V O L U M E 27, N O . 8, A U G U S T 1 9 5 5 readings are subtracted from the unknovm and standard readings to obtain net relative luminosities, which are averaged for each sample. The average net relative luminosities of the standards are plotted against concentration to give L - H calibration curves for both copper lines to which tlie average relative luminosities of the unknowns are referred. Or, if the internal standard method of calibration is employed. the ratio of the average net relative luminosities of the copper line to that of the silver line is plotted against concent,ration of copper in the standards on log-log graph paper to give the respective calibration curves. Procedure for Tin-Base Alloys. Weigh samples containing 1 t o 15 mg. of copper into 125-nil. Erlenmeyer flasks. Dissolve in 20 ml. of 487, hydrobromic acid containing 2 ml. of bromine. Cover and heat gently until dissolution of the sample is compiete. Add 10 ml. of 12%’ perchloric acid and heat in a well-ventilated hood, while swirling, over an open flame, until white fumes first appear. Then heat moderately and intermittently to decompose any lead bromide and t o expel all hydrobromic acid. A stream of compressed air passed into the flask materially hastens the removal of stannic bromide and antimony bromide. For large amounts of tin, it ma>- be necessar>- to repeat the volatilization step with an additional 5 ml. of hydrobromic acid. Treat the entire sample residue, or an aliquot portion, iri the same manner as the alloys.
1229
fianie rpectrophotonirtric procwlures should particularly :tpptal to anaa 1-vsts. Iiesults obtained for the samples enumerated in Table IV are apparently unaffected by the measurement method eniploycd. Those obtained by L - H measurements and by the iritc,rrial standard method show no significant differences. This is intrrpreted to mean that the measurement of the flame luminosity a t the base of the emission linea of copper will adequately c~ompcnsate for variations in the flame background, and that furthrrniorc, the copper emission lines are essentially unaffected by the prcsencc of relatively large amounts of diversc clement,i in the flame. Apparently, whenever the flame background is enhanced or depressed by other elements, the copper line emission remains rirtually unchanged, and is simply an additive factor onto thc flwme background. Consequently, a t least for the ci reported in this investigation, it is not necrswr trouble of incorporating an internal standard element in thr sitiiiple and observing the additional luminosity readings required. ACKNOW LEDGM E S T
D1 SCU SSION
Table IV summarizes the results obtaiiied on Bureau of Standards aluminum-, tin-, and zinc-base alloys. The reproducibility of the flame analyses wits very good. On replicate sample? the standard deviation from the mean was approximately 3%. I n many cases the results obtained irere within adequate agreement with the certifirate values. Because of the rapidity with which flame analyses can be accomplished, the procedure for copper offers a competitive method for the determination of copper i,ivaling the conventional colorinirtric or gravimetric niethodpl. Dissolution of the mmple is the only preliminary step required prior to the actual flame measurements, which themselves require only a fwr niiiiittcs of the operator’s time. The prwiiion of flame analyses is not quite a s high as for some c~olothietricmethods and for iiiost gravimetric methods, hut for many piirposeP the higher degree of prerision is not required. I t is in thip latter type of analyses that
The author wishes to thank Clarice L. Thompson for her vooperation in performing some of the cxperinient~in the early phase of this work. LITERATURE CITED
(1) rlhrens, L. H., “Spec.trocheinica1 Anal Wesley Press, Cambridge, 1950. (2) Cholak, J., and Huhbard, D. AI., ISD. Esc. C H E v . , AXIL. ED., 16, 728 (1944). (3) Churchill, J. R., Ibid., 16, 668 (1044). (4) Ells, V. R.,J . O p t . SOC.4 m e r . . 31,534 (1941 ( 5 ) Griggs, A I . A., Johnstin. R., and Elledge, B. E.. ISD. E r c . CHEM., Api.4~.ED., 13, 99 (1941). (6) Lundegbrdh, H., Lanthruks-Hogskol. A4nn.,3 , 49 (1936). (7) Lundeghrdh. H., ”Quantitatire Spektralanalyse der Elemeiite.” Part I, Gustav Fischer, Jena. 1929. (8) Singh, N. L., Proc. Indian Acad. Sci.. 25A, 1 (1947). ) .
