Spectrochemical Analysis of Bismuth Matrices Porous Electrode Technique JOSEPH C. DELANET'
AND
LOUIS E. OWEN2, Chemistry Section, iVEP'4 Project, Oak Ridge, Tenn.
Corrosion studies at the NEPA project required a spectrochemical procedure for the analysis of bismuth and lead-bismuth eutectic matrices containing 0.0001 to 0.5% beryllium, cobalt, chromium, iron, manganese, molybdenum, niobium, nickel, tantalum, titanium, vanadium, and/or tungsten. The procedure defined in this paper provides that a solution of the entire sample and an added internal standard be spark-excited in a porous cup electrode, the spectra photographed, the resulting film photometered, intensity ratios calculated, and the amount of contaminant determined by comparison of these intensity ratios and intensity ratios representing known concentrations. This method enabled NEPA and other participating organizations to determine the corrosion effects of molten bismuth and lead-bismuth eutectic on stainless steels and other metals.
I
NVESTIGATIONS at the KEPA Project required that a spectrochemical procedure be developed for the analysis of bismuth and lead-bismuth eutectic matrices containing 0.0001 to 0.5% beryllium, cobalt, chromium, iron, manganese, molybdenum, niobium, nickel, tantalum, titanium, vanadium, and/or tungsten. Study of the literature provided little indication of previous work on this problem. The amount of sample available for analysis was not sufficient to permit the preparation of homogeneous rods suitable for use as self-electrodes. Although it was possible to cast the samples into electrodes, these could not be produced with sufficient uniformity, and other mechanical processing to provide self-electrodes did not seem justified. It was concluded that a solution of the entire sample should be used, and various methods for the excitation of solutions were studied. The porous electrode technique proposed by Feldman (1)gave the most promise of applicability, and a satisfactory analytical procedure was developed. The general procedure can be adapted t o other solution excitation techniques. In the adopted analytical scheme, a solution or solutions of the entire sample is prepared with an internal standard, and the solution is spark-excited in a porous cup electrode. Analytical curves are obtained by means of standard solutions prepared and excited in a manner analogous to that of the samples. The porous cup electrode used consists of a purified graphite rod 0.25 inch in diameter and 1.5 inches long, drilled with an '/*-inch drill to within 1.2 mm. of the bottom. It is held as the upper electrode in a spark excitation stand and is opposed by a '/cinch carbon counterelectrode. The solution to be excited is introduced into the porous electrode with a glass dropper drawn to a long fine point. Excitation is started with the electrode cavity filled. The exposure is started after the solution has seeped through the unpierced end of the electrode and become available to the spark. Indication of the proper time for commencing the exposure is given by the appearance a t the analytical gap of a pinkish glow attributed to the hydrogen alpha line.
Sets of standard solutions were prepared which conformed to the analytical situations encountered. Principal individual element stock solutions consisted of high purity 1 or 0.1% concentration of the various elements of analytical interest and a solu1
A sufficient weight of bismuth metal was dissolved in nitric acid, or bismuth nitrate was put into a 10-ml. volumetric flask to give a concentration of approximately 20% bismuth on a weight-to-volume basis. Then 100 lambda of a 1% solution of iron ion, 100 lambda of a 1% solution of chromium ion, 1000 lambda of a 1% solution of nickel ion, 1000 lambda of a 0.1% solution of manganese ion, and 75 lambda of a 1% solution of platinum were added to the volumetric flask containing the bismuth. The solution was prepared with an acid concentration of nitric acid and/or hydrochloric acid of about 20% when the solution was dil ed to 10 ml. This provides a concentration in the solution of 10 p.p.m. iron, 100 p.p.m. chromium, 1000 p.p.m. nickel, 100 .p.m. manganese, 75 .p.m. platinum, and about 20% bismutg, all on a weight-to-voEme basis. In general, for this series the standard solutions containing lesser amounts of the analytical elements were prepared by diluting the initial solution with a stock solution containing 20% bismuth and 75 p.p.m. platinum. Usually these dilutions were made to obtain four points on the analytical curve. The possibility of error in preparing standards by dilution was realized, and a number of random standard series were prepared to validate the curves obtained by dilution methods.
