precipitated strontium nitrate. h second nitric acid treatment is usually required for adequate removal of calcium. The strontium is purified by the standard barium chromate and hydroxide precipitations, after which strontium carbonate is precipitated for determination of total radiostrontium. After total radiostrontium activity lias been counted, the strontium carbonate precipitate is stored for yttrium ingrowth and the yttrium is extracted and backextracted as described above. Loss of strontium in the magnesium precipitate is only about 2% and precipitation from the sodium hydroxidesodium carbonate so!ution is quantitative. Over-all chemical recovery is about 85%, most of the loss being in the nitric acid treatments. Determination of Radiostrontium in Soils and Sludges. T h e sample is
converted to ash in a muffle furnace and a n aliquot of 1 to 50 grams is taken for analysis. After addition of strontium and barium carriers and drying, t h e ash is mixed with 5 times its weight of sodium hydroxide and fused in a nickel crucible. Sodium carbonate is added t o t h e melt and the misture is heated again. The fusion mixture is taken up in hot water to complete disintegration of the solid, and centrifuged; the supernatant is discarded. The residual solid is dissolved in hydrochloric acid and the strontium precipitated as carbonate by addition of ammonia and sodium carbonate. Strontium is purified by nitric acid treatment, barium chromate precipitation, and hydroxide scavengings before counting for total radiostrontium content or separation of yttrium-90. I n these samples, nitric acid treatment usually is necessary, even if only strontium-90 is to be determined. If only leachable radiostrontium is
desired, the original sample is extracted several times with hot 6M nitric acid. After addition of carrier, the acid is neutralized and the strontium precipitated with carbonate, which is then treated in the same way as the carbonate obtained after fusion. Radiostrontium in Biological M a terials. T h e sample is converted t o ash, usually in a muffle furnace. For some samples, however, wet ashing with nitric acid followed b y a n equal-volume mixture of nitric and perchloric acids may be preferred. A suitable aliquot is taken, dissolved in hydrochloric acid, and evaporated t o near dryness. T h e chlorides are converted to nitrates by evaporation with nitric acid and strontium nitrate is precipitated from concentrated nitric acid. After a second precipitation from nitric acid, the precipitate is dissolved in water and made alkaline with ammonia to test for completeness of phosphate removal. Any precipitate (indicative of unremoved phosphate) is digested with sodium carbonate to convert the phosphate to carbonate. This is dissolved in acid, warmed to drive off carbon dioxide, and again made alkaline with ammonium hydroxide. If a precipitate forms, it is discarded. Sodium carbonate is added to the combined strontium-containing supernatants to precipitate the strontium as carbonate. This precipitate is collected and purified by barium chromate precipitation and hydroxide scavenging. In all cases with biological samples, treatment with nitric acid is necessary to prevent phosphate interference. ACKNOWLEDGMENT
The authors express indebtedness to
H. L. Krieger and L. P. Jarnagin, Robert A. Taft Sanitary Engineering Center, for assistance in developing some of the procedures described.
