ment interference is almost completely negligible in the samples tested here, and it is unlikely that interference is a serious limitation in analysis of most rocks and minerals on the earth’s crust. In every case, however, purity of thc radiation may be tested by decay curvc mcasurement, and supplementary tracc element analyses by other means may be the basis for calculated corrections. If in any ease a large cnlculated correction must be rrpplicd, it is recommended that a n experimentally determined interference factor F be used rather than estimated values from Table VI. If the procedure is applied to the routinc analysis of rock or mineral samplcs, the counting may bc carried out using a n automatic samplc changer, and many of the calculations may be programmed for a digital computer. I n this case, it may be ncccsssry to revise the order of counting samples with the various absorbers from that recommended here. However, if decay corrections are applied to all of the counting data, the result should be equally accurate. With properly automated operations, thc proccdure should
be rapid, and the cost per sample should be small. ACKNOWLEDGMENT
The participation of M. I. Goldstein, D. H. Anderson, L. H. Cohen, G. Faure, and G. L. Sehroeder and the support of P. hl. IIurley, W.H. Pinson, C. D. Coryell, and J. W. Irvine, Jr., are gratefully acknowledged. M. L. Bottino and E. M. Barrall carried out supplementary spectrographic and colorimetric analyses. Work supported in part by a grant from a National Science Foundation. LITERATURE CITED
(1) Abbey, S., Maxwell, J. A., Chem. in Can. 12. 37 (1960). (2) Ahrens, L: H.,’ Phys. Chem. Earth 2 , 30 (1957). (3) Aldrich, L. T.,Wethcrill, G. W., Ann. Rev. Nuclear Sci. 8, 257 (1958). (4) Bradley, J. E. S., Bradley, Olive,
Mineralogical Mag. 31, 164 (1956). (5) Brownell, G. M., Bramadat, IC, Knutson, R. A., Turnock, A. C., Trans. Roy. SOC. Can. 51, Scr. 111, Scct. 4, 1957, 19. (6) Fairbairn, H.W., Geochim. et Cosmochim. Acta 4, 143 (1953). (7) Hine, G. J., Burrows, B. A., Apt, L.,
Pollycove, M., Ross J. F., Sarkee, L. A., Nucleonics 13, d o , 2, 23 (1055). (8) Hughes D. J., Magurno, B. A., Brueeel, h. K., Suppl. No. 1 to U. 8. At. Energy Comm., Rept. BNL-325, January 1, 1960. (9) Hughes, D. J., Schwarta, R. B., U. S. At. Ener Comm., Rept. B N t 325,2nd ed. JuY 1, 1958. (10)Hurle b. M. U, 8. At. Ener Comm.&pt, NYd-3939, p. 48, M . 1 . K December 1958.
(11) Zbid., NYO-3940, p. 210, M.I.T., December 1959. (12) Ibid., Rept. NYO-3941, p. 274, hf.1.T.’December 1960. hf.1.T.’ (13) Reid, A. A. F., F. Caldwell, H. L., Vanatta, J. C.. Arch. hiochem. Biochem. Biouhvs. . I 84. 498 (195d). (14) Stevena, R. E.,et al., U . S. Geol. Szirv. Bull. 1113, p. 38 (1960). (15) Strominger, D., Hollander, J. hI., Swborg, G. T., Rev. M o d . Phys. 30, 585 (1858). 116) U. S.National Bureau of StnndardR. ‘ Circ. 398, Suppl., 1949. (17) U. S. National Bureau of Standards, Circ. 552, 3rd ed., 1959. (18) Winchestcr, J. W., J . Geophys. Res. 64, 1130 (1959). (19) Winchester, J. W., Prog. Inorg. Chem. 2 , 1 (1960). (20)Winchrster, J. W., Tsana. Am. Geophys. Union 39, 530 (1958). ’
RECEIVEDfor review December 2, 1960. Accepted April 17, 1961.
