Utilizing Recoil Protons SIR: Aumann and Born ( 1 ) have reported the use of protons recoiling from collisions with fission bred neutrons as a method for the determination of 0 1 8 isotope abundances in water samples. The OI8 in the wnt'er undergoes the reaction OL8(p,n)F1*,and the radiations from the decay of 110-minute F18 are measured. Their procedure uses comparator samples, and requires normalization of t'he deuterium content by exchange with gaseous ammonia. R e have been engaged in similar work with t'he Penn State reactor, a 200-kw. swimming pool type, undertaken to determine the degree of precision to which 0l8abundances can be measured by the recoil proton activation in samples recovered from chemical exchange kinetics studies a t relatively low enrichment, and the minimum size of sample required. TT7e have experienced large differences in the activation of paired samples; assuming identical recoil proton fluses in sample and comparator is a source of major and unacceptable error. The recoil proton flux induced in a water sample is a complex function of the fast neutron energy spectrum, the energy-dependent neutron-proton scattering cross section, the energy distribut'ion of the scattered protons from a given neutron energy, the stopping power oE the medium for protons, and the protium content of the sample. Variations in intensity and energy dist'ribution of the recoil proton flux from sample to sample can arise from a variety of factors, including neutron flux depression and tilt from other samples and experiments around the reactor core, flux shadowing, inability to reproduce sample irradiation geometry, and variations in sample size and shape. To avoid the errors caused by variations in the recoil proton flux, a method for monitoring the proton flux within the sample was developed and is described. X suitable proton flus monitor should be readily soluble in the water sample; should undergo a proton-induced nuclear reaction whose threshold and escitation function is similar to that of the 0 l 8 ( p , n)F'8 reaction; and should give a radioactive product of suitable half life and measurable radiation readily separable chemically, by decay analysis, or by spectrometry from the F'* activity. I n addition, the flux monitor should not give rise to extraneous neutron-induced activity in significant yield and should be available
in high purity. Ni(X0&.6H20 which undergoes the W4((p, n)Cu64 reaction satisfies these criteria and was selected as the proton flux monitor. Inclusion of nickel in the water does lead to some interfering nickel and cobalt activities, b u t the F I E and Cue4 activities are easily separated chemically by precipitation. The Ni(N03)2.6Hz0 also adds some 0l8which will result in additional F18 activity. Because the weight of the nickel nitrate added is known and the naturally occurring atom per cent of 0 1 8 may be safely assumed for this salt, the F18 activity caused by this extra 0l8may be subtracted from the total. EXPERIMENTAL
Sample Preparation. Distilled natural water and oxygen-18-enriched water supplied by Volk Radiochemical Co. were used. The enriched water contained 1.63 atom per cent 0 1 8 and 0.139 atom per cent 017,as reported by the supplier, and a n unknown but significant amount of deuterium. Dilutions of this enrichment were obtained by combining known amounts of enriched and natural waters. Reagent grade S i ( N 0 & . 6 H 2 0 , and the water samples, were weighed into screw-top polyethylene containers to *0.05 mg. The containers varied in capacity from 5 to 65 ml. and in diameter from 1.5 to 4.0 em. Sample Irradiation. All samples were irradiated for 2 hours inside a 0.040-inch thick tubular cadmium shield (to reduce thermal neutron activations) at t h e face of T h e Pennsylvania S t a t e University reactor core operating at 200 kw. iit this power level the fast neutron flux transversing t h e cadmium-lined cavity is approximately IO'? neutrons/sq. cm.see. The thermal neutron flux, from leakage through the cadmium liner and down the tube and from thermalization within the sample, was measured by sodium activation and is about 1010 neutrons/sq. cm.-see. Radiochemical Separation. T h e irradiated sample was opened and carrier solutions containing approximately 20 mg. of CuSOa, 10 mg. of K a F , a n d 2 mg. of CoClz (holdback carrier) were added t o t h e irradiation container. T h e same amounts of carrier from prepared stock solutions mere used for all samples, for convenience in the calculations. T h e t o p of t h e irradiation container was replaced, and the container was shaken well to ensure thorough mixing of the sample and carriers. The sample and two rinsings