LITERATURE CITED
( 1 ) Bryant, F. J., Chamberlain, A. C., Morgan, A . , Spicer, G. S., Brit. Atomic
Energy Research Establishment, Paper AERE-HP/R-2056(1956). ( 2 ) Federal Regzster 22, 19 (1957). (3) Glendenin, L. E., Paper 236, Xational Nuclear Energy Series IV, 9, “Radiochemical Studies: The Fission Products,” C. D. Coryell and N. Sugarman, eds., p. 1460, McGraw-Hill, Sew York,
1951. (4) Goldin, A. S., U. S. Atomic Energy Comm., Doc. TID-7517 (Pt. Ib), 323 (1F)R6I . ( 5 j Hahn, R. B., Straub, C. P., J . A m . Water Works Assoc. 47, 335 (1955). (6) Harley, J. H., U. S. Atomic Energy Comm., Doc. NYO-4700 (1957). (7) Kooi, J., ANAL.CHEM.30, 532 (1958). (8) blartell, E. A., U. S. Atomic Energy Comm., Doc AECU-3262 (1956). ( 9 ) Mawson, C. ii., Fisher, I., Natl. Research Council, Canada, Doc. CRM455. (IO) Milton, G. AI., Grummitt, W. E., Can. J . Chem. 35, 541 (1957). (11) Xational Committee on Radiation
Protection, Natl. Bur. Standards Handbook 52 (1953). (12) Sax, N. I . , Gabay, J. J., Revinson, D., Keisch, B., U. S. Atomic Energy Comm., NYO-4604 (1954). (13) Stanley, C. W., Kruger, P., Nucleonics 14 ( l l ) , 114 (1956). (14) Sunderman, D. N., Meinke, W.W-., A N ~ LCHEM. . 29, 1578 (1957). (15) Sverdrup, H. V., Johnson, N. W.,
Fleming, R. H., “The Oceans,” Prentice-Hall, New York, 1942. (16) Volchok, H. L., Kulp, J. L., Eckelmann, W. R., Gaetjen, J. E., Snn. N . Y Acad. Sci. 71, 295 (1957). (17) Willard, H. H., Goodspeed, E. W., 1x0. ENG. CHEM., ANAL. ED. 8, 414 (1936).
RECEIVED for review February 13, 1959. Accepted May 25, 1959. Division of Rater, Sewage, and Sanitation Chemistry, 134th Meeting, ACS, Chicago, Ill., September 1958. Part of the research reported was supported by the U. S. Atomic Energy Commission under Contract -4T(49-5)-1288.
Analysis of Thin Metal Films by Neutron Activation BARBARA A. THOMPSON General Engineering laborafory, General Electric Co., Schenecfady, N. Y.
b Gamma spectrometric and radiochemical methods previously used to measure trace impurities in very pure materials have been successfully a p plied to the analysis of very thin metal films deposited on glass. The sample weight is approximately 50 y and conventional analytical methods can measure only those impurities present at relatively high concentrations. Methods are described for the quantitative determination of 1 1 elements on a single sample. Additional elements can b e determined by modifying the chemical separation schemes. 1492
ANALYTICAL CHEMISTRY
N
activation analysis is ordinarily connected with the measurement of trace impurities where very high sensitivity is required. Occasionally, however, i t is necessary to analyze very small samples for impurities over a wide range of concentrations. Under these conditions, conventional methods can give only limited information. For example, certain magnetic applications utilize very thin metal films produced by vapor deposition. Even small concentrations of impurities may affect the magnetic properties, and thus a method of analysis for many elements at trace EUTRON
concentrations or higher is required. Because of the small sample size, conventional spectrographic analysis will detect only those impurities present a t concentrations of several per cent. As neutron activation analysis can detect extremely minute quantities of many elements, it is well suited to the analysis of these small samples. The procedures used and the results obtained in applying activation analysis to the measurement of a number of impurity elements in thin filnis consisting primarily of nickel and iron are presented.
