES

1. Marchante-Gayon, J. M., Sariego Muniz, C., GarciaAlonso, J. I. and Sanz-Medel, A., Analysis of biologi-cal materials by double focusing ICP-MS, in Advances inAtomic Spectrometry , Vol. 7, Sneddon, J. (Ed.), Elsevier,Amsterdam, 2001; Anal. Chim. Acta, 400, 307 (1999).

2. Krupp, E., Pecheyran, C., Pinaly, H., Montelica-Heino,M., Koller, D., Young, S. M. M., Brenner, I. B. andDonard, O. F. X., Spectrochim. Acta, Part B , 56, 1219(2001).

3. Krupp, E. M., Pecheyran, C., Meffan-Man, S. and Donard,O. F. X., Fresenius’ J. Anal. Chem., 370, 573 (2001).

4. Roboz, J., Introduction to Mass Spectrometry; Instrumen-tation and Techniques , Wiley-Interscience, 1968, NewYork, Chapters 2 and 3.

5. Septier, A., Focusing of Charged Particles , AcademicPress, New York, 1967.

6. Wollnik, H., Optics of Charged Particles , AcademicPress, San Diego, CA, 1987.

7. Kerwin, L., Ion Optics, in Mass Spectrometry , McDow-ell, C. A. (Ed.), McGraw-Hill, New York, 1963.

8. Becker, J. S. and Dietze, H.-J., J. Anal. At. Spectrom., 12,881 (1997).

9. Houk, R. S. and Niu, H. S., Spectrochim. Acta, Part B ,51, 779 (1996).

10. Bradshaw, N., Hall, E. F. H. and Sanderson, N. E., J.Anal. At. Spectrom., 4, 801 (1989).

11. Giessman, U. and Greb, U., Fresenius’ J. Anal. Chem.,350, 186 (1994).

12. Feldmann, I., Tittes, W., Jakubowski, N., Stuewer, D.and Giessmann, U., J. Anal. At. Spectrom., 9, 1007 (1994).

13. Gray, A. L., J. Anal. At. Spectrom., 1, 247 (1986).14. Appelblad, P. K., Rodushkin, I. and Baxter, D. C., J.

Anal. At. Spectrom., 15, 359 (2000).15. Freedman, P., NU Plasma, personal communication,

2000.16. Vaughan, M.-A., Horlick, G. and Tan, S., J. Anal. At.

Spectrom., 2, 765 (1987); Horlick, G. and Montaser, A.,Analytical characteristics of ICPMS, in ICPMS , Mon-taser, A. (Ed.), Wiley-VCH, New York, 1998, p. 527.

17. Vanhaecke, F., Dams, R. and Vandecasteele, C., J. Anal.At. Spectrom., 8, 433 (1993).

18. Jiang, S.-J., Stevens, M. A. and Houk, R. S., Anal.Chem., 60, 1217 (1988); Sakata, K. and Kawabata, K.,Spectrochim. Acta, Part B , 49, 1027 (1994); Tanner, S. D.,J. Anal. At. Spectrom., 10, 905 (1995).

19. Vanhaecke, F., Saverwyns, S., De Wannemacker, G.,Moens, L. and Dams, R., Anal. Chim. Acta, 419, 55(2000).

20. Gillson, G., Douglas, D. J., Fulford, J. E., Halligan, K. W.and Tanner, S. D., Anal. Chem., 60, 1472 (1988); Tan-ner, S. D., Spectrochim. Acta, Part B , 47, 809 (1992).

21. Nonose, N. and Kubota, M., J. Anal. At. Spectrom., 16,551 (2001).

22. Sakata, K., Yamada, N. and Sugiyama, N., Spectrochim.Acta, Part B , 56, 1249 (2001).

REFERENCES 415

23. Wollnik, H., J. Mass Spectrom., 34, 991 (1999).24. Bandura, D. R., Tanner, S. D., Baranov, V. I., Koyanagi,

G. K., Lavrov, V. V. and Bohme, D. K., Ion–moleculechemistry solutions to the ICP-MS analytical challenges,in Plasma Source Mass Spectrometry: the New Millen-nium , Holland, G. and Tanner, S. D. (Eds), Royal Societyof Chemistry, Cambridge, UK, 2001, p. 141.

25. Eiden, G. C., Barinaga, C. J. and Koppenaal, D. W.,Rapid Commun. Mass Spectrom., 11, 37 (1997).

26. Tanner, S. D., Baranov, V. I. and Bandura, D. R., Spec-trochim. Acta, Part B , 57, 136 (2002).

27. Douglas, D. J., J. Am. Soc. Mass Spectrom., 9, 101 (1998).28. Tanner, S. D., Baranov, V. I. and Bandura, D. R., Reac-

tion chemistry and collisional processes in multipoledevices, in Plasma Source Mass Spectrometry: the NewMillennium , Holland, G. and Tanner, S. D., (Eds), RoyalSociety of Chemistry, Cambridge, UK, 2001.

29. Bandura, D. R., Baranov, V. I. and Tanner, S. D., Frese-nius’ J. Anal. Chem , 370, 454 (2001).

30. Yamada, N., Takahashi, J. and Sakata, K., J. Anal. At.Spectrom., 17, 1213 (2002); Spectrochim. Acta, Part B,submitted (2002).

31. Bandura, D. R. and Tanner, S. D., At. Spectrosc., 20,69 (1999); Bandura, D. R., Baranov, V. I. and Tan-ner, S. D., J. Anal. At. Spectrom., 15, 921 (2000).

32. Moens, L. J., Vanhaecke, F., Bandura, D. R., Baranov,V. I. and Tanner, S. D., J. Anal. At. Spectrom., 16, 991(2001).

33. Marriott, P., Fletcher, R., Cole, A., Beaumont, I., Loft-house, J., Bloomfield, S. and Miller, P., J. Anal. At. Spec-trom., 13, 1021 (1998).

34. Olesik, J. W., Appl. Spectrosc., 51, 158A (1997).35. Houk, R. S., Winge, R. K. and Chen, X., J. Anal. At.

Spectrom., 12, 1139 (1997).36. Walder, A. J., Koller, D., Reed, N. M., Hutton, R. C.

and Freedman, P. A., J. Anal. At. Spectrom., 8, 1037(1993); Walder, A. J. and Furuta, N., Anal. Sci. (Jpn),9, 675 (1993); Halliday, A. N., Lee, D.-C., Christen-son, J. N., Walder, A. J., Freedman, P. A., Jones, C. E.,Hall, C. M., Yi, W. and Teagle, D., Int. J. Mass Spectrom.Ion Process., 146/147, 21 (1995).

37. Cromwell, E. and Arrowsmith, P., J. Am. Soc. MassSpectrom., 7, 458 (1996).

38. Nam, S., Montaser, A. and Cromwell, E. F., Appl. Spec-trosc., 52, 161 (1998).

39. Burgoyne, T. W., Hieftje, G. M. and Hites, R. A., J.Anal. At. Spectrom., 12, 1149 (1997); Anal. Chem.,69, 485 (1997); Solyom, D. A., Burgoyne, T. W. andHieftje, G. M., J. Anal. At. Spectrom., 14, 1101 (1999).

40. Solyom, D. A., Gron, O. A., Barnes IV, J. H. and Hieftje,G. M., Spectrochim. Acta, Part B , 56, 1717 (2001).

41. Barnes IV, J. H., Sperline, R., Denton, M. B., Bari-naga, C. J., Koppenaal, D. and Hieftje, G. M., Anal.Chem., 74, 5317 (2002).

42. Minnich, M. G. and Houk, R. S., J. Anal. At. Spectrom.,13, 167 (1998).

43. Scott, R. H., Fassel, V. A., Kniseley, R. N. and Nixon,D. R., Anal. Chem., 46, 75 (1974).

44. Wu, M., Madrid, Y., Auxier, J. A. and Hieftje, G. M.,Anal. Chim. Acta, 286, 155 (1994).

45. Tarr, M., Zhu, G. and Browner, R. F., Appl. Spectrosc.,45, 1424 (1992).

46. Gonzalez LaFuente, J. M., Marchante-Gayon, J. M., Fer-nandez Sanchez, M. L. and Sanz-Medel, A., Talanta , 50,207 (1999).

47. Olesik, J., Hensman, C., Rabb, S. and Rago, D., Sampleintroduction, plasma–sample interactions, ion transportand ion–molecule reactions: fundamental understandingand practical improvements in ICP-MS, in Plasma SourceMass Spectrometry: the New Millennium , Holland, G.and Tanner, S. D. (Eds), Royal Society of Chemistry,Cambridge, UK, 2001, p. 5.

48. Li, J., Umemura, T., Odake, T. and Tsunoda, K., Anal.Chem., 73, 1416 (2001).

49. Wang, L., May, S. W., Browner, R. F. and Pollock, S. H.,J. Anal. At. Spectrom., 11, 1137 (1997).

50. Elder, R. C., University of Cincinnati, personal commu-nication, 1988.


52. Garcia Alonso, J. I., Encinar, J. R., Muniz, C. S., Mar-chante Gayon, J. M. M. and Sanz Medel, A., Isotopedilution analysis for trace metal speciation, in PlasmaSource Mass Spectrometry: the New Millennium , Hol-land, G. and Tanner, S. D. (Eds), Royal Society of Chem-istry, Cambridge, UK, 2001, p. 327.

53. Wang, J., Houk, R. S., Dreessen, D. and Wiederin, D. R.,J. Am. Chem. Soc., 120, 5793 (1998).

54. Lopez Garcia, A., Blanco Gonzalez, E. and Sanz-Medel, A., Mikrochim. Acta, 112, 19 (1993).

55. Belen Soldado Cabazuelo, A., Montes-Bayon, M., BlancoGonzalez, E., Garcia Alonso, J. I. and Sanz-Medel, A.,Analyst , 123, 865 (1998).

56. Wang, J., Houk, R. S., Dreessen, D. and Wiederin, D. R.,J. Biol. Inorg. Chem., 4, 546 (1999).

57. Wang, J., Dreessen, D., Wiederin, D. R. and Houk, R. S.,Anal. Biochem., 288, 89 (2001).

58. Harrington, C. F., Elahi, S., Merson, S. A. and Ponnam-palavanar, P., Anal. Chem., 73, 4422 (2001).

59. Montes-Bayon, M., Soldado Cabazuelo, A. B., Gonza-lez, E. B., Garcia Alonso, J. I. and Sanz-Medel, A., J.Anal. At. Spectrom., 14, 947 (1999).

60. Ferrarello, C. N., del Rosario Fernandez de la Campa, M.,Sariego Muniz, C. and Sanz-Medel, A., Analyst , 125,2223 (2000).

61. Thomas, C., Jakubowski, N., Stuewer, D., Klockow, D.and Emons, H., J. Anal. At. Spectrom., 12, 1221 (1998).

62. Feldmann, I., Jakubowski, N., Stuewer, D. and Thomas,C., J. Anal. At. Spectrom., 15, 371 (2000).

63. Jakubowski, N., Feldmann, I., Stuewer, D. and Berndt, H.,Spectrochim. Acta, Part B , 47, 119 (1992); Berndt, H., Fre-senius’ J. Anal. Chem., 331, 321 (1988); Jakubowski, N.,


Jepkens, B., Stuewer, D. and Berndt, H., J. Anal. At. Spec-trom., 9, 193 (1994).

64. Jakubowski, N., Stuewer, D., Klockow, D., Thomas, C.and Emons, H., J. Anal. At. Spectrom., 16, 135 (2001).

65. Klueppel, D., Jakubowski, N., Messerschmidt, J., Stuewer,D. and Klockow, D., J. Anal. At. Spectrom., 13, 255 (1998).

66. Wind, M., Edler, M., Jakubowski, N., Linsheid, M.,Wesch, H. and Lehman, W. D., Anal. Chem., 73, 29(2001).

67. Wind, M., Wechst, H. and Lehmann, W. D., Anal. Chem.,73, 3006 (2001).

68. Siethoff, C., Feldmann, I., Jakubowski, N. and Linscheid,M., J. Mass Spectrom., 34, 421 (1999).

69. Hassellov, M., Lyven, B. and Beckett, R., Environ. Sci.Technol., 33, 4528 (1999).

70. Majidi, V. and Miller-Ihli, N., Analyst , 123, 863 (1998);Day, J. A., Sutton, K. L., Somon, R. S. and Caruso, J. A.,Analyst , 125, 819 (2000).

71. Prange, A. and Schaumloffel, D., J. Anal. At. Spectrom.,14, 1329 (1999); Fresenius’ J. Anal. Chem., 364, 452(1999).



74. Wang, Z. and Prange, A., Anal. Chem., 74, 626 (2002).75. Silva da Rocha, M., Soldado, A. B., Blanco-Gonzalez, E.

and Sanz-Medel, A., J. Anal. At. Spectrom., 15, 513(2000).

76. Lustig, S., De Kimpe, J., Cornelis, R., Schramel, P. andMichalke, B., Electrophoresis , 20, 1627 (1999); Lustig S.,

De Kimpe, J., Cornelis, R. and Schramel, P., Fresenius’J. Anal. Chem., 363, 484 (1999).

77. Alt, F., Messerschmidt, J. and Weber, G., Anal. Chim.Acta , 359, 65 (1998); Becker, J. S., Boulyga, S. F.,Becker, J. S., Damoc, N. -E. and Przybylski, M. Int. J.Mass Spectrom., submitted (2003).

78. Evans, R. D., Hintelmann, H. and Dillon, P. J., J. Anal.At. Spectrom., 16, 1064 (2001).

79. Rodriguez-Fernandez, J., Montes-Bayon, M., Pereiro, R.and Sanz-Medel, A., J. Anal. At. Spectrom., 16, 1051(2001).

80. Prohaska, T., Pfeffer, M., Tyulipan, M., Stingeder, G.,Mentler, A. and Wenzel, W. W., Fresenius’ J. Anal.Chem., 364, 467 (1999).

81. Ying, J.-F. and Douglas, D. J., Rapid Commun. MassSpectrom., 10, 649 (1996); Du, Z., Douglas, D. J. andKonenkov, N. V., J. Anal. At. Spectrom., 14, 1111 (1999).

82. Amad, M. H. and Houk, R. S., Anal. Chem., 70, 4885(1998); J. Am. Soc. Mass Spectrom., 11, 407 (2000).

83. Milgram, K. E., White, F. M., Goodner, K. L., Watson,C. H., Koppenaal, D. W., Barinaga, C. J., Smith, B. J.,Winefordner, J. D., Marshall, A. G., Houk, R. S. andEyler, J. R., Anal. Chem., 69, 3714 (1997).

84. Watson, C. H., Houk, R. S. and Eyler, J. R., unpublishedresults.

85. Guilhaus, M., Selby, D. and Mlynski, V., Mass Spectrom.Rev., 19, 65 (2000).

86. Infante, H. G., Van Campenhout, K., Blust, R. and Adams,F. C., J. Anal. At. Spectrom., 17, 79 (2002).

5.8 On-line Elemental Speciationwith Functionalised Fused Silica Capillariesin Combination with DIN-ICP-MS

J. BettmerJohannes Gutenberg-Universitat Mainz, Germany

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 4172 Experimental Set-up . . . . . . . . . . . . . . . . . 4173 Preparation and Procedure of Capillary

Modifications . . . . . . . . . . . . . . . . . . . . . . 4184 Characterisation of the Modified Capillaries 419

4.1 Stability and capacity . . . . . . . . . . . . 4194.2 Atomic force microscopy . . . . . . . . . 421

5 Applications of the Modified Capillaries inElemental Speciation . . . . . . . . . . . . . . . . 421

5.1 Modified capillaries for the separationof spectral interferences in ICP-QMS 421

5.2 Chromium speciation . . . . . . . . . . . . 4225.3 Mercury speciation . . . . . . . . . . . . . . 423

6 Conclusions and Future Developments . . . 4257 Acknowledgements . . . . . . . . . . . . . . . . . . 4268 References . . . . . . . . . . . . . . . . . . . . . . . . 426

1 INTRODUCTION

Hyphenated techniques are usually necessary toolsfor metal speciation analysis. The combination ofa chromatographic or electrophoretic separationunit with an element-selective detection system canprovide qualitative as well as quantitative informa-tion about the different forms in which an elementoccurs. Many papers have reviewed the frequentlyused techniques and their applications [1–11].

As the detection method in metal speciationICP-MS (inductively coupled plasma-mass spec-trometry) has become more and more popularowing its excellent properties concerning detec-tion limit, sensitivity, selectivity and the simplicityto couple both gas and liquid chromatographicsystems to it. Another important advantage isthe possibility of monitoring different isotopes ofone element which has been proved a successfor identifying artefact formations and degradation

processes during analysis [12–14]. In particular,isotope-labelled species nowadays contribute to abetter understanding for such processes combinedwith the possibility of applying isotope dilution foraccurate and precise quantification [13, 15].

This paper aims to present an alternative methodfor the direct elemental speciation using ICP-MS. The application of an on-line separationduring sample transport to the plasma unit will bedemonstrated for the examples of chromium andmercury speciation. Instead of chromatographiccolumns, chemically modified capillaries as usedin capillary electrophoresis (CE) allow a selectiveretention of one of the relevant species.

2 EXPERIMENTAL SET-UP

For introducing the liquid sample into the plasmaa direct injection nebulizer (DIN) was used. In

418 SPECIATION WITH FUSED SILICA CAPILLARIES AND DIN-ICP-MS

particular, for elements showing memory effectssuch as mercury this kind of sample introductionis favourable because of the minimisation ofthis negative effect. The sample transport by thedirect injection nebulizer was obtained throughfused silica capillaries (400–700 mm × 0.15 mmo.d., 50 µm i.d., encased in a Teflon tube). Thenebulizer gas was connected to the carrier gassupply (range 0–2 L min−1) while the nebulizerauxiliary gas was connected to the blend gassupply (range 0–2 L min−1) of the ICP-MS system.The liquid transport was performed using agas displacement pump with Milli-Q water ascarrier solution. A metal-free injection valvewas a component of the pump module of theDIN which allowed time-controlled injections. Anadditional argon supply (150 psi) was necessaryfor the DIN pump for the loading of the loopwith the sample. For the DIN two differentsample introduction capillaries are available, with50 and 75 µm i.d. For these investigations the50 µm i.d. capillaries were chosen for a moreefficient interaction between capillary surface andanalytes. The general operation conditions forthe ICP-MS in combination with the DIN areshown in Table 5.8.1, whereas the schematicadaptation of the DIN to the plasma torch ispresented in Figure 5.8.1. For these studies ICP-QMS was used in the time-resolved mode and

Table 5.8.1. ICP-MS (HP 4500) operating conditions withdirect injection nebulizer (DIN).

Parameter Value (DIN)

RF power 1320 WReflected power 0–8 WGDP pressure (DIN) 300–420 psiNebulizer pressure

(DIN)75 psi

Gas flow ratePlasma 13–14 L min−1

Auxiliary 0.8–1.5 L min−1

Carrier 0.4–0.7 L min−1

Capillary tip position 0.3 mm in front of nebulizerbody

Sample depth 9.5 mmSample cone Ni, 1.0 mm orificeSkimmer cone Ni, 0.4 mm orificeDwell time 300 ms

peak height measurements were chosen for dataevaluation.

3 PREPARATION AND PROCEDUREOF CAPILLARY MODIFICATIONS

In the field of capillary electrophoresis the surfacemodification of fused silica capillaries is a commontechnique for the suppression or reversal ofthe electroosmotic flow [16]. Different techniques,mostly based on the chemical binding of silane

Adapter tothe Torch

Modified Capillary

Nebulizer Gases

Auxiliary Gas

Plasma Gas

Torch

Figure 5.8.1. Adaptation of the DIN to the ICP torch.

CHARACTERISATION OF THE MODIFIED CAPILLARIES 419

reagents with the desired functional groups to thesurface have been carried out. This technique wasadapted to our purpose of modifying the innersurface of a fused silica capillary with a cationicand anionic exchanger material.

The activation of the inert surface was obtainedby rinsing the capillaries (50 µm i.d. × 360 µmo.d., length 40–70 cm) with 1 M NaOH. A min-imum of 30 min was necessary for a stable activa-tion of the capillary inert surface. Afterwards thecapillary was flushed with Milli-Q water (15 min)in order to remove the majority of sodium ionsand dried by purging with nitrogen (5 bar) at 80 ◦Cfor 1 h.

After the activation of the capillary surfacethe cationic exchanger coating was obtained byrinsing the capillaries with the coating reagent.The coating solution consisted of 5 % (v/v)2-(4-chlorosulfonylphenyl)-ethyltrichlorosilane inmethylene chloride. The capillary was rinsed withthis solution for 2 h at a temperature of 60 ◦C with a

SiO

SiO

SiO

SiO

SiO

SiO

SiO

OH OH OH OH OH OH OH

n

+

SiCl3ClO2S

1. 30 °C, CH2Cl22. H2O

SiO

SiO

SiO

SiO

SiO

SiO

SiO

O O O O O O O

n

OH OH OH

SO2OH SO2OH SO2OH

SiSi Si

Figure 5.8.2. Scheme of the coating procedure (cationexchanger).

NH3+Cl−

Si OH

O

NH3+Cl−

Si OHSi OH

SiO

SiO

SiO

SiO

SiO

Si

O O O O O

n

NH3+Cl−

Figure 5.8.3. Schematic surface of a modification with ananion exchanger.

pressure of 2 bar. This dynamic coating procedurewas stopped every 15 min so that the capillary wascoated for 5 min in a static mode. After the coatingthe reagent not bound to the surface was removedwith methylene chloride (20 min) and with water(15 min) at room temperature. Before use the cap-illaries were purged with nitrogen (5 bar) at 80 ◦Cfor 2 h. Figure 5.8.2 presents the scheme of thiscoating procedure for the example of a cationexchanger material [17].

In the case of an anion exchanger material,3-aminopropyltrimethoxysilane (Figure 5.8.3) wasused as the bifunctional reagent. The coatingprocedure was analogously performed as describedfor the cation exchanger.

4 CHARACTERISATION OF THEMODIFIED CAPILLARIES

4.1 Stability and capacity

The stability of the coatings was investigated tooptimise the efficiency of the capillaries withoutdestroying the surface while removing retainedspecies. Special attention was paid to the pHvalues of the introduced samples. These solutionswere prepared on a daily base to guarantee thebest possible accuracy of these data. Testing thepossible pH range was necessary because for someof the exchange resins no pH working range hasbeen determined so far. For this investigation shortpieces of 1.5–2.5 cm were connected between thenebulizer and the delivering pump flushed withsolutions of different pH ranging from 1.2 to10.5. For the sulfur-containing columns the m/z


Cation-exchanger [Cr(III)]

Anion-exchanger [Cr(VI)]

0 20 40 60 80 100 120 140

Concentration of Cr species [µg/L]

0

1

Rel

ativ

e si

gnal

inte

nsity

Figure 5.8.4. Capacity studies of the modified capillaries.

ratios 32 (S+) and 64 (SO2+) were constantly

measured. For the nitrogen-containing columnsm/z 54 (ArN+) was monitored. Although thesemasses have a naturally a high background level adestruction of the coating could be recognised bya significant increase of these levels. It turned outthat the coatings investigated suffered destructionat a pH lower than 1.8 and higher than 9.0. Hence,for the later experiments usually pH 2 or higherwas used as the most acidic solution to removeretained species from the capillary. The stable pH

ranges were later investigated for the most efficientpreconcentration or retention of the species.

The next step involved the evaluation of thebreakthrough capacity of the columns. The usedcoatings vary in their ion exchange strength. There-fore it was necessary to guarantee that the coatingswere not overloaded. Figure 5.8.4 shows a capacitystudy for chromium species. Each data point repre-sents a series of five injections (10 µL) with givenconcentration. The breakthrough limit for thesecapillaries can be set between 40 and 50 µg L−1.

(a)

1

2

3

100.

000

nm

µm

1

2

3

100.

000

nm

µm(b)

Figure 5.8.5. (a) AFM picture of an untreated surface; (b) AFM picture of an activated surface; (c) AFM picture aftermodification with an anion exchanger; (d) AFM picture after modification with a cation exchanger.

APPLICATIONS OF THE MODIFIED CAPILLARIES IN ELEMENTAL SPECIATION 421

(c) (d)

1

2

3

200.

000

nm

µm

1

2

3

4

200.

000

nm

µm


4.2 Atomic force microscopy

To be sure that every group has been derivatized,and then to be able to perform quantitativeretention of the species, a check of the internalsurface of the capillaries has been made via AFM(atomic force microscopy).

Figure 5.8.5(a) shows the internal surface ofsilica fused capillaries as they are before anytreatment. The surface is smooth, while someimpurities are present and represented as lightpeaks. Figure 5.8.5(b) shows the surface afterflushing it with NaOH: the darker parts of thesurface did not react, while the light struc-tured parts represent the removal of SiO2 lay-ers. Finally, the activated surface has been rinsedwith the derivatizing reagents: Figures 5.8.5(c)and (d) show the result of this final step. Whileusing 3-aminopropyltrimethoxysilane the surfacestill looks smooth and even (Figure 5.8.5(c)), andderivatization occurred on the major part; other-wise a large part seemed to remain unchangedwhen 2-(4-chlorosulfonylphenyl)-ethyltrichloro-silane was used (Figure 5.8.5(d)).

Impurities are anyway still present in everystep of the modification, and might be responsiblefor a nonquantitative coating of the capillaries.The second reason for this observation is that the

inability to activate the surface completely maylead to a nonquantitative retention of interferingspecies during the analysis, so that the exchangecapacity is limited.

5 APPLICATIONS OF THE MODIFIEDCAPILLARIES IN ELEMENTALSPECIATION

5.1 Modified capillaries for theseparation of spectral interferencesin ICP-QMS

In ICP-QMS analysis spectral interferences haveto be considered in order to guarantee the qual-ity of analytical measurements. A new way ofchemical separation of elements responsible for theexistence of spectral interferences such as clus-ter ions or doubly charged ions was developedby the use of the modified capillaries. Anionicand cationic exchanger coatings have proven toretain selectively anions and cations, respectively,which could strongly interfere with the determina-tion of selected elements. Examples were shownfor platinum group elements such as Pt, Pd, andRh (matrix separation concerning the interferingelements Cu, Pb, Rb, Sr, and Hf) and for As


(monoisotopic 75As is interfered with by the for-mation of 40Ar35Cl) [17–19].

5.2 Chromium speciation

The differentiation between the two dominantredox species of chromium (Cr(III) and Cr(VI))has gained importance during the last decades.Since the carcinogenic potential of hexavalentchromium has been realised, many analyticalmethods have been developed. They mainlyuse hyphenated techniques, the coupling of achromatographic separation and mostly element-selective detection unit. In the preceding chapterstechniques for applying ICP-MS as detector havebeen mentioned.

In contrast to conventional methods our devel-opment is without the adaptation of a chromato-graphic system to the ICP-MS detector. The sys-tem achieves the separation of both species bythe selective retention of one species dependingon the kind of modified capillary. In the case ofretaining Cr(VI), which is favourable in the case ofhigher Cr(III) concentrations, the use of an anionexchanger material (illustrated as an example inFigure 5.8.3) is preferred. For Cr(III) preconcen-tration a cation exchanger is consequently applied.

For developing a separation method the mostinfluential factor, the pH, was optimised in orderto retain one of the species, while the other can

quantitatively be eluted. The optimum pH for thecationic exchanger is 5.5, while the use of theanionic exchanger for the retention of Cr(VI) needsa pH of about 6.0. An example of the preconcentra-tion of Cr(VI) is given in Figure 5.8.6, demonstrat-ing that the relevant species will be quantitativelyretained in the capillary after six injections. How-ever, flushing the capillary with nitric acid releasesthe retained species and the signal intensity forthis accumulated chromium was compared with aninjection of 10 µg L−1 Cr(III). As shown, these tworesults are comparable and emphasise the possibil-ities given by this method.

In order to prove its reliability the method wasapplied for the determination of chromium in areference material CRM 545 (welding dust). Thesample preparation was carried following basicallythe instructions given in the EC certificationreport [20]. In brief: the dust was weighed, dilutedin a 2 % NaOH–3 % Na2CO3 solution and agitatedin a heated ultrasonic bath for 20 min. Prior toseparation and analysis the solution was dilutedby factor of 20 000 and the pH was adjusted to 5.5using dilute hydrochloric acid.

In Figure 5.8.7 a typical time-resolved pictureof the species-selective flow injection procedure isdemonstrated. Six replicate injections of the dis-solved sample show the signal for Cr(VI) whilethe injection of nitric acid releases the amount ofCr(III) initially preconcentrated in the capillary.

100 200 300 400 500

0

1

Injection of HNO3 (pH = 2)

Rel

ativ

e si

gnal

inte

nsity

53C

r

Injection of10 ng/mL Cr(III)

6 Injections of1.5 ng/mL Cr(VI)

Time [s]

Figure 5.8.6. Retention and elution of Cr(VI) using an anion exchanger capillary.

APPLICATIONS OF THE MODIFIED CAPILLARIES IN ELEMENTAL SPECIATION 423

Time [min]

20 000

0

40 000

60 000

80 000

100 000

120 000

140 000

160 000

0 2 64

Elution with HNO3

Sig

nal i

nten

sity

53C

r [c

ps]

Cr(VI)

Cr(III)

Figure 5.8.7. Time-resolved measurement of CRM 545 (six replicate injections of the dissolved sample [21]).

Table 5.8.2. Determination of chromium species in weldingdust CRM 545.

This work (g kg−1) Certified (g kg−1)

Total leachablechromium

38.3 ± 0.9 39.5 ± 1.3

Cr(VI) 38.1 ± 1.6 40.2 ± 0.6Cr(III) 0.27 ± 0.08 Not certified

The results obtained with this newly devel-oped method are summarised in Table 5.8.2 andshow excellent agreement with certified values.Recent publications have reported small amountsof Cr(III) in the sample which could also beobserved during our investigations.

These results show the applicability of thedeveloped method for the differentiation of Cr(III)and Cr(VI) without the necessity of a chromato-graphic system [21]. The detection limits obtainedare 8 ng L−1 for Cr(III) and 31 ng L−1 for Cr(VI),respectively.

5.3 Mercury speciation

Mercury is one of the elements for which analyticaldevelopments have been pursued most intenselyin order to speciate the organic forms of mercury,especially methylmercury. Hyphenated techniqueshave been applied to solve this analytical problem.In most cases a gas or liquid chromatographicseparation unit was coupled to an element-selective

detection system in order to obtain qualitativeand quantitative information about the differentbinding forms.

This work aims to apply modified capillaries toa screening method of mercury compounds. Thedifferentiation of organic and inorganic species asa preliminary diagnostic tool allows the estimationof the existence and concentration of the organicspecies. In the case that organic compounds couldbe detected a chromatographic separation couldprovide information about each single species.

For method development the behaviour of themost important organic species, methylmercury(MeHg+) was compared with the behaviour ofinorganic mercury (Hg2+). With the use of thecationic exchanger capillary modified with 2-(4-chlorosulfonylphenyl)-ethyltrichlorosilane, theretention behaviour of both compounds wasobserved as a function of the pH of the injectedsolution. Figure 5.8.8 demonstrates that the signalintensities of both compounds showed a local min-imum at a pH of 5.5. Our investigations provedthat the signal suppression of Hg2+ was causedby its complete retention inside the capillary. Inthe case of methylmercury the minimum in sig-nal intensity was derived from a signal suppres-sion in the plasma. No methylmercury could befound after rinsing the capillary with acid in orderto elute retained compounds. This result madethe differentiation between the two species after


5 6 7

0

1000

2000

3000

4000

10 µg/L Hg2+

10 µg/L MeHg+S

igna

l int

ensi

ty 20

1 Hg

[cps

]

pH

Figure 5.8.8. Influence of pH on the signal intensities of Hg2+ and MeHg+.

0 1 2 3 4 5

0

20 000

40 000

60 000

10 µg/L MeHg+ 10 µg/L Hg2+

Sig

nal I

nten

sity

202 H

g [c

ts]

Time [min]

Figure 5.8.9. Injection and elution profiles of Hg2+ and MeHg+ (five replicate injections of the analytes and elution with10−3 M HCl).

pH adjustment possible – whereas other alkylatedspecies like dimethylmercury behave like MeHg+.Typical injection signals for both species are shownin Figure 5.8.9. Elution of retained Hg2+ couldbe obtained by injecting 10−3 M HCl. The signalforms demonstrate that the use of the direct injec-tion nebulizer (DIN) minimises the memory effectsof mercury substantially.

Under optimised conditions the selectivity wasinvestigated. A comparison of signal intensitiesfrom MeHg+ and Hg2+ injections was carriedas demonstrated in Figure 5.8.10. A concentrationexcess of 100 ([Hg2+]/[MeHg+]) shows no signalfor inorganic mercury indicating that MeHg+can be selectively detected at levels higher than1 % of total mercury content in a sample. This

CONCLUSIONS AND FUTURE DEVELOPMENTS 425

1 10 100 1000 10 000

0

1

200

400

600

800

1000

1200

Rel

ativ

e si

gnal

inte

nsity

[202 H

g]

[Hg2+]/[MeHg]

Figure 5.8.10. Selectivity of MeHg+ versus Hg2+.

circumstance is particularly likely in biologicaland marine samples, for which this method canbe easily applied to prove the existence oforganomercury compounds.

The method was applied for the analysis ofa tuna sample which was foreseen as a refer-ence material for total mercury content (9/97).The sample was dissolved in 25 % (w/v) TMAH(tetramethylammonium hydroxide) by microwavedigestion (15 min, 45 W), and after filtration anddilution the solution was adjusted to pH of 5.5. Theresults obtained were compared with the resultsobtained using a hyphenated technique (cap-illary cold trap-GC-AAS, CCT-GC-AAS [22]).They were in a good agreement regarding themethylmercury content (Table 5.8.3) and empha-sised the potential of this method as a screeningmethod for organomercury compounds. The detec-tion limit for methylmercury was 100 ng L−1, suf-ficient for its determination in biological samples.

6 CONCLUSIONS AND FUTUREDEVELOPMENTS

These investigations showed that a lot of informa-tion about elemental species can be obtained by theuse of modified capillaries for sample introduction

Table 5.8.3. Determination of mercury species in referencematerial 9/97 (tuna fish).

Concentration (mg kg−1)

Modified DIN-ICP-MSConcentration (organic

mercury) 2.6 ± 0.2Comparative investigations

using CCT-AASConcentration (MeHg+) 2.4 ± 0.4

in ICP-MS, realised for the first time for metal spe-ciation analysis. On the example of redox speciesof chromium a differentiation between Cr(III) andCr(VI) is possible using the exchange capabilitiesof the modified surfaces. In the case of mercurya differentiation between inorganic and organicspecies is possible using a modification with acation exchanger. This can be useful as a screen-ing method. For further information about the exactspecies present an analysis using chromatographicseparation is necessary.

Further investigations are under way, espe-cially to evaluate a similar method for the screen-ing of organolead compounds. Furthermore, theapplication of more selective modifications, e.g.complex-forming reagents, is being studied to formspecies-selective interactions to the capillary sur-face. Another aim to be pursued is the application


of isotope dilution for precise and accurate quan-tification of the species of interest.

7 ACKNOWLEDGEMENTS

The author thanks the European Commissionfor the financial support (SMT-4-CT98-2233,MOSIS: Modified Sample Introduction Systems)and the project partners C. Camara, R. Garcia-Sanchez (Universidad Complutense de Madrid,Spain), L. Ebdon (University of Plymouth, UK),B. Rosenkranz, H.-G. Riepe (Westf. Wilhelms-Universitat Munster, Germany) and J. Schmitt(Johannes Gutenberg-Universitat Mainz, Germany)for excellent cooperation.

8 REFERENCES

1. Harrison, R. M. and Rapsomanikis, S. (Eds), Environ-mental Analysis Using Chromatography Interfaced withAtomic Spectroscopy , Horwood, Chichester, 1989.

2. Batley, G. E. (Ed.), Trace Element Speciation: AnalyticalMethods and Problems , CRC Press, Boca Raton, FL,1989.

3. Uden, P. (Ed.), Element-specific Chromatographic Detec-tion by Atomic Emission Spectroscopy , American Chem-ical Society, Washington, DC, 1992.

4. Kramer, J. R., Allen, H. E. (Eds), Metal Speciation: The-ory, Analysis and Application, Lewis, Chelsea, 1988.

5. Krull, I. S. (Ed.), Trace Metal Analysis and Speciation ,Elsevier, Amsterdam, 1991.

6. Szpunar-Lobinska, J., Witte, C., Lobinski, R., Adams,F. C., Fresenius’ J. Anal. Chem., 351, 351 (1995).

7. Donard, O. F. X., Martin, F. M., Trends Anal. Chem., 11,17 (1992).

8. Chau, Y. K., Zhang, S. Z., Maguire, R. J., Analyst , 117,1161 (1992).

9. Rosenkranz, B., Bettmer, J., Trends Anal. Chem., 19, 138(2000).

10. Ebdon, L., Hill, S., Ward, R. W., Analyst , 112, 1 (1987).11. Vela, N. P., Olson, L. K., Caruso, J. A., Anal. Chem., 65,

585 A (1993).12. Hintelmann, H., Falter, R., Ilgen, G. and Evans, R. D.,

Fresenius’ J. Anal. Chem., 358, 363 (1997).13. Demuth, N. and Heumann, K. G., Anal. Chem., 73, 4020

(2001).14. Encinar, J. R., Alonso, J. I. G. and Sanz-Medel, A.,

J. Anal. At. Spectrom., 15, 1233 (2000).15. Nusko, R. and Heumann, K. G., Fresenius’ J. Anal.

Chem., 357, 1050 (1997).16. Hjerten, S., J. Chromatogr., 347, 191 (1985).17. Riepe, H.-G., Gomez, M., Camara, C. and Bettmer, J.,

J. Mass Spectrom., 35, 891 (2000).18. Riepe, H.-G., Gomez, M., Camara, C. and Bettmer, J.,

J. Anal. At. Spectrom., 15, 507 (2000).19. Garcia, R., Gomez, M., Palacios, M. A., Bettmer, J. and

Camara, C., J. Anal. At. Spectrom., 16, 481 (2001).20. BCR Information, The Certification of the Contents of

Cr(VI) and Total Leachable Cr in Welding Dust Loadedon a Filter (CRM 545), EUR 18026 EN (1997).

21. Rosenkranz, B., Riepe, H.-G., Bettmer, J. and Ebdon, L.,J. Anal. At. Spectrom., in press.

22. Dietz, C., Madrid, Y., Camara, C. and Quevauviller, P.,Anal. Chem., 72, 4178 (2000).

5.9 Speciation Analysis by Electrochemical Methods

Raewyn M. TownThe Queen’s University of Belfast, Northern Ireland

Hendrik EmonsEC Joint Research Center IRMM, Geel, Belgium

Jacques BuffleCABE Geneva, Switzerland

Symbols . . . . . . . . . . . . . . . . . . . . . . . . . . 4281 Introduction . . . . . . . . . . . . . . . . . . . . . . . 4292 Overview of Electroanalysis . . . . . . . . . . . 429

2.1 Fundamentals . . . . . . . . . . . . . . . . . . 4292.2 Instrumentation . . . . . . . . . . . . . . . . 432

2.2.1 Potentiostats . . . . . . . . . . . . . 4322.2.2 Cell designs . . . . . . . . . . . . . 4332.2.3 Electrodes and electrode

materials . . . . . . . . . . . . . . . . 4332.2.3.1 Potentiometry . . . . . 4332.2.3.2 Dynamic techniques 433

2.3 Measuring techniques . . . . . . . . . . . . 4342.3.1 Potentiometry . . . . . . . . . . . . 4342.3.2 Fixed potential

methods – amperometry . . . . . 4352.3.3 Voltammetry . . . . . . . . . . . . . 435

2.3.3.1 Cyclic voltammetry(CV) . . . . . . . . . . . . 436

2.3.3.2 Linear sweepvoltammetry (LSV) 437

2.3.3.3 Normal (NPV) andreverse (RPV) pulsevoltammetry . . . . . . 438

2.3.3.4 Differential pulsevoltammetry (DPV) 439

2.3.3.5 Square wavevoltammetry (SWV) 440

2.3.3.6 AC voltammetry(ACV) . . . . . . . . . . . 441

2.3.3.7 Anodic strippingvoltammetry (ASV) 442

2.3.3.8 Strippingchronopotentiometry(SCP) . . . . . . . . . . . 443

2.3.3.9 Adsorptive strippingvoltammetry (AdSV) 443

2.3.4 Electrochemical detection inliquid chromatography andflow-injection analysis . . . . . . 444

3 Principles of Speciation by Electroanalysis 4453.1 Thermodynamic aspects . . . . . . . . . . 4453.2 Kinetic aspects . . . . . . . . . . . . . . . . . 446

3.2.1 Dependence of measuredparameters on lability . . . . . . 448

3.2.2 Dependence of lability onmeasurement time scale . . . . . 449

3.2.3 Lability at microelectrodes . . . 4493.3 Considerations for sample preparation

and experimental conditions . . . . . . . 4503.3.1 Measuring solutions . . . . . . . . 450

428 DETECTION

3.3.2 Adsorption effects . . . . . . . . . 4503.3.3 Avoiding saturation of ligands

at the electrode surface . . . . . 4504 Applications . . . . . . . . . . . . . . . . . . . . . . . 451

4.1 Aquatic systems . . . . . . . . . . . . . . . . 4514.2 Water/Sediment systems . . . . . . . . . . 4544.3 Biological matrices . . . . . . . . . . . . . . 454

5 New Concepts and Prospects . . . . . . . . . . 455

5.1 Selectivity . . . . . . . . . . . . . . . . . . . . 4555.2 Metal speciation dynamics and

bioavailability . . . . . . . . . . . . . . . . . 4555.3 Spatial resolution . . . . . . . . . . . . . . . 4555.4 Linking theoretical and experimental

developments . . . . . . . . . . . . . . . . . . 4565.5 Instrumentation . . . . . . . . . . . . . . . . 456

6 References . . . . . . . . . . . . . . . . . . . . . . . . 456

SYMBOLS

A electrode areaai activity of species i

AC alternating currentAdSV adsorptive stripping voltammetryACV AC voltammetryASV anodic stripping voltammetryαc cathodic transfer coefficientc bulk concentrationcL,T total ligand concentrationcL free ligand concentrationcM,T total metal concentrationcM free metal concentrationCd differential capacityCE-EC capillary electrophoresis with

electrochemical detectionCV cyclic voltammetryD diffusion coefficientδ steady-state diffusion layer

thicknessDC direct currentDPV differential pulse voltammetryE potentialEpa anodic peak potentialEpc cathodic peak potentialEeq equilibrium potentialEref potential of the reference

electrodeE

◦ standard potentialEo′ formal redox potentialE1/2 half-wave potential�Ep peak potential differenceEPPS N -2-hydroxyethylpiperazine-N ′-

3-propanesulfonic acidF Faraday constantf frequency

HEPES N -2-hydroxyethylpiperazine-N ′-2-ethanesulfonic acid

HEPPS N -2-hydroxyethylpiperazine-N ′-3-propanesulfonic acid

HMDE hanging mercury drop electrodei current (faradaic if)ic capacity currentil diffusion-limited currentip peak potentialipa anodic peak currentipc cathodic peak currentISE ion selective electrodek0 standard rate constantkc cathodic rate constantL lability criterion parameterLCEC liquid chromatography with

electrochemical detectionLSV linear sweep voltammetryµ reaction layer thickness,

(DM/k′a)

1/2

n number of electrons transferredOx oxidized form of the reactantPIPES piperazine-N -N ′-bis(2-

ethanesulfonic acid)r electrode radiusR gas constantRed reduced form of the reactantRs solution resistanceQ chargeN number of moles of electrolyzed

materialSCE saturated calomel electrodeSWV square wave voltammetrySV stripping voltammetryt timetm measuring timetp pulse time

OVERVIEW OF ELECTROANALYSIS 429

T temperatureTES N -tris(hydroxymethyl)methyl-2-

aminoethanesulfonic acidTMFE thin mercury film electrodev scan rateWE working electrode

Superscriptb bulk

Subscriptox oxidized speciesred reduced species

1 INTRODUCTION

It is well established that the total concentra-tion of an element provides no information aboutits bioavailability, toxicity, transport properties, orresidence time within a given system. The quanti-tative analysis of chemical species in real-worldsamples represents a complex process. It is afar more challenging task than that of determin-ing total concentrations. The ideal method wouldbe one that can be deployed in situ with theanalytical signal being directly interpretable interms of the species present in the medium. How-ever, in practice, measurements of element spe-ciation often involve a number of consecutivesteps, including general sampling, sample prepa-ration, possibly some additional species separationsuch as chromatography, the measurement step ofspecies quantification (detection) and data evalu-ation. IUPAC has published recommended guide-lines for nomenclature of chemical speciation andfractionation of elements [1]. This chapter presentsa description of detection methods which exploitelectrochemical processes at electrodes.

In comparison with many analytical techniques,electrochemistry is unique in that it is based oninterfacial phenomena. It finds its main appli-cation in the investigation of dissolved speciesbut can also be used in certain cases for thedirect measurement of solids. Electroanalyticaltechniques have certain features that are advanta-geous for speciation analysis. In contrast to atomicspectrometry or ICP-MS they belong to the lowenergy excitation techniques which is the reason

why they are species selective rather than elementselective. Many of them exhibit excellent detectionlimits coupled with a wide dynamic range. Mea-surements can generally be made on very smallsamples, typically in the microliter volume range.

If the main targets of speciation analysis aregrouped into redox states, metal(loid) complexesand organometal(loid) compounds, analytes in allthree areas can be determined by electroanalysis.

2 OVERVIEWOF ELECTROANALYSIS

2.1 Fundamentals

Electrochemical methods are based on the mea-surement of electrical signals associated withmolecular properties or interfacial processes ofchemical species. Owing to the direct transforma-tion of the desired chemical information (concen-tration, activity) into an electrical signal (potential,current, resistance, capacity) by the methods them-selves, they provide an easier and cheaper accessto automation, computer control and data handlingin comparison with the other analytical methodsrequiring an additional transducer to obtain elec-trical signals. Two major difficulties in the appli-cation of electroanalytical techniques to complexreal-world samples have been the lack of selec-tivity of electrochemistry and the susceptibility ofthe electrode surface to fouling by surface activematerial in the sample. The fundamentals of elec-trochemical methods which are suitable for thequantitative determination of chemical species aredescribed on various levels in a number of text-books and monographs [2–7]. However, the rel-atively high level of theoretical knowledge andinterdisciplinary understanding necessary to applymost of the techniques to new samples of complexcomposition or to develop electroanalytical proce-dures seems to be a barrier for the broader useof these methods. Moreover, the great number ofdifferent techniques can make the proper selectionfor solving a certain problem difficult not only forthe newcomer but even for more experienced sci-entists. In addition, there are still various names

430 DETECTION

for the same technique in the literature. The readeris referred to IUPAC recommendations for classifi-cation and nomenclature of electroanalytical tech-niques [8, 9]. In the following, we will concentrateon those methods with a broader applicability forspeciation analysis.

Electroanalytical techniques can be classifiedin various ways. An approach mainly based onthe character of the measured signals and theirexcitation is shown in Figure 5.9.1. Only majortechniques were selected for this ‘family tree’.Electrochemistry can be divided into ‘ionics’and ‘electrodics’. Conductance measurements areamongst the important analytical methods that arebased on electrochemical properties of chemicalspecies in the bulk phase. Because of the rathernonselective nature of the measurement it is rarelyused to determine certain species directly in thesample solution, but it provides an easily obtain-able parameter to estimate the total number of ionsin liquid samples. Conductometry does find useful

application in studying counterion association withpolyelectrolytes where the data obtained are com-plementary to those provided by voltammetricmethods [10, 11]. More important for speciationanalysis is the application of conductivity detectorsin chromatographic systems.

The majority of electroanalytical methods belongto the field of electrodics. That means the analyticalsignal is produced at an interface, mostly formed at ametal/solution contact, by heterogeneous processeswith the participation of charged species. Measure-ments in the state of electrochemical equilibrium ofelectrochemical cells, i.e. without any overall cur-rent flowing, are the basis of potentiometry. Theimportance of these methods for species analysisin liquids cannot be overestimated particularly dueto the exciting developments in the field of ion-selective electrodes in the 1960s and 1970s.

Dynamic electrochemical techniques involveexternally initiated electrolysis (oxidation or reduc-tion) at the surface of an electrode. Quantitation of

acvoltammetry

impedancemeasurement

cyclicvoltammetry

linear sweepvoltammetry

differentialpulse

voltammetry

square-wave

voltammetry

normalpulse

voltammetry

ac techniques potential scan potential step

amperometry voltammetry chrono-potentiometry

coulometry(galvanostatic)

controlledpotential

controlledcurrent

static (I = 0) dynamic (I ≠ 0)

electroanalysis

conductometry

ionicselectrodics

potentiometry

potentiostaticcoulometry

Figure 5.9.1. Family tree of main electroanalytical methods.


a species (the analyte) in the sample is achievedby measuring the current (or charge) generated bythe surface redox process.

The basic system of electrochemical experi-ments is the electrochemical cell. It consists ofat least two independent electrodes which areimmersed in an electrolyte solution. The electrodesare electrically connected through both the solu-tion and external wires via a measuring device.The current flowing through this electrical circuitcan result either from the occurrence of sponta-neous electrode reactions (potentiometry) or fromheterogeneous redox reactions which are externallydriven by the application of a potential differenceat the electrodes (electrolytic cell).

The fundamental relation between reactivespecies at electrodes and the electrode potentialunder equilibrium conditions is expressed by theNernst equation (5.9.1):

Eeq = E0 + RT/nF ln(∏

aoxν/ ∏

aredν)

(5.9.1)

or by inclusion of the activity coefficients intothe standard electrode potential E0 of the elec-trochemical cell reaction (leading to a formalpotential E0′):

Eeq = E0′ + RT/nF ln(∏

coxν/ ∏

credν)

(5.9.2)

where R, T , n, and F have their usual meaning, aox

and ared are the activities of Ox and Red, respec-tively, cox and cred are the corresponding concen-trations, and ν is the stoichiometric coefficient ofthe reaction.

The potential of the indicator (or working) elec-trode is measured or applied (for electrolytic cells)with respect to a reference electrode which servesas a ‘calibration point’ for the potential difference.

For electroanalytical purposes the mass transferof the analyte is driven either by diffusion aloneor by a combination of diffusion and convection.Convection must be avoided or carefully controlledduring the experiments because it influences thethickness of the diffusion layer δ (Figure 5.9.2)which must be constant during the analyticalmeasurement.

Another layer has to be considered at theelectrode/solution interface: the so-called electri-cal double layer. On the solution side it con-tains the first few species layers at the electrodewhere the electron transfer actually occurs andwhere the greatest potential difference appears andis composed mainly of electrically oriented solventdipoles and adsorbed electrolyte ions. Moreover,a more extended region of 1–3 nm thickness(depending on the electrolyte concentration) isincluded which is characterized by a potential

Solution

Analyte

Electrode

Mass Transfer

adsorption/desorption

inter-facialchem.

reaction

(surfacephase)

(bulk phase)

(diffusion layerthickness d)

Figure 5.9.2. Steps of an electrochemical reaction.

432 DETECTION

gradient and consequently a different spatial dis-tribution of charged species in comparison to thebulk phase of solution. From the point of viewof analytical measurements one has to take intoaccount that there is a charge separation in thisinterfacial region which gives rise to a capacitycurrent ic according to

ic = Cd (dE/dt) (5.9.3)

where Cd is the differential double layer capac-ity. It is potential dependent and varies usuallybetween 10 and 100 µF cm−2. The capacity cur-rent ic is part of the overall measured current(although it is not related to the electron trans-fer process) and contributes in most cases only tothe background signal of the analytical measure-ment. Therefore, ic should be separated from thefaradaic current if, which carries the concentration-dependent analytical information, or it should bekept at least very small and constant for trace anal-ysis. The latter approach can be fulfilled with theuse of microelectrodes; if and ic can be separatedin potential step techniques (e.g. pulse techniques)by exploiting their different time-decay character-istics, by taking the inverse of the time deriva-tive of potential in chronopotentiometry [12] andby using the difference in phase of ic and if inAC techniques.

The electron transfer process is usually the pro-cess of interest for most analytical determinationsand gives rise to the so-called faradaic current. Itis based on Faraday’s law:

Q = nFN (5.9.4)

where Q is the charge of electricity, F is the Fara-day constant, N is the number of moles of mate-rial electrolyzed, and n is the number of electronsper molecule involved in the electrolysis process.Many electroanalytical techniques measure current(i), which is obtained from the above equation bydifferentiation with respect to time (t) to give

i = dQ/dt = nF(dN/dt) (5.9.5)

which shows that the current is a measure of therate of electrolysis at the electrode surface. The

analytical use of electrochemistry relies on the pro-portionality of the rate of electrolysis, as measuredby current, and the concentration of the speciesundergoing electrolysis at the working electrode.An important aspect of dynamic electrochemistryis the potential that is applied to the electro-chemical cell. The applied potential provides thedriving force for the electrolysis reaction uponwhich the analysis is based. The rate constantof the electron transfer depends exponentially onthe electrode potential E and is described for areduction by

kc = k0 exp[−αcnF(E − E0′)/RT ] (5.9.6)

where k0 and αc are reaction-specific kinetic con-stants called standard rate constant and cathodictransfer coefficient, respectively. Obviously, elec-troanalysis has to obey and can exploit thesensitive control of the reaction rate by theelectrode potential. However, both the electrontransfer kinetics and the analyte transport, usu-ally by diffusion, have to be considered togetherfor the design and evaluation of electroanalyticalmeasurements.

2.2 Instrumentation

In general electroanalysis requires only relativelyfew and inexpensive instruments. Potentiometry isperformed with a potentiometer (pH/mV meter)with high input impedance and the dynamictechniques are using a potentiostat.

2.2.1 Potentiostats

The potentiostat applies a defined potential acrossthe working electrode and the reference elec-trode for the typical three-electrode configura-tion of modern electrochemical cells. Moreover,it measures the current at the working elec-trode. Commercially available instrumentation hasbeen reviewed [13]. There is increasing interestin the development of portable battery-poweredpotentiostats for field use [14, 15] and multiplexinstrumentation for measurements with multichan-nel array electrodes [16, 17].


2.2.2 Cell designs

The importance of mechanical, geometric, andhydrodynamic factors in electrode design have beensummarized [18, 19]. Flow-through cells allowautomation of electroanalytical determinations andare essential for in situ measurements. Systemsrange from on-line (e.g. shipboard) flow-injectiontype cells [20] to fully immersible units in whichpressure compensation is an integral component toallow measurements at depth and to overcome prob-lems associated with the liquid junction of the ref-erence electrode [21]. Key design aspects of suchcells have been discussed [19]. Special considera-tions are required for fabrication of microanalyticalsystems [22]. In all cases, the construction materialsshould be carefully chosen to minimize adsorptionof sample components [23].

2.2.3 Electrodes and electrode materials

2.2.3.1 Potentiometry

The measuring electrode (also called indicatorelectrode) for potentiometric speciation analysisconsists of an appropriately chosen ion-selectiveelectrode as discussed in Section 2.3.1. The var-ious types of ISEs have been reviewed [24, 25].The potential difference is measured versus areference electrode (usually saturated calomelelectrode = SCE, or Ag/AgCl electrode) whichprovides an invariant potential.

2.2.3.2 Dynamic techniques

Common three-electrode cells contain a referenceelectrode, a counter electrode (CE, representedtypically by a Pt wire or plate, sometimes alsoa glassy carbon electrode, etc.) and the workingelectrode (WE). A survey of available typesof reference electrodes, including those suitablefor miniaturization and elimination of internalsolution, has recently been published [19].

The analyte of interest will react (mostly byelectrolysis) at the WE whereas the counter reac-tion will take place at the CE. By that a currentflow through the reference electrode is avoided.

The working electrode should have a reproduciblesurface morphology and area, and a low resid-ual current.

Mercury remains the electrode material ofchoice for detection of metals due to its largehydrogen overvoltage and its remarkable repro-ducibility. However, solid electrode substratesincluding Au, Pt, Ag, and various forms of car-bon also find applications. Various reviews havesurveyed the range of electrodes used in voltam-metric analysis [19, 26, 27].