of the irradiation container mere transferred to a beaker and heated to about 50" C. The copper was separated as copper(I1) salicylaldoxime by precipitation at a pH of 3 (3) following two precipitations as CuSCN in a procedure adapted from one given by Bomen ( 5 ) . The F18 was separated as triphenyltin fluoride in a procedure adapted from one recommended by Ballczo and Schiffner ( 2 ) . The supernate from the first CuSCN precipitation containing the fluorine, nickel, and cobalt activities was pulled through a fine porosity Buchner filter funnel, adjusted to a p H of 4.j1 and added to a beaker which contained a 2.5-fold excess of (C6Hb)3SnC1dissolved in 50 ml. of chloroform. The mixture was stirred gently for about 30 minutes with a magnetic stirrer. The formation of the solid (C6Hs)&F takes place a t the interface, and the precipitate suspends in the chloroform layer. The contact between the precipitate and the nickel and cobalt activities is thus minimized, reducing the chance of occlusion and coprecipitation. To avoid pouring the nickel and cobalt activities through the filter upon which the (C6H5)3SnF precipitate was to be mounted, the aqueous layer was pipeted off prior to filtration. [This separation of fluoride as (C6H5)3SnFfails in the presence of lithium.] The precipitates were mounted by suction filtration through 2.9-em. paper circles (S and S Nr. 589, Blue Ribbon), previously treated with wash solutions and weighed, held in position on a fritted glass funnel by a 2.54-em. removable polyethylene filter chimney. The (C6H5)&hF was washed 15-ith 25 ml. of water, 10 ml. of chloroform, and 10 ml. of acetone. The Cu(C-iH60nN),was washed with 25 ml. of water, 10 ml. of alcohol, and IO ml. of acetone. After final ryashings, 1 ml. of acetone cont.aining approximately 0.2 mg. of collodion was pulled slowly throuah the precipitates. Sample Counting. After being mounted on aluminum sample mounts, t,he activity was measured on a n end-window flow counter, operated in t h e G. hf. region, for about three half-lives, and decay plots were extrapolated back to the time of removal from t h e reactor. T h e radiochemical purity was checked b y half life determination and b y gamma spectrum analysis. Radiochemical purities in excess of 99% were consistently achieved for both the F'* and Cue4 activities. Radiochemical yields varied from 70 to 99% for (C6H&SnF and from 66 to 82% for the Cu(C7HaO&)2. Careful control of the (C6Hj)3SnF filtration is necessary to ensure a VOL. 37,NO. 10, SEPTEMBER 1965
1269
consistent and smooth deposition of the precipitate. The copper precipitate is exceedingly easy to handle. RESULTS A N D DISCUSSION
The measure of particular interest is the mole fraction of 0 1 8 (moles of 0 I 8 / mole of oxygen) in the water sample. The derivation of an expression for the mole fraction in terms of the experimental observables involves considerable algebraic manipulation and is only outlined here. The measured activity of F 1 8 , when corrected and normalized for growth during irradiation, chemical yield, variations in the proton flux through the Cue4 activity and its decay and chemical yield, and for the activity arising from 0 1 8 introduced by the flux monitor, is directly proportional to the number of moles of 0 1 8 in the water sample. The true disintegration rates are related to observed activities through the radiochemical yields and a number of factors making up the counting efficiencies. The two activation equations for the two radioactive isotopes produced may be combined to eliminate the flux, and the true distintegration rates expressed in terms of observed counting rates. Involved in this combined equation are a large number of factors, some of which are physical constants, such as the reaction cross sections and decay constants; some are parameters set by the choice of measuring instruments, such as counting efficiencies; and some are arbitrary choices of convenience, such as the amounts of carriers used. For a given time of irradiation and activity measurement procedure, the constant factors may be collected into a single term, and the relation between the measured activities and the amount of 0 1 8 in the sample may be expressed as