T S A M P L E A - 0,OT
M A K E 1N IN H C I
Figure 1. Chemical separation scheme for copper, cadmium, tin, molybdenum, tungsten, chromium, nickel, and iron
I ADD CARRIERS
I WARM
I
I
PRECIPITATE
SOLU TI 0 N
DISSOLVE IN N H 4 0 h P U R I F Y AND F I N A L L Y PRECIPITATE AS
PRECIPITATE
SOLUTION
I 0 Fe, N i , C r
NH4OH, H2S
WASH W I T H 1N NoOH I
1
I
I
~~
SOLUTION
PRECIPITATE
+ SOLUTION
I
A 7 EVAPORATE TO DRYNESS
I PRECIPITATE
h
DISSOLVE IN H 2 0 AND PRECIPITATE AS
&
Cu M E T A L
1
ACIDIFY TO RE-PRECIPITATE SULFIDES, DISSOLVE I N AQUA REGIA AND DILUTE ;DD Q - BE!ZOIN OXIME,
I
PRECIPITATE
S3,FeS,NiS
DISSOLVE IN AQUA REGIA AND BOIL DILUTE, ADD N H 4 0 H
ADD CONCENTRATED H C 1
D I S S O L V E IN "0) B O I L TO DRYNESS ADD HC1, DILUTE ADD CrC1Z
Cr
+
PRECIPITATE
tARRY
TO
FINAL PRECIPITATION
1- (
+
PRECIPITATE
SOLUTION
]*
SOLUTION
PURIFY AND FINALLY ELECTROPLATE
&I I
PRECIPITATE SnS2, DISSoLVE I N "3' DISSOLVE, PRECIPITATE 0 x 1 D ~ ~ D w ~ ~ 0 ~ ' 0 4 WITH CUPFERRON AND IGNITE
&
PRECIPITATE
& IGNITE
SOLUTION
PURIFY AND PRiCSIPITATE I
EXPERIMENTAL
Elements to B e Measured. Eleven elements in each sample were determined: cadmium, chromium, copper, iron, molybdenum, nickel, phosphorus, tantalum, tin, tungsten, and zinc. A number of other elements, such as oxygen, nitrogen, and carbon, were of great interest b u t could not be measured by this technique. Elements such as aluminum and magnesium, which yield short-lived nuclides on ir-. radiation, mere of interest and could be measured by making a special run a t the irradiation facility. This would have consumed the entire sample. X u c h more information per sample could be obtained by analyzing for the longerlived nuclides mentioned above, and, therefore, the analytical procedure was developed with these measurements as its objective. Two aliquots were used and the separation and purification procedures are shown schematically in Figures 1 and 2. I n general, the methods were adapted from those applied to silicon in this laboratory ( 7 ) . Many of the final purifica-
tion schemes were adapted from those of Kleinberg (4). I n the tantalum procedures, the carrier is prepared according to the method of Linder (6) and the tantalum is complexed as the oxalate, thus preventing its precipitation during the initial part of the procedure. Analysis of Samples. T h e samples t o be analyzed were thin metal films produced by vapor deposition on glass. T h e primary constituents were nickel and iron. Each sample was about '/4 inch in diameter and weighed approximately 50 y, exclusive of the glass backing. The samples were analyzed for nickel and iron using the General Electric microemission x-ray spectrometer (1). Typical results are discussed by Chu and coworkers ( 2 ) . The particular samples used for this study were all 80% nicke1-20% iron to a precision of approximately &3%. Xext, the samples were irradiated for 24 hours a t a flux of 2 X l o i 3 neutrons per sq. cm. per second in the CP-5 reactor at Argonne Kational Laboratory. T o prevent contamination by reagents, the samples were irradiated while still attached to the glass. As an example of the need for this precaution, if only two drops of nitric
acid were used to dissolve the film, 0.07 y of copper would be introduced. I n a sample weighing 50 y, this is 0.14%, much too high a blank. After irradiation, the samples were carefully rinsed with distilled water to remove surface dirt. The metal was then dissolved away from the glass by warming in 1 to 1 nitric acid; about 10 seconds was required for this dissolution. As there is a fairly high probability of leaching a t least traces of sodium from the glass during this operation, any sodium activity observed in the dissolved samples was ignored. The possibility of contamination of the sample by impurities in the glass was considered. However, as the glass contained appreciable amounts of zinc and zinc was not detected in the dissolved film, this source of error appears to be unimportant, a t least for the particular elements studied. More conclusive information about this source of error would have been obtained from a blank piece of glass, irradiated and leached under identical conditions: this step should, therefore, be included as a routine part of the procedure. The glass became extremely radioactive and it was necessary to perform the preliminary dissolving operations beVOL 31, NO. 9, SEPTEMBER 1959
