Anal. Chem. 1985, 57, 1427-1433
concentrations should be avoided. For 0.15 M and 0.20 M oxalate, s = 0.014 M for 32 titrations in 0.15 M and s = 0.021 M for 20 titrations in 0.20 M ammonium oxalate. Since the values of average bias of -0.007 M and -0.008 M for 0.15 M and 0.20 M oxalate, respectively, are small compared to the standard deviation, it is concluded that these titrations are bias free. The precision shown by the above standard deviations for replicate titrations is reasonable for the sample sizes (100 pL) and pipetting techniques used. This indicates that the improved oxalate method based on titrations in 0.15-0.20 M ammonium oxalate to a precisely determined potentiometric end point (the instrumentally determined inflection point in the titration curve) is precise and accurate for free acid determinations in concentrated plutonium and mixed plutonium-uranium solutions.
ACKNOWLEDGMENT The authors express their appreciation to R. C. Lloyd and R. T. Primm for encouragement and support in this work. Registry No. U02(N0J2,10102-06-4;Pu(N03)(,13823-27-3; “OB, 7697-37-2; Pu, 7440-07-5; U, 7440-61-1.
LITERATURE CITED Cleveland, J. M. “The Chemistry of Plutonium”; Gordon and Breach: New York, 1970; Chapter 4. Smith, M. E. “The Determination of Free Acld In Plutonium Solutions”; USAEC Report LA-1864, 1955. Dahlby, J. W.; Waterbury, G. R.; Metz, C. F. “Investigation of Two Methods for Measuring Free Acid In Plutonium Solutions”; USAEC Report LA-3876, 1968. Biihr, W.; Thieie, D. “Determination of Free AcM in Concentrated Nitric Acid Solutions of Plutonium(1V)”; Kernforschungszentrum, Karisruhe Rept, KFK-499, 1966. Thiele, D.; Bahr, W. “Acidimetric Determination of Free Acid and Uranium In Nitric Acids Solutions in the Presence of Pu(1V)”; Kernforschungszentrum, Karisruhe Rept, KFK-503, 1966. Toth, L.M.; Friedman, M. A.; Osborne, H. M. “The Polymerlzation of Pu(1V) In Aqueous Nltric Acid Solutions”; US. Department of Energy Report, ORNLITM-7 180, 1980.
1427
“1979 Annual Book of ASTM Standards, Part 45, Nuclear Standards”; American Society for Testing and Materials: Philadelphia, PA, 1979; pp 420-421. Private comrnunicatlon quoted in ref 12. Bulls, K.; DeDalmassy, B. C.; Maurice, M. J. “The Determination of Free-Acidity in Plutonium Containing Soiutlons and the Seml-Quantitative Acidimetric Estimation of Piutonlum”; Euratom Rept, EUR-3934% 1968. Williams, T. L.; Jordan, L. G.; Ouyts, L. L.; Huff, G. A. “The Determination of Free Acld in Plutonlum Solutions by Thermometric Titration”; US. Department of Energy Report, AGNS-35900-2.4-82, 1980. Rodden, C. J., ”Analyticai Chemistry of the Manhattan Project”; McGraw-Hill, New York, 1950; pp 214-216. Moore, R. L.; Schmidt, H. R. “The Determination of Free Acid in Solutions Containing Uranyl Nitrate, Aluminum Nitrate and Sodium Dichromate”; USAEC Report HW-14603, 1949. Marshall, J.; Bar-Nun, A. Anal. Chim. Acta 1963, 20, 22. Obrenovlc, I.D. And. Chlm. Acta 1959, 27, 560. Dizadar, 2. I.; Graddy, R. H.; Dorsett, R. S. Anal. Chem. 1968, 40, 429. Schmieder, H.; Kuhn, E. Wanfa 1969, 16, 691. Motojlma, H.; Izawa, K. Anal. Chem. 1964, 36,733. Booman, 0. L.; Elliot, M. C.; Kimball, R. B.; Cartan, F. 0.; Rein, J. E. Anal. Chem. 1956, 30,284. Baumann, E. W. “Determination of Free Acid by Standard Addition Method in Potassium Thiocyanate”; US. Department of Energy Report, DP-1632, 1982. Pflug, J. L.; Miner, F. J. And. Chlm. Acta 1860, 23,362. Damlen, N.; Cauchetier, P. Anal. Chlm. Acta 1968, 4 1 , 483. Bryan, G. H.; Thompson, J. K.; Van Tuyl, H. H.; Brown, C. L.; Ryan, J. L. “Results of Research to Evaluate Solld Plutonium Nitrate as a Safe Shipping Form”; U.S. Department of Energy Report, BNWL-1941, Revlslon i,1977. Drummond, J. L.; Welch, G. A. J . Chem. SOC. 1956, 2565. Staritzky, E. Anal. Chem. 1956, 28,2021 and 2022. Uekl, T.; Zalkln, A.; Templeton, D. H. Acta Crystallogr. 1966, 20,836. Taylor, J. C.; Mueiler, M. H.; Hltterman, R. L. Acta Crysta//ogr. 1966, 20,842. Gibson, G. J. Am. Chem. SOC. 1954, 76, Ferraro, J. R.; Katzin, L. I.; 909.