%
Standard solutions containing varying amounts of matrix element were also prepared in order to measure the effect of matrix concentration on intensity ratios. This study was made to test the validity of occasionally diluting sample solutions to adjust them for use with prepared analytical curves. Standard solutions containing varying concentrations of multiple elements were also prepared. The purpose of this series was to determine the direction and degree of analytical interference among and between the analysis elements. The magnitude of interference was, in general, very slight a t low concentrations and of little analytical significance in the useful portion of the ranges
PREPARATION OF STANDARDS
1
tion containing 1% platinum for use as an internal standard. Platinum was chosen as an internal standard because it was the only element not likely to be requested for analysis. The use of platinum as an internal standard for the elements considered would not ordinarily be easily defended from a spectroscopic viewpoint; however, one of the most useful features of the porous cup electrode technique is the unusual freedom it provides in the selection of a reference element. Additional materials available for the preparation of standards were stocks of high purity bismuth and lead, and salts of these metals. Pure lead-bismuth eutectic alloy was also available. Series of tests were run to determine the effect of various acid concentrations on the intensity ratio values. These tests indicated such effects to be of smaller magnitude than the precision to which the work was to be performed. Standard solutions were prepared by several different methods, so that extraneous element and other deleterious effects might be evaluated, and corrections attempted. Five series of solutions graded as to element Concentration were prepared and excited. One series consisted of single elements alone in acid solution; a second contained multiple analytical elements alone in acid solutions; the third series had the single elements in acid solution containing bismuth; the fourth was prepared with multiple elements in acid solution containing biemuth; while the fifth series was of multiple analytical components in acid solutions containing lead-bismuth eutectic alloy. The solutions of all series carried identical concentrations of platinum as an internal standard. The method of preparing the standard series containing several elements of analytical interest, the internal standard, and thp bismuth matrix was, for a typical case, as follows:
Present address, P.O. Box 494, Oak Ridge, Tenn. Present address, 245 Maple .4ve., Saratoga Springs, h'. Y
577
5-78
A N P. L Y T IC A L C H E M I S T R Y
studied; however, in the case of a few lines the interference effects were of sufficient magnitude t o require appreciable corrections. Figures 1 and 2 illustrate the relative freedom from extraneous element effect exhibited by the 2599.396 A. iron line and the 2i66.540 A. chromium line when referred to the 2659.454 A. platinum line. I n these two figures data have been plotted for the conditions indicated. The line drawn through the points does not purport to be an analytical curve; it is the best line to be drawn through a series of points related to each other only inasmuch as they represent the same concentration of analysis element. The curves are designed t o illustrate that while diglit changes in slope do occur for the 2599.396 Fe/2659.454 Pt arid 2766.540 Cr/2659.454 Pt analy4s lines, depending on the concentration of other elements present, no major shifting of the line is obvious.
9.0
I
II
I
I I I 1 1 1 1
5.0
0
F
2
2
92 1.0 z
I111l1
0.P 10
100
CONCENTRAT1ON,
Figure 2.
P.P.M.
2566..540 Cr/2659.454 Pt
0
Singlp elements 0 Multiple element sjatems 0 Bismuth matrices A Ph-Bi matrices
1
These solutions were made t o contain approximately 20% leadbismuth eutectic misture. -4 study of the effects of the presence of some lead chloride precipitate indicated that it did not represent an interfering factor in semiquantitative studies. The special chemical procedure for lead-bismuth samples is included in the section on sample preparation. SAMPLE PREPARATION
0
Single elements Multiple element systems 0 Bismuth matrices A Pb-Bi matrices
0
On the other hand, Figures 3 and 4 show cases where a major shift is definitely indicated. These two figures indicate the curve& obtained under the indicated conditions for the 3981.651 A. nickri line and the 3021.558 A. chromium line when referred t o thr 2659.454 A. platinum line. It can be seen that in both of these figures it is necessary t o draw two distinct curves, the lower curve appearing valid for the analysis element in solution alone and with small amounts of other elements, the upper curve valid for the analysis element in the presence of small quantities of other elements and a matrix, the same line applying both to solutions containing bismuth or lead-bismuth eutectic matrix. I n cases where analytical curves such as those indicated i n Figures 3 and 4 must be employed, it was feasible t o improve analytical accuracy by interpolating between the line for 0% bismuth and the line for 20% bismuth for all solutions containing intermediate concentrations of bismuth. The reasonableness of this technique was determined by experiment. I n actual analytical practice the preceding composite curves were never used; the analytical curve being utilized depended in each case only on the composition of the particular sample being esamined. The standard solutions prepared for use with lead-bismuth matrices differed slightly from those for the bismuth matrices.