Two types of mercury electrodes are commonlyused for stripping voltammetry (see Section 2.3.3):the hanging mercury drop electrode (HMDE) andthe thin mercury film electrode (TMFE). Com-mercial HMDE systems are able to produce mer-cury drops with very reproducible dimensions (theinternal diameter of the capillaries is typically60–200 µm). Narrow bore glass capillaries havebeen used for production of renewable micromer-cury drops [28, 29]. Thread electrodes, in which amercury column is enclosed within a hydrophilicdialysis membrane tube (ca 150 µm diameter),have been proposed for use in flow cells [30].TMFEs are produced by electrochemical deposi-tion of a thin mercury coating onto an electricallyconductive substrate, e.g. glassy carbon, platinum,gold or iridium. The film can be preformed ordeposited in situ after adding Hg2+ to the sam-ple solution; in situ formation should, however,be avoided when element speciation in the sam-ple is of interest. Many workers have preferred theTMFE over the HMDE, because it exhibits greatersensitivity, lower detection limits and sharper strip-ping peaks as a result of the greater electrode sur-face/volume ratio and shorter diffusion distancesfor the deposited metal. In addition, a TMFE canbe rotated to produce more reproducible solutionhydrodynamic conditions than can be achieved bysolution stirring. However, the HMDE is morereproducible and easier to operate with respectto the repetitive formation of the electrode sur-face and is better defined from a theoretical pointof view. The deposition of mercury onto carbonproduces droplets of the metal, rather than a truefilm [31]. There are numerous publications on pre-treatment protocols for the carbon surface [32]

434 DETECTION

and various deposition procedures [33] which arepurported to result in useful mercury ‘films’. Somemetal supports, e.g. gold and platinum, form amal-gams which have different electrochemical prop-erties and thus lower the reproducibility of theprocedure. It has been shown that iridium offersan attractive alternative as a conductive support forthe TMFE due to its good wettability by mercurythat allows formation of a perfect mercury surface(rather than the separate droplets formed on glassycarbon), and its very low solubility in mercurythat prevents the drawbacks of amalgam formationassociated with Pt, Au, and Ag substrates [34, 35].

Microelectrodes. Microelectrodes are charac-terized by possessing at least one geometricaldimension in the low µm range [36]. Single micro-electrodes have been fabricated as disk, band,fiber or microdrop and offer several advantagesfor speciation measurements in real-world samples,including the very small influence of ohmic dropwhich facilitates application in low ionic strengthmedia [37] (e.g. freshwaters), their greater stabilityand reproducibility, their insensitivity to bulk solu-tion convection [38] (a consequence of sphericaldiffusion), and the potential for analyses at highspatial resolution arising from their small physi-cal dimensions. There is an increasing number ofreports of application of microelectrodes to speci-ation analysis [15, 37, 39–41]; measured concen-trations correspond to the mobile and labile metalfraction [19].

Care must be taken in fabrication of microelec-trodes to ensure reproducible results. For example,perfect electrical contact, Ir–glass sealing, andIr-disk morphology were found to be the keyaspects for obtaining reproducible mercury-coatedIr microelectrodes [42]. Mercury deposits on suchelectrodes can be used for several days withoutrenewal of the mercury layer and the electrodesthemselves have a lifetime of at least severalyears [42].

Various microelectrode arrays, typically basedon photolithographic procedures, for example thedeposition of Ir followed by electroplating withmercury, have also been reported for electroanaly-sis of trace metals [17, 43, 44].

Modified electrodes. Chemically modified elec-trodes designed for diverse applications havegained increasing interest in recent years [45].For element speciation purposes, modificationprocedures are typically aimed towards either(i) increased selectivity towards target species [46,47] or (ii) protection of the electrode surface fromfouling due to adsorption of organic materialsin the sample matrix. It is presumed that a thinsemipermeable layer will exclude potential foul-ing materials by size and/or charge exclusion.For example, cellulose acetate may protect againstadsorption of proteins [48] and humic acids [49];Nafion is purported to prevent adsorption of humicsubstances [50]. None of these approaches areideal because the coatings themselves are not inerttowards the target elements, e.g. Nafion is an ion-exchange polymer [51], and they are often difficultto prepare in a reproducible and controlled manner.

Recently, gel integrated microelectrodes (GIME)have been reported in which a microelectrode isintegrated into a relatively thick agarose gel layer.The GIME has been shown to possess many advan-tages in terms of long-term operation, antifoulingproperties, and reproducibility. In particular, it hasbeen shown to selectively exclude colloids andmacromolecules from the electrode surface whilebeing inert towards the target elements and ensur-ing diffusion-controlled transport due to its anticon-vective properties [52]. The GIME is commerciallyavailable (Idronaut, Italy).

2.3 Measuring techniques

The following sections discuss a variety of elec-troanalytical techniques that differ in the mode ofexcitation signal-response characteristics.

2.3.1 Potentiometry

In potentiometry, the equilibrium potential of anindicator electrode is measured against a selectedreference electrode using a high impedance volt-meter, i.e. effectively at zero current. The measuredsignal is a function of the activity of such species insolution that influence the potential at the indicator


electrode (see equation 5.9.1). A broad range ofion-selective electrodes (ISE) has been developedby placing a separating membrane on the tip of amore or less conventional second reference elec-trode. The potential measured is the difference ofpotential across this membrane which is influencedby the activity of the species on either side of themembrane. If the activity of the ionic species xwith the charge nx remains constant in the mem-brane the measured potential difference is

E − E0 = �E = (RT /nxF) ln axb (5.9.7)

Therefore, a potential variation of 59/nx mVshould be observed per decade of variation in bulkactivity ax

b at 298 K. In this way, activities typi-cally in the range of 10−1 –10−6 mol dm−3 can bemeasured. Recently, much lower detection limitsof the order of 10−11 mol dm−3 have been claimedby metal ion buffering in the internal referencesolution of polymer membrane ISEs [53]. A per-fect ISE would respond to only one kind of ionicspecies in solution containing any ion at any con-centration. In practice, this goal cannot be achievedand one has to consider interference effects of otherions, as quantified by their selectivity coefficients.

Various recommended methods have been pro-posed for the calibration of ISEs [54] and for thedetermination of selectivity coefficients [55, 56].

Ion-selective electrodes can be classified accord-ing to their membrane characteristics as shown in

Figure 5.9.3. Examples for types of ISEs are sum-marized in various review articles [57] and IUPAChas reported recommended nomenclature [58].

2.3.2 Fixed potential methods – amperometry

The simplest dynamic electroanalytical techniqueis the application of a fixed potential to an appro-priately chosen working electrode and measure-ment of the current due to electrolysis of theanalyte. If this potential is conveniently chosenor steady-state convection in the measuring cell isemployed, the magnitude of the current is directlyproportional to the concentration of the speciesof interest. This technique is also termed amper-ometry and plays a role as flow-stream detectiontechnique for speciation analysis (see below).

2.3.3 Voltammetry

More information from dynamic potential-control-led measurements is available if the potentialis changed in a well-defined manner during theexperiment. The measurement of correspondingcurrent–potential curves is called voltammetry(derived from volt–ampere–metrology) and canbe exploited to characterize electrochemical sys-tems also qualitatively as well as for increasing theselectivity. It should be noted that polarography

Ion-selective Electrodes with

solid membrane liquid (polymer)membrane

glassmembranee.g. pH electrode

crystallinemembrane

single crystalmembranee.g. F − electrode

polycrystallinemembranee.g. S2− electrode

ion-exchangemembranee.g. Ca2+ electrode

carriermembranee.g. K + electrode

multilayer ISE: - gas sensing ISE- enzyme ISE

e.g. CO2 electrodee.g. urea electrode

Figure 5.9.3. Ion-selective electrodes classified according to their membrane composition.

436 DETECTION

is frequently used in the literature instead ofvoltammetry for current–potential measurements.But the term ‘polarography’ is recommended nowfor the specific case of voltammetric measure-ments at working electrodes with constantly orperiodically changing surfaces, mainly the drop-ping mercury electrode (DME). In the followingthe general term voltammetry includes polarogra-phy, and selected voltammetric techniques withanalytical importance for speciation are brieflyintroduced. The characteristic parameters for thesemethods for reversible and irreversible systems aretabulated in the literature [7, 18].

2.3.3.1 Cyclic voltammetry (CV)

Cyclic voltammetry is perhaps the most versatileelectroanalytical technique for the study of elec-troactive species in quiescent solution. It can beused to observe rapidly their chemical reactionsand redox behavior over a wide potential range,i.e. in an extended energy region.

The excitation signal for CV is a linear poten-tial scan (or sweep) with a triangular waveform, asshown in Figure 5.9.4 together with a correspond-ing response curve. The important parameters ofa cyclic voltammogram are the magnitudes of theanodic peak current (ipa) and cathodic peak cur-rent (ipc) as well as the anodic peak potential (Epa)and cathodic peak potential (Epc). It should bementioned that the measurement of peak currentsis often complicated by problems in establishingcorrect baselines, particularly for more compli-cated systems.

The data for the oxidation and reduction peaksare suited to characterize redox reactions andpartially also of preceding or following chemicalprocesses. If the species undergo a reversibleelectrochemical reaction at the electrode withoutany chemical complications, i.e. electron transferoccurs very fast with respect to the other stepsespecially mass transport, the measurement ofEpa and Epc allows the estimation of the formalredox potential

E0′ = 12 (Epa + Epc) (5.9.8)

t

−E

E2

E1

(a)

(b)

E1E2 −E

i

Epa

Epc

Ox + e Red

Red Ox + e

Figure 5.9.4. Cyclic voltammetry: principal excitation (a) andresponse (b) curves.

Diagnostic criteria for such reversible processesare fulfilled if

�Ep = Epa − Epc = 59 mV/n

at T = 298 K (5.9.9)

ipa/ipc = 1 (5.9.10)

ip ∝ v1/2 (5.9.11)

where v represents the potential scan rate (dE/dt).In addition, the peak potentials should be

independent of the scan rate. The peak cur-rent for a reversible couple at a normal-sizedplanar electrode is described (at 298 K) by theRandles–Sevcik equation:

ip = (2.69 × 105)n3/2AD1/2v1/2c (5.9.12)


where ip is in amperes, the diffusion coefficient D

is in cm2 s−1, v is in V s−1 and the bulk concentra-tion of the reactant c is in mol cm−3. Despite thelinear correlation between the peak current and theconcentration, cyclic voltammetry is rarely usedfor quantitative analysis. Its detection limit is ofteninsufficient (in the range of 10−5 mol dm−3) andthe accurate determination of ip is complicated bynonlinear baselines.

In practice, most systems show some degree ofirreversibility in their CV. As a result of a slowerelectron transfer or coupled chemical reactions,the oxidation and reduction signals are diminished,broader peaks appear, the peak potential differenceincreases and the most marked feature of a CV ofa totally irreversible system is the absence of areverse peak. The equations (5.9.8)–(5.9.12) arenot applicable for such systems.

2.3.3.2 Linear sweep voltammetry (LSV)

This method can be described as one half ofa CV experiment. The excitation signal is alinear potential ramp usually in the range of10–1000 mV s−1 which causes a peak-shaped cur-rent–potential curve as response. The peak currentdepends on the bulk concentration of the elec-troactive species and is described for a reversibleprocess at a planar electrode by equation (5.9.12).In general, detection limits for many systems areabout 10−5 mol dm−3 due to the capacitive current.Application of LSV is limited due to the relativelyhigh detection limit and the often insufficient peakresolution in systems containing multiple redoxspecies. However, LSV still finds important appli-cation in stripping voltammetry (see below). Inaddition this excitation technique is used for thedevelopment of amperometric/voltammetric detec-tion schemes in flow analysis. In this case, so-called hydrodynamic voltammograms are recordedby applying the LSV excitation (Figure 5.9.5(a))to an electrochemical cell with stirred or flowinganalyte solution. The resulting current–potentialcurve (Figure 5.9.5(b)) has a sigmoid shape whichis known from classical DC polarography at thedropping mercury electrode. The mass transportby controlled convection causes the establishment

E

t

(a)

(b)

i

E

i1

E1/2

Figure 5.9.5. Hydrodynamic voltammogram: principal excita-tion (a) and response (b) curves.

of a diffusion layer with constant thickness inthe vicinity of the working electrode. Therefore,the current reaches a plateau that corresponds tothe condition of the analyte being reduced asrapidly as it is transported to the surface by diffu-sion in the solution near the electrode. The poten-tial at which the current is half of its plateau valueis termed the half-wave potential (E1/2) and theplateau current is termed the limiting current il.E1/2 and il are the two paramount parameters inthe analytical application of electrochemistry. il isused for quantitation by means of its relationshipto the bulk concentration of the analyte

il ∝ c (5.9.13)

and E1/2 is used for qualitative identificationthrough its relationship to the formal electrode

438 DETECTION

potential E0′ of the redox couple

E1/2 = E0′ − (RT /nF ) ln (DOx/DRed) − Eref

(5.9.14)

where DOx and DRed are the diffusion coefficientsfor the oxidized and reduced forms of the couple,respectively, and Eref is the potential of thereference electrode.

2.3.3.3 Normal (NPV) and reverse (RPV)pulse voltammetry

NPV can be described as a combination ofchronoamperometric measurements which registeri–t curves of potential step experiments. Inchronoamperometry the electrode potential is oftenimmediately changed from a region without anyelectrode reaction to the diffusion-limited rangeand the resulting current–time response at a planar

electrode in quiescent solution is given by theCottrell equation

i = nFAD1/2c/(π1/2t1/2) (5.9.15)

For NPV increasing potential step heights areapplied as indicated by the excitation signal inFigure 5.9.6(a). A typical normal pulse voltammo-gram exhibits a sigmoidal current–potential wave(Figure 5.9.6). The current is measured only for afew milliseconds (tm) at the end of each pulse toreduce the capacitive current. The initial potentialEi is normally fixed in a range without faradaicreactions and the pulse time tp should be short(commonly about 50 ms). The sampled current onthe plateau il can be calculated for a reversiblereaction according to

il = nFAD1/2cb/[π1/2(tp − tm)1/2] (5.9.16)

(a)

(b)

−E

t

tm

i

−E

−E

t

tm

i

+E

Figure 5.9.6. Principal excitation and response curves for normal pulse (a) and reverse pulse (b) voltammetry.


For polarographic experiments the electrode is heldat a base potential at which negligible electrolysisoccurs for most of the lifetime of each mercurydrop. After a fixed waiting period following dropformation the potential is rapidly switched to avalue E for ca 50 ms and the current is samplednear the end of this pulse time. This potential cycleis repeated with successive mercury drops withE being shifted progressively to more reducingvalues. If the waiting time between the pulsesis chosen properly (typically 1–4 s) to allowrestoration of the initially uniform character of theconcentration distribution in solution at Ei, thelimiting current for NPV is larger than that forsampled DC polarography by a factor of about 6.Therefore, detection limits for NPV are in therange of 10−6 mol dm−3.

In reverse pulse voltammetry (RPV) the basepotential is selected to be in the limiting currentregion for the reduction (or oxidation) and increas-ingly oxidizing (or reducing) values are appliedduring the pulse time (Figure 5.9.6(b)).

NPV at the DME is very sensitive to adsorptionprocesses, which give rise to maxima at theonset of the polarographic wave, the magnitudeof which increases with decreasing pulse time.This phenomenon and its use as a tool fordetecting the presence of adsorbing substanceshave been described in detail [59]. In contrast,the current measured at the DME by RPV isless susceptible to adsorption of electroinactiveligands because at extremely reducing potentialvalues the diffusion-limited current depends onlyon the properties of the bulk solution and atextremely oxidizing potentials it depends only onthe properties of the amalgam [60]. Nevertheless,the current–potential curves of RPV can also beaffected by adsorption [61] and thus caution shouldbe exercised in application of the DeFord–Humeapproach for calculation of stability constants fromsuch experiments (Section 3.1).

2.3.3.4 Differential pulse voltammetry (DPV)

An improvement of the if/ic ratio for sim-ple electrode reactions can be achieved withdifferential pulse voltammetry. The excitation

signal (Figure 5.9.7(a)) consists of a staircase (orramp) potential with a pulse train. In contrastto the former methods the output is now thedifference between individual currents measuredbefore the pulse is applied and at its end, respec-tively. Therefore, the response is a peak-shapedsignal (Figure 5.9.7(c)). Typical parameters forDPV measurements of simple faradaic reactionsare pulse heights (�pE) between 10–100 mV, apulse width (tp) of approximately 50 ms, timesbetween pulses of 0.5–5 s and ‘scan rates’ between1 and 10 mV s−1.

The peak current ip can be estimated for areversible faradaic reaction according to

ip = nFAc(D/πtp)1/2 1 − σ

1 + σ(5.9.17)

with

σ = exp(nF�pE/2RT ) (5.9.18)

The DPV peak potential Ep lies close to thevoltammetric half-wave potential E1/2 for small�pE:

Ep = E0′ + RT/nF ln(Dred/Dox)1/2

− �pE/2 = E1/2 − �pE/2 (5.9.19)

The �pE values of ca 50 mV used in most prac-tical analyses result from a compromise betweenmaximum peak current (ip increases with �pE

according to equation 5.9.17) and sufficient peakresolution (peak width increases as the pulse heightgrows larger). The detection limit of DPV is oftenof the order of 10−7 mol dm−3. The capacitivebackground in DPV is flat in the range −0.2 to−1.0 V versus SCE due to the fact that the doublelayer capacity does not vary much in this range.Therefore, DPV curves are often easier to evaluatethan wave-shaped curves.

It is important to note that the concentration-dependent signal, i.e. the peak current, canbe significantly lower for irreversible reactionsthan that predicted by equation (5.9.17). Suchprocesses are causing also broader peaks andequations (5.9.17)–(5.9.19) are not applicable.One has to consider that the time scale ofDPV experiments is usually shorter than that for

440 DETECTION

(a) (b)

(c)

∆pEE

t

tm1 tm2

E

t

1/f

ESW

tm1

tm2

E

∆i

Ep

ip

Figure 5.9.7. Excitation signals for differential pulse (a) and square wave (b) voltammetry and the principal response curve (c).

linear sweep voltammetry. Therefore, the degreeof reversibility toward the pulse methods maydiffer from that shown toward LSV and kineticeffects can play a more important role for DPVmeasurements.

2.3.3.5 Square wave voltammetry (SWV)

The excitation signal of this large-amplitude differ-ential technique consists of a symmetrical squarewave superimposed on a staircase, as shown inFigure 5.9.7(b). The current is sampled at theend of the forward pulse as well as at the endof the reverse pulse during each square wavecycle and the difference between the two val-ues is plotted versus the staircase potential. Theresulting response curve is comparable to DPV

(Figure 5.9.7(c)) and as for DPV the peak heightof the signal can be increased by application oflarger amplitudes of the excitation function, whichcauses on the other hand also a peak broadening.

Most experimental parameters are comparablewith those of DPV. However, the potential scanrate determined by the square wave frequency(5–500 Hz) is much faster. Detection limits forfavorable systems are usually slightly better, i.e.in the range of 10−8 mol dm−3. SWV is oftenmore sensitive than DPV because both forward andreversed currents are measured.

From an analytical point of view a majoradvantage of SWV is its speed, which allowsthe recording of a complete voltammogram withina few seconds. Certainly, the shorter measuringtime has only a negligible effect on analysis time


and sample throughput for practical applications,because the total analytical procedure requiresmuch more time than the actual measurement.However, there are several analytical aspects,where faster voltammetry is desired. The workingelectrode surface is exposed for less time tothe detector reactions as well as to interferenceprocesses, which could result in less pronouncedsurface alterations especially at solid electrodes.Such a fast technique can be used to obtainthree-dimensional i–E–t plots in flow injection orchromatographic systems during the residence timeof the injected sample in the flow cell detector.Moreover, a kinetic discrimination can be achievedagainst less reversible interfering reactions such asthe oxygen reduction. The theoretical backgroundfor SWV is much more well established than thatfor DPV [62].

2.3.3.6 AC voltammetry (ACV)

There are several electrochemical methods thatare based on the concept of impedance. Foranalytical purposes the most important is ACvoltammetry, where a constant sinusoidal ACpotential is superimposed upon a DC potentialramp (Figure 5.9.8(a)). Typically, the AC potentialhas a frequency f of 10–1000 Hz and a peak-to-peak amplitude of 4–20 mV. The role of theDC potential is to set mean surface concentrationsfor both redox states of the reactant, which thenface an excitation signal of low amplitude. Thecurrent flowing through the cell contains bothAC and DC components. The registered responsesare either the total AC current as a function ofthe DC potential, as shown in Figure 5.9.8(b),or preferably the current components in phase(or out of phase) with the AC potential as afunction of the DC potential. The latter methodis called phase-selective AC voltammetry andis based on the different electrical behavior ofohmic resistors and capacitors in AC circuits.Therefore, it allows the effective discriminationbetween faradaic (in phase) and capacity (outof phase) currents and leads to detection limitsof the order of 5 × 10−7 mol dm−3 for reversiblesystems.

(a)

(b)

iac

EEp

ip

E

t

1/f

∆E

Figure 5.9.8. AC voltammetry: principal excitation (a) andresponse (b) curves.

The bulk concentration can be determined fromthe peak current of the response curve. Thecorresponding correlation for reversible systems is

ip = n2F 2A(2π fD)1/2 �Ec/4RT (5.9.20)

Equation (5.9.20) is only valid if the AC timewindow is much shorter than the DC one and small

442 DETECTION

AC amplitudes are applied, i.e. �E should be lessthan 10/n mV.

It is important to note that AC techniquesare very sensitive with respect to slow electron-transfer kinetics resulting in smaller signals forsuch processes than predicted by equation (5.9.20).This feature can, however, be exploited for kineticdiscrimination against certain electroactive inter-ferences such as dissolved oxygen. It may beadvantageous for the analysis of fast-reacting ana-lytes to measure the AC current at a frequencyof 2f instead of the fundamental. This so-calledsecond-harmonic AC voltammetry can providebetter separation from capacitive current lead-ing to lower detection limits for fast electron-transfer reactions.

The AC voltammetry can also be used to ana-lyze electroinactive compounds which are sur-face active by measuring their adsorption sig-nals [63–65]. Such tensammetric measurementsshould be performed in a potential range withoutinterferences from faradaic reactions.

2.3.3.7 Anodic stripping voltammetry (ASV)

Voltammetric stripping (SV) techniques are nowwidely recognized as powerful tools for trace anal-ysis of metal ions and certain organic compoundsin solutions. They offer excellent detection limits(down to 10−12 M for certain metals!) coupled withinherent species selectivity. These features arisefrom the two-step nature of this technique: precon-centration of the analyte at the electrode, followedby generation of the analytical signal by strippingfrom the electrode.

The trace determination of many metal ionscan be performed by anodic stripping voltamme-try (ASV). Figure 5.9.9 illustrates determinationof Pb2+ at a HMDE by ASV. The accumulationprocess comprises the reduction of metal ions atconstant potential for several minutes, mostly facil-itated by convection. Thus, Pb2+ is electrochemi-cally extracted as elemental lead into the mercuryelectrode forming an amalgam. The resulting leadconcentration in mercury is substantially higherthan that in the solution of metal ions being ana-lyzed because of the much smaller electrode vol-ume in comparison with the solution volume. After

0−0.6 −0.4 E/V

50

100

150

i/nA

−E

stirring offstirring ont

(a)

(b)

Figure 5.9.9. Anodic stripping voltammetry: excitation signal(a) and measuring curve (b) for the determination of Pb2+ atan HMDE with DPV detection.

discontinuing the stirring, the potential is changedto more positive potentials by linear sweep, dif-ferential pulse or square wave voltammetry. Thiscauses the oxidation of the amalgamated lead backinto the solution registered by the current peakin Figure 5.9.9(b). The peak height is determinedby the concentration of Pb in the mercury elec-trode which is in turn proportional to the amountof Pb2+ in the sample if appropriate experimen-tal parameters such as electrode area, depositiontime, stirring etc. are held constant and there areno secondary effects such as peak broadening duethe presence of heterogeneous ligands (see below).The necessary accumulation time depends on theanalyte concentration and can reach up to 20 min atthe 10−9 mol dm−3 level. With this ASV approachabout 15 amalgam-forming metals can be deter-mined, including Tl, Cd, Zn, Cu, Bi, In and Ga.


The trace analysis of other metal ions (Hg, Au,As, Se) can be performed after their electrolyticdeposition as a metal film on bare solid electrodesmade from carbon or gold.

2.3.3.8 Stripping chronopotentiometry (SCP)

Stripping chronopotentiometry (also referred to as‘potentiometric stripping analysis’) [9], analogousto SV, is a two-step technique. The first, deposi-tion, step is identical to that for ASV, but reoxida-tion of the accumulated metal is then achieved byapplication of a constant oxidizing current or byconstant flux of a chemical oxidant (usually Hg2+or O2). The electrode potential is recorded as afunction of time during the stripping step, and theanalytical signal is the time taken for reoxidation(the transition time, τ ). The analytical signal canbe enhanced simply by application of a smallerstripping current, or by a lower flux of chemi-cal oxidant (a major advantage over voltammetricstripping techniques), with the practical detec-tion limit being determined by the presence ofredox active impurities, e.g. dissolved oxygen, inthe sample (which represents the major limitationcompared to voltammetric techniques, particularlySWV and ACV). The instrumentation required forSCP is simpler than that for voltammetric methods.

There are many claims in the literature (basedon empirical observations) that SCP is less suscep-tible to interferences from adsorption of organiccompounds on the electrode surface than isSV [66]. A more rigorous understanding of thisbehavior has recently been developed: determi-nation of transition times from the area underpeaks in the inverse of the time derivative ofpotential (dt/dE) versus E plots is the correctstrategy for effective elimination of charging cur-rents [12]. The area under the baseline corre-sponds to the time necessary for charging, whichis thus effectively eliminated from the analyti-cal signal by this approach. Some workers reportpoor baselines for dt/dE versus E plots, ascribethis to ‘adsorption’, and apply some arbitrary‘background correction’ protocol [67]. However,correct interpretation requires knowledge of thestripping time regime under which measurements

are made. When high stripping currents are usedwith a HMDE, the accumulated metal is not com-pletely stripped from the electrode during thetransition plateau and poor baselines result fromthe ongoing faradaic processes that follow thisincomplete depletion. When experimental condi-tions are chosen such that measurements are madein the complete depletion regime, a limit of detec-tion directly comparable with that for DP-SVis achieved with the advantage that discrimina-tion against capacitive charging is achieved by anapproach that avoids the adsorption complicationsassociated with pulse SV waveforms [12].

2.3.3.9 Adsorptive stripping voltammetry (AdSV)

In recent years a fast growing number of nonelec-trolytic accumulation procedures using the adsorp-tion of the species of interest at the electrodesurface has been developed for metal ion com-plexes as well as for an increasing number oforganic compounds. Tabulations of experimentalconditions for a range of elements have beenreported [68, 69]. This so-called adsorptive strip-ping voltammetry (AdSV) allows the trace deter-mination of metals such as Al, Fe, Co, Ni, Mo,V, Cr, Ti, U, La, which cannot be measuredby ASV due to nonfavorable reversible reactionsor the absence of amalgam formation at mer-cury electrodes. The technique involves addition ofsurface-active ligands (La) with an affinity for thespecies of interest M, followed by accumulationof the resulting MLa complexes as a monolayeron the electrode surface (either in open-circuit,or by application of an appropriate accumulationpotential), and finally quantification (stripping) bya reducing potential scan (for which a range ofwaveforms can be employed, e.g. DPV, SWV).The analytical signal obtained during the strippingstep may arise from reduction of either the ele-ment or the ligand in the adsorbed complex, orfrom catalytic effects [70–72]. This approach canalso improve the analytical signal for other met-als which are conventionally measured by ASV.AdSV measurements are particularly susceptibleto interference from other surface-active materialin the sample which will compete with the chelatecomplexes for coverage of the electrode surface.

444 DETECTION

Note that the term ‘cathodic stripping voltam-metry (CSV)’, is often used to denote AdSV.We recommend that this terminology be avoidedbecause it can cause confusion with other strip-ping methods which use a change of the oxidationstate during the analyte accumulation followed byits reductive determination [9].

Application of SV and SCP to measurementsof very low concentrations requires use of spe-cial guidelines for trace analysis. All precautionsshould be obeyed to prepare carefully the standardsand to avoid contamination of the sample or lossof analyte, e.g. by adsorption at cell walls. More-over, such two-step procedures at the trace levelare subject to several interferences. The adsorptionof foreign surface-active compounds can influencenot only the accumulation during the adsorptivestripping mode, but also the measuring step of theASV for metals, particularly when pulse modesare employed. Complications can also arise fromthe formation of intermetallic compounds in the

electrode or electroactive interferences with com-parable redox potentials to the analyte. It hasbeen frequently shown that such problems can beavoided by the careful development of an appro-priate stripping method, for which both electro-chemical and chemical parameters were optimizedtaking into account the sample matrix [14, 73, 74].

2.3.4 Electrochemical detection in liquidchromatography and flow-injection analysis

Electrochemistry offers a number of advantagesfor the trace determination of certain species inliquid chromatography (LCEC) and flow-injectionanalysis (FIAEC). One commonly used electro-chemical detector is a thin-layer cell in whichthe working electrode is positioned in a thinchannel through which the mobile phase flows.A detector of this type is shown schemati-cally in Figure 5.9.10, which illustrates also the

RE

WE

CE

solution

flowing

Ox

Red

OxOx

Red

RedRed

Red

Red

Red

Red

Figure 5.9.10. Scheme of an electrochemical thin-layer cell and the corresponding flow-through detection.

PRINCIPLES OF SPECIATION BY ELECTROANALYSIS 445

detection of a ‘plug’ of analyte Red that haseluted from the column being detected by oxi-dation to Ox as it sweeps over the electrode inthe thin channel. For maximum sensitivity thepotential of the electrode is held on the limit-ing current region of a hydrodynamic voltam-mogram (see Figure 5.9.5) for oxidation of Red.The resulting chromatogram shows a peak currentresponse for the detection of Red. Its concentra-tion can be quantified by measuring either the peakheight or integrating the peak and measuring thecharge, both of which are proportional to the bulkconcentration.

Thin-layer detectors of this type are mostlyoperated as fixed potential (amperometric) devicesand the current is measured as a function oftime. Typically about 5 % of the analyte is elec-trolyzed as it sweeps past the electrode. Amper-ometric detectors can achieve detection limits aslow as 10−8 –10−9 mol dm−3 of injected analyte.Coupled with an injection sample of 10 µL, theycan detect as little as 10−16 mol (or 0.1 fmol).Amperometry is advantageous in comparison tothe voltammetric techniques because the charg-ing current is minimized by operating at afixed potential.

Interest in voltammetric detectors in which thepotential is scanned to provide a voltammogram,however, is increasing because they provide quali-tative information about the species being detectedin the form of the half-wave potential. This canbe especially important in speciation studies asan additional parameter to be used in conjunctionwith chromatographic retention time for qualita-tive identification of a species. As for voltammetrictechniques, the development of acceptable voltam-metric detectors cannot be based on linear scanvoltammetry due to the poor sensitivity of thismode. Pulse techniques such as differential pulseand square wave voltammetry are the most com-monly used methods to diminish charging-currentcontributions. Recently, microelectrodes have beenapplied in electrochemical detectors to alleviatecharging-current problems [75]. Coulostatic detec-tors have also been designed to obtain voltammet-ric information [76].

3 PRINCIPLES OF SPECIATIONBY ELECTROANALYSIS

3.1 Thermodynamic aspects

The most straightforward information for deter-mination of thermodynamic equilibria parameters(K values) is provided by analytical signals fromISEs (i.e. potentiometric E values). In this tech-nique equilibrium is assumed to exist throughoutthe measurement system (i → 0) and the analyticalsignal depends only on the free metal ion activity(which is fixed by its equilibrium across all bind-ing sites in the system). Voltammetric signals arealso sensitive to K , but since the signal depends onthe flux of species to the electrode surface, correctinterpretation requires concomitant considerationof kinetic factors (D, kd) (see below).

ISEs provide a direct measure of the freehydration ion activity, {M} = cM,T γ/α, where γ

is the activity coefficient for M, and α its degreeof complexation (= cM,T/cM). α is related to thestability constants for metal complex species; forthe simplest case of a single well-characterizedsimple ligand forming a single complex, ML,α = 1 + KcL. Determination of α, and thus K

values, requires knowledge of cM,T, γ , and theEo and slope (calibration) parameters for theISE (equation 5.9.7). At constant ionic strength,equation (5.9.7) can be written as

�E = s log cM,T − s log α (5.9.21)

where s = 2.303RT/nF .Complexation parameters are usually deter-

mined from titrations of ligand with metal (or viceversa) [18]. For increasing cM,T at constant cL,T,α for each point in the titration can be calculatedfrom equation (5.9.21). The most useful part of thetitration curve is the region in which cM,T � cL,T,under these conditions cL ≈ cL,T and thus α =1 + KcL,T for the simple case. In the case of het-erogeneous ligands (typical of real-world samples),various data interpretation models are applied toextract K values (or their distribution) from thetitration curves [14]. Despite their limitations (inparticular interference from other ions and low sen-sitivity), ISEs are a useful tool for studying metal

446 DETECTION

ion complexation, e.g. the influence of parameterssuch as the metal ion loading (cM,T/cL,T) on α,which is an important factor for natural hetero-geneous ligands, is more readily tested with ISEpotentiometry than via other methods [77].

For voltammetric methods and labile complexes(see Section 3.3 for definitions), α can be calcu-lated from current and potential values obtained inthe absence and presence of ligand via the DeFordand Hume expression

ln α = nF

RT(E1/2 − E1/2

L) + ln

(il

ilL

)(5.9.22)

with ilL/il =

√D/DM on macroelectrodes (see

Section 3.2.3 for definitions), where il denotesthe polarographic limiting current, E1/2 the half-wave potential, superscript L denotes values inthe presence of ligand(s) and D is the meandiffusion coefficient for M and the complex. Underconditions where each ligand concentration is suchthat cL,T � cM,T, α is related to the stabilityconstants, Ki , for metal complexes MLi , by

α = 1 +∑

i

KicLi

and D is given by: D = DM/α + DMLifi , where

fi is the fraction i of complex MLi with respectto cM,T. These expressions are directly applicableto other voltammetric techniques by substitution ofEp and ip for E1/2 and il.

For a system containing M and a single metalcomplex species, ML

ilL

il=

√1

α+ DML

DM

(α − 1

α

)(5.9.23)

(For microelectrodes (Section 3.2.3) the sameequation holds but without the square root.)

There are three useful limiting cases of equation(5.9.23):

(i) small simple ligands for which DML =DM, thus il

L/il = 1, and ln α = nF(E1/2 −E1/2

L)/RT

(ii) α � DM/DML typical of strong complexa-tion by small ligands. In this case il

L/il =√DML/DM

and α can be computed from equation (5.9.22)if DML is known.

(iii) DM/DML � α − 1 typical of weak complex-ation by large ligands. In this situation ML isessentially immobile and the current is deter-mined primarily by the free metal ion, and

ilL

il=

√1

α=

√cM

cM,T

Application of this methodology has been dis-cussed in more detail for different types of metalcomplex species [18, 19]. The DeFord–Humeexpression (equation 5.9.22) has been extended tothe case for heterogeneous ligands [78]; variousdata interpretation models are applied to extract K

values (distributions) from the voltammetric curvesfor such systems (e.g. Freundlich isotherm).

Many publications on element speciation as-sume that the system under consideration is atequilibrium even at the electrode surface, i.e. thatthe complexes are labile (see Section 3.2) How-ever, this depends strongly on the measurementtime scale of the technique and must be tested andverified in each case [14, 75]. A more completeunderstanding of element speciation must incor-porate knowledge of the interconversion rate ofspecies with respect to diffusion rate, and thus astudy of chemical and physical kinetic propertiesis required.

3.2 Kinetic aspects

With the exception of potentiometry, electroan-alytical techniques are nonequilibrium (dynamic)techniques. Thus sound theoretical concepts arerequired to relate the analytical signals to theunderlying properties of the system under study.The important concepts involved in voltammet-ric measurements are redox reversibility, chemicallability, and physical mobility.

Reversible behavior results when the chargetransfer rate, k0, is much higher than the dif-fusion rate which corresponds to k0 � D/δ

for macroelectrodes (linear diffusion) and k0 �D/r for microelectrodes (spherical diffusion,r = electrode radius). Interpretation of data is


simplified for reversible systems and this conditionis assumed throughout the following discussion.Tests for reversibility using various electrochemi-cal methods are tabulated (see also Section 2.3.3.1)[18, 19].

The dynamic behavior of metal complexes atthe voltammetric interface has been discussed indetail by several authors [18, 19, 79–81], andonly a brief outline is included here. Complexesare considered as dynamic when (i) their mobilityis high enough, i.e. they can move towards theelectrode surface by diffusion, at a rate that isnonnegligible compared with that of free M, and(ii) their lability is also high enough, i.e. they canassociate/dissociate a large number of times duringtheir diffusive transport. Since the latter dependson the measurement time scale of the techniqueused, lability also depends on this time scale.Based on these concepts, the following definitionsare used:

• Complexes are said to be immobile, when theirdiffusion rate (mobility) is negligible comparedwith that of the free metal ion, M.

• Complexes are said to be inert when the numberof association/dissociation steps is negligibleduring the time scale of interest (usually the timeof diffusion through the diffusion layer, whichalso corresponds to the measurement time ofthe technique).

• Complexes are said to be labile, when theydissociate/associate a very large number of timeduring their diffusive transport to (or from)the electrode.

• Complexes are semi-labile when they exhibitbehavior borderline between inert and labilecomplexes.

• Complexes which are both mobile and eitherlabile or semi-labile are called dynamic.

It is most important to note that only dynamiccomplexes can contribute to the measured voltam-metric current.

The above definitions can be defined on amathematical basis as follows. Consider the sim-ple reaction

M + Lka←→kd

ML (5.9.24)

At one extreme, the inert situation arises when k′at ,

kdt � 1, where t is the measurement time. Thismeans that complex dissociation during this timeis negligibly small and the flux (i.e. the current)is thus controlled by the diffusion of free metalspecies in the bulk solution.

The flux J (t) in solution containing free metalions, M, and a single dynamic complex, ML, hasbeen computed in the presence of an excess ofligand L, for semi-infinite diffusion as the soletransport process towards the electrode and thecondition that this latter is a perfect sink for M(i.e. the potential is negative enough to ensure acomplete reduction of M). For a dynamic systemat a macroelectrode, J (t) is given by [82–84]:

J (t) = kd1/2DM

1/2c∗T(1 + εK ′)3/2

ε3/2K ′(1 + K ′)

× exp(�2t)erfc(�t1/2) (5.9.25)

where c∗T = c∗

M + c∗ML, ε = DML/DM, and

� = kd1/2(1 + εK ′)

ε3/2K ′(1 + K ′)1/2(5.9.26)

k′a = kac

∗L, and K ′ = k′

a/kd = Kc∗L.

A lability criterion (based on the magnitude of theterm �t1/2) is used to describe the contributionof metal complex species to the overall flux ofmetal at the electrode surface. The lability of metalcomplex species will decrease with increasingvalues of K ′ and decreasing values of ε for a givenmeasurement time scale. Two kinetic limiting casescan be identified:

(i) �t1/2 � 1, labile case. The flux is diffusioncontrolled and reduces to:

J (t) = D1/2

c∗T

(πt)1/2(5.9.27)

It corresponds to purely diffusion controlled cou-pled transport of M and ML (i.e. the asso-ciation/dissociation kinetics are fast comparedto diffusion).(ii) �t1/2 � 1, nonlabile case. The flux is entirely

controlled by the chemical kinetics of dissoci-ation of ML (i.e. electron-transfer kinetics are

448 DETECTION

slow relative to diffusional transport), and forlarge K ′(εK ′ � 1), the flux is given by:

J (t) = kdcT∗(

DM

k′a

)1/2

(5.9.28)

The case of inert complexes can be seen as thelimit of equation (5.9.28), when J (t) is negligiblysmall compared to the equivalent flux which wouldbe obtained in absence of L.

This approach has been extended to analysis ofthe steady-state case (δ as the variable) [79, 85],chemically heterogeneous systems that involve a

dissociation rate constant distribution [86], and tomicro uptake surfaces at which spherical diffusionmust be considered [79].

3.2.1 Dependence of measured parameterson lability

Characteristic changes in the current–potentialcurves are observed according to the lability ofthe metal complex species being measured relativeto the curves for free metal ions. This is shownschematically for DC polarography (or NPV) andfor DPV (or ACV) in Figure 5.9.11.

i = iL

E1/2 E1/2 Ep EpL

i

−E −E

(a)

(b)

(c)

L

E1/2 E1/2L

∆E1/2

iL

i

i

iL

i

i

ip = ip

ip

ip

ip

ip

(A)(B)

∆E1/2 ∆Ep

Ep EpL

∆Ep

−E

E1/2 = E1/2 Ep = EpL −E−E

−E

L

L

L

L

Figure 5.9.11. Comparison of i = f (E) curves obtained by (A) DC polarography or NPV and (B) AC polarography or DPV,for (a) labile, (b) nonlabile, and (c) inert metal complex species (solid lines) with those for the free metal ion (dashed lines).


The corresponding curves for other electroan-alytical methods follow a similar pattern [18]. Asexplained above, inert complexes do not dissociateduring the measurement time and thus do not con-tribute to the flux (and thus current). The currentmeasured for such systems is thus proportional tothe free metal ion concentration and there is nochange in the potential characteristics. In contrast,dynamic fully labile complexes do contribute to themeasured current, the potential is shifted relativeto the ligand-free case and the magnitude of thisshift is a measure of the thermodynamic stability ofthe complexes. Furthermore, for labile complexes,with DML ≈ DM, the limiting current is the samein the presence and absence of ligand; a reductionis observed when DML < DM. Dynamic nonlabilecomplexes display intermediate behavior, i.e. somereduction in limiting current (but less than that forinert complexes) and some shift in E (but less thanthat for labile complexes).

3.2.2 Dependence of lability on measurementtime scale

As discussed above and shown in particular inthe lability criteria expression, the lability of agiven complex is not an intrinsic property, butrather an operationally defined concept whichdepends on the measurement time scale of theanalytical technique employed. Typical time scales,tm, for electroanalytical techniques are shown inFigure 5.9.12.

3.2.3 Lability at microelectrodes

In recent years there has been an upsurge inthe application of microelectrodes [36] to element

speciation measurements. Rational interpretationof experimental data obtained at these electrodesrequires a sound theoretical understanding ofprocesses occurring at the microsurface. Transitionfrom a macro- to a microelectrode influences boththe nature of the flux to the surface and alsothe extent of lability (and thus bioavailability,Section 5.2) of metal complex species. Below isa brief discussion of the application of the labilityconcepts to microsurfaces. More details are givenin recent publications [87, 88].

Diffusion regimes range from a linear profile atmacroelectrodes, to spherical one at a microelec-trode. The limit between these regimes is deter-mined by the size of the electrode radius relative tothe diffusion layer thickness, i.e. the ratio r/δ. Thethickness of the diffusion layer in quiescent solu-tion at a planar electrode is given by δ =

√πDt ,

while under convective conditions (electrode rota-tion or solution stirring at rotation rate ω) δ �(D/ω)p (p ≈ 0.3–0.5).

Irrespective of the electrode size, the current(or metal flux) to the electrode surface at constantpotential is given by

i = nFADcb

(1

δ+ 1

r

)(5.9.29)

For a macroelectrode, i.e. a planar electrode atwhich linear diffusion occurs, 1/r � 1/δ; i is thusprimarily controlled by δ and therefore sensitiveto solution convection or time, since δ depends ontime in purely diffusive transport.

In contrast, for microelectrodes 1/r � 1/δ,therefore i is determined principally by r andconsequently is independent of both time andsolution convection (a major advantage for in situapplications, Section 4).

−3 −2 −1 0 1log tm/s

reduction step inSV and SCP

NPV and DPV

AC voltammetry

cyclic voltammetry

Figure 5.9.12. Effective measurement times for electroanalytical techniques.

450 DETECTION

Because lability is relative to transport rate, anddiffusive transport depends on microelectrode size,it follows that lability of complexes will them-selves depend on electrode size. This aspect hasbeen discussed by de Jong and van Leeuwen [84].

3.3 Considerations for samplepreparation and experimental conditions

3.3.1 Measuring solutions

The electroanalytical methods described above areused for measurements in solution. Therefore, allaspects of the preparation of nonliquid samplesfor speciation analysis, which are discussed inChapter 2.1 of this Handbook, have to be care-fully considered. Moreover, the specific demandsof electroanalysis may give rise to additional prob-lems. Many of the techniques require a sufficientelectrical conductivity of the measuring solutionprovided by ionic components that are not partic-ipating in any electrochemical process influencingthe analytical signal. The addition of such a sup-porting electrolyte, however, can change the orig-inal species distribution in solution. The use ofmicroelectrodes is a way to overcome this problem,since they can perform measurements in solutionwith high resistivity. Indeed, due to their smallsize, the measured current is small, as is the cor-responding ohmic drop. Usually, the pH of themeasuring solution cannot be modified, since oth-erwise, the whole of dynamic complexation equi-libria will be shifted. When the pH can be changed,great care must be taken to choose appropriatebuffers, as they usually also contain potential lig-ands that can change the speciation of metal ionsby complexation or adsorption at interfaces (elec-trodes, cell wall). The complexing ability and sur-face activity of several commonly used buffershave been characterized and compared [89, 90].During the reduction of metal ions, the potentialis sufficiently negative to also reduce the oxy-gen present in the solution. When elimination ofoxygen is not sufficient, its reduction leads to anincrease in pH at the electrode surface that maydrastically change the speciation if the solution

is not sufficiently buffered. Therefore, the sam-ple preparation and manipulation must be designedvery carefully in accordance with the chemicalproperties of the analyte species and the require-ments of the measuring method. As a general rule,sample handling should be minimized.

3.3.2 Adsorption effects

Adsorption of foreign material, in particular col-loidal and macromolecular organic compounds [91],from the sample at the working electrode usu-ally drastically interferes with speciation analy-sis [18, 19]. When the test metal species itself isadsorbed, special signals can arise. Pulse and ACmodes of polarography and voltammetry are use-ful tools to check for the presence of such adsorp-tion processes [59]. The impact of adsorption willdepend on the time scale of the technique, the con-centration of adsorbing material, and the electricalpotential at the interface [18, 92].

Many of the applications of electrochemi-cal techniques to speciation analysis (Section 4)involve measurements in matrices containing sig-nificant amounts of organic compounds that havea propensity for adsorption on electrode surfaces.Various approaches have been proposed to over-come interference from adsorption of organic com-pounds ranging from protective coating of theelectrode surface (Section 2.2.3), to removal ofthe organic matter, e.g. by addition of fumed sil-ica [93]. Most of these methods have some draw-backs, and any methodology which may perturbthe initial solution composition should be avoided.

3.3.3 Avoiding saturation of ligands at theelectrode surface

During the stripping step of ASV, the surface con-centration of oxidized metal ions is far greater thanthat in the bulk solution and thus secondary reac-tions may be induced at the electrode surface [94].An appropriate choice of experimental conditions,including the use of a sufficiently large excessof ligand, is required to minimize the impact ofthese effects [95]. The use of a medium exchange

APPLICATIONS 451

has been proposed to overcome this problem [96],but reproducible results have proved difficult toachieve in practice. The construction of stepwisestripping voltammograms (so-called ‘pseudopo-larography’), in which the magnitude of the strip-ping signal is recorded as a function of depositionpotential, may be a more appropriate approach toobtaining speciation information than attempts tointerpret single ASV measurements [97] providedconditions are carefully chosen to avoid the prob-lem of ligand saturation during the stripping step.

Note that a large excess of ligand over metalalso simplifies data interpretation (Section 3.2).Under these conditions a constant ratio betweencomplexed and free metal can be assumed overthe diffusion layer and thus the mean diffusioncoefficient operating from the bulk solution to theelectrode surface can be taken as constant.

4 APPLICATIONS

During recent years speciation analysis has beenmainly directed at the quantification of either var-ious redox states of an element, its organometallicspecies or metal complexes in equilibrium witheach other. These various analytes of interest pos-sess different chemical stability and volatility thathas to be considered for the selection and optimiza-tion of the analytical procedures. In principle, theideal method for any application is one that can bedeployed in situ with minimal sample perturbation.At present, such measurements are in the minority,but rapid advances in this field are anticipated.

The vast majority of literature to date onthe application of electroanalytical methods todetermination of speciation has been directed atspeciation of metal complexes in solution andreports very empirical data, e.g. percent ‘ASV-labile’ metal, with experimental details often verypoorly defined. Such data have no real meaning inthemselves, and certainly cannot be extrapolated toconditions other than those under which they weredetermined (e.g. to predict the impact of a changein pH, metal ion loading, etc.).

As is evident from the discussion in Section 3,(and see refs [18, 19]) reliable determination of

useful speciation parameters (stabilities of metalcomplex species and kinetic association/dissocia-tion rate constants) from electroanalytical mea-surements requires knowledge of the time scaleand diffusion regime of the technique employed.Furthermore, in real-world samples, the stabilityand rate constant parameters are typically repre-sented by a (continuous) distribution of values.This factor, together with other possible interfer-ences (Section 3.3) means that correct interpre-tation of electroanalytical measurements can bedifficult, and all experimental conditions must befully reported to facilitate utility of results toothers. Despite these difficulties, the sensitivityof electroanalytical techniques to many importantspeciation parameters makes them very powerfultools for many applications.

Examples of applications of electroanalyticalmethods to determination of element speciation invarious media are given below.

4.1 Aquatic systems

There are several reviews on application of elec-troanalytical methods to element speciation inaquatic systems [14, 19, 98, 99]. Redox spe-ciation is an important topic where voltamme-try/polarography and/or ASV could in princi-ple be used to distinguish between Fe(III)/(II),Cr(VI)/(III), Tl(III)/(I), Sn(IV)/(II), Mn(IV)/(II),Sb(V)/(III), As(V)/(III), and Se(VI)/(IV). On theother hand, there are only a few papers dealingwith the determination of organometallic speciesby direct electrochemical detection. For instance,Schwarz et al. [100]. have described the analysisof butyltin species in surface water from a har-bor by AdSV with tropolone. The three butyltincompounds are electroactive and can be detecteddown to 0.5–5 µg dm−3. A further LOD improve-ment by a factor of 10–20 has been obtained bypre-concentration of the analytes at a solid-phaseextraction column. Bond et al. [101] have found aninteraction of the reduction processes of butyltinspecies (without tropolone complexation) at themercury electrode which pointed to the necessityof additional separation steps for speciation analy-sis (see below).

452 DETECTION

ISE. An ISE has been directly applied to thedetermination of the free metal concentration inseawater [102]. Most studies have used ISEs tofollow metal titrations of water samples; variousmodels are then applied to the data for extraction ofapparent binding site affinities and correspondingcomplexation capacities [14, 103].

Laboratory and on-site polarography of O2,Fe(II), Mn(II) and S(−II). It has been shown,more than 20 years ago, that DC polarography,sampled DC polarography, DP polarography andCV can be used on board a ship to determinethe speciation of Fe(II), Mn(II), and S(−II) ineutrophic waters. This subject has been reviewedin several publications [19, 104, 105]. The methodallows determination of the total dissolved S(−II)(which is usually dominated by S(−II) and itsprotonated forms) and the total dissolved Fe(II)and Mn(II) (which in most waters are dominated(95 %) by the aquated Fe2+ and Mn2+ ions) [106].In addition, a special peak occurs in the presenceof adsorbable small polynuclear FeS complexes.The polarography of the Fe(II)/S(−II) system hasbeen studied in detail both in the laboratory andon site [107]. By taking the difference betweenthe colorimetric measurement of Fe(II) and S(−II)after 0.45 µm filtration, and the direct DPP signalsfor these ions, the concentration of colloidalFeS can also be obtained. The formation ofcolloidal FeS is time dependent and can befollowed by this method. It also allows them to bediscriminated readily from the colloidal forms ofMn(IV) and Fe(III) oxides. Detailed concentrationprofiles of these various species and their seasonalevolution with time have been reported by usingthis method and their redox interactions werediscussed [107]. Recently this approach has beenused to measure specifically the free Mn2+ by anin situ voltammetric probe, at depths in the range−80 to −95 m below the surface of Lake Lugano(Tessin, CH), and to discriminate it from colloidalMn upsurged by storm events from the sedimentpresent 2–3 m below [19].

AdSV. AdSV is probably one of the most widelyused methods for measurement of metal com-plexation parameters (stability constants and com-plexation capacities) in natural waters [108]. This

approach is based on competition between theadded ligand and complexants present in the sam-ple for the elements of interest, and therefore can-not be employed in a truly in situ form, although‘on-line’ procedures have been proposed [109,110]. It should be noted that the pH of the sam-ple is often adjusted in order to achieve optimalcomplexation conditions for the added ligand, e.g.the following buffers (at ca. 0.01 mol dm−3 con-centration) have been used in analysis of sea-water samples by AdSV: Tris [111], EPPS [111,112], HEPES [113, 114], HEPPS [115, 116, 117],borate [118, 119, 120], PIPES [121]; and infreshwaters: TES [122], EPPS [123], PIPES [124],HEPPS [125], HEPES [126–128], acetate [129].Any pH adjustment (and reagent addition) willundoubtedly perturb the sample equilibria. Adifferent ligand and/or conditions are usuallyrequired for each element of interest. The ligandsand conditions employed for determination of arange of elements in environmental and biolog-ical matrices by AdSV have been tabulated [69,70, 130].

Interpretation of AdSV data remains veryempirical and data are of questionable value dueto a number of artifacts whose importance isstill poorly identified. For example, it is assumedthat the added ligand reaches equilibrium withthe sample components even though very sta-ble complexes are purported to be detectable byAdSV (log K ≈ 15) which may have very lowrates of association/dissociation, necessitating verylong equilibration times with the added ligand.In addition, many of the ligands used for AdSVare unstable towards oxidation (e.g. catechol) andthus their concentration cannot be assumed to beconstant over the time course of the experiment.An important interference is competitive adsorp-tion of the added complex and natural surface-active sample components, e.g. humic substances,on the electrode surface. This aspect is usu-ally poorly characterized and made worse by thefact that analyses typically involve the use of arange of concentrations of added ligand (to deter-mine the so-called ‘complexation capacity’) suchthat the effective competition is not equal acrossall measurements. These ‘titration’ protocols are

APPLICATIONS 453

routinely used in determinations of metal complex-ation parameters (log K and complexation capac-ity) by AdSV in both seawaters [113, 117, 120]and freshwaters [124, 125, 128]. Considering thisfactor, attempts to develop multielement AdSVprotocols, involving several added ligands, mustbe interpreted with caution [69].

The analytical detection window of AdSVdepends on the concentration of added ligandand its stability constant with the element ofinterest [131]. When this factor is taken intoaccount, the so-called ‘strong’ binding ligandsreported by many workers are seen to be merelya consequence of the heterogeneity of ligands inthese systems [132].

Microelectrodes. Microelectrodes are necessaryfor speciation measurements on low ionic strengthfreshwaters when no perturbation of the water isallowed, even the addition of a non complexingelectrolyte. In addition, their signal is not influ-enced significantly by colloidal complexes withsize larger than ∼4 nm, so that they provide a morewell-defined distinction between truly dissolvedand colloidal (size limit ≈1 nm) species [19]. Forexample, measurements by square wave modulatedstripping voltammetry with an agarose gel-coatedmercury-coated iridium microelectrode showedthat most of the Pb(II) and Cd(II) (80–90 %) inriver waters heavily loaded with suspended par-ticles was associated with colloidal material andenabled determination of binding site concentra-tions and corresponding stability constants for met-als and proton [41].

Separation/detection. Several approaches havebeen reported for the exploitation of the capabil-ity of electrochemical techniques to detect verylow analyte concentrations after species separa-tion [133]. An off-line procedure for the separationof dibutyltin and triphenyltin by ion-exchangechromatography followed by ASV detection ofthe tin species has been described and testedwith the extract of a sediment reference mate-rial [134]. Conductometric detection in capillaryelectrophoresis (CE) was used for the on-linedetermination of arsenic and selenium species(As(III), As(V), DMA, Se(IV), Se(VI)) withlimits of detection of about 50 µg dm−3 [135].

The method has been applied to water sam-ples from a tailing of tin ore processing. Theadvantages of microelectrodes for amperometricCE detection [136] have been used for the spe-ciation of mercury (Hg2+, monomethylmercury,monoethylmercury) [137]. After electrophoreticseparation the Hg species are reduced at a goldmicroelectrode at −0.2 V. The procedure pro-vides detection limits of 0.2 µg dm−3 (Hg2+) and5 µg dm−3 (Me–Hg) with a dynamic range ofthree orders of magnitude. Also extracts of sedi-ments have been analyzed by this approach. Field-portable CE instrumentation with electrochemicaldetection has been reported [138].

In situ. A more realistic understanding of ele-ment behavior in aquatic systems requires contin-uous real-time monitoring. Various voltammetricprobes have been reported for in situ deploymentin waters [19, 139]. It has been shown that micro-electrodes are very important for such applications,for various reasons. Most of them are prototypeswith a limit of detection of about 10−8 mol dm−3,which is useful mostly for polluted waters. Acommercial one, however, is available [19] with adetection limit down to 10−10 – 10−11 mol dm−3.Although use of SW-ASV can minimize oxygeninterference by kinetic discrimination of metal andoxygen currents (the direct SWV mode does noteliminate the O2 component efficiently), in lowpH buffer capacity freshwaters oxygen removal isanyway necessary to prevent pH changes inducedby reduction of O2 during the deposition stepin ASV. An on-line system based on permeationof oxygen through silicone tubing surrounded byan enzymatic cross-linked O2-scavenging gel hasbeen recently developed and successfully deployedin situ coupled to a submersible voltammetricprobe for determination of trace metals in oxygen-saturated freshwater [140].