RECEIVED for review November 5, 1984. Accepted February 5,1985. This work was performed for the U.S.Department of Energy’s Consolidated Fuel Reprocessing Program under Contract DE-AC05-840R21400.
Application of Penning Mixture Proportional Counters for Gas Turbine Engine X-ray Fluorescence Spectrometer Wear Metal Monitor L. Leonard Packer* United Technologies Research Center, Silver Lane, East Hartford, Connecticut 06108 Mar ja-Leena Jarvinen and Heikki Sipila Outokumpu Oy, Institute of Physics, P.O. Box 27, SF-02201 Espoo 20, Finland The X-ray fluorescence spectrometry/anaiysls detection sensltivitles for titanium, iron, copper, and sliver In a lubrlcant oli matrlx have been determined by uslng argon-neon, argon-xenon, and argon-isobutane Pennlng mlxture proportlonai counters and curium-244, iron-55, americium-241, and cadmium-109 Isotope source excltatlons. Three u wear metal detectlons under static condltlons of 3 ppm iron, 1.2 ppm copper, 5.4 ppm tltanlum, and 9.6 ppm silver were achleved for a measurement time of 1000 8. The detection sensitivities of flowing anaiysls of “used” gas turbine engine lubrlcant were 3 ppm for Iron and 1.8 ppm for copper. These values can enable the detection of abnormal wear metals in gas turbine englnes.
U.S.Air Force operational procedures employ the analysis of engine lubricant oil to detect abnormal wear in gas turbine 0003-2700/85/0357-1427$01.50/0
engines. This Spectrometric Oil Analysis Program (SOAP) requires lubricant samples to be periodically taken from the engines and analyzed in a laboratory for various wear metal concentrations. The concentration of the wear metals is usually determined by emission spectrographic or atomic absorption analysis. Close support of an analysis laboratory is required to produce timely wear metal data. A continuous in-line monitoring system installed on aircraft engines would augment the current SOAP program. The use of in-line monitoring would minimize the immediate need for laboratory analyses during aircraft deployment. Previous research (1,2) demonstrated the capability of an X-ray fluorescent (XRF) system for monitoring the concentration of iron in lubricating oil. Correlation analysis of XRF data with atomic absorption analyses yielded an iron sensitivity equivalent to 9 ppm. This laboratory investigation was directed at evaluating the lowest XRF measurement limits for iron, copper, titanium, 0 1985 American Chemical Society
1428
ANALYTICAL CHEMISTRY, VOL. 57,
NO. 7, JUNE 1985
OIL SAMPLING CHAMBER
Table 11. Gain Shift vs. Count Raten OIL
SiGNAL PROCESSING
EXCITATION SOURCE
% shift
2 297 4 208
231 230 229 227 226 226 225
0 0.4
9 486 15253 18153 21 151
Figure 1. XRF engine mounted components.
press, % resolut window cathode atm (iron-55)' diam,* cm diam, cm voltage 3.7
14.3
1.9
2.5
800
7 7 2
13.0 13.1 13.7
1.9
2.5 3.0 2.5
620
2.2
1.9
0.9 1.7 2.2
2.2 2.6
Seven atmosphere of neon-argon proportional counter (3-cm diameter), iron-55 source to detector distance changed to effect count rates, ORTEC PA 242 preamplifier, full window illumination. and Cr filter.