The samples received for analysis were accompanied 113’ an analytical request containing information which indicated to the spectroscopist the elements likely to be present and those of analytical int,erest. The initial preparation usuii1l~-consisted of casting the sainl)les into rods 0.125 to 0.25 inch in diameter and from 0.5 to 4 inches ill length. These rods wwe first sparked as selfelectrodes to give an indication of the concentration of analyt,ical element,s. This step also provided informat,ion which suggested the proper approach for dissolving the sample and the solution concentration which should be excited. The excitation of the material as self-electrodes also pernlitt,ed an estimation of the content of trace analysis constituents and impurities not suited to the general procedure. The entire sample was then weighed and transferred to a beaker or volumetric flask of suitable size. -4beaker was used for samples expected t.o dissolve with difficulty, while the flask served for those which experience indicated would dissolve readily. CHEMICAL PROCEDURE
Bismuth Samples. A completely general method for dissolving the samples mas never found; most samples received individual treatment. The bismuth matrix dissolved readily under attack with nitric and hydrochloric acids. The amount of nitric acid was always sufficient t o dissolve independently all of the bismuth present. The nitric and hydrochloric acids also dissolved some or all of the elements sought. The nature of further attempts t o dissolve the sample completely were dictated by the elements sought.
573
V O L U M E 2 3 , NO. 4, A P R I L 1 9 5 1 h s an example of such treat,ment, wlutions of samples containing high concentrations of chromium were either evaporated nearly to dryness and then treated with hydrochloric acid or were filtered and the residue was treated with hydrochloric acid in an attempt to ensure complete solution of the chromium content. In many instances a highly magnetic residue was encountered which was passive even to aqua regia. This residue consisted mainly of chromium with some iron and small quantities of manganese. The action of various solvents used to dissolve this highly magnetic residue could be followed by occasional testing of the cooled residue with a small magnet. The solution was swirled in a beaker, which was then held in contact with the magnet until the solution came to rest. \Then the magnet was withdrawn, the out,line of the magnetic poles could be observed when extremely minute quantities of magnetic residue remained undissolved. Prolonged attack by aqua regia occasionally supplemented with small additions of hydrofluoric acid or perchloric acid produced eventual solution. The presence of excess fluoride or perchlorat,e ion often caused enhancement of various spectral lines, and care was always taken to minimize this effect bg removing as much as practicable of these anions. 9.0
1 1 I1
1
I 1 1 1 1 1 1
5.0
e 4
E
E
4 1.0
is the greater solution concentration and, therefore, greater sensitivity obtainable in the absence of the matrix element. Lead-Bismuth Samples. Analytical samples of lead-bismuth eut,ectic alloy matrices required chemical preparation for spectrochemical analysis differing from that for simple bismuth matrices. The steps in their preparation are outlined below. The sample, prepared as a cast pin of about 15 grams, was weighed and transferred to a 150-ml. beaker. Nitric acid was added in sufficient quantity apparently to dissolve the entire ?ample. The resulting solution ordinarily contained a residue of undissolved metallic impurities and often a slight preci itate of white salt. The clear solution was decanted as compfetely as practical into a 400-ml. beaker. Concentrated hydrochloric acid was then added to the rcsidue arid solution remaining in the 150-ml. beaker. This mixture was evaporated until crrstals formed. The solution was cooled and examined for evidence of remaining metallic residue. If any metallic residue remained, the treatment was repeated, with aqua regia replacing the concentrated hydrochloric acid. A sample of sperific genre usually received initial treatment with aqua regia a result of analytical experience. The liquid and precipitate resulting from the treatment with concentrated hydrochloric acid or aqua regia s e r e combined with the initial solution in the 400-ml. beaker. The chloride ion introduced caused partial precipitation of lead as lead chloride. Basic bismuth salts also precipitated if the acid concentration was lowered excessively. Acid was added to return such bismuth salts to solution. The total mixture was heated to redissolve the lead chloride, EO that the clear solution could be re-examined for traces of undissolved sample. A magnet was used here as previously described. The total solution was evaporated to a volume permitting the preparation of a solution containing 15 to 20% sample on a weight-to-volume basis. The lead chloride which reappeared upon cooling was transferred with the clear solution and treated as a part of it.