Recently, a field-deployable instrument forthe speciation analysis of As(III) and As(V) inpotable water has been described [141]. It isbased on ASV of As(III) on a gold electrodein 4.5 mol dm−3 HCl. Unfortunately, the As(V)content can only be calculated by differencefrom the total As concentration determined afteroxidation of the sample.

454 DETECTION

4.2 Water/sediment systems

Processes occurring at the sediment–water inter-face involve fluxes of material over very smalldimensions (10–100 µm) which are often redoxsensitive [142]. Nonperturbing in situ techniquesare thus the method of choice and miniatur-ized potentiometric and amperometric electrodeswith tip sizes down to 10 µm have been devel-oped. Detailed descriptions of such electrodes isavailable [143]. Potentiometric glass and liquidmembrane ISEs have been developed in par-ticular for recording pH and pCO2 concentra-tion gradients [144, 145], Ca2+ and NH4

+ [146],while amperometric chemical microsensors havebeen developed for O2 [147] and H2S [148], andamperometric biosensors have been developed forBOD (biological oxygen demand), NO3

− andCH4. Luther and coworkers have used a mercury-coated gold microelectrode to measure submil-limeter scale profiles of O2, Fe(II) and S(−II)at the sediment–water interface [149] by usingthe same polarographic waves as those describedabove for water column analysis. In this envi-ronment, however, fouling, especially by sulfides,can be problematic for unprotected electrodes andin particular calibration in synthetic solutions cangive rise to large errors when applied to mea-surements inside the sediment. This has beenhighlighted by means of a voltammetric systememploying an individually addressable microsen-sor array (150 µm spacing between each microdiskelectrode, each with a radius of 5 µm) [41]. Thisdevice was used to measure in one run Pb andCd (both 5 µmol dm−3) concentration gradientsacross an artificial sediment (silica beads)/waterinterface, with 200 µm resolution, and to fol-low their evolution with time during the diffu-sion of the metals from the water to the solidphase. Results could be interpreted in terms ofmolecular diffusion, complexation of the metalsby the silica particles, and porosity and tortu-osity of the ‘sediment’ phase [150]. The resultsobtained with this ‘synthetic’ well-controlled sed-iment strongly suggest that interpretation of directmeasurements in real sediments must be madewith caution.

4.3 Biological matrices

The effects and toxicity of an element and itsmetabolic behavior depend on its physicochemicalform within an organism. But speciation analysisin the complex organic matrix of biologicalsamples, even in liquids, is a very challengingtask, and to date the vast majority of publicationsin this area have used chromatographic methods,usually coupled with atomic or mass spectrometricdetection [151–153]. For example, a review onmetal speciation in biological fluids cited 151references, only one of which referred to use ofvoltammetry [154].

In addition to the ex situ analysis of biolog-ical body fluids direct in vivo measurements ofchemical species are becoming more feasible dueto the development of microanalytical tools andmethods. Electroanalytical methods are in princi-ple well suited for miniaturization [155] and havebeen among the first analytical in vivo appli-cations. But one has to consider that a devicedeveloped for in vivo use must be biocompati-ble (nonperturbing of the local environment) andresistant to interference from biofouling such asprotein adsorption.

ISE. Miniaturized ISEs have been developedfor in vivo use and applied to e.g. monitoring ofK+ levels during cardiac events [156]. To dateapplications have been limited to major cations(e.g. K+, Na+, Ca2+) with the intention to measuredirectly the ‘free’ ions [157].

ASV. There are some empirical reports of usingASV (typically at polymer-coated electrodes) fortrace metal determination in body fluids suchas urine, blood, and sweat [158], and in foodssuch as wine [159] and milk [160]. However,rigorous interpretation in terms of speciation hasnot been attempted.

SCP. As for ASV, application of SCP hasbeen largely empirical with no real attempts todetermine speciation parameters. The purportedadvantage of SCP over SV methods is that lesssample pretreatment is required. For instance, thedetermination of labile and total copper and leadconcentrations in wines has been reported [161].

NEW CONCEPTS AND PROSPECTS 455

5 NEW CONCEPTS AND PROSPECTS

5.1 Selectivity

One of the key requirements of speciation anal-ysis consists in the selective detection of thevarious analytes in complex matrices. Electro-analytical methods are often very powerful forsingle-species analysis, but it is difficult to deter-mine a range of species of the same elementsimultaneously. Moreover, many of the real-worldsamples of environmental, food and health con-trol contain surface-active and/or electroactivematrix interferences. Therefore, two strategies forimproving the selectivity in electroanalysis are fol-lowed: development of electrochemical sensors forselective single-species analysis, and the on-linecoupling of chromatographic/electrophoretic sep-aration with electrochemical detection. The firstroute is described in detail in Chapter 5.11 ofthis Handbook. The developments of LCEC andCE-EC are currently undergoing rapid changesbecause of the progress in micro- and nanotech-nologies. It allows not only the miniaturizationof the working electrode or the complete detec-tion unit (microcells etc.), but even that of thewhole analytical system after sampling (µTAS,micro total analysis system) [162]. The combina-tion of microstructuring and microfluidics opensa new horizon also for the application of interfa-cial detection techniques based on electrochemicalprinciples. For instance, capillary electrophoresisis well suited for the efficient separation of dif-ferently charged redox species in liquid samplesand the analytes can subsequently be quantified byamperometry/voltammetry at a microelectrode oran array if they are electroactive. It is envisagedthat new lab-on-a-chip developments will not onlyrevolutionize the analysis of biomolecules such asDNA, but also the speciation analysis of dissolvedmetal(loid) species in the future by very efficientcombinations of separation and detection methods.

5.2 Metal speciation dynamicsand bioavailability

Much of the work on the relationship betweenmetal speciation and bioavailability persistently

adopts a thermodynamic approach, notably thewidely used free ion activity model (FIAM) [163].There is a plethora of disparate empirical reportson the relationship between electrochemicallydetermined metal and that which is bioavailable.This situation has arisen from the lack of an appro-priate theoretical framework for rational data inter-pretation. A few publications [164–166], however,have shown that the FIAM is limited to cases inwhich mass transfer is not flux determining. In gen-eral, dynamic aspects must be taken into accountby quantifying the role of association/dissociationrate parameters for complexes in the medium inthe supply of free metal towards the consumingbiological interface [167]. This theoretical analy-sis has identified the conditions under which themetal species detected by voltammetric methodscan be appropriately compared with that whichmay be available for biouptake; in particular thesignificance of measurements performed at micro-electrodes vis-a-vis biouptake by microorganismswith sizes of the order of the operational diffusionlayer thickness (see Section 3) [79, 89].

5.3 Spatial resolution

Most real-world samples are spatially heteroge-neous, and indeed knowledge of this aspect may bekey to understanding their behavior/functionality.Realistic stratification or heterogeneity data canonly be obtained when the dimensions of the sens-ing element are much smaller than the heterogene-ity dimensions of the medium. In addition, the totalphysical dimension of the sensing system shouldbe such that it is nonperturbing of the medium;this is of particular concern for solid samples. Forvalid data interpretation, the relationship betweenthe dimensions related to the system being probedand those of the sensor-related processes must beestablished.

Electrochemical measurements at microelec-trodes have pushed forward into spatial reso-lution at the micrometer and even nanometerscale during recent years. Two directions ofdevelopment are being followed: the placement(and often also movement) of single micro-electrodes with micropositioning devices (leading

456 DETECTION

to techniques such as scanning electrochemi-cal microscopy [168, 169]), and measurements atmicroelectrode arrays with individually address-able electrodes. An example for the latter approachhas been reported with the simultaneous record-ing of 64 complete voltammograms at an iridium-based microelectrode array [170]. In addition, thecapability of electrochemical detection to be appli-cable in ultrasmall sample volumes (down to1 pL) [171] offers the opportunity to obtain dataabout the spatial distribution of chemical speciesin a matrix after locally resolved microsampling.

5.4 Linking theoretical and experimentaldevelopments

Progress in understanding of element speciationrequires concomitant theoretical and experimentaldevelopments. A sound theoretical basis can pointthe way for rational design of improved analyticalsystems. It is of fundamental importance that theanalytical signals should be directly linked to theunderlying properties of the system under studyin a rigorous scientifically sound manner. A majordisadvantage of many of the sample pretreatmentand electrode modification protocols which areclaimed to give an ‘improved electrochemicalresponse’ is that the possibility of any suchrigorous link is lost, and results are thus renderedmeaningless or at best only empirical.

Therefore, the present opportunities to studyinterfaces, in particular solid–liquid interfaces,at a molecular/atomic level should be used fora better understanding of the structure–propertyrelations of electrodes as the key component ofelectrochemical detection systems. On this basistailor-made sensing surfaces can be designed forelectrochemical speciation analysis. In addition, apoint of view that is more oriented towards specia-tion dynamics will be necessary and measurementsat very different time scales can conveniently beperformed using electrochemical techniques (seeSection 2.3).

5.5 Instrumentation

There is a need for further development of insitu sensors to allow measurements to be made

under the most relevant conditions and to obvi-ate the need for tedious, and probably perturb-ing, sampling procedures [172]. Electrochemicalmicrotechnologies can contribute to the creation ofsuch in-field monitors for speciation analysis in theenvironment, in particular for water analysis. Butthe development of bedside monitors for speciesdeterminations in health care units can also beenvisaged. Another opportunity will be the integra-tion of amperometric/potentiometric flow-streamcells into detection units, consisting also of opti-cal measuring devices, for modular multidetectorspeciation analyzers.

Overall it has to be realized that the experimen-tal and theoretical potential of electrochemistryfor speciation analysis has not been extensivelyexploited in many respects. In principle, electro-chemical detection could contribute more to theanalysis of redox states and their dynamics, as wellas to probing electrochemical potentials at inter-faces relevant for species immobilization or trans-formation. These approaches, however, require thecombination of different scientific and technicaldisciplines, including the corresponding multidisci-plinary teaching and further developments of reli-able electrodes and measuring devices.

6 REFERENCES

1. Templeton, D. M., Ariese, F., Cornelis, R., Daniels-son, L. G., Muntau, H., van Leeuwen, H. P. and Lobin-ski, R., Pure Appl. Chem., 72, 1453 (2000).

2. Bond, A. M., Modern Polarographic Methods in Ana-lytical Chemistry , Marcel Dekker, New York, 1980.

3. Bard, A. J. and Faulkner, L. R., Electrochemical Meth-ods. Fundamentals and Applications , 2nd edn, JohnWiley & Sons, Inc., New York, 2000.

4. Southampton Electrochemistry Group, InstrumentalMethods in Electrochemistry , Ellis Horwood, Chich-ester, 1985.

5. Kissinger, P. T. and Heineman, W. R. (Eds), Labora-tory Techniques in Electroanalytical Chemistry , 2nd edn,Marcel Dekker, New York, 1996.

6. Heyrovsky, J. and Kuta, J., Principles of Polarography ,Academic Press, New York, 1966.

7. Galus, Z., Fundamentals of Electrochemical Analysis ,Ellis Horwood, Chichester, 1976.

8. IUPAC Commission on Electroanalytical Chemistry,Pure Appl. Chem., 45, 81 (1976).

REFERENCES 457

9. Fogg, A. G. and Wang, J., Pure Appl. Chem., 71, 891(1999).

10. van Leeuwen, H. P., Cleven, R. F. M. J. and Valenta, P.,Pure Appl. Chem., 63, 1251 (1991).

11. van Leeuwen, H. P., Colloids Surf., 51, 359 (1990).12. Town, R. M. and van Leeuwen, H. P., J. Electroanal.

Chem., 509, 58 (2001).13. Bond, A. M. and Svestka, M., Collect. Czech. Chem.

Comm., 58, 2769 (1993).14. Williams, G. and D’Silva, C., Analyst , 119, 2337

(1994).15. Bond, A. M., Luscombe, D. L., Tan, S. N. and Wal-

ter, F. L., Electroanalysis , 2, 195 (1990).16. Hintsche, R., Albers, J., Berndt, H. and Eder, A., Elec-

troanalysis , 12, 660 (2000).17. Tercier-Waeber, M.-L., Pei, J., Buffle, J., Fiaccabrino,

G. C., Koudelka-Hep, M., Riccardi, G., Confalonieri, F.,Sina, A. and Graziottin, F., Electroanalysis , 12, 27(2000).

18. Buffle, J., Complexation Reactions in Aquatic Systems.An Analytical Approach , Ellis Horwood, Chichester,1988.

19. Buffle, J. and Tercier-Waeber, M.-L., In situ voltamme-try: concepts and practice for trace analysis and specia-tion, in In Situ Monitoring of Aquatic Systems. ChemicalAnalysis and Speciation , Buffle, J. and Horvai, G. (Eds),John Wiley & Sons, Ltd, Chichester, 2000, Chapter 9,pp. 279–405.

20. Achterberg, E. P. and Braungardt, C., Anal. Chim. Acta,400, 381 (1999).

21. Tercier-Waeber, M.-L., Buffle, J., Confalonieri, F., Ric-cardi, G., Sina, A., Graziottin, F., Fiaccabrino, G. C.and Koudelka-Hep, M., Meas. Sci. Technol., 10, 1202(1999).

22. Fiaccabrino, G. C., de Rooij, N. F., Koudelka-Hep, M.,Hendrikse, J. and van den Berg, A., Microtechnologyfor the development of in situ microanalytical systems,in In Situ Monitoring of Aquatic Systems. ChemicalAnalysis and Speciation , Buffle, J. and Horvai, G. (Eds),John Wiley & Sons, Ltd, Chichester, 2000, Chapter 12,pp. 571–610.

23. Dıaz-Cruz, H. M., Esteban, M., van den Hoop, M. A.G. T. and van Leeuwen, H. P., Anal. Chem., 64, 1769(1992).

24. Pranitis, D. M., Teltingdiaz, M. and Meyerhoff, M. E.,Crit. Rev. Anal. Chem., 23, 163 (1992).

25. Thomas, J. D. R., Pure Appl. Chem., 73, 31 (2001).26. Kalcher, K., Kauffmann, J.-M., Wang, J., Svancara, I.,

Vytøas, K., Neuhold, C. and Yang, Z., Electroanalysis ,7, 5 (1995).

27. Bond, A. M. and Scholz, F., Z. Chem., 30, 117 (1990).28. Town, R. M., Tercier, M.-L., Parthasarathy, N., Bujard,

F., Rodak, S., Bernard, C. and Buffle, J., Anal. Chim.Acta , 302, 1 (1995).

29. Novotny, L., Fresenius’ J. Anal. Chem., 362, 184(1998).

30. Jayaratna, H. G., Bruntlett, C. S. and Kissinger, P. T.,Anal. Chim. Acta, 332, 165 (1996).

31. Wu, H. P., Anal. Chem., 66, 3151 (1994).32. Gross, M. and Jordan, J., Pure Appl. Chem., 56, 1096

(1984).33. Cara, R. G.-M., Sanchez-Misiego, A. and Zirino, A.,

Anal. Chem., 67, 4484 (1995).34. Kounaves, S. P. and Buffle, J., J. Electroanal. Chem.,

216, 53 (1987).35. Kounaves, S. P. and Buffle, J., J. Electroanal. Chem.,

239, 113 (1988).36. Stulık, K., Amatore, C., Holub, K., Marecek, V. and

Kutner, W., Pure Appl. Chem., 72, 1483 (2000).37. Ciszkowska, M. and Stojek, Z., J. Electroanal. Chem.,

466, 129 (1999).38. Matysik, F.-M. and Emons, H., Electroanalysis , 4, 501

(1992).39. Daniele, S., Baldo, M.-A., Ugo, P. and Mazzocchin,

G.-A., Anal. Chim. Acta, 219, 19 (1989).40. Feinberg, J. S. and Bowyer, W. J., Microchem. J., 47,

72 (1993).41. Pei, J., Tercier-Waeber, M.-L. and Buffle, J., Anal.

Chem., 72, 161 (2000).42. Tercier, M.-L., Parthasarathy, N. and Buffle, J., Electro-

analysis , 7, 55 (1995).43. Feeney, R. and Kounaves, S. P., Electroanalysis , 12,

677 (2000).44. Le Drogoff, B., El Khakani, M. A., Silva, P. R. M.,

Chaker, M. and Ross, G. G., Appl. Surf. Sci., 152, 77(1999).

45. Kutner, W., Wang, J., L’Her, M. and Buck, R. P., PureAppl. Chem., 70, 1301 (1998).

46. Kalcher, K., Electroanalysis , 2, 419 (1990).47. Arrigan, D. W. M., Analyst , 119, 1953 (1994).48. Dam, M. E. R., Thomsen, K. N., Pickup, P. G. and

Schrøder, K. H., Electroanalysis , 7, 70 (1995).49. Christensen, M. K. and Hoyer, B., Electroanalysis , 12,

35 (2000).50. Capelo, S., Mota, A. M. and Goncalves, M. L. S., Elec-

troanalysis , 7, 563 (1995).51. Bagel, O., Degrand, C. and Limoges, B., Anal. Chem.,

71, 3192 (1999).52. Tercier, M.-L. and Buffle, J., Anal. Chem., 68, 3670

(1996).53. Sokalski, T., Ceresa, A., Zwickl, T. and Pretsch, E., J.

Am. Chem. Soc., 119, 11347 (1997).54. Buck, R. P. and Cosofret, V. V., Pure Appl. Chem., 65,

1849 (1993).55. Umezawa, Y., Umezawa, K. and Sato, H., Pure Appl.

Chem., 67, 507 (1995).56. Ren, K., Fresenius’ J. Anal. Chem., 365, 389 (1999).57. Whitfield, M., Ion Selective Electrodes for the Analysis

of Natural Waters , AMSA Handbook No. 2, AustralianMarine Association, Sydney, 1971.

58. Buck, R. P. and Lindner, E., Pure Appl. Chem., 66,2527 (1994).

458 DETECTION

59. van Leeuwen, H. P., Buffle, J. and Lovriæ, M., PureAppl. Chem., 64, 1015 (1992).

60. Galceran, J., Salvador, J., Puy, J., Mas, F., Gimenez, D.,Esteban, M., J. Electroanal. Chem., 442, 151 (1998).

61. Galvez, J. and Park, S.-M., J. Electroanal. Chem., 263,269 (1989).

62. Osteryoung, J. and O’Dea, J. J., Square-wave voltam-metry, in Electroanalytical Chemistry , Bard, A. J. (Ed.),Vol. 14, Marcel Dekker, New York, 1986, pp. 209–308.

63. Jehring, H., Elektrosorptionsanalyse mit der Wechsel-strompolarographie, Akademie-Verlag, Berlin, 1974.

64. Emons, H., Werner, G., Haferburg, D. and Kleber, H.-P., Electroanalysis , 1, 555 (1989).

65. Emons, H., Schmidt, Th. and Stulik, K., Analyst , 114,1593 (1989).

66. Ostapczuk, P., Stoeppler, M. and Durbeck, H. W., Fre-senius’ Z. Anal. Chem., 332, 662 (1988).

67. Jagner, D., Sahlin, E. and Renman, L., Talanta , 41, 515(1994).

68. Kalvoda, R., Fresenius’ J. Anal. Chem., 349, 565(1994).

69. Paneli, M. G. and Voulgaropoulos, A., Electroanalysis ,5, 355 (1993).

70. van den Berg, C. M. G., Anal. Chim. Acta, 250, 265(1991).

71. Emons, H., Schmidt, Th. and Werner, G., Anal. Chim.Acta , 228, 55 (1990).

72. Leon, C., Emons, H., Ostapczuk, P. and Hoppstock, K.,Anal. Chim. Acta, 356, 99 (1997).

73. Wang, J., Stripping Analysis , VCH, Deerfield Beach,1985.

74. Brainina, Kh. and Neyman, E., Electroanalytical Strip-ping Methods , John Wiley & Sons, Inc., New York,1993.

75. You, T., Yang, X. and Wang, E., Electroanalysis , 11,459 (1999).

76. Acworth, I. N., Naoi, M., Parvez, H. and Parvez, S.(Eds), Coulometric Electrode Array Detectors forHPLC , VSP, Utrecht, 1997.

77. Buffle, J., Tessier, A. and Haerdi, W., Interpretation oftrace metal complexation by aquatic organic matter,in Complexation of Trace Metals in Natural Waters ,Kramer, C. J. M. and Duinker, J. C. (Eds), MartinusNijhoff/Dr W. Junk, The Hague, 1984, pp. 301–316.

78. Filella, M., Buffle, J. and van Leeuwen, H. P., Anal.Chim. Acta , 232, 209 (1990).

79. van Leeuwen, H. P., J. Electroanal. Chem ., 99, 93(1979).

80. van Leeuwen, H. P., Cleven, R. and Buffle, J., PureAppl. Chem., 61, 255 (1989).

81. van Leeuwen, H. P. and Buffle, J., J. Electroanal.Chem., 296, 359 (1990).

82. de Jong, H. G., van Leeuwen, H. P. and Holub, K., J.Electroanal. Chem., 234, 1 (1987).

83. de Jong, H. G. and van Leeuwen, H. P., J. Electroanal.Chem., 234, 17 (1987).

84. de Jong, H. G. and van Leeuwen, H. P., J. Electroanal.Chem., 235, 1 (1987).

85. van Leeuwen, H. P., Anal. Proc., 28, 66 (1991).86. Pinheiro, J. P., Mota, A. M. and van Leeuwen, H. P.,

Colloids Surf. A, 151, 181 (1999).87. Galceran, J., Puy, J., Salvador, J., Cecilia, J. and van

Leeuwen, H. P., J. Electroanal. Chem., 505, 85 (2001).88. Pinheiro, J. P. and van Leeuwen, H. P., Environ. Sci.

Technol., 35, 894 (2001).89. Soares, H. M. V. M., Conde, P. C. F. L., Almeida,

A. A. N. and Vasconcelos, M. T. S. D., Anal. Chim.Acta , 394, 325 (1999).

90. Soares, H. M. V. M. and Conde, P. C. F. L., Anal.Chim. Acta , 421, 103 (2000).

91. Damaskin, B. B., Petrii, O. A. and Batrakov, V. V.,Adsorption of Organic Compounds on Electrodes ,Plenum, New York, 1971.

92. Pinheiro, J. P., Mota, A. M., Goncalves, M. S. and vanLeeuwen, H. P., Environ. Sci. Technol., 28, 2112 (1994).

93. Kubiak, W. and Wang, J., J. Electroanal. Chem., 258,41 (1989).

94. Buffle, J., J. Electroanal. Chem., 125, 273 (1981).95. Buffle, J., Mota, A. M. and Goncalves, M. L. S., Port.

Electrochim. Acta , 3, 293 (1985).96. Florence, T. M. and Mann, K. J., Anal. Chim. Acta,

200, 305 (1987).97. Town, R. M. and Filella, M., J. Electroanal. Chem.,

488, 1 (2000).98. Florence, T. M., Electrochemical techniques for trace

element speciation in waters, in Trace Element Speci-ation: Analytical Methods and Problems , Batley, G. E.(Ed.), CRC Press, Boca Raton, FL, 1989, Chapter 4,pp. 77–116.

99. Taillefert, M., Luther, G. W. and Nuzzio, D. B., Elec-troanalysis , 12, 401 (2000).

100. Schwarz, J., Henze, G. and Thomas, F. G., Fresenius’J. Anal. Chem., 352, 479 (1995).

101. Bond, A. M., Turoczy, N. J. and Carter, R. J., Anal.Chim. Acta , 310, 109 (1995).

102. Zirino, A., van der Weele, D. A., Belli, S. L., DeMarco, R. and Mackey, D. J., Mar. Chem., 61, 173(1998).

103. Verweij, W. and Ruzic, I., Croat. Chem. Acta, 70, 419(1997).

104. Davison, W., Buffle, J. and De Vitre, R., Pure Appl.Chem., 60, 1535 (1988).

105. Davison, W., Anal. Proc., 28, 59 (1991).106. DeVitre, R. R., Buffle, J., Perret, D. and Baudat, R.,

Geochim. Cosmochim. Acta, 52, 1601 (1988).107. Davison, W., Buffle, J. and DeVitre, R. R., Anal. Chim.

Acta , 377, 193 (1988).108. Town, R. M. and Filella, M., Aquat. Sci., 62, 252

(2000).109. Achterberg, E. P., Colombo, C. and van den Berg,

C. M. G., Cont. Shelf Res., 19, 537 (1999).110. Achterberg, E. P. and van den Berg, C. M. G., Mar.

Pollut. Bull., 32, 471 (1996).

REFERENCES 459

111. Apte, S. C., Gardner, M. J. and Ravenscroft, J. E., Mar.Chem., 29, 63 (1990).

112. Gardner, M. J. and Ravenscroft, J. E., Chem. Spec.Bioavail., 3, 22 (1991).

113. van den Berg, C. M. G., Anal. Chim. Acta, 164, 195(1984).

114. Braungardt, C., Achterberg, E. P. and Nimmo, M., Anal.Chim. Acta , 377, 205 (1998).

115. van den Berg, C. M. G. and De Luca Rebello, A., Sci.Tot. Environ., 58, 37 (1986).

116. Bruland, K. W., Rue, E. L., Donat, J. R., Skrabal, S. A.and Moffett, J. W., Anal. Chim. Acta, 405, 99 (2000).

117. Donat, J. R. and Bruland, K. W., Mar. Chem., 28, 301(1990).

118. van den Berg, C. M. G. and Donat, J. R., Anal. Chim.Acta , 257, 281 (1992).

119. Campos and C. M. G. van den Berg, M. L. A. M.,Anal. Chim. Acta, 284, 481 (1994).

120. Donat, J. R. and van den Berg, C. M. G., Mar. Chem.,38, 69 (1992).

121. Gledhill, M., van den Berg, C. M. G., Nolting, R. F.and Timmermans, K. R., Mar. Chem., 59, 283 (1998).

122. Fischer, E. and van den Berg, C. M. G., Anal. Chim.Acta , 432, 11 (2001).

123. Gardner, M. and Ravenscroft, J., Chemosphere, 23, 695(1991).

124. Gardner, M., Dixon, E. and Comber, S., Chem. Spec.Bioavail., 12, 1 (2000).

125. Witter, A. E., Mabury, S. A. and Jones, A. D., Sci. Tot.Environ., 212, 21 (1998).

126. Jin, L. and Gogan, N. J., Anal. Chim. Acta, 412, 77(2000).

127. Xue, H., Oestreich, A., Kistler, D. and Sigg, L., Aquat.Sci., 58, 69 (1996).

128. Xue, H. and Sunda, W. G., Environ. Sci. Technol., 31,1902 (1997).

129. Jones, M. J. and Hart, B. T., Chem. Spec. Bioavail., 1,59 (1989).

130. Zuhri, A. Z. A. and Voelter, W., Fresenius’ J. Anal.Chem., 360, 1 (1998).

131. van den Berg, C. M. G., Anal. Proc., 28, 58 (1991).132. Town, R. M. and Filella, M., Limnol. Oceanogr., 45,

1341 (2000).133. Buchberger, W., Fresenius’ J. Anal. Chem., 354, 797

(1996).134. Ochsenkuhn, K. M., Ochsenkuhn-Petropoulou, M.,

Tsopelas, F. and Mendrinos, L., Fresenius J. Anal.Chem., 369, 633 (2001).

135. Schlegel, D., Mattusch, J. and Wennrich, R., FreseniusJ. Anal. Chem., 354, 535 (1996).

136. Matysik, F.-M., Electroanalysis , 12, 1349 (2000).137. Lai, E. P. C., Zhang, W., Trier, X., Georgi, A., Kowal-

ski, S., Kennedy, S., MdMuslim, T. and Dabek-Zlotorzynska, E., Anal. Chim. Acta, 364, 63 (1998).

138. Kappes, T., Schnierle, P. and Hauser, P. C., Anal. Chim.Acta , 393, 77 (1999).

139. Wang, J., Larson, D., Foster, N., Armalis, S., Lu, J.,Rongrong, X., Olsen, K. and Zirino, A., Anal. Chem.,67, 1481 (1995).

140. Tercier-Waeber, M.-L. and Buffle, J., Environ. Sci. Tech-nol., 34, 4018 (2000).

141. Huang, H. and Dasgupta, P. K., Anal. Chim. Acta, 380,27 (1999).

142. Tessier, A., Carignan, R. and Belzile, N., Processesoccurring at the sediment–water interface: emphasis ontrace elements, in Chemical and Biological Regulationof Aquatic Systems , Buffle, J. and De Vitre, R. R. (Eds),Lewis, Boca Raton, FL, 1994, Chapter 4, pp. 139–197.

143. Buffle, J. and Horvai, G. (Eds), In Situ Monitoringof Aquatic Systems. Chemical Analysis and Speciation ,IUPAC Series on Analytical and Physical Chemistryof Environmental systems , John Wiley & Sons, Ltd,Chichester, 2000.

144. Cai, W.-J. and Reimers, C. E., Sensors for in situpH and pCO2 measurements in seawater and at thesediment–water interface, in In Situ Monitoring ofAquatic Systems. Chemical Analysis and Speciation ,Buffle, J. and Horvai, G. (Eds), John Wiley & Sons,Ltd, Chichester, 2000, Chapter 3, pp. 75–119.

145. Muller, B., Buis, K., Stierli, R. and Wehrli, B., Limnol.Oceanogr., 43, 1728 (1998).

146. De Beer, D., Potentiometric microsensors for in situmeasurements in aquatic environments, in In SituMonitoring of Aquatic Systems. Chemical Analysis andSpeciation , Buffle, J. and Horvai, G. (Eds), John Wiley& Sons, Ltd, Chichester, 2000, Chapter 5, pp. 161–194.

147. Glud, R. N., Gundersen, J. K. and Ramsing, N. B., Elec-trochemical and optical oxygen microsensors for in situmeasurements, In Situ Monitoring of Aquatic Systems.Chemical Analysis and Speciation , Buffle, J. and Hor-vai, G. (Eds), John Wiley & Sons, Ltd, Chichester, 2000,Chapter 5, pp. 19–73.

148. Kuhl, M. and Steuckart, C., Sensors for in situ analysisof sulfide in aquatic systems, In Situ Monitoring ofAquatic Systems. Chemical Analysis and Speciation ,Buffle, J. and Horvai, G. (Eds), John Wiley & Sons,Ltd, Chichester, 2000, Chapter 5, pp. 121–159.

149. Luther, G. W., Reimers, C. E., Nuzzio, D. B. andLovalvo, D., Environ. Sci. Technol., 33, 4352 (1999).

150. Pei, J., Tercier-Waeber, M.-L., Buffle, J., Fiaccabrino,G. C. and Koudelka-Hep, M., Anal. Chem., 73, 2273(2001).

151. Florence, T. M., Trace element speciation in biologicalsystems, in Trace Element Speciation: Analytical Meth-ods and Problems , Batley, G. E. (Ed.), CRC Press, BocaRaton, FL, 1989, Chapter 9, pp. 319–341.

152. Caroli, S. (Ed.), Element Speciation in BioinorganicChemistry , John Wiley & Sons, Inc., New York, 1996.

153. Caruso, J. A., Sutton, K. L. and Ackley, K. L. (Eds),Elemental Speciation , Elsevier, Amsterdam, 2000.

154. Das, A. K., Chakraborty, R., Cervera, M. L. and de laGuardia, M., Mikrochim. Acta, 122, 209 (1996).

460 DETECTION

155. Emons, H., Electrochemical microanalysis, in Electro-chemical Microsystems Technologies , Schultze, J. W.,Osaka, T., Datta, M. (Eds), Taylor and Francis, London,2002, pp. 371–383.

156. Buck, R. P., Cosofret, V. V., Lindner, E., Ufer, S.,Madaras, M. B., Johnson, T. A., Ash, R. B. and Neu-man, M. R., Electroanalysis , 7, 846 (1995).

157. Lindner, E. and Buck, R. P., Anal. Chem., 72, 336A(2000).

158. Hoyer, B. and Florence, T. M., Anal. Chem., 59, 2839(1987).

159. Azenha, M. A. G. O. and Vasconcelos, M. T. S. D., J.Agr. Food Chem., 48, 5740 (2000).

160. Vidal, J. C., Vinao, R. B. and Castillo, J. R., Electro-analysis , 4, 653 (1992).

161. Green, A. M., Clark, A. C. and Scollary, G. R., Frese-nius’ J. Anal. Chem., 358, 711 (1997).

162. Manz, A., Graber, N. and Widmer, H. M., Sens. Actua-tors B , 1, 244 (1990).

163. Morel, F. M. M. and Hering, J. G., Principles andApplications of Aquatic Chemistry , John Wiley & Sons,Inc., New York, 1993.

164. Whitfield, M. and Turner, D., Critical assessment ofthe relationship between biological thermodynamic andelectrochemical availability, in Chemical Modeling inAqueous Systems , Jenne, E. A (Ed.), ACS Symp. Ser.93, American Chemical Society, Washington, DC, 1979,Chapter 29, pp. 657–680.

165. Tessier, A., Buffle, J. and Campbell, P. G. C, Uptakeof trace metals by aquatic organisms, in Chemicaland Biological Regulation of Aquatic Systems , Buffle, J.and DeVitre, R. R (Eds), Lewis, Ann Arbor, MI, 1994,Chapter 6, pp. 199–232.

166. van Leeeuwen, H. P., Environ. Sci. Technol., 33, 3743(1999).

167. van Leeuwen, H. P., J. Radioanal. Nucl. Chem., 246,487 (2001).

168. Bard, A. J., Fan, F. R. F., Kwak, J. and Lev, O., Anal.Chem., 61, 132 (1989).

169. Wittstock, G., Emons, H., Ridgway, T. H., Blubaugh,E. O. and Heineman, W. R., Anal. Chim. Acta, 298, 285(1994).

170. Pei, J., Tercier-Waeber, M.-L., Buffle, J., Fiaccabrino,G. C. and Koudelka-Hep, M., in Electroanalysis ,Emons, H. and Ostapczuk, P. (Eds), ForschungszentrumJulich GmbH, Julich, 2000, p. B 04.

171. Clark, R. A., Hietpas, P. B. and Ewing, A. G., Anal.Chem., 69, 259 (1997).

172. Batley, G. E., Collection, preparation, and storage ofsamples for speciation analysis, in Trace ElementSpeciation: Analytical Methods and Problems , Batley,G. E. (Ed.), CRC Press, Boca Raton, FL, 1989,Chapter 1, pp. 1–24.

5.10 Future Instrumental Developmentfor Speciation

Andrew N. Eaton and Fadi R. Abou-ShakraMicromass UK Ltd, Manchester, UK

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 4612 Requirements for Elemental Detection

Systems Coupled to ChromatographyTechniques . . . . . . . . . . . . . . . . . . . . . . . . 4622.1 Control of chromatograph . . . . . . . . . 4622.2 Data acquisition . . . . . . . . . . . . . . . . 4632.3 Data processing . . . . . . . . . . . . . . . . 463

3 Structure Elucidation . . . . . . . . . . . . . . . . 4654 Combinations . . . . . . . . . . . . . . . . . . . . . . 466

4.1 Solvent compatibility . . . . . . . . . . . . 4664.2 Sensitivity of the techniques . . . . . . . 4674.3 Instrument control . . . . . . . . . . . . . . 4674.4 Multiple combinations . . . . . . . . . . . 468

5 New Techniques . . . . . . . . . . . . . . . . . . . . 4685.1 Alternatives to the ICP . . . . . . . . . . . 4685.2 Glow discharge . . . . . . . . . . . . . . . . 469


1 INTRODUCTION

It is evident from the IUPAC definition given at thestart of this book that speciation is really a combi-nation of two distinct aims. Firstly to identify all ofthe species in a sample which contain a given ele-ment or elements, and secondly to quantify them.The plethora of papers published on the subject ofspeciation suggests two things to the authors: thatthere is a tremendous amount of interest in spe-ciation, and that none of the analytical techniquesthat are currently available provides a completesolution to the challenges involved.

Elemental analysis techniques such as atomicabsorption spectrophotometry (AAS), atomic fluo-rescence spectrometry (AFS), inductively coupledplasma atomic emission spectrometry (ICP-AES)and inductively coupled plasma mass spectrometry(ICP-MS), and others, provide excellent elementspecific information. A discussion of the relativemerits and demerits of each of these is outside the

scope of this chapter, and for the purposes of thiswork, such techniques will be collectively referredto as ‘elemental detection techniques.’

The ability of these techniques to provide ele-ment specific detection is their strength, and alsotheir major weakness for speciation work, since, bydefinition, they measure the amount of an elementpresent irrespective of the ligands to which it iscomplexed or the compound – the exact oppositeof the information which we are trying to glean.

The usual approach to circumvent this problemhas been to use a separation system, usually chro-matography, to separate the various compoundsprior to analysis. Until recently, elemental detec-tion techniques were only capable of acquiringsteady state signals. However, systems are nowbecoming commercially available with data acqui-sition systems which allow them to be coupledto the gamut of separation systems available.With speciation analysis being driven by formerelemental analysts who already have elemental

462 DETECTION

detection techniques, such instruments are likelyto feature heavily in the future of speciation, andthe requirements for integrating such systems withthe appropriate separation system will be discussedlater in the chapter.

However, coupling a separation system, such asa chromatograph, to an elemental detector givesus only half of the solution, as there remainsthe issue of identifying the species present in thechromatogram. Most workers involved in suchresearch take the approach of spiking the samplewith an authentic standard of one of the expectedelement-containing species, and by noting whichpeak in the chromatogram gets bigger, an identitycan be assigned. There are, however, a number ofdrawbacks to this approach.

There is the possibility of two or more speciesco-eluting at a given retention time. If bothspecies contain the element of interest, then theelemental detector cannot distinguish between thetwo species. The only safe options are to re-run thesample using a different separation, the chances ofthe same two peaks co-eluting under two differentsets of conditions being very small, or to usetechniques such as isotope dilution which involvesignificant manual sample preparation and repeatedanalyses. This increases the method developmenttime and, at least, doubles the analysis time. Inaddition, it is necessary to spike the sample witheach individual standard, or isotopic tag, and runeach spiked solution through the separation. Forsamples containing even a modest number ofspecies, this represents a huge amount of repetitionand a greatly increased method development time.While this may be acceptable for a researchproject, it would definitely not be practical ina high throughput, routine analysis laboratory,even if this spiking and re-running process weremade completely automatic, something which nocommercially available system can do at present.

This technique relies upon the availability ofauthentic standards for each species in the sam-ple. If the sample is relatively simple, this maynot be a problem, but as the technique is appliedto more and more complex samples, it becomesimpossible to predict all of the possible species,let alone obtain authentic standards for each.

A good example of this is in the work byMarchante-Gayon et al. [1] on the determinationof selenium species in nutritional commercial sup-plements. Although many of the peaks in theirchromatograms could be identified by spiking withauthentic standards, such as selenite, selenate,selenocystine, selenomethionine and selenoethio-nine, there remain a number of peaks in eachchromatogram which do not correspond to any ofthese compounds and therefore cannot be identi-fied. Worse still, the technique offers no possibleclues whatsoever as to how to proceed with theidentification of such unknowns.

Elemental detection therefore is not the wholesolution and it becomes essential to look to othertechniques or combinations of techniques to pro-vide the necessary structural elucidation capability.

2 REQUIREMENTS FORELEMENTAL DETECTION SYSTEMSCOUPLED TO CHROMATOGRAPHYTECHNIQUES

In general, and provided that the chromatographyused is adequate to the task, the detection andquantification of a specific metal ion as it elutesfrom the column should be a very simple task.The usual parameters, such as detection limit, andstability, apply, just as they do in all such analy-ses, but the task is not a particularly demandingone. Having identified those peaks in the chro-matogram which contain the metal, the process ofidentifying the species comes into play, and forthis the elemental detection techniques are uselessand alternative methodologies and techniques arerequired, especially as samples of increasing com-plexity are investigated.

2.1 Control of chromatograph

In the majority of work published on the couplingof chromatography systems to elemental detectors,there has been only minimal control of thechromatograph by the elemental detector datasystem, or vice versa, usually just the provisionof sending a trigger signal to either start the

REQUIREMENTS FOR ELEMENTAL DETECTION SYSTEMS 463

chromatographic run, or to start the acquisition.This means that both parts of the system haveto be set up independently of each other, withthe result that the entire set of experimentalconditions are not stored in any single location.In research projects, this is not a serious problem,but when speciation becomes routine and suchanalysis results are being used as evidence inlegal proceedings, it is essential to have a clearaudit trail for all of the method and results. Forthis reason alone, it will become necessary tocontrol the entire system from a single set ofsoftware.

Control will need to take the form of definingthe gradient to be used, if any, starting and stop-ping the pump, injecting samples and controllingthe autosampler, and recording all of these settingsin addition to those of the elemental analyser.

2.2 Data acquisition

The use of chromatography imposes on theelemental detection system a requirement to beable to monitor continuously variable signals,depending upon the nature of the technique used.

For essentially single element techniques suchas AAS or ‘reaction cell’ type ICP-MS instru-ments, there are no special requirements otherthan the ability to constantly monitor and storethe output from the detector, the time intervalbetween successive measurements being relativelyunimportant. Similarly, techniques which providesimultaneous detection of a number of elementssuch as simultaneous ICP-AES, or ICP-MS with atime-of-flight (TOF) analyser, also have no signif-icant restriction on the data acquisition rate, otherthan the ability to continuously monitor and storethe output from each detector.

The difficulty comes for multielement scanningtechniques, such as sequential ICP-AES and ICP-MS. Here the time taken to perform each mea-surement and to move the analyser to the nextmeasuring position, termed its ‘duty cycle’, meansthat there is often a trade-off between the num-ber of elements which can be determined, and theresulting number of data points per chromatogrampeak which can be acquired.

Table 5.10.1. Typical peak widths for various common sepa-ration techniques.

Separation technique Peak width Time per scanfor ten points/peak

Capillary electrophoresis 2–5 s 200 msHPLC 5–20 s 500 msGC 2–3 s 200 msFast GC <100 ms 10 ms

While a large number of points per peakincreases the quality of the peak shape, it israrely the case that this is the most importantconsideration. Often it is more desirable to havea number of elements (analytes plus internalstandard) and, if possible, multiple lines or isotopesof these elements, and this requirement usuallymeans restricting the number of points per peak.For quantitative purposes, a minimum of ten pointsper peak is required in order to define the peakadequately, because, unlike spectroscopic peaks,the chromatography peak width may vary quitesignificantly.

In the main, the time available to take measure-ments is governed by the choice of chromatogra-phy used, which dictates the width of the resultingpeaks. Table 5.10.1 shows a range of chromato-graphic techniques and typical peak widths, and themaximum time in which a scan can be performed,assuming ten points per peak.

In order to combine a chromatography tech-nique with a particular elemental detector, it isessential that the acquisition system of the detectorcan acquire and store a scan fast enough to main-tain the minimum number of points per peak forthe type of chromatography used.

2.3 Data processing

The processing of chromatographic data is verydifferent to processing spectra. Firstly, the datasystem needs to be able to display chromatograms,based on successive element scans, and to displaysuch chromatograms using the total signal obtainedfor all peaks in the spectra, using single-elementor line signals, and combinations of lines. Anexample of this, based on ICP-MS data formetabolites of a platinum-containing drug in urine,is shown in Figure 5.10.1.

464 DETECTION

0

1.65E5

0

4.32

4.34

4.34

4.34

4.34

4.00 6.00 8.00 10.00 12.00 14.00

Retention Time (Minutes)

16.00 18.00 20.00 22.00 24.00

6.23

(a) 198Pt

(b) 194Pt

(c) 192Pt

(d) 196Pt

(e) 195Pt

6.21

6.29

6.20

6.25

17.25

17.27

17.29

17.27

17.27

7.08E5

0

1.44E4

0

5.59E5

0

7.40E5

Figure 5.10.1. ICP-MS mass chromatograms showing the separation of three unknown metabolites present in urine after taking adrug which contained platinum: (a) mass chromatogram for m/z 198; (b) mass chromatogram for m/z 194; (c) mass chromatogramfor m/z 192; (d) mass chromatogram for m/z 196; (e) mass chromatogram for m/z 195.

36.0 36.5 37.0 37.5 38.0 38.5 39.0 39.5 40.0


40.5 41.0 41.5 42.0 42.5 43.0 43.5 44.0

0

1.5E5

Cou

nts

per

seco

nd

Figure 5.10.2. Integration of chromatogram peaks with sloping baseline.

STRUCTURE ELUCIDATION 465

Another difference is in the way peaks areintegrated. Unlike elemental spectra, the base-line in a chromatogram can vary significantlyduring a run, as can be seen in Figure 5.10.2.Therefore, for routine work, the data systemmust be capable of automatically handling slop-ing baselines if accurate quantitation is to be car-ried out.

There are many specialist chromatography dataprocessing packages available which carry outthis type of data processing, usually designedfor organic mass spectrometry applications, andthese could be used to carry out this task,provided that the elemental detector data filesare readable, or can be made readable, by thesoftware. However, this increases the time andeffort required to carry out the analysis, thusreducing throughput.

Most elemental data systems offer a qual-ity control option, whereby data are processedduring a run and checked against user definedquality control (QC) standards, and immediatelyre-run, or the instrument re-calibrated, if there isa problem. This facility is essential to maintain-ing throughput in routine laboratories. To carryout such QC checking, the data processing must,of necessity, be carried out on the instrumentdata system. Since, as far as the authors areaware, only one of the commercially availableelemental detector data systems is capable ofcarrying out both chromatography data process-ing and QC checking, this is an area whichwill have to develop in the future, if elementaldetectors are to be used for speciation in a rou-tine manner.

3 STRUCTURE ELUCIDATION

It is clear that very different analytical techniquesare required to identify the species, than areused only to measure the amount of the elementpresent: the two requirements being somewhatmutually exclusive.

Although not exclusively so, the majority ofthe elements of interest are contained withinorganic compounds or bound by organic lig-ands. It is logical therefore to look at organic

analysis techniques, such as nuclear magnetic res-onance spectrometry (NMR), infrared spectrom-etry (IR), ultraviolet spectrophotometry (UV) andmass spectrometry (MS), which have been devel-oped specifically to obtain structural informationor elemental composition for organic compounds.

With the exception of mass spectrometry, allof these techniques provide information on thefunctional groups present and the types of bondwhich a compound contains, but they cannotpositively identify the actual component underanalysis. Only mass spectrometry provides bothstructural information, through fragmentation stud-ies, and the possibility to identify the molecularweight of the compound, thereby allowing it tobe positively identified in a single run. Organicmass spectrometry has developed dramatically inthe last 10 years, primarily as a result of thedevelopment of the electrospray ionisation (ESI)technique, which is both a soft ionisation tech-nique, allowing molecular weight information to begleaned from even very labile compounds, and isapplicable across a wide range of polarities of com-pounds, thereby increasing the range of amenableanalyte compounds.

The technique of electrospray, as applied tospeciation, is discussed elsewhere in this volume,and so will not be covered here. However, anydiscussion about the future of instrumentation forspeciation needs to consider the implications ofdevelopments in organic mass spectrometry.

Organic mass spectrometry development hadbeen driven largely by the requirements of thepharmaceutical and biotechnology industries toidentify very quickly large numbers of compoundspresent in complex biological samples such asurine and plasma. In this respect, the complex-ity of the samples can be compared to thoseof environmental samples, where degradation ofcompounds by the biosystem leads to very com-plex mixtures.

The primary tool in such analysis is tan-dem mass spectrometry, or MS/MS which usestwo mass analysers, the first to isolate a par-ent ion, chosen on the basis of its molecularweight, which is then fragmented, by means ofa collision cell, and the fragments mass analysed

466 DETECTION

by the second mass analyser. The instrumenta-tion used, until recently, has been the ‘triplequad’, which uses two quadrupole analysers and anRF multipole (quadrupole, hexapole or octapole)encased in a gas tight canister to form the colli-sion cell.

Recent variants on this approach have beenthe use of a quadrupole ion trap. Here, a singletrap is used to retain all of the ions, to selec-tively eject all but the parent ion, fragment it,and eject the fragment ions to the detector accord-ing to mass, thereby producing a mass spectrum.Theoretically, this approach allows multiple col-lection, fragmentation and mass analysis, termed‘MSn’ (MS to the n), although the complexityinvolved in setting up such analyses has seen thistechnology confined to the research laboratories,rather than being used instead of the over 500 triplequad systems which are sold each year throughoutthe world.

Another more widely applicable variant onthe triple quad has been to replace the secondquadrupole with an orthogonal TOF analyser. Theadvantage of the TOF analyser is that it producesfull mass scans, thereby counting all of the ionswhich pass through it, unlike a quadrupole massfilter which allows ions of a single mass throughat any time. By admitting all of the ions to thedetector, rather than just one mass, for a massrange of 500 Da, the sensitivity could theoreticallybe as much as 500 times greater, although a factorof 200 × is commonly found in practice.

4 COMBINATIONS

An increasing number of workers have taken theapproach of combining more than one techniquein order to obtain both structural and quantita-tive information. Typically, elemental detectorsare used in parallel with organic mass spectrom-eter. The former offer in general the sensitiv-ity, selectivity and specificity required to identifyand possibly quantify the presence of a speciescontaining the element of interest. The latter onthe other hand, can be used to provide accuratemass information or structural information via par-ent–daughter fragmentation studies. An example

of such an attempt is the work of Nicholson et al.where ICP-MS is used in parallel to ESI-TOF-MS [2].

The considerations that must be taken intoaccount to ensure a successful use of a combinationof techniques are: solvent compatibility, sensitivityof the techniques, and instrument control.

4.1 Solvent compatibility

A quick review of the current literature on theseparation of metal/semimetal-containing speciesfor detection by element-sensitive detectors showsthat the great majority of the mobile phasesused contain high levels of salts or ion-pairingreagents. This will lead a substantial reduction inthe sensitivity for ESI based systems. However,advances in ICP generators and matching circuitryhave made it possible to use highly organic mobilephases without affecting the stability and theperformance of the ICP source. This in turn shouldopen the road for the use of more ESI friendlychromatographic conditions, which would enablesimultaneous ICP/ESI – MS detection on the sameeluents. Figure 5.10.3, shows the separation ofdiphenyltin, dibutyltin, triphenyltin, and tributyltinusing HPLC ICP-MS using a 65 % acetonitrilemobile phase.

This is paralleled with advances on the ESIfront with the use of nebuliser assisted electrosprayand ‘Z-type’ orthogonal interfaces which mean thatthose systems are currently more tolerant of saltsin the mobile phase.

Another exciting advance is in the use ofsuperheated water as a mobile phase [3]. Thismethod is based upon the observation that waterat temperatures in excess of 100 ◦C and which iskept in a liquid from as a result of a high backpressure can function as a suitable solvent forreverse-phase chromatography, without the needfor salts or organic modifiers. This technique iscurrently in its infancy. However, should it proveto be successful for elemental speciation it wouldoffer the ideal solution for the compatibility issuesbetween the hard and the soft ionisation sources.

COMBINATIONS 467

0

1.6E+4

Cou

nts

per

seco

nd

DPhT

DBT

TBT

TPhT

1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0


9.0 10.0 11.0 12.0 13.0 14.0 15.0

Figure 5.10.3. Separation of diphenyltin, dibutyltin, triphenyltin, and tributyltin using HPLC ICP-MS with a 65 % acetonitrilemobile phase.

4.2 Sensitivity of the techniques

Co-eluting material, which may not significantlyaffect the elemental detector, can be a major sourceof a high background signal intensity that limits thedetection capabilities of an ESI-MS system. How-ever, the use of MS-MS can provide a more selec-tive way for identifying the species of concern andhas been reported to offer a significant improve-ment in signal to background ratio. Advancedacquisition systems currently enable time sched-uled acquisitions whereby the mass spectrome-ter is set to carry out specific MS-MS (precur-sor–product) monitoring at times coinciding withthe retention time of the species of interest, theability to trigger these experiments upon the detec-tion of such peaks on the elemental system wouldbe a very powerful tool indeed.

In addition, it is well known that with ESI,molecular clusters are formed during the ionisationand extraction stage. This does in turn reducethe sensitivity of the instrument by reducing the

number of ions of a species that are available to themass spectrometer. Declustering techniques, viaexcitational or collisional dissociation are offeredas a standard in modern ESI-MS instrument.Therefore, method development work is requiredto ensure that a declustering mechanism is usedto eliminate adducts of the species of interest.This declustering mechanism, however, must notbe strong enough to cause any fragmentation ofthe species.

4.3 Instrument control

Parallel operation of elemental and structuraldetectors can be made more efficient using asingle control system. The main benefit of sucha system lies in its ability to detect the presenceof a peak on the elemental detector and tosubsequently trigger specific experiments such asMS-MS on the second detector, thereby improvingits detection/identification power. Furthermore,additional information from the elemental detector

468 DETECTION

regarding the presence of other elements such asS, P, Cl, etc. in the detected peak, can be usedto speed up library searching in order to identifythe subsequently detected peak on the structuraltype detector.

Finally for routine analytical work the use ofa single control system can dramatically improvethroughput through ease of use. It does alsoprovide an easier platform for compliance withQC/QA and other regulatory protocols.

4.4 Multiple combinations

Although the combination of an elemental detec-tor and organic mass spectrometry yields mostof the necessary information required for speci-ation, it is likely that the complexity of systemsunder investigation will increase and more com-plex organic structures will be studied. This willmean that the fragmentation and accurate massinformation may not be sufficient to fully deter-mine the structure of the organic compounds. Insuch cases, multiple hyphenation is a possibilityand Nicholson et al. [2] have shown that it is pos-sible to couple ICP-MS, ESI-MS, NMR and diodearray detectors to the eluent from a single HPLCinstrument. The additional structural informationprovided, primarily by the NMR, greatly increasesthe chances of being able to correctly assign astructure and obtain quantitative measurements ina single run.

5 NEW TECHNIQUES

With the ultimate aim being a combination of anelemental-specific quantification technique and astructural identification technique, in a single sys-tem, several workers are investigating new tech-niques largely, although not exclusively, centredon new ion sources for mass spectrometry.

5.1 Alternatives to the ICP

Many groups are working on alternatives tothe ICP [4]. These include development of the

helium ICP and microwave induced plasma (MIP)sources, and also the use of high power nitrogenmicrowave induced plasmas [5, 6]. While themain focus of this work has been to overcomethe problems associated with the argon plasma,primarily the formation of interfering argides inICP-MS, in recent years the He-MIP has beenapplied to the analysis of halogens, usually coupledto gas chromatography which uses a heliummobile phase.

These alternative plasmas, however, are notintended to address the requirement for an instru-ment which combines elemental and molecularinformation in a single run, nor is there any indi-cation that they have the potential to do so.

One area of research which does show potentialis the development of the low pressure, lowpower plasma [7–9], and several workers havedeveloped what is termed a ‘tuneable plasma’,(Marcus et al. provide a recent review [10]). Suchsystems, coupled to mass spectrometry, have beenapplied to the analysis of organometallics andorganohalides introduced via a gas chromatograph,and have been used to produce either elementalor molecular information, yielding spectra whichare very similar to those produced by an electronimpact (EI) source mass spectrometer.

While showing significant potential, these tune-able plasmas suffered from linearity and sensitivityproblems. Some of these have been alleviated bythe use of reagent gases to modify the ionisationprocess [11]. This technique enabled the detectionof molecular spectra for chlorobenzene, iodoben-zene and dibromobenzene, with detection limits of100, 140 and 229 pg, respectively.

It must be stated that work with such tunableplasmas has concentrated on obtaining both ele-mental and molecular information, but under dif-ferent conditions; therefore each sample has to berun twice. Therefore the only advantage offeredby this solution over the use of dedicated ele-mental and ESI analyser is a lower capital cost.In addition, so far the work has concentratedon the analysis of volatile species introduced bya gas chromatograph. It is questionable whethersuch tunable plasmas will be robust enough tofunction with aqueous samples in HPLC solvent

REFERENCES 469

systems [12]. This is a very serious limitation tothe applicability of the technique.

5.2 Glow discharge

One of the most exciting areas of research hasemerged from developments in the glow dischargesource. Glow discharge has largely been applied tothe analysis of conducting solid samples, althoughthe development of the radio frequency glowdischarge (RF-GD) has further extended the rangeto nonconducting materials [13].

You et al. [14] have demonstrated the use of aparticle beam interface organic mass spectrometerfitted with an RF-GD source instead of theconventional EI source, and were able to produceboth elemental and molecular species in thesame spectra, although with poorer detectionlimits compared to conventional elemental analysistechniques such as ICP-MS.

