Methods of Analysis by the U.S. Geological Survey National ... · USGS U.S. Geological Survey <...

Methods of Analysis by the U.S. Geological SurveyNational Water Quality Laboratory— Determination ofWhole-Water Recoverable Arsenic, Boron, and VanadiumUsing Inductively Coupled Plasma–Mass Spectrometry

Open-File Report 99–464

U.S. Department of the InteriorU.S. Geological Survey

Methods of Analysis by the U.S. Geological SurveyNational Water Quality Laboratory— Determination ofWhole-Water Recoverable Arsenic, Boron, and VanadiumUsing Inductively Coupled Plasma–Mass Spectrometry

By John R. Garbarino

U.S. GEOLOGICAL SURVEY

Open-File Report 99–464

Denver, Colorado2000

U.S. DEPARTMENT OF THE INTERIOR

BRUCE BABBITT, Secretary

U.S. GEOLOGICAL SURVEY

Charles G. Groat, Director

The use of brand, firm, and trade names in this report is for identificationpurposes only and does not constitute endorsement by the U.S. Government.

For additional information write to: Copies of this report can be purchased from:

U.S. Geological Survey U.S. Geological SurveyChief, National Water Quality Laboratory Branch of Information ServicesBox 25046, Mail Stop 407 Box 25286Federal Center Federal CenterDenver, CO 80225-0046 Denver, CO 80225-0286

Contents III

CONTENTS

Abstract ........................................................................................................................ 1Introduction ................................................................................................................... 1Analytical method.......................................................................................................... 2

Application............................................................................................................... 2Summary of method ................................................................................................. 3Interferences............................................................................................................. 3Apparatus, instrumentation, and operating conditions ............................................... 4Reagents and calibration standards ........................................................................... 4Sample preparation................................................................................................... 5Analytical procedure ................................................................................................ 5Calculations.............................................................................................................. 6Reporting results ...................................................................................................... 6

Discussion of results ...................................................................................................... 6Results for Standard Reference Material ................................................................... 6Spike recoveries in natural-water samples ................................................................ 8Comparison of inductively coupled plasma–mass spectrometry to formermethods of analysis .................................................................................................. 9

Conclusions ................................................................................................................... 14References cited ............................................................................................................. 14

FIGURES

1–3. Graphs showing statistical results for natural whole-water in-bottle digestates frominductively coupled plasma–mass spectrometry and former methods of analysis:

1. Statistical analysis of arsenic results for 50 natural whole-water in-bottledigestates from inductively coupled plasma–mass spectrometry and hydridegeneration–atomic absorption spectrophotometry.......................................... 11

2. Statistical analysis of boron results for 66 natural whole-water in-bottledigestates from inductively coupled plasma–mass spectrometry andinductively coupled plasma–atomic emission spectrometry........................... 12

3. Statistical analysis of vanadium results for 21 natural whole-waterin-bottle digestates from inductively coupled plasma–mass spectrometryand inductively coupled plasma–atomic emission spectrometry .................... 13

TABLES

1. Inorganic constituents and codes......................................................................... 22. Former methods and inductively coupled plasma–mass spectrometric method

detection limits for new elements determined in whole-water digestate............... 33. Statistical analysis of long-term inductively coupled plasma–mass

spectrometric results for U.S. Geological Survey’s Standard ReferenceWater Sample T145 ............................................................................................ 7

4. Short-term analytical variability as a function of elemental concentrationfor inductively coupled plasma–mass spectrometry............................................. 7

Determination of Whole-Water Recoverable Arsenic, Boron, and VanadiumUsing Inductively Coupled Plasma –Mass Spectrometry

IV

TABLES— Continued

5. Average percent spike recoveries in reagent-water, surface-water, ground-water, and in-bottle digest matrices by using inductively coupled plasma–massspectrometry....................................................................................................... 8

6. Chemical characteristics of natural-water samples used to evaluate inductivelycoupled plasma–mass spectrometry .................................................................... 9

7. Statistical analysis of inductively coupled plasma–mass spectrometry andformer methods of analysis for the determination of whole-water recoverablearsenic, boron, and vanadium.............................................................................. 10

Conversion Factors and Abbreviated Water-Quality Units V

CONVERSION FACTORS AND ABBREVIATED WATER-QUALITY UNITS

Multiply By To obtainliter (L) 2.64 x 10-1 gallon

microgram (µg) 3.53 x 10-8 ounce, avoirdupoismilligram (mg) 3.53 x 10-5 ounce, avoirdupoismilliliter (mL) 2.64 x 10-4 gallon

Degree Celsius (°C) may be converted to degree Fahrenheit (°F) by using the followingequation:

°F = 9/5 (°C) + 32.

Abbreviated water-quality units used in this report are as follows:

mg/L milligram per literµg/L microgram per literµS/cm microsiemens per centimeter at 25°C

Other abbreviations also used in this report:

amu atomic mass unitASTM American Society for Testing and MaterialsDCP–AES direct current plasma–atomic emission spectrometryFEP fluorinated ethylene propylene (Teflon)GF–AAS graphite furnace–atomic absorption spectrophotometryHCl hydrochloric acidHG–AAS hydride generation–atomic absorption spectrophotometryHNO3 nitric acidICP–AES inductively coupled plasma–atomic emission spectrometry, also known as

inductively coupled plasma–optical emission spectrometry (ICP–OES)ICP–MS inductively coupled plasma–mass spectrometryMDL(s) method detection limit(s)MRL(s) minimum reporting level(s)MPV(s) most probable value(s)NWQL National Water Quality LaboratorySRWS(s) U.S. Geological Survey Standard Reference Water Sample(s)USGS U.S. Geological Survey< less than≤ less than or equal to± plus or minus


VI

Definitions:

MDL The method detection limit (MDL) is defined as the minimum concentration of anelement that can be measured and reported with 99-percent confidence that theconcentration is greater than zero and is determined from analysis of a sample in agiven matrix that contains the element of interest (U.S. Environmental ProtectionAgency, 1997).

