values ascertained are in good agreement with NAA and results reported in the literature. Typically, accuracies of 5% or better are obtained. Figures 7 and 8 show the spectra obtained on Granite GA. Spectral background for the elements with K x rays in the range of 31 to 32 keV (Cs, Ba) contain a fraction of coherent scattered excitation radiation, which causes larger errors in the determination of these elements. These errors could essentially be eliminated and much higher x-ray linehpectral background intensity ratios realized for these elements by operating the x-ray tube at approximately 60 kV and inserting a thicker external Mo filter. Doing so would shift the scattered radiation to higher energies. However, our excitation system is limited to a maximum voltage of 50 kV. Using 100-mg specimens pressed into 0.32-cm i.d. cylinders, we have calibrated the system and determined the concentrations of elements with characteristic x rays of energies less than 16 keV. T o do so required highly collimated excitation and detection radiation beams. Sensitivities realized were approximately a factor of 2 less than those attained with 2-g specimens. Although the results obtained were dependent upon the density of the prepared specimens, typically accuracies of 10% or better were achieved.
LITERATURE CITED H. A. Liebhfsky, H. G. Pfeiffer, E. H. Winslow, and P. D. Zemany, "X-Rays, Electrons, and Analytical Chemistry. Spectrochemical Analysis with XRays", Wiley-Interscience New York, 1972: L. S. Birks, "X-Ray Spectrochemical Analysis", Wiley-lnterscience, New York, 1969. R. C. Reynolds, Jr., Am. Mineral., 48, 1133 (1963). R. C. Reynolds, Jr., Am. Mineral, 52, 1493 (1967). C. E. Feather and J. P. Willis, X-Ray Spectrom., 4, 41 (1976). G. Andermann and J. W. Kemp, Anal. Chem., 30, 1306 (1958). R. Jenkins, "An Introduction to X-Ray Spectrometry", Heyden, London, 1974. W. ti. McMaster, N. K. Del Grande, J. H. Mallett, and J. H. Hubbeli, "Compilation of X-Ray Cross Sections", University of California, Lawrence Livermore Lab. Report UCRL-50174, Section Ii, Revision I(1969). F. J. Flanagan, Geochim. Cosmochim. Acta, 37, 1189 (1973). I. Perlman and F. Asaro, in "Science and Archaeology", R. H. Brill, Ed., MIT Press, Cambridge, Mass., 1971. R. D. Giauque, R. B. Garrett, and L. Y. Goda, "Calibration of Energy Dispersive X-Ray Spectrometers for Analysis of Thin Environmental Samples", Lawrence Berkeley Lab. Report LBL-4481 (1976). Etude Cooperative, Analusis, 2, 59 (1973). W. Bambynek et al., Rev. Mod. Phys., 44, 716 (1972). J. S. Hansen, H. U. Freund, and R. W. Fink, Nucl. Phys. A,, 142, 604 (1970). M. Roubault, H. de La Roche, and K. Govindaraju, Sci. Terre, 15, (4), 351 (1970). H. de La Roche and K. Govindaraju, Bull. SOC. Fr. Ceram., 85, 35 (1969).
ACKNOWLEDGMENT
RECEIVEDfor review July 28, 1976. Accepted October 11, 1976. This work was supported by the U.S. Energy Research
The authors thank Frank Asaro, Helen Michel, and Harry Bowman for providing the French geochemical reference standards and the neutron activation analysis data. We are grateful t o them and to Joe Jaklevic and Carl Blumstein for their comments on the preparation of this paper.
and Development Administration. Any conclusions or opinions expressed in this report represent solely those of the author(s) and not necessarily those of the Lawrence Berkeley Laboratory nor of the U.S. Energy Research and Development Administration.
Determination of Uranium in Natural Water by Preconcentration on Anion-Exchange Resin and Delayed-Neutron Counting R. J. N. Brits" and M. C. 6. Smit Chemistry Division, Atomic Energy Board, Private Bag X256, Pretoria, South Africa
A method is proposed for the determination of uranlum in water. The uranium is extracted by sorption of the uranyl thiocyanlde complex on an anion-exchange resin. The preconcentration step is largely independent of water quallty and insensitive to experimental parameters. Uranium on the resin is directly determined by delayed-neutron counting, which Is a specific and sensitive analytical method for uranium. The accuracy is 4% and the detection limit for a 1-1. sample is 0.04
w1.
Uranium is known to be leached out of rock and to be present in water as uranyl carbonate complexes, e.g., UOZ(CO&2H202-. The uranium content of natural water can therefore be used to delineate target areas for more intensive exploration. The origin of the water, the rainfall, and geological factors may all influence the uranium content of such water, and consequently its concentration varies in the range of 0.1-10 pg/l. Since an anomalous uranium concentration may be only slightly above the background level, refined analytical methods must be employed to detect small concentration variations. Since the use of hydrogeochemical exploration techniques implies the analysis of large numbers of samples, the analytical method must be both rapid and inexpensive.
