ANALYTICAL CHEMISTRY, VOL. 50, NO. 7, JUNE 1978 (4) C.L. Luke, Anal. Chim. Acta, 41, 237 (1968). (5) R. Pueschel, Talanta, I S , 351 (1969). (6) T. Fukusawa and T. Yamane. Anal. Chim. Acta, 84. 195 (1976). (7) R. I. Martens and R. E. Githens, Sr., Anal. Chem., 24, 991 (1952). (8) F. Haase and R. Wolfenstein, Ber. Msch. Chem. &s., 37, 3228 (1904). (9) R. Plesch. GIT fachz. Lab., 17, 677 (1973).
899
(10) D. H. Anderson, Anal. Chem., 48, 117 (1976). (1 1) Perkin-Elmer, Norwalk, Conn., "Analytical Methods for Atomic Absorption Spectroscopy", 1971.
RECEIVEDfor review March 28,1977. Accepted March 6,1978.
Energy Dispersive X-ray Fluorescence Spectrometric Determination of Trace Elements in Oil Samples Hideo Kubo' and Robert Bernthal Nuclear Physics Laboratory, University of Colorado, Boulder, Colorado 80309
Thomas R. Wildeman" Department of Chemistry and Geochemistry, Colorado School of Mines, Golden, Colorado 8040 1
A method is described for the determination of trace elements in petroleum by energy dispersive x-ray fluorescence spectrometry. Minimum sample preparation is required. This is achieved by making small targets and spiking the sample with a solution of Cr and Rh in H2S04or organo-Rh in mineral oil. Use of two spiking elements with different x-ray energies facilitates the determination of x-ray absorption corrections. I n the NBS fuel oil (SRM 1634), V, Fe, Ni, and Mo were detected and the results of the analyses correspond well with the NBS certified values. I n the shale oil, Fe, Ni, Zn, As, and Se were detected. The one-element splklng method works well for samples of low viscosity where the sample can be spread on the supporting foil thin enough so that the absorption of measured x-rays can be ignored, whereas the two-element spike is needed for high viscosity samples (NBS fuel oil) where samples cannot be made thin enough to ignore sample absorption.
Recently, there have been a number of studies on analysis of petroleum and petroleum products for trace elements (1-4). This interest is because elements such as As, Se, and P b may poison the catalysts used in refineries and automobiles. Also elements such as Cd, As, Se, Hg, a n d P b can cause accumulative detriment to the environment when they are present in fossil fuels. T h e presence of P b in gasoline is a n example of both situations. T h e question of trace elements in petroleum is of more concern to t h e development of synthetic fuels made from coal or oil shale. In these processes, the solids must be heated (usually to temperatures of around 500 "C) and this allows the possibility of elements bound in the rock t o be transferred t o the synthetic crude oils (5-7). The trace element analysis of petroleum products is heavily dependent on atomic absorption analyses (8). However, recent studies by cooperating laboratories have shown that digestion procedures are needed for the atomic absorption analysis of heavy crude oils and vacuum residues (2-4). Typically, these techniques require a good deal of experience before they become routine (2-4, 7). Multielement analyses of wear metals in oils have been done by plasma-source emission spectrometry and apparently interelement and matrix effects are minimal (8). Also, neutron activation has been used for multielement analyses of petroleum (9). However, this technique is not *Onleave from the Department of Radiology, Kitasato School of Medicine, Sagamihara, Kanagawa, Japan 228. 0003-2700/78/0350-0899$0 1.OO/O
readily available and reactor operators are especially worrisome about sample leakages or explosions when these types of samples are subjected to intense neutron bombardment. Consequently, t h e study of other multielement analysis procedures for petroleum is attractive. X-ray fluorescence spectrometric (XRF) analysis of oils would be useful, especially if methods could be found so digestions would not be necessary. This paper is a report of the results of the development of an analytical procedure for trace elements in petroleum by energy dispersive X R F t h a t requires minimum sample preparation. It has been demonstrated that if proper attention is paid to instrument parameters, a n energy dispersive X R F system can detect trace elements in the part-per-million range in typical samples (1&12). I t has also been shown that with good matrix correction programs, multielement analyses on such systems are accurate (10,13). Energy dispersive instruments have been designed to analyze either thin targets (12)or thick targets (10, 11). The system used here was the thin target type primarily designed for the analysis of trace elements in biological samples (12). In this instrument the samples are circular targets 3-4 mm in diameter weighing 300 pg. Weighing and uniformly depositing such a small quantity of petroleum was found to be too difficult. Instead, a n internal spike of an appropriate metallic element was used to eliminate uncertainty over the target thickness. Just how t h e internal spike was used in the determination of sample thickness is shown later in t h e paper.