where A F and A C u are the observed activities of F18 and C d 4 in counts/ minute at the end of the irradiation. N 1 is the total number of 0l8atoms in the water sample; N z is the total number of 0 1 8 atoms in the Ni(?rT03)2.6Hz0 added; N a is the total number of Nis4 atoms in SF and the N i ( N 0 ~ ) ~ . 6 H z added. 0 S C u are the F 1 8 and Cu'34 self-absorption correction factors. IVF and WCu are the weights of the F1*and Cua4containing precipitates; K is the lumped collection of constants referred to above. Nl is a function of the average molecular weight of the water sample, H20. This is given by,
H T
=
2x
+ f1eM'6+ f17M17
where 1270
+
p M ' 8
(2)
is the average atomic weight ANALYTICAL CHEMISTRY
Figure 1 . Regression line for natural water samples
12
of hydrogen in the water sample; and f18 are the mole fractions of Ole, 017,and 0 1 8 in the sample; and M16, W7, and J P are the atomic weights of 0 I 6 , O I 7 , and 0'8. It may be assumed within the accuracy of the measurement that for natural water and water of moderate 0'8 enrichments, f17 is approximately zero. By substituting Equation 2, with f17 = 0, into Equation 1 and, for convenience, by setting S F T Y F / S C ~ T V C U= R, a n explicit expression for the mole fraction of 0 ' 8 in the water sample may be obtained :
sample measured \vas unity. I n unknown water determinations, the use of a self-absorption correction factor corrects the observed activity only to the activity of the lightest self-absorption sample. The correction from this point to zero sample weight is a constant which, while unknown, may be included in K . Thus, K is redefined to include the ratio of the constant parts of the self-absorption correction factors of FI8 and CuC4. The value of K , a collection of both known and unknown constants, may be determined from Equation 1 by measurements on a known sample or water. The mean value of K for a set of 12 natural water samples (natural abundance of 0 ' 8 and H2 assumed) was determined to be 0.03006 + 0.00016. The internal consistency of the data and the assumption that K is a constant can be checked by plotting the data from the 12 known natural water samples according to Equation 1. Thus a plot of
f16, f17,
where fl is the nautrally occurring mole fraction of 0 I 8 in the Ni salt; fi is the naturally occurring mole fraction of Ni54; n is the number of moles of I\Ji(NOs)z.6Hz0 added to the water sample; and W 1 is the weight of the water sample in grams. By substituting numerical values, using the C12 scale, and assuming a normal deuterium abundance, justified in the following discussion, the expression reduces to :
should be a straight line passing through the origin and having a slope of K . This plot is shown in Figure 1. An analysis of the data plotted in Figure 1 using a least-squares line fitting technique (6) which assumes the possibility of error in both variables as well as a non-zero intercept gave a value of 0.03002 for K and a system blank-Le.,
-4 the intercept on the A F axis-of c.p.m., which is negligible. The volumes of water used were varied from 0.5 ml. to 60 ml. for the natural water samples. Volumes of 2 ml. seem adequate and convenient to work with and were used for the enriched water samples. Three sizes of irradiation containers were used, and the ratio of the weight of water to the weight of Ki(?;03)2.6H20 x-as varied from 10 to 70. Linear regression was
A F - 2.266 R K Acu = 1495 W1 R K A C u - 0.1113 (AF - 2.266 R K ACU) The self-absorption factors are obtained from an experimentally determined plot of self-absorption factors as a function of sample weight. Because it is not feasible to precipitate samples of zero or near zero weight and because the unknown shape of the self-absorption curve in the vicinity of zero sample weight makes extrapolation inaccurate, the data was normalized so that the self-absorption factor for the lightest
(4)
significant at the 99% confidence level, indicating that K is indeed a constant, independent of the size and shape of the water sample, and of the ratio of sample to flux monitor. Because the least-squares analysis involved an estimation of the ratio of the true variances along the two coordinate directions and the assumption that this ratio is constant, the mean value of K , which includes the mean correction for
1.6
1 -
RECOIL
1.5
PROTON
1.4-
FLUX is
1.3-
arbitrary
units
1.2
-
,,I
.
1.0
Table 1.
Analysis of Enriched Water Samples Known f l 8 Measured f18 Error, yo 0.0163 0.0161 -1.2 0.0123 n .n12n - 2- . 40 0090 0.0092 +2.2 0.00695 0.00679 -2.3 0.00326 0.00344 +5.5 0.00251 -2.3 0.00257
2-
-
1
1
2
3
1
4
,
5
6
1
7
1
B
N A T U R A L WATER
Figure 2.