MPV The most probable value (MPV) is equal to the median value for numerousinterlaboratory analyses that use multiple analytical methods.

Whole-water recoverable. Pertains to the constituents in solution after a representativewater-suspended-sediment sample is digested (usually by using dilute acid solution).Complete dissolution of particulate matter often is not achieved by the digestiontreatment, and thus the determination represents something less than the “total”amount (that is, less than 95 percent) of the constituent present in the dissolved andsuspended phases of the sample. Equivalent digestion procedures would be requiredof all laboratories that perform such analyses to achieve comparability of analyticaldata, because different digestion procedures are likely to produce different analyticalresults.

Introduction 1

Methods of Analysis by the U.S. Geological SurveyNational Water Quality Laboratory— Determinationof Whole-Water Recoverable Arsenic, Boron, and VanadiumUsing Inductively Coupled Plasma–Mass Spectrometry

By John R. Garbarino

ABSTRACTAnalysis of in-bottle digestate by using

the inductively coupled plasma–massspectrometric (ICP–MS) method has beenexpanded to include arsenic, boron, andvanadium. Whole-water samples aredigested by using either the hydrochloric acidin-bottle digestion procedure or the nitric acidin-bottle digestion procedure. When thehydrochloric acid in-bottle digestionprocedure is used, chloride must be removedfrom the digestate by subboiling evaporationbefore arsenic and vanadium can beaccurately determined. Method detectionlimits for these elements are now 10 to 100times lower than U.S. Geological Survey(USGS) methods using hydride generation–atomic absorption spectrophotometry (HG–AAS) and inductively coupled plasma–atomic emission spectrometry (ICP–AES),thus providing lower variability at ambientconcentrations. The bias and variability ofthe methods were determined by using resultsfrom spike recoveries, standard referencematerials, and validation samples. Spikerecoveries in reagent-water, surface-water,ground-water, and whole-water recoverablematrices averaged 90 percent for sevenreplicates; spike recoveries were biased from25 to 35 percent low for the ground-watermatrix because of the abnormally high ironconcentration. Results for reference materialwere within one standard deviation of themost probable value. There was no

significant difference between the resultsfrom ICP–MS and HG–AAS or ICP–AESmethods for the natural whole-water samplesthat were analyzed.

INTRODUCTION

The U.S. Geological Survey (USGS)National Water Quality Laboratory (NWQL)offers several methods for the determinationof recoverable arsenic, boron, and vanadiumin whole-water samples. Former USGSmethods use single-element quantification,such as graphite furnace– (GF–AAS), andhydride generation–atomic absorptionspectrophotometry (HG–AAS), direct currentplasma–atomic emission spectrometry (DCP–AES), or the simultaneous multielementtechnique of inductively coupled plasma–atomic emission spectrometry (ICP–AES).This report provides data that validates theaddition of the aforementioned elements tothe existing inductively coupled plasma–massspectrometric (ICP–MS) method.

Elements that are being added to theexisting multielement ICP–MS method andtheir corresponding former USGS methods ofanalysis are listed in the following table.Arsenic, boron, and vanadium have beenadded to method I-4471-97 (see Garbarinoand Struzeski, 1998).

Determination of Whole-Water Recoverable Arsenic, Boron, and VanadiumUsing Inductively Coupled Plasma–Mass Spectrometry

2

Element GF–AAS

HG–AAS

DCP–AES

ICP–AES

Arsenic 4 4Boron 4 4Vanadium 4

ICP–MS is compared to one formermethod from the list in the preceding table.In all comparisons, the most current (as ofJanuary 1998) former method is used.Whole-water recoverable arsenic by ICP–MSis compared to the former HG–AAS method.The digestion of whole-water samples forrecoverable arsenic by the former HG–AASmethod uses an online sulfuricacid/potassium persulfate digestion. Incontrast, the ICP–MS method uses an in-bottle digestion procedure to determinewhole-water recoverable concentrations.Whole-water recoverable boron andvanadium by ICP–MS is compared to theformer ICP–AES method.

Ambient concentrations of recoverablearsenic and vanadium cannot be accuratelymeasured by ICP–MS in digestates preparedby using the hydrochloric acid (HCl) in-bottledigestion procedure (Hoffman and others,1996) because of interferences from chloride.When whole-water recoverable arsenic andvanadium are being determined by ICP–MS,either the HCl must be removed from the HClin-bottle digest through subboiling evapora-tion (see Appendix in Garbarino andHoffman, 1999) or nitric acid (HNO3) in-bottle digestion must be used (Garbarino andHoffman, 1999).

The expanded method was developedby the USGS for use at the NWQL. Theexpanded method supplements other officialUSGS inorganic methods (Fishman, 1993;Fishman and Friedman, 1989; Garbarino, andStruzeski, 1998). The new elements will beavailable in the whole-water recoverableICP–MS schedules.