EXPERIMENTAL Apparatus. The small sample XRF system at the Nuclear Physics Laboratory of the University of Colorado has been previously described (12). In this study, the direct x-ray beam was provided by a tungsten anode tube. Three sets of primary filters (between the x-ray tube and sample) and tube voltages were chosen to change the minimum detection limit (MDL) over various element regions. At the 95Y0 confidence level, the MDL is equal to 3 &/sensitivity (with units of ppm) where B is background counts under the x-ray peak. For adequate MDL over a broad energy range, a 0.12-mm thick Lu filter and 55-kV tube voltage were used. To improve MDL for lighter elements (K through Zn), a 0.51-mm A1 filter and a tube voltage of 40 kV were used. To improve MDL for heavy elements (As through Sr), a 0.10-mm Sn filter and 55-kV tube voltage were used. Table I gives sensitivity in counts per nanogram per coulomb of integrated anode current for various vacuum deposited thin metal films for the three system parameters. The unit of coulombs is used rather than time or power because in this system the x-ray tube current changes from sample to sample. The value for V is obtained by inter1978 American Chemical Society
900
ANALYTICAL CHEMISTRY, VOL. 50, NO. 7, JUNE 1978
Table I. Results of Calibration Runs o n Thin Film Targetsa Lu at 55 kV
__
Sn at 55 kV
AI at 40 kV
MD L, cts/ng.C RJ cts/ng.C RJ PPm ... ... ... 7.62 0.0476 ... 10.3 0.0676 ... 3.65b 0.0415 22 10.8b 0.0675 8.7 14.3b 0.0938 4.7 4.08 0.0566 11 14.9 0.0931 4.4 19.3 0.127 2.6 8.61 0.0978 5.0 25.6 0.160 2.0 33.1 0.217 0.9 13.4 0.152 2.7 39.9 0.249 1.2 51.3 0.337 0.8 16.0 0.182 1.9 46.7 0.292 0.8 59.8 0.392 0.5 18.0 0.215 1.7 54.3 0.339 0.6 68.8 0.451 0.4 As 28.5 0.324 0.6 80.6 0.504 0.3 102.0 0.669 0.5 Se 32.0 0.364 0.6 90.9 0.5 68 0.3 114.9 0.754 0.6 Mo 71.2 0.809 0.3 169 1.06 1.2 175 1.15 2.2 0.6 160 1.00 1.9 152.4 1.00 3.3 Rh 88.0 1.00 a For each element, the sensitivity is given in counts per nanogram per coulomb and RJ is the ratio of the sensitivity for element J t o that for rhodium. The minimum detection limit (MDL) in units of ppm is also given for each filter. Value obtained bv interuolation. Element Ti V Cr Fe Ni cu Zn
cts/ng.C
RJ
MDL, ppm
polation. The MDL in the present analysis strongly depends on the mass ratio of the sample oil to the spiking solution, since addition of the latter simply deteriorates the MDL. Therefore, the results in Table I are a summary of typical values of the MDL given in ppm for the oils that were studied. To avoid possible losses by vacuum distillation, the analyses were performed a t atmospheric pressure. Layered collimators made of lead, copper, and aluminum collimate the x-ray beam and determine the field of view of the Si(Li) x-ray detector. A PDP-8/E computer changes the sample, starts analysis for a preset accumulated anode charge, serves as the multichannel analyzer, and stores the resulting x-ray spectrum on a DEC magnetic tape when the analysis is finished. The anode current is varied from a few mA to 15 mA depending on the sample size and choice of filter to keep the system dead time to about 35%. Data-taking time is normally about 2 h. Longer runs are made to obtain reasonable statistics on the background peaks. The data are analyzed on a PDP-9 computer using a peak fitting program, SPECTR. SPECTR unfolds the overlapping peaks in the spectrum using a nonlinear least square method. The x-ray peak shape is assumed to be a Gaussian and the underlining background was chosen to be either linear, parabolic, or exponential. Calculated peak counts were compared t o the standard reference results to obtain the element concentrations. Samples and Reagents. Two samples were analyzed: (1) NBS standard fuel oil (SRM 1634) and (2) a shale oil retorted from 10 kg of shale by Fischer assay (14). Two spiking standards were used; (1)a solution of 80 ppm Rh and 200 ppm Cr in dilute H2S04 and (2) a solution of 90 ppm organo Rh in mineral oil. Sulfuric acid was used in the first spike solution because it is used to extract oil from