,
1
SAMPLE
0
1
1
1
I
1
I
Z
NUMBER
Variations in recoil proton flux
any system blank, was chosen as the best value of K . The choice makes essentially no difference in the accuracy of the measurement. However, K , which is similar t o an instrumental constant, is dependent on the particular set of analysis conditions used and is valid only when those conditions are reproduced. The results of the measurement of the mole fraction of 0 ' 8 in six enriched water samples are given in Table I. I n these calculations, the naturally occurring average atomic weight of hydrogen was assumed, although a moderate enrichment was known to be present. Vncertainty in the deuterium abundance in a n 0'8-enriched water sample is a possible source of two kinds of error. First, the expression for 15.9949), the includes the term (2H average molecular weight of H:OL6. The error introduced by the assumption that the deuterium concentration is the natural abundance is less than the intrinsic precision of the determination for any but water samples of high deuterium content. Error introduced b y this uncertainty is calculated to be approximately 1% of the 0l8abundance measurement for each 10 mole yGH2 in the sample. A second kind of error arising from an uncertainty in the deuterium content may be introduced by recoil deuteron initiated nuclear reactions which produce the same activities, FI8 and C U ~ ~ as the primary recoil proton reactions. Possible reactions are 0l8(d,2n)F18, NP4(d, 2 4 ~ 4 017(d, , T L ) F ~OW, ~, y)~18, and Ni62(d, ~ ) C U ~The ~ . ( d , 272) reactions have reaction Q values 2.2 m.e.v. more negative than the ( p , n) reactions and are not favored. The (d, y ) and ( d , n) reactions have positive Q values but have cross sections several orders of magnitude lower than the cross sections for the ( p , n) reactions. I n addition, the recoil deuteron spectrum is less in magnitude and decreases more rapidly with energy than the recoil proton spectrum because of the smaller scattering cross section and the lower efficiency
+
1
9
of energy transfer to the heavier nucleus in the scattering process. One would not, therefore, expect any appreciable contributions from recoil deuteron induced reactions. *in experiment to determine whether deuteron reactions did make an observable contribution was carried out with two water samples containing 15.44 and 28.27yG deuterium and normal oxygen isotope abundances. Values for K of 0.0292 and 0.0316 were obtained, which are in general agreement with the value obtained from natural water samples (0.015% deuterium), indicating that significant enrichments do not contribute noticeably. For the evaluation of the average molecular weight of the water sample, i t has been assumed that the abundance of 0 1 7 is low enough to disregard (or more accurately, that all OI7 atoms may be taken b y number and weight as 0l6 atoms), even in an 018-enriched water sample. However, OI7 could also give rise to F18 activity through the ( p , y) reaction, which has a positive Q value and, therefore, will have a finite cross section for reaction over the whole proton energy spectrum. A value of K determined solely from natural water samples allows the accurate determination of the 0 1 8 abundance in 0I8-enriched water samples without calibration of the procedure using known heavy oxygen water; thus 017( p , y ) reaction is not contributing significantly to the ,yield of FIE. I n addition, Blanchard (4) has shown theoretically that the 017 ( p , y) reaction would not be expected to contribute significantly. No interference from the F19(n, 2n)F18 reaction on fluoride impurities in the water or nickel nitrate was found, principally because there are relatively few reactor neutrons above the 10-m.e.v. threshold for this reaction. A small interference was detected from the Cu63(n,y)C~6~ reaction on copper impurities in the nickel nitrate. Irradiation of samples of the nickel nitrate in a thermal neutron flux approximately the same as that inside the cadmium shield showed that the CuO4 activity resulting
