presence of all the impurities in question: none of the impurities were detectable when electrode blanks with buffer were arced under the conditions described. RESULTS AND DISCUSSION
Moving plate studies clearly illustrate the relative volatilization rates of the matrix and the impurities (Figures 3 and 4). The spectral intensity of molybdenum is low initially, while the impurity spectra, whether in synthetic standards or unknown samples, exhibit peak intensities during this part of the burn. The continuum exhibits the same behavior as the molybdenum spectrum. Consequently interferences from background, halation, or even weak spectral lines are significantly reduced. The net effect is that all elements which have a higher rate of volatilization than the matrix are greatly enhanced. This includes all those elements which are generally classified as volatile and medium volatile. The sensitivities, for example, of copper, calcium, magnesium, and manganese in this method are lome? than might be inferred from the analytical ranges (Table 11). However, the method W&J not set up for lower estimations because of the relatively high residuals in the blank, and possible contribution from the graphite in the buffer and electrodes. The assumption that carbide formation accounts for the marked depreseion of the molybdenum spectrum was confirmed by Debye-Scherrer x-ray patterns of pulverized residues from samplebearing electrodes. There was little or no evidence of metallic molybdenum, yet strong diffraction lines attributable to molybdenum carbide were present. Accordingly, other refractory metalstungsten, for example-would likewise Be depressed under these conditions. A qualitative estimate of the sensi-
tivity of tungsten in graphitediluted molybdenum under high voltage alternating current excitation revealed a detection limit of approximately 0.5 to I .O%. However, with direct current excitation of zinc oxide-diluted molyb denum samples, a sensitivity of 0.01% was easily attained. The impurity elements in this method are roughly grouped according to their sensitivities and spectrochemical behavior. Two spectrographs were used q a convenience. The elements merely a run on the Littrow spectrograph could be run just as easily on the dual grating, although the superior dispersion and resolution of a grating instrument in the visible region were desirable for the determination of tungsten and the alkalies. By superimposing the spectra from both gratings and filtering out the higher orden it was possible to determine barium, potassium, sodium, and strontium simultaneously on the same plate space required for single spectra. Other impurities could easily be added to this system of analysis. .The method has also been applied successfully to the analysis of massive samples such as sintered rods and pellets. One can drill into the surface of these pieces with a tungsten carbide bit and pulverize the shavings with a tungsten carbide mortar and pestle. This method of pulverizing the molybdrnum contaminates it with tungsten and a trace of cobalt, and therefore precludes a tungsten determination; however, the cobalt eontamination is not high enough to interfere with the internal standardization. PRECISION AND ACCURACY
The accuracy of the method could be evaluated only in relation to synthetic standards, because chemical analyses were not available for any of these elements in the trace range. A comparison
Table 111.
Data on Precision and Accuracy
(20efficient of
I’:lement, 70 + Found; residual av.
\-an-
-Added
Elemrnt GI-
0.0024
bin“ riia 1’bO
0.00095 0.011 0.0092 0 0019 0.0042
Fen Sia
Snn Alb Cah
Cub Mgb
Bac
w
Nac SrC We a
0.002‘2
0.0047 0,0019 0.0020 0.0017 0.0016 0.0016 0.0016 0.0063 0.060
tion, GI
/o
0.0024 0 0034 o.oO091 0.011 0.0094 0.0020 0.0044 000.16 0.0021 0.0021 0.0018 0 0017
0 Ooii
0.0017 0.0060
0.060
11 15 10 12 I6 11
11 14 11 17 11 Y
13 14 1@
-,
Thirteen analyses.
* Fourteen analyses. Ten analyses.
of analytical values with theoretical values of synthetic standards is given in Table 111. Precision estimates are also given. LITERATURE CITED
( 1 ) Dufiendack, 0. S., Wolfe, I ( . A., IND. I ~ GCHBN.. . ANAL.E n . 10, 161 (1938). (2) Dyck, R., Vrleker, T. J., ASAI.. Ciiml. 31,3%2 (1959). (3) McPherson, J. D., “Effect of Rcsidud Elements on the Properties o~,hlctaIs,” Lecture V, “Newer Metals, p. 183, American Society for Metals, Clcveland, 1956. (4) “M.I.T. Wavelength T:tblrs,” G. R. Harrison, ed., Wiley, Ncw York, 1939. (5) Olds, I,. E.. Rmgstorfr. G. W. P., J. Metals 8, 150-5 (1956). (6) Pickman, D. O., Alloy . W e l d s Rev. 8, NO.82,l-8 (1956). RECEIVEDfor review :\[nil Accepted June 22, 1959.