sewage. The organo Rh standard was prepared by dissolving Rh acetylacetonate in mineral oil. Procedure. The oil samples were heated to 50 to 60 "C for easy pipetting, weighed on a Mettler balance, and mixed with a weighed amount of spiking solution. From 0.05 to 5.0 mg of target material was drawn from the mixture using an Eppendorf pipet and was dropped onto double-layered Formvar film. Table I1 lists the various combinations of sample and spiking standards which were made. There was a question of whether the sulfuric acid solution of the one spiking standard adequately mixed with the oil samples. To answer this, tests were made on Mixture 2 of Table 11. First, the mixture was vigorously shaken for 10 min and two targets were extracted from the top of the mixture. The As to Rh peak counts ratios were measured in these samples and were found to be 0.204 and 0.214: Then, the mixture was left standing for 5 min and two samples each from the top and bottom portions were taken. The As/Rh values were 0.218, 0.208, 0.223, and 0.223. These results indicate that sulfuric acid is an adequate solution to mix with the oil.
CALIBRATION METHOD For the typical XRF analysis of thin films, the target mass and atomic number (Zeffmust be known to calculate the mass absorption correction of t h e radiation. By using a spiking
MDL, ppm
Table 11. Data on the Mixture of Sample Oil with Spiking Solution Sample Spiking solution Weight, Weight, TYpe g Type g Standard 1.21 Cr + Rh 1.42 oil Shale 1.29 Cr + Rh 0.796 oil 0.481 Organo Rh 0.540 Standard
Mixture NO.^ 1
2 3
oil
4
Shale oil
0.608
Organo Rh
0.656
The first column shows the mixture number which is referred to in Table 111. standard, the values of these parameters are not needed. However, the sensitivity ratio (RJ)of element J to the Rh spike must be known, where RJ is defined as
counts of dement J per unit concentration counts of standard Rh per unit concentration (1) These values of RJ are determined by dividing the sensitivity of element J by t h a t of Rh. They are listed in Table I. Specifying the sample and spiking solution weights t o be Ws
RJ =
and WspK, respectively, the spiking element concentration in the mixture can be expressed as "SPK
=
wSPK
wS -k wSPK
' cSPK
where CspK is the spiking element (Rh and/or Cr) concentration in the spiking solution. Similarly, t h e element J concentration in the target sample is
WS
C'J = wS
+ wSPK
.CJ (J f spiking elements)
(3)
where CJ is the concentration in the original sample prior to mixing. For t h e sample-spike mixture, Equation 1 can be written as
where N is the number of peak counts. Inserting Equations 2 and 3 into the above equation, t h e element concentration CJ in the original sample is determined as
ANALYTICAL CHEMISTRY, VOL. 50. NO. 7 , JUNE 1978
c,=-NJ NRh
.- wSPK .-
'%'
ws
RJ
I 000000. (J f spiking elements)
(4)
The determination of RJ takes into account all the absorption in the beam path except that in the sample. There are two possible methods to take into account sample selfabsorption. One is to make thin-layer target samples where absorption can be neglected, so that only one standard element is needed. The other is to calculate the absorption explicitly. To determine the explicit absorption, the known concentration of an element besides R h in the mixture is needed. This is the reason for the addition of 200 ppm Cr as well as Rh to the oil sample. Since a Cr peak is not detected in either oil sample, Equation 4 for Cr in the case of no self-absorption reduces to
C,(measured)
=
Na cRh .-
R,
NRh
-
0
e
Channe I Number
SHALE
3
OIL 4-11-77 #99
(5) 1
oooooo,
: Rh
-
I 00 . 0 U
By treating ( p x ) as a parameter of an effective target thickness, errors due to nonuniform mass thickness can be minimized. By choosing an appropriate Zeff,one can obtain the corresponding mass absorption coefficient ( p / p ) c r (15). Then Equation 6 can be analytically solved for ( p x ) using Newton's method (16). In analogy with Equation 6, Equation 4 is modified to include sample absorption as:
[
-
Ccr(comected) = C,(measured) .