RECEIVED for review January 21, 19.53. .Iccei)ted May 2 , 193.5, I’rcsentcd \SociEry. a t the Southeastern Regional hieeting of the A M E R I C ACHEIIICL Birmingham, 41a.. 1954. Contribution S o . 140 from the Departiricnt of Chemistry, Cniversity of Tennessee, Knoxrille.
Flame Photometric Determination of lithium in Silicate Rocks R. B. ELLESTAD, Lithium
Corp. o f America, Inc., Minneapolis, M i n n .
E. L. HORSTMAN, Rock
Analysis Laboratory, University o f Minnesota, Minneapolis, M i n n .
Wet chemical methods are unsatisfactory for the determination of small amounts of lithium in rocks, so that geochemical studies hate depended on spectrographic methods. Trace amounts of lithium in silicate rocks and minerals are rapidly determined by flame photome t r y . The alkali metals and magnesium are separated from other rock constituents by an acid decomposition followed by a single precipitation with basic lead carbonate. A Beckman DU spectrophotometer is used and the test solution is burned in an inexpensive flame attachment using natural gas and compressed air. Sodium and potassium interferences are compensated for by the use of appropriate additions to the lithium standards. The method is sensitive to 5 p.p.m. of lithia in the original silicate sample, with a maximum deviation of 0.0005% of lithia in the range 0.001 to 0.03%. The method can he applied to a wide variety of silicate materials.
.
F
L4ME photoriiet,ric methods are graduallj- displacing [vet
methods for the determination of the alkali metals. The J . Lawrence Smith method, although uscful for the deterinination of sodium and potassium in silicate rocks and niincrals, is not suitable for the determination of lithium. Killiams and hdaiiir (10) arid Broderick and Zack ( 3 )have applied the flame photometer to the determination of lithium in gl Fanus (4)and Beer ( 2 ) to the analysis of spodumene; and .\IrCoy and Chrietiansen ( 7 )to portland cement. Sone of these mctliod?, or the recent fluoroniet1,ic method of Khite. Fletcher, and I’arks ( { I ) , has been applied to tlie determination of trace amounts of lithium in silicate rocks. The determination of trace amounts of lithium in silic*;ttch rocks and minerals hits been confined to spectrographic nietlioli+. In this field the procedures of Strock (81 and Lundeghrdh ( p i ) have heen most widely used. The flame spertrophotonietcr gives results for trace aniounts of lithium that comp:are favor:tl)ly with determinations made with the spectrograph.
A N A L Y T I C A L CHEMISTRY
1230 Table I. Intensit) of Radiation a t 671 M,u for Different Concentrations of Sodium and Potassium (0.4-mm. slit: full sensitivity. A t same setting, 1 p.p.m. of Liz0 will give a reading of approximately 4.0) Reading on T scale NazO or KzO, P. P. M . Xa20 K?O n 0 0 (water! 0 0.2 120 0.2 0.4 240 0.2 0,s 300 360 0.3 0.6 420 0.7 0.4 0.5 480 0.8 0.6 0.9 540 0.7 1.0 600 730 0.9 1.2 ~
paper (S. & S.No. 505, Whatman S o . 1). Wash thoroughly with hot water until the volume of the filtrate and washings is about 125 ml. Concentrate to about 30 ml., cool, and filter through a small paper into a 50-ml. volumetric flask, washing with water. Make u p to the mark and mix. The solution prior to the addition of the lead carbonate will contain all of the aluminum, iron, sodium, potassium, lithium, and magnesium as sulfates, with a little free sulfuric acid. Lead carbonate neutralizes the free sulfuric acid and precipitates iron and aluminum without increasing the salt concentration in the solution. Excess lead carbonate is insoluble, and does not raise the p H much above 6 . The precipitate is easily filtered and washed.