X-Ray Absorption Edge Determination of Uranium in Complex Mixtures E. A. HAKKILA University of California, Los AIamos Scientific laboratory, 10s Alamos, N.
b Uranium in the concentration range between 0.5 and 40 mg. per mi. is determined by an x-ray absorption edge procedure. In this procedure a molybdenum-niobium target is bombarded b y the radiation from a tungsten-target x-ray tube and the intensities of the fluorescent x-rays of molybdenum and niobium are measured after they have passed through an absorptian cell filled with the sample solution. The measured intensities are compared to the intensities transmitted through the same cell filled with water. The Llil absorption edge for uranium occurs at a wave length between the wave lengths of the fluorescent x-rays from the target, and the transmitted intensities are related to uranium concentration by standard absorptian principles. For determining between 0.5 and 40 mg. of uranium per ml. the relative standard deviation of the method varies from 6 to 0.34%. Defining sensitivity as that concentration of uranium equivalent to three times the standard deviation of determining a blank, 0.06 mg. 1012
ANALYTICAL CHEMISTRY
M.
of uranium per ml. can b e measured when a cell of 3-cm. path length i s used. Only yttrium interferes seriously with the procedure. Approximately 5 minutes are required to convert the instrument from normal x-ray fluorercence operation to absorption edge analysis. Twenty to forty analyses can be performed daily.
T
IIE NEED for a rapid procedure for determining uranium in various complcx mixtures led to thc invcstigation of the application of x-ray spectrographic methods for the determination of uranium, but these procedures are not applicable to a wide variety of samples unless either the uranium is first separated from other elements or separate calibration curves arc prepared for each type of sample. A further disadvantage of x-ray fluorescence is the nonlinearity of these calibration curves at high uranium concentrations. Peed and Dunn (10) and Wright and Barringer (19)have described the x-ray absorption edge determination of ura-
nium. The absorption edge technique is theoretically less susceptible to interference by other elements and is more precise for higher concentrations than is x-ray fluorescence. In x-ray absorption rdge analysis the transmitted intensities for two x-ray energies located on each side of a suitable absorption edge of the element being determined are related to the concentration of that element. The absorption edge technique was first applied to quantitative analysis by Glocker and Frohnmeyer (8) in 1925, but few practical applications were made in the next two decades because no suitable detection systems were available. More recently, Engstrom (6) presented an excellent review of the application of the technique to analysis of biological tissues. He used a microfocus x-ray tube and utilized either the x-ray tube continuum or secondary radiation from selected pure elements (6) to obtain desired x-ray energies. Various authors (1, S, 4, 10, 1 1 , IS) have described the use of the continuous radiation from conventional x-ray tribes in conjunction with com-
/
X-RAY TUBE
I-""' /
SPRING
,-, LIF CRYSTAL
Figure 1. Optical arrangement for x-ray absorption edge determination of uranium
mercial crystal spectrometers to oubain the desircd x-ray energies. Hughes and Hochgesang (9) suggested the use of a thorium-target x-ray tube to utilize the intense primary thorium spectrum for the determination of lead in gasoline, but noted difficulties in preparation and maintenance of these tubes. Because the use of secondary radiation requires a minimum of instrument modifications, and provides high line intensities with no detectable secondorder interference, its application to x-ray absorption edge analysis ww further investigated. The secondary K a radiation of molybdenum and niobium emitted by a 50% molybdenumniobiurii alloy mounted in the sample drawcr of a commercial I'hilips Electronics x-ray spectrograph was utilized to obtain the desired radiation for the absorption edge determination of uranium. The LIII absorption edge for uranium falls between the K a linea for molybdenum and niobium. Although the discusaion is limited to the determination of uranium, the procedure can also be applied to the determination of yttrium. APPARATUS
A Phillips Electronics three positionhead x-ray spectrograph with a lithium fluoride analyzing cr stal and a scintillation detector wit{ associated high speed electronics was modified for absorption edge analysis (Figure l ) , The exit collimator was replaced by a l/lCinch thick lead sheet having a 15by 15-mm. window centered to coincide with the collimator port. The lead window and spring clips to hold either a 1- or 3-cm. path length absorption cell in front of the lead window are hrld in place by the collimator screws. The cell of 1-cm. path length is used for analyzing solutions containing up to 40 mg. of uranium per nil. The longer ccll is used for analyzing uranium concentrations in the range of 0.06 to 2 mg. prr ml. The cells are constructed of Lucitc with l/le-inch thick polystyrene windows remented in place with chloroform. Polystyrene is preferred to Lucite or beryllium as window matrrial brcaupe of its greater stability toward nitric acid. The cells are 50 mm. wide and 60 mm. high with a rescrvoir 20 mm. wide and 50 mm. high. A minimum of 5 ml. of sample is required for analysis using the 1-cm. path length cell.