Majidi et al. [15, 16] observed that ‘glowdischarge plasmas are steady state but highlyheterogeneous plasmas with a number of dynamicprocesses occurring simultaneously.’ They used amicrosecond pulsed power supply, instead of aDC source, to produce a transient glow dischargeplasma. As a result, the nature of the plasma ishighly time dependent, and therefore the effect ofdifferent amounts of energy imparted to the analytecan be determined by sampling from the plasmaat the appropriate time during or after the pulse.They used a TOF mass spectrometer to givethem the ability to ‘time gate’ different partsof the plasma, thereby achieving anything from‘soft’ chemical ionisation, giving molecular ioninformation, through EI type fragmentation to fullelemental ionisation. Results have been presentedfor a number of volatile organic compounds, suchas ethylbenzene, toluene and p-xylene, and theorganometallic tungsten hexacarbonyl.

It could be postulated that this apparatus,coupled with a particle beam interface suchas the one used by You et al. may result inan instrument that meets the requirement for atechnique capable of measuring both elemental andmolecular information in a single run, although

much work would be needed before the systemcould be commercialised.

6 CONCLUSIONS

The fact that there is a scientific need to developnew instrumentation for speciation analysis ishopefully now well established. The current stateof play involving many different approaches sug-gests that we are a long way from our single‘Speciator’ instrument. It would be foolish to tryto predict at this stage which area will dominatein the future, as this will depend not only uponscientific factors.

The extent to which speciation becomes aroutine technique will depend on the need forit. Although we would like to think that thescientific requirement was enough, in the realworld, instrument companies will only develop aproduct for which they can see a well definedmarket. This will not happen until the majority ofWestern countries have established environmentallegislation which requires speciation, rather thantotal element measurements to be made. This ishappening, but legislation proceeds at a snail’space and it will be a number of years before itbecomes the standard rather than the norm.

7 REFERENCES

1. Marchante-Gayon, J. M., Thomas, C., Feldmann, I. andJakubowski, N., J. Anal. At. Spectrom., 15, 1093 (2000).

2. Nicholson, J. K., Lindon, J. C., Scarfe, G. B., Wilson,I. D., Abou-Shakra, F., Sage, A. B. and Castro-Perez, J.Anal. Chem., 73, 1491 (2001).

3. Wilson, I. D., Chromatographia , 52, S28 (2000).4. Evans, E. H., Giglio, J. J., Castillano, T. M. and Caruso,

J. A., Inductively Coupled and Microwave Induced PlasmaSources for Mass Spectrometry , Barnett, N. W. (Ed.),Royal Society of Chemistry, Cambridge, 1995.

5. Deutsch, R. D. and Hieftje, G. M., Appl. Spectrosc., 39,214 (1985).

6. Ohata, M. and Furuta, N., J. Anal. At. Spectrom., 13, 447(1998).

7. Waggoner, J. W., Belkin, M., Sutton, K. L., Caruso, J. A.and Fannin, H. B., J. Anal. At. Spectrom., 13, 879 (1998).

8. Rodriguez, J., Pereiro, R. and Sanz-Medel, A., J. Anal.At. Spectrom., 13, 911 (1998).

470 DETECTION

9. Belkin, M., Waggoner, J. W. and Caruso, J. A., Anal.Commun., 35, 281 (1998).

10. Marcus, R. K., Evans, E. H. and Caruso, J. A., J. Anal.At. Spectrom., 15, 1 (2000).

11. O’Connor, G., Ebdon, L. and Evans, E. H., J. Anal. At.Spectrom., 12, 1263 (1997).

12. Pack, B. W., Broekaert, J. A. C., Guzowski, J. P., Poehl-man, J. and Hieftje, G. M., Anal. Chem., 70, 3957 (1998).

13. Marcus, R. K., Harville, T. R., Mei, Y. and Shick, C. R.Jr., Anal. Chem., 66, 902A (1994).

14. You, J., Dempster, M. A. and Marcus, R. K., Anal. Chem.,69, 3419 (1997).

15. Steiner, R. E., Lewis, C. L. and Majidi, V., J. Anal. At.Spectrom., 14, 1537 (1999).

16. Majidi, V., Moser, M., Lewis, C. L., Hang, W. andKing, F. L., J. Anal. At. Spectrom., 15, 19 (2000).

5.11 Biosensors for Monitoring of Metal Ions

Ibolya Bontidean and Elisabeth CsoregiLund University, Sweden

Wolfgang SchuhmannRuhr Universitat Bochum, Germany

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 4712 Whole Cell Biosensors . . . . . . . . . . . . . . . 4723 Enzyme and Apoenzyme Based Biosensors 4784 Antibody or Protein Based Biosensors . . . . 478

5 Protein Based Capacitive Biosensor . . . . . . 4786 Conclusions . . . . . . . . . . . . . . . . . . . . . . . 4817 Acknowledgements . . . . . . . . . . . . . . . . . . 4828 References . . . . . . . . . . . . . . . . . . . . . . . . 482

1 INTRODUCTION

Out of the 90 naturally occurring elements, 21 arenonmetals, 16 are light metals and the remaining53 are heavy metals. Most of the metals are tran-sition elements with densities above 5 g cm−3 andhave incompletely filled d orbitals. Some of thecations formed by these metals have the abilityto form potentially redox-active complexes withsuitable ligands. Therefore metal cations play animportant role as trace elements in many sophis-ticated biochemical reactions. At higher concen-trations, however, metal cations form unspecificcomplexes and have toxic effects in the cell. Ofthe 53, only 17 metals are available to the livingcell and form soluble cations, thus showing biolog-ical influence: Fe, Mo, Mn, Zn, Ni, Cu, V, Co, W,Cr, As, Ag, Sb, Cd, Hg, Pb and U. Some of theirmain properties are presented in Table 5.11.1 [1].

The toxicity of metals is based not only on theiroxidation state, but also on the form in which theyoccur, i.e. whether it is elemental, inorganic ororganometallic. For example organomercurials aremore toxic than Hg2+, ethylated lead is extremely

toxic and the same is valid for arsenic. Somemetals (e.g. Ni, Co, Zn, Cu) are essential tomicroorganisms as trace nutrients, in contrast toothers (e.g. Hg, Cd, Pb) which are extremely toxiceven at trace levels; however, all metals are toxicin µM to mM concentrations.

Because of the extreme toxicity metals displayfor various forms of life, and their broad distribu-tion in nature, detection of metals has evolved frombeing an analytical task into a necessity in variousareas such as medicine, food industry, environ-ment, etc. Recognizing the importance of metalmonitoring, several methods have been developedover the years. Powerful methods such as atomicabsorption and emission spectroscopy [2] or induc-tively coupled plasma mass spectroscopy [2, 3]have been described and are commercially avail-able. These methods show good selectivity, sensi-tivity, reliability and accuracy; however, they oftenrequire very expensive instrumentation operated byhighly skilled personnel.

Electrochemical detection methods includingion-selective electrodes, polarography and other

472 DETECTION

Table 5.11.1. Metals with biological influence and their oxidation forms.

Metal Toxicity Available inorganic ionic forms

V Mostly toxic V(V) in vanadateCr Cr(III) has physiological role in man, Cr(VI) is toxic Cr(VI) in chromate and Cr(III)Mn Very low toxicity, except in human where it acts on

the central nervous systemAny oxidation state between

Mn(II) and Mn(VII), Mn(II)predominant

Fe Biologically most important metal, not toxic Fe(II) and Fe(III)Co Co(II) has medium toxicity, but Co dust causes lung

diseaseAlmost only Co(II), rarely Co(III)

Ni Medium toxicity, may cause nickel allergy to humans Ni(II) and very unstable Ni(III)Cu Toxicity is based on the ability of Cu to easily

interact with radicalsCu(II) and Cu(I)

Zn Very important physiological function, ‘no lifewithout Zn’, very low toxicity

Only as Zn(II)

As Well-known toxin even at trace concentrations As(V) in arsenate and As(III) inarsenite

Mo Molybdate is the biologically most importantoxyanion

Mo(VI) in molybdate

Ag Very toxic due to its strong complex with sulfur Ag(I) in Ag2SCd Toxic due to thiol binding and protein denaturation Cd(II) in CdSHg Metal with the strongest toxicity, no beneficial

functionHg(II) in HgS

Pb Toxic for humans, acts on the nervous system, onblood pressure and on reproduction

Pb(II)

U Radioactive toxin, no beneficial function is known U(VI) in UO2−2

voltammetric methods [4] are less expensive buttheir main disadvantage is their inability to detectmetals at extremely low concentration.

As environmental concentrations of metals arereduced, increasingly sensitive analytical methodsare required to monitor their distribution. More-over, none of the above mentioned methods is ableto selectively detect the amount of metal which isbioavailable and therefore likely to present a riskto living organisms.

In this respect, biosensors are useful analyticaltools since they are able to monitor that part of thetotal metal concentration that is available for thebiological component used as recognition element.A large variety of biological recognition elementsand transducers have been used in biosensor con-struction for metal detection. Different biosensorarchitectures are described briefly below and theirmain characteristics (dynamic range, DR; limit ofdetection, LOD) are presented in Table 5.11.2.

2 WHOLE CELL BIOSENSORS

Whole cell biosensors can be constructed usingcertain microorganisms, e.g. bacteria, yeasts, fungi,

lichens, mosses, and water plants due to theirability to accumulate metals [5]. The advantages ofusing intact cells as sensing element in a biosensorare: (i) microorganisms are usually more tolerantto assay conditions than isolated biomolecules, dueto mechanisms that enables them to regulate theirinternal composition, (ii) microorganisms provideinformation about the bioavailability of the analytebecause the analyte must be taken up beforethe response is produced, (iii) microorganisms arerelatively cheap since large quantities of theseliving organisms can be prepared comparablyinexpensively.

Among the disadvantages of using microor-ganisms are: (i) increased response times of thebiosensors (ii) difficulties of the regeneration of thesensor (iii) variation of the response when usingcells from different batches (iv) influence of thebiosensor response by culture age, temperature,cell density and aeration during induction. Someof the microorganisms have learned to survive andgrow in environments containing metals, develop-ing resistances to metals such as Zn, Cu, Ni, Co,Hg, Cr, and Pb.

WHOLE CELL BIOSENSORS 473

Table 5.11.2. Metal biosensors and their properties.

Biological molecule Transducer type Operating conditions M2+ LOD DR Ref.

Wholecell

Mosses Sphagnum Sp. StrippingDifferential pulseVoltammetry

Acetate pH 6.0,IS 0.7, 10 %moss, carbonpaste electrodes

Pb2+ 2 µg L−1 5–125 µg L−1 [45]

Bacteria Nitrosomonas Amperometric Oxygen uptake rateis measured

Cd2+Cu2+

8.3 mg L−1

173 mg L−1[46]

E. coli + merpromoter + luxgenes fromVibrio fischeri

Optical detection Bioluminescence ismeasured at28 ◦C, 30 minresponse timeunder aeration

Hg2+ 0.1 µM 20 nM–4 µM [8]

E. coli + lacZgene

Optical detection Electrochemicalassay ofβ-galactosidaseactivity

Cd2+ nM [47]

Synechococcussmt + lux fromVibrio fischeri

Optical detection Luminescence ismeasured

Zn2+ 0.5–4 µM [48]

E. coli + luxgenes fromVibrio fischeri

Optical detection Hg2+Cu2+

0.1 µM0.1 µM

[49]

R. silverii + luxoperon fromVibrio fischeri

Optical detection 23 ◦C, 0.2 %acetate, 20 mMMOPS, pH 7.0,20 µg mL−1

tetracycline

Cu2+Zn2+Cd2+Cr6+Pb2+Tl+Ni2+

2 µM5 µM5 µM1 µM1 µM

2–40 µM5–250 µM5–200 µM1 µM–40 µM1 µM–40 µM

[10, 50]

R. silverii + luxoperon fromVibrio fischeri

Optical detection Microorganismsimmobilized inpolymermatrices, 25 ◦C

Cu2+ 1 µM [6]

E. coli + mer-luxplasmid

Optical detection Luciferase activityis detected

Hg2 1–10 000 nM [51]

E. coli + luxoperon

Optical detection 30 ◦C, M9 medium Hg2+Cu2+

10 nM1 µM

[52]

E. coli + fireflyluciferase gene

Optical detection Luminescence ismeasured inmicrotiter platesafter 60 min, at30 ◦C

Hg2+ 0.1 fM 0.1 fM–0.1 µM [7]

Staphylococcusaureus + fireflyluciferase gene

Optical detection Luminescence ismeasured inscintillationcounter after60 min;

AsO43−

Cd2+1 µM1 µM

1–5 µM1–20 µM

[12]

Staphylococcusaureus + fireflyluciferase gene

Optical detection Luminescence ismeasured inmicrotiter plates,30 ◦C

Cd2+Pb2+Hg2+

10 nM–1 µM33 nM–330 µM33–100 nM

[14]

Bacillussubtilis + fireflyluciferase

Optical detection Luminescence ismeasured inmicrotiter plates,30 ◦C

Cd2+Zn2+

3.3 nM–1 µm1–33 µM

[14]

Yeast Fluorescencedetection

Light emitted at509 nm ismeasured whenthe system isexcited at 395 nm

Cu2+ 0.1 µM [53]

(continued overleaf )

474 DETECTION



Tissue Cucumber leaves Amperometricdetection

Effect of metal onthe hydrolysis ofcysteine ismeasured in thepresence ofL-cysteinedesulfhydrolase

Pb2+Cd2+

30 nM10 nM

[54]

Protein Apophytochelatin UV spectrophoto-metric detectionat 215 nm

270 mMapophytochelatinis used

Cd2+ 1–6 mg L−1 [43]

MerR-LacZα:M15complex

Spectrophotometricdetection

Microtiter platescoated withBSA–divinyl-sulphone–glutath-ion were treatedwith Hg2+concentrationsand after washingthe protein wasbound to it

Hg2+ µg L−1 level [44]

glutathione UV spectrophoto-metric detectionat 215 nm

160 mM glutathionis used

Cd2+ 1–8 mg L−1 [43]

pH measurement Protein crosslinkedwithglutaraldehydeand entrappedbehind a dialysismembrane

Cd2+ 10–80 mg L−1 [43]

GST-SmtA Capacitive Protein immobilizedwithcarbodiimide ona thiol modifiedgold electrode

Cu2+Hg2+Cd2+Zn2+Pb2+

1 fM1 fM1 fM1 fM1 fM

1 fM–1 mM1 fM–1 mM1 fM–1 mM1 fM–1 mM1 fM–1 mM

[11, 74]

MerR Capacitive Protein immobilizedwithcarbodiimide ona thiol modifiedgold electrode

Cu2+Hg2+Cd2+Zn2+

1 fM1 fM1 fM1 fM

1 fM–1 mM1 fM–1 mM1 fM–1 mM1 fM–1 mM

[11, 74]

MerP Capacitive Protein immobilizedwithcarbodiimide ona thiol modifiedgold electrode

Hg2+ 1 fM 1 fM–1 mM [68]

Phytochelatins Capacitive Protein immobilizedwithcarbodiimide ona thiol modifiedgold electrode

Cu2+Hg2+Cd2+Zn2+Pb2+

1 fM1 fM10 fM<1 fM1 fM

1 fM–10 mM1 fM–10 mM10 fM–10 mM<1 fM–10 mM1 fM–10 mM

[70]

Antibody Antibody againstCd-EDTAcomplex


Microtiter plateswere coated withCd-EDTA-BSAconjugate andthen the antibodywas added,HEPES bufferpH 7.0–7.2

Cd2+ 7 µg L−1 10–2000 µg L−1 [42]

Enzyme Urease ISFET Inhibition of ureaseimmobilized ondifferentmembranes,0.02 M HEPES,25 ◦C, batchmode

Cu2+Hg2+Cd2+Pb2+

1–10 mg L−1

0.25–5 mg L−1

3–10 mg L−1

2–10 mg L−1

[31]




ISFET Inhibition of ureaseimmobilized in aNafion film,20 ◦C,

Hg2+Cu2+

1 µM3 µM

[39]

Thermometricdetection

Acid urease isimmobilised oncontrolled-poreglass

Cu2+ 5–100 µM [37]

Ammonia sensor Inhibition of urease,cuvette test withammoniasensitive coatingon the wall, 0.1 Nmaleate bufferpH 6

Cu2+Hg2+Zn2+Pb2+

0.25 mg L−10.07 mg L−1

50 mg L−1

100 mg L−1

0.4–0.7 mg L−1

0.07–1 mg L−1

50–70 mg L−1

100–350 mg L−1

[38]

Ammonia sensor Enzyme reactorwith ureaseinhibited bymercury, enzymeimmobilized onglass beads

Hg2+ 0–15 nM [29]

pH sensor Inhibition of ureaseimmobilized withthymol bluecovalently boundto aminopropylglass at the tip ofan optical fiber

Cu2+Hg2+

2 µg L−1 [34]

Conductometricdetection

Enzyme oninterdigitatedgold electrodes,residual activityof urease ismeasured, 5 mMTris-HNO3pH 7.4, 50 mMurea

Hg2+Cu2+Cd2+Pb2+Co2+

1–50 µM2–100 µM5–200 µM0.02–5 mM10–500 µM

[36]

Conductometricdetection

Inhibition of ureaseis monitored witha standingacoustic wavedevice

Hg2+ 20 µg L−1 [35]

Fluorimetricdetection at340/485 nm

Flow system,enzymeimmobilized oncontrolled poreglass, 0.005 Mphosphate bufferpH 6.5

Hg2+ 0.5–100 µg L−1 [32]

Potentiometricdetection

Urease entrapped inPVC membraneat the surface ofiridium oxideelectrode

Hg2+ [40]

Fluorescencedetection at340/455 nm

Flow system,inhibition ofurease detectedusingo-phthalaldehyde

Hg2+ 2 µg L−1 [33]

IrTMOS Ammonia detectionby IrTMOS,0.05 M Tris-HCl,pH 8.3

Hg2+ 0.005 µM [30]

(continued overleaf )

476 DETECTION



L-Lactatedehydrogenase

Amperometricdetection

Enzymecoimmobilizedwith L-lactateoxidase on thetop of an oxygenelectrode

Hg2+Cu2+Zn2+

1 µM10 µM25 µM

[55]

Glycerophosphateoxidase

Oxygen electrode Inactivation ofenzyme by metalions, enzymeimmobilized byreticulation ingelatine film orcovalent bindingon a membrane

Hg2+ µM 20–500 µM [56]

Pyruvate oxidase Oxygen electrode Hg2+ 10 nM [56]L-Lactate

dehydrogenaseAmperometric

detectionEnzyme

coimmobilizedwith L-lactateoxidase on thetop of an oxygenelectrode

Hg2+Cu2+Zn2+

1 µM10 µM25 µM

[55]

Cholinesterase Voltammetricdetection

Flow system,enzymeimmobilized onnitrocellulosefilm withglutaraldehyde

Pb2+Cu2+Cd2+

5 µM50 nM5 µM

[41]

Alkalinephosphatase


Chemiluminescencefrom enzymecatalyzedhydrolysis of aphosphatederivative of1,2-dioxetane ismeasured

Zn2+ 0.17 mg L−1 [19]

Horseradishperoxidase


Inactivation of theenzyme ismeasured

Hg2+ 0.1 µg L−1 [55]


Inhibition ofenzymeimmobilized onsolid supports ismeasured

Hg2+ 0.1 ng L−1 Four orders ofmagnitude

[57]

Invertase Amperometricdetection

Enzyme inhibitionby mercury ismeasured

Hg2+ 10–60 µg L−1 [58]


Inhibition ofenzymeimmobilized on amembrane ismeasured

Hg2+ 1 µg L−1 [59]

Alcohol, sarcosineor glycerol-3-Poxidase


H2O2 production ismeasured with aRu/graphiteworkingelectrode at+700 mV versus.Ag/AgCl

Hg2+Ni2+Cu2+V5+

0.05–0.5 mg L−1 [60, 61]

Urease +acetylcholin-esterase

ISFET Inhibition of theenzymes by HMis measured

Hg2+Cu2+Cd2+Co2+

µM range [62]

Acetylcholin-esterase


Inhibition ofenzyme by metalions is measured

Cu2+Cd2+Fe2+Mn2+

0.01 pM1 pM10 pM100 pM

[63]




Apoenzyme Alkalinephosphatase

Calorimetricdetection

Enzymeimmobilized onepoxide acrylicbeads, 100 mMTRIS-HCl pH 8.0

Zn2+Co2+

0.01–1.0 mM0.04–1.0 mM

[17]

Alkalinephosphatase


Flow injectionsystem, change inabsorbance at405 nm ismeasured

Zn2+Co2+

sub-µM 0.1–10 µM1–200 µM

[15, 18]

Potentiometricdetection

Flow-throughISFET, pH shiftdetected

Zn2+ 0.01–1.0 mM [16]

Optical detection Chemiluminescencefrom enzymecatalyzedhydrolysis of aphosphatederivative of1,2-dioxetane ismeasured

Zn2+ 0.5 µg L−1 0.5–50 µg L−1 [19]

Ascorbate oxidase Calorimetricdetection

Flow system,enzymeimmobilized onporous glassbeads

Cu2+ 1–50 µM [25]

Ascorbate oxidase Spectrophotometricdetection

Absorbance at265 nm ismeasured

Cu2+ 0.1–10 µM [27]


Polarographicoxygen electrodeis used

Cu2+ 0.5–2 µM [26]

Carbonicanhydrase

Fluorescencelifetime detection

The affinity of theapoenzyme fordifferent HM isused

Cu2+Zn2+Cd2+Co2+Ni2+

pMnMnM

[24]

Calorimetricdetection

Flow system,enzymeimmobilized onporous glassbeads

Zn2+Co2+

25–250 µM50–200 µM

[20, 21]

Optical detection at326/460 and560 nm

Recognition ofmetal ion byapoenzymetransduced by thedansylamidefluorescent probe

Zn2+ 40–1000 nM [22, 23]

Galactose oxidase Calorimetricdetection

Cu2+ 5–20 mM [28]


Detection withoxygen electrode

Cu2+ 0.1–10 mM [26]

Alkalinephosphatase +ascorbateoxidase


Enzymescoimmobilizedon a polymermembraneattached to apolarographicoxygen electrode

Cu2+Zn2+

2–100 µM2–200 µM

[64]

Tyrosinase Amperometricdetection

Flow system withoxygen electrode

Cu2+ up to 0.05 mM [65]

478 DETECTION

The general approach to constructing biosensorsbased on intact cells is to fuse an inducible pro-moter from one of the metal resistance operons toa reporter gene that includes genes for biolumines-cent proteins such as luciferase. The light producedby luminescent proteins can be measured with pho-tometers or luminometers. When the bacteria emitlight, optical fibers can be used to transmit thelight to the detection device [6]. Several biosen-sors were developed using the promoter–reportergene concept. Different bacteria were geneticallymodified with lux genes from Vibrio fischeri or lucgenes from firefly luciferase used as reporter genesfused to promoter regions of operons responsiblefor resistances to various metals: mer operon fordetection of mercury [7–9] and copper [6, 10, 11],ars operon for detection of arsenic [12, 13], or cadoperon for detection of cadmium [12, 14].

3 ENZYME AND APOENZYMEBASED BIOSENSORS

The use of enzymes and apoenzymes as recogni-tion elements in biosensors for metal detection isbased on the fact that the metal may act as catalyst(cofactor) or inhibitor. There are enzymes that havebinding selectivity for the metal ions which partic-ipate as cofactors in catalysis, a fact that has beenexploited, e.g. in biosensors based on apo formsof alkaline phosphatase [15–19], carbonic anhy-drase [20–24], ascorbate oxidase [25–27], andgalactose oxidase [26, 28].

A more frequently used approach is based onthe inhibition of the enzyme activity in the pres-ence of metals allowing the correlation of thedecrease in enzyme activity with the metal con-centration. Most such biosensors are based onthe inhibition of urease [29–40], but also car-bonic anhydrase [23], cholinesterase [41], alkalinephosphatase [19] were used for this purpose. Ascan be derived from Table 5.11.2 enzymes orapoenzymes have been coupled to various trans-ducers such as amperometric and potentiomet-ric electrodes, optical fibers, conductometric orpiezoelectric devices. As compared to cell-basedbiosensors, heavy metal biosensors using enzymes

or apoenzymes as biological recognition elementare easier to construct and can be regenerated to acertain extent.

4 ANTIBODY OR PROTEIN BASEDBIOSENSORS

There have been a few attempts to use anti-bodies [42] or proteins [43, 44] as recognitionelements for metal biosensors. Since immuno-assays are quick, inexpensive, easy to perform,and portable, these assays are becoming increas-ingly accepted for environmental applications. Pro-teins, e.g. phytochelatins or metallothioneins, canbe used as biological components for metal bind-ing by their immobilization at the surface ofan appropriate transducer, e.g. pH-sensitive field-effect transistor, resonating piezoelectric crys-tal or optical devices. These proteins selectivelybind metal ions via thiolate complex formation.Changes within the layer of the immobilized pro-tein (e.g. release of protons, changes in mass andoptical properties) are transformed into measurablesignals by the transducer.

A common drawback of the abovementionedbiosensors, excepting whole cell and protein basedones, is the impossibility to detect bioavailableconcentrations of metals. Moreover, cell basedbiosensors can only detect metals down to micro-molar level, while the protein based capacitivebiosensors are sensitive to concentrations in thefemtomolar range. A more detailed description ofsuch biosensor design is presented below.

5 PROTEIN BASED CAPACITIVEBIOSENSOR

Proteins from three different classes of metalbinding proteins have been used as biologicalcomponent of a biosensor, namely synechococ-cal metallothionein, SmtA [66], mercury resistanceproteins, MerR [67] and MerP [68], and a phy-tochelatin [69, 70].

The SmtA protein is a metallothionein (MT)from the cyanobacterium Synechococcus. Metal-lothioneins are small proteins which sequester

PROTEIN BASED CAPACITIVE BIOSENSOR 479

metal ions in a ‘cage’ structure. In animalmetallothioneins there are two domains, each ofwhich can sequester three or four metal ions.Metal binding is associated with a large confor-mational change in the protein, as the sulfhydrylgroups of cystein residues coordinate the metalions. The SmtA protein was overexpressed asa glutathione-S-transferase-metallothionein fusionprotein, which was subsequently used as the bio-logical recognition element of a biosensor.

MerR is the regulatory protein responsible forinducible expression of mercury resistance pro-teins [67]. Hg(II) binds to the dimeric MerR pro-tein, and there is genetic evidence that this resultsin a conformational change in the protein [71, 72].

MerP is a 72 amino acid periplasmic mercurybinding protein containing two cysteine residues,Cys-14 and Cys-17, both necessary for specificbinding of Hg2+. MerP contains a GMTCXXCmetal binding domain that also can be foundin Menkes copper transporting ATPase, a protein

responsible for Menkes syndrome, which is alethal X-chromosome hereditary disease of copperstarvation. The amino acid sequence similarity forthe two proteins is shown below and the exactmatches were denoted with stars (*):Due to its similarity with Menkes copper trans-porting ATPase, a thorough study of MerP pro-tein is important to elucidate the mechanism ofmetal transport in the cell, identifying the residuesresponsible for the specific metal binding. Thedeveloped MerP based biosensor seems to be apromising tool in this respect [68].

Plants, algae and some fungi are capable ofsynthesizing, on exposure to metals, thiol-richpeptides such as (1-glutamylcysteinyl)n-glycinewith n = 2 to 11, also known as phytochelatins.Phytochelatins (PCs) seem to be involved indetoxification and homeostasis of trace metalsin plants and thus serve functions analogousto metallothioneins in animals. Phytochelatinsare enzymatically synthesized by a specific

Menkes ATPase LTQETVINID GMTCNSCVQS IEGVISK-KPG VKSIRVSLANMerP ATQTVTLAVP GMTCAACPIT VKKALSK-VEG VSKVDVGFEK

** **** * *** * * *

1-glutamylcysteine-dipeptidyl transpeptidase (phy-tochelatin synthase), which is activated by thepresence of metal ions and uses glutathioneas a substrate. Phytochelatins bind metal ionsby thiolate coordination yielding intracellularmetal complexes.

These different types of metal-binding proteinshave been coupled to a highly sensitive capaci-tive transducer. The principle of capacitance mea-surements is based on the modulation of theelectrical double layer and was described ear-lier [11, 73–75]. The total capacitance measuredis given by the sum of the capacitances of the dif-ferent layers covering the electrode surface: (i) theself-assembled nonconductive layer needed forprotein binding, (ii) the protein layer, and (iii) theionic space charge formed by the hydrated ions ofthe buffer.

When metal ions bind to the protein, a confor-mational change in the protein’s structure invokesa change of the capacitance which is detected

and correlated with the metal concentration. Theassumed conformational change is schematicallydepicted in Figure 5.11.1.

Proteins were covalently immobilized on thesurface of gold electrodes after carbodiimideactivation of the carboxylic acid head groups ofa thioctic acid monolayer. The protein-modifiedbiosensors were then used as working electrode ina three(four)-electrode electrochemical flow cell.A Pt foil was used as auxiliary electrode, anda Pt wire and a commercial Ag/AgCl electrodeserved as quasi and real reference electrodes. Theimportance of using a second Ag/AgCl referenceelectrode was explained previously [73].

Measurements were made by applying a poten-tial pulse of 50 mV at the working electrode afterinjection of 250 µL sample solutions into the10 mM borate buffer carrier flow, kept at a pHvalue of 8.75. The current transients invoked bythe application of the potential pulse are recorded,and the decrease in current is evaluated according

480 DETECTION

SS

O

NH

S S

O

NH

S S SS

O

NH

S S

O

NH

S S

M2+

M2+

M2+

M2+M2+

Figure 5.11.1. Schematic representation of conformational changes occurring upon binding of metal ions to proteins.

−1400

−1200

−1000

−800

−600

−400

−200

0−16 −14 −12 −10 −8 −6 −4 −2

Cap

acita

nce

chan

ge/n

Fcm

−2

log([M2+]/M)

−200

−160

−120

−80

−40−15 −14 −13 −12 −11 −10

Figure 5.11.2. Typical calibration curves obtained for Pb (�), Hg (� ), Cd (Ž), Zn (�), and Cu (�) obtained with a GST-SmtAbased capacitive biosensor (10 mM borate buffer, pH 8.75, flow rate 0.5 ml min−1, room temperature).

to the following equation:

i(t) = u

Rsexp

(− t

RsCt

)

where i(t) is the current at time t , u is the amplitudeof the potential pulse applied, Rs is the resistancebetween the gold and the reference electrodes, Ct

is the total capacitance over the immobilized layerand t is the time elapsed after the potential pulsewas applied.

Typically, all these protein based capacitivebiosensors detected metals with high sensitivity

and a very broad detection range starting at fMlevel up to mM concentrations. However, dif-ferences were noticed depending on the pro-tein used and the metal ion to be detected.Typical calibration curves obtained for the deter-mination of different metal ions using the GST-SmtA based electrode are shown in Figure 5.11.2,while Figure 5.11.3 depicts the signals obtainedfor the determination of Hg2+ with four differ-ent biosensors.

The SmtA-based sensors show highest sensitiv-ity and extended linear range for Hg2+ at lower

CONCLUSIONS 481

−50

−100

−150

−200

0−14−15 −13 −12 −11 −10

−200

−14 −12 −10 −8 −6 −4 −2 0−16

−400

−600

−800

−1000

−1200

−1400

0

Cap

acita

nce

chan

ge/n

Fcm

−2

log([Hg2+]/M)

Figure 5.11.3. Calibration curves obtained for the determination of Hg2+ using GST-SmtA (•), MerR (�), MerP (� ) andphytochelatin (�) based capacitive biosensors.

Table 5.11.3. Concentrations at which the shape of the capacitive signal changes forcapacitive biosensors based on GST-SmtA, MerR and phytochelatin.

Protein Metal ion

Cd Hg Cu Zn Pb

GST-SmtA 10−9 M 10−10 M 10−6 M 10−7 M 10−7 MMerR 10−6 M 10−7 M 10−5 M 10−6 M No resultsPhytochelatin 10−6 M 10−7 M 10−5 M 10−5 M 10−4 M

concentrations and a broad selectivity pattern,sensing Pb2+, Hg2+, Cd2+, Zn2+, and Cu2+ ions(Figure 5.11.2), while the MerR and phytochelatinbased ones were more selective towards one spe-cific metal ion, Hg2+ and Zn2+, respectively(results not shown).

Biosensors based on all four proteins (GST-SmtA, MerR, MerP and PC) are able to mea-sure Hg2+, with the GST-SmtA based one beingthe most sensitive. The shape of the calibra-tion curves may be explained by conformationalchanges related to the biological roles of the dif-ferent proteins, namely binding and regulating thetransport of Hg2+ in the animal cells (SmtA, MerP,MerR) or plant cells (PC). Obviously, dependingon the metal concentration, there are two distinct

responses of the biosensors. It can be assumedthat in the low concentration range the signal isdue to the titration of the cysteine and histidineresidues in the amino acid sequence of the proteinswith the metal ion. The accentuated capacitancechanges occurring at higher concentrations (seeTable 5.11.3) may be correlated with the confor-mational changes in the protein’s structure. For allelectrode types, the regeneration of the electrodesis possible by injecting EDTA.

6 CONCLUSIONS

Biosensors have been shown to be a versatile toolfor monitoring metal ions with high sensitivity

482 DETECTION

and selectivity. However, their bioanalytical char-acteristics (e.g. lifetime, linear range, sensitivity,selectivity) vary significantly for the different sen-sor types and fields of application. Therefore, oneshould always consider the particular applicationarea before choosing the most adequate sensor con-figuration. On the other hand, biosensors based onmetal-binding proteins are able to detect bioavail-able concentrations of metals with a very highsensitivity and within an extremely broad con-centration range. Considering present progress inbiomolecule engineering, a further improvement intheir selectivity is expected, and thus, they seemto be very promising tools for further fundamentaland applied studies.

7 ACKNOWLEDGEMENTS

The European Commission (CEMBA contractEVK1-1999-0008) and the Swedish ResearchCouncil (NFR) supported this work financially.Professor Nigel Brown (University of Birming-ham, UK) and Professor Ashok Mulchandani (Uni-versity of California, Riverside, CA, USA) areacknowledged for their kind gift of SmtA, Mertype proteins and phytochelatins, respectively.

8 REFERENCES

1. Nies, D. H., Appl. Microbiol. Biotechnol., 51, 730 (1999).2. Jackson, K. W. and Chen, G. R., Anal. Chem., 68, R231

(1996).3. Burlingame, A. L., Boyd, R. K. and Gaskell, S. J., Anal.

Chem., 68, R599 (1996).4. Anderson, J. L., Bowden, E. F. and Pickup, P. G., Anal.

Chem., 68, R379 (1996).5. Wittmann, C., Riedel, K. and Schmidt, R. D., Microbial

and enzyme sensors for environmental monitoring, inHandbook of Biosensors and Electronic Noses: Medicine,Food and the Environment , Kress-Rogers, E. (Ed.), CRCPress, Boca Raton, FL, 1997, pp. 299.

6. Leth, S., Maltoni, S., Simkus, R., Mattiasson, B., Cor-bisier, P., Klimant, I., Wolfbeis, O. S. and Csoregi, E.,Electroanalysis , 14, 35 (2002).

7. Virta, M., Lampinen, J. and Karp, M., Anal. Chem., 67,667 (1995).

8. Tescione, L. and Belfort, G., Biotechnol. Bioeng., 42, 945(1993).

9. Selifonova, O., Burlage, R. and Barkay, T., Appl. Envi-ron. Microbiol., 59, 3083 (1993).

10. Corbisier, P., van der Lelie, D., Borremans, B., Pro-voost, A., de Lorenzo, V., Brown, N. L., Lloyd, J. R.,Hobman, J. L., Csoregi, E., Johansson, G. and Mattias-son, B., Anal. Chim. Acta, 387, 235 (1999).

11. Bontidean, I., Lloyd, J. R., Hobman, J. L., Wilson, J. R.,Csoregi, E., Mattiasson, B. and Brown, N. L., J. Inorg.Biochem., 79, 225 (2000).

12. Corbisier, P., Ji, G., Nuyts, G., Mergeay, M. and Sil-ver, S., FEMS Microbiol. Lett., 110, 231 (1993).

13. Tauriainen, S., Virta, M., Chang, W. and Karp, M., Anal.Biochem., 272, 191 (1999).

14. Tauriainen, S., Karp, M., Chang, W. and Virta, M.,Biosens. Bioelectron., 13, 931 (1998).

15. Satoh, I. and Aoki, Y., Denki Kagaku , 58, 1114 (1990).16. Satoh, I. and Masumura, T., Flow injection biosensing

of zinc(II) ions with use of an immobilized alkalinephosphatase reactor, in Technical Digest of the 9th SensorSymposium , Sasaki, A. (Ed.), Tokyo, Japan, 1990, p. 197.

17. Satoh, I., Biosens. Bioelectron., 6, 375 (1991).18. Satoh, I., Ann. N. Y. Acad. Sci., 672, 240 (1992).19. Kamtekar, S. D., Pande, R., Ayyagari, M. S., Marx,

K. A., Kaplan, D. L., Kumar, J. and Tripathy, S., Anal.Chem., 68, 216 (1996).

20. Satoh, I., Ikeda, K. and Watanabe, N., Microanalysis ofzinc(II) ion by using an apoenzyme thermistor, inProceedings of the 6th Sensor Symposium , Takahashi, K.(Ed.), Tokyo, Japan, 1986, p. 203.

21. Satoh, I., Continuous biosensing of heavy metal ionswith use of immobilized enzyme-reactors as recognitionelements, in Proceedings of MRS International Meetingon Advanced Materials , Karube, I. (Ed.), Pittsburgh, PA,Vol. 14, 1989, p. 45.

22. Thompson, R. B. and Jones, E. R., Anal. Chem., 65, 730(1993).

23. Thompson, R. B., Maliwal, B. P., Feliccia, V. L., Fierke,C. A. and McKall, K., Anal. Chem., 70, 4717 (1998).

24. Thompson, R. B., Patchan, M. W. and Ge, Z., U.S.patent, 1999, p. 25.

25. Satoh, I., Kimura, S. and Nambu, T., Biosensing ofcopper(II) ions with an apoenzyme thermistor containingimmobilized metalloenzymes in flow system, in Digestof Technical Papers of the 4th International Conferenceon Solid-State Sensors and Actuators, Transducers ’87 ,Matsuo, T. (Ed.), Tokyo, Japan, 1987, p. 789.

26. Satoh, I., Kasahara, T. and Goi, N., Sens. Actuat. :B , 1,499 (1990).

27. Satoh, I. and Nambu, T. Flow-injection photometricbiosensing of copper(II) ions with the use of an immo-bilized ascorbate oxidase column, in Technical Digest ofthe 10th Sensor Symposium , Nakamura, T. (Ed.), Tokyo,Japan, 1991, p. 77.

28. Satoh, I., Ann. N. Y. Acad. Sci., 613, 401 (1990).29. Ogren, L. and Johansson, G., Anal. Chim. Acta, 96, 1

(1978).

REFERENCES 483

30. Winquist, F., Lundstrom, I. and Danielsson, B., Anal.Lett., 21, 1801 (1988).

31. Sakai, H., Kaneki, N., Tanaka, H. and Hara, H., Sens. andMater., 2, 217 (1991).

32. Bryce, D. W., Fernandez-Romero, J. M. and Luque deCastro, M. D., Anal. Lett., 27, 867 (1994).

33. Narinesingh, D., Mungal, R. and Ngo, T. T., Anal. Chim.Acta , 292, 185 (1994).

34. Andres, R. T. and Narayanaswamy, R., Analyst , 120,1549 (1995).

35. Liu, D. H., Yin, A. F., Chen, K., Ge, K., Nie, L. H. andYao, S. Z., Anal. Lett., 28, 1323 (1995).

36. Zhylyak, G. A., Dzyadevich, S. V., Korpan, Y. I., Sol-datkin, A. P. and Elskaya, A. V., Sens. Actuators B ,24–25, 145 (1995).

37. Preininger, C. and Danielsson, B., Analyst , 121, 1717(1996).

38. Preininger, C. and Wolfbeis, O. S., Biosens. Bioelectron.,11, 981 (1996).

39. Volotovsky, V., Nam, Y. J. and Kim, N., Sens. ActuatorsB , 42, 233 (1997).

40. Krawczynski, T. K. V., Moszczynska, T. and Trojanow-icz, M., Biosens. Bioelectron., 15, 681 (2000).

41. Budnikov, H. C., Medyantseva, E. P. and Babkina, S. S.,J. Electroanal. Chem., 310, 49 (1991).

42. Khosraviani, M., Pavlov, A. R., Flowers, G. C. andBlake, D. A., Environ. Sci. Technol., 32 137 (1998).

43. Hilpert, R., Zenk, M. H. and Binder, F., Biosensors forthe detection of heavy metal ions, in GBF Monogr., 367(1989).

44. Klein, J. and Mattes, R., Anal. Biochem., 260, 173 (1998).45. Ramos, J. A., Bermejo, E., Zapardiel, A., Perez, J. A.

and Hernandez, L., Anal. Chim. Acta, 273, 219 (1993).46. Verschure, L., Gernaey, K. and Verstraete, W., Water

(Wijnegem, Belg.), 14, 163 (1995).47. Biran, I., Babai, R., Levcov, K., Rishpon, J. and Ron,

E. Z., Environ. Microbiol., 2, 285 (2000).48. Erbe, J. L., Adams, A. C., Taylor, K. B. and Hall, L. M.,

J. Ind. Microbiol., 17, 80 (1996).49. Holmes, D. S., Dubey, S. K. and Gangolli, S., Environ.

Geochem. Health , 16, 229 (1994).50. Corbisier, P., Thiry, E., Masolijn, A. and Diels, L., Con-

struction and development of metal ion biosensors in Pro-ceedings of the 8th Bioluminescence ChemiluminescenceInternational Symposium, (1994) p. 151.

51. Lyngberg, O. K., Stemke, D. J., Schottel, J. L. and Flic-kinger, M. C., J. Ind. Microbiol. Biotechnol., 23, 668(1999).

52. Corbisier, P., Thiry, E. and Diels, L., Environ. Toxicol.Wat. Qual., 11, 171 (1996).

53. Liu, Y., Ensor, M. and Daunert, S., Copper ion detectionusing genetically engineered yeast, in Book of Abstracts

of The 215th ACS National Meeting , Dallas, TX, 1998,032-ANYL, Part 1.

54. Kremleva, N. V., Medyantseva, E. P., Budnikov, G. K.and Bormotova, Y. I., J. Anal. Chem., 54, 151 (1999).

55. Fennouh, S., Casimiri, V., Geloso-Meyer, A. and Burs-tein, C., Biosens. Bioelectron., 13, 903 (1998).

56. Gayet, J. C., Haouz, A., Geloso-Meyer, A. and Burs-tein, C., Biosens. Bioelectron., 8, 177 (1993).

57. Shekhovtsova, T. N. and Chernetskaya, S. V., Anal. Lett.,27, 2883 (1994).

58. Bertocchi, P., Ciranni, E., Compagnone, D., Magearu, V.,Palleschi, G., Pirvutoiu, S. and Valvo, L., J. Pharm.Biomed. Anal., 20, 263 (1999).

59. Amine, A., Cremsini, C. and Palleschi, G., Mikrochim.Acta , 121, 183 (1995).

60. Compagnone, D., Bugli, M., Imperiali, P., Varallo, G.and Palleschi, G., NATO ASI Ser., 38, 220 (1997).

61. Compagnone, D., Palleschi, G., Varallo, G. and Imperi-ali, P., Amperometric biosensors for the determination ofheavy metals, in Proc. SPIE–Int. Soc. Opt. Eng., 2504,141 (1995).

62. Starodub, N. F., Torbicz, W., Starodub, V. M., Kanjuk,M. I. and Ternovoj, K. S., Transducers ’97 , 2, pp. 1383(1997).

63. Stoytcheva, M., B. Soc. Chim. Belg., 103, 147 (1994).64. Satoh, I., Sens. Actuatuators B , 13, 162 (1993).65. Mattiasson, B., Nilsson, H. and Olsson, B., J. Appl.

Biochem., 1, 377 (1979).66. Turner, J. S. and Robinson, N. J., J. Ind. Microbiol., 14,

119 (1995).67. Hobman, J. L. and Brown, N. L., Metal ions in biological

systems, in ‘Mercury and its Effects on Environment andBiology , Vol. 34, Sigel, A. and Sigel, H. (Eds), MarcelDekker, New York, 1997 p. 527.

68. Bontidean, I., Hobman, J. L., Brown, N. L. and Csore-gi, E., Anal. Chem., submitted (2003).

69. Bae, W., Chen, W., Mulchandani, A. and Mehra, R. K.,Biotechnol. Bioeng., 70, 518 (2000).

70. Bontidean, I., Ahlqvist, J., Mulchandani, A., Chen, W.,Bae, W., Mehra, R. K., Mortari, A. and Csoregi, E.,Biosens. Bioelectron., in press (2003).

71. Parkhill, J., Ansari, A. Z., Wright, J. G., Brown, N. L.and O’Halloran, T. V., EMBO J., 12, 413 (1993).

72. Parkhill, J., Lawley, B., Hobman, J. L. and Brown, N. L.,Microbiology (Reading, UK), 144, 2855 (1998).

73. Berggren, C. and Johansson, G., Anal. Chem., 69, 3651(1997).

74. Bontidean, I., Berggren, C., Johansson, G., Csoregi, E.,Mattiasson, B., Lloyd, J. R., Jakeman, K. J. and Brown,N. L., Anal. Chem., 70, 4162 (1998).

75. Berggren, C., Bjarnason, B. and Johansson, G., Instrum.Sci. Technol., 27, 131 (1999).

5.12 Possibilities Offered by Radiotracersfor Method Development in Elemental SpeciationAnalysis and for Metabolic and EnvironmentallyRelated Speciation Studies

Rita CornelisLaboratory for Analytical Chemistry, University of Ghent, Belgium

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 4842 Modes of Radioactive Decay Relevant for

Speciation Purposes . . . . . . . . . . . . . . . . . 4853 Elements Suitable for Radiotracer Studies 4864 Radionuclide Measurements . . . . . . . . . . . 488

4.1 Gamma-detection . . . . . . . . . . . . . . . 4894.1.1 NaI(Tl) scintillation detector 4894.1.2 Semiconductor detectors . . . . 489

4.2 Beta detection . . . . . . . . . . . . . . . . . 4904.2.1 Liquid Scintillation Detector 490

4.3 Autoradiography . . . . . . . . . . . . . . . . 490

4.3.1 X-ray film autoradiography 4904.3.2 Phosphor imaging . . . . . . . . . 490

5 General Sources of Error in RadioactivityMeasurement . . . . . . . . . . . . . . . . . . . . . . 491

6 Application of Radiotracers to SolveSpecific Chemical Speciation Problems . . . 4916.1 Life sciences . . . . . . . . . . . . . . . . . . 4916.2 Environmental sciences . . . . . . . . . . . 496


1 INTRODUCTION

The development of methods for chemical speci-ation analysis can be facilitated to a great extentby incorporating a suitable radioisotope of the ele-ment into the system and measuring the radiationof the isolated species [1]. The most basic assump-tion is that the radiotracer behaves exactly in thesame way as the stable isotopes of the element itcharacterises. Therefore, the radiotracer has to beincorporated under the identical form as the speciesendogenous to the system on the basis of completeisotopic exchange.

The separation steps are common to the nonra-dioactive procedures, and the only difference liesin the final detection of the analyte. The advantage

in ease of detection of trace element amountsof radioactive isotopes is unsurpassed for gammarays, and useful in many a case of positron andbeta emitters. Detection of gamma rays consists ofa simple measurement of the radioactivity withoutrequiring the type of sample preparation and cal-ibration that is necessary with other techniques.Even when the method seems effortless, suchas electrothermal atomic absorption spectrome-try (ET-AAS), inductively coupled plasma atomicemission spectrometry (ICP-AES), inductivelycoupled plasma mass spectrometry (ICP-MS) orelectrochemistry, it will be more tedious than forgamma spectrometry. This is due to the fact thatthe emission of the radiation by the radionuclide isindependent of physical and chemical influences.

MODES OF RADIOACTIVE DECAY RELEVANT FOR SPECIATION PURPOSES 485

Additionally, the method of using radiotracerscircumvents the hazard of detecting contaminationfrom other sources of the trace element studied, asonly the radioactive tracer is measured.

A major warning is needed. The sample to beanalysed for its different species should never besubjected to neutron or particle activation in anuclear reactor, cyclotron or accelerator prior tospeciation analysis. This is totally inappropriatebecause when a chemical compound undergoesnuclear bombardment, the chemical valency andthe chemical bonds might change through the Szi-lard–Chalmers effect, as this causes the radioactiveatom in the process of its formation to break loosefrom its molecule with loss of the original chemicalspecies. In 1934 Szilard and Chalmers showed thatafter the neutron irradiation of ethyl iodide mostof the iodine activity formed could be extractedfrom the ethyl iodide with water. They used a smallamount of iodine carrier (nonradioactive), reducedit to iodide, and finally precipitated it as AgI. Evi-dently the iodine–carbon bond was broken whenan 127I nucleus was transformed by neutron captureto 128I [2].

A detailed study of the Szilard–Chalmers effecton the decomposition of six different organoarseniccompounds during neutron irradiation has beenreported by Slejkovec et al. [3]. For aqueous solu-tions of monomethylarsonic acid, dimethylarsinicacid, arsenobetaine, arsenocholine, tetramethylar-sonium ion, trimethylarsine oxide the degree ofdecomposition was high (>80 % for 10 min ofirradiation at 3.8 × 1016 s−1 m−2, yielding mainlyAs(V)) whereas irradiation of solid arsenobe-taine for 60 min gave low decomposition yields(<10 %). What makes these experiments evenmore interesting is that the irradiation resulted invery high specific activities for the decompositionproducts in the samples irradiated in aqueous solu-tion. For As(V) specific activities about 1000 timeshigher than those expected from direct irradiationof As(V) were found (>3800 kBq µg−1).

Whereas radiotracer techniques are simple, itis not practicable to envisage neutron activationanalysis of the different fractions after separationof the species, owing to its time-consuming nature.First of all, the induced radioactivity is no longer

specific for the radioisotope of the element, butis now due to the mixture of isotopes that becameradioactive in the process of neutron activation. Asa consequence elaborate radiochemical separationwould be needed to separate the radioisotope ofthe species under investigation from the matrixradioactivity. Secondly the number of fractionsafter, e.g., a chromatographic separation, are aboutfiftyfold as numerous, making it an endless task.

Radiotracers are most useful in two ways:

(1) During the exploratory phase of methoddevelopment for extraction, chromatographicand electrophoretic techniques.

(2) For in vitro and in vivo, and environmentallyrelated studies about mobility, storage, reten-tion, metabolism and toxicity of trace elementspecies. It allows to follow the behaviour andtransformation of a newly administered radio-tracer, as the ‘cold’ or nonradioactive share ofthe element cannot be measured. Such radio-tracer experiments can often be done using acarrier-free radioisotope, this means all iso-topes added are radioactive, and they will notsignificantly add to the mass already present.This is an additional advantage as the origi-nal concentration of the trace element in thesystem remains unaffected.

2 MODES OF RADIOACTIVE DECAYRELEVANT FOR SPECIATIONPURPOSES [4, 5]

Radioisotopes undergo radioactive decay by emit-ting alpha, beta or gamma rays.

Alpha rays are monoenergetic and consist of anHe nucleus (mass 4, consisting of two neutrons andtwo protons); beta emitters emit either a positron(β+) or an electron (β−), and simultaneously aneutrino, beta rays are not monoenergetic, the totaldecay energy being divided over the particle andthe neutrino; gamma emitters emit discrete gammaenergies as electromagnetic waves.

Alpha emitters are mainly found among theheavier elements. As they need much more cautionto handle and require more elaborate measurementtechniques, they are not commonly used as tracers

486 RADIOTRACERS IN ELEMENTAL SPECIATION ANALYSIS

for elemental speciation purposes. It may beinteresting to mention that all elements found innatural sources with atomic number greater than83 (bismuth) are radioactive. They belong to theuranium, thorium and actinium series.

Many radioisotopes undergo decay by succes-sive β- and γ -emission. A limited number ofradioisotopes are solely β emitters, such as 3H,14C, 32P.

The decay of a radioisotope follows the expo-nential law, Nt = N0e−λt , where Nt is the (large)number of radioactive atoms at time t , N0 is thenumber present at time t = 0 and λ is a constantcharacteristic of the particular radioisotope, relatedto the half-life (t1/2) or the time required for an ini-tial large number of atoms to be reduced to halfthat number by radioactive decay (t1/2 = ln 2/λ =0.693/λ). Radioactive decay is considered to bea purely statistical process. In practical work thenumber of atoms is not measured. The radioactiv-ity emitted by a radioisotope follows the equationAt = A0e−λt . Even the radioactivity is usually notmeasured in an absolute way. In order to do anabsolute measurement it would be necessary todetermine the detection efficiency, which dependson the nature of the detection instrument, the sen-sitivity of recording the particular radiation in thatparticular instrument, and the geometrical arrange-ment of sample and detector. This is circumventedby doing relative measurements by measuring a‘standard’ solution at the beginning of the exper-iment and then comparing the radioactivity of allthe subsequent samples in identical geometry anddensity of solution to that of the ‘calibrant’. Thisprocedure is suitable for the purpose for radiotracerbased method development in elemental specia-tion studies.

3 ELEMENTS SUITABLEFOR RADIOTRACER STUDIES

Although a radioisotope exists for almost everyelement in the periodic table, only a limitednumber are suitable for studies in this way. Firstlythe half-life of the isotope must be adequate forthe duration of the experiment. As a rule, half-lives of less than a few hours are not acceptable

for this purpose. Whenever more than one isotopeof the same element is eligible, the isotope with theshortest half-life is to be preferred as this reducesthe problem of waste disposal. This option can,of course, only be considered when this isotope ismeasurable under the experimental conditions withas high a sensitivity as possible.

Gamma-emitters are more suitable because self-absorption of the radiation in the medium is eithernegligible or can be easily corrected for. Any liquidor solid form is convenient for measurement with aNaI(Tl) scintillation detector or a Ge detector unit(see next paragraph).

Electron and positron emitters are second choicebecause they have a poor penetration range (theradiation is absorbed to a large extent by themedium and the walls of the counting vial), andtherefore require labourious sample preparation.The detection method that may be envisaged fora flat sample is a Geiger–Muller counter or aproportional counter. In case of speciation workthese historic detection systems would never beconsidered because they are too impractical forserial analysis. The other detection method thatwill be considered is liquid scintillation counting,where the sample is intimately mixed with aspecial scintillation cocktail. As this techniqueis prone to quenching and luminescence error,which adversely affect the counting efficiency,suitable correction protocols must be included foreach set of experiments. The sample preparationis rather time consuming and the correctionsmay add significantly to the overall error of themeasurement.

Alpha emitters are rarely used for trace exper-iments, because they are too radiotoxic and havevery poor penetration (the radiation has a shortrange and can be stopped by the species of whichthey form an integral part, by the medium and bythe walls of the vial in which the sample is col-lected). The exceptions are the actinides and 210Po,a decay product of radiolead, which emit usefulalpha particles. These radioisotopes need to becarefully monitored in the environment. Alpha par-ticle detection can be carried out by 4π counting,2π counting, scintillation counting, alpha particlespectrometry with an ionisation chamber or with

ELEMENTS SUITABLE FOR RADIOTRACER STUDIES 487

the more common silicon surface barrier semicon-ductor, all requiring extensive sample preparationprior to counting. Elements with potentially usefulradionuclides are given in Figure 5.12.1.

Today a major drawback in doing radiotracerresearch is that fewer and fewer radioisotopes arecommercially available. Many nuclear facilitiesare closed. The demand for radioisotopes, isdwindling, except for those routinely used formedical diagnostic purposes. Whenever a specialradioisotope is needed, the manufacturers maynevertheless be willing to produce it, albeit atexorbitant costs.

The most common production mode forradioisotopes will be the irradiation of the parentisotope in a nuclear reactor at a high neutronflux. The parent isotope undergoes (n,γ ) reactionand gives rise to a radioisotope of the sameelement, but with the atomic weight increasedwith 1 unit mass. Examples are: 63Cu(n,γ )64Cu,50Cr(n,γ )51Cr, 59Co(n,γ )60Co, 74Se(n,γ )75Se, . . .

[4]. Isotopes produced in this way will not becarrier free; this means they will consist of amixture of mainly inactive carrier (nonradioactive,naturally occurring isotopes of the element) andthe radioactive isotope. The specific activity of theradioisotope is defined by the counting rate pertotal mass of the element. In case of ultratraceelement studies, it is necessary to know the

amount of carrier added. Irradiation of the elementenriched in the parent isotope of the radioactivedaughter increases its specific activity, albeit atan additional cost. This was done, e.g., by Parentet al. [6], for the production of 191Pt. The irradiatedPt was enriched in 190Pt up to 4.19 %, which meansabout 400 times higher than the natural isotopicabundance (θ = 0.01 %). It is also possible tomake noncarrier added radioisotopes with fastneutrons via (n,p) reaction, e.g., 64Zn(n,p)64Cu [7].On the assumption that the zinc target is free fromcopper impurities, and that no stable copper atomsare added during the post-irradiation radiochemicalseparation originating from copper impurities inthe reagents and recipients, this is an efficient wayto produce a carrier-free copper isotope.