Most of the methods proposed for the determination of natural levels of uranium in water entail a concentration step. Both coprecipitation with ferric hydroxide ( I ) and adsorption on charcoal ( 2 ) suffer from disadvantages ( 3 ) .Evaporation by freeze drying ( 4 ) and sorption on an ion-exchange resin (5-9) have also been reported. Subsequent uranium determination is usually done fluorometrically or spectrophotometrically. A rapid nuclear method (9, IO), which does not require preconcentration is based on the measurement of the 239Uisotope produced by the 238U(n,y)239Ureaction during a short reactor irradiation. The extremely sensitive method of counting the fission tracks ( 4 , l l )from the 235U(n,f)reaction has also been used. Analysis of water using delayed-neutron counting has been reported ( I ); however, preconcentration was achieved through coprecipitation. The extraction of the uranyl thiocyanide complex on a strong basic anion-exchange resin ( k d = 65 000) has been described by Korkish (5-7). Preconcentration by this ion-exchange procedure followed by direct determination of the uranium on the resin by delayed-neutron counting, was thus investigated as a possible alternative procedure.
EXPERIMENTAL Ion Exchange. The extractions are performed with the apparatus shown in Figure 1. The glass (or quartz) tube with a supporting ANALYTICAL CHEMISTRY, VOL. 49, NO. 1, JANUARY 1977
67
1 t-21F1a“
T a b l e I. Analyses of Solutions Containing 2 pg U r a n i u m in 250 m l a n d Stored in Polythene f o r Different P e r i o d s of Time Time stored, days 1 2 4 8 12 19 30
&--
88
pg
U recovered
Acidified
Neutral
2.07 2.03 1.99 2.04 1.99 1.98 1.94 2.05
1.99 2.00 2.06 2.07 1.97 2.02 2.03 2.06
k r e w thread joint and latex seal
Quartz tube
Quartz w w l
Figure 1. Extraction apparatus
quartz-wool plug is loaded with about 400 mg AG 1-X8 resin (200-400 mesh; Bio-Rad Laboratories), which has been converted to the thiocyanide form prior to loading. The usual procedure is to add 2.5 ml of concentrated HCl, 5 g of KSCN and 1.2 g of ascorbic acid to a 250-ml filtered water sample. This solution is shaken and allowed to stand for a few hours before it is passed through the column. After the column has run dry, the tube is disconnected from the apparatus and the remaining liquid sucked out of the resin. The resin is then quantitatively transferred to a polythene irradiation vial (i.d. = 8 mm) by pushing the resin and the plug out with a glass rod. The container with the plug andresin is then heat-sealed. If necessary, water samples can be filtered through a 0.8-gm membrane filter (Sartorious SM 12804) to prevent blocking of the resin column, and to eliminate the uranium contribution of particulate matter. Uranium Determination. The uranium content of the resin is determined with FRITS (Fast Reactor Irradiation and Transfer System), an automatic delayed-neutron counter developed a t the Atomic Energy Board (12). In the design, special attention was given to fail-safe safety precautions and back-up systems. The sample, in a polythene rabbit, is irradiated for 30 s in the core of SAFARI-1 (&h = 2 X 1019m-2 s-l), an ORR-type experimental reactor. After a delay time of 20 s, the rabbit is counted for 50 s is a counter with four BF3 tubes. The sensitivity (42400 countslpg U) is determined every 20 min to compensate for short-term flux variations. Since the background is about 400 counts, the a priori detection limit (131, LD is 0.04 pg. Throughput is about 30 sampledh. Blank Value. The observed uranium content of the unused resin, plug and container was less than 0.03 fig. Addition of the chemicals (KSCN and “&EN; analytical and technical grades) to 1-1. deionized water samples also gave a blank value less than 0.03 pg. For ease of handling, the quartz-wool plug can be replaced by polypropylene or Nylon wool, both of which are “uranium-free”. Glass wool is unsuitable. “Uranium-free” water for experimental purposes is prepared by extracting the uranyl thiocyanide complex with a large amount of resin. Influence of Some Experimental Parameters. Filtration prior to acidification is preferred since acidic solutions can leach uranium out of particulate matter. Deionized, neutral water samples (500 ml) to which 0.3 pug U was added, were filtered through cotton wool, glass-fiber filters, and the membrane filter. None of these filters shows any adsorption of uranium and any one of them can be used for filtering. The effect of resin mass on the uranium yield was investigated by passing 1-1. seawater samples through columns containing different (150-500 mg) amounts of resin. No significant differences could be observed. By changing the height of the flask above the extraction tube and adjusting the tap, the flow rate could be varied from 2-30 ml min-’ without any significant loss of uranium. The flow speed adopted, which sufficed for practical purposes, was about 8 ml min-I. The effect of water volume was studied by first extracting 0.3 pg U from a 500-ml solution onto the column, followed by 2 1. “uranium-free” water containing the usual concentration of chemicals. The experiment was repeated with a 10-1. sample. Subsequent U determination yielded 0.28 pg and 0.32 pg, respectively. Uranium de68
ANALYTICAL CHEMISTRY, VOL. 49, NO. 1, JANUARY 1977