The source of secondary radiation is a 50% molybdenum-niobium alloy polished flat on one side and machined to fit into the standard sample cell supplied with the x-ray spectrograph. Either prepared alloys or pelleted oxide mixtures can be used as sources of secondary radiation, but the former are preferred because scattering of primary radiation from the x-ray tube is minimized. REAGENTS
Analytical reagent grade chemicals and low conductivity water were used in all reagents. The uranium standard solution was prepared as follows: Clean approximately 12 grams of high purity uranium metal with nitric acid to remove oxide film. Wash with water, then acetone, and air dry. Weigh 1O.OOO grams of the metal into a 100-ml. volumetric flask and dissolve carefully with nitric acid. Dilute to volume with distilled water. PROCEDURE
Samples received as solutions are analyzed as received, or the uranium concentration may be adjusted by dilution or evaporation. Solids are dissolved in acid, preferably nitric, and are diluted with water to give uranium concentrations of 10 to 20 mg. per ml, If the uranium concentration is high, or if impurity elements of high mass absorption coefficients are present in large amounts, it may be necessary to dilute the sample further so that sufficient count rates may be obtained. If the yttrium concentration is greater than 0.01 mg. per ml., it should be rtmoved by precipitation as the fluoride, A blank solution is prepared to contain approximately 1 to 5 mg. per ml. of each of the major elements expected in the sample, and the solution is diluted to a total volume of 10 ml. with a 2 t o 5 N solution of the acid used in sample dissolution. In the same manner a standard is prepared to contain 20.00 mg. of uranium per ml. The x-ray tube, which is operated a t 50 kv. and 30 ma., is turned on 1 hour before analyses are performed. Then the absorption cell is filled with water, and duplicate mcasurements are made of the time to accumulate 106 counts at the Zia lines for molybdenum and niobium. The measurcmente are repeated with the sample cell filled with the blank, then with the standard solution. Measurements are made for all
samples, accumulating 106 or 105 counts, the number chosen so that counting time for any one line does not exceed 10 minutes. At the conclusion of the analyses, the water and blank solution mettsurements are repeated, and the average times to accumulate 106 counts are calculated. All times are corrected for coincidence loss hy fiubtinctiiig 4.0 seconds per 106 counts ttlkan Count rates lower than 5000 C.P.S. we also corrected for background. Appropriate values for kl tlnd kl for Equation 4 are ddcrmined from the blank and standard soiutions, and the uranium concentration of each sample is calculated. The procedure can also be applied without modification to the determination of yttrium in the absence of uranium. DISCUSSION
X-rays passing through a sample are absorbed according to the equation I
= I@-[P/P)cL
( 1)
where
Io is the incident intensity I is the transmitted intensity (p/p)
is the mam absorption coefficit,nt
c is the concentration expressed in grams per cc.