ANALYTICAL METHOD

Application

The determination of whole-waterrecoverable arsenic, boron, and vanadiumhave been added to ICP–MS methodI-4471-97. Details of this method areprovided in Garbarino and Struzeski (1998).Whole water is digested using either the HClin-bottle procedure described by Hoffmanand others (1996) or the HNO3 in-bottleprocedure described by Garbarino andHoffman (1999). Whenever whole-waterrecoverable arsenic and vanadium are beingdetermined in the HCl in-bottle digest,however, the HCl must be removed from thedigest by subboiling evaporation (seeAppendix in Garbarino and Hoffman, 1999).The new laboratory code, parameter code,method code, and reporting unit for eachelement are listed in table 1.

Table 1. Inorganic constituents and codes

Metals, Acid Digestion, Whole-WaterRecoverable Method I-4471-97

[µg/L, microgram per liter]

Element Lab codeParameter

and methodcodes

Arsenic, µg/L 2500 01002F

Boron, µg/L 2501 01022C

Vanadium, µg/L 2502 01087B

The short-term method detection limits(MDLs) and analytical concentration rangesare listed in table 2. The elemental lineardynamic range for ICP–MS is greater than1 mg/L when calibrating both the pulse andanalog measurement modes. Short-termMDLs were calculated by using U.S.Environmental Protection Agency’s (1997)definition and represent pooled averages onthe basis of four MDLs determined on

Analytical Method 3

Table 2. Former methods and inductively coupled plasma–mass spectrometric method detection limitsfor new elements determined in whole-water digestate

[All concentrations are in micrograms per liter; MRL, minimum reporting level; MDL, method detection limit;DCP–AES, direct current plasma–atomic emission spectrometry; GF–AAS, stabilized temperature graphite furnace–atomic absorption spectrophotometry; HG–AAS, hydride generation–atomic absorption spectrophotometry; ICP–AES, inductively coupled plasma–atomic emission spectrometry; ICP–MS, inductively coupled plasma–massspectrometry]

Former methods ICP–MS

Element Technique MRLUpper

concentration limit(without dilution)

Short-termMDL

Arsenic GF–AAS 0.9 50 0.07HG–AAS 1 20

Boron DCP–AES 10 10,000 1 0.5ICP–AES 13 10,000

Vanadium ICP–AES 10 10,000 0.081Method detection limit for boron is limited by reagent blank concentration.

different days over several weeks. TheMDLs have not been established for formermethods, therefore, minimum-reportinglevels (MRLs) are listed that are probablywithin a factor of 5 of the MDL.

Summary of Method

The ICP–MS method has beendescribed previously in Garbarino andStruzeski (1998). The following sectionsonly provide additional information specificto the elements that are being added to themethod.

Interferences

Physical and spectral interferenceassociated with the determination of arsenic,boron, and vanadium by ICP–MS aredocumented in Horlick and Shao (1992),Garbarino and Taylor (1994), Garbarino andStruzeski (1998), and Garbarino (1999).

3.1 Physical interferences. Internalstandards are used to minimize effects fromsample transport, instrumental drift, and

matrix-induced fluctuations in plasmacharacteristics. Typical internal standardelements are 72Ge+, 115In+, and 209Bi+ (seeGarbarino and Struzeski, 1998; Garbarino,1999). Alternative isotopes may besubstituted after ensuring that there are nospectral interferences associated with thenew selections.

Memory effects related to sampletransport are negligible for most elementsnormally present in whole-water digestate.Carryover from samples that have arsenicand boron concentrations less than or equalto 200 µg/L is negligible when the sampleintroduction described in Garbarino andStruzeski (1998) is used. Vanadium,however, did not recover to reagent-blankintensity levels within the rinse period;vanadium intensities were 2 times greaterthan reagent-blank level. Fortunately,vanadium concentrations rarely exceed 100µg/L. Nevertheless, the analyst must reviewall analytical results to ensure that errorsfrom carryover are minimized.


4

Sample matrix composition could alsoaffect the bias and variability of ICP–MSdeterminations. The use of internalstandardization compensates for most matrixeffects, however, some matrix interferencesremain problematic. Matrix compositioncan suppress the ionization efficiency of theplasma and result in negatively biasedelemental concentrations, especially forlighter elements (see Garbarino, 1999). Thissuppression can be significant for whole-water matrices because of the level ofdissolved solid concentrations.

3.2 Spectral interferences. There areno isobaric interferences on arsenic, boron,or vanadium. Chloride-associatedmolecular-ion interferences have beendocumented on arsenic (75As+) andvanadium (51V+). The molecular ions40Ar35Cl+ and 35Cl16O+ interfere with 75As+

and 51V+, respectively. The standardequation used to correct for chlorideinterference on arsenic uses a term based ona measurement made at 82 amu. If thebromide concentration in a sample exceedsabout 100 µg/L, the standard correctionequation will give positively biased arsenicresults. All the equations used to correctmolecular-ion interference on arsenic andvanadium are described by Garbarino(1999). Interference corrections on arsenicand vanadium have been shown to beacceptable for sample solutions that havechloride concentrations less than 5,000mg/L. There are no significant interferenceson arsenic, boron, or vanadium from doublycharged ions.

Apparatus, Instrumentation, andOperating Conditions

Instrumentation previously describedin a whole-water recoverable method byGarbarino and Struzeski (1998) has beenreplaced with a Perkin-Elmer Elan 6000ICP–MS. Details of instrument operation

are fully documented in the NWQLStandard Operating Procedure IM0011.1(T.M. Struzeski, U.S. Geological Survey,written commun., 1998) and by Garbarino(1999). The standard Perkin-Elmer cross-flow nebulizer and spray chamber is used tointroduce samples. The cross-flow nebulizerresists clogging (≤ 0.5 percent totaldissolved solids) and is chemically inert.Other nebulizer designs can be used butmust be resistant to clogging and capable ofproviding MDLs that are within a factor oftwo of those listed in table 2.