Table I. Penning Gas Proportional Counter Detector Specifications Outokumpu Type 490
argon-isobutane neon-argon neon-argon argon-xenon
peak channel
12 091 PROPORTIONAL COUNTER
counter gas
count rate, cps
700
720
'Measurements performed at 8000 cps Cr filter - full window illumination by using a modified EG&G ORTEC PA 242 H Hybrid preamplifier and a Canberra 2020 amplifier (2 ps). *All detectors are 14.5-cm long and contain a 0.5-mm beryllium window, except Ar-Xe contains a 0.25-mm window.
and silver contained in a lubricant oil by using a detector and isotope excitation source applicable to the future application in an in-flight wear metal monitor system.
EXPERIMENTAL SECTION Wear Metal Monitor. This experimental work reflects the need to operate an X-ray fluorescent spectrometer in the hostile vibrational and heat-producing environment (approximately 200 "C) of a gas turbine engine. The engine-mounted XRF monitor (Figure l), detector, excitation source, oil flow chamber, and preamplifier are the key components determining the detection sensitivity of the in-line wear metal analyzer. The signal processing and data reduction component would be located in an air-conditioned aircraft instrument compartment, and it is assumed that modification of existing commercial instrumentation would provide adequate demonstration instrumentation. Detector. The ultimate sensor performance is determined by the X-ray detector performance. The X-ray detector converts the X-ray photons into ion pairs that are collected as a series of short current pulses. The fundamental electrical ion pair conversion, the statistics of ion pair production, the intrinsic electronic noise, and a myriad of other competing interactions determine the particular X-ray detector quality. Stability and good resolution in the 3-25-keV X-ray energy range are required for a wear metal elemental analyses. Energy resolution sufficient to identify the specific X-ray photons emitted by the silver, titanium, iron, and copper wear metal debris within the excitation backscattering X-ray background and detector effectiveness (efficiency and size) are the essential requirements for the X-ray detector. Solid state X-ray detectors such as germanium and silicon require cryogenic cooling. A mercuric iodide detector was considered, but the approximate 4-mm2 detector area, fabrication uncertainties, and temperature limitations were the reasons for not pursuing the higher X-ray resolution capabilities of this type of detector. The gas-filled proportional counter detector provides internal gas gain, minimal temperature dependency, resistance to vibration, and mechanical reliability, and it was operated successfully during gas turbine testing ( 2 ) . The use of Penning gas mixtures of Ne-Ar, Ar-Xe, and ArC4HI0(3-6) (Table I) in the proportional counter reduces the average ionization energy of the gas mixture. This increase in the number of ion-pairs per X-ray events improves the energy resolution. The high f i t Townsend ionization coefficient of these mixtures permits the use of anode voltages in the 600-800 range with a fill gas pressure of 7 atm. The use of above 1 atm gas filling provides increased detector efficiency and a decrease in the
low-energy background caused by wall effect. The use of rare gases can provide virtually unlimited lifetime (7). In addition, the undesirable count rate gain decrease with increasing counting rate feature of standard gas-filled proportional counters is reduced by using Penning gas mixtures. For the count rate increases anticipated for gas turbine lubricant oil wear metal concentration ranges, Table 11, the gain shelf is within acceptable limits. Preamplifier. The electrical and mechanical design of the preamplifier and the physical connection to the detector are critical to the resolution, gain stability, linearity, and count rate capability of the X-ray spectrometer. Of particular concern in obtaining the best energy resolution is the fact that electrical noise modulates the statistically varying detector signal, causing apparent energy spread. Low noise is also essential because operating the proportional counter under decreased gas gain conditions requires that the noise contribution be at a spectral energy position of less than about 1 keV. Initial wear metal monitor test stand engine testing ( I ) was performed by using a modified discrete component preamplifier (Tennelec TC-133). Preamplifier engine noise pick-up limited monitor operation to the engine idle condition. Wear metal monitor full engine power test stand operation (approximately 45 000 Ibs of thrust) was achieved (2)by using a hybrid preamplifier (LeCroy TRA-510). This hybrid preamplifier uses solidstate amplifier components and short-leg firm mechanical mounting of the input field effect transistor. In this laboratory investigation, two additional hybrid preamplifiers were evaluated Amtec A203 and the Ortec H242A. When a 3.7-atm argon-isobutane proportional counter was used the 5.9-keV X-ray (Cr filter) full-width half-maximum percent resolution at 10000 counts per second was determined to be preamplifier LeCroy TRA-510 AMPTEC A203 ORTEC H242A ORTEC H242 modified