CRh . -
51 1
Figure 1. X-ray spectrum of shale oil using an AI filter and 40-kV tube voltage. Counting time was 1.5 h
Because of sample absorption, Ca(rneasured) may not be 200 ppm. Assuming a target of uniform mass thickness of ( p x ) and a mass absorption coefficient of (PIP), the absorption corrected Cr concentration can be written as ( 1 1 ) :
NJ . -wSPK CJ = ,
901
1
NRh wS RJ 1 - e x P { - @. /( P P x) )J ' ( p ~ ) ~ J # spiking elements (7)
where ( p x ) is determined in Equation 6. In the above procedure, the absorption of primary bremstrahlung and R h K a x-rays in the target is assumed to be zero. For these oil samples, a Zeff= 6 was used in Equations 6 and 7 since the oil matrix is primarily carbon and hydrogen. It should be noted, however, that the change in the absorption correction caused by using a different value of Zeffamounts to a few percent at most. This is because the absorption correction factor depends on the product of ( ~ / p ) and ~ , (pr). Changes in ( p / p ) J due to different Zeffare compensated by the effective target thickness as is seen in Equation 6. For example, the difference in the absorption correction for Zeff = 7 and 12 is about 1.5% for V , while the values for ( p / p ) for V are 31.65 and 168.3 cm2/g a t Zeff= 7 and 12, respectively (15).
RESULTS AND DISCUSSION Typical x-ray spectra of shale oil and standard oil spiked with organo Rh are shown in Figures 1 and 2. These spectra were obtained with a 0.51-mm thick A1 filter at anode voltage of 40 KeV. In standard oil, Ar, V, Fe, Ni, Cu, Zn, W Lp, W Ly + Se K a , Kr, and R h x-rays (all K x-rays except W) are observed. Similarly, Ar, Fe, Ni, Cu, Zn, W Lp, As, W Ly + Se, Kr, and R h are identified in shale oil. Among these, Ar and Kr x-rays are produced in the beam path. When mineral oil spiked with organo R h was analyzed to check for contamination Ar, Fe, Cu, Zn, W LP, W L? + Se, and P b Lcu lines
LI:
3
Zn I IIII I I IIt
l d u u L III I I II I 1 II II I I L
l l d
Channel Number 51 1 STD OIL 4-1 1-77 ~ 2 0 5
Figure 2. X-ray spectrum of NBS SRM 1634 using an AI fitter and 4 0 k V tube voltage. Counting time was 1.5 h
Table 111. Results of the Background Peak Counts per Scattered X-Ray Counts (Summed over Channels 300 to 309) for Three Different Filters"
Filter A1 Lu Sn
Background counts/scattered countsb Target material Fe Cu Zn W Lp Formvarpowder 3.17 1.51 2.47 3.52 Mineral oil 2.88 1.83 1.70 4.20 Formvar powder 24.6 Formvar powder 13.5 3.18 4.47
" The filters are (1)0.51 mm thick AI filter at 40 kV, ( 2 ) 0.12 mm thick Lu filter at 55 kV, and ( 3 ) 0.10 mm thick Sn filter at 55 kV. The ratios are times the value. Le.. 3.17 X lo-). are observed (Figure 3). Similarly, Fe, Cu, and W Lp were observed with a Sn filter, and W LP with a Lu filter. It is, therefore, necessary to determine the source of these background contributions. If they are coming from the spiking solution, they should be proportional to the Rh counts, whereas if they are coming from the x-ray analysis system they should be approximately proportional to the target mass. Unfortunately, the target mass was not measured in the present study. However, it is known that under the same experimental conditions, sample scattered x-ray counts reflect the amount of mass present if the matrix is the same (17). Formvar powder and mineral oil, both organic compounds, can be assumed to have Zeff= 7 . Then, both samples should yield the same intensity ratio of background peak to scattered x-rays (summed over channels 300 to 309 in spectrum). Using the A1 filter, the ratios for Fe, Cu, Zn, and W Lp x-rays in Formvar powder and in mineral oil are given in Table 111. Though the statistics in the peak counts are poor, the Fe, Cu, and W Lb' results imply that these peaks are coming from the
ANALYTICAL CHEMISTRY, VOL. 50, NO. 7 , JUNE 1978
902
Table IV. Results of the XRF Analysis of Standard Fuel Oil and Shale Oila Case
Filter
No. of
Mixture
samples
No.