Primary isotopes used to determinerecoverable concentrations for the newelements are as follows (amu, atomic massunit):

Arsenic 75 amu Boron 11 amuVanadium 51 amu

Reagents and Calibration Standards

A report by Garbarino and Struzeski(1998) describes the preparation ofcalibration standards, internal standardsolution, and the performance checksolution. New elements are calibrated byincluding them in the multielementstandards described in method I-4471-97 atthe same concentrations. Multielementstandards are prepared in a matrix that ismatched to the acid concentration of thesamples being analyzed. Arsenic andvanadium must not be included in amultielement standard that contains bromideor chloride. The interference solution hasbeen added to verify the accuracy ofinterference corrections on arsenic andvanadium.

ASTM Type I reagent water(American Society for Testing andMaterials, 1995, p. 122–124), spectroscopicgrade commercial standards, and ultrapureacids must be used to prepare all solutions.

Analytical Method 5

All percentages represent volume-to-volumeratios. All concentrated acids andcommercial standards must be verified tocontain concentrations of concomitantelements that are less than the MDLs afterthe prescribed dilution. Every solution mustbe stored in a designated FEP Teflon bottle.Use clean Type A glass volumetric flasks toprepare all solutions except for thosecontaining boron. Regularly verify theaccuracy of all pipets and volumetric flasksfor preparing standard solutions.

5.1 Nitric acid (HNO3):Concentrated, specific gravity 1.41.

5.2 Hydrochloric acid (HCl):Concentrated, specific gravity 1.19.

5.3 Calibration blank: Reagent wateris acidified to either 3 percent HNO3 whenusing the HNO3 in-bottle procedure or 1percent HNO3 when using the HCl in-bottleprocedure.

5.4 Commercial single-elementstandard solutions, 1.00 mL = 10 mgpreserved in HNO3 for each of thefollowing: As, B (in water), and V.

Chloride: 1.00 mL = 100 mg in water andBromide: 1.00 mL = 1.0 mg in water

5.6 Multielement stock solution I, 1.00mL = 0.010 mg of As, B (in water), and V:Dilute 1.0 mL of each commercial single-element standard to 1,000 mL in a volumetricflask with 1 percent HNO3. Store in a cleandesignated FEP Teflon bottle.

5.7 Multielement calibration standard I,1.00 mL = 0.025 µg of As, B (in water), andV: Include new elements in an existingcalibration standard by diluting 0.250 mL ofmultielement stock solution I in a 100-mLvolumetric flask with the suitable calibrationblank.

5.8 Multielement calibration standardII, 1.00 mL = 0.100 µg of As, B (in water),and V: Include new elements in an existingcalibration standard by diluting 1.0 mL ofmultielement stock solution I in a 100-mLvolumetric flask with the suitable calibrationblank.

5.9 Multielement calibration standardIII, 1.00 mL = 0.200 µg of As, B (in water),and V: Include new elements in an existingcalibration standard by diluting 2.0 mL ofmultielement stock solution I in a 100-mLvolumetric flask with the suitable calibrationblank.

5.10 Interference check standard,1.00 mL = 0.50 mg Cl, 0.005 mg Br, and0.025 µg As and V: Dilute 5.0 mL of thecommercial chloride standard solution,0.50 mL of commercial bromide standardsolution, and 2.5 mL of multielement stocksolution I in a 1,000-mL volumetric flaskwith the suitable calibration blank. Store ina clean designated FEP Teflon bottle.

Sample Preparation

Whole-water recoverable arsenic,boron, and vanadium. Nonfiltered, acidifiedwhole-water samples analyzed by ICP–MSfor recoverable arsenic, boron, vanadium,and other elements must be digested byusing either the HCl in-bottle procedure(Hoffman and others, 1996) or the HNO3 in-bottle procedure (Garbarino and Hoffman,1999). When the HCl in-bottle digestion isused, the HCl must be removed bysubboiling evaporation if arsenic andvanadium are determined (see Appendix inGarbarino and Hoffman, 1999).

Analytical Procedure

Refer to Perkin Elmer (1997a, 1997b)and NWQL Standard Operating ProcedureIM0011.1 (T.M. Struzeski, U.S. Geological


6

Survey, written commun., 1998) for detailsof the analytical procedure. In addition,verify the accuracy of interferencecorrection equations by analyzing theinterference check standard (see section5.10) with every batch of samples. The Elansoftware automatically verifies that theresults meet acceptance criteria.

Calculations

No additional calculations are requiredin this method.

Reporting Results

The number of significant figuresreported varies with element and is afunction of concentration. Whenever theconcentration is less than the MDL for anelement, the result is reported as less thanthe MDL (< MDL). All other elementalresults should be reported by using thecriteria listed below. These criteria arebased on the uncertainty suggested in thefollowing Discussion of Results section.Alternatively, the variability in the meanconcentration could be used to establish theappropriate number of significant figures toreport for each individual sample matrix.This procedure would provide the mostaccurate estimate of the uncertaintyassociated with each sample.

For arsenic and vanadium—

• If the concentration is greater than theMDL, but less than 10 µg/L, reportresult to two decimal places.

• If the concentration is greater than 10µg/L, but less than 100 µg/L, reportresult to one decimal place.

• If the concentration is greater than 100µg/L, report result to threesignificant figures.

