Parts-per-Million Determinations of Uranium and Thorium in Geologic Samples by X-ray Spectrometry Gerard W. James Kansas Geological Survey, University of Kansas, 1930 Ave. " A ", Lawrence, Kansas 66044
Low levels (0 to 100 ppm) of uranium and thorium in geologic samples may be determined rapidly and accurately by wavelength-dispersive x-ray emission spectrometry. Three sigma detectlon llmlts of 1.2 ppm U and 1.5 ppm Th can be achleved with a total countlng tlme of 5 mln on a powdered 104 sample. The estlmated standard errors of the analytlcal calibration curves are 0.7 ppm U and 1.2 ppm Th. The method Is applicable to a broad variety of sedlmentary and Igneous rock types; matrix effects are sufflciently corrected for by a peak-tebackground ratlo method of data reduction. The mean difference between values obtalned fluorimetrically and by x-ray spectrometry for 36 exploratlon samples containing 1.0 to 20 ppm U was 0.8 ppm.
Table I. Uranium Analytical Calibration Curve Rp
100
50 25 10 5 0
u PPm, x-ray
lRb
2.306 k 0.008 1.653 i: 0.006 1.330 * 0.007 1.136 k 0.003 1.071 * 0.004 0.994 i: 0.001
99.9 k 0.6 49.9 k 0.5 25.1 k 0.5 10.3 i: 0.3 5 . 3 t 0.3 -0.6 i: 0.1
Table 11. Matrix Variations
s (Counts/ (counts/ SI
PPm
Rb
Rapid and accurate low level determinations of uranium and thorium in geologic samples are essential to large-scale uranium exploration programs. Many government laboratories are utilizing neutron activation techniques, but most industrial laboratories do not have access to reactor facilities. Of the other instrumental methods of analysis, only wavelength-dispersive x-ray emission techniques have rapid analytical times, adequate sensitivity, and simple sample preparation (1). Previous work has demonstrated the utility of x-ray methods for the analysis of ore-grade rocks and ore concentrates, but most investigations indicated x-ray methods had inadequate sensitivities for determining low levels (less than 100 ppm) of uranium and thorium and that quantitative analyses required the addition of an internal standard (2-4). More recent work (5) has suggested satisfactory matrix corrections for exploration samples could be obtained by the peak-to-background ratio method (6), which utilizes the relative intensity of scattered tube radiation as an internal standard. The objective of this study is to provide an experimental assessment of the potential applications of x-ray spectrometry for the low level (0 to 100 pppm) determinations of uranium and thorium in a variety of rock types. EXPERIMENTAL Apparatus and Operating Conditions. Analyses were performed using a Philips Model 1410 vacuum spectrograph with a XRG-3000 generator and a Mark I11 data controller and processor. Instrument parameters are as follows: x-ray tube, molybdenum; power, 50 kV and 50 mA; analyzing crystal, LiFzzo; detector, scintillation; PHA, baseline 2.0 V, window 2.0 V; goniometer, uranium La 37.30" 28, thorium La 39.23" 28, background 36.88' 28; collimation, fine; spectral path, air. Procedures. Ten-gram samples, ground to -200 mesh, were analyzed directly as loose powder in a 50-mm diameter sample holder covered with poly(propy1ene)film.Analytical count times were 50 s at the analyte and background positions. Uranium calibration standards were prepared by spiking Ottawa sand with a carnotite ore containing 0.152% uranium. Uranium matrix variation samples were prepared by utilizing reagent-grade chemicals and the carnotite ore. The thorium
Matrix SiO, Si0,-Fe,O,
U
P P ~ )
Rp/Rb
150
5.1 1.9
2.31 2.15
98 220 1395 304
0.9 2.4 13.6 3.8
1.87 2.01 2.34 2.23
s)
390
(x-ray) 100 87
(1:l) Fez03
CaCO, C
KAlSi,O,
66 77 102
94
analytical calibration curve was determined from analyses of US. Geological Survey igneous rock standards. RESULTS AND DISCUSSION U r a n i u m Analytical Calibration Curve. Results of triplicate analyses of the uranium sandstone standards, containing 0 to 100 ppm U, are presented in Table I. The predicted concentrations were determined by a least squares fit of the peak-to-background count-rate ratios (R,/Rb). The goodness of fit of the regression line was 0.9999, and the estimated standard error for a single determination was 0.7 PPm In order to determine the validity of the KGS sandstone standards, a second set of uranium standards was obtained from the Kerr-McGee Corporation. The recommended values for these samples were 0.5, 19.6, and 44.1 ppm U; using the KGS calibration curve, the predicted x-ray values were 0.7 f 0.3, 19.9 f 0.4, and 46.8 f 0.6 ppm U. S p e c t r a l Interferences. In most geologic samples, spectral interferences with the uranium L a peak position (37.30' 20) and the background position (36.88' 28) are likely to be negligible. Ottawa sand samples spiked with 1000 and 2000 ppm strontium (Sr K a 35.85' 20) had no influence on the uranium peak-to-background ratios. Samples spiked with 400, 1000, and 20000 ppm rubidium (Rb KLY 37.99" 28) indicated a Rb peak tailing correction factor of 0.1 ppm U per 100 ppm Rb. Most igneous rocks and sandstones contain less than 250 ppm Rb and 500 ppm Sr. M a t r i x Considerations. As would be predicted from tables of mass absorption coefficients, matrix absorption characteristics may vary considerably for different rock types. Table I1 presents data obtained from various sample blanks and blanks spiked with 100 ppm uranium. Even though the
u.