~-
.iPPARAl
A Beckman Model I>U spectrophotometer was used with a flame attachment employing an air-natural gas flame. The atomizer is similar to that of Barnes and others ( I ) , coiisisting of hypodermic needles for air and sample solution inlet, rigidly mounted on a stainless steel plate. This plate is set in n rubber stopper which in turn is placed in the mouth of a 2-liter, roundbottomed flask serving as the spray chamber. The latter has a Lv-ater-sealed continuous drain a t the bottom and an exit sidearm tube near the top. Air and aerosol leaving the flask are mixed with nat>nralgas in a Venturi miser and proceed to the burner. The latter is a Fisher burner top mounted on a borosilicate glass tube, which is set in a 500-ml. Erlenmeyer flask fitted with a side tube for the introduction of the air-gas mixture. ;211 the air used for combustion passes through the atomizer. Gas consumption is about 5 cu. feet per hour at a pressure of about 8 inches of water. T h e air requirement is about 37 cu. feet per hour, supplied a t 6 pounds per square incall. Careful ront,rol of both air and gas flow, especially the latter, is important. Constancy of gas flow is checked by passing the gas through an orifice flow meter equipped with a water manometer having one leg inclined 3 t o 4 degrees from the horizontal. The flow of gas is held to a maximum deviat'ion of less than 0.Z70b. The settings of the Beckman DU spectrophotometer are as follows: wave length, 671 mp; selector sivitch a t 0.1 ; slit a t 0.4 mm. ; sensit,ivity control a t counterclockwise limit,
The following technique was used in running sample solutions on the photometer. Approximate sodium, potassium, and lithium determinations are made using calibration curves for these elements. This information is used to prepare lithium standards approximately equivalent t o the unknown in sodium and potassium concentration, and bracketing the approximate lithium content. The sample is compared with these standards. EXPERIMENTAL
Interferences. In the determination of trace amounts of lithium in silicate rocks the interference of macro quantities of sodium, potassium, and magnesium must) be considered. Continuum and radiation interferences for these three elements were investigated. A series of stnndard solutions of sodium and potassium sulfate was prepared from the pure salts. These solutions were run on the instrument a t the settings previously given, and the emission a t 671 mp was measured to obtain the continuum effect as given in Table I. (Magnesium up to 360 p,p,m. showed no effect, and is omitted from the table.) This interference is positive and additive to the lithium emission.
Table 11. REAGENTS
Standard Lithium Sulfate Solution, Approximately 0.1N. LVeigh out approximately 8.5 grams of lithium hydroxide monohydrate (purified by recrystallization if necessary), dissolve in 300 ml. of distilled water, and neutralize with a slight excess of 1 S sulfuric acid, using methyl red indicator. Boil several minutes, and make just alkaline to methyl red with dilute lithium hydrowde solution, Filter the solution and make up t o 2 liters. To standardize, pipet 25-1111. samples into platinum dishes, acidify with a drop of 4 S sulfui.ic acid, and evaporate t o dryness. Bring the residue t o constant weight by repeated heating t o a dull red heat over a low flamc. Calculate the normalit>- of the solution from the weight of lithillin suliate found. Basic Lead Carbonate. The quality of the reagent is important. Some brands of analyticd reagent grade have been foufd unsuitable owing t o a very high sodium content. hIerck s reagent grade n-as used without purification. PROCEDURE
iVeigh out a 0.5-gram sample of the very finely ground material and transfer t o a 50-ml. platinum dish. >loisten with LT-ater and add 0.4 ml. of concentrated sulfuric acid, folloxed by 15 1111. of hydrofluoric acid and a drop of nit,ric acid. Heat t o a temperature just below boiling by either a hot plate or infrared light until sulfnric acid fumes appear. This u-ill require from 1 to 2 hours. Cool, moisten with 2 ml. of water and 0.1 ml. of sulfuric acid, and again evaporate t o fumeq. Make a third evaporation with 5 ml. of added water to enellre removal of all the fluoride. Take up the residue in 25 to 30 nil. of water and heat, with occasional stirring, until soluble solids are in solution. Transfer t o a 150ml. beaker and dilute t o about 70 ml. At this point solution is essentially c,omplete csccpt for a small residue of zircon, corundum, or other minerals iiot attacked by the acid decomposition. Heat t o near boiling and add basic lead carbonate, with stirring, until the solution is alkaline to methyl red, keeping the beaker covered as much as possible during this addition. Boil several minutes, rinPe down, and stir well. Filter on a 11-cm.