and L is the path length of the cell The concentrations and mam absorption coefficients for all elements in the sample must be considered. Mass absorption coefficients for any element can be related to wave length by the equation
-
h" (2) In thin equation k is a constant for the element over a wave-length region where no absorption edge occulg but is ditrerent for all elements, and n is constant for the element over a limited wavelength region. If absorption measurements are made at two wave lengths selected on either side of an absorption edge for the element being determined, the mass absorption coefficients for that element a t the two wave lengths differ by a large factor, while absorption coefficients for other elements differ according to the equation (PlP)
where A ( p / p ) is the difference in mass absorption coefficients at the two wave lengths. If A1 and A, are selected sufficiently close together, A ( p / p ) becomes negligibly small. Experimentally, using the x-ray tube continuuni as the source of x-rays, the proximity of the two wave lengths is limited by the resolving power of the crystal in the spectrogoniometer, usually of the order of 0.02 A. with a lithium fluoride analyzing crystal. It is not always possible to obtain x-ray lines with the optimum separation, and jn the present study a separation of 0.038 A. occurs. As has been shown VOL. 33, NO. 8, JULY 1961
1013
by Dodd (4), Engstrom (6),and Peed and Dunn (IO), the concentration of the element sought can be related to measured intensities by the equation
IW
kl log 1 - kz log
c
IW
(4)
. I1
In this equation 2303 X
k, = L[h
;( M l P h - '1 2303 A
(5) (P/Ph]
7
c is the concentration in milligrams per milliliter and IT, IT, 1 2 , and I! are intensities transmitted through the sample and water a t the respective wave lengths. Peed and Dunn (IO), Stewart (If), and Wright and Barringer (IS) agsumed n to be constant regardless of the impurity element present, and, therefore, kl and kt to be the same for all impurity elements. However, Victoreen (II) has shown that n is dependent on wave length and atomic number, especially a t wave lengths just
Table 1. Maximum Allowable Concentrations of Various Impurities in Determination of 1 Mg. per Milliliter of Uranium in 2N Nitric Acid Solution
MiniMaximum mum UThat Concen- Can Be tration Detected, (PIP)=Mg./Mi. %b
Element Al
5.22 38.5 48.1 58.0
Fe
Ni Zn Cd Th
380 57 45 38 68 24 17
);;(
Hg
(130)
* After
0.7
0.8 0.4 1.3 1.8
Allen (8). Values in () are
estimated. * Based upon
Table II.
0.1 0.5
u =
0.03mg. per ml.
Determination of
kl/kz
Atomic Element c1 Ca Fe Ni
Zn
Rb
Sr
Cd Nd
3
a
Th After Allen
No. 17 20 26 28 30 37 38 48 60 80 82 90 (g).
0
INTERFERENCES
Interferences in absorption edge analysis can be classified into three categories: absorption edge interference, limitations due to high mass absorption coefficients of impurity elements, and variation of the exponent n in Equation 2. Each of these factors was studied as it applies to the determination of uranium, and to absorption edge analysis in general.
in Presence of Various Elements Dissolved in Nitric Acid
Region Where Mo and Nb K a Lines Occur Below K edge
At K edge Between K and LI edges Between LII and LIIIedges Values in () are eatimated.
-.
1014
below an absorption edge. The present investigation has shown that n can be assumed constant for all elements when the product of the mass absorption coefficient and the concentration in milligrams per milliliter does not exceed approximately 500. The effect of higher concentrations of impurity elements is minimized or eliminated by adding these elementa to blank and standard solutions used in determining values for kl and kr in Equation 4. Because counter response is not linear above approximately 5OOO c.p.s. due to coincidence low, all measured count times must be corrected for the resolving time of the counter (7). In this work, the resolving time of the scintillation counter was determined to be 4 pseconds using the multiple foil technique. In addition, for count rates be,low approximately 5000 c.P.s., the electronic noise of the counter (approximately 10 c.p.8.) becomes significant and a correction must be applied if precision of the order of a few tenths of a per cent is sought. The relative sensitivity of the absorp tion edge technique for various elements has been discussed by Engstrom (6). Generally, the LIII edge is more sensitive than the K, LI, or LII edges. Although better sensitivity is theoretically possible using the M Y edge, this long wave-length radiation would be almost completely attenuated by the sample cell and matrix when applied to solution analysis. Therefore, the LIII edge is the most suitable for the determination of uranium.