An interesting alternative to produce an isotopewith a high specific activity is to irradiate acompound where the element is bound to acarbon atom. Due to recoil, the radioactive elementis knocked off and set free (similar to thearsenic example, given previously). To continuethe example of copper, via recoil of 64Cu fromorganocopper compounds high specific activitycopper radiotracer may be produced.

The other interesting way to produce aradioisotope is through particle activation, followedby a radiochemical separation of the radioisotope

1

3 4 5 6 7 8 9 10

13 14 15 16 17 18

11 12

IIIb IVb Vb VIb VIIb

IIIa IVa Va VIa VIIa 0

VII Ib IIb

Sc Ti

Y Zr

La

Ce

Th

Pr

Pa

Nd

U

Pm

Np

Sm

Pu

Eu

Am

Gd Tb Dy Ho Er Tm Yb Lu

Hf

V

Nb

Ta

Cr

Mo

W

Mn

Tc

Re

Fe

Ru

Os

Co

Rh

Ir

Ni

Pd

Pt

Cu

Ag

Au

Zn

Cd

Hg

Ga

In

T1

Ge

C

Sn

Pb

As

Sb

Bi

Se

Te

Po

Br

P S C1

I

Kr

Xe

Ia

H

Na

K

Rb

Cs

Ca

Sr

Ba

IIa

2

Lanthanides

Actinides

Figure 5.12.1. Elements with potentially useful radionuclides.


from the matrix and from other interfering iso-topes. An example is the production of 48V throughthe nuclear reaction 48Ti(p,n) 48V, and the sepa-ration of the 48V from the Ti matrix and inter-fering Sc isotopes. The advantage here is thatthe radioisotope is in principle carrier free, whichmeans that all V atoms present are radioactive [8].This assumes, however, that there was no V impu-rity either in the Ti target, or in the reagents used todo the radiochemical separation of the V. This willnever be true. Nevertheless the radioisotope maybe claimed to be carrier free, because the mass ofV attributable to the nonradioactive contaminantmay be orders of magnitude below or similar tothe mass of the radiotracer.

Although it is feasible to do tracer experi-ments on a mixture of radionuclides with suitablenuclear characteristics, only one radioisotope isnormally used, so that the isolation and countingof the species can be straightforward. Sometimesit may be very tempting to use two radioisotopesof the same element to follow the behaviour oftwo important species, e.g., two oxidation states.At the start of the experiment, e.g., an aquaticmedium, each oxidation state of the element islabelled with its specific isotope. Depending on thechemical environment of the medium, the dura-tion since the start of the experiment, the tem-perature, . . ., the oxidation state will be stable,reduced or oxidized. This can be unravelled with aspecific radiotracer per oxidation state. It requiresradioisotopes with suitable nuclear characteristics:comparable half-life and nonoverlapping photo-peaks. This situation, however, is seldom met.The element As is the exception, with a choiceof three radioisotopes: 74As (t1/2 = 17.78 days, γ :634.8 keV), 76As (t1/2 = 26.4 h, γ : 559.1 keV) and77As (t1/2 = 38.8 h, γ : 239.0 keV). The individ-ual isotope can then be converted into a specificspecies, e.g., oxidation state (arsenate, arsenite, andin living organisms also into methylated species(monomethylarsenic acid or dimethylarsonic acid).

A dual radiotracer technique, with an emphasisof probing artefacts, was used in a case study oftechnetium and spinach [9]. The authors wantedto study the uptake of Tc by the plant, butwere also on the alert for possible errors during

their homogenisation procedure (possible lossesand redox conversions). For the uptake by theplant they used 99TcVIIO4

− (spike in the substrate(99Tc: t1/2 = 2.13 × 105 years, β−: 294 keV)) thatwas transformed into 99TcX by the plant (plant-formed species). For testing the homogenisa-tion procedure they added another isotope, either95mTcO4

− (95mTc: t1/2 = 61 days, γ : 204.1, 582.1,835.1 keV) or 99mTcO−

4 tracers (99mTc: t1/2 =6.02 h, γ : 140.6 keV). The tracing of the differ-ent isotopes allowed them to correct for procedurallosses and redox conversions. Similar radioisotopematches can be obtained for other elements, e.g.,Mn, Co, Se, Sn, Sb and I.

In some cases researchers use a ‘substitute’tracer, i.e., a radiolabel foreign to the compound,such as 125I-labelled (125I : t1/2 = 59.4 days, decaymode: electron capture) or a 11C-labelled (11C :t1/2 = 20.3 min, positron emitter) methyl group ona protein.

A small group of elements have naturalradioisotopes. The most important ones are 3H,14C, 40K, 210Pb, 226Ra, 232Th, 235U and 238U, all ofwhich have very long half-lives, requiring speciallow radioactivity measurement facilities.

4 RADIONUCLIDEMEASUREMENTS [5]

As a rule radioactivity measurements will have tobe normalised to equal counting times, correctedfor decay during counting and decay against thecounting of the reference.

The mathematical equation to correct for decayduring counting (tcounting time), for decay versustime zero (twaiting time) and to normalise per timeunit is the following:

Normalised radioactivity

= measured radioactivity

× λ exp(+λtwaiting time)

1 − exp(−λtcounting time)

with λ being the decay constant (λ = 0.693/t1/2),exp (+λtwaiting time) the correction for radioactivedecay between the arbitrary time zero and the start

RADIONUCLIDE MEASUREMENTS 489

of the measurement, and 1 − exp (−λtcounting time)

the correction for decay during the measurement.The time unit of the normalised radioactivity willbe that of the decay constant, e.g., when the half-life is expressed in hours, then this equation givesthe normalised radioactivity per hour.

Besides the measurement of the radioactivesample, it is necessary to measure the radioac-tivity due to the background radioactivity of thesurroundings and the electronic noise of the equip-ment. The net radioactivity is that measured for thesample minus that of the background. This back-ground activity is always present and due, amongothers, to naturally occurring 40K in concrete andother building materials, to remnants of fissionproducts from fall-out of atomic bombs and/ornuclear disasters, and to electronic noise.

The counting time will be adjusted to theradioactivity of the sample (A∗) and to its ratioversus the background radioactivity (B). Theminimum is a counting rate equal to that of thebackground measured during the same time, butif feasible a counting rate ten times that of thebackground is advisable. The precision of thecounting rate depends on the counting statisticsof the sample and on that of the backgroundmeasurement. It is calculated in the following way:

The radioactivity A∗ has an error due tocounting statistics equal to

√A∗ (square root of

A∗), i.e., with a confidence level of 1 σ (65 %)and twice this value for 2 σ . The net radioactivityhas a counting error equal to

√(A∗ + B) (square

root of the sum of the activity measurements A∗and B) with a confidence level of 1 σ (65 %) andtwice this value for 2 σ . Relative counting errorsexceeding 30 % are the upper limit.

This section will consider detection of γ

emitters, and that of positron and electron emitters.

4.1 Gamma-detection

4.1.1 NaI(Tl) scintillation detector

The most important detector for γ counting, offer-ing a high sensitivity, is the thallium activated NaI

crystal scintillation detector. The detection is basedon the fluorescence produced in the NaI (Tl acti-vated) monocrystal, coupled to a photomultiplier,a power supply and an amplifier–analyser sys-tem. When the ionising radiation passes throughthe scintillator it produces photons in the UV–VISrange. The number of photons produced is propor-tional to the energy absorbed in the scintillator.Light falling on the photomultiplier is convertedinto a number of electrons that is proportional tothe number of photons falling on the photocath-ode. The final electrical pulse is amplified andanalysed. It is proportional to the energy of thedetected γ -ray.

The resolution of the NaI(Tl) detector is about7 % (full width at half-maximum (FWHM) of the662 keV photopeak of 137Cs. This means that thisγ -peak is completely resolved from γ -rays withenergies below 562 keV and above 762 keV, atequal intensity of the γ -peaks.

4.1.2 Semiconductor detectors

In a semiconductor the atoms are arranged closelytogether in a crystal lattice. At absolute temper-ature (0 K) the electrons fill up completely thelowest energy levels, called the valence band. Atany other higher temperature there will be somethermal excitation of electrons from the valenceband to the conduction band, leaving some emptyplaces or ‘holes’, carrying a positive charge. Theelectron in the conduction band is free to move andif an electrical field is applied, it moves towards thepositive electrode. When radiation is absorbed inthe crystal electron–hole pairs are created and thecollection of these charge carriers gives rise to anoutput signal proportional to the amount of energyabsorbed. The detection efficiency of a semicon-ductor can be equal to or an order of magnitudelower than that of an NaI(Tl) scintillation detector,depending on the size of the crystal. The reso-lution, however, is superior. The FWHM of the1.33 Mev photopeak of 60Co is 2 keV or below.This means that γ -energies of 4 keV below or4 keV above this energy are completely resolved.The most common semiconductor detectors aremade of high purity germanium.


4.2 Beta detection

4.2.1 Liquid Scintillation Detector

In case of β−-emitters a special type of detectoris needed, because the shielding of, e.g., asolid scintillator, prevents most of the radiationfrom penetrating. Therefore the radioactivity isintimately mixed with the scintillator, to which theenergy of excitation is transferred. The scintillationvial is ‘viewed’ by one or two photomultipliers,power supplies and an amplifier–analyser system.The efficiency of detection is very high, andcan reach 100 % for very low energy β-emitters,e.g., tritium. This type of measurement, however,is very prone to quenching of the fluorescencedepending on the composition of the aqueousphase. There exist, however, adequate methods tocorrect for this drawback.

4.3 Autoradiography

This method is suitable for the detection of β-particles and γ -rays in flat samples, e.g., in a gelused for electrophoresis or a section of organictissue cut with a microtome in cytology.

There exist two different techniques: the classi-cal autoradiography, using X-ray films and sincethe late seventies autoradiography using phos-phor technology.

4.3.1 X-ray film autoradiography

Provided that the X-ray film makes close contactwith the flat sample, autoradiography makes excel-lent spatial resolution. The exposure of the filmmay take many weeks. Silver halide crystals in theemulsion respond directly to the β-particles andγ -rays emitted from the sample. Each emissionconverts several silver ions from a particular silverhalide crystal to silver atoms to produce a stablelatent image. When the film is subsequently devel-oped these few silver atoms catalyse the reductionof the entire silver halide crystal (grain) to metal-lic silver to produce an autoradiography image

of the radioisotope distribution. Under carefullycontrolled conditions, a densitometer can be usedto quantify image formation. The dynamic rangeof the film is limited to about two orders of mag-nitude, and is complicated by the characteristicsigmoidal density versus log exposure responsecurve [10].

In order to enhance the effect an intensifyingscreen can be placed at the opposite side of thefilm. Use of, e.g., a hyperfilm, i.e., a plasticbase with an emulsion on both sides, sandwichedbetween the sample and the intensifying screen,exposes the film not only to the direct radiation butalso to the fluorescence emitted by the intensifyingscreen when hit by remaining radiation aftercrossing the film. Atoms of the intensifyingscreen are excited through this radiation, which isfollowed by fluorescence and conversion of silverions to silver atoms.

4.3.2 Phosphor imaging [10]

Phosphor imaging is a newer (late seventies)and faster method, replacing the traditional X-rayfilms (and eliminating developing chemicals) byreusable storage phosphor screens that are 10 to250 times more sensitive than film. The sample isplaced in close contact with a phosphor screen forhours up to weeks, depending on the nature (β−-,γ -rays), the half-life and the number of radioactiveatoms. The atoms of the screen that receive theradiation of the radioisotopes are excited. At theend, the screen is scanned by a laser beam. Asthe screen is scanned luminescence is collectedfrom the excited areas by a fibre optic bundle,channelled to a photomultiplier and convertedto an electrical signal. The amount of light isproportional to the amount of radioactivity in thesample in a wide linear dynamic range, coveringfive orders of magnitude and a sensitivity 10 to 250times that of X-ray film. This phenomenon can bedigitised and results in a two-dimensional pictureof the location of the radioactivity. In the case ofgelelectrophoresis this picture is then combinedwith that of the proteins, made visible throughstaining (e.g., silver staining, or Coomassie blue).

APPLICATION OF RADIOTRACERS TO SPECIATION PROBLEMS 491

The results give a quantitative estimate of thepresence of the trace element species.

5 GENERAL SOURCES OF ERRORIN RADIOACTIVITY MEASUREMENT

The proper function of the detector must beunder adequate control to maintain stability anddetect deviations. Although the dynamic range ofthe detection is very large, there is a practicallimitation due to deviation from normality at highcounting rates. A change in geometry (putting thesample at a greater distance from the detector) isoften sufficient to eliminate errors of this type.

A major, unexpected source of error in usingradiotracers can be the presence of a radioactiveimpurity. This has once been the situation for a51Cr (t1/2 = 27.7 days, γ : 320 keV) source pro-vided by a commercial supplier. At the time ofdelivery only the 320 keV γ peak of 51Cr couldbe measured on the intrinsic Ge detector, but afterdecay of about 280 days the radioactivity of thissource was reduced 1000-fold and became equalto that of the 57Co contaminant (t1/2 = 271.8 days,main γ 121.6 keV), which was originally maskedby the Compton radiation caused by the 51Cr320 keV γ ray. A regular check with a high res-olution Ge detector of the radiotracer in differentfractions precludes such errors.

6 APPLICATION OF RADIOTRACERSTO SOLVE SPECIFIC CHEMICALSPECIATION PROBLEMS

As said previously the procedures to separate thespecies are the same as those in use with mostof the other analyte detection systems describedin this book, but the compounds are measuredthrough the radioactivity emitted by the labelledspecies, on the assumption that they underwentcomplete isotopic exchange with the nonradioac-tive share of the molecules. In contrast to mostof the other systems, the detection of the radioac-tivity occurs mostly off line, which means thetechnique is nonhyphenated. The separation tech-niques that are liable to profit most from the ease of

radioactivity detection are liquid chromatography,liquid–liquid extraction, ultrafiltration and gelelectrophoresis. Applications in method develop-ment are documented in the life and environ-mental sciences. A minor share of applicationsconcerns species that carry natural radioactivity,or radioactivity due to nuclear fission experimentsor nuclear disasters.

6.1 Life sciences

Radiotracer examination of a particular step ofan analytical procedure is a most useful toolin method validation. Hereafter follow a coupleof examples.

A very delicate step in the measurementof arsenic by hydride generation-atomic absorp-tion spectrometry (HG-AAS) is the reproducibleand quantitative on-line transformation of thearsenic compounds from the aqueous aliquotsinto hydrides by NaBH4 in the reaction coil fol-lowed by removal of the generated hydrides in agas–liquid separator. In the coil the reduction ofthe traces of arsenate, arsenite, monomethylarsonicacid (MMA) and dimethylarsinic acid (DMA)can be adequately led to completion. In order toevaluate two different gas–liquid separators usedin arsenic speciation by HG-AAS, van Elterenet al. [11] applied an 74As(III) spike to establishthe yield of hydride generation with respect to bothgas–liquid separators and concluded that only oneof the two types was satisfactory. The hydride pro-cedure prior to gas–liquid separation was essen-tially quantitative for all species investigated, butonly the ‘classical separator’ in which gas and liq-uid are separated by gravity was reliable to strip thearsine completely from the solution, independentof back-pressure. The second type of gas separatorwhere gas diffuses through a permeable tube didnot allow the complete stripping from the arsineand back-pressure worsened things.

An area that is prone to many artefacts is that ofthe metalloproteins in biological fluids and tissues.The easiest case is without any doubt that of acovalently bound metal. Nevertheless, the tediousoptimisation study of the separation parameters,


retention behaviour and recovery of metalloproteincomplexes in serum by SEC was greatly facili-tated by using metalloprotein complexes labelledwith radiotracers (59Fe: t1/2 = 44.51 days, γ : 1099and 1291 keV and 65Zn: t1/2 = 244.3 days, γ :1115.5 keV) [12].

Another example concerns the separation of thevery labile, noncovalently bound protein–vana-dium complexes in human serum. De Cremeret al. [8] went through a long process of find-ing a suitable column and elution media for48V-labelled proteins (48V t1/2 = 15.98 days, γ :984 keV) that guaranteed the preservation of theoriginal species. Ultrafiltration was a most use-ful tool for preliminary investigations about thestability of the complexes in the various mediatested, because it allowed in a simple procedureto find out if the radioactivity remained with theproteins, which are bigger than the cut-off of thefilter, or dissociated from the protein in that par-ticular medium and ended up in the filtrate [13].Similarly during the chromatographic experiments,it was easy to find out if the vanadium elutedwith the expected mass of the protein in caseof SEC, or as free vanadate. It could be shownthat affinity chromatography was not applicablebecause the medium immediately destroyed thevanadium–protein binding.

One step beyond the development and opti-misation of analytical methods with the aid of aradiotracer, is its use for in vitro and in vivo stud-ies, to determine the distribution of added tracerto a living system or to a particular compound.Besides testing the procedures on ‘real samples’it also provides basic knowledge on the behaviourof the species in a living system, which paves theway to subsequent ‘cold’ studies. The terminology‘cold’ and ‘warm’ designates respectively ‘stable’and ‘radioactive’ isotopes.

In vitro experiment consists in adding a radio-tracer, e.g., 51Cr radiotracer as CrCl3 in physi-ological medium to serum, leave it to incubatefor, e.g., 1 day at 37 ◦C. In vivo labelling is then,e.g., the intraperitoneal injection of 51Cr radio-tracer as CrCl3 in physiological medium intoa rat, wait an appropriate time to allow thecompound to distribute over the organism, and

measure the absorbed radiolabel in different bodycompartments (blood, soft and hard tissues). Inboth cases it is important to study under whichform the 51Cr is present. Of particular interestwill be the fraction of 51Cr bound to proteins.Ultrafiltration, gel filtration and ion exchange chro-matography are the relevant tools to use [14]. Theradiolabelling allowed to study the kinetics ofnewly added Cr to be investigated. Such an experi-ment with serum reveals how in case of Cr the 51Crbinds mainly to transferrin and to a lesser extentto albumin [15].

A warning is needed. The application of a com-pound, whether in vivo or in vitro, whether radioac-tive or not, to study the distribution of added tracerto a living system or a particular substance canentail many artefacts. During the initial step ofincubation, the tracer should be enabled to becomepart of the system or medium so as to reachequilibrium. Care should be taken to administerdoses reflecting physiological levels of the ele-ment, so that fortuitous linkages to those compo-nents with many binding sites are avoided. When-ever feasible, carrier-free radioisotopes should beadministered.

Radiotracer labelling has been widely used tostudy the behaviour of many essential and toxicelements in body fluids and tissues. Here followsa number of examples to illustrate the potential ofradiotracer method.

Recently a study was conducted to determinethe behaviour of 114mInCl3 intraperitoneallyinjected in rats (114mIn : t1/2 = 49.5 days, γ :192 keV), how it distributed over the body, andwhat kind of protein bound In species couldbe discerned in the different body fluids andtissues [16]. It may be interesting to follow thecomplete scheme, starting with the in vivo organdistribution. The results are given in Table 5.12.1,listed as percentage of the radioactivity pergram organ (wet weight) and percentage perorgan (wet weight). Large amounts of indiumare stored in kidney, liver and spleen. This istypical for ionic indium. Other studies applyingcolloidal indium oxide, reported that indiumaccumulates preferentially in liver, spleen and thereticuloendothelial system [17].


Table 5.12.1. In vivo organ distribution of 114mIn in ratsexpressed as per cent per gram organ of total activityin all measured organs and as per cent per organ oftotal activity in all measured organs [16]. Reproducedby permission of the Royal Society of Chemistry.

Organ % per g organ % per organ

Liver 16.5 ± 4.5 69.2 ± 7.3Kidney 17.1 ± 2.8 10.8 ± 2.8Heart 2.7 ± 0.5 0.7 ± 0.3Testes 2.8 ± 0.6 2.3 ± 0.6Thyroid glands 3.7 ± 0.5 0.5 ± 0.1Lung 4.4 ± 1.1 2.0 ± 0.9Spleen 21.8 ± 2.1 4.2 ± 0.6Stomach 6.6 ± 2.1 3.0 ± 1.1Bladder 13.4 ± 5.6 0.9 ± 0.3Small intestine 6.1 ± 2.9 6.4 ± 3.1Bone 4.9 ± 1.0 a

aTotal bone mass not measured [16].

Table 5.12.2. In vivo intracellular distribution of 114mIn inkidney, liver and spleen. A, pellet of nucleus; B, pelletof mitochondria; C, pellet of lysosomes and peroxisomes;D, pellet of ribosomes and small vesicles; E, homogenatecontaining cytoplasma. Results denoted as percentage of totalactivity within each tissue homogenate [16]. Reproduced bypermission of the Royal Society of Chemistry.

Kidney Liver Spleen

% in A 9.7 ± 4.1 6.3 ± 0.3 3.4 ± 1.4% in B 22.5 ± 1.0 14.1 ± 2.2 18.7 ± 2.9% in C 5.7 ± 2.6 1.7 ± 0.5 7.7 ± 1.6% in D 15.3 ± 1.0 4.6 ± 0.4 11.8 ± 1.5% in E 46.8 ± 3.6 73.3 ± 3.3 58.4 ± 5.3

One step further, the intracellular distribution ofthe trace element in tissues of animals was inves-tigated after in vivo treatment with a radiolabel.The example with indium continues. Van Hulleet al. [16] studied the partition of nontoxic tracedoses of carrier free 114mIn as InCl3 in the Wistar rat.Table 5.12.2 shows the results for the in vivo intra-cellular distribution of 114mIn in kidney, liver andspleen by differential centrifugation. In all cases thecytosolic fraction accounts for the highest activityof 114mIn, followed by the mitochondrial fraction.Some reflections have to be made about the outcomeof such intracellular distribution studies. Althoughthe fractions are described as nucleus, mitochon-dria, lysosomes,. . ., none of these fractions containssolely one specific organelle. The presence of otherorganelles can be determined by the analysis of theirspecific enzymes, the so-called marker enzymes.

The distribution of the organelles has to be carriedout separately on a small aliquot of each fraction,the remaining part being used for trace element anal-ysis. At the end, a matrix of x equations with thex unknown concentrations of the trace element ineach organelle has to be solved. When this workis undertaken using nonradiolabelled tissue, signif-icant, even unsurmountable, problems due to thepossible contamination of the cell fractions with thetrace element under study by the many reagents andrecipients can be anticipated. As the working con-clusions may be approximately the same as thosedrawn from the simplified model of supposedly pureorganelle fractions, no effort was made to work inthe strict way. Working with in vivo labelled tis-sue, where a complete isotopic exchange with thenatural isotopes has been achieved, circumvents theproblem related to contamination hazards. The onlyinstance where it may be feasible to work with anonlabelled compound, is when it does not occur innature and cannot be formed in the organism. Forinstance the study of the distribution of cisplatinis feasible in the cold way. To add to the degreeof difficulty of this work, little is known about thepreservation of the original trace metal – organellebinding throughout the tedious intracellular frac-tionation, a problem common to ‘cold’ and ‘warm’procedures. It is not surprising that apparently veryfew attempts have been made to study speciation of‘cold’ trace elements in this way.

When studying blood, the first aspect to con-sider is the partition of the species between serumand packed cells. In the case of indium themajor share of the 114mIn activity in the bloodis located in serum, 90.2 ± 4.1 % of total indiumcontent of blood. Next, serum is subjected todifferent types of chromatographic separations.Figure 5.12.2 shows the elution profile of 114mInin rat serum separated by SEC. There is onlyone peak of indium and it coincides with the elu-tion of transferrin. Figure 5.12.3 shows the chro-matogram of the in vivo speciation of 114mInin blood lysate. The In elutes slightly ahead ofhaemoglobin, so further investigations are neededto determine whether or not haemoglobin is themain carrier of indium. The separation of rat urine


absorbance 280 nmactivity 114 m-In

1 2 3 4 5 6 7 8 9 10 11 12

elution volume/ml

abso

rban

ce 2

80 n

m/m

AU

activ

ity 1

14 m

-In/

coun

ts

13 14 15 16 17 18 19 20 21 22 23 24

50000

100000

0

150000

200000

250000

300000

0

250

500

750

1000

1250

1500

1750

2000

0

Figure 5.12.2. Elution profile of [114mIn]InCl3/rat serum separated with size exclusion chromatography on Asahipak 520-GS 7G.Buffer: 10 mM hepes +5 mM NaHCO3 + 0.15 M NaCl, pH 7.4; UV absorption at 280 nm [16] (reproduced by permission ofThe Royal Society of Chemistry).

600

01 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 300

1200

1800

2400

3000

3600

4200

4800

elution volume/ml

abso

rban

ce 2

80 n

m/m

AU

1000

0

2000

3000

4000

5000

6000

7000

8000

activ

ity 1

14m

-In/

coun

ts


Figure 5.12.3. Elution profile of [114mIn]InCl3/rat blood cell lysate separated with size exclusion chromatography on Superose 12HR. Buffer: 15 mM hepes + 0.15 M NaCl, pH 7.2; UV absorption at 280 nm [16] (reproduced by permission of The Royal Societyof Chemistry).


01 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 390

200

400

600

800

1000

1200

1400

1600

1800

elution volume/ml

abso

rban

ce 2

80 n

m/m

AU

activ

ity 1

14 m

-In/

coun

ts

0

1000

2000

3000

4000

5000

6000


Figure 5.12.4. Elution profile of [114mIn]InCl3/rat urine separated with size exclusion chromatography on Superdex Peptide.Buffer: 15 mM hepes +0.15 M NaCl, pH 6.5; UV absorption at 280 nm [16] (reproduced by permission of The Royal Societyof Chemistry).

by SEC is shown in Figure 5.12.4. In urine, indiumis mostly bound to the low molecular mass frac-tion, with a maximum corresponding to a molec-ular mass of 300–4000 Da. The results of livercytosol by SEC are shown in Figure 5.12.5. Itappears that indium preferentially or exclusivelybinds to the high molecular mass (HMM) frac-tion. The large amount of 114mIn might be partlydue to the presence of serum transferrin in theorgan after homogenisation. As the elution peakis much broader than expected, there is ground tothink that there is one or more other HMM compo-nents present in the fractions. Only in the case ofkidney cytosol (figure not shown), does a consider-able amount of indium appear in the low molecularweight (LMM) fraction. Further research consistsin identifying the HMM and LMM compounds car-rying the element In.

Very extensive research on the in vivo behaviourof radiotracer Se in Wistar rats has been done byBehne and coworkers [18] over the last 15 yearsusing 75Se tracer with very high specific activity

allowing the determination of the species in thepmol to fmol range. The wide array of bio-chemical separation techniques, combined with γ -scintillation spectrometry and also autoradiographyin the case of gel electrophoresis allowed them toreveal most interesting findings about the diverseSe species in the organism, as well as about theirtransformation in the living system. They con-cluded that all of the selenium present in the mam-malian organism is protein bound, and thereforespeciation is mostly concerned with the determina-tion of the different selenium-containing proteins.A method was developed for the determinationof selenocysteine and selenomethionine in theselenium-containing proteins. The identification ofspecific selenoproteins was achieved by analysisof their selenoamino acid residues and by stud-ies on their characteristics and possible biologicalfunctions. This is being followed by the develop-ment of methods for the quantitative analysis ofthe selenoproteins in question in the tissues of ani-mals and man.


01 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 320

100

200

300

400

500

600

700

800

elution volume/ml

abso

rban

ce 2

80 n

m/m

AU

activ

ity 1

14 m

-In/

coun

ts

0

5000

10000

15000

20000

25000

30000

35000

40000


Figure 5.12.5. Elution profile of [114mIn]InCl3/rat liver cytosol separated with size exclusion chromatography on Superose 12HR. Buffer: 15 mM hepes +0.15 M NaCl, pH 7.2; UV absorption at 280 nm [16] (reproduced by permission of The Royal Societyof Chemistry).

Method development for the identification ofthe different chemical forms of Se proteins in75Se-enriched yeast was successfully done byChery et al. using 2D electrophoresis, followed byphosphor imaging [19] (see Chapter 4.4).

Another study of which the successful out-come is solely due to the use of radiotracer, con-cerns the basal metabolism of intraperitoneallyinjected carrier-free 74As-labelled arsenate in rab-bits. [20]. It was very interesting to follow theprocess of the reduction of arsenate to arsenite,its binding to transferrin in serum, to haemoglobinin the red blood cells, and its transformation intomonomethylarsonic acid, and next to dimethy-larsinic acid, and the binding of these differentcomponents to tissue proteins. In healthy rabbitsand also in man the addition of first one and thena second methyl group to the arsenic is a natu-ral detoxification process. This process is, how-ever, seriously inhibited when either toxic doses ofinorganic arsenic (arsenate or arsenite) are admin-istered, or in case of uraemia.

Therefore the next step consisted in studying theeffect of toxic doses of arsenate and observe manyvariations in the metabolic pattern of inorganicarsenic [21]. In a further scenario the arsenatemetabolism was studied in uraemic rabbits, show-ing the dramatic inhibiting effect of this pathologyon arsenate metabolism, including the inhibitoryeffect on the detoxifying methylation mechanismtransforming inorganic arsenic species into far lesstoxic methylated compounds [22].

6.2 Environmental sciences

In environmental sciences there is the search forthe chemical species and where this appears to beimpossible, it becomes the science of fractionationof elements, which means classification of an ana-lyte or a group of analytes from a certain sampleaccording to physical (e.g., size or solubility) orchemical (e.g., bonding, reactivity) properties [23].

The investigation of the different chemicalforms of elements in natural waters, sediments


and sludges is a matter of great difficulty. Thespecies belong to chemically very different cate-gories [24] and include different oxidation states(e.g., Cr(III) and Cr(VI)), inorganic compounds andtheir complexes (hydroxicomplexes, e.g., Fe(OH)3,Al13O4(OH)24

7+, complexes with SO42−, Br−, F−,

CO32−, organic complexes, macromolecular com-

pounds and complexes (e.g., the element beingbound to humates, clays). To mimic the behaviourof elemental species as they occur in nature is notevident. It is difficult to create a laboratory scalesystem as complex as prevails in nature. Thereforethe experimental set-up is usually simplified. Theradiotracer must be of the highest specific activityand it has to be assumed that the tracer added in theionic form will exchange with the natural system. Inpractice it seems that at least some exchange takesplace, but that it is a slow process with a half-timeof hours, up to days and weeks.

A radiolabel proved to be very useful for inves-tigating the stability of Cr(III) and Cr(VI) speciesin water in a feasibility study to produce a ref-erence material for intercomparison of specia-tion methodology. Unfortunately, only one usefulradioisotope of Cr exists: 51Cr (t1/2 = 27.7 days,γ : 320 keV). The absence of a specific isotopefor each oxidation state is a problem common toany other conventional analytical method, wherethe measurement will detect any form of Crpresent. The changes that might be anticipatedin such a solution are the conversion of Cr(VI)into Cr(III) and the hydrolysis of Cr(III). Dyget al. [25] studied the effect of different param-eters on the stability of Cr(III) and Cr(VI) inwater. These included a study of the stabilityas a function of time, possible losses caused byadsorption, temperature dependence and choice ofthe material of the container. The method usedto differentiate between Cr(III) and Cr(VI) wasbased on the cationic behaviour of the Cr(III) andthe anionic behaviour of Cr(VI) as chromate. Inthis work the Cr(III) and Cr(VI) were separatedthrough extraction of Cr(VI) with Amberlite LA-1 or LA-2 diluted with isobutylmethylketone. Thebehaviour of the radiolabelled Cr(III) and Cr(VI)were checked in separate runs, with detection ofthe 51Cr label.

Radiotracers in controlled laboratory experi-ments are a very tempting tool to study thebehaviour of metal ions and other contaminants innatural waters, because they allow researchers inan uncomplicated manner to learn about the role ofindividual parameters on kinetics and sorption toparticulate material. This was adopted by McCub-bin and Leonard [26] for the element Th with theaid of 234Th (t1/2 = 24.1 days, β−: 198 keV, weakγ ’s) but the outcome of their experiments was onlymoderate as they concluded that caution shouldbe exercised in extrapolating information obtainedfrom investigations using radiotracers to predicttrace element behaviour in natural waters.

Another interesting example deals with cop-per in the environment. The group of van Eltereninvestigated a chromatographic technique to inves-tigate the lability of copper complexes understeady-state conditions using high specific activ-ity 64Cu (t1/2 = 12.7 h, γ : 511 keV) [27]. Alongthe same lines they examined the usefulness ofsolid-phase extraction (SPE) cartridges for cop-per speciation screening [28]. In the latter studyfive SPE cartridges were used to extract cop-per species from a sample. Each cartridge isexpected to extract a specific (group of) copperspecies, depending on the chemical properties ofthe extracting phase. In theory this would leadto the extraction of free copper ions (Cu2+) andlabile inorganic copper complexes (like CuCl+) bychelex cartridges, cationic species by SCX car-tridges, anionic species by SAX cartridges andhydrophobic species by C18 and RP cartridges.However, in practice it may be possible thatunwanted secondary interactions take place, mak-ing the cartridge less selective than desired. Theseartefacts were visualized by radiotracer studies.The composition of the test samples and calcu-lated speciation with regard to Cu species at pH 6.0are given in Table 5.12.3. The results are givenin Figures 5.12.6–5.12.8, proving that secondaryinteractions must play a role in the extraction ofcopper species. Both Chelex and SCX cartridgesare expected to retain free copper ions, whilethey should let negatively charged Cu–EDTA andCu–TACTDD pass through. Indeed, it is found thatfree copper is almost completely retained on both


Table 5.12.3. Composition of test samples and calculatedspeciation with regards to Cu species at pH 6.0 [28].

Sample Sample composition calculatedspeciation

Test 1 1 × 10−6 mol L−1 Cu 76 % Cu2+21 % CuCl+2 % Cu(OH)2(aq)1 % CuCl2

Test 2 1 × 10−6 mol L−1 Cu 25 % Cu2+7.5 × 10−7 mol L−1 EDTA 75 % Cu–EDTA2−

Test 3 1 × 10−6 mol L−1 Cu 25 % Cu2+ orinorganic Cu

7.5 × 10−7 mol L−1 TACTDD 75 % Cu–TACTDD−

EDTA: ethylenediaminotetraacetic acid, disodium salt.TACTDD: 1,4,8,11-tetraazacyclotetradecan-5,7-dione.

cartridges as can be seen in Figure 5.12.6. How-ever, Figure 5.12.7 and 5.12.8 indicate that part ofCu–EDTA and Cu–TACTDD complexes is alsoretained on these columns. In the absence of sec-ondary reactions 25 % of the total copper should beretained, but the chelex cartridge retains more than50 % and the SCX cartridge even more: about 70 %in case of Cu–EDTA and almost all copper in case

of Cu–TACTDD. The SAX cartridge shows somesecondary interactions as well. It retains 63 % ofthe total copper when only Cu2+ is present, whileit should retain any cationic species. It seems toperform better with Cu–EDTA, as about 80 % ofthe total copper is retained when 75 % is expected.For Cu–TACTDD the performance is worse, only42 % being retained. The C18 and RP cartridgesshow opposite behaviour with respect to free cop-per ions. While the C18 cartridge retains only32 %, the RP cartridge retains more than 70 %of the total copper. The presence of Cu–EDTAreduces these values by a factor of two (17 %and 35 %, respectively), while with Cu–TACTDDsimilar responses are found, both cartridges retainabout 50 % of the total copper.

In particular, free copper ions seem to beextracted effectively by cartridges that shouldnot have any affinity for this species: SAX,C18 and RP. In general, these results makeclear that secondary interactions must play arole in the extraction of copper species. Withrespect to this the C18 cartridge performs best,

98.3

1.7

68.471.6

28.431.6

37.2

62.8

5.4

94.6

0

20

40

60

80

100

120

%

cartridge eluate

Chelex SCX SAX C18 RP

Figure 5.12.6. Extracted and eluted amounts of copper species (as percentage of total copper) of test sample 1, containing Cu2+and inorganic Cu species [28] (reproduced by permission of Akademiai Kiado, Budapest).


58.6

70.3

79.7

16.8

34.4

41.4

29.7

20.3

83.2

65.6

0

10

20

30

40

50

60

70

80

90

SCX SAX C18 RP

cartridge eluate

Chelex

Figure 5.12.7. Extracted and eluted amounts of copper species (as percentage of total copper) of test sample 2, containing 75 %Cu–EDTA [28] (reproduced by permission of Akademiai Kiado, Budapest).

55.8

96.2

41.8

50.845.744.2

3.8

58.2

49.254.3

0

20

40

60

80

100

120

Chelex SCX SAX C18 RP

%

cartridge eluate

Figure 5.12.8. Extracted and eluted amounts of copper species (as percentage of total copper) of test sample 3, containing 75 %Cu–TACTDD [28] (reproduced by permission of Akademiai Kiado., Budapest).


followed by SAX and RP. The anionic speciesCu–EDTA and Cu–TACTDD seem to undergosecondary interactions as well. They are notcompletely passed through Chelex, SCX, C18 andRP cartridges and Cu–TACTDD is not completelyretained on the SAX cartridge. It seems thatCu–EDTA behaves more like an anionic speciesthan Cu–TACTDD which has a more hydrophobiccharacter. This study clearly shows the advantagesof applying a highly specific activity radiotracerin speciation studies. The high specific activityof the radiotracer allowed the researchers to addonly a negligible amount of mass to a sample,which ensures that initial chemical equilibriaare maintained.

At the beginning of this section, fractionationwas mentioned. Part of such work consists inapplying sequential extraction steps. There exists aprocedure for soils, sediments and sludges devel-oped on a European scale in the frame of the ECStandards, Measurement and Testing programme(SM&T). These procedures do not provide infor-mation about chemical species, but fractionate themetals according to operationally defined selec-tive dissolution of geochemical phases. Gilmoreet al. [29] studied the readsorption and redistribu-tion of lead in the SM&T sequential extraction pro-cedure using a spike of 212Pb tracer (t1/2 = 10.64 h,γ : 238.6, 300.0 keV).

The possibilities offered by radiotracers tostudy dynamic speciation are really unique. Theyallow researchers to study not only the equilib-rium distribution of different species, but alsothe kinetics of their interconversion. This partic-ular aspect of speciation analysis is nicely illus-trated by the work of Achterberg et al. [30] onspecies kinetics and heterogeneous reactivity ofdissolved copper in natural organic-rich freshwa-ters under steady-state conditions, i.e., with mini-mum disturbance of existing equilibria, using highspecific activity 64Cu2+. Study sites with contrast-ing suspended particulate matter (SPM) character-istics were investigated. The analytical protocolallowed the differentiation between the followingCu species: SPM-associated Cu, dissolved reactive(free and labile) Cu, and organically complexedCu. The data obtained were successfully evaluated

by compartmental analysis, which showed theimportance of organically complexed Cu in fresh-waters, and the dominant role of the interac-tions between organically complexed Cu and SPMin SPM-rich water. The kinetic 64Cu measure-ments indicated that the attainment of equilibriumbetween dissolved reactive and organically com-plexed Cu took about <1–2 h, and 4–15 h forthe interaction between dissolved organically com-plexed and SPM-associated Cu. The kinetic studywas augmented by voltammetric measurements ofthe dissolved (stable) Cu equilibrium speciationconditions in the natural waters. These measure-ments showed that the waters contained very lowcupric ion concentrations (10 −12 –10 −15 M), withmore than 99.9 % of the dissolved Cu complexedby strong organic ligands (conditional stabilityconstants: 1013.4 –1015.4).

Last but not least there are the extensivestudies on the speciation of radionuclides in theenvironment [31]. The presence of trace amountsof fission radioisotopes such as 90Sr, 99Tc, 137Csand actinides in nature as a consequence offall-out of the fission products of 235U and239Pu used in nuclear weapon programmes, thestorage of nuclear waste and also from theChernobyl legacy (April 26, 1986) has been thesubject of many studies. The Chernobyl accidentcan be considered as the ultimate global scaleunauthorised radiotracer experiment of the pastcentury, giving workers in the nuclear field ampleopportunity to collect environmental samples tofollow the pathways of certain elements in thegeosphere. The relatively long half-lives (up tothousands of years) of many of the fall-outradionuclides are well suited for use in the studyof these long-term processes.

The ultimate fate and effects of these long-livedradionuclides in the ecosystem are mainly depen-dent on their chemical forms, which determinetheir partitioning and chemical transport in vari-ous locations. These parameters together with thephysical transport of water masses and associatedsuspended particles, play an important role in theavailability of the radiation to living matter. Manystudies cover this domain [32–37].


The assumption that the behaviour of theradionuclides can be inferred from their stableanalogues does not necessarily hold true, as hasbeen observed by Joshi [33] for the behaviourof Co, Cs and Pb in water from Lake Ontario.The concentrations of 60Co, 137Cs and 210Pb areseveral orders of magnitude lower than thoseof the stable isotopes or members of the samesubgroup of elements. Joshi also pointed out thatthe chemical forms of the stable analogues ofseveral radionuclides in the Great Lakes have notyet been investigated. For the actinides, true stableanalogues do not exist. This very important issueof the actinides in the environment will be dealtwith in great detail in a special chapter of the nextvolume of the handbook.

Interesting radionuclide and mobility studiesin Norwegian and Soviet soils were studied byOughton et al. [34] 3 years after the Chernobylaccident. A sequential extraction procedure hasbeen applied to study the speciation of Chernobyl-derived radionuclides (137Cs and 90Sr) in soils fromNorway, and from Byelorussia and the Chernobylregion in the Ukraine. They used six different

extraction media, starting with very mild (water),NH4OAc (exchangeable), to NaOAc (carbonates),to NH2OH.HCl (easy reducible Fe/Mn oxides),H2O2 (oxidizable organic matter), to nitric acid(acid digestible), the remaining part being definedas residue. Figure 5.12.9 shows the relative dis-tribution of 137Cs (t1/2 = 30.17 years, γ : 662 keV)and Figure 5.12.10 for 90Sr (t1/2 = 28.64 years,β− : 546 keV) in the sequential extraction fractionsfor four types of soil samples collected in Norway.Most of the 137Cs was strongly associated withsoil components (retained in the residue), whereas90Sr was more mobile, up to 70 % being found inthe easily extractable fractions. Amano et al. [36]have reported on the transfer capability of long-lived Chernobyl radionuclides from surface soil toriver water in dissolved forms.

Sanada et al. [37] have studied the accumula-tion of 137Cs,90Sr,239+240Pu, and 241Am in sedi-ment of the Pripyat River in the exclusion zoneof the Chernobyl nuclear plant and addressedthe dissolution of these radionuclides from thesediment 12 years after the accident. The peakarea of the concentrations of the radionuclides is

0

10

20

30

40

50

60

70

137-

Cs

(%)

Nor

way

198

9

soil 1

soil 2

soil 3

soil 4

H2O NH4OAc NaOAc NH2OH.HCl H2O2 HNO3 Residue

Fraction

Figure 5.12.9. Relative distribution of 137Cs in sequential extraction fractions. Samples (n = 4) were collected from Lierne,Norway, July 1989 [34] (reproduced by permission of The Royal Society of Chemistry).


10

20

30

40

50

60

70

NH4OAc NaOAc NH2OH.HCl H2O2 HNO3 Residue

90-S

r (%

) N

orw

ay 1

989

soil 1soil 2soil 3soil 4

0

H2O

Fraction

Figure 5.12.10. Relative distribution of 90Sr in sequential extraction fractions. Samples (n = 4) were collected from Lierne,Norway, July 1989 [34] (reproduced by permission of The Royal Society of Chemistry).

at 20–25 cm depth. The composition was com-pared with that of the radioparticles released dur-ing the accident. An analysis using a selectivesequential extraction technique was applied for theradionuclides in the sediments. It comprised acidsoluble, reducible, oxidisable and residual frac-tions. The results are given in Figure 5.12.11. 137Csand 239+240Pu were concentrated in the residualphase, suggesting an effective fixation of these par-ticles after the dissolution of the fuel particles toambient environmental matrices. The higher dis-tribution of 90Sr over the acid soluble, reducibleand oxidisable phases suggested a higher poten-tial mobility for this element compared with 137Csand 239+240Pu. These results suggest that the pos-sibility of release of 137Cs and 239+240Pu from thebottom sediment was low compared with 90Sr. Theauthors concluded that the potential dissolution andsubsequent transport of 90Sr from river bottomsediment should be taken into account with respectto the long-term radiological influence on theaquatic environment.

7 CONCLUSIONS

The previous examples are not exhaustive. Theyare meant to give an idea of how radio-tracers are most handy in method develop-ment for elemental speciation analysis and inresearch aiming at metabolic and environmentallyrelated studies.

Analytical chemists have been very resourcefulin developing separation and measurement systemsfor many element species. Every time a properanalytical method has been developed, it requiresvalidation to make sure that the data producedby different laboratories are comparable and trace-able [38]. Radiolabelled species are most useful toassist in controlling and validating separate stepsor the whole procedure. Unfortunately the useof radioactivity has fallen into disgrace with thegeneral public and the politicians, so that manyresearch centres have had to close their nuclearfacilities. This slows down and may even halt fur-ther use of radiotracers for method development

CONCLUSIONS 503

Fra

ctio

n (%

)F

ract

ion

(%)

Fra

ctio

n (%

)

0

50

10090Sr

0

50

100

acid soluble reducible oxidizable residue

0

50

100

S2 0−5 cm S2 6−7 cm S2 16−18 cm P1 28−30 cm

137Cs

239,240Pu

3.3Bq/g 4.2Bq/g 3.5Bq/g 68Bq/g

0.45Bq/g 0.33Bq/g 1.2Bq/g 9.6Bq/g

0.017Bq/g 0.0027Bq/g 0.013Bq/g 0.018Bq/g

Figure 5.12.11. Fractions of radionuclides over different phases segregated by sequential selective extraction applied to selectedsamples of bottom sediment of the Pripyat river near the Chernobyl Nuclear Power Plant. The value is total radionuclideconcentration [37] (Reprinted from Applied Radiation and Isotopes , Vol. 56, Sanada et al., pp. 751, 2002, with permission fromElsevier Science).

in elemental speciation analysis and for metabolicand environmentally related speciation studies.Pessimism, however, is not a good attitude for fur-ther development.

Overall, the integration of the use of radiotracersoffers an invaluable, complementary, analyticaltool in elemental speciation analysis in the life andenvironmental sciences.


8 REFERENCES

1. Cornelis, R., Analyst , 117, 583 (1992).2. Friedlander, G., Kennedy, J. W., Macias, E. S. and Miller,

J. M., Nuclear and Radiochemistry , 3rd edn, Wiley-Interscience, New York, 1981.

3. Slejkovec, Z., van Elteren, J. T., Byrne, A. R. and deGoeij, J. J. M., Anal. Chim. Acta, 380, 63 (1999).

4. Adams, F. and Dams, R., Applied Gamma-ray Spectrom-etry , Pergamon, Oxford, 1970.

5. Brune, D., Forkman, B. and Persson, B., Nuclear Analyt-ical Chemistry , Chartwell-Bratt, 1984.

6. Parent, M., Strijckmans, K., Cornelis, R., Dewaele, J.and Dams., R., Nucl. Instrum. Methods. B , 86, 355(1994).

7. van Elteren, J. T., Kroon, K. J., Woroniecka, U. D. andde Goeij, J. J. M., Appl. Radiat. Isotopes , 51, 15, 1999.

8. De Cremer, K., Cornelis, R., Strijckmans, K., Dams, R.,Lameire, N. and Vanholder, R., J. Chromatogr. B , 757,21 (2001).

9. Harms, A. V., van Elteren, J. T., Wolterbeek, H. Th. andde Goeij, J. J. M., Anal. Chim. Acta., 394, 271 (1999).

10. Johnston, R. F., Pickett, S. C. and Barker, D. L., Electro-phoresis 11, 355 (1990).

11. van Elteren, J. T., Das, H. A. and Bax, D., J. Radioana-lyt. Nucl. Chem., 174, 133 (1993).

12. Raab, A. and Bratter, P., J. Chromatogr. B , 707, 17(1998).

13. De Cremer, K., De Kimpe, J. and Cornelis, R., Fresenius,J. Anal. Chem., 363, 519 (1999).

14. Cornelis R, R., De Kimpe, J. and Zhang, X., Spectrochim.Acta, Part B , 53, 187 (1998).

15. Borguet, F., Cornelis, R., Delanghe, J., Lambert, M. C.and Lameire, N., Clin. Chim. Acta , 238, 71 (1995).

16. Van Hulle, M., De Cremer, K., Cornelis, R. and Lameire,N., J. Environ. Monit., 3, 86 (2001).

17. Castronovo, F. P. and Wagner, H. N., J. Nucl. Med., 14,677 (1973).

18. Behne, D., Hammel, C., Pfeifer, H., Rothlein, D., Gess-ner and H., Kyriakopoulos, A., Analyst , 123, 871 (1998).

19. Chery, C. C., Dumont, E., Cornelis, R. and Moens L.,Fresenius’ J. Anal. Chem., 371, 775 (2001).

20. De Kimpe, J., Cornelis, R., Mees, L. and Vanholder, R.,Fund. Appl. Toxicol., 34, 240 (1996).

21. De Kimpe, J., Cornelis, R., Wittevrongel, L. and Van-holder, R., J. Trace Elem. Bio. Med , 12, 193 (1998).

22. De Kimpe, J., Cornelis, R., Mees, L., Vanholder, R. andVerhoeven, G., J. Trace Elem. Bio. Med , 13, 7 (1999).

23. Templeton, D. M., Ariese, F., Cornelis, R., Danielsson,L.-G., Muntau, H., van Leeuwen, H. P. and Łobiski, R.,Pure Appl. Chem., 72, 1453 (2000).

24. Bowen, H. J. M., Page, E., Valente, I. and Wade, R. J., J.Radioanal. Chem., 48, 9 (1979).

25. Dyg, S., Cornelis, R., Griepink, B. and Quevauviller, Ph.,Anal. Chim. Acta, 286, 297 (1994).

26. McCubbin, D. and Leonard, K. S., Sci. Total Environ.,173, 259 (1995).

27. Van Doornmalen, J., van Elteren, J. T. and de Goeij,J. J. M., Analyt. Chem., 72, 3043 (2000).

28. Van Doornmalen, J., Kroon, K. J., van Elteren, J. T. andde Goeij, J. J. M., J. Radioanalyt. Nucl. Chem., 249, 349(2001).

29. Gilmore, E. A., Evans, G. J. and Ho, M. D., Anal. Chim.Acta , 439, 139 (2001).

30. Achterberg, E. P., van Elteren, J. T. and Kolar, Z. I.,Environ. Sci. Technol., 36, 914 (2002).

31. Salbu, B. and Steinnes, E., Analyst , 117, 243 (1992).32. von Gunten, H. R. and Benes, P., Radiochim. Acta, 69, 1

(1995).33. Joshi, S. R., Sci. Total Environ., 100, 61 (1991).34. Oughton, D. H., Salbu, B., Riise, G., Lien, H., Ostby, G.

and Noren, A., Analyst , 117, 481 (1992). Konoplev,A. V., Bulgakov, A. A., Popov, V. E. and Bobovnikova,T. I., Analyst 117, 1041 (1992).

35. Salbu, B., Krekling, T. and Oughton, D. H., Analyst 123,843 (1998).

36. Amano, H., Matsunaga, T., Nagao, S., Hanzawa, Y.,Watanabe, M., Ueno, T. and Onuma, Y., Org. Geochem.,30, 437 (1999).

37. Sanada, Y., Matsunaga, T., Yanase, N., Nagao, S.,Amano, H., Takada, H. and Tkachenko, Y., Appl. Radiat.Isotopes 56, 751 (2002).

38. Ebdon, L., Pitts, L., Cornelis, R., Crews, H., Donard,O. F. X. and Quevauviller, P., Trace Element Speciationfor Environment, Food and Health , Royal Society ofChemistry, Cambridge, 2001.

CHAPTER 6

Direct Speciation of Solids

6.1 Characterization of Individual Aerosol Particleswith Special Reference to Speciation Techniques

H. M. OrtnerDarmstadt University of Technology, Germany

Acronyms . . . . . . . . . . . . . . . . . . . . . . . . 5061 Introduction: The Significance of Single

Particle Characterization in Science andIndustry . . . . . . . . . . . . . . . . . . . . . . . . . . 507

2 Semiquantitative Particle Evaluation +Morphology Often = Speciation . . . . . . . . 5092.1 The key to a profound particle

characterization: a multimethodapproach . . . . . . . . . . . . . . . . . . . . . 5092.1.1 Particle collection with a

cascade impactor on glassycarbon disks . . . . . . . . . . . . . 513

2.1.2 Electrostatic particlecollection for nanometreparticles . . . . . . . . . . . . . . . . 515

2.1.3 Topochemical characterizationof solid aerosols by SEM andEPMA . . . . . . . . . . . . . . . . . 5152.1.3.1 Scanning electron

microscopy . . . . . . . 5152.1.3.2 Electron backscatter

diffraction, EBSD 5162.1.3.3 EPMA mapping . . . . 517

2.1.4 Quantitative bulkcharacterization on glassycarbon disks by TXRF formain, minor and traceelements . . . . . . . . . . . . . . . . 517

3 Explicit Methods of Particle Speciation . . . 5183.1 Valence band X-ray spectrometry by

EPMA-WDX . . . . . . . . . . . . . . . . . . 5183.2 TEM of particles smaller than 0.5 µm

in diameter . . . . . . . . . . . . . . . . . . . . 5183.2.1 Experimental procedure . . . . . 5183.2.2 Information content of TEM

investigations . . . . . . . . . . . . 5193.2.2.1 Energy filtering TEM

(EFTEM) or electronspectroscopicimaging (ESI) . . . . . 519

3.3 X-ray induced photoelectronspectrometry (XPS) . . . . . . . . . . . . . 519

3.4 Auger electron spectrometry (AES) 5203.5 Mossbauer spectrometry . . . . . . . . . . 5203.6 Micro-Raman spectrometry . . . . . . . . 5203.7 X-ray absorption techniques:

µ-EXAFS and µ-XANES . . . . . . . . . 5204 Trace and Isotopic Characterization of

Single Particles by Mass SpectrometricMethods and PIXE . . . . . . . . . . . . . . . . . . 5214.1 Secondary ion mass spectrometry

(SIMS) . . . . . . . . . . . . . . . . . . . . . . 5214.2 Laser micro(probe) mass spectrometry

(LAMMS) and laser ablation-inductively coupled plasma-massspectrometry (LA-ICP-MS) . . . . . . . . 521


506 CHARACTERIZATION OF INDIVIDUAL AEROSOL PARTICLES

4.3 Laser desorption/ionization-time-of-flight-mass spectrometry(LDI-TOF-MS) . . . . . . . . . . . . . . . . 521

4.4 Proton induced X-ray emission(spectrometry) (PIXE) . . . . . . . . . . . 522

4.5 Three-dimensional SIMScharacterization of large particles . . . 522

5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . 5226 Acknowledgements . . . . . . . . . . . . . . . . . . 5237 References . . . . . . . . . . . . . . . . . . . . . . . . 523

ACRONYMS

AAS Atomic absorptionspectrometry

AES Auger electron spectrometry(this is the commonacronym in topochemicalanalysis. Unfortunately it isidentical with the acronymfor atomic emissionspectrometry in bulkanalysis. In this paper,AES is used exclusivelyfor Auger electronspectrometry)

AFM Atomic force microscopyBSE Backscatter electronCPC Condensation particle counter3D Three dimensionalDMA Differential mobility analyserEBSD Electron backscatter diffractionEELS Electron energy loss

spectrometryEDX Energy-dispersive X-ray

(fluorescence spectrometry)EFTEM Energy filtering transmission

electron microscopyEPMA Electron probe microanalysisESEM Environmental scanning

electron microscopyESI Electron spectroscopic imagingEXAFS Extended X-ray absorption fine

structureFEG Field emission gunFWHM Full width at half-maximumGC Gas chromatographyGIF Gatan imaging filterHPLC High performance liquid

chromatographyHR High resolution

ICP-OES,MS Inductivity coupledplasma – optical emissionspectrometry, massspectrometry

ICSD Inorganic Crystal StructureData Base

JCPDS Joint Committee of PowderDiffraction Standards

LA-ICP-MS Laser ablation-inductivelycoupled plasma-massspectrometry

LAMMS Laser micro(probe) massspectrometry

LC Liquid chromatographyLDI-TOF-MS Laser

desorption/ionization-timeof flight-mass spectrometry

MRP Mass resolution powerMSC Molecular sieve

chromatographyPIXE Proton induced X-ray

emission (spectrometry);µ-PIXE is frequently usedfor PIXE with focusedproton beams

PSE Periodic system of theelements

PTFE Poly(tetrafluoroethylene)SAED Selected area electron

diffractionSE Secondary electronSEM Scanning electron microscopySIMS Secondary ion mass

spectrometrySMPS Scanning mobility particle

sizingSTEM Scanning transmission electron

microscopyTEM Transmission electron

microscopy

INTRODUCTION 507

TLC Thin layer chromatographyTXRF Total reflection X-ray

fluorescence (spectrometry)UHV Ultrahigh vacuumWDX Wavelength-dispersive X-ray

(fluorescence spectrometry)XANES X-ray absorption near edge

structure (spectrometry)XPS X-ray induced photoelectron

spectrometryXRS X-ray spectrometryZ Atomic number of an elementZAF Atomic number –

absorption – fluorescencecorrection procedure inEPMA

1 INTRODUCTION: THESIGNIFICANCE OF SINGLE PARTICLECHARACTERIZATION IN SCIENCEAND INDUSTRY

The characterization of particles, especially aerosolparticles, is of great importance to a broad varietyof scientific and industrial fields [1].