termination in which 200-, 500- and 1000-ml seawater samples were used yielded 3.00, 3.04, and 2.87 ppb, respectively. Extraction of 500-ml water samples containing 3 pg U a t pH values 0, 1, and 2 yielded 2.99, 2.97, and 2.94 pg, respectively. Interferences. The uranyl carbonate complexes are very stable, and even minute amounts (3) of C0a2- can compete with SCN-. This might prevent the quantitative extraction of U from the solution. Although the relative amount of free C0z2- a t a pH value of 1 is extremely small, the possible effect of carbonate was studied by adding 0.4 g Na2C03 to 100-ml deionized water samples containing 5 fig U. After adjusting the pH to 1,one of the samples was shaken to free the solution of COS,while the other sample was extracted directly. Both extractions yielded a value of 4.95 gg U. It is recommended, however, that the solution be shaken and then be left standing for a few hours, since dissolved CO2 tends to form bubbles on the resin, thus blocking the column. Water samples from iron-rich areas or from boreholes with rusted pipes or lines, may contain relatively large amounts of iron. Since the ferric thiocyanide complex is co-extracted, it might saturate the resin and prevent quantitative sorption of the uranium complex. Korkish ( 5 )suggested reduction with ascorbic acid to decrease the extraction of ferric thiocyanide. Quantitative reduction of ferric ions can easily be established by observing the disappearance of the intense red color of the solution. In spite of this reduction, the resin usually becomes red during the extraction, possibly because of re-oxidation of the FeZf by the resin in the presence of thiocyanide. Seawater samples (1 1.) containing 1000 ppm Fe3+ and 2.9 fig U were extracted with and without the ascorbic acid. The resins contained 2.5 and 0.05 pg U, respectively. Although ascorbic acid and the sulfate ion can both complex the uranyl ion, no effect from either of these was expected or could be experimentally detected. Thorium constitutes a minor interference in the uranium determination with FRITS. In the unlikely event of T h being present in an appreciable concentration in natural water, the extraction of T h under our experimental conditions should be rather insignificant (14). A deionized water sample (11.) containing 1000 p of T h was extracted and yielded an apparent U content of 0.1 gg. If this result cannot be ascribed to U impurity in the Th, it is clear that interference from T h is negligible. Depletion of the 235Ucontent of U, e.g. in samples from Oklo, Gabon, would lead to a low result for U. Accordingly this method might cause one to miss a U deposit depleted in 235U. Stability of Solutions. A knowledge of the stability of the various solutions is useful when scheduling the analysis of large numbers of water samples on a routine basis. It is usually recommended that water samples be acidified and kept in polythene containers for prolonged storage. The acidification of unfiltered samples, on the other hand, should be avoided to prevent the leaching of U from particulate matter. Meeting both these requirements would mean the filtration of a sample on collection, which, for a large exploration effort, is impracticable. In order to establish some guidelines, tap-water samples (neutral and acidified with HCl), containing 2 fig U were stored in 250-ml polythene bottles and analyzed over a period of time. The results are summarized in Table I. Storage of acidified and neutral solutions containing 0.4 pg U/1. for up to 40 days also did not result in a detectable loss of U. A more detailed study of such dilute solutions, especially if the sample is not filtered in the field should yield interesting results. Filtered borehole water submitted for analysis showed no significant difference between neutral and acidified samples after storage of four months. These
~~
~
Table 11. Precision and Accuracy of Uranium Determinations Sample
Ten replicates, Fdl.
Intercept, wg11.
Sensitivity countslpg
Slope countslpg
Borehole HH Borehole HO Seawater
8.32 f 0.16 7.47 f 0.11 3.08 f 0.12
8.46 f 0.15 7.80 f 0.28 3.20 f 0.09
2317 2327 2321
2309 f 19 2298 f 38 2261 f 25
results do not support those of Robertson (15) or Reimer (16), but are in accordance with the observations of Hashimoto (17). Possible loss of U from filtered acidified thiocyanide solutions stored in laboratory glassware may occur as a result of adsorption on the glass, but for storage periods up to ten days no such effect was detected for solutions containing 10 pg U/1. The resin is pre-treated with a solution similar in composition to the extraction solution, and columns are left in this solution until required. This solution was found to be unstable: after ten days it became milky because of the formation of sulfur, and the smell of HgS also became noticeable. Although no change in the efficiency of the solution could be detected in experiments designed for the purpose, the sulfur precipitate might block the resin; prolonged storage of prepared solutions is therefore not recommended. Radiolytic Stability of the Resin. Radiolytic decomposition of the resin by the intense neutron and y-ray bombardment in the reactor core can cause an increase in pressure inside the container, and create a safety hazard for FRITS. This effect was studied by irradiating aliquots of the dry resin (ca. 500 mg) for periods up to 30 min in the pneumatic irradiation facility of SAFARI-1.The pressure built up in the airtight polythene container was measured with a sensitive water manometer. The very small increase in pressure that was detected for the longer irradiation periods appeared to be due to heating effects only, and proved that the irradiation of dried resin posed no explosion hazard.