ANALYTICAL CHEMISTRY, VOL. 49, NO. 7, JUNE 1977
967
Table 111. Uranium in Altered Sandstones
u, PPm Sample
Fluorimetric
X-ray
Reduced ss. Reduced ss. Reduced ss. Oxidized ss. Oxidized ss. Oxidized ss. Oxidized ss.
11.9 14.5 25.5 4.5 8.3 12.8 45.9
14.8 i 0.4 16.8 i 0.3 28.1 i 0.2 7.4 i 0.5 9.2 i 0.5 16.1 i 0.5 57.3 i 0.3
Table IV. Uranium in Igneous Rocks USGS Standard G-1 Granite G-2 Granite AGV-1 Andesite W-1 Basalt
u PPm (7)
3.4 2.0
1.9 0.6
PPm u (x-ray) 3.2
f
0.2
1.9 i 0.2 1.9 * 0.2 0.7 i 0.3
background count-rates (Rb) and sensitivities (8)vary by an order of magnitude or more, the predicted uranium concentrations (based on the KGS sandstone analytical calibration curve utilizing the peak-to-background ratio method of data reduction) are fairly consistent. For samples with basically the same matrix, large variations in SOz, Fe203,CaC03, and organic material are not likely to have significant influence on the predicted uranium concentrations. Thus, the KGS sandstone calibration curve should be directly applicable to all types of sandstones (unaltered, reduced, and oxidized) and most igneous and metamorphic rocks. Applications t o Exploration Samples. An exploration suite of 36 unaltered grey sandstone samples containing 1 to 20 ppm uranium was obtained from the Kerr-McGee Corporation. Results obtained from duplicate x-ray analyses agreed very well with the fluorimetric values obtained by Kerr-McGee; the average difference was 0.8 ppm, with a range of differences of 0.0 to 2.2 ppm. Results of duplicate analyses of three dark grey-black “reduced” sandstones and four orangish tan “oxidized” sandstones are presented in Table 111. The results of the x-ray analyses are consistently higher than the fluorimetric values; the acid-extractiontechniques utilized in the fluorimetric determinations may not be completely recovering all of the uranium present in these types of samples. Results of triplicate analyses of four US.Geological Survey igneous rock standards are presented in Table IV. A suite of 21 grey, greenish grey, and dark grey-black shale samples, which contained 0.3 to 21 % organic carbon, were analyzed by y-ray spectrometry in the Keith-Weiss Geological Laboratories at Rice University. The y-ray results indicated the shales contained 4 to 70 ppm uranium; the average difference between duplicate x-ray analyses and the y-ray values was 1.6 ppm, with the differences ranging from 0.1 to 6.0 ppm. Twelve siliceous limestone “sil-Crete” samples were submitted to Hazen Research Laboratories for fluorimetric determinations of uranium. X-ray analyses had predicted concentrations of 2 to 102 ppm uranium. The initial fluorimetric results were considerably lower than the x-ray values; the average difference between the two data sets was 28 ppm with a range of differences between 2 and 68 ppm. The samples were re-analyzed using a HF digestion to obtain a higher percentage of recovery of the uranium. The second set of fluorimetric determinations agreed much better with the x-ray values; the average difference was 4 ppm uranium, with a range of differences between 1 and 13 ppm. Determination of Thorium. Results of triplicate analyses of U.S. Geological Survey igneous rock standards and the Ottawa sandstone blank were used to establish a thorium 968
ANALYTICAL CHEMISTRY, VOL. 49,
NO. 7, JUNE 1977