• In atmospheric sciences, the characterization ofindividual aerosol particles, their size distri-bution and chemical composition is of greatrelevance to modelling atmospheric processesand for environmental control purposes [2,3]. Compared to conventional bulk techniques,individual particle analysis provides additionaland complementary information concerning ori-gin, formation, transport and chemical reactions[4–6].

• Occupational health monitoring relies on parti-cle collection and subsequent particle character-ization to evaluate health hazards for workersexposed to dusts from foundries, calcinationovens, powder handling, milling etc. [4, 7].

Particle characterization is an important sourceof information for compound identification forcleanroom control [6, 8]: in microelectronic com-pound fabrication or in ultratrace analytical labo-ratories for quality control of ultrapure chemicals,

it is essential not only to monitor particle concen-trations in air – which is done routinely – but alsoto identify the particles for possible elimination ofsources of particulate contamination [6]. The iden-tification of particles which caused malfunction ofhighly integrated microelectronic devices is alsovery important [8].

• Powders are the basis of powder technologies.Usually, single particle characterization is notnecessary in powder technologies and bulk char-acteristics are determined in routine quality con-trol. However, single particle characterizationis essential for the determination of so called‘heterogeneous impurities’ in raw and interme-diate products in powder metallurgy [9, 10].Heterogeneous impurities are particulate impu-rities in the raw and intermediate products ofpowder metallurgy with particle diameters aboveapproximately 5 µm. They are introduced intothe raw or intermediate powders at variousstages of production as e.g. ore or gangue par-ticles, abraded particles from grinding, millingand mixing operations or by careless powdermanipulation or storage (e.g. cigarette ash, tex-tile fibres, hair, rubber particles, aerosol dust,etc.). Owing to the sintering process in powdermetallurgy, such impurities are not homogenizedas in melting operations. Hence, the particlesremain unaffected or react partially with thematrix material to form inclusions which usu-ally act as centres for crack formation if thematerial is mechanically stressed. In fine wiredrawing or foil production, wires break or thefoils become perforated at the place of a het-erogeneous inclusion. The analytical determina-tion of heterogeneous impurities is therefore animportant procedure in the quality control of rawand intermediate products of powder metallurgy[9, 10].

• The characterization of wear particles, e.g. inpolymer extrudates or in motor lubricants [11], isanother importantareaof technological relevance.

• The quality control of composite particles desig-ned for coating operations by plasma sprayingor of particles used in powder technology andmetallurgy is a further very important field [12].


• The characterization of complex particles isof relevance in some applications, e.g. forpigments. Heterogeneous catalysts are usu-ally also applied in particulate form andtheir characterization is accomplished usingdiverse methods of solid state analysis [13].Another important group of particles of rel-evance in analytical chemistry are chromato-graphic materials for column fillings for gaschromatography (GC), high performance liquidchromatography (HPLC), molecular sieve chro-matography (MSC) (e.g. macroreticular or pel-licular resins or materials for the preparation ofthin layer chromatography).

• A rather exotic but nevertheless important roleis played by cosmic [14–16] and terrestrial par-ticles in the degradation of material surfacesexposed to space in the low Earth orbit (LEO)range, i.e. in an altitude of 400–600 km abovethe Earth’s surface. The most alarming obser-vation was that more than 80 % of these parti-cles are of anthropogenic origin, i.e. man madedebris (e.g. paint particles from space shuttles,particulate residues from solid state rockets etc.).The differentiation between cosmic and anthro-pogenic particles is only possible by isotoperatio measurements of particles or what remainsof such particles on degraded material surfaces[15, 17]. Because of speeds of typically sev-eral km s−1 on impact, practically no particlessurvive collisions with space-exposed materialsurfaces and what remains is a thin layer ofrecondensed matter after being vaporized dur-ing impact. This layer can only be qualitativelyanalysed by secondary ion mass spectrometry(SIMS). Material degradation by the combinedeffect of erosion by particles and oxidative pro-cesses on such eroded surfaces by the atomicoxygen which is still present in the LEO regionis becoming a serious problem for satellite life-times in this region [15, 18].

• A similarly exotic field is the forensic character-ization of the diverse particulate material fromcrime scenes, which is used as physical evidenceand is a major task in forensic science [19].

• Last but not least, volcanic dust particles andinterplanetary dust particles are studied intensely

in geology: explosive volcanic eruptions caneject vast amounts of gaseous and particulatematerial into the atmosphere within a briefperiod. Much of this material remains in thetroposphere for a considerable length of time[20]. These particles can greatly reduce theradiation of the sun reaching the Earth’s surfaceand thus decisively influence weather patterns.In addition, interplanetary dust particles arestudied intensely since they often stem from thevery beginning of the formation of our solarsystem 4.5 billion years ago, and some of themmight even stem from pre-solar system times[16, 17].

Taking into account this broad range of interest inparticle characterization, a multimethod approachfor the comprehensive characterization of particleshas been developed for a particle diameter rangefrom 10 nm to 100 µm, which is relevant in theabove fields of study.

A thorough evaluation of the particle size dis-tribution as well as of the lifetime of particles ofvarying size in different domains of the Earth’satmosphere showed that particles smaller than100 nm in diameter exhibit very short lifetimesmainly due to agglomeration [20]. Furthermore,most sources of particles introduced into the atmo-sphere emit particles of 100 nm and more in diame-ter, many natural sources emit particles even in the1 µm range [21]. This is the reason why our mainscheme of particle characterization is designed forparticles >100 nm. However, particles with diame-ters down to 10 nm are of relevance for health haz-ard studies to evaluate the toxicity of dust in manyindustries [22]. In this case, the application oftransmission electron microscopy (TEM) is neces-sary for particle size distribution measurements inthe nm range and for phase and/or compound iden-tification of such particles. Alternatively, high res-olution scanning electron microscopy (HR-SEM)can also be applied [1, 21].

Nanometer sized particles have also becometechnologically important in materials science forthe production of materials with new and inter-esting properties. They have also revolutionizedestablished technologies and led to the develop-ment of materials with greatly improved high-tech

PARTICLE EVALUATION + MORPHOLOGY 509

properties [23]. The characterization of such parti-cles is also accomplished by the combined use ofTEM and HR-SEM.

The large diameter end of aerosols is again lim-ited by the increasingly short lifetimes of largeparticles in the range of 25–100 µm in diameterdepending – as can be expected – on particle den-sity and morphology. This upper limit is equallysignificant for health hazard investigations, sincecoarse particles are usually trapped in our respi-ratory system before the particles can reach thedeeper bronchial system and the lungs. For vari-ous reasons, this is also the upper limit of particlesof technological interest (heterogeneous particlesin powder metallurgy, wear particles, harmful par-ticles in microelectronics technology) [8–10].

2 SEMIQUANTITATIVE PARTICLEEVALUATION + MORPHOLOGYOFTEN = SPECIATION

It should be emphasized that besides special meth-ods of single particle speciation, which are still lesscommon and will be discussed later, the primarilydescribed procedure of the routinely used HR-SEMand electron probe microanalysis (EPMA) shouldbe regarded as a most important step towards sin-gle particle speciation. In many cases semiquanti-tative single particle analysis by EPMA, togetherwith an often typical particle morphology as stud-ied by HR-SEM, unambiguously answer the ques-tion of the compound making up the particle. Inother cases speciation is possible by selected areaelectron diffraction (SAED) in the transmissionelectron microscope or by the modern technique ofenergy filtering transmission electron microscopy(EFTEM). In still other cases only the appropri-ate combination of such techniques will give aconclusive answer to particle speciation, especiallyof compound particles which occur very oftenin aerosols.

Figure 6.1.1 demonstrates this possibility of anHR-SEM identification of a particle found in anaerosol from the largest aluminium productionfacility of Norway [21]. Such cryolite coatedAl2O3 particles were found to make up about 50 %(in particle numbers) of aerosols collected in the

electrolysis hall at the Norsk Hydro Aluminiumplant, Haugesund, Norway. They are the mostlikely reason for the so-called ‘potroom asthma’,which occurs in such sites of the electrolyticproduction, of aluminium metal [23].

Figure 6.1.2(a) shows a TEM micrograph ofa platinum catalyst particle, which was collectedtogether with nickel-containing particles in the‘reduction hall’ of the secondary nickel refineryat Monchegorsk (Penninsula of Kola, Russia)[24]. Figure 6.1.2(b) gives the corresponding EDXspectrum. The copper lines stem from the coppergrid of the TEM polycarbonate foil on whichthe particles were collected. The C Kα line israther intense and stems from the soot whichis obviously combined with the finely dispersedPt. Figure 6.1.2(c) shows the electron diffractiongrainy ring pattern. This is indicative of the verysmall grain size of the Pt particles, of the orderof tens of nanometres. It also unambiguouslyidentifies the particle as platinum.

Figure 6.1.1 demonstrates the principal possi-bility of compound identification by qualitativeEDX analysis together with particle morphologywhich yields additional valuable information, e.g.related to the toxicity of certain aerosol parti-cles. This information is even superior to merespeciation and this is the reason why the wholescheme of particle characterization that has beendeveloped is elucidated. Figure 6.1.2 demonstratesthe very valuable combination of direct speciationby electron diffraction in the transmission elec-tron microscope. In addition to mere speciation, ityields a complete compound identification of crys-talline phases which includes the relevant crystalstructure. Again, this information is in many casesimportant since identical compounds with varyingcrystal structure might be of varying toxicity as isthe case for nickel sulfides. Their toxicity falls inthe following compound sequence [25]: α-Ni3S2,β-NiS > NiO � metallic Ni � amorphous NiS.

2.1 The key to a profound particlecharacterization: a multimethod approach

Table 6.1.1 presents an overview of the multi-method approach to particle characterization of


(a)

(b)

Energy (keV)

0.80 1.60 2.40 3.20 4.00 4.80 5.60 6.40 7.20 8.00

O Ka

NiLa

AlKa

C Ka

F Ka

NaKa

SiKa

S KaNiKa

NiKb

Figure 6.1.1. HR-SEM study of aerosol particles collected in the hall of one of the largest aluminium production sites ofNorsk Hydro Aluminium at Haugesund, Norway. (a) Al2O3 particle coated with Na3[AlF6]. (b) qualitative EDX spectrum. Maincomponents: Al, O. Coating: Na, Al, F. Further elements present: C, Si, S; the Ni lines stem from the sample holder.


(a) (c)

200 nm

(b)

0

1 2 3 4 5

Energy (keV)

6 7 8 9 10

20

40

80

60

100

120Cou

nts 140

160

Cu

SiO

C

Pt

180

200

220

240

260

NiCu

Cu

Pt

Figure 6.1.2. TEM characterization of a platinum-containing particle, an example of particle identification by TEM/EDX/electrondiffraction. (a) Diffraction contrast micrograph. (b) Qualitative EDX analysis: the Cu lines stem from the filter material.(c) Electron diffraction ring pattern typical for Pt metal.

solid aerosols which was developed by us [1]. Ofcourse, such a scheme does not work without cer-tain limitations: since practically all topochemicalmethods applied are working under vacuum condi-tions (some of them even under ultrahigh vacuum(UHV) conditions), the inspection of droplets isnot feasible and moisture and volatile compoundsare thereby removed. Hence, many aerosol par-ticles undergo a certain change before they canbe inspected.

Figure 6.1.3 demonstrates the possibility of spe-ciation and – more than that – the evaluation ofthe crystal structure and thus the phase identifica-tion by TEM. This is especially important for casessuch as nickel compounds: the determination ofcarcinogeneity of such particles depends not onlyon compound identification but also on the deter-mination of the crystal structure. Besides Millerite,Heazlewoodite (Ni3S2), NiO and NiSO4 · H2Ohave been found in aerosols of Monchegorsk [24].


Table 6.1.1. Scheme of single particle characterization with special reference to speciation.

Sampling Particle collection with a three- or five-stage impactor onto high purity glassy carbondisks or by a one-stage impactor for very low aerosol concentrations. Carbon coating byevaporation (if necessary).

SEM survey (preferably byHR-SEM)

• Selection of samples with the appropriate particle density (if sampling with varyingsampling times was performed).

• First survey of particle size distribution.• First survey of particle morphologies.• Check on particle homogeneity or heterogeneity and on particle aggregates.• Qualitative analysis of individual particles by EDX for selection of elements to be studied

by EPMA.

Semiquantitative EPMAinvestigation (down to0.5 µm particle diameter)

• Elemental mapping (usually by WDX) for elements selected by the SEM survey.• Semiquantitative particle evaluation yielding the following information:

– elemental composition of particles (including C, N, O but no trace contents);– number and size distribution of particles of specific composition;– systematic assignment of size distribution and morphology to identified classes of

particles;– assessment of elemental homogeneity or heterogeneity of individual particles.

Quantitative bulkcharacterization by TXRFfor main, minor and traceelements

• Quantitation is accomplished by use of Sc as internal standard for low Z elements and ofY for high Z elements.

ESEM work • For particle inspection under moisture saturated conditions and investigations concerningmorphological changes with decreasing humidity.

• For the investigation of insulating particles without coating.

AFM work • For quantitative volumetric particle investigations, under varying ambient conditions inthe tapping mode.

EBSD • Only applicable for well-developed crystals larger than 5 µm and with flat surfaces (e.g.platelets). Phase identification is then possible for single particles.

Explicit methods of particle speciation

Valence bond X-rayspectrometry byEPMA-WDX

• Single particle speciation is often possible for low Z elements and for particles > 5 µmin diameter. Time consuming.

TEM/SAED • For particles smaller than 0.5 µm in diameter, compound identification is possible forcrystallized species usually performed in combination with elemental analysis by EDXand/or EELS.

EFTEM or ESI • For particles smaller than 100 nm or for particle domains not thicker than 100 nm,mapping of elemental binding states is feasable with single nm-lateral resolution infavourable cases as long as the particle endures the high intensity of the electronbombardment.

XPS • Most commonly used method of solid-state speciation. Synchrotron-XPS allows singleparticle characterization.

AES • Speciation possible especially for low Z elements with excellent lateral and depthresolution (both in the low ten nm range).

Moßbauer spectrometry • Unfortunately limited to the mg range (with special apparatus). No single particlespeciation possible. Restricted to iron compounds and some other compounds.

Micro-Raman spectrometry • Functional group analysis in organic particles, ≥1 µm possible with modern laser Ramanmicroprobes.

µ-EXAFS and µ-XANES • Requires synchrotron X-ray radiation. EXAFS yields a series of valuable information alsofor amorphous particles such as:– type(s) of nearest neighbour(s);– distances to nearest neighbours;– coordination number.XANES gives information on the valency state and/or chemical state of the probed atom(e.g. oxidic or metallic etc.) With appropriate X-ray optics, mapping of this informationwith resolution in the low ten µm range is feasible.



Trace and isotopic characterization of single particles by mass spectrometric methods and PIXE

SIMS The most sensitive topochemical method. The NANOSIMS has been constructed withdedication to particle analysis (cosmic particles) with diameters in the tens of nm range.Speciation is sometimes possible from the observed mass pattern. 3D SIMS is possiblefor particles with certain geometries (e.g. platelets). Lateral resolution with conventionalsector-field mass spectrometers is in the single µm range, depth resolution in the low10 nm range.

LAMMS and LA-ICP-MS LAMMS is extensively used for single particle characterization. Speciation is feasible infavourable cases and especially for inorganic compounds. Isotopic composition is alsodeterminable, also with LA-ICP-MS which is the more sensitive technique of the two.

LDI-TOF-MS Can be used for the on-line characterization of single particles from a sampled air stream.Very fast. Quantitation possible in favourable cases.

µ-PIXE Single particle trace analysis with focusing nucleoprobes possible. Lateral resolution around5 µm.

2.1.1 Particle collection with a cascadeimpactor on glassy carbon disks

Taking into account the broad range of interestin particle characterization as outlined above,a multimethod approach for the comprehensivecharacterization of particles was developed for aparticle diameter range from 10 nm to 100 µm,which is relevant in the above fields of study. Athorough evaluation of particle size distributionsand of lifetimes of particles of varying sizein different domains of the Earth’s atmosphereshowed that particles smaller than 100 nm indiameter exhibit very short lifetimes, mainly dueto agglomeration [26]. Furthermore, most sourcesof particle emission emit particles of 100 nm andmore in diameter. Many natural sources introduceparticles into the atmosphere even in the 1 µmrange [26]. In contrast, the presently and generallyaccepted ranges of particle sizes of relevancefrom a health related point of view are thefollowing [27]:

• Inhalable fraction: is defined as the mass frac-tion of total airborne particles which is inhaledthrough the nose and/or mouth. It comprisesparticles smaller than approximately 100 µm inaerodynamic diameter.

• Thoracic fraction: is the mass fraction ofinhaled particles penetrating the respiratorysystem beyond the larynx. It is given by acumulative lognormal curve with a median aero-dynamic diameter of 10 µm and a geometricstandard deviation of 1.5.

• Respirable fraction: is the mass fraction ofinhaled particles which penetrates to the uncil-iated airways of the lung (alveolar region). Itis given by a cumulative lognormal curve witha median aerodynamic diameter of 4 µm and ageometric standard deviation of 1.5.

• ‘High risk’ respirable fraction: is a definition ofrespirable fraction for the sick and infirm, orchildren. It is given by a cumulative lognormalcurve with a median aerodynamic diameterof 2.5 µm and a geometric standard deviationof 1.5.

These are the reasons why our main scheme ofparticle characterization is designed for particleslarger than 100 nm. In contrast, particles withdiameters down to 10 nm are of great relevancefor health hazard studies to evaluate the toxicityof dust in many industries [22] and of diesel soot[27]. It should be mentioned that the influenceof fine and ultrafine particles on human healthis attracting the greatest current attention. Often,adverse health effects seem to be linked withsmaller particles [28]. In addition, it seems thatnearly all of the mass emitted by modern dieselengines is in the nanometre diameter range [28].The application of TEM is necessary for particlecharacterization in the nanometre range and forphase and/or compound characterization of suchparticles, as will be discussed later.

Alternatively, HR-SEM can also be appliedwhich is experimentally much faster and eas-ier. Particle size distribution measurements in


(b)

0

1

O

Si

Ni

SNi

2 3 4 5

Energy (keV)

6 7 8 9 10

50

100

150

200

250

300

350

400

Cou

nts

450

500

550

600

650

700

CuCu

Ni

Cu

1µm

(a) (c) (d)

Figure 6.1.3. Speciation by electron diffraction in the transmission electron microscope: (a) diffraction contrast micrograph of anNiS particle; (b) qualitative EDX analysis; (c) electron diffraction, zone axis (21 − 2); (d) Electron diffraction, zone axis (211).

the nanometre range are today usually performedby scanning mobility particle sizing (SMPS).The SMPS instrument of TSI Inc., USA, isbased on droplet growth by condensation nucle-ation, followed by particle size distribution anal-ysis of the droplets grown into the micrometrerange [29]. It consists of a differential mobil-ity analyser (DMA) and a condensation particlecounter (CPC).

Aerosol collection with a one-stage impactorhas been used for particles with a minimum diam-eter of 100 nm for relatively long periods (hours)in order to obtain the necessary sensitivity for ele-ment detection in the pg m−3 range in clean-roomatmospheres [6]. Collection times must, of course,be shorter (minutes or even seconds) for heavilycontaminated atmospheres in industry in the rangeof mg m−3 for health hazard evaluation [30].


2.1.2 Electrostatic particle collectionfor nanometre particles

Particles with diameters below 100 nm cannot bedeposited by impaction. Electrostatic depositionmethods are in use for this purpose [29, 31, 32].Electrostatic deposition is a well-known techniquefor collecting airborne particulates as well as artifi-cially produced particles for subsequent investiga-tion with various analytical systems [32]. For thistechnique high deposition efficiencies up to 100 %have been reported for particulates in a size rangeof 3.5 nm to 3 µm [32]. The theory of electrostaticdeposition predicts high deposition efficiency forlarger and smaller particles as well [33]. A suitablecommercially available instrument is also offeredby TSI, as was mentioned above [29, 31, 32].

2.1.3 Topochemical characterization of solidaerosols by SEM and EPMA

2.1.3.1 Scanning electron microscopy

The first step in our scheme of particle charac-terization is always an SEM survey, in whichimportant information is collected on the follow-ing aspects:

(a) Particle abundance. This information is impor-tant for optimizing the particle collection param-eters for an appropriate number of particles perunit area of the glassy carbon disks. In the caseof remote sampling, this is not possible and sam-ples are collected with various sampling times andthe optimally loaded ones are selected for furtherinvestigations.

(b) Size distribution. This is an important parame-ter of its own. It also yields information on whetherEPMA can be applied as next step or whether TEMor HR-SEM would have to be selected in the caseof particles with diameters below 0.5 µm.

(c) Particle morphologies. In many cases, certainparticle morphologies already give the possibilityto identify certain particle types, e.g. soot particlesexhibit such a typical morphology that they canusually be identified in this step. The same is truefor most biological particles (pollen) salt particlesfrom the sea and many mineral particles.

(d) Particle homogeneity or heterogeneity. Formany aerosol particles, especially for aggregates,their heterogeneous structure often becomesevident.

(e) Sample charging. In most practical cases, itis not necessary to coat the glassy carbon diskswith the collected particles because of the sufficientelectrical conductivity of the glassy carbon. How-ever, if larger amounts of highly insulating parti-cles are present, coating with carbon is performed.

Several particles (the number depends on theextent of morphological and constitutional varia-tion of the inspected sample) are then qualitativelyanalysed by EDX for the selection of elementswhich need to be mapped by EPMA.

It should be emphasized that particle morpholo-gies and morphological details of particles in thenanometre range can only be properly studied bySEM instruments with a field emission gun oran LaB6 electron source [34]. This is the reasonwhy HR-SEM instruments are used throughout ouraerosol investigations. It is also essential to use amodern thin window EDX detector since classicalEDX detectors are not capable of analysing ele-ments with an atomic number Z < 11. Oxygen andcarbon are always included in our WDX elementmapping. Since B and F are usually not expected inparticles, their presence is only checked in specialcases. This leaves only nitrogen mapping ques-tionable for elements of the second period of theperiodic system of elements (PSE). Like B and F,it is only mapped in special cases.

Environmental scanning electron microscopyuses instruments that are able to work under lowpressures. These have been under development forquite some years. They have now reached matu-rity especially with the Philips ESEM series [35,36]. Water vapour can be introduced as a gasinto the specimen chamber so that saturated cham-ber conditions can be reached and maintained.Water can even be condensed onto as well asremoved from the sample in a controlled man-ner. This allows the morphological and analyticalinvestigation of samples under moist conditionswhich would create morphological artifacts in anormal SEM instrument under vacuum. In order


to create a water saturated atmosphere at low pres-sure, a Peltier cooling stage is available which can,of course, also be used for the removal of waterfrom samples under vacuum by freeze drying. Fur-thermore, a specimen chamber gas handling systemallows the introduction of various gases at lowpressure. Air can be introduced, e.g. for low orhigh temperature oxidation studies (in dry or moistair) since a hot stage is also available for tem-peratures up to 1500 ◦C. Such conditions are, ofcourse, especially interesting for material sciencestudies. The evolved gases of such experimentscan be introduced into a quadrupole mass spec-trometer via a small capillary and analysed. Gasconcentrations of <1 µg mL−1 up to 100 % arethus determinable.

This successful operation of scanning electronmicroscopes at low pressure is only possible withsome constructional developments [35, 36]:

(a) The beam gas path length, i.e. the distance thatthe primary electron beam has to overcomefrom the high vacuum system of the electronoptics column to the sample, is kept to aminimum. This is also important for thelateral resolution of X-ray detection which ishampered by primary electrons being scatteredby gas molecules.

(b) Secondary electron detectors have to be mod-ified to so-called ‘gas amplification detectors’,whose operation is analogous to the flow pro-portional gas detector used in WDX. Backscat-tered electrons suffer negligible energy loss inthe gas atmosphere and retain sufficient energyto activate large area scintillators without post-specimen acceleration.

(c) A great advantage of operation at elevatedpressure is the automatic discharge of the neg-atively charged surface of insulators due to gasions that are attracted by this charge and ‘neu-tralize’ it. The gas ions above the sample aregenerated by the primary electron beam and bysecondary electron ionization processes. Thisenables the morphological study of insulatingsamples without a coating. In a subsequentanalytical investigation under vacuum for lightelements there is no interference of a conduc-tive coating.

Variations of the morphology and volume ofparticles with e.g. varying humidity can alsobe advantageously investigated by atomic forcemicroscopy (AFM) in the tapping mode [37].Tapping mode means that the needle of the AFMinstrument is raised from and lowered onto thesample with a relatively high frequency (kHz)while the sample is slowly moved along in thex, y plane. In this way, particles are generallynot moved from their original position by thetransgressing needle. Particle morphologies canthus be determined without interference of theneedle which is usually led over the particle inthe noncontact mode.

Since no comparative studies are yet avail-able with the ESEM and the AFM it is dif-ficult to say whether the AFM approach bearsadvantages over respective ESEM observations.Certainly, quantitative volumetric observations ofexpanding or shrinking particle volumes undervarying ambient conditions are only possible byAFM. Single sub-µm particles have been inves-tigated with respect to changing humidity in thesurrounding atmosphere [37]. Volume calculationsallowed monitoring of these changes on a quanti-tative basis.

2.1.3.2 Electron backscatter diffraction, EBSD

The evolution of this technology has essentiallytaken place in the 1980s. Today it is availableas an add-on package to a scanning electronmicroscope [38]. The essential features of EBSDare its unique capabilities to yield crystallographicdata by imaging in real time with a spatialresolution of 0.5 µm, combined with the regularcapabilities of SEM, such as capacity for largespecimens and option of chemical analysis. Inother words, EBSD can automatically determinecrystallographic data of each grain that is analysedby electron diffraction to yield so-called Kikuchipatterns from which these data can be obtained.Hence, EBSD could also be used for phaseidentification of each analysed particle [38]. Thiswas the onset to our efforts to apply EBSDalso to the characterization of aerosol particles.Unfortunately, these efforts were unsuccessful


[24]. The reason for this is that EBSD needswell-developed crystals at least 1 µm in diameter(due to the spatial resolution of EBSD of 0.5 µm).Another restriction is the fact that EBSD onlyworks well for flat and polished surfaces. For theusually very irregularly shaped aerosol particles itis generally not possible to obtain the respectiveKikuchi patterns. In addition, aerosol crystallitesvery frequently contain many imperfections, whichconsequently blur the evolution of a Kikuchipattern. It is due to these circumstances thatEBSD cannot be applied generally to phaseidentification of aerosol crystallites unless theseare big enough (>1 µm in diameter) and wellcrystallized.

2.1.3.3 EPMA mapping

A qualitative method for the characterization ofa great number of individual particles based onelement distribution maps has recently been devel-oped [39]. Using this procedure, the size, shapeand qualitative chemical composition of each par-ticle can be deduced from a combination of sec-ondary electron (SE) and backscattered electron(BSE) images and element distribution maps forthose elements which have been chosen by theabove SEM survey. The knowledge of the qualita-tive chemical composition of individual particles issufficient for many applications. For example, wehave recently investigated the significance of ironin atmospheric processes [40]. The iron-bearingparticles were characterized by our analysis pro-cedure and were classified into several categories(i.e. metal, oxide, silicate particles). We are alsousing this procedure for source apportionment ofaerosols, where characteristic elements for eachsource are known from bulk measurements bytotal reflection X-ray fluorescence (spectrometry)(TXRF) (see below).

In addition to the qualitative determination, asemiquantitative estimate of the chemical compo-sition of each particle can also be obtained [39].For this purpose, the count rates for the measuredelements of each particle are derived from therespective element distribution maps. These arecorrected for matrix and geometric effects using

the particle ZAF procedures of Armstrong [41, 42].In contrast to the qualitative procedure, spectrome-ter defocusing cannot be neglected in semiquanti-tative analysis, even at high magnifications. Thecorrection is performed using algorithms whichwere also developed [43].

This semiquantitative approach turned out tobe necessary e.g. for particle characterization inoccupational health monitoring. In the course ofthe evaluation of samples collected at the largestnickel refinery in the world at Monchegorsk onthe Kola peninsula (Russia), it became apparentthat the differentiation of particles containing var-ious amounts of nickel oxides and nickel sul-fides in contrast to nickel sulfate particles wasonly feasable by semiquantitative characterization[30]. In contrast to a previously developed sim-ilar procedure [44] we use WDX rather thanEDX in order to include the very important ele-ments C, N and O. Recently, the EDX proce-dure [44] has been further developed to use win-dowless EDX detectors in EPMA [45]. In addi-tion, Monte Carlo calculations were developed toimprove the very difficult matrix correction pro-cedures for the soft X-rays of the elements ofthe second period of the PSE [45]. Whether ornot this leads to data comparable with the WDXEPMA procedure has not yet been investigated.It is doubtful because the second great draw-back of the use of EDX detectors still remains:the essentially worse spectral resolution whichinevitably leads to serious line interferences ofspectral lines of metallic elements of higher Z withthe analyte lines of the second period elementsto be determined. Unfortunately, the EPMA map-ping method is not capable of analysing particlessmaller than 0.5 µm in diameter for instrumen-tal and physical reasons. Therefore, other meth-ods and instrumentation have to be applied forsmaller particles.

2.1.4 Quantitative bulk characterizationon glassy carbon disks by TXRF for main,minor and trace elements

It is beyond the scope of this chapter to give anoverview of the wide range of methods used today


for the bulk characterization of particulate samples(e.g. atomic absorption spectrometry (AAS), X-rayspectrometry (XRS), inductively coupled plasma-optical emission spectrometry (ICP-OES), inducti-vely coupled plasma-mass spectrometry (ICP-MS),etc.) [46]. Only the use of TXRF is outlined here.TXRF is a trace and microanalytical techniqueand matches the specific requirements of impactor-collected particulate samples in an ideal way: itcovers the small area where particles are depositedbelow the respective impactor jets and allows asafe multielement quantification by internal stan-dardization. TXRF has been used extensively forquantitative aerosol analysis [47, 48], and is oneof the most important methods for this purpose.Quantitative determination of the elemental com-position of the collected aerosol is carried outwith two TXRF instruments on the glassy car-bon carriers of selected purity. Quantitation isaccomplished by use of Sc and Y as internal stan-dards for elements of low and high Z, respec-tively. Details of the procedure are described inrefs 5 and 6. Since the application of the inter-nal standard solution changes the aerosol par-ticles it has to be applied as the last step ofthe usual procedure after the morphological andEPMA survey.

3 EXPLICIT METHODSOF PARTICLE SPECIATION

We are now arriving at methods which can givedirect information on particle speciation. However,these methods are less frequently used in routineaerosol characterization than those discussed upto now.

3.1 Valence band X-ray spectrometryby EPMA-WDX

This method is already well established for theidentification of binding states in solids, especiallyfor elements of low and medium Z [49]. Thebasis is the ‘chemical shift’ of X-ray lines if theonset of the electron jump to an inner orbitalvacancy lies in the valence band of the respective

atom. Generally the line shift is small (about1–3 eV) in comparison with the peak width ofX-ray lines (full width at half-maximum (FWHM)≈40 eV). In geology and mineralogy, a methodhas been developed to determine the Fe(II)/Fe(III)ratio in solid samples by precision measurementof the variation of the position and shape of theFe Lα and Fe Lβ lines [50]. The Lα/Lβ intensityratio might also vary. This method has the greatadvantage over other established methods forsolid-state speciation (XPS, AES) that it exhibitsmuch greater detection sensitivity and/or lateralresolution (compared with XPS only in this case).Bulk speciation can be carried out directly onsamples collected on glassy carbon disks. Forparticles >5 µm, single particle speciation is alsopossible but time consuming as Hoflich et al. wereable to demonstrate recently [51]. Unfortunately,many particles exhibited instability by the longmeasuring times which are necessary for X-rayprofile precision measurements.

3.2 TEM of particles smaller than 0.5 µmin diameter

3.2.1 Experimental procedure

Samples for TEM investigations are usually obtai-ned on the last (fifth) stage of the cascade impactorwith a Formvar foil placed on the glassy carbondisk. The Formvar foil is a polycarbonate filterreinforced by a copper grid. Very fine particleswhich cannot be collected by impaction are eitherdeposited electrostatically or they are depositedby suction with a small pump on a Formvar foilwhich is placed on a ceramic filter support. In somecases, for ‘concentrated’ industrial aerosols, mereexposure of the Formvar foil to the contaminatedatmosphere for a few minutes is sufficient. TheFormvar foil is carbon coated after samplingand prior to TEM inspection. Particles can becharacterized in only a limited number sinceautomation comparable to that outlined above forthe EPMA procedure is not yet available. Particlesthicker than 50 nm cannot be penetrated by theprimary electron beam. However, thinner portions

EXPLICIT METHODS OF PARTICLE SPECIATION 519

of such particles can usually be inspected attheir edges.

3.2.2 Information content of TEMinvestigations

The following information can be obtained byTEM investigations:

(a) Particle size distribution and partly morphol-ogy can be studied by imaging with the transmit-ted beam.

(b) Phase identification is possible by selectedarea electron diffraction (SAED). If point pat-terns are obtained for single crystals, the observedreflections can be converted into the respective lat-tice constants. The latter are used for compoundidentification by search in the Inorganic Crys-tal Structure Data Base (ICSD) using the proce-dure ‘PIEP’ which was developed by Miehe [52].Other suitable programmes have also been devel-oped [53]. In the case of very small crystallitesand nanocrystalline particles, diffraction rings areobtained. Since atomic number dependences ofthe scattering power for electrons and X-rays aresimilar, the relative intensities in the correspond-ing electron and X-ray powder diffractograms arealso similar. It is, therefore, possible in such casesto carry out a compound identification with theJoint Committee of Powder Diffraction Standards(JCPDS) data bank for X-ray powder diffrac-tion data.

(c) Compound identification for amorphous parti-cles is possible in simple cases (if no multiphaseparticles are present) by the determination of the ele-mental composition by EDX or by electron energyloss spectrometry (EELS) or, preferably, by a com-bination of the two methods. In favourable cases,speciation is possible because EELS peaks oftencontain binding information in the respective nearedge fine structures [54]. EDX and EELS are com-plementary since EELS is most sensitive for lowZ elements whereas EDX is more powerful forelements of medium and high Z by the nature ofthe underlying physical process. Xhoffer et al. havegiven a more detailed account of EELS for singleparticle analysis [48].

3.2.2.1 Energy filtering TEM (EFTEM)or electron spectroscopic imaging (ESI)

There are some very promising developments inTEM-EELS which will have a substantial influ-ence on single particle characterization: up tonow, EEL spectra were usually acquired froma small selected area (point analysis). Now, thetwo-dimensional acquisition of any spectral fea-ture of an EEL spectrum has become possible withwhat is called energy filtering transmission electronmicroscopy (EFTEM) or electron spectroscopicimaging (ESI) [55, 56]. Energy filtering devicesfor TEM have become commercially availableonly recently and the Gatan imaging filter (GIF)can be attached to almost any 100–400 kV TEMinstrument [57]. With ESI, energy filtered imagesare acquired which can then be combined toshow the distribution of elements in the specimenwith nanometre resolution [55]. This procedure hasadvantages over the scanning transmission electronmicroscopy (STEM)–parallel EELS combinationsince acquisition times for high resolution mea-surements of the latter are essentially longer thanthose for ESI [58]. In favourable cases, ESI allowsthe mapping of elemental binding states with singlenanometre lateral resolution [55, 56].

3.3 X-ray induced photoelectronspectrometry (XPS)

XPS is the most commonly used method of solid-state speciation with excellent depth resolution inthe single nanometre range [34, 59]. Lateral res-olution, however, is limited to the low millime-tre range in older instrumentation. Modern XPSinstruments can achieve a lateral resolution in thelower micrometre range and can thus be usedfor single particle XPS characterization [1, 3, 4].The best lateral resolution in the single microme-tre range is achieved with XPS instrumentationadjoined to electron accelerators, e.g. in Greno-ble or Hamburg (Hasylab). The application of XPSto environmental particulate samples has been dis-cussed extensively by Xhoffer et al. [48]. It shouldbe mentioned that XPS is a UHV method andonly vacuum-stable particles can be studied. On


the other hand, the danger of particle disintegrationby X-rays is considerably lower than for elec-tron probe methods discussed so far. On accountof its excellent depth resolution, XPS in com-bination with SEM/EDX or EPMA/WDX allowsresearchers to distinguish between bulk composi-tion and thin coatings of larger particles.

3.4 Auger electron spectrometry (AES)

AES can be used especially for the speciationof low Z elements with excellent lateral anddepth resolution (in the 10 nm range). Auger signalshapes are frequently sensitive to relevant bindingsituations and allow e.g. the differentiation ofgraphitic, amorphous and carbidic carbon [34].

3.5 Mossbauer spectrometry

This method has been specifically applied for ironspeciation in large, integral aerosol samples [60].Iron is one of the most abundant elements insolid and aqueous atmospheric samples [61]. It isusually introduced into the atmosphere as soil dust,fly ash from power plants and waste incinerationfacilities, from exhaust of combustion enginesand generally from industrial operations [62]. Athorough study of the significance of iron foratmospheric redox reactions and on the presenceof iron in the above mentioned sources wasrecently carried out [63]. Magnetite, hematite,goethite and iron(III) silicates were found inthe inspected samples [60, 63]. Speciation byMossbauer spectrometry is also possible for anumber of other elements [64].

3.6 Micro-Raman spectrometry

With the advent of well focused laser beams assources for the excitation of Raman spectra andparallel developments in relevant instrumentation,the analysis of discrete particles ≥ 1 µm in diam-eter has become possible by Raman spectrometryyielding valuable information on functional groupsin organic particulates [34]. Although additional

vibrations might be caused by a well-definedgeometry of the inspected particles, this is usu-ally of no concern for irregularly shaped particleassemblies which are generally present in aerosolsamples. A more detailed discussion of micro-Raman spectrometry for particulate samples canbe found elsewhere [46].

3.7 X-ray absorption techniques:µ-EXAFS and µ-XANES

These methods have been in use in materials char-acterization and biological studies for quite sometime [34]. However, only recently has their lat-eral resolution been improved to the micrometrelevel by the use of polychromatic lenses in com-bination with an Si (111) channel-cut monochro-mator [65–67]. While EXAFS uses an energyregion extending from 50 to as much as 1500 eVabove the K-absorption edge of the probed ele-ment, XANES analyses the region within ± 50 eVof the absorption edge. EXAFS yields the follow-ing information [34]:

• nearest neighbour(s) distance(s);• type(s) of nearest neighbour(s);• coordination number;• Debye–Waller factors, which are indicative of

the degree of vibrational and static disorder.

XANES probes the shape of the absorption edge.This will change with the varying binding situationof the probed atom. Variations in the valence statecan thus be determined and chemical state mappinghas become feasable. The XANES spectrum isused as a ‘fingerprint’ spectrum for the elementof interest in the material under examination. TheEXAFS spectrum is mathematically manipulatedin the following way [68]:

(a) Isolation of the fine structure from the gen-eral absorption.

(b) Conversion to a reciprocal space representation(k-space).

(c) Application of a Fourier transform that convertsthe k-space EXAFS spectrum into a kind ofradial distribution function, which is calleda radial structure function (RSF). The RSF

TRACE AND ISOTOPIC CHARACTERIZATION OF SINGLE PARTICLES 521

describes the local structure about the atom thatabsorbs the X-rays. Peaks in the RSF representthe coordination shells of atoms.

A disadvantage of these methods is the neces-sity to use very intense X-ray beams which areonly available at large synchrotrons, e.g. the Euro-pean Synchrotron Radiation Facility (ESFR) atGrenoble (France) or the HASYLAB facilities atDESY in Hamburg (Germany). A great advantageof these methods is the fact that they can be per-formed under ambient conditions provided that theX-ray energy in the probed region of the absorptionedge of the element investigated is high enough(Z > 11). Only recently have these methods beenused for particle characterization and work is inprogress for Ni-containing particles [66].

4 TRACE AND ISOTOPICCHARACTERIZATION OF SINGLEPARTICLES BY MASSSPECTROMETRIC METHODSAND PIXE

4.1 Secondary ion mass spectrometry(SIMS)

Relatively few methods are available for the deter-mination of trace elements in single particles [1,7, 48]. SIMS is known to be the most sensi-tive topochemical method [34, 69]. It has beensuccessfully used for the trace characterization ofsingle particles [15, 18]. Since it uses minimalsample amounts during analysis, it can almostbe considered a nondestructive technique. A fur-ther unique characteristic of SIMS is its ability todetermine isotopic abundances and isotopic ratiosin single particles. This is important, e.g., fordistinguishing between cosmic and Earth debrisparticles in erosion studies of space-exposed mate-rials [14, 15, 18]. Owing to its good lateral res-olution in the beam scanning mode (better than0.5 µm), particles down to 0.5 µm in diameter canbe characterized with conventional instrumenta-tion. The newly designed NANOSIMS instrumentcan analyse particles with diameters in the 10 nmrange [70]. From numerous trace element measure-ments of Stadermann [15] on cosmic particles, it

can be deduced that at least the order of magnitudeof trace constituents can be reliably determined bySIMS in single particles.

It is interesting that isotopic ratio measurementsof single grains of cosmic dust by SIMS haveyielded important information on pre-solar mat-ter and on nucleosynthesis as the theory of ele-ment formation in stars [71]. In favourable cases,conclusions can be drawn on speciation from thepattern of various masses in a SIMS mass spec-trum [34].

4.2 Laser micro(probe) massspectrometry (LAMMS) and laserablation-inductively coupled plasma-massspectrometry (LA-ICP-MS)

LAMMS has been used extensively for singleparticle characterization and is a powerful toolespecially for the characterization of poorly vac-uum resistant organic particles [44, 48]. Equivalentin performance to SIMS, LAMMS is an off-linemethod because particles must be collected andmounted on a substrate. Detection of trace met-als is feasible at the µg g−1 level and speciationof inorganic compounds (especially those contain-ing nitrogen and sulfur) is possible. Distinctionbetween surface and volume compositions of a par-ticle is also possible, in addition to the detectionof trace organics (especially aromatics). However,large pulse-to-pulse and also particle-to-particlevariations of the ion signal intensity inhibit quan-tification [72].

LA-ICP-MS is probably a more powerful tech-nique owing to the much more efficient ioniza-tion of material in the inductively coupled plasma.Laser ablation of the sample, in contrast, shouldonly produce a very fine aerosol which is quan-titatively transported into the inductively coupledplasma by an argon carrier gas.

4.3 Laser desorption/ionization-time-of-flight-mass spectrometry(LDI-TOF-MS)

Laser desorption/ionization (LDI) coupled withtime-of-flight MS (TOF-MS) was used by Johnston


and Wexler [72] for the on-line characterization ofsingle particles from a sampled air stream. Severalparticles can be sampled and analysed per second,allowing the compilation of large data sets in avery short time. This allows an improvement inthe precision of analysis of particles exhibitingsimilar size and composition by averaging therecorded spectra. The average composition ofeach group is then quantitatively determined bycomparison with spectra of relevant standardparticles having known size and composition.Since quantitation is frequently synonymous withspeciation, the latter is also possible with thismethod in favourable cases.

4.4 Proton induced X-ray emission(spectrometry) (PIXE)

PIXE is another method with a trace characteriza-tion capability for particles with detection limits ofthe order of 1–10 mg g−1 [48, 73]. Other variantsof nuclear microprobe techniques are still restrictedto a few laboratories worldwide but can provide awealth of information for individual particles [48].µ-PIXE uses a well-focused proton beam and hasbeen used for the single particle characterizationof giant North Sea aerosols and other samples formajor, minor and trace element analysis [74].

4.5 Three-dimensional SIMScharacterization of large particles

In recent years it has been shown that three-dimensional (3D) imaging is a powerful newapplication of SIMS [75]. This method combinesthe surface imaging capabilities of SIMS withdepth profiling, creating layer-by-layer imagesof elemental distributions as the primary ionbeam sputters deeper and deeper into the sample.The amount of data produced during a typicalmeasurement of this type can easily reach severalhundred megabytes or even exceed one gigabyte.This is the reason why 3D SIMS has onlybecome practicable with the advent of fast meansof management of large amounts of data [34].With suitable modern imaging software it is

possible to convert this information into easyto understand 3D visualizations of elementaldistributions within a given sample volume nearthe original surface [76]. When the images arecreated by rastering the primary ion beam, thelateral resolution achievable is only limited by thebeam diameter, which can be significantly smallerthan 1 µm. The depth resolution is only limited bythe thickness of the ion beam mixing layer (at bestseveral nanometres).

In general, particles represent one of the leastsuited categories of samples for analysis by 3DSIMS. Not only do particles often exhibit a het-erogeneous composition, leading to a series ofartifacts, but also their morphologies often makemeaningful 3D SIMS impossible without specialsample preparation and methods of corrections(e.g. eliminating effects of varying sputter ratesin changing matrix compositions) [76]. Anotherimportant requirement for SIMS measurements,a flat sample surface at the beginning of anal-ysis, is generally not complied with. However,in favourable cases, the 3D SIMS technique canbe successfully applied to particle characteriza-tion [1]. Although SIMS primarily gives informa-tion on elemental (and isotopic) distributions theseare very valuable for a further interpretation withregard to speciation.

5 CONCLUSION

The main incentives to this chapter were the fol-lowing. An overview of modern methods of sin-gle particle characterization and speciation was, ofcourse, the first aim. However, there is addition-ally a wealth of topochemical methods of particlecharacterization which can be used in combinationto yield a profound means of particle characteriza-tion and simultaneously also give valuable answerswith respect to speciation. Crystal structures ofcrystalline aerosol particles may also contributesignificantly to the toxicity of certain aerosolcompounds such as nickel sulfides. Therefore acorresponding TEM inspection seems essential,since only electron diffraction patterns will yieldthe necessary information on crystal structures

REFERENCES 523

with the necessary lateral and depth resolutionuseful for single particle analysis. It is the mul-timethod approach outlined here which producesvaluable synergistic effects in an in-depth interpre-tation of relevant results. Of course, it is neitherpossible nor meaningful to use all the methodsdescribed together to provide solutions to questionsrelated to particle characterization. However, cer-tain methodological combinations have proved tobe very successful in our experience, e.g. the com-bination of the bulk analytical method TXRF withthe topochemical methods of HR-SEM, EPMA andTEM. The procedure developed by us for semi-quantitative particle characterization by EPMA hasrevealed complex compositions of particles col-lected in the metallurgical industries [21, 24, 30,77]. This has called for further characterizationof the phase composition and crystal structures ofsuch particles by TEM-EDX-SAED-EELS. Somerelevant examples have been presented. SIMS, incontrast, can reveal the 3D compositional structureof larger particles with certain geometries [76] orthe isotopic pattern of trace elements of particles,e.g. to distinguish between terrestrial and cosmicparticles in near-Earth space – an important issuein modern space technology [14–16].

It is to be expected that certain combina-tions of the methods of solid-state characterizationaddressed here will lead to many new insightsinto the very diverse fields of science and technol-ogy for which particle characterization was shownhere to be of relevance. The field of instrumentalsolid state particle characterization including spe-ciation is a truly interdisciplinary one since suchinstrumentation is generally found in institutionsof material science and/or solid state chemistryand physics.

6 ACKNOWLEDGEMENTS

The results presented here were obtained onthe instrumentation of various working groups,mainly at the Institute of Material Science ofthe Darmstadt University of Technology but alsoat other institutions. I would, therefore, like tothank all our respective collaborative partnersfor excellent cooperation. Table 6.1.2 gives an

Table 6.1.2. Collaborative partners in single particlecharacterization.

Dr. FrankStadermann

McDonnel Center for the SpaceSciences, Washington University,St. Louis, MO, USA

Prof. Dr. Stephan Dept. of Environmental Mineralogy,Weinbruch Institute of Material- and

Dr. Ing. Martin Geosciences, DarmstadtEbert University of Technology,

Germany

Prof. Dr. G. Dept. of Combustion Engines,Hohenberg Institute of Mechanical Engineering,

Dr. Ing. Michael Darmstadt University ofWentzel Technology, Germany

Dipl.-Ing.BarbaraZelenka

Dr. Peter Dept. of Chemical Analytics,Hoffmann Institute of Material- and

Dipl. Ing. Geosciences, DarmstadtBurkardHoflich

University of Technology,Germany

Prof. Dr. YngvarThomassen,

National Institute of OccupationalHealth, Oslo, Norway

Asbjorn Skogstad

Dr. V. P. Kola Research Laboratory forTchashchin Occupational Health, Kirovsk,

Dr. M.Tchashchin

Russia

Prof. Dr. EvertNieboer

Dept. of Biochemistry andOccupational Health, McMasterUniversity, Hamilton, Ontario,Canada and Institute ofCommunity Medicine, Universityof Tromso, Tromso, Norway

Dr. G. Helas Dept. of Biogeochemistry, MaxPlanck Institute for Chemistry,Mainz, Germany

Dr. Ing. W.Schulmeyer

Plansee AG, Dept. of RefractoryMetals, Reutte, Tirol, Austria

overview on our numerous collaborative partnersin particle characterization.

7 REFERENCES

1. Ortner, H. M., Hoffmann, P., Stadermann, F. J., Wein-bruch, S. and Wentzel, M., Analyst , 123, 833 (1998).

2. Pruppacher, H. R. and Klett, J. D., Microphysics ofClouds and Precipitation , Reidel, Dordrecht, 1978.

3. Andreae, M. O., in World Survey of Climatology , Vol. 16:Future Climates of the World: A Modelling Perspective,Henderson-Sellers, A. (Ed.), Elsevier, Amsterdam, 1995,p. 347.

4. Willeke, K. and Baron, P. A. (Eds), Aerosol Measure-ment , Van Nostrand Reinhold, New York, 1993.


5. Ebert, M., Dahmen, J., Hoffmann, P. and Ortner, H. M.,Spectrochim. Acta, Part B , 52, 967 (1997).

6. Ebert, M., Hoffmann, P., Ortner, H. M. and Dahmen, J.,GIT Fachz. Lab., 10/96, 982 (1996).

7. Ortner, H. M., J. Environ. Monit., 1, 273 (1999).8. Grasserbauer, M. and Werner, H. W. (Eds), Analysis of

Microelectronic Materials and Devices , John Wiley &Sons, Ltd, Chichester, 1991.

9. Ortner, H. M. and Wilhartitz, P., Fresenius’ J. Anal.Chem., 337, 686, (1990).

10. Ortner, H. M., The Influence of Trace Elements on theProperties of Hard Metals , COST 503, Material SciencesSeries: Powder Metallurgy – Powder Based Materials,Vol. IV, Valente, I. (Ed.), European Communities, Brus-sels, 1997, p. 70.

11. Hunt, T. M., Handbook of Wear Debris Analysis andParticle Detection in Liquids , Elsevier, London, 1993.

12. Ortner, H. M., Petter, H. and Birzer, W., in Proceedingsof the First Plasma-Technique Symposium , Vol. 2, Esch-nauer, H., Huber, P., Nicoll, A. R. and Sandmeier, S.(Eds), Plasma Technik, Switzerland, 1988, p. 237.

13. Nietmantsverdriet, J. W., Spectroscopy in Catalysis ,VCH, Weinheim, 1993.

14. Stadermann, F. J. and Jessberger, E. K., in Proceedingsof the European Conference on Space Debris, ESA SD-01, 1993, p. 185.

15. Stadermann, F. J., Measurement of isotopic and elementalabundancies in single interplanetary dust particles bysecondary ion mass spectrometry, Ph.D. Dissertation,University of Heidelberg, 1990 (in German).

16. McDonnel, A. M. (Ed.), Cosmic Dust , John Wiley &Sons, Ltd, Chichester, 1978.

17. Allamandola, L. J., Sandford, S. A. and Wopenka, B.,Science, 237, 56 (1987).

18. Stadermann, F. J. and Olinger, C. T., Meteoritics , 27,291 (1992).

19. Walls, H. J., Forensic Science, 2nd edn, Sweet andMaxwell, London, 1968.

20. Rampino, M. R., in World Survey of Climatology , Vol. 16:Future Climates of the World: a Modelling Perspective,Henderson-Sellers, A. (Ed.), Elsevier, Amsterdam, 1995,p. 95.

21. Hoflich, B. L. W. and Ortner, H. M., Final Report: Char-acterization of Single, Respirable Particles In AluminiumPotrooms at Norsk Hydro Aluminum, 2001.

22. Wilson, R. and Spengler, J. D. (Eds) Particles in Our Air:Concentrations and Health Effects , Harvard School ofPublic Health, 1996.

23. Siegel, R. W., in Material Science and Technology , Vol.15: Processing of Metals and Alloys , Cahn, R. W. (Ed.),VCH, Weinheim, 1991, p. 583.

24. Morlang, A., Phase analysis of single particles by electrondiffraction in SEM and TEM, Diploma Thesis, Institute ofMaterial Sciences, Darmstadt University of Technology,2000, (in German).

25. Sundermann, F. W., Jr., in Nickel in the Human Environ-ment , Sundermann, F. W., Jr (Ed.), International Agencyfor Research on Cancer, Lyon, 1984, p. 127.

26. Jaenicke, R., in Atmospharische Spurenstoffe, Jaenicke, R.(Ed.), VCH, Weinheim, 1987, p. 321.

27. Harrison, R. M. and Van Grieken, R. (Eds.), AtmosphericParticles , IUPAC Series on Analytical and PhysicalChemistry of Environmental Systems, Vol. 5, John Wiley& Sons, Ltd, Chichester, 1998.

28. Cahier, H., Carbonaceous combustion aerosols, in Atmo-spheric Particles , Harrison, R. M. and Van Grieken, R.(Eds), IUPAC Series on Analytical and Physical Chemistryof Environmental Systems , Vol. 5, John Wiley & Sons,Ltd, Chichester, 1998, p. 296.

29. Koropchak, J. A., Sadain, S., Yang, X., Magnusson, L.-E., Heybroek, M., Anisimov, M. and Kaufman, S. L.,Anal. Chem., 71, 386A (1999).

30. Hoflich, B. L. W., Wentzel, M., Ortner, H. M., Wein-bruch, S., Skogstad, A., Hetland, S., Thomassen, Y., Cha-schin, V. P. and Nieboer, E., J. Environ. Monit., 2, 213(2000).

31. Model 3089 Nanometer Aerosol Sampler, AdvanceProduct Information, TSI, St. Paul, MN, USA 2000.

32. Dixkens, J. and Fissan, H., Aerosol Sci. Technol., 30, 438(1999).

33. Hincks, W., Aerosol Technology , John Wiley & Sons,Inc., New York, 1982.

34. Ortner, H. M., in Analytiker Taschenbuch , Vol. 19, Gunz-ler, H. et al. (Eds), Springer, Berlin, 1998, p. 217.

35. The XL ESEM series, Company Brochure of PhilipsElectron Optics, Eindhoven, Netherlands.

36. The XL 30 SFEG Scanning Electron Microscope, Tech-nical Data Sheet, Company Brochure of Philips ElectronOptics, Eindhoven, Netherlands.

37. Kollensperger, G., Friedbacher, G., Kotzick, R., Niess-ner, R. and Grasserbauer, M., Fresenius’ J. Anal. Chem.,364, 296 (1999).

38. Randke, V., Electron Backscatter Diffraction , GuideBook Series, Oxford Instruments, High Wycombe, Eng-land, 1996.

39. Weinbruch, S., Wentzel., M., Kluckner, M., Hoffmann, P.and Ortner, H. M., Mikrochim. Acta, 125, 137 (1997).

40. Hoffmann, P., Dedik, A. N., Ensling, J., Weber, S., Sin-ner, T., Gutlich, P. and Ortner, H. M., J. Aerosol Sci., 27,325 (1996).

41. Armstrong, J. T., in Electron Probe Quantitation , Hein-rich, K. F. J. and Newbury, D. E. (Eds), Plenum, NewYork, 1991, p. 261.

42. Armstrong, J. T. and Buseck, P. R., X-ray Spectrom., 14,172 (1985).

43. Kluckner, M., Brandl, O., Weinbruch, S., Stadermann,F. J. and Ortner, H. M., Mikrochim. Acta, 125, 129(1997).

44. Jambers, W., De Bock, L. and Van Grieken, R., Analyst ,120, 681 (1995).

45. Chul-Un Ro, Osan, J. and Van Grieken, R., Anal. Chem.,71, 1521 (1999).

REFERENCES 525

46. Smith, R.-U. (Ed.), Handbook of Environmental Analysis ,2nd edn, AOAC, Mc Lean, VA, 1995.

47. Klockenkamper, R., in Total Reflection X-ray Fluores-cence Analysis , Winefordner, J. D. (Ed.), Chemical Anal-ysis Series , Vol. 140, John Wiley & Sons, Inc., New York,1997, p. 151.