RESULTS A N D DISCUSSION Two composite borehole samples, HH and HO, and fresh seawater were analyzed. Ten replicate analyses of each sample were performed and each sample was also analyzed by the standard-addition method, using U of natural isotopic composition. In the case of the borehole samples, the average counts of the replicate analyses were included in the regression line. Since the standard additions t o the seawater were performed 30 days after the replicate analyses, two seawater samples without added U were analyzed anew. In all cases 250-ml aliquots of the borehole samples and 500-ml aliquots of seawater were used. T h e results are summarized in Table I1 and Figure 2. Since the slopes of the regression lines for the boreholes samples agree with the sensitivity factor of FRITS, one concludes t h a t the method yields accurate results. T h e slightly smaller slope of the line for seawater may be due to experimental error. Alternatively, the high chloride concentration in seawater may cause U t o be eluted out of the column since the uranyl chloride complex is not extracted efficiently a t pH 1. T h e agreement between the replicate analyses and the intercept of the regression line for seawater also tends to confirm the observation t h a t U is not lost during storage. Analyses of a series of 3000 heterogeneous water samples indicated t h a t precision does not suffer in routine work.
CONCLUSION Pre-irradiation chemistry is usually not employed in nuclear methods of analysis, because of contamination risks. In this approach, however, relatively small quantities of chemicals are added which do not contain uranium as an impurity. Because of the simplicity of the preconcentration step, larger numbers of water samples may be extracted in a suit-
15 001
10 ooc
e
a
i
5 000
-
/
HH BOREHOLE 250ml HO BOREHOLE 250ml o500ml SEA WATER
1
2
3
4
19 Uranium added
Figure 2. Standard addition lines for three water samples
ably equipped field laboratory. Although delayed-neutron counting requires access t o a nuclear reactor with a suitable counting system, containers with resin can, if time is not of prime importance, be airmailed t o the nearest facility.
ACKNOWLEDGMENT The authors thank B. V. G. Raisey, F. M. Sunde, and J. N. van der Walt for their support with the experiments.
LITERATURE CITED ( 1 ) D. Ostle, R. F. Coleman, and T. K. Ball, "Uranium Prospecting Handbook", Proceed. NATO-sponsored Advanced Study Inst. London, Sept-Oct 1971, p 95. (2) H. A. van der Sloot, R. Massee. and H. A. Das, J. Radioanal. Chem., 25, 99 (1975). (3) J. Veselsky, Radiochim. Acta, 21, 151 (1974). (4) K . K. Bertine, L. H. Chan, and K. K. Turekian, Geochim. Cosmochim. Acta, 34, 641 (1970). (5) J. Korkish and L. Godl, Anal. Chim. Acta, 71, 113 (1974). (6) J. Korkish and I. Steffan, Anal. Chim. Acta, 77, 312 (1975). (7) J. Korkish and H. Krivanec, Talanta, 23, 295 (1976). (8) L. R. Hathaway and G.W. James, Anal. Chem., 47, 2035 (1975). (9) E. S. Gladney, J. W. Owens, and J. W. Starner, Anal. Chem., 48, 973 (1976). (IO) E. Steinnes, Radiochem. Radioanal. Lett., 16, 25 (1973). (1 1) R. L. Fleischer and A. C. Delany, Anal. Chem., 48, 642 (1976). (12) M. C. B. Smit, PEL-248, Jan. 1976. (13) L. A. Curie, Anal. Chem., 40, 586 (1968). (14) H. Hamaguchi, K. Ishida, and R . Kuroda, Anal. Chim. Acta, 33, 91 (1965). (15) D. E. Robertson, Anal. Chim. Acta, 42, 533 (1968). (16) G. M. Reimer, Radiochem. Radioanal. Lett., 21, 339 (1975) (17) T. Hashimoto, Anal. Chim. Acta, 56, 347 (1971).
RECEIVEDfor review August 30,1976. Accepted October 13, 1976.
ANALYTICAL CHEMISTRY, VOL. 49, NO. 1, JANUARY 1977
69