Element concn, ppm Fe
V
Ni
NBS fuel oil (SRM 1634) 1 2
Lu Sn Sn Sn A1
3 4b 5 6
NBS certified values 5 4 4
320 i 1 5 323 + 9 311 t 7 310+ 5 325 i 11 3 0 3 + 18
1 1 1
3 3
5 3
13.5 + 14.4t 15.1 i 16.2 + 10.8 i
16.9 i
1.0 1.7 2.4 2.8 3.3 2.5
36i 4 32 i 2 33+1 32i 1 35+2 36i 1
Shale oil Fe 4 4 4 4 2
Lu Sn Sn Sn A1
I 8
9b 10 11
Data are given as mean value upside down. I 000000.
i
2 2 2 4 4
Rh
-
P P
sL
d 3
I
100.0
0.6
17.3 i 0.4
2.1 0.5
18.1 + 0.4 18.5 i 0 . 1 18.9 i 0.1
0.84 i 0.16 0.83 + 0.03 0.81 i 0.04
1.0
14.7 + 0.6
Not analyzed
1.0
0.71 i 0.10
one standard deviation for the number of samples given in the 3rd column.
cil
14.6 i 10.5 + 12.2 i 8.4 i 25.0 i
se
As
Channel Number 51 I MINERAL OIL+RH WITH . S I MM AL
0
Figure 3. X-ray spectrum of mineral oil with the organo-Rh spike using an AI filter and 40-kV tube voltage. Couting time was 5 h to obtain good statistics on the small background peaks
system, Zn is thought to be contained more in Formvar powder. A similar experiment was performed with a Lu filter and a Sn filter using Formvar powder and the results are also give in Table 111. These values were used for background subtraction. The source of the W contamination is speculated to be the tungsten anode of the x-ray tube, where W L x-rays are coming through the primary filter. The results of the present analysis are summarized in Table IV. Data are given as mean value f l standard deviation for the number of samples given in the third column. For the NBS oil in cases 2, 3, and 4 and shale oil in cases 7, 8, and 9 of Table 111,the sample absorption for Cr ranged from 20% to 50%. Standard oil has consistently higher absorption than shale oil because of its higher viscosity. Cr absorption for cases 5 , 6 , 1 0 , and 11 was estimated to be a few percent, and thus neglected in analysis. NBS certified fuel oil data are also included in the table for V, Fe, and Ni. The certificate for SRM 1634 reports the abundances of other elements such as Be, S, Cr, Mn, Zn, As, Cd, Hg, and Pb. However, they were below the detection limit of the present system. T o test the uniformity of oils mixed with the Cr and R h spike, the back side (cases 4 and 9) of the target was irradiated almost 1 month after the front side (cases 3 and 8). Comparison of these two sets of data shows almost identical results in trace element concentrations and it can be said that the procedure involving the H z S 0 4solution is adequate for oil analysis. In the case of standard oil, V and Ni are in good agreement with the NBS certified values, regardless of which spiking method and which filter and voltage combination were used.
Target
Fe concentration was in agreement with the certified value within experimental error. However, the value varied from 10.8 f 3.3 ppm (case 5) to 16.9 & 2.5 ppm (case 6). The rather large variation of Fe concentration may be attributed to (1) possible contamination and (2) poor statistics associated with background subtraction. Mo is not reported in the NBS certificate, while we measured Mo concentration of 0.72 f 0.09 ppm in standard oil (case 2) using a Lu filter which was the best MDL for Mo in the present work. There are no previously determined values for trace elements in shale oil. For this oil, the As concentration has been determined to be 26 f 3 ppm by an atomic absorption method with a digestive sample preparation (13). The Se concentration has been determined to be 0.80 + 0.03 ppm using a fluorimetric analysis after sample digestion (13). Our experimental data agree reasonably well among the five different cases for As as well as Se concentration. The Fe level was also satisfactorily reproduced, except one case (case 11) which showed twice the concentration as our other measurements. In addition, analysis using the A1 filter revealed Ni and Zn. Their concentrations were estimated to be 0.65 k 0.40 and 0.54 0.35 ppm, respectively. In summary, the one-element spiking method seems to work very well for samples with low viscosity and the two-element spiking method reproduces the certified value for samples with high viscosity. Because of the simplicity in mixing and handling oil samples, the organic form of the spiking compound seems preferable, though it may be hard to find a suitable compound.