Table V. Thorium Analytical Calibration Curve Th PPm PPm Th USGS Standard (7) R,IRb (x-ray) GSP Granodiorite 99.5 1.933 f 0.016 99.7 i 1.5 G-1 Granite 50 1.405 f 0.005 48.9 i 0.5 G-2 Granite 24.2 1.158 f 0.003 25.1 i 0.4 AGV-1 Andesite 6.4 0.966 i 0.006 6.6 i 0.6 2.7 i 0.2 W-1 Basalt 2.4 0.926 i 0.002 0.0 0.892 * 0.002 -0.5 i 0.2 Sandstone blank analytical calibration curve; the x-ray predicted values are presented in Table V. The goodness of fit of the regression line was 0.9997, and the estimated standard error for a single analysis was 1.2 ppm thorium. The procedures employed in this study for the determination of thorium utilize the same background position as for the determination of uranium. Judging from the results of the analytical calibration curve, the peak-to-background data reduction method sufficiently corrects for matrix effects. The iron and silicon contents of the rock standards ranged from 1.9 to 11.1% Fe203 and 52.6 to 72.6% Si02; background count-rates ranged from 205 to 300 counts per second. Spectral interferences with the determination of thorium in most rocks are unlikely. Potential interferences with the thorium La peak position (39.23’ 20) include several weak lead Lp lines and the bismuth Lp, and Lpz at 39.06 and 39.19’ 28. The addition of 1000 ppm P b to an Ottawa sand sample had no effect on the thorium peak-to-background ratio; the addition of 1000 ppm Bi gave an apparent T h response of 2.4 ppm Th per 10 ppm Bi. Most geologic samples contain less than 50 ppm lead and less than 0.2 ppm bismuth. Analytical Precision a n d Detection Limits. In order to evaluate the reproducibility of the x-ray spectrometric method of analysis for uranium and thorium, splits from one bulk sandstone (containing 11.4 ppm U and 22.6 ppm Th) were loaded into six sample holders and analyzed as separate samples. The root-mean-square (rms) standard deviations of the six determinations of the three 20 positions (Rb = 336.2 f 2.4 counts/s; U R, = 386.7 h 2.8 counts/s; T h R, = 380.6 f 2.4 counts/s) are about the same or slightly less than the standard deviations predicted from counting error. Hence, the detection limits and precision of the method may be directly estimated on the basis of standard counting errors, which are a function of counting time. For a total analysis time of 5 min on a sequential spectrometer (two 50-s measurements at each of the three 28 positions) the 30 (99.6% confidence level) detection limits are 1.2 ppm uranium and 1.5 ppm thorium. CONCLUSIONS Modern x-ray spectrometers have the necessary sensitivity to detect part-per-million levels of uranium and thorium. The results of this study indicate problems associated with sample matrix variation can be reduced to an acceptable minimum by utilizing scattered tube radiation as an internal standard, thus allowing one calibration curve for each element for most types of exploration samples. The determination of uranium and thorium by x-ray spectrometry provides an attractive lower-cost alternative to neutron activation methods of analysis, and allows the rapid and accurate determinations necessary for establishing anomalous and background levels of uranium in large-scale exploration programs. LITERATURE CITED (1) G. W. James and L. R. Hathaway, “Exploration for Uranium Ore Deposits”, International Atomic Energy Agency, Vienna, 1976, pp 31 1-320. (2) I.Adler and J. M. Axelrod, Anal. Chem., 27, 1002 (1955). (3) H. F. Carl and W. J. Campbell, Anal. Chem., 27, 1884 (1955). (4) W. C. Stoecker and C. H. McBride, Anal. Chem., 33, 1709 (1961).
(5) N. H. Clark and J. G. Pyke, Anal. Chim. Acta, 58,234 (1972). (6) G. Andermann and J. W. Ksmp, Anal. Chem., 30, 1306 (1958). (7) F. J. Flanagan, Geochim. Cosmochim. Acta, 37, 1189 (1973).
RECEIVED for review January 24, 1977. Accepted March 7,
1977. Publication authorized by the Director of the Kansas Geological Survey. Presented in part a t the International Atomic Energy Agency Symposium on Exploration of Uranium Ore Deposits, Vienna, Austria, March 29-April2,1976.