Effect of Radiation Interference Liz0 Found in Rock, %
Sodium and Potassium Content of Sample Solution NazO, KZO. p.p.m. p. p .ni. 527 527 627 830 830 500 500 510 510
300 300 300 537 537 330 330 360 360
Corrected for continuum interference only 0.0015 0.0040 0.0051 0,0060 0.0099 0,0107 0.0156 0 0173 0 0205
Corfected for continuum radiationand interference 0,0015 0.0030 0.0048 0.0074 0.0114 0.0118 0.0165 0,0200 0.0243
The radiation inteiference effect of sodium and potassium on is, the presence of these alkalies delithium is negative-that presses the intensity of the lithium emission. This was studied by preparing sample solutions of rocks having a wide range of lithium content, comparing first with pure lithium standards, and then with lithium standards containing sodium and potassium. By bracketing the sample with pure lithium standards and deducting from the gross sample reading the background b l m k obtained on a solution iTith sodium and potassium equal to that of the sample solution, a lithium value is obtained which is corrected for continuum interference, but not for radiation interference. Next, by repeating this, but using lithium standards with sodium and potassium conceutrations equal to that of the sample solution, the lithium value obtained is corrected for both types of interference. These results are given in Table 11. Investigation of radiation interference of magnesium sulfate showed a negligible effect up to a concentration of 360 p.p.m. of magnesium.
.
V O L U M E 27, NO. 8, A U G U S T 1 9 5 5 Table 111.
Recover) of idditions of Lithia
% LIZ0
CO LliO Added
Total c; LIZ0
0.0015
0.0003 0.0003 0.0015 0.0030 0.00% 0.0060 0 0073 0 0003 0 0021 0 0039 0 0091 0 01313
0 0018 0 0018 0.0030 0 0043 0 0073 0 0073 0 0090 0 0116 0 0134 0 0132 0 0202 0 0247
in Rock
0 0113 0 0111
Table I\
.
1231
7c LIZ0 Found 0.0019 0.0020 0.0030 0.0048 0.0074
0.0065
0.0086 0 0118 0 0133 0 0149 0 0200 0 Q243
Coniparison of Flame Spectrophotometer and Spectrographic \.lethods YCL1.0 ~I laiiie
s~e~tri~pliotonietei
Standard granite G - 1
0 00JP 0 0049
Standard diabase W-1
0.0030 0.0033 0 001'7
Spectrograph 0 004 0 005 0,002 0,002
Aa an additional check a standard diab:tse and granite ( 5 ) were run several months apart. These results are compared in Table It' with spectrographic results (6) and indicate that the flame spectrophotometer compares favorably with the spectrograph. Replicate determinations on a variety of samples showed a maximum deviation of 0.0005% of lithia in the $ample or 0.05 p.p.m. of lithia in solution in the range 0.05 to 3 p.p.m. of lithia. Average deviation for multiple determinations is ahout O.OOO:37, of lithia. T h e method ip considered to be :wcurnte to 0.001% of lithia in thP sample. ACKNOWLEDGMENT
This investigation was made in the Rock hnul! University of Minnesota, at the suggestion of S. S. Goldich. The use of lead carbonate in the procedure given was originally sugge3ted by P. T. Gilbert, Jr., while in the employ of lIetallo>Corp. (now Lit,hium Corp. of America). This research WLP snpported by a grant from the Penrose Fund of the Geological Society of America for research in analytical methods applicable to rocks and minerals. The financial assistance of the Gulf Oil Fellowship P1:m t o 1;. I, Horstman is gratefully acknowledged LITERATURE CITED (1)
PRECISION AND ACCURACY
T h e semitivity of the instrument at the settings used (full aensitivit,y, 0.4-mm. slit, 671 nip) was 0.05 p.p.ni. of lithia in solution, corresponding to 0.00057, of lithia in the sample. A4difference of this amount of lithia coiild usually be detected (see Tahle 111). Because no ~tancl:irdmateriala are :tvail;thle for anal>.sis, the only check on the :tcc~ur.icyis I)!, actding known amounts of lithia to the rock powders tiefore dec.omposition and checking recovery. This W ~ Edone on rorks on which replicate determiriations had been made, the average value heing wed as the true valne. RewltP of thew deterniin:itions are show1 in Tnhle 111.