z
I
0.2
I
I
I111Il
0 1.0 CONCENTRATION, P.P.M.
Figure 3.
A
2
2981.651 Nil2659.454 Pt
0 Single elements 0 Multiple element systems 0
I-
Bismuth matrices Pb-Bi matrices
5
I-
z
Samples containing elements such as niobium, tantalum, and tungsten, which are only slightly soluble in nitric acid, required more extensive treatment. ~ I u l t i p l ecomponent samples containing any of these elements were dissolved as previously described and filtered, and both resi-
due and filtrate were analyzed for the elements, the residue being dissolved separately for analysis. Samples containing only one of these elements for analysis could be treated with greater ease. This type of sample was dissolved in nitric acid only and filtered, and the filtrate was analyzed for niobium, tantalum, or tungsten as indicated. Only about 1% of the total contaminant of this type of sample went into the initial filtrate. The filter paper containing the residue from this treatment was dry-ashed and treated with hot concentrated sulfuric acid, fuming sulfuric acid, or a mixture of hydrofluoric and nitric acids in order to dissolve niobium, tantalum, and tungsten, respectively. These solutions were diluted to known volumes after the internal standard was added. They were then excited as quickly as possible. I n no case wu9 a solution allowed to remain overnight before use. The solution containing tungsten was especially sensitive and tended to precipitate tungsten trioxide when it was diluted to facilitate handling. The analytical results of the separate solutions were combined in calculation for report purposes. One advantage of the two-step method of dissolving the samples
0
Single elements
A
Pb-Bi matrices
0 Multiple element systems 5 Bismuth matrices
Sfter the samples were dissolved, the solutions were transferred to suitable volumetric flasks and adjusted to contain 20% bismuth (or lead-bismuth) on a weight-to-volume basis and approximately 20% with respect to nitric acid and/or hydrochloric acid. Separated sample solutions not containing lead-bismuth were prepared with as small a volume a8 convenient in order to provide the highest possible concentration of analytical elements. Solutions of either of these types were then prepared for excitation by
580
A N A L Y T I C A L CHEMISTRY Table I
Analysis Element Be
Line,
Matrix Bi
co
Bi
Cr
Bi, Pb-Bi
Fe
Bi, Pb-Bi
Mn
Bi, Pb-Bi
MO
Nb
Bi, Pb-Bi a
Bi, Pb-Bi Ta
a
Ti
Bi
V
Bi
W
b
A. 3131.072 2650,781 3443.641 3474.022 3021.558 2766.540 3593,488 3605.333 2739.546 2599.396 2626.585 2794.817 2801.064 2810.200 2816.154 3132.594 3028.443 3032.768 3194.977 2950.878 2981,651 3054.316 3050.819 2675.901 3214.750 3465,562 3186.454 3254,250 2880.030 2952.075 2944.395 2946.981
Range, P.P.M.