percent fwhm resolution 15.5 17
14.5 14.3
For count rate shifts of 15000 counts per second to 20000 counts per second, the portion of the 5.9-keV X-ray peak channel shifted about 0.8% for the A203,0.5% for the H242A, and 2.4% for the TRA-510. The split metal case packaging, Microdot input connectors, and extra features of the standard ORTEC H242A were not suitable for flowing lubrication testing. The following modifications were performed: The timing channel was removed. The input field effect transistor was repositioned and the circuit tuned for maximum energy resolution. The preamplifier noise was measured a t zero capacitance to be 1.3 keV silicon equivalent and the gain is 438 mV/MeV silicon equivalent. The high gain resulted from the use of only stray capacitance (approximately 0.1 pF) in the feedback loop. This was done to reduce preamplifier noise to the absolute minimum. This high gain has resulted in sensitivity to temperature changes. Tests show that a 0.06% /"C change results from the high preamplifier gain configuration. Addition of 0.5-pF capacitance to the feedback circuit would reduce the temperature coefficient by about a factor of 5 . Radioactive X-ray Excitation Sources. There are practical limits in the selection of the radioactive material to provide the
ANALYTICAL CHEMISTRY, VOL. 57, NO. 7, JUNE 1985
1429
Table 111. Radioactive Material Excitation Source Characteristics princ X-ray emission, keV curium-244 americium-241 iron-55 cadmium-109
princ y emission, keV
14-23 12-22 5.9-6.5 22.1-24.9
half-life, year
spec activ, mC/mg
theoret yield per disintegration, %
18
81 3.4' 40
10" 36* 26c
none 59.6
433 2.7 1.24
88.7 (4%)
lOOd
y ray emissions from a decay of 241Am. Manganese K X-rays from electron capture decay a L X-ray emissions from (Y decay of 244 Cm. of @Fe. Silver K X-rays from electron capture decay of lWCd.
I
10
I
1
1
50
100
150
x RAY ENERGIES (keVi
Flgure 2. Metal foil curium-244 excltation.
20
100
20 0
30 0
%RAY ENERGIES (KeV1
Flgure 4. Metal foil americium-241 excitation. ---
I
I
ENERGY W V I
Flgure 5. Calculated 80% absorption In oil.
optimum X-ray excitation source. The excitation source must emit sufficient X or y radiation above the photoelectric absorption energies for the elements being analyzed but not emit interfering radiation in the energy range of the element being analyzed. The elemental fluorescent yield, excitation energy path length, and elemental photoelectric cross section also affect the trace analysis detection limits. Other factors such as a sufficiently long half-life to minimize source replacement and Nuclear Regulatory Commission procedures affect excitation source selection. The radioactive material source characteristics and X-ray detection factors are summarized in Tables I11 and IV for "Cm, =Fe, %lAm, and lo9Cd.
Curium-244. The X-ray excitation of iron in the lubricant a t the parts per million level was previously evaluated ( Z ) , using a 60 mC/Cmm-diameter disk "Cm (Isotope Products Laboratory PH244-100). Curium is a relatively clean source of L X-rays in the 12-23-keV energy range, suitable to fluoresce copper, iron, titanium, and silver X-rays. The K lines of titanium, iron, and copper and L line of silver spectra obtained by 244Cmexcitation from metal foils illustrate the X-ray resolution of the proportional counter (Figure 2). Iron-55. The excitation photoelectric cross section for titanium and silver is 10 times as great for the 5.9-keV X-rays from 55Fe than it is for the 12-23-keV 244CmX-rays (Table IV). Silver and titanium metal foil excited by an S5Fesource, Figure 3, shows spectra of silver and titanium and the backscattering of the source line. A 20-mC l i e configuration =Fe source (Isotope Product Laboratory PH-55) was used for these measurements. Americium-241. The photoelectric cross section times the fluorescence yield is about 2 times higher for K-shell silver X-ray
1430
ANALYTICAL
cHEMismy. VOL.