from thermal neutron activations was activity. about 1% of the total Because the thermal flux bears some relationship (but not a direct onej to the recoil proton flux, this source of error is probably somewhat less than 1%. No interference was found from the ZnG4(n, p)Cu64 reaction on zinc impurities. Although the consistency of the results indicates that the recoiling proton flux is adequately measured by the Ni64 flux monitor, significant variations in the proton flux from sample to sample were detected. A measure of the recoiling flux is given by the expression
A plot of these expressions for the 12 natural water samples is given in Figure 2. Variations up to 40% are apparent. These variations are, however, of little consequence in the determinations because the presence of the flux monitor allows the F18activity to be normalized to the recoiling proton flux. T h a t the flux monitor adds some oxygen of normal isotopic abundance to the sample being measured adds some degree of complexity to the functional relationship between the experimental observables and the OI8 abundance of the sample, but we were unable to find a n oxygen-free substance with the properties required of a proton flux monitor. The 0 1 8 contributed by the monitor in small samples of low enrichment is just outside the limits within which it could be ignored. For a 1-gram sample of water a t twice normal 0 1 8 abundance, to neglect the 0 ' 8 contributed by 0.1 gram of monitor salt would lead to an error of about 4% in the measurement. The procedure proposed here, then, allows the determination of the 0 ' 8 abundance in 1-gram water samples to about 3y0, and in terms of the 0l8 excess abundance, the parameter that is measured in 0 ' 8 tracer studies allows the measurement of 0.1 mole % 0l8 excess above normal to about 10%. The method is also applicable to other compounds of hydrogen and oxygen that do not contain other readily activated elements. Compounds of hydrogen, oxygen, carbon, or nitrogen VOL. 37, NO. 10, SEPTEMBER 1965
1271
would be suitable. Because the slowing down medium for the proton will not be the same, K will change in a way related to the change in electron density. Using the identical irradiation, separation, and counting system described, we have evaluated K for small samples of ethanol of normal 0 ’ 8 abundance and find a value of K about 0.0263, which is somewhat lower than in the aqueous determination. This value is in agreement with the first approximation one might make, that K will be inversely proportional t o the electron density of the medium.
ACKNOWLEDGMENT
Cooperation of the staff of the Xuclear Reactor Facility of The Pennsylvania State University and financial support by the Department of the Army of the graduate study under which this work was carried out are gratefully acknowledged. LITERATURE CITED
(1) Aumann, D. C., Born, H. J., ~Vaturwiss. 51, 159 (1964).
(2) Ballczo, H., Schiffner, H., Z. And. Chem. 152, 3 (1956). (3) Biefeld, L. P., Howe, D. E., IND. EKG.CHEM.,ANAL.ED.11, 251 (1939).
(4) Blanchard, C. H., Westinghouse Atomic Power Division, Rept. WAPDAlW(P)-51, December 1955. (5) Bowen, H. J. &I., Intern. J . A p p l . Radiation Isotopes 4: 214 (1959). (6) Villars, D. S., Statistical Design and Analysis of Experiments for Development Research,” p. 170-72, Wm. Brown Co., Dubuque, 1951. LARRYH. HUNT’ WARREN W. MILLER Department of Chemistry The Pennsylvania State University University Park, Pa. Present address: Department of Chemistry, United States Military Academy, West Point, N. Y.