For boron—

• If the concentration is greater than orequal to the MDL, but less than 100µg/L, report result to one decimalplace.

• If the concentration is greater than 100µg/L, report result to threesignificant figures.

DISCUSSION OF RESULTS

The bias of the ICP–MS methods forthe determination of whole-waterrecoverable concentrations is established bycomparing results to former USGS methodsof analysis. ICP–MS method variability wasdetermined from replicate analyses over arange of elemental concentrations preparedin a calibration-blank matrix. All whole-water samples analyzed by ICP–MS wereprepared by using subboiling evaporation ofthe HCl in-bottle digestate (Garbarino andHoffman, 1999). Results for standardreference material, spiked samples, andwhole-water digestates are used to evaluatethe determination of recoverable arsenic,boron, and vanadium by ICP–MS. Allformer methods are U.S. Geological Surveyapproved methods (Fishman and Friedman,1989; Fishman, 1993).

Results for Standard Reference Material

U.S. Geological Survey SRWS T145was analyzed repetitively for 3 weeks todetermine the long-term bias and variabilityof the ICP–MS method (see table 3). Resultsfor all elements are within one standarddeviation of the most probable value (MPV).The average long-term variability is 4±2

Discussion of Results 7

percent for elemental concentrations rangingfrom 10 to 46 µg/L. The paired Student t-Test was used to test the null hypothesis thatthe ICP–MS method gives mean elementalconcentrations that are not significantlydifferent from the MPVs. P-values werecalculated for each element to provide alevel of confidence in accepting the nullhypothesis. The larger the p-value thegreater the confidence in accepting the nullhypothesis. When the p-value exceeds 0.05,the null hypothesis is acceptable at the 95-percent confidence level. The Student t-Testindicated that only the boron experimentalresults are not significantly different fromthe MPV. Nevertheless, the difference

between the MPV and the experimentalmean for arsenic and vanadium isanalytically insignificant.

The short-term variability of ICP–MSover an extended concentration range islisted in table 4. The variability is based onthree replicate determinations (an acquisi-tion time of about 1 minute) at eachelemental concentration in the calibrationblank matrix. The short-term variabilitywas less than or equal to 5 percent at 0.5µg/L for arsenic and vanadium; variabilityfor boron was about 5 percent at 1 µg/L.

Table 3. Statistical analysis of long-term inductively coupled plasma–mass spectrometric resultsfor U.S. Geological Survey’s Standard Reference Water Sample T145

[ICP–MS, inductively coupled plasma–mass spectrometry; element results are in micrograms per liter; MPV,the published most probable value; ±, the plus or minus symbol precedes the F-pseudosigma in the MPVcolumn and the standard deviation at 1σ in the experimental mean column; n, number of replicates used tocalculate the experimental mean; p-value, level of significance; <, less than]

Element MPV Experimentalmean, n=12 t-Test statistic p-value

Arsenic 10 ± 1 10.3 ± 0.2 9.00 <0.0001Boron 46 ± 6 45 ± 3 -0.23 0.8253Vanadium 12 ± 2 10.6 ± 0.5 -7.86 <0.0001

Table 4. Short-term analytical variability as a function of elemental concentration for inductivelycoupled plasma–mass spectrometry

[ICP–MS, inductively coupled plasma–mass spectrometry; %RSD, percent relative standard deviation on thebasis of three sequential determinations in a 0.4-percent solution of concentrated nitric acid in deionized water;<MDL, less than the method detection limit; nd, not determined]

Concentration, in micrograms per literElement 0.05

%RSD0.1

%RSD0.5

%RSD1.0

%RSD10

%RSD50

%RSD100

%RSD250

%RSDArsenic <MDL 30 3 4 2 nd 1 0.6Boron nd 10 nd 5 2 0.4 0.2 ndVanadium 8 10 2 0.6 0.7 nd 0.5 0.5


8

Spike Recoveries in Natural-WaterSamples

Spike recovery percentages listed intable 5 were determined for the newelements in matrices representative ofreagent-water, surface-water, ground-water,and whole-water digests. Seven replicaterecoveries at 5 to 10 times the MDL (thelow-level spike) and 75 µg/L (the high-levelspike) were determined in each matrix overa period of about 1 week. Averagerecoveries in the reagent-water matrixranged from 77 to 105. Recovery variabilityfor the low-level spike ranged from 4 to 9percent, depending on the element.

Recoveries of the high-level spike inthe surface-water matrix were similar tothose of the reagent-water matrix. Ambientconcentrations of vanadium in the surfacewater, however, hindered the recovery of thelow-level spike. Recovery of low-level

arsenic and boron ranged from 88 to 112percent. The variability in the recovery of1 µg/L boron in the presence of 40 µg/Lboron was 52 percent; the variability forarsenic was less than 6 percent at about thesame spike concentration.

The ground-water matrix used forspike recoveries was selected to examine theeffects of interferent species on elementaldeterminations. The ground water had arelatively high concentration of iron(340 mg/L) that exceeds the concentrationfound in over one-half of the whole-waterdigestates analyzed by NWQL. Ambientconcentrations of boron precluded therecovery of the low-level spike, however,recoveries for arsenic and vanadiumaveraged 70 percent. High-level spikerecoveries averaged 67±3 percent. The lowpercent recoveries for the lighter elementsconfirm the effects from ionizationsuppression.

Table 5. Average percent spike recoveries in reagent-water, surface-water, ground-water, andin-bottle digest matrices by using inductively coupled plasma–mass spectrometry

[µg/L, micrograms per liter; number following the plus or minus symbol (±) is the standard deviation on thebasis of seven determinations accrued on separate days; high spike, 75 µg/L for all elements; na, not applicablebecause the difference between the spike concentration and ambient concentration was greater than a factorof 10; <MDL, concentration is less than the method detection limit; %, percent]

Reagent-water matrix Surface-water matrix

Element

Low-spike,

inµg/L

Ambientconcen-tration,in µg/L

Low-spike

recovery,in %

High-spike

recovery,in %

Ambientconcen-tration,in µg/L

Low-spike

recovery,in %

High-spike

recovery,in %

Arsenic 0.5 <MDL 102±4 101±1 1 112±6 105±2Boron 1 <MDL 77±9 103±2 40 88±52 102±4Vanadium 0.6 <MDL 98±4 105±4 60 na 104±4

Ground-water matrix Synthetic whole-waterrecoverable digest matrix

Arsenic 0.5 1 75±8 68±4 7 100±10 95±22Boron 1 350 na 70±20 8 na 97±9Vanadium 0.6 <MDL 64±20 64±10 7 na 97±10


The whole-water digest matrix wasprepared by weighing 200 mg of NationalInstitute of Standards and Technology’s2704 Buffalo River sediment into 400 mL ofcalibration blank, digesting the mixtureusing the HCl in-bottle digestion, removingthe HCl by subboiling evaporation, andreconstituting the residue in 3 percentHNO3. The boron and vanadium low-levelspikes were not recovered because thematrix concentration exceeded theirconcentration by about a factor of 10; thearsenic recoveries averaged 100±10 percent.High-level spike recoveries for arsenic,boron, and vanadium were nearly 100percent with variability of less than or equalto 10 percent.

Comparison of Inductively CoupledPlasma–Mass Spectrometry toFormer Methods of Analysis

The ICP–MS results are compared toformer methods of analysis, such as hydridegeneration–atomic absorption spectro-photometry (HG–AAS) and inductivelycoupled plasma–atomic emissionspectrometry (ICP–AES). Whole-water

digest samples were selected from thepopulation of such samples submitted to theNWQL. The samples have a wide range ofelement concentrations and specificconductance. Surface-water and ground-water samples are included in the sampleset; the number of each type is aboutproportional to its fraction of the totalsubmitted for analysis during an averageyear. Other chemical characteristics thatoften influence the performance of analyticalmethods, such as sulfate and chlorideconcentrations, were also considered in theselection process (see table 6 for thechemical characteristics of the samples).

Results were evaluated by usingseveral different approaches. Because thedata extend over a wide concentration range,it is inappropriate to use the paired Studentt-test to evaluate the null hypothesis becauseerrors, whether random or systematic, areindependent of the concentration.Consequently, linear regression analysis isused to calculate the slope, y-intercept, andcoefficient of determination (R2) for theequation that describes the relation betweenICP–MS and a former USGS method. A

Table 6. Chemical characteristics of natural-water samples used to evaluate inductivelycoupled plasma–mass spectrometry

[ICP–MS, inductively coupled plasma–mass spectrometry; µg/L, microgram per liter; mg/L, milligramper liter; <MDL, less than the method detection limit; SC, specific conductance]

Element, in µg/L 25th percentile Median 75th percentile Maximum

Arsenic 1.7 2.9 7.4 104Boron 52 120 210 1,700Vanadium 1.8 4.7 9.7 1,530

ConstituentChloride, in mg/L 9.8 68 501 9,176SC, in µS/cm1 455 944 2,490 52,600Sulfate, in mg/L 108 309 1,306 16,832

1Specific conductance in microsiemens per centimeter at 25°C (µS/cm); includes measurements forwhole-water samples prior to digestion.


10

slope coefficient of one and a y-intercept ofzero indicate exact correlation. Thecorresponding p-values indicate the degreeof confidence in each coefficient. Box plotsalso are provided to show the distribution ofthe results for each method. Thenonparametric Wilcoxon Signed Rank Testalso is used to evaluate whether there is asignificant difference between results fromthe ICP–MS and former USGS methods.Data that were less than the highest MDL orMRL were omitted from the data set prior tostatistical analysis. Statistical analysisresults are summarized in the followingparagraphs and are listed in table 7. Resultsof the statistical tests are shown in figures 1through 3.

Linear regression results showacceptable correlation between ICP–MS andformer USGS methods for the determinationof whole-water recoverable arsenic, boron,and vanadium (see table 7). The p-valuesfor all the slope coefficients indicate that theslope is not significantly different from 1.0at the 95-percent confidence level. Inaddition, the y-intercept p-values for allelements indicate that the intercepts are notsignificantly different from zero. The p-values from the Wilcoxon Signed Rank Testsupport the linear regression results forarsenic and vanadium. The p-value forboron was probably influenced by severalhigh-level concentrations in selected whole-water digestates.

Table 7. Statistical analysis of inductively coupled plasma–mass spectrometry and former methods ofanalysis for the determination of whole-water recoverable arsenic, boron, and vanadium

[Coef., the slope coefficient of the regression line; Const., the y-intercept constant of the regression line; p-value, level ofsignificance; R2, coefficient of determination; HG–AAS, hydride generation–atomic absorption spectrophotometry; ICP–AES, inductively coupled plasma–atomic emission spectrometry; <, less than]

Slope y-interceptElement Former

methodCoef. p-value a Const. p-value b

R2

WilcoxonSigned

Rank Testp-value c

Arsenic HG–AAS 0.96 <0.0001 0.42 0.2895 0.959 0.1660Boron ICP–AES 0.89 <0.0001 -5.0 0.6206 0.927 <0.0001Vanadium ICP–AES 1.1 <0.0001 -1.8 0.6436 0.988 0.1138

aThe null hypothesis: slope is not equal to one.bThe null hypothesis: y-intercept is equal to zero.cThe null hypothesis: the difference in concentration between the new inductively coupled plasma–mass

spectrometric method and the former USGS method is equal to zero.