*
ACKNOWLEDGMENT The authors thank Robert Meglen for providing the spike standards and Rod Smythe for stimulating discussions. LITERATURE CITED J. H. Runnells, R. Merryfield, and H. 8.Fisher, Anal. Cbem., 47, 1258 (1975). H. E. Knauer and G. E. Millirnan. Anal. Cbem., 47, 1263 (1975). W. K . Robbins and H. H. Walker, Anal. Cbem., 47, 1269 (1975). H. H. Walker, J. H. Runnels, and R. Merryfieid, Anal. Chem., 48, 2056 (1976). C. H. Prien in "Guklebook to the Energy Resources of the Piceance Crrek Basin, Colorado". Rocky Mountain Assoc. Geologists, Denver, Cob., 1974, pp 141-150. H. Schultz, G. A. Gibbon, E. A. Hattrnan, H. B. Booker, and J. W. Adkins Prepr., Div. Pet. Chem., A m . Chem. SOC.,22 (2), 589 (1977). J. S. Fruchter, J. C. Laul, M. R. Peterson, and P. W. Ryan, Prep., Div. Pet. Chem., Am. Chem. SOC.,22 (2), 793 (1977). V. A. Fassel, C. A. Peterson, F. N. Abercrornbie, and R. N. Kniseley, Anal. Chem., 48, 516 (1976). J. M. Fraser, Anal. Chem., 49, 231R (1977). R. D. Giauque, R. B. Garrett. and L. Y. Goda, Anal. Chem., 49, 62 (1977).
ANALYTICAL CHEMISTRY, VOL. 50, NO. 7 , JUNE 1978 (11) R. D. Giauque, R. B. Garrett, and L. Y. Goda, Anal. Cbem., 49, 1012 (1977). (12) A. C. Alfrey. L. L. Nunnelley, H. Rudolph, and W. R. Smythe, Adv. X-Ray Anal., 19, 497 (1976). (13) T. R. Wildeman and R. R. Meglen, unpublished work, Colorado School of Mines, 1977. (14) T. R. Wiueman. Prepr., Div. Pet. Cbem., Am. Cbem. Soc. 22 (2), 760 (1977). (15) W. H. McMaster, N. Kerr Del Grande, J. H. Mallett, and J. H. Hubbell, Unv. of Calif. Radiation Lab, Report No. UCRL-50174-SCE.2 R-1 (1969).
903
(16) J . Mathews and R. L. Walker, "Mathematical Methods of Physics", 2nd ed., W. A . Benjamin, Reading, Mass., 1970, p 360. (17) H. Kubo and W. R. Smythe, unpublished work, Univ. of Colorado, 1977.
RECEIVED for review August 15,1977. Accepted March 6,1978. This study was performed with financial support from ERDA under Grant No. c(xt4017-1 and is part Of the Environmental Trace Substances Research Program of Colorado.
Determination of Niobium in Geological Materials by Activation Analysis with Pre-Irradiation Separation Ralph 0. Allen' and Eiliv Steinnes" Instituff for Atomenergi, Isotope Laboratories, Kjeller, Norway
The solvent extraction of Nb from geological materials in the presence of carrier-free "Nb tracer prior to neutron irradiation Is described. The possibilities of contamination and interferences are discussed. Dissolution of samples and the solvent extraction of Nb are relatively simple and can be made simultaneously on a large number of samples and the subsequent activation procedure allows more efficient use of the reactor. For the measurement of l o 4 g of Nb, the uncertainty due to counting statistics is f2% and the limit of detection or of contamination is of the order of 5 X lo-' g. This technique was used to measure the Nb content in 15 U.S. Geological Survey standard rocks.