ANALYTICAL CHEMISTRY
kdkr 1.157 1.155 1.148 1.145 1.142 1.150 1.162 1.145 1.143 1.140 1.138 1.129
2N
All elements having an absorption edge between the two x-ray lines measured, in this case between the K a lines for molybdenum and niobium, will contribute to the concentration found. Only yttrium, polonium, protactinium, and radon have absorption edges in the region of interest, and only yttrium will normally be found in samples in concentrations large enough to be of concern. Yttrium produces an error of approximately 1.5 mg. of uranium per mg. of yttrium present. It can be removed readily by fluoride precipitation. From statistical considerations the precision of a determination based upon lo6, lo6, and lo4counts is, respectively, 0.1, 0.33, and 1.0%. Therefore, at least lo6 counts should be accumulated a t each of the x-ray lines if precision of 0.5% or better is anticipated, and transmitted intensities should yield count rates greater than 200 C.P.S. in order that this number of counts may be taken in a reasonable length of time, preferably less than 10 minutes. To attain these count rates, especially for high uranium concentrations, the mass absorption coefficients of other constituents in the sample should be low or the concentrations of elements of high mass absorption coefficient should be low. Low atomic weight solvents such as water, nitric acid, hydrofluoric acid, or hydrocarbons are to be preferred. In Table I are tabulated maximum allowable concentrations of various impurity elements in order that 1 mg. of uranium per ml. may be determined in 2N nitric acid with a minimum count rate of 200 C.P.S. Defining the limit of detection as three times the standard deviation of determining a blank (30 equals 0.09 mg. of uranium per nil.), detectable concentrations of uranium as an alloying element in the metals studied are calculated, Table I. Due to the variation of n in Equation 2 for different elements, kl and kz of Equation 4 are not expected to be the same for all impurity elements. This was experimentally shown by preparing solutions 2N in nitric acid and containing various elements in concentrations to give approximate ( p / p ) c values of 1000. Values for kl/k2 for each element were calculated and are sho c n in Table 11. Samples containing 0 and 20.00 mg. of uranium per ml. and various amounts of thorium, lead, zinc, iron, or chloride were prepared and analyzed by the recommended procedure. Uranium found was calculated using values for kl and kz as determined individually in the presence of each of the impurity elements, and using average values for the constants. As shown in Table 111, good recoveries were obtained using the average kl and kt values for impurity concentrations having ( p / p ) c values
less than approximately 500, but recovery of uranium in the presence of higher concentrations of impurities was erratic. When the kl and k.1 values aa determined in the presence of the individual impurity elements were used, uranium recoveries were within the precision of the procedure over the completc range of impurity concentrations studied. Each value shown is the average of duplicate analyses on a single sample.
RELIABILITY
The precision of the absorption edge method for determining uranium was calculated for known solutions containing various amounts of uranium dissolved in nitric acid, and the relative standard deviation, based upon 14 determinations at each concentration, is shown in Table IV for the cells of 1and 3-cm. path length. Forty-eight uranium carbide samples, previously analyzed gravimetrically by ignition to uranium oxide, were dissolved in nitric acid to give uranium concentrations between 8 and 18 mg. per ml. No attempt was made to remove the dispersed free carbon from the solution, Assuming gravimetric values to be correct, the solutions were analyzed by the x-ray absorption edge procedure with an average recovery of 100.11% and a relative standard deviation of 0.46%. In addition, six uranium-%% zirconium alloys, which had been dissolved in sulfuric acid, reduced in a zinc column and analyzed for uranium by cerate titration, were analyzed by the x-ray absorption edge procedure. Each solution waa diluted to 50 ml. to give uranium concentrations of approximately 6.5 mg. per ml. At this concentration the relative standard deviation of the procedure is approximately 0.8%, and the resulta, compared in Table V to values obtained volumetrically, agree within the error of the procedure. Each value shown is the result of a single determination. Defining sensitivity as that amount of uranium equivalent to three times the standard deviation of determining a blank, approximately 0.06 mg. of uranium per ml. can be measured using a cell of 3-cm. path length. The procedure is especially suitable for analysis of samples which normally require time-consuming separations or other chemical manipulations, and where precision of the order of 0.5 to 1% is satisfactory. Any laboratory equipped for x-ray fluorescence analysis can be set up for absorption edge analysis with no additional capital outlay, and without removing the instrument from normal x-ray spectrographic service.