0.5-0.06 5-0.6 100-10 100-10 100-10 100-10 20-2 20-2 100-10
100-10 100-10
10-1 100-10 100-10 100-10 10-5 100-10 ’ 100-10 100-10 100-10 1000-100 1000-100 100-10 1000-100
1000-100 1000-100 100-10 100-10
100-10 400-50 400-50
2659,454 2929.794 2997.967 b
CALCULATIONS
The photographed spectra were all photometered using an ARL film densitometer. Partial background correction was made by substraction of values reIated to the true intensity values through the plate calibration curve. Intensity ratios for analytical purposes were calculated for the wave-length pairs of Table I. The use of the platinum lines a t 2659.454, 2929.794, and 2997.967 A. was dictated by interference factors noted. OccasionaIly the 3132.594 A. molybdenum line was used as a refcrence line for the 3131.072 and 2650.781 -4. beryllium lines.
100-10
P t (reference lines)
.Z
former. The sample was in the positive electrode of the polarized spark which crossed a 2-mm. gap. Spectra were photographed in the 2480 to 3700 A. region with a Baird 3-meter grating spectrograph. A 50p slit was employed. In most cases, duplicate spectra were recorded on SA-1 film, which was processed under carefully controlled conditions in the conventional manner. Some consideration was given to the effect on the excitation of polarity in the analytical gap. A slight increase in sensitivity with the sample in the anode was noted. This is in agreement with the observation of Jolibois ( 2 ) . A 0-2 ampere r-f ammeter was installed in series with the analytical gap for use during porous electrode excitation. This meter provided sensitive electrical indication of conditions in the analytical gap. I t was observed that concentrated salt Rolutions were excited with increased electrical stability.
Pib, Ta, separated from matrix. Excited in solution containing H I S O I . W, separated from matrix. Excited in presence of fluoride ion.
using them to fill 10-ml. volumetric flasks to which 75 lambda of a 1%platinum solution had previously been added. The sohtion, ready for excitation, therefore contained 75 p.p.m. of platinum (weight-to-volume) as an internal standard. Samples containing elements requiring special treatment were handled in a manner analogous to that employed in the bismuth matrix samples.
DISCUSSION
Intensity ratios obtained from duplicate excitations of samples agreed within 20% for lines having transmission values from 5 t o 90%. In most cases, the agreement was within less than 5%. The authors’ experience employing the porous electrode technique substantiates the claim of Feldman ( 1 ) for an inherent precision of the method of about 2%. The requirements of their application were not, however, rigorous enough to demand limiting precision; and the authors took over-all advantage of the situation in a manner which permits them to claim only the precision given above. In many instances, for the analysis of samples from special experimental work for specific elements they obtained precision of the order claimed by Feldman. LITERATURE CITED
( 1 ) Feldman, C., ANAL.CHEM., 21, 1041 (1949); 5. S. Atomic Energy EXCITATION
The solutions were excited in porous cup electrodes according to the method proposed by Feldman (1). Excitation was furnished by the high voltage section of an ARL high precision source unit with 60 volts to the primary of the high voltage trans-
Commission, AECD-2392. (2) Jolibois, P., Compt. rend., 202, 400 (1936). RECEIVED August 12, 1950. Presented before the Division of Bnalytical Chemistry a t the 118th Meeting of the . ~ M E R I C A K CHEMICAL SOCIETY, Chicago, Ill.
Differential Spectrophotometric Determination of High Percentages of Nickel ROBERT BASTIAN, Sylvania Electric Products, Inc., Kew Gardens, N . Y .
R
ECENT papers (2-4) have emphasized the great accuracy and precision which are possible in colorimetry when a solution of high absorbancy (for nomenclature see S , 5 , 6 ) is compared to a similar standard rather than to the solvent. The terms ‘‘differential colorimetry” or “method of transmittance ratios” have been applied to this procedure. A general mathematical treatment has been given by Hiskey ( d ) , and a treatment applied specifically to the Model DU Beckman spectrophotometer has been made by Bastian, dWeberling, and Palilla ( 3 ) . Using a reference standard in place of the solvent, the absorbancy (optical density) scale on the Reckman instrument is set
for zero by making the slit aperture wider than usual. Somewhat more concentrated solutions are then read against this standard. For systems obeying Beer’s law at the wider aperture, the resultant absorbancy scale readings are directly proportional to the difference in concentration between the zero point standard and the solution in question. The term A: is used here for the absorbancy scale reading obtained against the zero point standard. Then:
A:
1
log -
T: