V O L U M E 23, NO. 12, DECEMBER 1951

been ~~ reoopi5ed by the present authors (8) rand are discussed by

1823

LITERATURE CITED

Am. Soo. Testing Materials, Philadelphia. "A.S.T.M. SCandards

Cdkins, L. E., and White, M. M., Nail. Perroleurn News, 38.

Feldman, C., ANAL. C ~ B M . , 21,1041-6 (1949). Gsmbrill, C. M., Gassmann, A. G., and O'Neill, R. 0.. I W , , 23.

Gassmann. A. G., and O'Neill, W. R., Ibid., 21,417-18 (1949). Harvey, C. E., "Swctroohemical Procedures," Glendele, Calif.,

Kolthoff, I. M., and Sandell, E. B., "Textbook of Qucmtitative

Pagliassot,ti, J. P., and Porsohe, F. W., ANAL. CHEM., 23,198-200

on Petroleum Products and Lubricants," November 1948.

R519-30 (1946).

1365 (1951).

Applied Research Laboratory, 1950.

Inorganio Analysis," New York. Maomillan Co.. 1938.

(1'351).

Harvey (e). Any tendency toward preferential adsorption of the additive

elements by the electrode carbon is minimized because the sample k carried on the electrode surface and does not involve the pene- tration of the samde throueh the electrode.

Solids, suspended in theoil, thou h they may he very Iinely divided and their presence not read& apparent, are not filtered out hut are carded into the discharge sone.

This nonfiltering characteristic suggests the application of rotating electrodes to the analysis of used oils. Gamhrill, Gass- msnn, and O'Neill ( 4 ) have recently made use of this sample- handling technique for the analysis of used oils. Extension of the present method to the determination of iron, lead, and copper 8s well as the additive metals in used oils is currently being in- vestigated in this Laboratory.

RECETVED May 31, 1951. Presented before the Pitbburzh Conference on Analytical Chemistry and Applied Speotrosoopy. March 7. 1951.

Spectrochemical Analysis 0% Lithium I

LOUIS E. OWEN AND JANUS Y. ELLENBURG' Chemistry Section, NEPA Projecta, Oak Ridge, Tenn.

Corrosion studies involving molten lithium metal and ferrous alloys made neces- sary the development of an analytical procedure for the determination of metallic impurities in lithium. The general method evolved calls for the spectrographic excitation in a porous oup electrode of the elements of analytical interest sepa- ratedfrom thelithium matrix. The analytioal technique has permitted thequan- titativestudyofeormsionratesnotonlyoflithium but alsoofother alkali metals and alkali metal hydroxides.

NVESTIGATIONS into the corrosion effects of molten lithium I metal on metallicsurfsces~dedesir~hl.ble thedevelopment of a spectrochemical procedure for the analysis of lithium matrices. It N&S originally specified that the method he general, as it was expected that determination of nearly all metallic elements would be needed. The required limits of analyticel precision were not n&11ON, as the experimental teohniques producing the original samples were known to be subject to wide variations.

Working under these criteria, the NEPA Project spectrographic hhoborstories evolved a general analytical scheme for lithium metal which called far the preparation of a solution containing an internal standard and representing the elements of analytical interest separated from the lithium matrix. The solution was excited in a porous cup electrode. Analytiod curves were obtained from standard solutions prepared and excited in a manner to compare with the solutions obtained from samples.

This paper is concerned with the analysis of lithium metal for beryllium, chromium, cobalt, iron, nickel, niobium, man- ganese, molybdenum, tantalum, titanium, uranium, vanadium, tungsten, and eirconium; most of these elements were determined in the range 0.0001 to O.ly0. Tantalum and tungsten did not fit the general scheme of andysis and are detailed separately.

In other project laboratories the andytical soheme developed a t NEPA has been extended to many other metallic elements with procedure modifications acoommodating the specific chemi- c d characteristics of the elements sought.

DEVELOPMENT

In the early stages of the investigzhon the lithium samples arrived in the lahorstory as mechanically sealed metallie tubes (Figure 1) containing contaminated lithium with analytical interest wholly in the lithium portion. With simulated samples

x Present address, Csrbide and Carbon Chemioah Co., Y-12 Plant, Oak Ridge. Ten% I Now ANP Ooeration of General Electric Co.

it was found feasible to open these capsules with a tubing cutter without introducing avert contamination. In view of the nature of the samples, it ,!-as decided that only a solution technique would suffioe to separate the lithium from the capsule material. Water and the alcohols were found suitable far this we. The true sample portion could he removed from' the capsule by re- action with these solvents and washing of the container. As the solvent step appeared necessary to gain the analysis sample, the development work was concentrated upon a procedure employing solutions.

Figure 1. Sample Capsules Left. Stainless steel . Right. Arm- iron

The local familiarity with the porous electrode technique (1,s) led naturally to its use. Solutions of lithium contaming the elements of interest were prepared and excited. The presence of the lithium salts reduced the sensitivity of the analysis elements

1824

by a factor as great as lo', so procedures avoiding the presence of lithium were developed. The general analytical scheme called far the excitation of a solution containing an internal stand- ard and representing the elements of analytical interest sepa- rated from the lithium matrix.

SAMPLE TREATMENT

Two types of cspsules (Figure 1) containing lithium were received for spectrochemical examination. One type consisted of lithium metal in stainless steel capsules and called for tho determination of the elements of the steels in the lithium.

This type was first handled in the wet analytical laboratory for the determination of oxide and nitride content. Total lithium content of these samples was also determined by ordinary analye ical methods to permit the calculation of spectrographic results,

A N A L Y T I C A L C H E M I S T R Y

combined with their respective original residues for conversion to analysis solutions.

reaction the solutions werefiltered and tKe lithiumconbminants

~~~~ 1 ~~

ketallic impurities of anslyiich interest was then transferred to the spectrographic laboratory for further analysis. The

These capsdes were prepared for bpening by reducing a section on a lathe until tubing cutters could sever them without con- tamination. A sample weight for calculation of results was ob- tained by the weight difference of the capsule just before opening and afber being rinsed and dried following the solution removal of the sample.

I

m UI t h Pr wyl su 61

6s ac ar

ti( 8"

PREPARATIC IN OF THE ANALY

ltaininlr tho rmirh,r

'SIS SOLUTION

e&ules were all digested to prepare solutions for excitation. The filter paper was wet-ashed by heating rvith 3 ml. of a mixture of 30% Concentrated sulfuric a~ id -707~ concentrated nitric acid by volume until the nitrogen .peroxide fumes were completely expelled. About 0.5 ml. of concentrated nitric acid was added dropwise until oxidation of the cellulose was complet,e. This treatment produced partial solution of the elements for analysis. Complete solution required individual srtmple treatment. Iron and chromium were dissolved by the careful addition of 0.5 ml. of concentrated hydrochlorio acid in drop increments. Sub- sequent additions of nitric, sulfuric, or hydrochloric acid rmre made as necessary to provide solution of the element or elements of interest in a particular sample. For samples high iu chromium the h drochlorio acid content was maintained at a level high enougx to prevent the salting out of dehydrated chromium sulfate or nitrate.

The final volume of any of these solutions was kept helow 5 ml. to permit quantitative transfer to 10-ml. volumetrio flasks. Preceding the trausfer, platinum for use as an internal standard was introduced as 75 microliters of a 1% platinum solution giving B concentration of 75 p.p.m. (weight/volume) in the final solution for excitation. Before these solutions were brought to find volume with water, 0.5 mi. of concentrated nitric acid was added t~ each, to stabilize their acid content and ensure flow through

~~ , ~1~

pigure 2. Solution Extraction of Lithium from Capsules

Water was used as the solvent for the whole group in the auner illustrated in Figure 2. Cobalt, molybdenum, titanium, ,anium, vanadium, and zirconium were separated easily from .e lithium matrix by filtration of the lithium hydroxide solution oduced by the reaction with water. All filtrations in this work :re performed with a 15-mm. disk of Whatman No. 42 paper pported by a fritted-glass Biichner-type funnel. Vacuum tration was generally employed. Beryllium, chromium, iron, nickel, niobium, and manganese Id appreciable solubility in the lithium hydroxide and required lditional chemical separation. Iron, chromium, manganese, Id nickel were precipitated as their sulfides by ammonium Ifide. Beryllium and niobium were recovered by a precipits, )n with gquinolinol (8). The individual precipitates were

IQUES . _.._ I_.._ ".-.." P._r _._I ._. jl.l ..ere excited Using the

porous electrode method developed by Feldman (8). The sample electrode was the mode of a. direct current spark disoharge ob- tained from Bin ARL high precision S O U I C ~ unit held at a power level just low enough to avoid boiling the solution. This corm sponded to electrical indications of BO volts acro8s the primary of the power transformer of the Source and a reading of 1.5 amperes on an R-F ammeter in series with the gap. The elec- trodes were oresparked until flow was established as indicated by color changes 'm the spark column. 3 minutes on S. A. No. 1 film. rigidly controlled Conditions in the conventional manner.

Exposures were made for The film was processed under

Tmnsmittances of the spectral lines and of their adjacent back- ground regions were read on a densitometer and converted to numbers related to intensities by means of a photographic black- ening curve. The log intensity of the background was subtracted from the log intensity of spectral line plus background as B

partial background correction. The wide range of elements for analysis necessitated the choice

of au element such as platinum for the internal standard. Had the analytical elements been limited to definite groups, an in- ternal standard would have been chosen with lines more suitable from a spectroscopic viewpoint.

Acceptable reproducibility of the porous electrode excitation conditions was indicated by the consistent transmittance values of between 45 and 50% for the platinum lines used. Duplica- tion of analytical runs produced values with less than 10% spread. The limiting precision of the method is about 2%; however, the precision achieved is an economic level based upon the precision of original experimental procedures.

An expression of accuracy for a procedure of this type is difficult to obtaiu, as no primary lithium standards exist for which tNe values of impurity content are known. Therefore the repro- ducibility of independently prepared synthetic standards prw vides the closest expression of merit, but it can be called no more than precision. For this work the precision was + 5%.

ERYLLlUM . . . ALTERNATIVE PROCEDURE FOR B

The actual samples containing beryllium tenaea M coucenwa- tional levels not requiring the sensitivity of the generalized method and a simplified procedure WBS instituted,

V O L U M E 23, NO. 1 2 , D E C E M B E R 1 9 5 1