15, 1959.
Spectrochemical Determination of Trace impurities in Plutonium Nitrate Solutions A. J. JOHNSON and EDWARD VUVODA
Rocky Flats Plant, The Dow Chemical Co., Denver, Colo.
b A direct spectrographic method has been developed for determining trace amounts of impurities in plutonium nitrate solutions. A 25-pl. volume of the solution cqntaining 500 y of plutonium is evaporated onto 800 y of sodium fluoride, previously evaporated in a shallow cratered electrode. The electrode is excited by an alternating current arc and the
spectrum recorded on a photographic plate. The method is appliccsbie for the determination of 34 elements.
T
amounts of impurities in plutonium nitrate solutions are usually determined spectrographically by direct spark methods (4, li, 10, !I) or spectrochemical procecium (1, 6-81, whereby the plutonium is separated UCE
from the impurities. The sensitivity of the spark methods is restricted by .the limited amount of sample which can be excited, since the spectral richness of plutonium ihterferes with the impurity spectrum. From experience, the amount of plutonium which can be tolerated on the electrode hxm been placed a t 50 y. To improve the detectability of the impurity elements, a method was needed VOL. 31, NO. 10, OCTOBER 1959
1643
timated by visual comparison with standard spectra. Cobalt was used as an internal standard when photometric techniques were employed.
which would tolerate larger amounts of plutonium on the electrode yet retain the simple sample addition employed by the direct spark methods. The separation methds were not considered applicable, because speed was of primary importance. The graphite electrode alternating current arc method introduced by Duffendack (9, S) and modified by Wilhelm (9) was adapted by this laboratory for the analysis of plutonium solutions. The arc method provided the detectability for an adequate impurity analysis. A portion of the solution was evaporated onto an electrode charge of sodium fluoride, which was excited by an alternating current arc. The impurity concentrations were es-
Table 1. Element
Equipment. The spectrograph was a fixed-position 3.Cmeter Jarrell-Ash Wadsworth mount instrument with a 15,000 lines-per-inch grating giving a reciprocal dispersion of 5.12 A. per mm. in the first order and having a 30inch plate holder. The alternating current arc used for eample excitation was obtained from a JACO Varisource. The visual comparison of spectral line densities was made on a Hilger projection spectrum comparator. A nonrecording Spec Reader micro-
Wave lengths of Spectral lines and lower limits of Detection Lower Limit Lower Limit of Detection, Wave of Detection, Wave Length Element Length y/G. Pu Y / G .Pu 3a79.06 200 10 Ge 3280.68 3256.09 100 In 25 3082.16
La Mg
2Ooo
As Au B
a
EXPERIMENTAL
2349.84 2427.95 2497.73 4554.04 Ba 2348.61 Be 3067.72 Bi 39Z7.67 Ca 2288.02 cd 3942.75 Ce 3405.12 co 3147.06a co 3082.62. CO 3021.35 Cr 3273.96 cu 2599.57 Fe 2874.24 Ga Internal standard line.
200 25 25 4 100 25 100
Mn Mo
Nb Ni Pb Pd Sb
loo0 lo00
Sn
Th
... 25
Ti V
25 -_ 50
in
W
Zr
100
Element Al 3082.16 Cu 3273.96 Fe 2599.57 Fe 3059.09
A A
C A B A
C
25 __
25 2Ooo 25 50
200 lo00 25 loo00 200 40 200 100
2Ooo
Reproducibility of log I. and Log /./\a
Sample No. B B B C
M n 2801.06 Mo 31X2.59 M o 3193.97 Ni 3002.49 Pb 2833.07 Sn 2879.99 V 3185.40 y / g . of p l u t o n i ~
3199.92 3185.40 4294.61 3282.33 3391.98
2Ooo 25
photometer was used to measure the transmittances of the spectral lines. hlicrophotometer readings were converted to log intensity values using a calculating board. Preparation of Standard Samples. Stock solutions for 34 elements were prepared by dissolving weighed quantities of their salts in known volumes to give a final concentration of 2 mg. per ml. for each element. Several master solutions were prepared by
Table II. Wave lengths of Analytical line Pairs Cnnrfintration .hlytical Internal Range, Line Standard Line P.P.M.. Fe 3059.09 CA 3147.06 100-1ooc) M o 3193.97 Co 3082.62 100-1OOO V 3185.40 CO 3147.06 100-1o00 r/g. of plutonium.
Table 111.
4333.73 2795.53 2801.06 3132.59 3094.18 3002.49 2833.07 3242.70 2598.06
Internal Standsrd
... ... ...