*
1493
hind a lead shield. After removal of the glass, very little activity remained associated with the samples and all further work was performed without shielding. Immediately after dissolution, the radiation from each sample was analyzed with a 100-channel gamma scintillation spectrometer. Those elements present at high levels or for which the sensitivity is very high were measured in this way. The recent catalogue of gamma-ray spectra issued by Heath (3) proved very useful in identifying the observed photopeaks. Figure 3 shows a typical gamma scan of a sample which contained iron, copper, and tungsten. The gamma peaks from the different radioisotopes overlap and thus chemical separations are necessary for a more accurate determination of the concentrations of many of the elements in the sample. Tungsten, iron, and sodium could be determined quite accurately without chemical separations, merely by following the decay of their gamma peaks and comparing to standards. Copper would be somewhat more difficult because the composite decay curve of the 0.5-m.e.v. peak would have to be resolved and the activity due to tungsten subtracted out. The precision of the net activity would be the square root of the sum of the two activities and, therefore, somewhat poorer than that obtained by measuring a separated copper fraction. For the other elements of interest, which are not detectable by gamma spectrometry, the maximum concentration levels can generally be established with greater sensitivity by employing chemical separations t o remove the other interfering radiations. This is particularly true of relatively shortlived impurities whose activity may be completely masked by longer-lived activities. Separations are, of course, mandatory for any elements of interest which are pure beta-emitters-e.g., phosphorus. I n view of these considerations, the other elements of interest which were not observed by gamma spectrometry mere measured by beta-counting following the radiochemical separations described above. I n each case, approximately 20 mg. of carrier was added from a standardized solution. The chemical yields varied from sample to sample, and, except for tungsten and zinc, were never below 35%. I n all cases, including tungsten and zinc, the gravimetric factors made i t possible to determine the yields to a precision of at least =k2Y0, usually about *0.5%. This was, in general, considerably better than the counting precision. The total time involved per batch of four samples, when i t is necessary to perform chemical separations for each element of interest, is about two weeks for a single operator. This time can be reduced approximately one day for each element that can be determined by gamma spectrometry alone without separations. The first measurements on the separated copper fractions were made at about 14 hours after discharge from the reactor. This delay is im1494 *
ANALYTICAL CHEMISTRY
M A K E 1 N I N HC1 I ADO CARRIERS
ADD ZrO:tt ,HEAT
PRECIPITATE
SOLUTXW
DISSOLVE IN H F P U R I F Y AND FINALLY PRECIPITATE
ADD N H 4 0 H
I,,,,,;]
Figure 2. Chemical separation scheme phosfor zFnc, phorus, a n d . tantalum
I I PRECIPITATE
SOLUTION
Q'
DISSOLVE IN HF PURWY, ADD ",OH AND IGNITE
NEUTRALIZE AND ADO H2S
DISSOLVE IN " 0 3 ADD ZINC,REAGENT
posed primarily by transportation limitations and is a minimum possible under the best conditions when either the Argonne or the Brookhaven reactor is used. This delay could, therefore, be reduced appreciably for laboratories located closer to such irradiation facilities. A standard end-window GeigerMuller counter was used for betacounting all fractions except nickel. Because of the low energy of tlie nickel63 beta-particles, it was necessary to measure these fractions with a gas-flow counter. For calibration purposes, standards consisting of pure oxides of the elements of interest were irradiated under conditions as nearly identical as possible to those under which the samples were irradiated. The radiations from the samples were then compared to those standards. Throughout the analyses every effort was made to minimize the various sources of error as discussed by Plumb and Lewis (6).
-1
w
z z a X
U
K W
a w I-3
E r K
a W I-v)
z
3 0 u
1
RESULTS
Element Cd Cu Cr 1210 P
Ta
w
Sn
Zn
Impurities in 2 0 7 0 iron-8070 Nickel Films
Concentration, Per Cent Sam- Sam- Sam- Sample 1
ple 2