For this technique the capsule was reacted with water as usual, rinsed, and removed. The total solution then was brought to neutrality with hydrochloric acid, and 5% hydrochloric acid plus 5% nitric acid calculated on the basis of a known volume was added. A portion of the known volume of solution was then poured into a 10-ml. volumetric flask containing 75 microliters of a 1% platinum solution. The solution from this IO-ml. flask was used for excitation. Separate analytical curves were pre- pared for use in analyzing beryllium solutions under this pro- cedure.

SPECIAL PROCEDURE FOR TANTALUM AND TUNGSTEN

I t was necessary to employ arc techniques foI tantalum and tungsten, as their limit of sensitivity when excited in a porous elwtrode was higher than tllrir concentration in the samples for analysis.

I I I I I [

1825

I I l l

5.0 4.0 3.0 2.0

03

I O

- -

1

0.5 0.4 0.3

0 2

O.' k / J /

0.05 k

WEIGHT/VOLUME OF SOLUTION, P. P. M. Analytical Curves for Manganese, Kickel, and Zirconium

Figure 3.

As tantalum IT as insoluble in lithium hydroxide, the sample n as dissolved from the capsule as usual and the tantalum retained on filter paper. The residue was ashed in a Vycor crucible, 5 mg. of cupric oxide were added to the ash, and the mixture was heated a t 800" C. for 30 minutes in order to oxidize the tantalum. Five milligrams of powdered graphite were then added to the cooled mix. This entire sample was arced in a cratered electrode.

Tungsten partially reacted with the lithium hydroxide when the sample was dissolved from the iron capsule. In an initial filtration, part of the tungsten content was retained in the ele- mental state on a filter paper disk while the remainder went into the filtrate as lithium tungstate. Ammonium sulfide was used to convert the lithium tungstate to ammonium tungstate. Tung- sten sulfide was precipitated by acidification with hydrochloric acid of the cold solution of ammonium tungstate. The filter paper carrying the tungsten sulfide was added to that with the elemental tungsten and ashed. The analytical scheme for the tantalum was then followed for the tungsten.

ANALYTICAL CALCULATIONS

Suitable spectral lines for analytical use were determined from evcitation of standard solutions. Working curves for analysis n ere prepared which related the intensity ratios of these lines to platinum lines against concentration of the element in solutions prepared by carrying out the chemical separations on synthetic standards. These synthetic standards were solutions of lithium to which known quantities of elements had been added. Thus the manner of calibration compensated for any reproducible in- complete recoveries. Figure 3 is typical of the analytical curves prepared.

Sample calculations were carried out in an analogous manner. Determined intensity ratios provided values from the working curves related to the concentration of the element sought in the solution used for excitation. Conversion to parts per million of original sample were then made.

Calculations for the tantalum and tungsten samples, that

were arced, were based upon the intensity ratios of their spectral lines compared to the lines of the internal standard, copper. Figure 4 shows the analytical curves for tantalum and tungsten.

PREPkRATION OF STANDARDS

Analytical curves for beryllium, chromium, cobalt, iron, nickel, niobium, manganese, molybdenum, titanium, uranium, vanadium, and zirconium were obtained by the excitation of standard solutions of these elements made to known concentra- tional levels. Stock solutions of 1 yo (weight/volume) concentra- tion were used to prepare spectrographic standards in the range 1 to 1000 p.p.m. Ten milliliters of a standard solution contained the desired amount of metallic element, 7 5 p.p.m. of platinum, and 10% volume of hydrochloric or nitric acid. Analytical curves for multiple element analysis were prepared from solu- tions containing elements in the concentration levels encoun- tered, as approximated by the use of curves prepared from single element solutions.

1 1 I I l l l l l I c

/* /

/ Ta3012.5 A: Cu3022.6 A'

2 1.0 ,/'W 29470 A: f 0.8 / Cu 3022.6 A.

0.6 z 0 5

04

4 /

/

4

/ /

/

Analytical curves for tantalum and tungsten were prepared from arc excitations of mixes produced by carrying known weights of their oxides through the special procedure outlined for tan- talum and tungsten. The oxides were placed on filter paper disks corresponding to the ones obtained in actual analysis.

EXTESSION OF TECHNIQUE

The analytical procedure outlined has been employed in many laboratories and has also formed the basis for the extension of this technique to the determination of metallic contaminants in other alkali metals and their hydroxides.

The procedure for the analysis of other alkali metal matrices followed that for lithium with modifications introduced by the sample container and sample solvent selection based upon per- missible speed of reaction. Samples composed of alkali metal hydroxide matrices were prepared for the general analytical scheme by simple water solution of the sample. The separation of the elements of analytical interest followed the generalized procedure used for lithium.

LITERATURE CITED

(1) Delaney, J. C., and Owen, L. E., ~ ~ N A L . CHEM., 23, 677 (1951). (2) Feldman, C., Ib id . , 21, 1041 (1949). (3) Mitchell, R. L., and Scott, R. O., Sprctrochim. Acta , 3, 389-70,

RECEIVED March 28, 1951. Presented s t the Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Pittsburgh. Pa., March 6 , 1951.

K. 3/4 (1948).