Cation Exchange Separation of Molybdenum, Tungsten, Niobium, and Tantalum from Other Metal Ions SIR: Hydrogen peroxide forms stable complexes in acidic aqueous solution with only a few metal ions. Fritz and Abbink ( 7 ) used a dilute solution of hydrogen peroxide to elute vanadium from a cation exchange column and thus to separate it from a number of other metal ions. Strelow (22) used hydrogen peroxide and sulfuric acid to separate titanium from more than 20 cations by cation exchange. Various authors have used solutions of hydrogen peroxide to elute molybdenum(VI), tungsten(VI), niobium(V), and tantalum(V) from cation exchange columns (1, 2, 8, 10, I I ) , but a very limited number of separations have been reported and the conditions for elution have varied considerably. Strelow (12) indicated a successful elution of molybdenum(V1) and niobium(V) from a cation exchange column with acidic hydrogen peroxide but stated that tungsten(V1) and tantalum (V) showed a tendency to hydrolyze. The purpose of the present work was to study the cation exchange separation of molybdenum(VI), tungsten(VI), niobium(V), and tantalum(V) as a group from other metal ions. EXPERIMENTAL
Apparatus. Conventional 1.2-cm. i.d. ion exchange columns with coarse glass frits were used. A slurry of resin and eluting solution was added t o t h e column until t h e bed had a height of 12 cm. Sample solutions and eluting solution were added dropwise from a separatory funnel inserted in the top of t h e column through a one-holed rubber stopper. Resin. Dowex 5OW-X8 cation exchange resin, 100 to 200 mesh, was used in the hydrogen Form. Reagents. Except for t h e metal ion solutions listed, all metal ion stock solutions were 0.05M solutions 1272
ANALYTICAL CHEMISTRY
of the nitrate or perchlorate salt in dilute nitric or perchloric acid. Chromium(II1) was a 0.05-11 solution of chromium chloride in dilute nitric acid. Zirconium(1V) was a 0.05V solution of zirconyl chloride in 0.3J1 hydrochloric acid. The zirconium(1V) salt was dissolved in concentrated hydrochloric acid and diluted to volume. Titanium(1V) was a 0.05J1 solution of titanium tetrachloride in 0.2.11 sulfuric acid and 0.3% hydrogen peroxide. Tin(1V) was a 0.0511 solution of tin tetrachloride in 0.3X hydrochloric acid. Molybdenum(V1) was a 0.0551 solution of molybdic acid which was made slightly basic (pH 8.3) with sodium hydroxide. Tungsten(V1) was a 0.05M solution of potassium tungstate in distilled water (pH 8.5). The potassium tungstate was prepared and purified in Ames Laboratory. I t was analyzed for tungsten by hydrogen reduction and was found to be pure. The niobium(V) and tantalum(V) 0.0556 stock solutions were prepared as follows: weighed amounts of the high purity metal (99.97%) were dissolved in hydrofluoric and nitric acids in plastic beakers provided with plastic covers. Dissolution was complete in about 5 minutes a t room temperature. The resulting solution was evaporated to about 5-10 ml. in the plastic beaker and next was transferred to a platinum evaporating dish. After addition of 10 ml. of concentrated sulfuric acid, the solution was further evaporated to fumes of sulfur trioxide to remove traces of fluoride. The vessel was cooled and the solution was diluted with approximately equal quantities of concentrated sulfuric acid and 30% hydrogen peroxide. The solution was transferred to a 500-ml. flask, a total of 30-55 ml. of concentrated sulfuric acid and 5C-65 ml. of 30y0 hydrogen peroxide were added, and the solution was diluted to volume with distilled water. I n all instances these stock
solutions mere stable for at least a month. I n the fuming step it was found that if heating was continued much beyond the first appearance of sulfur trioxide fumes, i t was difficult to dissolve the Xb2O5and Ta2O6which formed. The eluting solution consisted of 0.25M sulfuric, perchloric, or nitric acid containing 1% hydrogen peroxide. Separation Procedure. Synthetic sample mixtures for separation were prepared b y mixing known quantities of molybdenum (-0.25 mmole), tungsten, niobium, or tantalum (-0.5 mmole for each of the last three) with approximately a n equal quantity of a second metal ion. T h e mixture of niobium or tantalum and other metal ion was already 0.5-1-11 in sulfuric acid and 1.5-2% in hydrogen peroxide (from stock solutions), so t h a t further addition of sulfuric acid and hydrogen peroxide was unnecessary. The sample was added to the ion exchange column, and molybdenum was eluted with 125 ml. of 0.2531 sulfuric or perchloric acid containing 1% hydrogen peroxide. About 80 ml. of a 0.2531 nitric acid, 1% hydrogen peroxide solution was used to elute tungsten, niobium, and tantalum. The use of nitric acid for the last three metals is desirable because recoveries in the gravimetric analytical method are reported to be slightly low (at least for tungsten), when much sulfuric acid is preaent in the precipitation medium ( 5 ) . I n each case, after elution the column was washed with about 25 ml. of distilled water. Then the other metal ion was stripped from the resin with the amount of eluting solution indicated in Table I. The maximum flow rate obtainable was used in all of the elutions. Analysis of Column EfRuents. T h e column effluents containing t h e other metal ion were evaporated almost t o dryness and then diluted t o 100 ml. with distilled water for titration.