Iiaruer. 11. R..Richardson, D.. Berry. J. IV., and Hood, 11. L.,
I N D . E N G . CHEM., A%X.kL. E D . ,17, 605 (1945). ( 2 ) Beer. H. L., The Precambrian. 24, 8 (19613. (3) Broderick, E. J., and Zack, P. G., 4 x - a ~CHEJI., . 23, 1455 (1951). ( 4 j Brumbaugh. R. J., and Fanus, IT, E., I b i d . , 26, 463 (1954). (6) Fairbarn, H. W., and others, U. S. Geol. Survey, Bull. 980
(1961).
(6)
LundegSrdh, P. H., A r k i c Kemi, M i ' m r a l . G r o l . . A23, S o . 9 (1946).
( 7 ) 3IcCoy. W,J., and Christiansen, G. G., Ani. SOC.Testiug Alaterktls, Special Tech. Pub. 116, 44 (1951). (8) Strock, L. K.,S a c h r . Ges. W i s s . Gottiiigen, M a t h . - p h y s i k . Klasse, Fachgruppe, I V , 171-204 (1936). (9) White, C. E., Fletcher, 31. H., and Parks, J., Axar.. CHEU.,23, 478 (1951).
(10) Williams, J. P., and Adams. P. B., J . A m . C'eram. Soc., 37, 306 (1954).
RECEIVED for review January 24, 1955. Accepted . i p r i l 14, 1955.
Amperometric Titration of Iron with 1-Nitroso-2-naphthol RAY F. WILSON and HUBERT G. LOVELADY Department o f Chemistry, Texas Southern University, Houston, r e x .
A n amperonietric titration of iron(IIl), in acetic acidsodium acetate hufrer, \+ ith 1-nitroso-%naphthol gal e satisfactor) results. It vas the purpose of this inrestigation to determine optimum conditions for ohtaining reproducible and acciirate resd ts in the direct titration of iron w i t h this reagent. The titration is more rapid than the gra\ imetric methods, and necessitates fewer operations than other ampernmetric methods for the determination of iron. Of the cations studied, onlj lend interferes. Iron i n iron ores was determined rather accuratelj, after it had heen separated bj ether extraction. This method should be especiall? suitahle for determination of iron in selected steels.
S
solutions nith cupferron, the ammonium s d t of phenylnitrosoh!.drosylamine, in tartrate and cit'rate buffers. They recommended that, in the presence of t,artrate or citrate. thr iron solutions be titrated in a cell covered with a coat of black paint, since iron(II1) under these conditions is readily redured by light to iron(I1). These authors report an error of 1% for solutions not less than l m 3 1 in iron, and an error of 2 to 3$4 for solutions Iyhich are 0.5m3fin iron. This investigation was undertaken to ascertain the possihilitv of obtaining reproducible and sufficiently accurate stoichiomcstrjc results in the direct' titrations of iron(II1) JT-ith l-nitroso-2n:iphthol in acetic acid-sodium acetate buffer. The precipitate formed in the reaction between iron and 1-nitroso-2-naphthol, i n acetic ncid-w:iter medium, has the composition (C,OHGSOYIaFe
h S D B E R G L.;) hae performod nmperometric titrations of iron( 111) with broniosine (5,7-dibrorno-S-hydrosyquinoline) EXPERIXIENTAL :it 50" C. In his work the titration error is reported to be 2% for a Reagents and Solutions. The stock solution of iron was pre0.1ndI concentration of iron. and not more than 0 . i % for concenpared by dissolving 27 grams of Baker,s analyzed iron(III trations of iron greater than 0 . 4 n d f . Kolthoff and Liberti ( 2 ) chloride hexahydrate in difit,illed\vatey. This solution treated hnve carried out satisfactory amperometric. titrntione of iron(II1) with 10 nil. of concentrated hydrochloric acid to prevent hyrol?-sis