57. NO. 7. JUNE 1985
Table IV. X-ray Fluorescence Analysis Factors ormc
min e x i t energy. Lev iron
silver 'K"
?-ray emission energy, keV
7.1 8.9
6.4
25.5
22.1
fluoresc yield 3.5
X
8.3 x
10.'
excitation photoelectric awn section, ml/g 6 keV 14 keV 23 keV 80keV 82.5
lo-'
79.7
21.6
~
re1 detect eftiieienw 7 atm of 3.7 atm of neon.% argon, 7'0
oil depth for 80w absorp? mm
0.95
46
81
2
5.43
2
10
45
.Assume 2.7-cm path le+ for neon and 2.3 for argon and a 0.5-mm beryllium detector window. 'Assumed carbon absorption coefficient with a densitv of 0.987 (nlcm9.
Flouv (1. scalter.
'% pure metal XRF superimposed over lubrlcam back-
fluorescence hy the 59.6-keV %'Am y than for L-shell silver excitation by Wcm CTablea III and TV and Figure 4). The longer
/ npuv 7. XRF lubricant stalk sample howK.
lubricant path length for the 22.1-keV silver K X-ray, Figure 5, increases the probability for silver atom interactions. A 45-mC 3-mm-diameter spherical source (Amemham Corp. AMC 25) was used to evaluate the applicability of ='Am excitation for silver detectability. Cadmium-109. The 2 2 h V X-ray emitted by 'W can excite copper, iron, titanium, and silver X-rays (Table 111and Figure 6). The photoelectric cross section is about one-third that of curium, but cadmium haa about 10 times more X-ray output per millicurie than curium (Table IV). The major disadvantage is the unsuitability of ita 1.24 year half-life for practical field usage. A 3-mC bum-diameter disk '%murce I was used (Amemham CUC. 13055). Measurement Geometry. The lowest limit of detection for iron, copper, titanium, and silver in MIL-L-7808 lubricant WBB measured by using four different proportional counter detectom and three different X-ray excitation radioactivematerial sources. The deteaion levels, counts per parts per million, were determined hy using a series of Conostan organometallic standards in MIL L-78082-based oil. Six-milliliter lubricant standard samples (3.4-mm depth) were used for iron, copper, and titanium, and 25-mL lubricant standard samples (15-mm depth) were used for silver measurements. The sampling measurements were performed in side murce geometry, Figure 7, hy using an aluminum sample holder containing a 0.13-mm Kapton window. 'Used" Gas Turbine Engine Lubrication Flowing Measurements for Iron and Copper. Flowing lubricant sampling measurements were performed by using a nozzle configuration
ANALYTICAL CHEMISTRY, VOL. 57. NO. 7. JUNE 1985
1491
REGIONOF-INTEREST
H
Figure 8. oil sampling chamber. 1
5
10
15
X-RAY ENERGIES [keV)
Figuv IO. Curium244 lubrication backscatter spectrum.