Concn., P.P.XI.*
1m
100 100
3147.06
246
CO & 2 . 6 2 ... ...
200 25 248 200 1W 25 242
...
co
3i47.06
Ci1e5cient
of Variation,
% 10.7 10.1 9.7 3.2 9.7 10.0 5.4 10.0 10.4 9.5 6.0
combining compatible stock solutions.
A series of standard solutions was prepared from the master solutions by successive dilutions to give the following concentrations: 50, 20, 10, 5, 2, 1, and 0.5 y per ml. To prevent the hydrolysis of certain cations, the final standard solutions were prepared in 10% nitric acid. The internal standard solution was prepared by dissolving enough cobalt nitrate to give a cobalt concentration of 640 y per ml. Procedure. The lower electrode (UCP Type 4925) was watrrproofed with an ethereal solution (5 grams per liter) of Type N Apiezon grease. A pointed graphite rod, inch in diameter and 11/4 inchcs long (UCP Type 1346B), served RS the counterelectrode. Twenty microliters of a 40 mg. per ml. sodium fluoride solution was evaporated in the shallow crater of the lower electrode. leaving a residue of 800 y of sodium fluoride. The impurity elementa and internal standard were added to the lower electrode by 25-pI. increments. The electrode was dried after each addition. The electrode was then removed to a glove box designed for handling plutonium. Twenty-five microliters of an impurityfree plutonium nitrate solution was subsequently evaporated on the electrode, leaving a plutonium concentration of 500 y . To check for residuai impurities in the Apiezon grease, sodium fluoride, and the plutonium m t r i x solution, separate electrodes were prepared for these compounds and their spectra were recorded on the same plate with the standards. No s i m cant amounts of impurities have been found in these materials, which would interfere with the impurity analysis. The electrode charge "as thoroughly dried before being arced. A variable hot plate with a graphite elertrode block was used to evaporate and dry the electrode charge. Preparation of Samples. The sample electrodes were prepared in the same manner aa the standard electrodes, omitting the impurity elements. Because the samples varied in plutonium concentration, which is determined by alpha counting procedures, it was necessary to dilute or concentrate the sample so as to have approximately 500 y of plutonium in the 25 pl. evaporated on the electrode. The electrodes were excited with a 2400-v0lt, 4.5-ampere altcrnating current arc using an electrode qaration of 4 mm. The exposure conditions were: Preburn, seconds Expiire, seconds Analytical gap, mm. Slit width, microns Slit height, mm.
4 4
4 30 1.5
Three Eastman SA-1 plates were used to record the spectrum over a spectral region of 2100 to 4100 A. second order. Standard photoprocessing procedures were followed using a &minute developing time. When photometric
methods mere used, the plate emulsions were calibrated by the tmc-step method.
The influence of calcium on the stability of log I. appeared to be small for the elements studied. Calcium appeared to have little effect on the spectral line stability of the other impurity elements not included in Table IV. Effect of Plutonium on Line Intensities. In many of the analyses the final impurity concentrations were reported in milligrams per liter. For this type of reporting, it would be convenient to run the samples by removing a fixed volume of the solution, neglecting the plutonium concentration. Because the sodium fluoride served as a buffer and a carrier, ignoriug the plutonium concentration seemed feasible. The effect of vsrying concentrations of plutonium on the analytical line intensities n-as investigated by the procedure used for the calcium study (Tahle V).
RESULTS
Because accuracy and precision were less important than speed, the majority of the impurity concentrations mere determined by visually comparing mmple spectra with standard spectra. The results are rvported in micrograms per gram of plutonium or in milligrams per liter. The wave lengths of the spectral h e s used and the lower limits of detection are listed in Table I. The principal analytical line pairs used are shown in Table 11. Precision. For reproducibility data, three samples were analyzed a number of times. Sample A was repeated 16 times on the same plate, while samples B and C were analyzed 10 times on two plates. The transmittance values for nine elements mere obtained with a micropbotometer. The microphotometer readings mere converted to log intensity values by means of an H and D curve obtained from calibrated spectra. The data are presented in Table III. Accuracy. To ascertain the accuracy of the method, several plutonium solutions were prepared with known amounts of impurities present. The visual results on %he known solutions mere aithin a factor of 2 of the amount present. The photometric results for iron, molybdenum, and vanadium were Within +20% of the amount present for a range of 100 to 1000 p.p m. DISCUSSION
Hutonium Spectrum. Tbe analysis of plutonium solutions was complicated by the spectral richness of plutonium, which interfered with the inpurity lines and produced a high spectral background. It was found kbat the intensity of the plutonium spectrum could be reduced by employing sodium fluoride as a spectroscopic buffer. For many of the impurity lines, a higher line-background ratio was obt3med with the sodium Buoride p m n t , which suggrsted that it served as a speetrograpbic carrier aa well as a buffer. The quenching effect of sodium fluoride on the s p ~ c t d intensity of plutonium is clearly demonstrated in Figure 1. The impurity lines of Cr 2843.25, Mg 2852.13, and P b 2833.07 represent a concentration of 500 p.p.m. in ench SpeCtNm. The direct spark methods mentioned earlier :imited the amount of plutonium vhich could be excited to 50 7. Consequently, by increasing the amount of piutonium which could be exrited to 5'30 7. an increaee in detectability WMI
Table IV. Stability of Log 1. with Different Calcium Concentrations Ca Concn., P.P.M.' Log [AI Log Icu Log IN{ 2.58 0.553 40 1.20 3.14 0.626 1.29 0.591 1.27 2.95 1.21 2.89 0.556 10,000 Av. 1.25 2.89 0.582 Av. dev. 1 0 . 0 3 1.0.155 +0.027 AV. %dev. 1 2 . 4 1.5.4 zt4.6 rig. of plutonium. 2.OOo 5.000
Figure 1. Quenching effect of sodium fluoride A.
Spark rpemum of SO y of Pu
B.
Alternoling current arc spectrum of 500 1 of Pu C. Alternating anent arc rpedmm of 500 y of Pu 4- 800 y of NlrF
Efieet of Calcium on Line Intensities. In the plutonium solutions analyzed calcium mas the most prominent impurity element. An investigation mas conducted to learn the effect of calcium concentration on some of the impurity lines. A plutonium solution maa spiked with four different concentrations of calcium and ten impurity analyses mere made for each d c i u m concentration. The log intensities for the analytical lines were obtained by the usual photometric pmcedures. Log I. for each calcium concentration is compared in Table IV.
Table V.
The influence of plutonium ou the stability of log I. is apparent from Table V. However, the effect varies for different elements, as illustrated by nickel, lead, and tin. To minimize this effect, the plutonium conceutration on the electrode was held to 500
150 7. ACKNOWLEDGMENT
The authors are indebted to A. I(. Williams for preparing the impurityfree plutonium nitrate solution and to C. W. Bamck for the original electrode design.
Stability of Log I. with Different Plutonium concentrations
0.965 0.70 0.727 1.05 1.040 0.374 0.809 1.21 0.630 1.29 O.TJ9 0.365 1.71 1.47 0.550 O.iS5 520 0.795 0.277 0.645 2.70 0.933 2.36 0.992 0.353 1040 1.58 1.54 0.742 0.892 0.342 0.769 Av. rt0.20510.066 fO.12410.033 10.090 zt0.41 ztO.115 *O.@Z Av.dev. Av. %dev. f11.6 f8.4 f13.9 f9.6 111.7 126.6 f19.5 139.2
130
260
2.16 1.68 1.49 1.69 3.76
0.753
0.788 0.693 0.922 0.789
realized. VOL 31, NO. 10, OCTOBER 1959
1645
LITERATURE CITED
J. I(. Faris, J . P., Buchanan, X. F., ANAL.&EM. 30, 1909 (1958). (2) Duffendack, 0. S., Thompson, X. B., ; l j Brody,
Proc. Am. SOC. Testing Muteriab 36,
?art 11, 301 (1936). ( 3 ) Duffendack, 0. S., Wolfe, R. A., IND. -ZNG. CHEM.,ANAL.ED.10. 161 (1938). '4) Fowler, C. A.. Atomic Energy of Canada Ltd., Chalk River, Rept. ?DB 92 (1953).
(5)Fred, M. S., Nachtrieb, N. H., Tomkins, F. S., J. Opt. SOC.Am. 37, 279 (1947). (6LK0, R. K.,V . S. Atomic EnergS. ,omm. Rept. HW-57873 (1958). (7) Reichreiber ;Rein), J. E., Langhorst, A. L., Jr., Elliott, M. C., Zbid., LA-1354 (1952). (8)Van Tuyl, H. H., Ibid., HW-28530 (1953). (9)Wilhelm, H. A., IND.ENG.CHEW, ANAL.ED.10, 211 (1938).