Figure 1. Statistical analysis of arsenic results for 50 natural whole-water in-bottle digestates frominductively coupled plasma–mass spectrometry and hydride generation–atomic absorptionspectrophotometry.

Percentile

10th25th

50th

75th90th

8.16

11.13

1.56

51

0

50.30

0

8.27

10.94

1.53

51

0.01

49.57

0

Mean

Std. Dev.

Std. Error

Count

Minimum

Maximum

# Missing

As by HG-AAS, in µg/L A s b y ICP-MS, in µg/L

Descript ive Stat ist ics

-10

0

10

20

30

40

50

60

As

by IC

P-M

S, i

n µg

/L

-10 0 10 20 30 40 50 60As by HG-AAS, in µg/L

R e g r e s s i o n P l o t

-10

0

10

20

30

40

50

60

As by HG-AAS, in µg/L A s b y ICP-MS, in µg/L

Box Plot

1

0

-1.39

0.1660

-1.39

0.1660

# 0 Dif ferences

# Ties

Z-Value

P-Value

Tied Z-Value

Tied P-Value

Wilcoxon S igned Rank Test for As by HG-AAS, in µg /L , As by ICP-MS, in µg /L

33 781.00 23.67

17 494.00 29.06

Count Sum Ranks Mean Rank

# Ranks < 0

# Ranks > 0

Wilcoxon Rank In fo for As by HG-AAS, in µg /L , As by ICP-MS, in µg /L

E X P L A N A T IONHydride generat ion-atomic absorpt ion spectrophotometry (HG-AAS);Inductively coupled plasma-mass spectrometry ( ICP-MS); microgramsper l i ter (µg/L); Std. Dev., standard deviat ion; Std. Error, standard error;<, less than; >, greater than; #, number

Line of regression


12

Figure 2. Statistical analysis of boron results for 66 natural whole-water in-bottle digestates frominductively coupled plasma–mass spectrometry and inductively coupled plasma–atomic emissionspectrometry.

209.72

248.34

30.57

66

14.90

1511.20

0

182.18

230.30

28.35

66

9.57

1612.00

0

Mean

Std. Dev.

Std. Error

Count

Minimum

Maximum

# Missing

B by ICP-AES, in µg/L B by ICP-MS, in µg/L

Descriptive Statistics

-200

0

200

400

600

800

1000

1200

1400

1600

1800

B b

y IC

P-M

S, i

n µg

/L

-200 0 200 400 600 800 1000 1200 1400 1600 1800B by ICP-A ES, in µg/L

Regression Plot

-200

0

200

400

600

800

1000

1200

1400

1600

1800

B by ICP-AES, in µg/L B by ICP-MS, in µg/L

Box Plot

0

0

-4.90

<0.0001

-4.90

<0.0001

# 0 Differences

# Ties

Z-Value

P-Value

Tied Z-Value

Tied P-Value

Wilcoxon Signed Rank Test for B by ICP-AES, in µg/L, B by ICP-MS, in µg/L

9 338.00 37.56

57 1873.00 32.86


# Ranks < 0

# Ranks > 0

Wilcoxon Rank Info for B by ICP-AES, in µg/L, B by ICP-MS, in µg/L

EXPLANATIONInductively coupled plasma-atomic emission spectrometry (ICP-AES);Inductively coupled plasma-mass spectrometry (ICP-MS); microgramsper liter (µg/L); Std. Dev., standard deviation; Std. Error, standard error;<, less than; >, greater than; #, number

Line of regression

Percentile

10th25th

50th

75th90th


Figure 3. Statistical analysis of vanadium results for 21 natural whole-water in-bottle digestates frominductively coupled plasma–mass spectrometry and inductively coupled plasma–atomic emissionspectrometry.

72.69

120.32

26.26

21

7.24

518.58

0

81.39

138.61

30.25

21

10.35

573.30

0

Mean

Std. Dev.

Std. Error

Count

Minimum

Maximum

# Missing

V by ICP-A ES, in µg/L V by ICP-MS, in µg/L

Descriptive Statistics

-100

0

100

200

300

400

500

600

V b

y IC

P-M

S, i

n µg

/L

-100 0 100 200 300 400 500 600V by ICP-AES, in µg/L

Regression Plot

-100

0

100

200

300

400

500

600

V by ICP-AES, in µg/L V by ICP-MS, in µg/L

Box Plot

0

0

-1.58

0.1138

-1.58

0.1138

# 0 Differences

# Ties

Z-Value

P-Value

Tied Z-Value

Tied P-Value

Wilcoxon Signed Rank Test for V by ICP-AES, in µg/L, V by ICP-MS, in µg/L

13 161.00 12.38

8 70.00 8.75


# Ranks < 0

# Ranks > 0

Wilcoxon Rank Info for V by ICP-AES, in µg/L, V by ICP-MS, in µg/L

EXPLANATIONInductively coupled plasma-atomic emission spectrometry (ICP-AES);Inductively coupled plasma-mass spectrometry (ICP-MS); microgramsper liter (µg/L); Std. Dev., standard deviation; Std. Error, standard error;<, less than; >, greater than; #, number

Line of regression

Percentile

10th25th

50th

75th90th


14

CONCLUSIONS

Results from reference material, spikerecoveries, and the analysis of natural-watersamples were used to evaluate the overallbias and variability of the determination ofwhole-water recoverable arsenic, boron, andvanadium by inductively coupled plasma–mass spectrometry (ICP–MS). All testresults provide an accurate estimate of theexpected analytical performance. Thefollowing list outlines the major conclusionsof this report. In addition to analyticalperformance comparisons, suggestions areprovided for selecting appropriatemethodology and the potential impact of theuse of ICP–MS on long-term trend analysisin water-quality studies.