48. Xhoffer, C., Wouters, L., Artaxo, P., Van Put, A. and VanGrieken, R., in Environmental Particles I , Vol. 1, IUPACSeries on Environmental Analytical Chemistry and Phys-ical Chemistry, Buffle, J. and van Leeuwen, H. P. (Eds),Lewis, Boca Raton, FL, 1992, p. 107.

49. Meisel, A., Leonhardt, G. and Szargan, R., X-ray Spectraand Chemical Binding , Springer Series in ChemicalPhysics, Vol. 37, Springer, Berlin, 1989.

50. Hofer, H. E., Brey, G. P., Schulz-Dobrick, B. and Ober-hausli, R., Eur. J. Mineral., 6, 407 (1994).

51. Hoflich, B. L. W., Rausch, A., Weinbruch, S., Helas, G.,Ortner, H. M. and Ebert, M., Speciation of sulfur inindividual aerosol particles by precisive measurement ofchemical shift of the SKα X-ray line in the electronmicroprobes, submitted to Mikrochim. Acta.

52. Miehe, G., in Referate, 3. Jahrestagung der DeutschenGesellschaft fur Kristallographie, Oldenbourg, Munich,1995, p. 51.

53. Zaefferer, S., J. Appl. Crystallogr., 33, 10 (2000).54. Bischoff, E., Campbell, G. H. and Ruhle, M., Fresenius’

J. Anal. Chem., 337, 469 (1990).55. Hofer, T. and Warbichler, P., Ultramicroscopy , 63, 21

(1996).56. Hofer, T., Warbichler, P. and Grogger, W., Ultrami-

croscopy 59, 31 (1995).57. Kriwanek, O.-L., Gubbens, A. J., Bellby, N. and

Meyer, C. E., Microsc. Microanal. Microstruct., 3, 187(1992).

58. Grogger, W., Hofer, T. and Kothleitner, G., Mikrochim.Acta , 125, 13 (1997).

59. Grasserbauer, M., Dudek, H. J. and Ebel, M. F., Ange-wandte Oberflachenanalyse, Springer, Berlin, 1985.

60. Weber, S., Hoffmann, P., Ensling, J., Dedik, A. N., Wein-bruch, S., Miehe, G., Gutlich, P. and Ortner, H. M., J.Aerosol Sci ., 31, 987 (2000).

61. Sedlak, D. L. and Hoigne, J., J. Atmos. Environ., Part A,27, 2173, (1993).

62. Seinfeld, J. H., Atmospheric Chemistry of Air Pollution ,John Wiley & Sons, Inc., New York, 1986.

63. Hoffmann, P., Dedik, A. N., Deutsch, F., Ebert, M.,Hein, M., Hoffmann, H., Lieser, K. H., Ortner, H. M.,Schwarz, M., Sinner, T., Weber, S., Weidenauer, M. and

Weinbruch, S., in Dynamics and Chemistry of Hydro-meteors , Jaenicke, R. (Ed.), Deutsche Forschungsgemein-schaft, Wiley-VCH, Weinheim, 2001, p. 440.

64. Gutlich, P., Linck, R. and Trautwein, A., MossbauerSpectroscopy and Transition Metal Chemistry , Springer,Berlin, 1978.

65. Janssens, K., Vince, L., Wie, F., Proost, K., Vekemans, B.,Vittiglio, G., Yan, Y. and Falkenberg, G., Feasibility of(trace-level) micro-XANES at Beamline L, in HASYLABJahresbericht 1999 , Materlik, G. et al. (Eds), Hamburg,1999.

66. Salbu, B., Janssens, K., Irekling, T., Simionovici, A.,Drakopoulos, M., Raven, C., Snigireva, I., Snigirev, A.,Lind, O. C., Oughton, D. H., Adams, F. and Kashparov,V. A., Micro XANES for characterization of fuel par-ticles, in ESRF Highlights 1999 , European SynchrotronRadiation Facility 2000, Grenoble, France, p. 24.

67. Janssens, K. H. A. (Ed.), Microscopic X-ray Fluores-cence Analysis , John Wiley & Sons, Ltd, Chichester,2000, p. 370.

68. Fariborz Goodarzi and Huggins, F. E., J. Environ. Monit.,3, 1 (2001).

69. Ortner, H. M. and Wilhartitz, P., Mikrochim. Acta II , 177(1991).

70. Schuhmacher, M., Rasser, B., De Chambost, E., Hill-ion, F., Mootz, Th. and Migeon, H. N., Fresenius’ J.Anal. Chem., 365, 12 (1999).

71. Bernatowicz, T. J. and Walker, R. M., Phys. Today , 50,25 (1997).

72. Johnston, M. V. and Wexler, A. S., Anal. Chem., 67,721A (1995).

73. Injuk, J., Breitenbach, L., Van Grieken, R. andWatjen, U., Mikrochim. Acta, 114/115, 313 (1994).

74. Injuk, J. and Van Grieken, R., Spectrochim. Acta, Part B ,50, 1787 (1995).

75. Gara, S., Stingeder, G., Tian, C., Hutter, H., Fuhrer, H.and Grasserbauer, M., in Secondary Ion Mass Spectrom-etry: SIMS VIII , Benninghoven, A., Janssen, K. T. F.,Thumpner, J. and Werner, H. W. (Eds), John Wiley &Sons, Ltd, Chichester, 1992, p. 537.

76. Stadermann, F. J. and Ortner, H. M., in Secondary IonMass Spectrometry: SIMS X , Benninghoven, A., Hagen-hoff, B. and Werner, H. W. (Eds), John Wiley & Sons,Ltd, Chichester, 1997, p. 325.

77. Gunst, S., Weinbruch, S., Wentzel, M., Ortner, H. M.,Skogstad, A., Hetland, S. and Thomassen, Y., J. Environ.Monit., 2, 62 (2000).

6.2 Direct Speciation of Solids: X-ray AbsorptionFine Structure Spectroscopy for Species Analysisin Solid Samples

Edmund WelterHamburger Synchrotronstrahlungslabor (HASYLAB) at Deutsches Elektronen-Synchrotron (DESY),Hamburg, Germany

1 Introduction . . . . . . . . . . . . . . . . . . . . . . 5262 Basic XAFS Theory, the Reason Why

XAFS Spectroscopy is Element andSpecies Selective . . . . . . . . . . . . . . . . . . 527

3 EXAFS Data Evaluation . . . . . . . . . . . . . 5294 The XAFS Experiment . . . . . . . . . . . . . . 530

4.1 Transmission . . . . . . . . . . . . . . . . . 5304.2 Fluorescence yield . . . . . . . . . . . . . 5324.3 Sample preparation and handling . . . 535

5 Examples of Contemporary Use inSpeciation Analysis . . . . . . . . . . . . . . . . . 535

5.1 Environmental analysis . . . . . . . . . . 5355.2 Catalysis research . . . . . . . . . . . . . . 538

6 Limitations . . . . . . . . . . . . . . . . . . . . . . . 5397 µ-XANES . . . . . . . . . . . . . . . . . . . . . . . 540

7.1 Mapping technique . . . . . . . . . . . . . 5417.2 Single particle technique . . . . . . . . . 541

8 X-ray Raman Spectroscopy . . . . . . . . . . . 5429 Summary and Outlook . . . . . . . . . . . . . . 544

10 References . . . . . . . . . . . . . . . . . . . . . . . 545

1 INTRODUCTION

Many of the known analytical methods for ele-mental speciation analysis are restricted to fluidsamples, for instance the whole range of chromato-graphic methods. Consequently these methods areapplicable to solid samples only if it is possibleto dissolve the analyte without destruction of thespecies information. This is often a tedious anderror-prone operation. An analytical method cir-cumventing this problem by enabling the directspecies determination in solids is thus highly desir-able. Such a technique is provided by X-rayabsorption fine structure (XAFS) spectroscopy.

Another positive property of XAFS spec-troscopy is that it yields the species informationwithout being significantly disturbed by the matrix

due to its high degree of element selectivity. Thisis in contrast to other methods that enable thedetermination of the chemical speciation only in apure or at least highly concentrated solid sample.Examples for the latter methods are X-ray diffrac-tion (XRD) and IR spectroscopy. Their applica-bility is limited to more or less pure compoundsbecause they cannot differentiate between analyti-cal signals from the analyte and the matrix, a factthat makes it impossible to identify single speciesof the analyte in the sample.

The basic theoretical background, the experi-mental implementation and further enhancementsof the applicability by introduction of newexperimental techniques such as measuring XAFSspectra with µm spatial resolution and X-rayRaman spectroscopy will be discussed in the

BASIC XAFS THEORY 527

following. Some examples of contemporary use inspecies analysis will be given to demonstrate thepotential of the method.

2 BASIC XAFS THEORY, THEREASON WHY XAFS SPECTROSCOPYIS ELEMENT AND SPECIESSELECTIVE

XAFS theory is discussed in detail in severalmonographs [e.g. 1–3]. In the following only thebasic elements required to understand the analyti-cally interesting element and species selectivity ofXAFS spectroscopy can be discussed.

As in other absorption spectroscopical meth-ods, for example in the UV/Vis or IR region, inX-ray absorption spectroscopy (XAS) the depen-dence of the absorption coefficient µ on thewavelength of the incoming X-ray beam is mea-sured (equation 6.2.1). In X-ray spectroscopy theabscissa is usually scaled in energy units (eV), notwavelength or frequency.

µd = lnI1

I2(6.2.1)

with µ being the absorption coefficient in units ofcm−1, d the sample thickness in cm and I1, I2

the photon beam intensities in front of and behindthe sample.

Figure 6.2.1 shows a simplified X-ray absorp-tion spectrum of a sample that contains Fe, Cuand a small amount of Zn. Unlike the absorptionspectra in UV/VIS or IR spectroscopy the X-rayabsorption spectrum shows no peaks but edgesat which the absorption coefficient µ increasesabruptly. The reason for the difference betweenabsorption spectra in the UV/Vis and in the X-ray region is that in XAS an electron is excitedto the continuum whereas the transitions in opticalUV spectroscopy take place between energeticallywell defined orbitals, see Figure 6.2.2.

The edge position corresponds to the energy thatis necessary to lift a core electron from an innershell to the continuum. This energy is specific forevery element, thus making XAFS spectroscopy anelement-selective method. The edges that are most

Fe K

Cu K Zn K

Zn L3

0 2000 4000 6000 8000 10000 12000

X-ray photon energy/eV

100

1000

10 000

log

µ/cm

−1

Figure 6.2.1. Absorbance coefficient (µ) over X-ray photonenergy calculated for a sample with the hypothetical sumformula Fe2Cu20Zn (spectrum calculated by use of XOPprogram package) [4].

Photon

Electron

Unoccupied orbital

UV/Vis photon

X-ray photon

Photoelectron

Figure 6.2.2. Schematic visualization of the reasons for thedifferent shape of UV/Vis (peaks) and X-ray absorption spectra(edges). UV/Vis spectroscopy probes the small but discreteenergy differences between the highest occupied and the lowestunoccupied atomic or molecular orbitals. Absorption of anX-ray photon lifts a core shell electron to the continuum.

often used are the K-edges. They result from theexcitation of a K-shell electron to the continuum.For elements of higher Z the L-edges, normally theL3-edge, are also used for XAFS spectroscopy. K-edge positions range from some 100 eV (K-edgespectra of low Z elements such as C, N and O)to ∼115 keV (K-edge of uranium). The distancesbetween edges of consecutive elements are of theorder of some 100 eV for the 3d transition metalsand ∼3000 eV for the K-edges of the actinides.

Small perturbations of the core ground statesdue to the redox state lead to a ‘chemical shift’of the edge position depending on the oxidationstate of the absorbing atom. This chemical shift

528 DIRECT SPECIATION OF SOLIDS

Cr2O3CrNa2CrO4

5940 5960 5980 6000 6020 6040


0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

md n

orm

Figure 6.2.3. XANES spectra of three different Cr compoundswith different Cr oxidation states, showing the chemical shiftof the edge position. Note the large pre-edge peak of the Cr(VI)compound; this peak can be used to identify and quantifyCr(VI) in mixtures.

is normally of the order of some eV. This isexemplified in Figure 6.2.3, which shows theedge region of the XAS of three chromiumcompounds (Cr, Cr2O3 and Na2CrO4) that containCr in three different oxidation states. In generalthe edge position of the more highly oxidizedcompounds is shifted to higher energies, becausethe core electrons are bound more strongly inthese compounds. However, Figure 6.2.3 not onlyshows the small shifts in the edge position causedby the oxidation state of the absorbing element.It can also be seen that the region above theedge is not as smooth as it is in the simplifiedspectrum in Figure 6.2.2 and instead shows wellvisible oscillations of the absorption coefficientabove the edge. These oscillations can be found upto 1000–2000 eV and even higher above the edgedepending on the element, the chemical compoundof the element and experimental conditions. XAFSspectroscopy is based exactly on these oscillationsof the absorption coefficient.

The fine structure of the absorption edge wasfirst detected in 1920, but only with the inventionof synchrotron radiation as a source of veryintense X-rays to XAFS spectroscopy in 1974did it become possible to measure XAFS spectrawithin a reasonable time and in diluted samples.The region above the edge is normally dividedinto two subregions, the first 50–100 eV abovethe edge are called X-ray absorption near edge

structure (XANES), the region above the XANESregion is called extended X-ray absorption finestructure (EXAFS).

A first idea for the understanding of the occur-rence of the fine structure can be drawn from thecomparison of XAS spectra from a monoatomicgas such as Kr and a diatomic gas such asBr2 [5]. The spectrum from the monoatomic gasshows only some structures in the actual edgeand absolutely no EXAFS oscillations, whereasthe spectrum from a diatomic gas shows bothsome structures in the actual edge and oscillationsin the EXAFS region. This finding indicates thatthe EXAFS oscillations might result from effectscaused by neighboring atoms whereas the struc-ture in the edge is caused by ‘internal’ effects inthe absorber atom. The latter is in fact a resultof transitions of a core shell electron to higherunoccupied orbitals, whereas the oscillations in theEXAFS region are caused by a different effect thatwill be qualitatively discussed in the following.

The fine structure, XANES as well as EXAFS,contains much more information about the sam-ple than simply detecting the oxidation state ofthe absorber atom. It contains information aboutnature, number and distance of the next neigh-boring atoms. Figure 6.2.4 shows the XAS spectraof four different Pb compounds. It is clearly seenthat the fine structure above the edge is uniquefor every single Pb compound. The reason for theoccurrence of the fine structure oscillations and

PbSO4PbCO3basic-PbCO3PbOyellow

X-ray photon energy

13000 13100 13200

md (

norm

)

0

20

40

60

80

100

120

Figure 6.2.4. Normalized near-edge XAFS spectra of four inor-ganic Pb(II) compounds; spectra like these are used as referencespectra for fingerprint analysis of Pb-containing samples.

EXAFS DATA EVALUATION 529

Sum

Distance

Atom ABS1 BS2

D1 D2

Figure 6.2.5. Qualitative explanation for the occurrence of theXAFS oscillations. Three atoms are shown, the absorbing atomA and two backscattering neighbor atoms (BS1 and BS2) at twodifferent distances D1 and D2 with D2 being larger than D1and (D2 – D1) equal to half the wavelength of the outgoingphotoelectron wave (note that the sinusoidal curves neglectthe influence of the sample specific phase shifts during thescattering process).

their species selectivity can be understood quali-tatively by looking at Figure 6.2.5.

If the energy of the absorbed X-ray photon(Ehν) is higher than the energy (E0) that isnecessary to lift a core shell electron to thecontinuum, an electron from an inner shell is liftedto the continuum. This photoelectron leaves theatom with a kinetic energy (E) given by:

E = Ehν − E0 (6.2.2)

To understand the occurrence of the fine structureoscillations it is necessary to look at the waveproperties of the photoelectron. The wavelength(λ) of the outgoing spherical photoelectron wave islinked to the energy of the absorbed X-ray photon(Ehν) by:

λ = 2π

k(6.2.3a)

where

k =√

2m

h2 (Ehν − E0) (6.2.3b)

with m = mass of the electron, Ehν and E0

in joules.

Equation (6.2.2) shows that the energy of thephotoelectron increases with increasing energyof the incident X-ray photons during the XAFSscan. This leads to a variation of the wavelengthof the photoelectron wave (equations 6.2.3a andb). The outgoing photoelectron wave is partiallyreflected by the neighboring atoms. The inter-ference between outgoing and reflected electronwaves, which can either be constructive or destruc-tive at the origin, causes the oscillations of theabsorption cross-section, which is measured as afunction of the energy of the incoming photonbeam. A measurable reflection of the outgoingphotoelectron wave is caused only by the near-est neighbor atoms. That means the fine structureprobes the immediate chemical environment, theshort range order (typically 2–3 nearest neigh-bors), of the absorbing atom species and givesinformation about the nature, distance and numberof the nearest neighbors of the absorber atom.

The restriction to the determination of the shortrange order means that XAFS spectroscopy isalso applicable in amorphous samples, whereasXRD probes the average position on lattices andis therefore normally performed on crystallinesolid samples.

3 EXAFS DATA EVALUATION

The process most frequently used to evaluate theinformation contained in the EXAFS spectra is amultistep process, which is based on the ‘shortrange, single electron, single scattering theory’.The model presented before is the basis of thistheory. Major steps of the evaluation procedure arein the order they have to be done [6]:

• Background correction (subtraction of the back-ground absorption that is caused by Rayleighscattering, Compton scattering and by precedingabsorption edges).

• Normalization of the edge jump to 1.• Conversion of the initial energy axis to the pho-

toelectron wave vector scale (‘k-scale’) usingequation (6.2.4).

k = 0.152√

E − E0 (6.2.4)


0

20

40

60

80

100

120

12950 13000 13050 13100 13150

md(

norm

)/A

U


Pure PbCO3

Pure PbO

Figure 6.2.6. Linear combinations of normalized PbO and PbCO3 XAFS spectra. The spectrum of pure PbO is shown in frontthat of pure PbCO3 as the last spectrum. Linear combinations like these can be compared with experimental spectra from samplescontaining mixtures of compounds to determine the relative amounts of the constituents.

where E and E0 are given in eV. Equation (6.2.4)can be calculated from equation (6.2.3b) byemploying the values of m and h and convertingthe energy units from joules to electronvolts.

• Fourier transformation of the spectrum.• Fitting (modeling) of theoretical spectra by use

of several computer codes.

All steps are performed by use of special computerprograms, see for instance refs 7 and 8.

This classical EXAFS data analysis is mostuseful, however, to analyze the spectra of purecompounds of the absorber atom or of single com-pounds of an element in a matrix consisting ofother elements. It is not applicable to mixtures ofseveral components of the same absorber atom.The calculated results like the measured spec-tra would be weighted linear combinations of therespective parameters of the different substances.This is exemplified in Figure 6.2.6, which showsthe spectra of pure PbCO3 and pure PbO fram-ing spectra of mixtures of the two compounds.The spectra shown were calculated using linearcombinations of the two pure substance spectra.Mixed spectra as they are often obtained fromnatural samples make the standard evaluation pro-cedure inapplicable for species analysis in thesetype of samples.

Alternative evaluation strategies for XAFSspectra which can be used in samples that con-tain more than one species, will be presented withthe examples for the use of XAFS spectroscopy inspecies analysis, especially in those from environ-mental species analysis.

4 THE XAFS EXPERIMENT

4.1 Transmission

The principal experimental setup that is needed tomeasure XAFS spectra is shown in Figure 6.2.7.It is basically the same as in any other absorption

Storagering

Sen

Sex

MC IC1 IC2 IC3

FD(optional)

Sample

~ 25–40 m

Source

Figure 6.2.7. Schematic drawing of a synchrotron radiationXAFS experiment in transmission (black sample) and in flu-orescence mode (grey sample and additional fluorescencedetector) at a bending magnet beamline (Sen, monochromatorentrance slit; Sex, monochromator exit slit; MC, monochro-mator, IC1 – IC3, ionization chambers 1–3; FD, fluorescencedetector). IC3 is used to measure an energy reference simulta-neously.

THE XAFS EXPERIMENT 531

spectroscopy method. One small difference is thatthe sample is usually positioned between a detectorthat measures the incoming intensity and a detectorthat measures the transmitted intensity. The basicelements of an XAFS experiment are the X-ray source, a monochromator, two detectors thatmeasure the incident and the transmitted intensityand some kind of sample holder device to handlethe sample itself.

In almost all XAFS experiments synchrotronradiation is used as the X-ray source, althoughthere are concepts for laboratory XAFS spec-troscopy equipment using X-ray tubes. The mainreasons for the use of synchrotron radiation are thehigher flux and the easy tunability of these sources.The small divergence of the synchrotron beam addsto the advantages for spectroscopy.

Synchrotron radiation is emitted during theacceleration of charged particles such as electronsor positrons. The first sources of synchrotronradiation were the bending magnets that are usedto force the charged particles that are acceleratedin a synchrotron ring into the orbital track. Thesynchrotron light from bending magnets is emittedin a narrow cone tangentially to the circumferenceof the synchrotron ring. The dipole radiationemitted by a synchrotron source covers a wideenergy (or wavelength) range, starting in theinfrared and going up to the hard X-ray and evenγ -ray region.

The demand for higher flux led to the inventionof special devices for the production of even moreintense light. These insertion devices are called‘wigglers’ and ‘undulators’. Wigglers increase theflux by adding the dipole radiation of severalalternating magnets which force the charged parti-cles into a sinusoidal trajectory. At undulators theemission of photons at every dipole occurs – incontrast to wigglers where the emission of syn-chrotron radiation happens independently at everydipole – with a fixed phase relation. The emissionspectrum of wigglers shows a broad energy spreadlike the emission spectrum of a bending magnet.In the case of an undulator the emission spectrumis overlaid by several small emission lines withvery high intensity. This emission characteristicmakes undulators less suited for EXAFS scans,

which cover an energy range much broader thanthe emission lines. If one accepts the higher exper-imental effort associated with the use of undulatorsthey are nevertheless very valuable sources forexperiments which need an extremely high photonflux, for instance the registration of XAFS spectrafrom highly diluted samples.

However in many cases the flux of the sourceis not the limiting factor for an XAFS experiment.That is the reason why even today many XAFSexperiments are still performed at bending magnetbeamlines. They simply offer a sufficiently highintensity to measure spectra with good signal/noiseratio from pure or moderately diluted samples.

An XAFS scan requires monochromatic (mono-energetic) X-ray light, which must be scanableover a range of 1000–2000 eV. The monochro-matic X-ray light is produced by the use of crystalmonochromators. The two-crystal design shown inFigure 6.2.8 is most widespread. The major advan-tage of this design compared to a single-crystaldesign is that the monochromatic beam leaves themonochromator horizontally, with only a smallvertical offset to the white beam. The material inmost widespread use for the analyzer crystals is Si,because crystals of the required size and purity areavailable relatively cheaply. Furthermore, Si crys-tals have appropriate plane distances and a goodreflectivity for measurements in the X-ray region.Crystals for different energy regions and spectralresolutions can be produced by cutting monocrys-talline silicon along certain crystal planes, thusproducing crystals with different distances (d) ofthe crystal planes. The transmitted wavelengths canbe calculated using Bragg’s law given by

nλ = 2d sin θ (6.2.5)

with n = 1, 2, 3, . . ., λ = wavelength, d = crystallattice spacing and θ = incidence angle.

Transmitted radiation with n > 1 is called‘higher harmonic’; this higher harmonic radiationmust be excluded, because it produces a largebackground. There are two common techniquesto achieve this. First, the use of mirrors. Thisstrategy makes use of the fact that the angle underwhich total reflection occurs on a surface rapidlydecreases with wavelength, so higher harmonics


Higher energy (shorter wavelength => smaller Bragg angle)

Fixed exit

Lower energy (longer wavelength => bigger Bragg angle)

Figure 6.2.8. Principal setup of the two double crystal monochromator designs in most widespread use for XAFS spectroscopy.Energy scans are performed by rotating the crystals around axis 1, thus changing the angle θ (Bragg angle) under which theincident beam hits the first crystal. The fixed exit setup (left) needs two goniometers for the parallel rotation of both crystals anda linear movement for the translation of the second crystal. The alternative design (right) needs only one goniometer on whichboth crystals can be mounted but makes it necessary to move the sample and the exit slit accordingly.

with half the wavelength can be excluded almostcompletely using appropriate glancing angles. Thesecond technique is to detune the double crystalmonochromator slightly, which means tilting oneof the crystals slightly, so that both crystals areno longer perfectly parallel. This works becausethe bandwidth of the transmitted radiation is muchgreater for the first harmonic than for that ofthe higher harmonics. Furthermore, the use ofcertain crystal planes such as Si(111), wherethe second order reflection is prohibited, furtherreduces higher harmonic contents [9].

The energy resolution of a crystal monochro-mator is determined by two main factors. Firstlythe width of the reflected Bragg peak, which is afunction of the chosen crystal material, of the cho-sen reflex and of the quality of the crystal. Thesecond important factor is the divergence of theincident beam, because it determines the range ofincident angles on the crystal. The divergence oflaboratory sources such as X-ray tubes is normallydecreased by use of collimators. Because of thesmall divergence of the synchrotron beam this isnot necessary for XAFS measurements for whichan energy resolution of 0.5 to some eV is suffi-cient. A higher spectral resolution would on the

contrary be counterproductive, because the higherthe resolution of the monochromator is, the loweris the transmitted flux and consequently the statis-tical quality of the data.

The spatial emission properties of synchrotronradiation also make it possible to work withoutfocusing optical elements. Without focusing opticalelements, a typical size of the beam on the samplewould be 10 mm horizontal and 1 mm vertical. Thebeam shape is finally defined by slit systems, whichalso have the function to minimize stray light.The beam size of an unfocused beamline can bereduced using slits, but again only for the priceof statistical quality. Focussing mirrors (toroidalor elliptical) are used in many XAFS beamlinesto increase the flux density on the sample andto cut off higher harmonic (n in equation (6.2.5)>1) radiation.

4.2 Fluorescence yield

XAFS measurements in transmission work well inpure or highly concentrated samples. With increas-ing dilution the problem that the small edge ofinterest is located on a high background absorp-tion becomes more and more important. This can

THE XAFS EXPERIMENT 533

be seen in Figure 6.2.1 where a small Zn K-edge ispositioned on the much larger background whichis mainly caused by Cu and Fe.

Therefore, diluted samples cannot be measuredwith the basic transmission mode XAFS exper-iment. Fortunately there are alternatives to thedirect measurement of the absorption. One ana-lytical signal, whose height is proportional toabsorbance but much less affected by the back-ground, is the fluorescence. Fluorescence photonsare generated during the relaxation process of theionized absorber atom. There are several possiblerelaxation processes but the emission of a fluo-rescence photon is the most important relaxationprocess for all elements with Z > 32 (Ge) [10] (forlighter elements the emission of an Auger electronis the preferred relaxation process). Fluorescencephotons are emitted when an electron from a highershell decays down to the hole in the K or L shellthat is left after the emission of the photoelectron.The energy of the fluorescence photon is equal tothe energy difference between the two orbitals, andis thus characteristic of the emitting element and

always lower than the energy of the absorbed inci-dent X-ray photon.

To measure XAFS spectra in fluorescencemode means that the intensity of the respectivefluorescence line, which is proportional to thenumber of absorbed photons, is measured whilethe energy of the incident beam is scannedover the edge and the EXAFS region of theabsorbing element as is done in transmission mode.Figure 6.2.9 shows the development of the Pb LX-ray fluorescence spectra during a scan of theincident energy between 12 800 and 13 200 eV.Clearly visible are the Pb Lα1,2 and the Pb Lβ2.15

emission lines that appear after the energy of theincident photons is high enough (13 035 eV) for theexcitation of an L shell electron to the continuum.The intensity of these lines shows the first XAFSoscillations. The third line, which is visible beforethe Pb emission lines appear and which disappearsunder the Pb Lβ2.15 line, is caused by elasticallyscattered incident photons.

The fluorescence technique obviously makes ahigher experimental effort necessary. The major

Figure 6.2.9. Development of the X-ray fluorescence spectra of a PbCO3 sample during an XAFS scan over the Pb L3-edge.The peaks are: Pb Lα1,2 (10 500 eV), Pb Lβ2,15 (12 622 eV) and elastically scattered photons very weak but visible before theedge (equal to incident beam energy). Spectra measured with a Ge detector optimized for high count rates.


addition to the standard transmission experimentis the detector system which consists of the actualdetector and the signal processing electronics. Ithas to fulfill two major, partially contradictory,requirements. Firstly it must offer a good energyresolution, in most real samples at least 1000 eV,preferably much better. Secondly the detector sys-tem must handle high count rates (>100 kcps),because otherwise the time needed for measure-ments with satisfactory signal/noise ratio would beunacceptably long.

Today energy-dispersive semiconductor detec-tors which are optimized for high count ratesare used routinely for this task. They make itpossible to measure at count rates of 100 kcpswith an energy resolution of 250–500 eV. Withthese detector systems the separation of the emis-sion lines from the different elements in the sam-ple (analyte/matrix) is normally large enough tomeasure the intensity of the fluorescence line ofthe analyte with often only negligible interfer-ence by the matrix-caused emission. Figure 6.2.10shows raw Cr K XAFS spectra measured on asoil sample that contained 200 µg g−1 Cr. While

the transmission spectrum shows a small edge sit-ting on a high background the fluorescence spec-trum shows a much better signal/background ratioand the first oscillations of the XANES region,although it is plagued by statistical noise.

It must be mentioned that fluorescence detectionof XAFS spectra suffers from some inherentproblems itself. The first problem is that ofstatistical noise. Because the count rates are muchsmaller than they are in transmission mode thedata often suffer from statistical noise. The reasonfor the low count rates is simply a geometricalproblem. A detector with a diameter of the entrancewindow of 2.5 cm, positioned in a distance of5 cm, covers only 1.6 % (∼0.2 sr) of the totalsolid angle. Approaches to solve this problem areobvious: the number of detectors can be increasedand the distance between sample and detector canbe decreased. Both approaches lead to an increaseof the solid angle covered and a correspondingincrease in the total count rate.

Another problem that must be considered iscalled ‘self-absorption’; it leads to a decrease inthe amplitude of the XAFS oscillations compared

5800 6000 6200

200

400

600

Flu

ores

cenc

e in

tens

ity/c

ps

X-ray photon energy

1.1

1.2

1.3

1.4

1.5

1.6

1.7

1.8

1.9

2.0

md

/AU

Figure 6.2.10. Comparison of a raw fluorescence and a raw transmission Cr XAFS spectrum measured simultaneously on a soilsample that contained 200 µg g−1 Cr. Note the absence of a pre-edge peak, indicating the absence of Cr(VI). The spectra weremeasured at a bending magnet beamline with a five pixel Ge-detector for fluorescence detection. They demonstrate the muchhigher signal/background ratio achieved with the fluorescence technique.

USE IN SPECIATION ANALYSIS 535

with those measured in transmission mode. Themagnitude of this effect depends among other fac-tors on the concentration of the absorber in thesample [11, 12]. Especially if the standards aremeasured in transmission mode or have stronglydiffering sample composition, the spectra mustbe corrected before they can be used in any ofthe XAFS spectra evaluation procedures. Sincethe factors that lead to the self-absorption phe-nomenon are known the spectra can be correctedmathematically [12].

4.3 Sample preparation and handling

Sample preparation is probably the one step ofan analytical method bearing the highest risk ofloss of species information. Notably the ability toinvestigate wet samples is of utmost importancefor species analysis. As briefly mentioned beforeit is one of the major advantages of XAFSspectroscopy that it can be performed in theoriginal sample. The XAFS experiment is inmost cases, dependent on the energy of theabsorption edge, very flexible with respect tosample properties such as size, form, humidity ortemperature. The XAFS spectra of most elementscan be measured in wet samples under ambientconditions. The critical factor is the energy of theabsorption edge; the lower the energy, the smalleris the penetration depth of the photons. At energiesbelow 5–6 keV measurements are better performedunder reduced pressure. This corresponds to theCr K edge (5989 eV). However, even in thesecases it is possible to find solutions, for instanceby building special sample holders that make itpossible to keep at least the actual sample underambient conditions.

Either way it is necessary to choose a suitablesample thickness. If using transmission modethe sample thickness should be chosen so thata µd value between ∼2.5 and 3.0 is achievedjust above the edge. If fluorescence detection isused the sample should – but does not necessarilyhas to be – thicker to obtain the highest possiblecount rates.

Nevertheless, one restriction which might inter-fere with species analysis should be kept in mind.

The samples must be homogenous (thickness, ele-mental and species composition) over the irradi-ated area. If this is not the case the sample has to behomogenized or the beam size has to be reduced toa length scale on which the sample is homogenous(see µ-XANES section). Homogenization of solidsamples is normally done by grinding. In prac-tice this can produce problems for species analysis,because the resulting enormous increase of surfacecan lead to artifacts, for example changes in oxi-dation state. In all cases where this risk exists caremust be taken to avoid these changes in samplecomposition, for instance by simply working undera protective atmosphere.

Once ground the sample might be mixedwith a light element bonding agent, for instancepolyethylene or boron nitride powder, and pressedinto pellets of appropriate thickness in a suitablepress. Another possibility is to spread a thin layerof the powder on an adhesive tape, or mix thepowder with glue.

The homogeneity of paste-like or liquid samplesis normally less problematic. These samples can behandled in small containers with thin entrance andexit windows made from a light element materialof appropriate mechanical strength.

5 EXAMPLES OF CONTEMPORARYUSE IN SPECIATION ANALYSIS

A large number of applications of XAFS spec-troscopy for analytical investigations that can besubsumed under the term ‘species analysis’ can befound in the literature. The chosen examples canby no means be a complete overview of this fieldof research. They are intended to give an idea ofthe many possible fields of application and to givesome indications of applications. The examples aretaken from two fields of research, environmentalanalysis and catalysis research.

5.1 Environmental analysis

Chemical speciation of metals plays an importantrole in the assessment of contaminated environ-mental compartments. Examples are the long-term


behavior of heavy metals in soils and sedimentsor the safe disposal of waste containing heavymetals or radioactive elements. The typical sam-ple material for these investigations is a soliddiluted sample.

Two different approaches that use XAFSspectroscopy can be distinguished: firstly thephysico/chemical characterization of single impor-tant components, for example the interaction ofmetals with humic acids or naturally occurringminerals; secondly the identification of chemicalcompounds in a (complex) natural sample. While itis often suitable to use the described classical pro-cedure of EXAFS spectra interpretation in the firstcase this is not a possible strategy in the secondcase. The reason is that in the latter case the sam-ples usually contain more or less complex mixturesof different compounds of the element of interest.

An example for the first case is the determina-tion of the binding form of actinides and fissionproducts in humic acids and certain clay and ironminerals. These investigations are aimed at thedevelopment of safe nuclear waste disposal andwill serve as a way to predict the behavior of theseelements in the environment if ever released [13].In one of these studies the binding form of uranylions (UO2+

2 ) to wood degradation products wasinvestigated. For this purpose known complexingwood degradation products were added to solutionscontaining uranyl ion under strict control of pH,temperature and redox potential. The binding formof the uranium to these substances was determinedby EXAFS spectroscopy and the results were com-pared with model calculations [14].

An example of an investigation of the chemicaltransformations that occur during the sorption ofmetals to clay minerals is a study of the sorptionof Ni to clay minerals and aluminum oxides [15].The kinetics of this process was investigated usingtime-resolved XAFS spectroscopy. On a time scaleof several hundred hours major changes of thechemical binding form could be detected and werefollowed by the changes of the fine structure in theXAFS spectra. In either case the classic strategywas chosen to evaluate the spectra from these well-defined synthetic samples.

In contrast to the investigations on syntheticmodel compounds it is often not possible tointerpret the EXAFS spectra in the usual waywhen investigating natural samples, because thesample contains more than one compound ofan element. Then one can try to investigate thespectra by use of a fingerprint approach. This isdone by measuring the spectra of pure referencecompounds which might be constituents of thesample and calculating the best fitting linearcombination of these reference compound spectra[16–19]. The working principle of this method canbe seen in Figure 6.2.11, which shows some of thespectra already presented in Figure 6.2.6 togetherwith a spectrum measured for a mixture of PbOand PbCO3 that contained 25 % of the lead as PbOand the remaining 75 % as PbCO3.

Often it is more useful to work with the nearedge region than the actual EXAFS region forthis type of analysis. This is because this regionof the spectrum exhibits the largest features. Themajor steps of such a fingerprint analysis areshown in the flow chart in Figure 6.2.12. Severalfactors influence the reliability of the resultsdrawn from the fingerprint method. In particular,when the actual edge region is used, a verycareful energy calibration that eliminates the smallshifts (typically 0.5–1.5 eV) of the measured edgeposition between two spectra is essential, becauseat this point µd changes drastically within smallenergy differences.

By far the greatest risk of erroneous resultsusing the fingerprint method arises from the selec-tion of inappropriate reference compounds. Toomany reference compounds make it possible tofind a good fit, although compounds contained arein fact missing. It is therefore mandatory to bewell informed about the chemical behavior of theanalyte and the possible reactions with matrix com-ponents, to minimize the number of reference com-pounds tested. Particularly problematic here arethe numerous poorly defined compounds that areformed by interaction of the analyte with matrixcomponents such as clay minerals and humic acids.These substances must be synthesized in the labo-ratory under strict control and adjustment to natural

USE IN SPECIATION ANALYSIS 537

0

20

40

60

80

100

13000

13050

13100

Figure 6.2.11. Working principle of a fingerprint analysis using linear combinations of XAFS spectra. The highlighted spectrumis measured on a sample that contained a mixture of PbCO3 (72.9 % of total Pb) and PbO (27.1 % of total Pb). The surroundingspectra are linear combinations with weights of 24 % or 26 % for PbO and 76 % or 74 % for PbCO3.

Data reduction:

•background correction

•energy correction

• multipoint energy calibration•normalization

• edge jump -> 100%

• pre-edge - polynomial

/ cubic-splines

• post-edge - polynomial

Refining:

• weighted sum of several fluorescence spectra• deglitching / smoothing

comparable spectra

Calculation of best fittinglinear combination of thechosen reference spectraby use of suited computeralgorithms

+ additional information likeSEM/EDX spectra or

historical informationis used to reduce the number

of spectra of referencecompounds

compounds&

amounts

Reference spectrainorganic compounds•metals sorbed to

•clay minerals•manganese oxides•iron oxides

•metals complexed by humic acids

Spectrum of sample

Raw spectra

absorption spectra

fluorescence spectra

pre-correction of fluorescence spectra:

• deadtime correction• weighted average pixel• correction of selfabsorption

Figure 6.2.12. Flow chart depicting the major steps during a fingerprint analysis of near-edge XAFS spectra.


conditions of parameters such as pH, redox poten-tial, temperature etc.

At this point the analysis of ‘natural’ samplesstrongly interferes with the characterization of sin-gle compounds. Preparing adsorbates of metals onclay minerals as reference materials for examplemakes it necessary to take the results of the inves-tigations on the kinetics of the sorption process,which were mentioned before, into account [15].

Sometimes it is possible to work around theproblem to choose a suitable not too large set ofreference compounds by use of certain statisticalmethods. Principal component analysis (PCA)[20] is a method that was previously used forthis purpose. Fundamental to this method is theavailability of an adequate number of sampleswhich contain differing amounts of the samechemical compounds. The number of samplesmust be at least one more than the number ofcompounds. Under this suppositions it is possibleto calculate from the set of sample spectra virtualmain compounds. The number of significant maincompounds corresponds to the minimum numberof real compounds that are contained in at leastone of the samples. One major advantage ofthis technique is that up to this point PCAyields its information without the use of referencecompound spectra.

In a further step called ‘target transformation’PCA not only provides information about thenumber of main components, it also tests whethera certain pure compound is a constituent of at leastone of the mixtures. Thus it is possible to limit thenumber of reference compound spectra used in thefingerprint method to those that are identified asbeing contained in at least one of the samples.

PCA has been used in several recent studiesof chemical binding forms. Examples are thedetermination of iron binding forms in eightcoal samples from different mines [21] or theidentification of changes in the speciation of Cuduring a chemical reaction [22]. In the first studyat least four main components were found tocontribute to the spectra of the mixtures. FeS2

could be identified by target transformation as oneof these species.

The Cu spectra for the latter example weremeasured in transmission mode on a copper metalfoil that was placed in a high temperature furnace(T > 700 K) under a continuous gas flow. Themetal was oxidized by varying concentrations ofoxygen to obtain either Cu2O or Cu and thechanges of the spectra followed over a longerperiod of time. The occurrence of Cu2O duringthe oxidation could be verified by using PCA andtarget transformation.

5.2 Catalysis research

Another field of science where analysis of speci-ation in solids is important and in which XAFSspectroscopy is frequently employed is catalysisresearch. Following the alterations of chemicalspeciation of the catalytically active side of a cat-alyst yields valuable insight into the mechanismof the catalytic process. It is used as a basis fora more precisely aimed development of new cat-alysts. In the context of this chapter it is used asan example of the ability of XAFS spectroscopyto perform species analysis in solid samples in situduring an ongoing chemical reaction.

The high flexibility in the design of XAFSsamples and sample holders enabled by the largepenetration depth of X-rays is of special impor-tance for in situ measurements during catalyticreactions. This will be demonstrated here usingexamples from investigations where XAFS spec-troscopy was coupled with other analytical meth-ods for special investigations. All these examplesare not standard techniques at any beamline butwill demonstrate the high flexibility of XAFSspectroscopy with respect to the sample environ-ment. Catalysis research benefits especially fromthe fact that XAFS spectra can be measured underphysico/chemical conditions (temperature, pres-sure, composition) similar to those in real use.Examples can be found in the literature whereXAFS spectra were measured under high pressure[23], high temperature [24] or a reactive atmo-sphere [25].

Tracing the formation and transformation ofchemical species during chemical reactions makes

LIMITATIONS 539

it necessary to decrease the time needed forthe measurement of a single spectrum. The timeneeded for an average XAFS scan at a bendingmagnet beamline is 15–60 min, which is much toolong to trace chemical reactions. Special experi-mental techniques make it possible to decrease thetime needed to no more than some seconds or evenms, especially when the scan range is limited tothe edge region (�E about 100 eV) which is oftensufficient and high flux sources are used. Two dif-ferent techniques are used to achieve such shortmeasuring times, as follows.

In the first approach a polychromatic incidentbeam is used. With an optical element that dis-perses the transmitted polychromatic X-ray lightspectroscopically after it has passed the sampleit is possible to map the whole spectrum on adetector array, analogous to the well-known diodearray detectors in UV/Vis spectroscopy. A draw-back of this technique is that the spectra can-not easily be measured in fluorescence mode.This limits the use of the so-called dispersiveEXAFS (DEXAFS) technique to higher concen-trated samples.

The second approach is to ‘simply’ increasethe scan speed of a conventional sequential XAFSscan. This makes it necessary to register thedetector signals continuously thus avoiding theoverhead caused by the stepwise registration ofthe spectra, as is usually employed in classicXAFS spectroscopy. This technique is namedQEXAFS (quick scanning EXAFS). The majorproblem of this technique is obvious: the highscan speed leads to poorer counting statisticsresulting in noisier spectra. QEXAFS experiments,particularly those with very high repetition rates,must be performed at high flux beamlines oninsertion devices. However, using fluorescencedetection and a scan range of 100 eV it waspossible to measure the Cu Kα XANES spectraof samples containing 2.5 mmol L−1 Cu in waterwith repetition rates of 9 Hz with reasonablequality [26].

The combination of XAFS spectroscopy withother analytical methods offers the opportunity toget additional information, for example thermo-dynamical data, simultaneously. For this purpose

XAFS was combined in one experiment with DSC[27]. Reactions of the general type

M–OOC–CH2 –X −−−→ MX

+ 1/n[–OOC–CH2 –]n

were studied, with M being an alkali metal or silverand X a halogen. A commercial DSC apparatusthat was mounted in the beam was used. Thesample powders were mixed with inert boronnitride, so that the desired edge jump could beachieved. This mixture was placed in thin walledAl crucibles. A hole with 1 mm diameter allowedthe beam to pass through the sample to measure themetal spectra. QEXAFS scans were then measuredat increasing temperatures. They showed clearlythe change in the binding form of the metal,coinciding with the DSC signal.

Other examples are the combination with FTIRspectroscopy [28], which was used to investigatethe molecular structure of RbReO4, or withX-ray diffraction (XRD) [29, 30], which addslong range order information to the short rangeorder information about the immediate chemicalenvironment of the absorber atom that is yieldedfrom XAFS spectroscopy.

6 LIMITATIONS

The limitations of a method are probably moreuseful for a decision on whether a certain methodcan be helpful to solve an analytical problem or notthan an enumeration of positive properties. As ananalytical method that is (most often) performedon solid samples XAFS spectroscopy is affectedby all the problems that are typical of this typeof sample, especially representativeness of thesamples for the bulk material, homogeneity ofthe actual sample etc. These problems will notbe further discussed here, because they are notspecific to this experimental technique.

Beside the positive aspects there are several fac-tors limiting the usefulness of XAFS spectroscopyfor elemental speciation analysis in solid sam-ples. Some limitations, such as the general needfor homogenous samples, the limited availabil-ity of synchrotron radiation source beamlines for


XAFS measurements or the problem of choosingsuitable reference substance spectra for the finger-print method, have already been mentioned.

For all samples which contain only oneunknown species of an element the most reliableinformation is yielded by the ‘classical’ EXAFSevaluation procedure. A detailed discussion oferror sources and how to minimize the errors intro-duced by the evaluation procedure can be found inthe literature [31].

The fingerprint approach is not limited to a purecompound but works more reliably with a smallnumber of compounds. Using this approach onlymain compounds with shares of more than ∼5 % ofthe total amount can be identified with reasonablecertainty [17].

Although the actual detection limits dependstrongly on the matrix properties it can be saidthat XAFS spectroscopy is not a technique thatis suited for trace analysis. If the measurementsare performed in transmission mode the detectionlimit is around 20 g kg−1. The use of fluorescencedetectors decreases the detection limit dependingon the matrix properties by a factor between 10and 1000 and in extreme cases even further. Ingeneral the detection limit is lower in samples thatcontain a heavy element as analyte in a light ele-ment matrix. This can be demonstrated with someexamples. For U in water it was possible to mea-sure EXAFS spectra in solutions containing only1 mmol L−1 of U [14]. The sorption process of Npat the α-FeOOH/water interface was investigatedin samples that contained some 10 µg g−1 Np [32].

Figure 6.2.10 shows a raw Cr K edge spectrummeasured with fluorescence detection in a samplethat contained 200 µg g−1 of Cr in soil. Thisspectrum was measured at a bending magnetbeamline (Beamline A1 at HASYLAB at DESY,Hamburg, Germany) using an energy-dispersiveGe detector.

An experimental setup such as the one thatis used for X-ray Raman spectroscopy (seeFigure 6.2.16) can also be used to increase thesignal/noise ratio in fluorescence detection of‘conventional’ XAFS spectroscopy. The analysercrystal used splits spectroscopically the radiationemitted from the sample. Thus it is possible to

exclude from the actual detector the unwanted pho-tons that are emitted by the matrix. The use of sucha secondary monochromator makes it possible tomake full use of modern synchrotron sources withvery high flux, because the detector and electron-ics must no longer count all the photons but onlythose originating from the element of interest.

In the following, two newer experimental devel-opments that can be helpful to overcome some ofthe limitations mentioned will be presented. Theneed for homogenous samples can often be over-come by reduction of the sample and beam sizeto a length scale on which the sample is homoge-nous. The necessity to measure the XAS spectraof lighter elements such as C, N and P under highvacuum conditions can be overcome by a tech-nique that uses higher energy radiation to gatherthe desired information.

7 µ-XANES

The distribution of an analyte in a solid sample isoften not homogenous or the distribution of differ-ent chemical binding forms is not homogenous. Inthese cases analytical methods with a high spatialresolution in the µm range can be a way to yieldvaluable additional information.

A technical prerequisite to perform XAFS spec-troscopy with a spatial resolution in the µm range(µ-XANES or µ-EXAFS) is the availability offocusing X-ray optics. With the development ofeffective X-ray optics during the last decade ithas become possible to perform X-ray absorptionspectroscopy on a µm scale. The µ-focus tech-nique minimizes the need for homogenous sam-ples, while at the same time raising the question

Source

Mirror 1

Mirror 2 Sample atfocal point

Figure 6.2.13. Scheme of a widely used setup of two ellip-tical focusing mirrors that enables horizontal and verticalfocusing of an X-ray beam under grazing incidence (Kirk-patrick–Baez geometry). A setup like this is used at severalµ-focus beamlines.

µ-XANES 541

of whether the results are representative of thebulk material.

Three types of optical devices are used forfocusing to µm spot sizes: first refractive opticselements such Fresnel zone plates and lenses, sec-ond two different types of reflective optics. Thefirst of the latter are toroidal or elliptical mir-rors which are operated under grazing incidence.Figure 6.2.13 shows as an example the ‘Kirk-patrick–Baez’ setup that is often used to producea small focal point with X-ray mirrors. Complexexperimental setups like this have to be employed,because X-ray photons are reflected only undergrazing incidence. The second approach usingreflective optics are special glass capillaries whichguide and concentrate the X-ray photons. Each ofthese methods has its special merits and the deci-sion which of the systems is used must be takenindividually, according to the X-ray source, sampleproperties and scientific question.

If a µ-focus beam is available two differentexperimental strategies can be employed, themapping technique in which a larger area ofthe sample is scanned and a two-dimensionalmap of the sample (surface) is produced andsecondly the analysis of single particles. In eithercase the sample must be mounted on an x/yscanning device with sufficient spatial resolution.The position of the beam is normally fixed sinceit is a much higher experimental effort to changethe position of the beam.

7.1 Mapping technique

Problems with the representativeness of the resultsyielded by µ-XANES can be partly overcome bymapping of larger areas on the sample. The sameapproach is chosen in the case of elemental map-ping, using for example electron microprobes andX-ray fluorescence detection. µ-XANES adds theability to discriminate between different chemi-cal forms of the element under investigation. Themajor advantage of this approach, however, is thatit markedly increases the relevance of the resultsobtained by adding the information about the spa-tial distribution of the analyte and its chemicalforms to the information about the chemical form

of the analyte at one spot, which is achieved witha nonscanning technique. The chemical parameterwhich is easiest to map using µ-XANES is the dis-tribution of different oxidation states of the analyteon the sample surface.

To perform oxidation state mappings the ele-mental distribution, e.g. the corresponding fluores-cence intensity, is measured at several differentenergies of the incident monochromatic beam. Thecorresponding energies of the incident beam arechosen so that certain oxidation states of the ana-lyte are excited preferentially [33]. In case of Cr,images recorded at incident beam energies of about5997, 6001 and 6007 eV would be used to distin-guish between elemental Cr, Cr(III) and Cr(VI);see Figure 6.2.3. The distribution of the differentoxidation states can then be calculated from theresulting elemental mappings.

This technique can be used for the determi-nation of redox states and mineralogical associa-tions of toxic or essential species in natural andcontaminated soils, sediments, waste encapsula-tion materials or minerals. An example is thedepiction of annual ring-like structures of alter-nating oxidation states of manganese in Mn/Fenodules from the Baltic sea [34]. Another exampleis the investigation of the redox chemistry ofmetals at the root – soil interface of plants andits role in agriculturally relevant plant diseases,for instance the measurement of Se diffusionand reduction at the water–sediment boundarythat was performed recently [35]. Further inter-esting applications would be measurements ofthe distribution of elements in different oxidationstates in microelectronic devices or the mappingand chemical characterization of metals within sin-gle cells and at the binding domains of bio films.These films are really ubiquitous at water/mineralsurfaces and are believed to play an important rolein many different processes such as the binding ofmetals in soils and sediments or the corrosion offresh and wastewater supply systems.

7.2 Single particle technique

The single particle technique yields essentiallythe same information as can be drawn from the


standard XAFS technique, but with much higherspatial resolution. Three important reasons forusing this experimental approach and not a bulktechnique are as follows.

(1) For small samples such as single dust particlesor colloidal particles. Examples are the charac-terization of single colloidal particles which arebelieved to play an important role in the trans-port of toxic metals in groundwater aquifers [36]or the determination of the redox state of theEarth’s interior based on valence determinations onmicrocrystals within diamonds or volcanic glasses.One example of the latter is the investigationof Fe inclusions in volcanic glasses which isof interest in geochemistry [37]. Focusing wasachieved using a Kirkpatrick–Baez mirror sys-tem. The samples investigated originated from twodifferent volcanic sites, the glass inclusions hadan Fe content of 10.86–9.31 %. The beam sizewas 15 µm × 15 µm. Only the actual edge region,−30 to +40 eV around the edge, was measured.Using standards with known Fe(II)/Fe(III) ratiosfor the calibration it was possible to determine theFe(II)/Fe(III) ratio in the samples. In general it canbe said that, if the sample contains regions (par-ticles) that contain higher concentrations of theanalyte than the bulk, the analyte/matrix signalratio can be increased by the µ-XAFS techniquecompared with investigations on bulk samples.(2) Different binding forms of the analyte arecontained in discrete particles. If the analyte iscontained in discrete particles in different bindingforms the problem that arises from the overlayof different XAFS spectra can be avoided sothat the ‘classical’ EXAFS spectrum interpretationbecomes possible. An example is the identificationof different uranium oxidation states in nuclearfuel particles, emitted during different phases ofthe Chernobyl incident [38].(3) High radioactivity or toxicity of the samplesmakes working with small amounts of samplematerial desirable or necessary. An example froma recent investigation is a µ-XAFS study of sorbedPu on tuff [39]. It was performed at an undulatorbeamline, using Kirkpatrick–Baez optics to focusthe beam. The spatial resolution was limited by thebeam size of 4 µm × 7 µm.

Further applications originate from all areas ofresearch where small samples and/or inhomoge-neous samples are involved.

8 X-RAY RAMAN SPECTROSCOPY

Elements such as C, N and O have an enormousimportance in chemistry. Unfortunately XAFSspectroscopy is not easily usable for the determi-nation of the local structure around light absorberatoms. This is because of the small energy of theabsorption edges, less than 1000 eV. The pene-tration depth of X-ray photons with such a lowenergy is very small, so that the samples haveto be very thin. For the same reason the mea-surements on light elements have to be performedunder high vacuum conditions, thus making itvery difficult if not impossible to measure wetsamples under realistic conditions without speciesalterations.

In X-ray Raman spectroscopy (or nonresonantX-ray Raman scattering spectroscopy) a sampleis irradiated with a fixed energy beam. The scat-tered radiation is investigated spectroscopically. Incontrast to resonant edge absorption spectroscopy,which is normally used to measure XAFS spec-tra, the photon energy can be much higher thanthe edge energy. In the case of a nonresonant edgeabsorption the scattered photon will lose part of itsenergy. This energy loss corresponds to the energyof the respective absorption edge. The process isequivalent to electron energy loss spectrometry(EELS), which is a well-known analytical tech-nique in transmission electron microscopy, or theclassical Raman spectroscopy in the IR region.Figure 6.2.14 depicts the X-ray Raman processschematically.

A plot of the distribution of the scattered pho-tons over the energy axis shows the same oscil-lations that can be found in the conventionallymeasured absorption spectrum. The major advan-tage of this technique, compared with the resonantmeasurement of the XAS spectra in the soft X-rayregion, is as mentioned above that the X-ray pho-ton energy can be chosen so high (∼10 keV) thatlow Z elements in thick samples can be measuredunder environmental conditions.

X-RAY RAMAN SPECTROSCOPY 543

Photoelectron

Scattered photonIncident photon

Photon

Electron

Figure 6.2.14. Schematic visualization of the X-ray Ramanprocess. The incident photon is inelastically scattered wherebyit loses a part of its energy. The photoelectron leaves the atomwith an energy that is equal to this loss minus the thresholdenergy E0, which is necessary to lift a core shell electron tothe continuum. The outgoing photoelectron wave undergoes thesame scattering processes as in XAFS spectroscopy.

For these elements X-ray Raman spectroscopywill become to a valuable alternative way to obtainthe desired information. Up to now the majordrawback has been the extreme weakness of theRaman scattering process. This makes it difficultto achieve good spectra with high resolution anda good signal/noise ratio with contemporary X-raysources within reasonable time spans.