Flgum B. Cbs&bop sampllng system
sampling chamber used during engine testing, Figure 8. The aluminum chamber is designed to disperse the lubricant flow uniformly over the 0.13-mm-thick beryllium window. Measurements were performed by using the 3-cm-diameter 7-atm neonargon proportional counter and the BO-mCi W!m excitation source. The closed-loop sampling system, Figure 9, includes a water bath and a circulation pump. A water bath is required to reduce the localized heating effect on the temperature-limited organometallic standards. The 4-L capacity system was calibrated by using iron and copper organometallic standards in MILL7808 base oil. Three approximately 2.5-L 'used" gas turbine engine lubricant samples (USAF furnished) were measured for iron and copper concentrations. The engine lubricant samples were heated to 110 "F and circulated at a flow rate of 0.5 L/min at 5 psig. Pump suction was from the bottom of the beaker, and return was above the beaker lubricant level. X-ray analysis data acquisition time was loo0 8. R E S U L T S AND DISCUSSION 1. Lowest Limits of Detection-Static Sampler. Curium-244 Excitation-Iron, Copper, a n d Titanium Measurements. Measurements of the wear metal elements were initially performed by wing the W m excitation source. The X-ray lubricant backscatter spectrum, as detected by the 7-atm N e A r counter, Figure 10, shows the background con-
tribution from the excitation source, the gold source holder, and the iron contamination in the beryllium detector window. The 18.3-keV and 14.3-keV principal X-ray excitation energies provide a continuous incomplete energy capture low-energy backgnund over the 3-&keV region of interest. The spectrum shows the regions of interest for silver, titanium, iron, and copper that are inkgated to determine the X-ray fluorescence responses. The silver background consists of low-energy background from the principal curium excitation energies. The copper X-ray background consists of backscattering from the gold source holder and from the curium backscatter. The iron background consists of both iron contamination in the beryllium detector window and curium low-energy backscatter. For titanium X-rays, the WCm backscatter is the major background contributor. Use of lower level contaminated beryllium would not significantly increase the iron detection sensitivity in as much as the sensitivity is a function of the square root of the background. The complete elimination of iron background contribution from beryllium contamination would increase the iron detection sensitivity by about 18%. The gold X-ray background effect on copper detection is compensated in part by the higher copper photoelectric excitation e1088 section of the gold X-ray energies. The lowest limit of detection for these measurements was defined an being the 3 x (background counts)'/* divided by the counts per parts per million. The background counts are the total counts in each spectral region of interest for blank lubricant samples. The copper and iron lower limit of detection wan evaluated for each of the four detectors (Table I), and the 2-3 ppm 30 detection sensitivity was essentially the same, Table V. Because of the similarities of detector responses, titanium and silver were measured by using the 3-cm-diameter NeAr detector. The titanium 30 detection level was 15 ppm, but silver was not detectable by using a 100 ppm silver standard. The curium-244 excitation intensity could be increased by using three or four additional disk sources. With additional sources, the multichannel analyzer process dead time would be minimized by using the neon-argon detector, Table V. Iron-55 Excitation-Titanium Measurements. The =Fe lubricant backscattering speetra, Figure 11,of both 0 and 100 ppm titanium illustrate the close proximity of the 4.5-keV titanium X radiation to the base of the 5.9-keV manganese
1432
ANALYTICAL CHEMISTRY, VOL. 57, NO. 7, JUNE 1985
Table V. Lowest Level of Detection 3u from Background” gas proport counter
iron, ppm
copper, ppm
titanium, ppm
silver, ppm
count rate, cps
Curium-244 Excitation Ne-Ar (3.0-cm diameter) Ne-Ar (2.5-cm diameter) Ar-Xe Ar-isobutane
3.0 3.0 3.0 3.9
1.8 1.8
15.0
3550 2000 8150 9450
5.4
1050
1.2
1.8 Iron-55 Excitation
Ne-Ar (3.0-cm diameter) Americium-241 Excitation Ar-isobutane
9.6
’1000-s data acquisition (live time).
180
Static sample holder Figure 7.
Table VI. XRF Analysis of Used Gas Turbine Engine Oil (ppm)
sample no.
USAF’ SOAP
XRF flowing
XRF static undisturbed top middle
homogeneous
bottom
Fe OP-201-13 OP-201-14 OP-201-15
10
14 22
4.9 5.5 12.4
2.2 4.5 19.2
5.8 19.1
0 0
0 0.7
0.6
3.0
2.0
2.5
5.6
4.0 18.5
3.2 4.5 21.5
2.1
cu OP-201-13 OP-201-14 OP-201-15
0
2 1
2.4 1.8 3.7
1.2
0 0 2.7
‘Atomic emission.