Anaiysis of Petroleum Products in the
(10) Zotov, G., Fowler, C. A., Atomic Energy of Canada LM., Chalk River Rept. PDB 45 (1951). (11)Ibid., PBD 91 (1953). RECEIVEDfor review December 10, 1958. Accepted July 10,1959. Presented in part at the Pittsburgh Conference on Analytical Chemist and Applied Spectroscopy, March 1959,?hsburgh, Pa. Vork done under AEC Contract No. AT(29-1)-1106.
C12
to
C20
Range
jOIpplication of FIA Separatory and Low Voltage Mass S pectromet ric Techniques 6.1. KEARNS, N. C. MARANOWSKI, and G. F. CRABLE Gulf Research & Development Co., Pittsburgh, Pa. ,A procedure has been developed The fluorescent indicator adsorption for determining the composition of (FIA) technique (2,6 ) was found to be petroleum products in the C12 to C ~ O the most satisfactory separatory tool {ange using FIA separations and standbecause of its availability in most petroard and low voltage mass spectroleum laboratories, the ease with wllich metric techniques. The variation of the hydrocarbon types can be identified, low voltage sensitivities of aromatics speed, and freedom of fractions from with the number of substituents is used eluents. The FIA column was evaluio determine the average number of ated with respect to repeatability, substitutions per benzene ring. sample recovery, and ability to separate
production of high quality and commercially valuable petroleum uroducts from low grade charge stocks :mi .become an increasingly important ;art of Tehing technology. Such prodeuures, t o be economically feasible, rewme detailed information concerning tne composition of the material and the :orrelation of composition with chemical m u physical properties. These results an be obtained only from analytical procedures capable of providing detailed .omposition data. Standard and the -ecently developed low voitage mass -oectrometric techniques in con~unction sath suitable separation procedures proide the means for obtaining the dexrea information. This report describes i n anaiytical scheme for materiais in ++e approxlmate carbon number range ,:to Cn,and the results of such anaiyses %muse of che complexity of products n this range, it is desirable to separate he samples chromatographically bc"ore performing mass spectroinetnc .na:vses. Zach separated fraction muid contain predominantly one hy.rocarbon type and &ompound types + .uch have the same molecular formula ,mid be resolved. Tse of these concentrated fractions also results in in:reased peak heights for the low ioniz~ n voltage g analyses. HE
I
1646
0
ANALYTICAL CHEMISTRY
higher molecular weight compounds adequately and was found satisfactory for use with products in this range. The separated fractions are analyzed by mass spectrometric techniques. The compound types and the distribution of condensed cycloalkanes in the saturate fraction are determined using standard mass spectral data and analyses (3,IO). Low ionizing voltage techniques are used to analyze the oiefin and aromatic fractions. Field and Hastings (8) showed the advantage of using low energy electrons to simplify the mass spectrum and Lumpkin (11) present4 calibration data for aroniatic hydrocarbons of higher molecular weight. These data showed that the sensitivity idcreased with the degree of condensation of aromatic nuclei and decreased with molecular veight. A study of the low voltage rnass spectra of a series of alkyl benzenes has resulted in a relationship between sensitivity and the number of substitutions on the benzene ring. This relationship can be applied to aromatic fractions to dekrmine approximately the average number of substitutions per aromatic molecule. This procedure. can be applied only to aromatic fractions free from olefins and saturates. The information so obtained. although of a semiquantitative nature, can be of considerable value in understanding refin-
ing processes and in correlating properties with composition. ANALYTICAL PROCEDURES
Fluorescent Indicator Adsorption.
The fluorescent indicator adsorption technique (2, 5 ) is used for the analysis and separation of total saturates, oIefins, and aromatics. This method is a simple procedure for the chromatographic separation of hydrocarbons using silica gel as the adsorbent and isopropyl alcohol as the displacing agent. Fluorescent dyes are used to make the hydrocarbon-type boundaries clearly visible under ultraviolet light. The technique used for this work diEers from the standard FXA procedure only in the e l i i tion of the capillary tubing between the charger and the separator sections. This modification not only facilitates glass blowing, but also permits the sample to pass more rapidly through the charger section without changing the effectiveness of separation. The three iractions are collected from the tip of the capillary constriction a t the bottom of the analyzer section. With the aid of an ultraviolet light source, the emerging colored droplets can be identified 85 olefinic or aromatic once the colorless saturate fraction has been collected. By this method, eight to ten samples can be processed by one operater per day without difficulty. The repeatability limits for this method as stated in the ASTM procedure are: aromatics, 2.0 volume yo;olefins, 2.0 volume %; and saturates, 1.5 volume yo. Euplicate and, in one caw, triplicate analyses of a Diesel fuel and four diEerent jet fuels showed that the data were well within these limits. A question of importance in the use of