Table 111. Comparison of Average and Individual Values for kl and kl for Determination of 0 and 20 Mg. of Uranium per Milliliter in Presence of Various Foreign Elements
0 Mg. U/Ml. Added
Element Th Th Th Pb Pb Pb Pb Zn Zn Zn Zn Fe Fe Fe
Fe
c1 e1 c1
Concentration Mg./Mf. 24 12 6 20 15 10 4
40 20 10 5 40 20 10 5 130 86 43
(Ir/p)c
2180 1090 550 2720 2040 1360 540 2320 1160 580 290 1540 770 385 190 1500 1000 500
U Found, Mg./Ml. Using Using average individual kl and k, kl and kl 0.47 0.20 0.05 0.09 0.16 0.09 0.06 -0.19 -0.07 -0.06 -0.03 -0.46 -0.18 -0.02
0.10 0.02 -0.04 -0.04 0.00 -0.03 0.02 -0.16 0.00 -0.02 -0.01 0.00 0.00 0.12 0.08 -0.07 0.00 0.10
0.03 -0.32 -0.27 -0.15
Std. dev.
0.22
20.00 Mg. U/Ml. Added U Found, Mg./Ml.
Using average kl and kg
Using individual kl and kr
20.38 20.14 20.12 20.39 20.22 20.15 19.91 20.08 20 .OO 20.00 20.00 19.69 19.69 19.99 20.00 19.82 19.92 19.94
19.95 19.97 20.07 20.10 20.01 20.04 19.89 20.11 20.00 20.00 19.99 19.99 19.89 20.09 20.04 19.99 20.03 19.97
0.20
0.07
0.06
ACKNOWLEDGMENT
The author acknowledges the helpful suggestions of G. R. Waterbury during the investigation, and in manuscript review. Thanks are also expressed to G. C. Heasely and R. L. Carpenter for performing the gravimetric and volumetric uranium analyses, respectively, and to J. M. Dickinson for preparation of the molybdenum-niobium alloy. The work was performed under the supervision of C. F. Metz.
Table IV. Precision of X-Ray Absorption Edge Determination of Uranium in Nitric Acid Solutions
U Concentration Mg./Mi. 0.60
Relative Standard Deviation, 70 1-cm. cello 3-cm. cellb ... 6.0
1.00 5.0 5.00 0.96 10.00 0.65 20.00 0.57 40.00 0.34 6 1,024,000counts. b 204,800 counts.
1.9
... ... ... ...
LITERATURE CITED
(1) Barieau, R. E., ANAL. CHEM. 29,
348 (1957). (2) Compton, A. H., Alliaon, S. K., “X-Ra a in Theory and Experiment,” 2nd eJ., pp. 800-6, Van Nostrand, New York, 1935. (3) Dietrich, W. C., Barringer, R. E., U. S. Atomic Energy Comm. Rept. Y-1153 (1957). (4) Dodd’ C. G., Proc. Eighth Annual Conf. Applications of X-Ray Analysis, 11, Denver Research Institute, benver, Colo., 1959. (5) Engstrom, A., Acta Radiol. Suppl. 6 3 , (1946). ( 6 ) Engstrom, A., Rev. Sci. Instr. 18, 681 (1947). (7) Friedlander, G., Kennedy, J. W.,
Table
V.
Sample No.
Determination of Uranium in Complex Mixtures
Uranium Found, Mg. R+ Volucovery, X-ray % metric
1 336.5 2 321.4 3 344.1 4 330.7 5 329.7 6 345.8 a 204,800 counts. b 1,024,000 counts.
343a 320.5” 342b 32gh 330s 347.5b
101.9 99.7 99.4 99.5 100. 1 100.5
“Nuclear and Radiochemistry,” pp.
265-6, Wiley, New York, 1949. (8) Glocker, R., Frohnmeyer, W., Ann. Physik 76,369 (1925). (9) Hughes, H. K., Hochgesang, F. P., ANAL.CHEM.22,1248 (1950). (10) Peed W. F. Dunn, H. W., U. S. Afomic hnergy Cbmm. Rept. ORNG1265 f 1962). \----
(11) Stiwart J. H., Jr., ANAL.CIIEM.32, 1090 (1960).
(12) Victoreen, J. A., J . Appl. Phys. 14, 95 ( 1943). (13) Wright, W. B., Barringer, R. E., U. S. Alomic Energy Comm. Rept. Y-1095 (1955).
RECEIVEDfor review January 13, 1961. Accepted April 3, 1961. Work done under the auspices of the U. 5. Atomic Energy Commission. VOL. 33, NO. 8, JULY 1961
1015