T h e determination of niobium by neutron activation analysis is not common because of t h e time restrictions imposed by t h e short half life ( t l l P= 6.6 min) of the 94mNb formed from this monoisotopic element. Another problem with this measurement is t h e high degree of internal conversion in the decay of 94mNb,which makes the 16.7-keV N b x-ray the only practical peak to use in y spectrometry. In order to gain resolution and efficiency for measuring these x-rays, it is advantageous to use low energy photon detectors (e.g., 1 ) . Nevertheless, the Compton and bremsstrahlung background from other elements in a sample limit the sensitivity and force one to resort to a rapid radiochemical separation (e.g., 2, 3). One of the advantages of these procedures over other techniques is t h e relatively small risk of contamination and/or loss of Nb because the pre-irradiation treatment of the sample is minimized. The chemical separations are carried out after the activation of the sample and in t h e presence of carrier to determine any losses. However, rapid separations on very radioactive samples (like activated geological materials) are difficult, inefficient, as well as a hazard t o the analysts' health. While a pre-irradiation separation of Nb or any other element does risk contamination and/or loss, this approach can improve the sensitivity and efficiency of some neutron activation measurements. In this work the dissolution of geological samples and the separation of N b was made in t h e presence of carrier-free g6Nbtracer which eliminated the need for a quantitative recovery of the Nb. The separation procedures did not need to be as complex 'On leave from the Department of Chemistry, University of Virginia, Charlottesville, Va.
and selective as those used for some spectrophotometric methods ( 4 - 7 ) because interferences by elements like T a are not a problem in the neutron activation measurement. The simple separation procedures helped eliminate some of t h e possible sources of Contamination and the contamination levels were below t h e detection limit of the method.
EXPERIMENTAL Procedure. Samples of the US.Geological Survey standard rock samples were weighed into 100-mL Teflon beakers which had previously been washed with HF. The sample size was not critical although for most of the geological standards a 0.1-g sample was sufficient to ensure an uncertainty due to counting statistics of less than 2 % . To each beaker a measured amount (0.5 WCi) of 95Nbtracer (carrier free "Nb in 0.5'70 oxalic acid; Radiochemical Center, Amersham, U.K.) was added. The tracer was also added to several irradiation vials which were to contain the Nb standards. Sample dissolution was carried out by adding 10 mL of concentrated HF and 1 mL of concentrated HN03 (both reagent grade) to each beaker. In addition to dissohing the matrix, the presence of fluoride ion helped ensure dissolution and exchange of the Nb which is strongly complexed by fluoride. The beakers were placed on a hot plate until the acid was evaporated. Then 10 mL of aqua regia was added to dissolve any phases not soluble in HF, and this too was evaporated to dryness. The residue was dissolved in 15 mL of an aqueous solution which was 5.6 N in HF and 9 N in H2S04. Kb was extracted from this solution into 10 mL of methyl isobutyl ketone (hexone) and the aqueous phase was removed. The hexone was then washed with a fresh 15-mL portion of the HF/H2S04solution. Care was taken in each separation to prevent any physical carryover of the aqueous solutions as the same separatorg funnel was used for the back extraction. The Nb was extracted into 4 mL of a 1.5% H 2 0 2 solution, and then transferred into a clean polyethylene irradiation vial. Standards were prepared by measuring 4 mL of a Nb standard solution (1 ppm Nb in 0.5% oxalic acid) into the irradiation vials containing the 95Nbtracer. Irradiations. After all samples and standards had been prepared, they were irradiated sequentially in the "rabbit" facility of the JEEP I1 reactor (Kjeller, Norway) for 5 min each (flux = 1.5 X 1013n s-l). The samples arrived in the isotope laboratory 30 s after the end of irradiation and the dominant activity was the 190( t l j z= 27 s) from the water. After inserting the next sample into the reactor, the irradiated sample was opened and transferred to a clean polyethylene counting bottle. The irradiation vial was washed with 1 mL of water bringing the total volume of the counting solution to 8 mL which was sufficient to cover the bottom of the counting bottle in a reproducible manner. Radioassay. The samples were counted exactly 3 min after the end of the irradiation on a planar, intrinsic germanium detector (200 mm2 surface area and 7 mm active depth) coupled to a
0003-2700/78/0350-0903$01.00/0 @ 1978 American Chemical Society