Figure 6.2.15 shows examples of the K-edgespectra from boron and nitrogen in different BNsamples measured at a wiggler beamline [40]. Theflux during these measurements was 1013 pho-tons s−1, the spectral width (FWHM) was 1.1 eV at6 keV. The time needed for data collection of onespectrum was some hours. In an earlier study [41]measuring times between 24 h and 3 days wereused to obtain X-ray Raman spectra of pure car-bon compounds (diamond and graphite) at a bend-ing magnet beamline. The flux was estimated tobe 1011 photons s−1, the linewidth of the excit-ing monochromatic X-rays was estimated to be2 eV (FWHM). Today this technique is obviouslystill very time consuming, but with the predictableincrease of the spectral flux from X-ray sources inthe near future it will be possible to measure thesespectra within reasonable time.

Examples for possible applications of X-rayRaman spectroscopy include the following origi-nating from three totally different fields of researchwhere the chemical form of a low Z element isof interest:

180

Int.[

arb.

uni

ts]

200

Energy loss [eV]

220

(a)

240

Int.[

arb.

uni

ts]

Energy loss [eV]

(b)

380 400 420 440 460

Figure 6.2.15. K-edge X-ray Raman spectra from (a) boronand (b) nitrogen measured in a BN sample, spectra weremeasured at a wiggler beamline (BL16X of Photon Factory,KEK, Tsukuba, Japan), from ref. 40. Reproduced by permissionof the American Institute of Physics and the authors.

• Measurement of the XAS spectra of lightelements in semiconductor materials. At presentthese studies have to be performed in thesoft X-ray region. Materials of interest areIII–V semiconductor materials such as Mg-or Si-doped GaN [42] or AlGaN and InGaNalloys [43].

• An interesting experiment would be the com-bination of X-ray Raman spectroscopy withthe imaging techniques mentioned above. This


Incoming beam

Circle radius 2Rrow

(Johann-Geometry)

Rowland circle radius = Rrow

Bent crystal

Energy X

SampleEnergy Y

Energy Y > X

Figure 6.2.16. Schematic drawing of the ray paths in an experimental setup (Johann geometry) that employs a focusing bentcrystal for the spectroscopic dispersion of the X-ray photons emitted from the sample. A setup like this can be used for themeasurement of X-ray Raman spectra or to increase the signal/background ratio in a ‘conventional’ fluorescence mode XAFS scan.

would enable a three-dimensional mapping oflight elements and their chemical binding formsin thick particles where the penetration depthof edge energy photons is much too small.Again, a scientifically and economically inter-esting goal for this method is semiconductors,especially B- and N-containing III–V semicon-ductors (see above).

• The use of X-ray Raman spectroscopy incatalysis research will make it possible toperform in situ experiments and look from theother side, not from the reactive (metal) centerof a catalyst but from the substrate molecule onthe chemical reaction mechanism.

Figure 6.2.16 shows the principle experimentalsetup that can be used for X-ray Raman exper-iments. It is based on a focusing, sphericallyor cylindrically bent analyser crystal and a two-dimensional detector such as a CCD chip. Both aremounted on the circumference of a Rowland cir-cle. The crystal is bent to a radius twice that of theRowland circle. This so-called Johann geometryenables measurements with high energy resolution

(better 1 eV) for the X-ray Raman experiment andfocuses the spectroscopically dispersed light on thedetector, thus increasing the count rate by increas-ing the used solid angle.

9 SUMMARY AND OUTLOOK

Summarizing the foregoing it can be concludedthat XAFS spectroscopy is an analytical methodwhich yields all the necessary information toundertake species analysis in solid samples froma large number of different scientific fields. Par-ticularly advantageous is the negligible amount ofsample preparation that is required owing to theflexibility of the XAFS experiment with respect tosample properties such as humidity, size, shape,state of aggregation etc. However, XAFS spec-troscopy requires homogenous samples and suit-able synchrotron radiation sources or beamlinesand the detection limits are often higher than thelimits that are set in legal guidelines.

Recent and future experimental developmentssuch as the µ-methods and X-ray Raman scattering

REFERENCES 545

techniques described will most probably extend thepossible applications in species analysis markedly.The development of sources with even higher fluxtogether with improved detectors will enable afurther reduction of the detection limits.

At the moment a larger number of conventionalXAFS beamlines and µ-XAFS beamlines areavailable for experiments. All types of synchrotronsources, from bending magnets to undulators,are used at one or the other storage ring forXAFS spectroscopy beamlines. Bending magnetbeamlines have the advantage of easier operationand that more beamtime is available.

Undulator or wiggler beamlines, however, arethe first choice for well-planned experiments onhighly diluted samples. Because of the complexexperimental techniques and data evaluation thenewcomer in this field would be well advised tostart with his experiments at a bending magnetbeamline. All synchrotrons offer user support tohelp in planning and conducting experiments andchoosing the best suited beamline for a particularapplication. The high experimental effort thatis associated with the measurement of XAFSspectra appears to be appropriate in all caseswhere the analyte cannot be separated fromthe matrix without destruction of the speciesinformation or where measurements must beperformed in situ.

10 REFERENCES

1. Teo, B. K., EXAFS: Basic Principles and Data Analysis ,Springer, Berlin, 1986.

2. Koningsberger, D. C. and Prins, R. (Eds.), X-Ray Absorp-tion: Principles, Applications, Techniques of EXAFS, SEX-AFS and XANES , John Wiley & Sons, Inc., New York,1988.

3. Fay, M. J., Proctor, A., Hoffmann, D. P. and Hercules,D. M., Anal. Chem., 60, 1225A (1988).

4. Sanchez del Rıo, M. and Dejus, R. J., XOP: Recentdevelopments, SPIE Proc., 3448 (1998); see also: http://www.esrf.fr/computing/scientific/xop/intro.html.

5. Kincaid, B. M. and Eisenberger, P., Phys. Rev. Lett., 34,1361 (1975).

6. Sayers, D. E., Lytle, F. W. and Stern, A. E., Phys. Rev.Lett., 27, 1024 (1971).

7. Ankudinov, A. L., Ravel, B., Rehr, J. J. and Conrad-son, S. D., Phys. Rev. B , 58, 7565 (1998).

8. Stern, E., Newville, M., Ravel, B., Yacoby, Y. and Has-kel, D., Physica B , 208–209, 117 (1995).

9. Koch, E. E. (Ed.), Handbook on Synchrotron Radiation ,Vol. 1, North-Holland, Amsterdam, 1983.

10. Lide, D. R. (Ed.), CRC Handbook of Chemistry andPhysics: A Ready-reference Book of Chemical and Physi-cal Data , 80th edn., CRC Press, Boca Raton, FL, 1999.

11. Iida, A. and Noma, T., Jpn. J. Appl. Phys., 32, 2899(1993).

12. Troger, L., Arvanitis, D., Baberschke, K., Michaelis, H.,Grimm, U. and Zschech, E., Phys. Rev. B , 46, 3283(1992).

13. Scheidegger, A. M., Strawn, D. G., Lamble, G. M. andSparks, D. L., Environ. Sci. Technol., 30, 548 (1996).

14. Denecke, M. A., Reich, T., Pompe, S., Bubner, M., Heise,K. H., Nitsche, H., Allen, P. G., Bucher, J. J., Edelstein,N. M. and Shuh, D. K., J. Phys. IV Fr., 7, C2–637(1997).

15. Scheidegger, A. M., Strawn, D. G., Lamble, G. M. andSparks, D. L., Geochim. Cosmochim Acta, 62, 2233(1998).

16. Manceau, A., Boisset, M. C., Sarret, G., Hazemann, J. L.,Mench, M., Cambier, P. and Prost, R., Environ. Sci.Technol., 30, 1540 (1996).

17. Welter, E., Calmano, W., Mangold, S. and Troger, L.,Fresenius’ J. Anal. Chem., 364, 238 (1999).

18. Szulczewski, M. D., Helmke, P. A. and Bleam, W. F.,Environ. Sci. Technol., 31, 2954 (1997).

19. Hesterberg, D., Sayers, D. E., Zhou, W., Plummer, G. M.and Robarge, W. P. Environ. Sci. Technol., 31, 2840(1997).

20. Malinowsky, E. R. and Howery, D. G., Factor Analysisin Chemistry , John Wiley & Sons, Inc., New York, 1980.

21. Wassermann, S. R., J. Phys. IV Fr., 7, C2–203 (1997).22. Hilbrandt, N., Wasserman, S. R. and Martin, M., Solid

State Ionics , 101–103, 431 (1997).23. Frenkel, A. I., Wang, F. M., Kelly, S., Ingalls, R., Has-

kel, D., Stern, E. A. and Yacoby, Y., Phys. Rev. B , 56,10 869 (1997).

24. Koningsberger, D. C. and Vaarkamp, M., Physica B , 208& 209, 633 (1995).

25. Kappen, P., Grunwaldt, J. D., Hammershoi, B. S., Tro-ger, L. and Clausen, B. S., J. Catal., 198, 56 (2001).

26. Lutzenkirchen-Hecht, D., Grundmann, S. and Frahm, R.,J. Synchrotron Radiat., 8, 6 (2001).

27. Epple, M., Troger, L. and Hilbrandt, N. SynchrotronRadiat. News , 10, 11 (1997).

28. Wilkin, O. M. and Young, N. A., J. Synchrotron Radiat.,6, 204 (1999).

29. Topsoe, H., Stud. Surf. Sci. Catal., 130, 1 (2000).30. Grunwaldt, J.-D., Molenbroek, A. M., Topsoe, N. Y.,

Topsoe, H. and Clausen, B. S., J. Catal., 194, 452(2000).

31. Krappe, H. J. and Rossner, H. H., Phys. Rev. B , 61, 6596(2000).

32. Combes, J. M., Chisholm-Brause, C. J., Brown, G. E.,Parks, G. A., Conradson, S. D., Eller, P. G., Triay, I. R.,


Hobart, D. E. and Miejer, A., Environ. Sci. Technol., 26,376 (1992).

33. Sutton, S. R., Bajtz, S., Delaney, J., Schulze, D. andTokunaga, T., Rev. Sci Instrum., 66, 1464 (1995).

34. Kersten, M. and Wroblewski, Th., ‘Two dimensionalXAFS Topography of Amorphous Oxyhydroxide Lay-ers in Mn/Fe nodules’, in HASYLAB Annual Report1998 , HASYLAB at DESY, Hamburg, Germany, 1998,pp. 875–876; also available on the World Wide Web:http://www-hasylab.desy.de/science/annual reports/1998/part1/contrib/26/2052.pdf.

35. Tokunaga, T. K., Sutton, S. R., Bajt, S., Nuessle, P.and McCarthy, G. S., Environ. Sci. Technol., 32, 1092,(1998).

36. Kammer, F. v. d. and Forstner, U., Water Sci. Technol.,37, (6–7) 173 (1998).

37. Mosbah, M., Duraud, J. P., Metrich, N., Wu, Z., Delaney,J. S., San Miguel, A., Nucl. Instrum. Methods B , 158, 214(1999).

38. Salbu, B., Radiat. Prot. Dosim., 92, 49 (2000).39. Duff, M. C., Newville, M., Hunter, D. B., Bertsch, P. M.,

Sutton, S. R., Triay, I. R., Vaniman, D. T., Eng, P. andRivers, M. L., J. Synchrotron Radiat., 6, 350 (1999).

40. Watanabe, N., Hayashi, H., Udagawa, Y., Takeshita, K.and Kawata, H., Appl. Phys. Lett., 69, 1370 (1996).

41. Tohji, K. and Udagawa, Y, Phys. Rev. B , 39, 7590(1989).

42. Katsikini, M., Moustakas, T. D., Paloura, E. C., J. Syn-chrotron Radiat., 6, 555 (1999).

43. Katsikini, M., Fieber-Erdmann, M., Holub-Krappe, E.,Korakakis, D., Moustakas, T. D. and Paloura, E. C. J.Synchrotron Radiat., 6, 558 (1999).

CHAPTER 7

Calibration

7.1 Calibration in Elemental Speciation Analysis

K. G. HeumannInstitut fur Anorganische and Analytische Chemie, Mainz, Germany

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 5472 Special Features of Calibration for

Elemental Species Analysis . . . . . . . . . . . . 5493 External Calibration . . . . . . . . . . . . . . . . . 5514 Internal Calibration . . . . . . . . . . . . . . . . . . 552

4.1 Standard addition method . . . . . . . . . 5524.2 Mass spectrometric isotope dilution

technique . . . . . . . . . . . . . . . . . . . . . 554

4.2.1 Fundamentals . . . . . . . . . . . . 5544.2.2 Species-specific and species-

unspecific calibration . . . . . . . 5554.2.3 Validation of analytical

procedures by the isotopedilution technique . . . . . . . . . 559

5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . 5606 References . . . . . . . . . . . . . . . . . . . . . . . . 561

1 INTRODUCTION

Similar to methods for the determination of totalelement concentrations, calibration methods forelemental speciation analysis can be classified intocategories of absolute and relative methods [1].An absolute method produces a result that isdirect traceable to SI units. Gravimetry, titrimetry,and coulometry are such absolute determinationmethods. Absolute methods can be evaluatedwithout any comparative measurement, but theydepend on physicochemical constants such asFaraday’s constant in the case of coulometry.In the absence of interferences, the amount ofan elemental species determined by coulometrycan be calculated using the measured amount ofcharges if the corresponding redox reaction is welldefined. A well-known example of a coulometric

determination of an elemental species is the anodicoxidation of As(III) to As(V) at a platinumelectrode. The basis of titrimetric determinationsof elemental species is a well-defined chemicalreaction, where the amount of the analyte is relatedto the stoichiometric coefficients of the reaction.Titrimetric redox reactions are well established sothat different oxidation states of an element canbe determined, e.g. the determination of Fe(II) bypermanganate in acidic solutions. The selectiveprecipitation of sulfate, even in the presence ofother sulfur species, is a well-known example ofgravimetric elemental species determinations fromstudent courses.

All these absolute methods must be suffi-ciently selective if other compounds are presentin the sample besides the elemental species tobe determined. For coulometric determinations


548 CALIBRATION

selectivity can be achieved in many cases by apply-ing a fixed potential difference to the electrodes,which does not allow a redox reaction for othercompounds. Selectivity is normally an especiallycritical point in gravimetry because many com-pounds can usually be precipitated by the sameprecipitator. Possible interferences must thereforebe eliminated before precipitation takes place. Forexample, gravimetric determination of nitrate ispossible by precipitation as nitron nitrate. Inter-ferences by nitrite, often also present in sampleswhere nitrate occurs, can be eliminated by selectivereduction of nitrite to elementary nitrogen, e.g. byamidosulfuric acid. Coprecipitation and adsorptionof other species can also occur during gravimet-ric determinations. These interferences are usuallydifficult to prevent.

However, all absolute methods are extremelysensitive with respect to possible interferences sothat they are normally only useful for the deter-mination of pure samples. In addition, gravimetricand titrimetric methods can usually not be appliedat trace levels, where it is very often necessaryto analyze elemental species. Thus, absolute meth-ods are not important at the moment for elementalspeciation analysis. On the other hand, determina-tions of pure stock solutions of elemental speciesstandards, which are used for calibration of rel-ative methods, are often carried out by titrimet-ric methods.

Relative methods are those where detection ofthe elemental species in a sample to be analyzed isachieved by comparison with a set of calibrationsamples of known content. Reference materials(RMs) or certified reference materials (CRMs)are applied for calibration with identical, or atleast similar, matrix composition. In contrast toCRMs that are certified for their total elementconcentration, there are only a few CRMs availablefor elemental species analysis. Aspects which needto be taken into consideration when calibratingby using CRMs, are summarized in Chapter 7.2.Most of the analytical methods applied today forelemental speciation are relative methods so thatthe corresponding type of calibration procedurewith a set of calibrants is fully discussed in thefollowing text.

Because of the practical advantages of linearcalibration graphs they are always favored inanalytical chemistry. Linear calibration graphs canbe obtained by measuring only a few calibrationstandards and, in addition, are easily described bya simple mathematical function:

S = kc + b, (7.1.1)

where S is the signal response of the detectionmethod, c the concentration of the calibrant, k thecalibration factor, and b the intersection with they-axis. The k value reflects the sensitivity, andsensitivity is therefore better the higher the k valueis for a given concentration range.

The quality of a calibration is, in principle,controlled by the repeatability of the measurement,the trueness of the standards used, and the validityof the comparison between the calibrant and thesample. Whereas the repeatability influences theprecision of the analytical result, the two lastfactors mentioned influence the accuracy. Thevalidity of the comparison between the calibrantand the sample is one of the most criticalpoints in calibration techniques for trace analyses.Therefore different calibration modes are useddepending on how critical this comparison is.External calibration modes, where the sampleand the corresponding calibrant are separatelymeasured, do not fit identical measuring conditionsfor the standard and the sample in the same wayas internal calibration techniques. The followingdifferent modes of internal calibration have beenused, up to now, for elemental speciation analysis:

(a) For the standard addition technique each sam-ple is split into several subsamples and anincreasing but known amount of the analyteis added to the different subsamples. This cal-ibration technique usually eliminates possibleinfluences of the matrix composition on thesignal intensity of the detection system [2].If this is a critical point during analysis, thestandard addition method can successfully beapplied for the determination of elementalspecies. Some of the possible errors duringanalytical steps prior to the final detection,for example, by extraction of an analyte from

SPECIAL FEATURES 549

solid samples, can usually not be corrected forthis calibration mode unless total equilibrationbetween analyte and standard is guaranteed.

(b) The species-specific isotope dilution techniqueuses a known amount of a spike, which con-tains the elemental species to be determinedin a different isotopic composition from thatof the sample. For isotope dilution the spikeis added to the sample and homogeneouslymixed prior to all other analytical sampletreatment steps. Because the isotope ratio ofthe isotope-diluted sample must then be mea-sured, mass spectrometry has to be used asdetection method. Isotope dilution mass spec-trometry (IDMS) fits excellently the princi-ples of an internal calibration because isotopesare, within an uncertainty of usually less than0.1 %, identical in their chemical behavior andthey can easily be detected by mass spectrom-etry. In addition, only a ratio of the amountof isotopes and not an absolute amount ofthe analyte must be measured, which allowsloss of substance during sample pretreatmentsteps without any influence on the analyticalresult. This is due to the fact that, even inthe case of loss of substance, the isotope ratioof the isotope-diluted elemental species doesnot change. These characteristics of IDMS nor-mally lead to highly accurate results [3]. Theisotope dilution technique is also a one-pointcalibration method, which saves time com-pared with other calibration methods. Theseadvantages are the reason why an increasingnumber of elemental speciation analyses arecarried out by IDMS. Under certain conditionsother chemical forms of an element, not iden-tical with the elemental species to be deter-mined, can be used (species-unspecific IDMS;see Section 4.2.2).

2 SPECIAL FEATURES OFCALIBRATION FOR ELEMENTALSPECIES ANALYSIS

A lack of availability of calibrants, problemswith the stability of elemental species standards,possible species transformations during the sample

treatment procedure, and a total separation of onespecies from the other in the case of a species-unspecific detection, are special features of calibra-tion in elemental species analysis compared withthe determination of total element concentrationsin a sample. Whereas most of the relevant inor-ganic elemental species are commercially avail-able today, e.g. selenite and selenate or iodide andiodate, there is often a lack of organoelementalspecies such as dimethylthallium. However, mostof the anthropogenic organoelemental species dis-tributed worldwide, e.g. tributyltin and tetraethyl-lead, are available on the market. There is a totallack of isotopically labeled elemental species sothat in all these cases synthesis of the correspond-ing spike compounds must be carried out.

Literature procedures are available for the syn-thesis of most of the stable elemental specieswhich are of actual interest in elemental speci-ation analysis. Nevertheless, an exact character-ization has to be carried out after synthesis ofan elemental species which is used as a cali-brant. With regard to the use of such a cali-brant the species-specific purity is an importanttopic to be taken into account. When synthesizingelemental species with natural isotopic compo-sition substantial amounts of educts can usu-ally be applied. In contrast to this only smallamounts, usually in the milligram range, must beapplied for the synthesis of isotopically labeledelemental species because of the costs of theisotope-enriched initial material. In these cases thechemical procedure must be optimized with respectto a high synthesis yield for the isotope-enrichedcompound but not for the other reactant(s). A cou-ple of descriptions for the synthesis of isotopi-cally labeled elemental species can be found inthe literature. Corresponding procedures for 82Se-enriched selenite and selenate as well as those for129I-enriched iodide and iodate are described byHeumann and coworkers [4, 5]. A 206Pb-enrichedtrimethyllead spike was first used by Ebdon‘sgroup for the determination of the correspond-ing lead species in rainwater by species–specificIDMS [6]. Recently, procedures for the synthesisof isotopically labeled monomethylmercury havealso been published [7, 8], where methylation of

550 CALIBRATION

isotope-enriched Hg2+ ions was carried out byreaction with a Grignard reagent in organic solventand with methylcobalamin in aqueous solution,respectively.

The stability of elemental species standard solu-tions used for calibration is a much more criticalpoint than for standard solutions used for totalelement determinations. In addition to problemsobserved already with trace element standards,where adsorption at walls of storage vessels orevaporation of the solvent may influence the con-centration, decomposition of the elemental speciescan also lead to time-dependent variations of theconcentration in solutions. Different storage con-ditions, such as composition of the solvent, pHvalue, oxidants, temperature, but also the mate-rial of the storage vessel, are important parameterswhich can cause a possible decomposition of theelemental species. The storage conditions for ele-mental species standard solutions must thereforebe carefully checked and they are more critical thelower the concentration level is.

For example, it was found that an aque-ous monomethylmercury solution was stable formonths when it was stored in a closed PFA ves-sel at 4 ◦C in the dark [8, 9], which could not beconfirmed for PE vessels. A mixture of Cr(III)and Cr(VI) remains stable over months if it isstored in PFA bottles at 5 ◦C in an HCO3

−/CO2

buffer solution at pH 6.4 under a CO2 blanket [10,11]. Different stabilities were found for iodide andiodate solutions in the concentration range of afew µg L−1 during their storage in distilled water.Whereas iodate was stable over months, the iodideconcentration decreased by about 10 % after 5 daysand by more than 25 % after 1 month [12], so thatiodide solutions must be freshly prepared for cal-ibration purposes. The instability of iodide is alsothe reason why iodine doping of food is often car-ried out with iodate and not with iodide. Contraryto the iodide instability, bromide, also as a mixturewith bromate, was found to be stable for monthsin an HCO3

−/CO2 buffer solution [13].Transformation of the elemental species dur-

ing the sample treatment procedure can be anotherimportant source of error in elemental speciation. Ifcalibration by a species-specific standard is carried

out for the detection step, possible transforma-tions of the corresponding elemental species atother analytical steps prior to detection are notreflected in the result obtained. The total analyt-ical procedure should therefore be validated forpossible species transformations to be sure thatspecies transformation do not play a role. Such avalidation of analytical methods can be carried outby applying the isotope dilution technique [8] (seeSection 4.2.3). During the last few years doubts,especially on the accuracy of methylmercury deter-minations, arose when Hintelmann et al. observedadditional formation of this mercury species duringits water vapor distillation from sediments [14].By determining the distilled methylmercury afterethylation and subsequent analysis with a GC-ICP-MS system, they found that substantial amountsof this species can be formed during the distil-lation process from inorganic Hg2+ ions. In con-trast, Demuth and Heumann observed formationof elementary mercury from methylmercury dur-ing ethylation of this species in the presence ofhalide ions [8]. Species transformations have beenbest investigated for methylmercury, but they canalso occur for other alkylated elemental species,e.g. of lead and tin.

Because most of the detection methods appliedtoday in elemental speciation analysis, e.g. atomicabsorption spectrometry, atomic emission spec-trometry, and ICP-MS, respectively, are notspecies specific in their signal response, a com-plete separation of the different species one fromthe other must be carried out prior to detection.In addition, the recovery must be determined foreach of the species analyzed if less than 100 % ofthe elemental species is isolated after the differ-ent sample treatment steps. However, under cer-tain conditions electroanalytical methods enabledirect detection of elemental species in solutioneven if other species of the same element arepresent (see Chapter 5.9). An interesting exampleis the determination of monomethylcadmium inaquatic samples in the presence of inorganic Cd2+by differential pulse anodic stripping voltammetry,which allows the determination of this biogenicallyproduced cadmium species at concentration levelsdown to about 0.5 ng L−1 [15].

EXTERNAL CALIBRATION 551

3 EXTERNAL CALIBRATION

External calibrations can be carried out with aspecies-specific or a species-unspecific standardsolution. In the case of a species-specific calibra-tion the chemical form of the calibrant is iden-tical with the analyte to be determined, whereasfor species-unspecific calibration the calibrant isdifferent in its chemical form. Species-unspecificcalibration can never be used for a complete ana-lytical procedure including, for example, extractiveor chromatographic separations. There will alwaysbe a fractionation between the different speciesduring such a separation procedure which preventscalibration of these analytical steps by an elemen-tal species not identical in its chemical form withthe analyte.

Under certain conditions species-unspecific cal-ibration is therefore only acceptable for thedetection step. However, in this case it must beguaranteed that the response of the detector isidentical for the different elemental species in thesample and in the standard solution. It can beassumed that detector systems operating at hightemperatures, such as those with inductively cou-pled plasma excitation, should not show significantdifferences in the detector response for various ele-mental species. Table 7.1.1 shows a comparisonof the ICP-MS response for a 0.5 % HNO3 acidicsolution containing inorganic lead ions (Pb2+) andtrimethyllead (Me3Pb+), respectively, for differentnebulizer systems [16]. Both solutions are iden-tical in their lead content (12 µg L−1). As can

Table 7.1.1. ICP-MS response for a Me3Pb+ and an inorganicPb2+ solution of identical lead content measured with differentnebulizer systems (0.5 % HNO3 acidic solutions containing12 µg Pb L−1; measured isotope 208Pb).

Nebulizer system Pb species ICP-MSresponse (cps)

Cross-flowa Pb2+ 6400 ± 400Me3Pb+ 6360 ± 380

µ-flowb Pb2+ 8960 ± 220Me3Pb+ 9000 ± 210

Ultrasonic withmembranedesolvatora

Pb2+Me3Pb+Blank(0.5 % HNO3)

13 850 ± 1200475 ± 55255 ± 50

Quadrupole ICP-MS instrument used: aELAN 5000; bHP 4500.

be seen from the results listed in Table 7.1.1the same detector response (counts per second,cps) is obtained for both lead species within thegiven standard deviations using a cross-flow and aµ-flow nebulizer system, respectively. From thisit follows that under these conditions the sameamount of lead is introduced into the plasma torch,using either trimethyllead or inorganic lead, andthat there is also no difference in the ionizationefficiency for both species. A similar result, withidentical response for both species, was also foundfor an iodide and iodate solution at the 4 µg L−1

concentration level by applying a quadrupole ICP-MS with a cross-flow nebulizer [12].

Identical absorbances were measured by flameatomic absorption spectrometry for the above men-tioned trimethyllead and inorganic lead solutionscontaining identical lead contents. This also did notchange when the temperature of the flame was var-ied using different air–acetylene gas mixtures. Incontrast, when using an ultrasonic nebulizer with amembrane desolvator (CETAC U-6000AT) as theintroduction system for ICP-MS, a totally differ-ent result was observed. Whereas a high ICP-MSresponse was obtained for inorganic Pb2+ ions, theresponse for Me3Pb+ was only a few counts persecond above the HNO3 blank (Table 7.1.1). Themembrane desolvator obviously eliminates methy-lated lead effectively from the sample before reach-ing the plasma so that introduction of alkylatedheavy metal species by such a system is not a goodchoice because fractionation between different ele-mental species can easily occur.

From determinations of the total element con-centration it is well known that the matrix canstrongly influence the detector response. It hastherefore to be proved if such matrix effects arealso relevant for different elemental species. Fromthe results summarized in Table 7.1.1 it followsthat a 0.5 % HNO3 acidic solution, for example,does not cause differences in the detector responsefor the two lead species analyzed (see results forcross-flow and µ-flow nebulizer). However, thisis not necessarily also valid for all other matricesand/or concentration ranges.

Figure 7.1.1 represents the signal intensity ofPb2+ and Me3Pb+, respectively, measured with

552 CALIBRATION

0.0 0.2 0.4 0.6 0.8 1.0

2500

3000

3500

4000

4500

5000

5500

6000

6500

Me3Pb+

Pb2+

inte

nsity

[cps

]

chloride concentration [%]

Figure 7.1.1. ICP-MS response of two different lead species containing the same lead content (12 µg Pb L−1) dependent onthe matrix influence of a diluted seawater sample (the concentration of the diluted seawater is represented by its chlorideconcentration).

an ICP-MS instrument (ELAN 5000), equippedwith a cross-flow nebulizer, and its dependenceon the salt concentration of a diluted syntheticseawater sample [16]. The salt concentration isexpressed as its chloride content, with 1.8 % forundiluted seawater. Me3Pb+ can be biogenicallyproduced in the ocean so that the correspondingmatrix is relevant for trimethyllead determinations.For both lead species the signal intensity decreasessubstantially with increasing chloride concentra-tion, affecting Me3Pb+ much more than Pb2+. Thisleads to a difference of 15 % in the signal intensityat a chloride concentration of about 0.9 % betweenboth species by a total response reduction of about60 % for Me3Pb+ compared with the species solu-tion without any matrix. From this it follows thatcalibration of species with a calibrant, different inits chemical form, can be strongly influenced bythe matrix and also depends on its concentration.Summarizing the possibilities of species-unspecificcalibration, it must be stated that the validity ofsuch a calibration must always be proved for dif-ferent matrix concentrations even if an identicalresponse of different species is obtained for thematrix-free solution.

On-line coupling of chromatographic separationwith atom spectrometric detection methods is one

of the most powerful tools for speciation. In allthese cases transient signals are produced whichmust be compared for calibration with correspond-ing signals of a standard solution. For quantifica-tion either the peak height or the peak area canbe used for calculating the analytical result. Pre-cision and accuracy are usually better when peakareas are evaluated whereas evaluation of the peakheight is much simpler. In the case of symmetricpeaks the peak height fits much better the result bypeak areas compared with asymmetric peaks.

4 INTERNAL CALIBRATION

4.1 Standard addition method

Calibration by the standard addition method elim-inates matrix interferences during detection andcan also be used to correct for possible losses ofthe analyte during sample treatment procedures ifsample and standard have been allowed to equi-librate. Because of possible matrix effects, whichaffect the detector response to different elemen-tal species in a different way (see Section 3),species-specific standard addition should usuallybe applied for detector calibration. Calibration of

INTERNAL CALIBRATION 553

the other analytical steps prior to detection canonly be carried out by a species-specific standard.This means that the calibrant, added in increasingamounts to the different subsamples, should beidentical in its chemical form with the elementalspecies to be determined. Under these conditionsanalytical results, obtained with the standard addi-tion method, are usually more precise and accuratethan those obtained by external calibration. Onedisadvantage of this calibration method is the factthat at least three to five aliquots (subsamples)need to be prepared because an increasing amountof the calibrant must be added to these differentsubsamples.

Figure 7.1.2(a) shows a standard addition cal-ibration curve obtained for the determination ofbromate in a mineral water sample by detectionof the 79Br isotope with a quadrupole ICP-MSafter species separation by anion exchange chro-matography, which is shown in Figure 7.1.2(b) forthe original sample. The standard addition calibra-tion curve resulted in a bromate concentration of1.64 µg L−1, whereas 14.6 µg L−1 was obtained forbromide by applying an analogous calibration pro-cedure also for this bromine species [13]. Linearcalibration curves by the standard addition methodare represented by equation (7.1.1). The corre-sponding mathematical expression for the bromate

0 0.5 1 1.5 2 2.5 3 3.5 54.54

Pea

k ar

ea [a

rb. u

nits

]

−0.5−1−1.5

concentration of sub-sample by added amount of analyte [µg/L]

(a)

result:1.64 µg/L

S = 59.47c + 97.48

R2 = 0.9976

50

200

150

250

300

100

350

0

20

40

60

80

100

inte

nsity

[x 1

03 cps

)

120

140

0 5 10 15 20 25 30 35 40


(b)

BrO3− Br−

Figure 7.1.2. Determination of bromate by ICP-MS with the standard addition method in a mineral water sample after separationby anion exchange chromatography: (a) calibration curve; (b) chromatogram of separated bromate and bromide detected byICP-MS.

554 CALIBRATION

determination discussed is given in Figure 7.1.2(a).The quality of the calibration curve of the standardaddition method (linear regression) should alwaysbe judged by the correlation coefficient R, or bet-ter by its squared value R2, which is given for thebromate determination in Figure 7.1.2(a). The cor-relation coefficient can vary between −1 and +1and values close to −1 or +1 indicate a good fitof the regression curve. For a good calibration theR2 value should be significantly larger than 0.95.The relative uncertainty of the result obtained bythe standard addition method is best if the slope ofthe calibration curve is around 45◦ and becomeshigh for extremely gradual or steep curves.

4.2 Mass spectrometric isotope dilutiontechnique

4.2.1 Fundamentals

Isotope dilution mass spectrometry is internation-ally accepted as a definitive method of proven highaccuracy and precision where the possible sourcesof error are understood and usually also under con-trol [3, 17]. One of the greatest advantages ofIDMS is that loss of the analyte has no effect on theresult after the isotope dilution process has takenplace, which also means that no recovery must bedetermined for sample treatment procedures of theisotope-diluted sample. The isotope dilution stepmust guarantee total equilibration between the iso-topically labeled spike and the sample species. Thisis usually no problem in aqueous systems, wheremany of the analyses for elemental species inthe environment take place, or for samples totallydissolved, e.g. by an acid. IDMS of elemental spe-ciation analysis can be particularly limited in caseswhere sample matrices cannot be totally dissolvedbecause of possible species transformations duringthis process. The principles of IDMS are describedin various textbooks [3, 18].

In principle, all different types of mass spec-trometers can be used for the isotope dilutiontechnique. However, in the case of using ioniza-tion methods where molecules or molecular frag-ments are produced and analyzed, as is true for

applying electron impact or electrospray ionization(ESI), the natural isotopic pattern of the moleculescan cause severe interferences. This is a specialproblem for elemental species containing largeorganic molecules, e.g. for metal complexes ofbiomolecules, because the 12C/13C pattern of suchcompounds causes many isotopic peaks. In thiscase mathematical corrections must be appliedto obtain reliable analytical IDMS results. It istherefore much easier to apply ionization meth-ods where preferably atomic ions are produced anddetected, such as thermal ionization mass spec-trometry (TI-MS) or ICP-MS. In addition, theseionization methods are mostly more sensitive thanthose producing molecular ions. The disadvantageis that the elemental species to be determined mustbe known and well defined because no structuralinformation is obtained by these ionization tech-niques. The combined use of different ionizationtechniques will therefore certainly become a futuretrend for the quantification and characterization ofelemental species of unknown composition. Forexample, ICP-IDMS can quantify the elementalspecies by a selected element in the molecule andESI-MS may be able to identify its structure andcomposition.

TI-IDMS and ICP-IDMS have usually beenapplied to determine elemental species. Becausethese mass spectrometric methods are not ableto differentiate between different chemical formsof an element, a total separation of all elementalspecies must be carried out prior to the mass spec-trometric detection. In this case isotope-labeledspikes, identical in their chemical composition withall elemental species to be determined, must firstbe added to the sample. After equilibration ofthe elemental species of the sample with thoseof the spike, which is best done in a homoge-neous solution, separation of the isotope-dilutedspecies can be carried out. Then, the isotope ratioR of the spike isotope over a reference isotopeof the element, used for labeling the elementalspecies, is measured for all separated fractionsof the different elemental species. The schematicFigure 7.1.3 shows the principles of such an iso-tope dilution technique using a copper species asan example.


sample

63 65

63

spike

65

63

m/z

m/z

ion

inte

nsity

65

isotope diluted sample

Figure 7.1.3. Schematic figure of the principles of isotopedilution mass spectrometry by the example of measuring thecopper isotopes in a copper containing species.

The isotope ratio R of the isotope-diluted sam-ple is the only quantity that needs to be measuredfor calibration by the species-specific IDMS tech-nique. If the mass spectrometer produces isotopepeaks with a flat top, as is the case for mag-netic sector field instruments, the peak heights ofthe spike and reference isotope are used for cal-culation of the isotope ratio. In all other cases,e.g. by applying quadrupole instruments, the peakareas of the isotopes must be measured for thedetermination of the isotope ratio. As can be seenfrom Figure 7.1.3, the ion intensities of the lighter(63Cu; in the following equations index 1) and theheavier (65Cu; index 2) isotopes of the isotope-diluted sample are identical to the sum of thesample portion and spike portion used. Hence,equation (7.1.2) enables the isotope ratio R to becalculated:

R = (NShS2 + NSph

Sp2 )/(NShS

1 + NSphSp1 )

(7.1.2)

where NS,Sp is the number of molecules of ele-mental species in the sample and in the spike, andh1,2 are the isotopic abundances of isotope 1 andisotope 2. After transformation of equation (7.1.2)for the content GS of the elemental species in thesample one obtains

GS = 1.6610−24(M/W S)NSp[(hSp2 − Rh

Sp1 )/

(RhS1 − hS

2)] [g g−1] (7.1.3)

where M is the molecular weight of the elemen-tal species and W S is the sample weight.

The isotope abundances in the elemental speciesof the sample are usually identical with the well-known natural isotopic compositions, which arelisted in corresponding IUPAC tables [19]. Theisotope abundances of the spike solution and itsconcentration are determined separately where theconcentration is usually obtained by a reverseIDMS technique using a standard solution of theelemental species with natural isotopic composi-tion. Possible mass bias effects of the applied massspectrometric techniques can best be eliminated byanalyzing the isotopic compositions of the sam-ple and the spike under the same conditions asthe isotope ratio R, or they have to be correctedby independent measurements with standards [20].Calibration by IDMS should also take into accountan optimization of the ratio of the amount of sam-ple species over spike species to minimize theerror multiplication factor, which influences theprecision of the IDMS result. Depending on theenrichment of the spike isotope this ratio shouldnormally be between 0.1 and 10 [3].

4.2.2 Species-specific and species-unspecific calibration

Whereas TI-IDMS was the preferred techniquebefore 1995 [3–5, 11], the first ICP-IDMS anal-yses of elemental species appeared in 1994 [6,21]. Today, the number of investigations usingICP-IDMS for elemental species analysis exceedsthose with TI-IDMS. This is due to several advan-tages of ICP-MS compared with TI-MS even ifthe selectivity is much better for TI-MS becauseof significantly reduced problems with spectromet-ric interferences. The multielement capability ofICP-MS is one of these advantages. However, themuch more complicated sample preparation tech-nique for TI-IDMS with its relatively high timeconsumption as well as the fact that only ICP-MSoffers on-line coupling with separation techniquesare the major reasons why ICP-IDMS is now morefrequently used. The great difference in the sampletreatment procedures between TI-IDMS and ICP-IDMS is demonstrated in Figure 7.1.4, showing

556 CALIBRATION

Sample + 79Br− spike + 79BrO3− spike

Separation of isotope diluted bromine speciesby anion exchange chromatography

Collection ofBr− fraction

Precipitation of AgBr and filtration

Dissolution of AgBr in NH3 soultion

Deposition on filament andevaporation to dryness

79Br/81Br ratio measurementby NTI-IDMS

79Br/81Br ratio measurement in theBr− and BrO3

− fraction by ICP-MS

NTI - IDMS ICP - IDMS

off-line

BrO3− fraction

reduced to Br−

on-line

Figure 7.1.4. Comparison of the sample pretreatment techniques for the determination of bromide and bromate in an aquaticsample by negative thermal ionization isotope dilution mass spectrometry (NTI-IDMS) and inductively coupled plasma isotopedilution mass spectrometry (ICP-IDMS), respectively.

the determination of bromide and bromate byspecies-specific calibration with the isotope dilu-tion technique [13]. Whereas the isotope-dilutedfractions of bromate and bromide, separated byan anion exchanger column (Figure 7.1.2(b)), areintroduced on-line into the ICP-MS instrumentwithout any additional sample treatment, variousoff-line steps of elemental species/matrix separa-tion and species isolation must be carried out inorder to measure the bromine isotope ratio of thetwo species by negative thermal ionization massspectrometry (NTI-MS).

Volatile and thermally stable elemental speciesare usually well defined with respect to their chem-ical composition and structure and can best beseparated by gas chromatography. Species-specificIDMS calibration using on-line coupling of GCwith ICP-MS is therefore preferably applied forthis type of elemental species. The determina-tion of dimethylselenide and dimethyldiselenideis a relevant example [22]. For GC-ICP-IDMSdeterminations the isotope dilution step is always

carried out in the beginning of the analyti-cal procedure so that the isotope-diluted sam-ple passes the GC separation column. Isotopeeffects during this separation are usually verysmall so that they can be ignored. In addition,some nonvolatile elemental species can easily beconverted into a volatile compound by derivatiza-tion which also offers the possibility of analyz-ing them by a GC-ICP-IDMS coupling system.For example, monomethylmercury (MeHg+) isconverted into the volatile ethylated compoundMeEtHg by sodium tetraethyloborate [8, 14] andselenite is specifically converted into piazselenolby 1,2-diaminobenzenes [23].

Besides the species-specific spiking mode, aspecies-unspecific spiking mode is also possibleunder certain conditions. This is the only way toapply IDMS calibration to either elemental specieswhere the chemical composition and structure isnot exactly known (so that an isotope-labeledspike cannot be synthesized) or to all specieswhere the synthesis of a labeled compound is


too complicated. These are the reasons whymetal complexes of humic substances or thoseof biomolecules have only been quantified, upto now, by a species-unspecific spiking modewhen using IDMS calibration in connection witha HPLC-ICP-MS coupling system [24–26]. Whenapplying the species-unspecific spiking mode,where the spike may exist in any chemical form,the isotope dilution step cannot be carried outbefore a complete separation of the differentspecies has taken place. No loss of substance ofthe different species is therefore allowed up tothis analytical step. It must also be guaranteedthat the sample and the spike species do notproduce different ICP-MS responses and that thereis no discrimination between the sample and spikespecies by the sample introduction system (seeSection 3; Table 7.1.1 and Figure 7.1.1).

Figure 7.1.5 shows the schematic diagram ofan HPLC-ICP-IDMS system which is applied forspecies-specific but also for species-unspecific cal-ibration [21]. An HPLC system, including a highpressure pump, sample injection valve, guard andseparation columns, and a UV flow-through detec-tor, is used for coupling liquid chromatography

with ICP-MS. All different types of separationcolumns can be applied, such as size exclu-sion chromatography, normal and reversed-phasechromatography, and ion chromatography. If thespecies-specific isotope dilution mode is applied,the isotope-diluted sample is injected into the sys-tem, separated and then the separated elementalspecies are directly introduced into the ICP-MSinstrument for measuring the isotope ratio R of allseparated fractions. In this case the isotope ratio ofa peak containing only a single elemental specieshas identical values at all peak positions. It istherefore possible to determine the isotope ratio,which can be directly converted into a concentra-tion using equation (7.1.4) with the correspondingisotope intensities at a single peak position or, forbetter precision and accuracy, by measuring thetotal peak area of the reference and spike isotope.If IDMS calibration is carried out in the species-unspecific spiking mode, the spike solution is con-tinuously added by a pump to the separated frac-tions after they have passed the UV detector. Forprecise species determinations it is very importantthat the pump produces a constant spike flow with-out pulsation. The spike flow is calibrated by a

Unspiked sample

Sample injection valve

Standard injectionfor calibration of

spike flow

ICP-MSUV detector

Units additionally used for species-unspecificspiking mode

Pump for the continuousaddition of a species-

unspecific spike

HPLC pump

Guard column

Separation column

Sample spiked with correspondingspecies-specific spikes

Figure 7.1.5. Schematic figure of an HPLC-ICP-IDMS system for elemental species analysis by species-specific andspecies-unspecific spiking modes.

558 CALIBRATION

corresponding standard solution injected by a sep-arate valve into the system. The separated elemen-tal species and the spike solution are completelymixed within a Y-shaped connection capillary andthen the isotope diluted fractions are introducedinto the ICP-MS instrument.

The isotope ratio R is measured at the appropri-ate retention time as represented in the top part ofFigure 7.1.6 for a separated copper species whenusing a 65Cu-enriched spike solution. So long asno copper species are eluted from the separationcolumn the isotope ratio of the spike solution ismeasured. If a separated copper species appears themeasured isotope ratio shifts towards the naturalisotopic composition, which is proportional to theamount of the copper species. Using the calibrated

spike flow and the known eluent flow the isotoperatio chromatogram can be directly converted intoa mass flow chromatogram, which is shown at thebottom of Figure 7.1.6. More details about thisHPLC-ICP-IDMS system and the mathematicalconversion of the isotope ratio chromatogram intoa mass flow chromatogram are given in ref. 21.This type of calibration offers the possibility ofcarrying out analysis by IDMS even for elemen-tal species which are not sufficiently character-ized. It also allows the determination of ‘real-timeconcentrations’ of chromatographically separatedelemental species. These advantages make HPLC-ICP-IDMS one of the most powerful analyticaltools for the quantification of trace amounts of ele-mental species.

0

1

2

3

4

5

0 10 20 30


6

8

10

12

14spike isotope ratio

isotope ratiochromatogram

mass flowchromatogram

65C

u/63

Cu

ratio

mas

s flo

w C

u [p

g/s]

Figure 7.1.6. Measured isotope ratio chromatogram of a copper species by HPLC-ICP-IDMS with the species-unspecific spikingmode and its transformation into a mass flow chromatogram.


The HPLC-ICP-IDMS system represented inFigure 7.1.5 was used in connection with an ionchromatographic column and the species-specificcalibration mode, for example, to determine iodideand iodate in aquatic samples [26]. Fractions ofheavy metal complexes with humic substances,separated by size exclusion chromatography, havebeen quantified by species-unspecific IDMS cal-ibration [21, 24, 26]. Recently, capillary elec-trophoresis was also successfully coupled withICP-IDMS using species-unspecific spiking for thedetermination of sulfur in metallothionein frac-tions [25]. The corresponding mass flow chro-matogram of a rabbit liver sample is representedby Figure 7.1.7. In this case the known amino acidsequence of the metallothionein fractions MT1 andMT2 also allowed calculation of the amount ofmetallothionein in these fractions by measuring theamount of sulfur.

Recently, electrothermal vaporization (ETV)was coupled with ICP-IDMS using species-unspecific calibration for the determination ofmercury species in biological samples [27]. ThisETV-ICP-IDMS system allows the direct deter-mination of mercury species by solid samplingwithout dissolution of the sample, which is agreat step forward with respect to a time-effectivespecies analysis at a low contamination level. Thetemperature program which was used evaporatedmonomethylmercury at a much lower tempera-ture than inorganic mercury so that these two

mercury species could be separated by the ETVsystem. Species-unspecific calibration was carriedout with 200Hg-enriched elementary mercury (Hg0)by mixing a well-defined gas flow of the ele-mentary mercury spike, produced by an exactlytempered permeation tube, with the gaseous mer-cury species emitted by the ETV system. Whenmethylmercury and inorganic mercury in a bio-logical reference material (TORT-2) were mea-sured with this ETV-ICP-IDMS system, the resultsobtained agreed very well with the certified values.

4.2.3 Validation of analytical proceduresby the isotope dilution technique

An intensive discussion of possible systematicerrors in analyses for elemental species arose aftertransformation of mercury species during sampletreatment was first identified by using the isotopedilution technique with isotopically enriched Hg2+ions [14]. By determining the distilled MeHg+species after ethylation with NaBEt4 and sub-sequent analysis by GC-ICP-MS, it was foundthat substantial amounts of methylmercury canbe formed during the distillation process fromHg2+. The resulting overestimation of methylmer-cury in sediments was as high as 80 %. This clearlydemonstrated that validation of analytical methodsmust always be carried out to guarantee accurateresults in elemental species analysis. In contrast to

0

400

800

1200

1600

MT1: 13.9 ng(= 1.3 ng S)

MT2: 10.7 ng(= 1.0 ng S)

F1

F2


met

allo

thio

nein

[pg/

s]

0 3 6 9 12

Figure 7.1.7. Mass flow chromatogram of sulfur from separated metallothioneins of a rabbit liver by CE-ICP-IDMS (MT1 andMT2 are known metallothionein fractions with 20 cysteine units; the mass flow of sulfur could therefore be used to calculated thecorresponding amount of protein; F1 and F2 are unknown metallothionein fractions and the first peak at about 3.5 min migrationtime may be due to a sulfur impurity).

560 CALIBRATION

other atom spectrometric methods only mass spec-trometry is able to verify such a transformationof elemental species during the sample treatment.The isotope dilution technique has therefore beenincreasingly applied during the last few years tovalidate analytical procedures for possible speciestransformations. As a result it was also found thatmethylmercury is transformed to elementary mer-cury during ethylation by NaBEt4 in the presenceof halide ions [8, 28].

Speciation of Cr(VI) and Cr(III) in solid envi-ronmental samples is a great challenge becausebidirectional species transformation is possibleduring recommended sample treatment procedures,e.g. those by EPA method 3060A [29]. To deter-mine the degree of transformation of Cr(VI) intoCr(III) and also the opposite reaction during sam-ple pretreatment, a double-spiking IDMS methodwas developed using a 53Cr(VI)-enriched and a50Cr(III)-enriched spike. Recently possible trans-formations of butyltin species were also investi-gated using the IDMS technique [30].

Even if the IDMS technique is able to iden-tify possible transformations of elemental speciesduring sample treatment procedures and is there-fore able to offer validation of analytical meth-ods, the question nevertheless remains whetherunder these conditions IDMS calibration resultsin accurate analytical data or not. The determi-nation of MeHg+ in a river water sample byapplying a GC-ICP-IDMS system is used to showthe corresponding result [8]. After the samplewas spiked with 201Hg-enriched MeHg+ (iso-tope ratio 201Hg/202Hg = 6.45), monomethylmer-cury was converted into the volatile MeEtHgby reaction with NaBEt4. Under these condi-tions Hg2+ ions were also ethylated to becomevolatile Et2Hg. After a purge and trap procedurethe volatile mercury species were then separatedusing a capillary GC column. The correspond-ing chromatograms of the isotopes 201Hg (spikeisotope) and 202Hg (reference isotope), detectedwith a quadrupole ICP-MS instrument, are rep-resented at the top of Figure 7.1.8. The MeHg+peak shows an isotope ratio 201Hg/202Hg some-what between that of the spike solution and thenatural isotopic ratio of 0.44, which is due to

the presence of monomethylmercury in the riverwater sample. Also the Hg0 peak does not showthe natural isotopic composition, which is a clearindication that transformation of methylmercuryinto elementary mercury took place. By apply-ing NaBPr4 for propylation instead of ethylationof MeHg+, no transformation of methylmercuryinto elementary mercury could be observed (bot-tom of Figure 7.1.8). However, by evaluating themeasured isotope ratio of the MeHg+ peak forethylation and propylation using equation (7.1.3)identical analytical results within the limits of errorof (3.8 ± 0.1) ng mL−1 and (3.6 ± 0.1) ng mL−1,respectively, were obtained. This means that trans-formation of methylmercury during the ethylationprocess has no effect on the analytical result.This is due to the fact that loss of methylmer-cury during the ethylation process takes placeonly after a total equilibration between the spikeand sample species, which demonstrates the greatadvantage of an IDMS calibration for elementalspecies analysis.

5 CONCLUSION

External as well as internal calibration can beapplied for elemental species analysis. Even ifsome atom spectrometric detectors can also becalibrated under certain conditions by elemen-tal species not identical with the species to bedetermined, the best way to obtain accurate ana-lytical results is calibration with the elementalspecies of the analyte. This is absolutely nec-essary in all cases where separation techniquesare involved in the analytical method such asfor hyphenated techniques most frequently appliedin elemental speciation. Internal calibration usu-ally results in more precise and accurate resultsthan does external calibration and can be car-ried out either by the standard addition or bythe IDMS method. IDMS has the advantages ofbeing a one-point calibration method and thatloss of the analyte after the isotope dilutionstep usually has no effect on the result. How-ever, IDMS needs the application of an isotopi-cally enriched spike so that for more frequent

REFERENCES 561

0

50

100

150

200

250201

202

0

50

100

150

200

250

0 2 4 6 8 10

inte

nsity

[×10

3 cps

]

propylation

ethylation

202Hg

201Hg

201Hg

201Hg

202Hg202Hg

MeHg+

(201/202 = 1.61)

MeHg+

(201/202 = 1.69)


Hg2+

Hg2+

Hg0

Hg0

(201/202 = 1.31)

Figure 7.1.8. GC-ICP-IDMS chromatograms of a river water sample for methylmercury determination by 201Hg-enriched MeHg+after derivatization with sodium tetraethylborate and sodium tetrapropylborate, respectively. (For all elemental species deviatingfrom the natural isotopic ratio 201Hg/202Hg = 0.44 the corresponding value is given in brackets.)

applications today’s total lack of commerciallyavailable isotope-labeled elemental species mustbe overcome.

6 REFERENCES

1. Marshal, A., Calibration of Chemical Analysis and Useof Certified Reference Materials, Draft ISO Guide 32,ISO/REMCO N 262, International Organization of Stan-dardization, Geneva, 1993.

2. Skoog, D. A., West, D. M. and Holler, F. J., AnalyticalChemistry – an Introduction , Saunders College Publish-ing, Philadelphia, PA, 1994, p. 431.

3. Heumann, K. G., Isotope dilution mass spectrometry inInorganic Mass Spectrometry , Adams, F., Gijbels, R. and

van Grieken, R. (Eds), John Wiley, & Sons, Inc., NewYork, 1988, pp. 301–376.

4. Heumann, K. G. and Grosser, R., Fresenius’ J. Anal.Chem., 332, 880 (1989).

5. Reifenhauser, C. and Heumann, K. G., Fresenius’ J.Anal. Chem., 336, 559 (1990).

6. Brown, A. A., Ebdon, L. and Hill, S. J., Anal. Chim.Acta., 286, 391 (1994).

7. Snell, J. P., Steward, I. I., Sturgeon, R. E. and Frech, W.,J. Anal. At. Spectrom., 15, 1540 (2000).

8. Demuth, N. and Heumann, K. G., Anal. Chem., 73, 4020(2001).

9. Leermakers, M., Lansens, P. and Baeyens, W., Frese-nius’ J. Anal. Chem., 336, 655 (1990).

10. Dyg, S., Cornelis, R., Griepink, B. and Verbeeck, P.,Stability study of Cr(III) and Cr(VI) in water for

562 CALIBRATION

production of an aqueous chromium reference material, inMetal Speciation in the Environment , Broekaert, J. A. C.,Gucer, S. and Adams, F. (Eds), NATO ASI Series G:Ecological Sciences, Vol. 23, Springer, Berlin, 1990,pp. 361–376.

11. Nusko, R. and Heumann, K. G., Anal. Chim. Acta, 286,283 (1994).

12. Vogl, J., Iodine species analysis in aquatic systems byHPLC/ICP-MS and application of on-line isotope dilutiontechnique, Diploma Thesis, University of Regensburg,Germany, 1994.

13. Diemer, J. and Heumann, K. G., Fresenius’ J. Anal.Chem., 357, 74 (1997).


15. Pongratz, R. and Heumann, K. G., Anal. Chem., 68, 1262(1996).

16. Helfrich, A. and Heumann, K. G., unpublished.17. De Bievre, P., Fresenius’ J. Anal. Chem., 350, 277 (1994).18. Smith, D. H., Isotope dilution mass spectrometry, in

Inorganic Mass Spectrometry – Fundamentals and Appli-cations , Barshick, C. M., Duckworth, D. C. and Smith,D. H. (Eds), Marcel Dekker, New York, 2000 pp. 223–240.

19. Rosman, K. J. R. and Taylor, P. D. P., Pure Appl. Chem.,70, 217 (1998).

20. Heumann, K. G., Gallus, S. M., Radlinger, G. andVogl, J., J. Anal. At. Spectrom., 13, 1001 (1998).

21. Rottmann, L. and Heumann, K. G., Fresenius’ J. Anal.Chem., 350, 221 (1994).

22. Gallus, S. M., Development of a GC/ICP-MS system forselenium species determinations by isotope dilution massspectrometry, PhD Thesis, University of Regensburg,Germany, 1998.


24. Heumann, K. G., Gallus, S. M., Radlinger, G. andVogl, J., Spectrochim. Acta , B53, 273 (1998).

25. Schaumloffel, D., Prange, A., Marx, G., Heumann, K. G.and Bratter, P., Anal. Bioanal. Chem., 372, 155 (2002).


27. Gelaude, I., Dams, R., Resano, M., Vanhaecke, F. andMoens, L., Anal. Chem., 74, 3833 (2002).

28. Lambertsson, L., Lundberg, E., Nilsson, M. andFrech, W., J. Anal. At. Spectrom., 16, 1296 (2001).

29. Huo, D. and Kingston, H. M., Anal. Chem., 72, 5047(2000).

30. Alonso, J. I. G., Encinar, J. R., Gonzalez, P. R. andSanz-Medel, A., Anal. Bioanal. Chem., 373, 432 (2002).

Date post:	24-Jan-2021
Category:	Documents
Upload:	others
View:	3 times
Download:	0 times

Handbook of Elemental Speciation: Techniques and...

Documents