Spectrophotometric Microdetermination of Alkaline Earth and Lanthanide Elements with Hydroxy Naphthol Blue and Ethylenediaminetetraacetic Acid Harry G. Brittain Department of Chemistry, Ferrum College, Ferrum, Virginia 24088
A new spectrophotometric reagent for the quick and easy determination of alkaline earth and lanthanlde elements is described. These elements may be determined uslng a reagent composed of 0.001 M hydroxy naphthol blue (HNB) and 0.01 M EDTA, in either phosphate or acetate buffer that has been adjusted to a pH of 6. At low metal concentrations, linear relationshlps between HNB absorbance and metal Ion concentration are observed for alkaline earth and most lanthanide elements. The sensitivity Is metal dependent, but ranges from 1.6 pg/mL for Sm3+ to 19.5 pg/mL for Dy3+.
The determination of total water hardness by titration with ethylenediaminetetraacetic acid (EDTA) is a very important analytical technique ( I ) and continues to enjoy widespread use as a method to obtain concentrations of calcium and magnesium, either together or separately. Many metallochromic indicators exist and are used to observe titration end points, with eriochrome black T, murexide, calmagite, calcein, and others being most commonly used. An excellent review which details the use of many of these reagents has been published by Diehl ( 2 ) . One azo dye that has not received as wide an application as some of the others is hydroxy naphthol blue, 1-(2-naphthalaz0-3,6-disulfonicacid)-2naphthol-4-sulfonic acid (HNB):
3g
="Go3H
H 0
\ /
\ / S03H
The use of HNB as an EDTA indicator in the determination of calcium was first described by Goettsch (3), and later applied to the measurement of Ca2+in serum and urine ( 4 , 5 ) . HNB has also been used to successively titrate Ca2+and Mg2+ in river waters and lakes (6). No work has been published which details the use of HNB in the determination of elements other than calcium and magnesium, or examining the possibility of using HNB as a colorimetric reagent. This paper describes the use of hydroxy naphthol blue in the spectrophotometric microdetermination of alkaline earth and
lanthanide elements, and represents the first use of HNB to determine elements other than calcium and magnesium.
EXPERIMENTAL Apparatus. Absorbance readings were taken at room temperature (25 f 2 "C) on a Beckman model 25, double-beam, UV/visible spectrophotometer. All measurements were taken in matched quartz cells with a 1-cm pathlength. Reagents. Hydroxy naphthol blue was obtained from Mallinckrodt Chemical Works and was recrystallized once before use. Beryllium sulfate, calcium nitrate, magnesium nitrate, barium chloride, and strontium chloride (AR grade) were also obtained from Mallinckrcdt and used without further purification. Samples of terbium, samarium, europium, dysprosium, gadolinium, and praseodymium were all purchased as the 99.9% chlorides from Alfa Inorganics, while lanthanum nitrate came from Fisher Scientific. The 99.9% oxides of erbium, ytterbium, and neodymium were donated by R. B. Martin of the University of Virginia, and were subsequently converted into the chloride salts. Na2EDTA.2H20was purchased from J. T. Baker and used as obtained. Procedure. A stock solution which was 0.01 M in Na2EDTA.2H20and 0.001 M in hydroxy naphthol blue was prepared in both 0.1 M phosphate and acetate buffers, but the final results were independent of the buffer used. The final pH of all stock solutions was adjusted to 6.0. These stock solutions were found to be completely stable for at least 1month, with no detectable changes in absorbance being observed during this time. These findings are in contrast to previous reports ( 4 ) that the indicator solution will not last longer than 2 weeks. All solutions were made up in class A accuracy volumetric glassware, and were prepared using doubly distilled water. The concentration of added metal ions was varied by pipetting small amounts (less than 0.1 mL) of a concentrated metal stock solution directly into the spectrophotometer cuvette and then adding 3.0 mL of the HNB-EDTA stock solution. Each solution was scanned over the 750 to 450 nm spectral region and necessary corrections to the absorbance values were made to account for the dilution of the original HNB-EDTA solution. Each calibration curve consisted of 11points, and covered the concentration ranges of 1 to 600 Fg/mL for the alkaline earth elements and 1to 300 pg/mL for the lanthanide elements. The interaction of each metal with the HNB reagent was examined in both acetate and phosphate buffers, but no difference were detected. Linearity in the absorbance changes was not maintained above the levels stated above and was never observed for Er3+ and Yb3+. The addition of Be2+resulted in precipation of the element and apparently was unable to bind to the HNB reagent. The height of the uncomplexed dye peak at 650 nm was measured, corrected for any baseline drift, and then compared to the abANALYTICAL CHEMISTRY, VOL. 49, NO. 7, JUNE 1977
* 969