• Method detection limits (MDLs) forICP–MS are between about 10 and100 times lower than hydridegeneration–atomic absorptionspectrophotometry (HG–AAS) andinductively coupled plasma–atomicemission spectrometry (ICP–AES)methods. Therefore, ICP–MS is themethod of choice wheneverelemental concentrations are lessthan 10 µg/L.

• The short- and long-term accuracy forthe determination of the newelements by ICP–MS wereacceptable; all the elements werewithin one standard deviation of themost probable value.

• Data for whole-water digestatesindicated that there was nosignificant method bias for thedetermination of arsenic, boron, andvanadium by ICP–MS. However,matrix interferences from highconcentrations of concomitantconstituents, for example iron, canaffect the determination of someelements. Concomitant

concentrations that cause significantinterference are usually much greaterthan concentrations found in mostsamples submitted to NWQL.

• Preparation of whole-water samplesrequires using either the HCl in-bottle digestion procedure or theHNO3 in-bottle digestion procedure.Whenever the HCl in-bottleprocedure is used, the HCl must beremoved from the digest by usingsubboiling evaporation. This step isnot required if the HNO3 in-bottledigestion procedure is used.However, there are potentialdifferences between the HNO3 in-bottle and HCl in-bottle proceduresin the solubilization of sedimentmaterial. Such differences dependon the mineral composition of thesediment in the whole-water sample.

• Data from ICP–MS will impact long-term trends in water-quality studiesbecause of the improved bias andvariability at elemental concen-trations less than 10 µg/L.

REFERENCES CITED

American Society for Testing and Materials,1995, Annual book of ASTMstandards, Section 11, Water (D1193,Standard specification for reagentwater): Philadelphia, v. 11.01,p. 122–124.

Fishman, M.J., ed., 1993, Methods ofanalysis by the U.S. Geological SurveyNational Water Quality Laboratory—Determination of inorganic andorganic constituents in water andfluvial sediments: U.S. GeologicalSurvey Open-File Report 93-125,217 p.

References Cited 15

Fishman, M.J., and Friedman, L.C., eds.,1989, Methods for determination ofinorganic substances in water andfluvial sediments: U.S. GeologicalSurvey Techniques of Water-Resources Investigations, book 5,chap. A1, 545 p.

Garbarino, J.R., 1999, Methods of analysisby the U.S. Geological SurveyNational Water Quality Laboratory—Determination of dissolved arsenic,boron, lithium, selenium, strontium,thallium, and vanadium usinginductively coupled plasma–massspectrometry: U.S. Geological SurveyOpen-File Report 99-093, 31 p.

Garbarino, J.R., and Hoffman, G.L., 1999,Methods of analysis by the U.S.Geological Survey National WaterQuality Laboratory— Comparison of anitric acid in-bottle digestionprocedure to other whole-waterdigestion procedures: U.S. GeologicalSurvey Open-File Report 99-094, 23 p.

Garbarino, J.R., and Struzeski, T.M., 1998,Methods of analysis by the U.S.Geological Survey National WaterQuality Laboratory— Determination ofelements in whole-water digests usinginductively coupled plasma–opticalemission spectrometry and inductivelycoupled plasma–mass spectrometry:U.S. Geological Survey Open-FileReport 98-165, 101 p.

Garbarino, J.R., and Taylor, H.E., 1994,Inductively coupled plasma–massspectrometric determination ofdissolved trace elements in naturalwater: U.S. Geological Survey Open-File Report 94-358, 28 p.

Hoffman, G.L., Fishman, M.J., andGarbarino, J.R., 1996, Methods ofanalysis by the U.S. Geological SurveyNational Water Quality Laboratory—In-bottle acid digestion of whole-watersamples: U.S. Geological SurveyOpen-File Report 96-225, 28 p.

Horlick, Gary, and Shao, Youbin, 1992,Inductively coupled plasma–massspectrometry for elemental analysis, inMontaser, Akbar, and Golightly, D.W.,eds., Inductively coupled plasmas inanalytical atomic spectrometry (2d ed.):New York, VCH Publishers, Inc.,p. 551.

Perkin-Elmer, 1997a, Elan 6000 inductivelycoupled plasma–mass spectrometersoftware guide: Norwalk, Connecticut,Perkin-Elmer part number 0993-8968Rev. F.

_______1997b, Elan 6000 inductivelycoupled plasma–mass spectrometerhardware guide: Norwalk, Connecticut,Perkin-Elmer part number 0993-8969Rev. E.

U.S. Environmental Protection Agency,1997, Methods for the determination ofmetals in environmental samples: EPA200 Series, Supplement I, EPA-600/R-94/111, Revision 2.8, May 1994.

Date post:	13-Aug-2019
Category:	Documents
Upload:	lamdang
View:	216 times
Download:	0 times

Methods of Analysis by the U.S. Geological Survey National ... · USGS U.S. Geological Survey <...

Documents