K energy emitted by the Fe-55 excitation source. Although the 7-atm neon-argon proportional counter contains less than 1% argon, a 2.9-keV argon escape peak is positioned directly at the silver L 3-keV spectrum position. Use of 100 ppm silver standards indicated no silver sensitivity. The calcium peak was due to the material used for source positioning. The 5.4 ppm titanium sensitivity is a 2.7 times improvement over the detection levels achieved with the 244Cm source. A properly designed annular iron-55 source should be capable of reducing the detection levels to about 3 ppm. Also, the use of the thinner beryllium window would increase the transmission for titanium and thus increase the detection sensitivity. The close proximity of the 5.9-keV backscatter to the titanium peak will require careful attention to gain shift stability. Americium-241 Excitation-Silver Measurements. The 241Amlubricant backscatter spectrum, Figure 12, for 0 ppm silver illustrates the low interference in the region of interest for the 22.1-keV silver X-ray energy. The background at the region of interest is essentially incomplete energy capture from the approximately 50-keV inelastic Compton scattering of the 59.6-keV y energy. The high level of background in the range below 10 keV is due to Compton electrons resulting from the 50-keV photon escaping from the gas volume after undergoing Compton scattering. This high background and the lower excitation efficiency of 241Am-24160-keV photons for iron, copper, and titanium, Table IV, when compared with the excitation efficiencies of 55Fe,244Cm,and lo9Cdsources prevents the use of 241Amfor parts per million analysis of these elements. The 9.6 ppm lower limit of silver detection could be improved to about 6 ppm with a properly designed annular 241Amsource. 2. “Used” Oil Analyses. XRF “flowing oil” analyses for iron and copper done on samples of “used” lubricant yielded results considerably different from the values reported by the
20
30
40
50
X-RAY ENERGIES (keV)
Figure 11. Ti calibration in oil; 55Fesource.
USAF SOAP analyses for these samples (see Table VI). The XRF iron analyses were only about half the value reported by SOAP, and the XRF copper analyses were consistently higher than the SOAP values. Discrepancies of this type have previously been reported by other investigators (8, 9) when comparing results between emission and atomic absorption spectroscopy analyses of lubricant oil samples. Additional analyses were done to determine if particle settling (10) in the samples was a factor in these results. Samples (6 mL) were taken immediately after shaking the 1-gal storage containers and analyzed by using “static” XRF techniques. Additional
1433
Anal. Chem. 1985, 57, 1433-1436 Am-241 50 keV BACKSCATTER
in these results preclude a judgement a t this time as to whether the SOAP or the XRF results are correct. Additional studies in this area are necessary. Registry No. Ar, 7440-37-1;Ne, 7440-01-9;Xe, 7440-63-3;Fe, 7439-89-6; Cu, 7440-50-8; Ti, 7440-32-6; Ag, 7440-22-4; 244Cm, 13981-15-2;55Fe,14681-59-5;241Am,14596-10-2;ImCd,14109-32-1; isobutane, 75-28-5.
LITERATURE CITED
X-RAY ENERGIES (keV)
Figure 12. Americium-241 lubricant backscattering spectrum.
analyses were done on samples carefully removed from the containers after standing undisturbed for 15 h a t 75 OF (see Table VI). These results indicate that particle settling was not a problem, but the scatter and inconsistencies reflected
(1) Packer, L. L.; Miner, J. R. AFAPL-TP-75-6, AFAPL Contract No. F33815-744-2024, Jan 1975. (2) Internal UTRC Memo, L. Packer to H. Zickwolf (P&WA/CPD), 10 Jan 1980. (3) Sipila, H. Nucl. Instrum. Methods 1976, 733, 251-252. (4) Sipila, H.;Kiuru, E. Adv. X-ray Anal. 1978, 27, 187-192. (5) Jarvinen, M.-L.; Sipila, H. Nucl. Instrum. Methods 1982, 193, 53-56. (8) Jarvinen M.-L.; Sipila, H. I€€€ Trans. Nucl. S d . 1984, NS-31 356. (7) Sipiia, H.; Jarvinen, M.-L. Finnish Patent Applications 83-3547 and 83-353. (8) Kittirlgs, D. C.; Ellis, J. ASD-TR-68-2, 1984. (9) Brown, J. R.; et al. Anal. Chem. 1980, 52,2385-2370. (10) Schrand, J. B., et al. AFAPL-TR-75-77, Feb 1976.
RECEIVED for review November 28,1984. Accepted February 4, 1985. This work was supported in part by U.S. Air Force Wright Aeronautical Laboratory Contract F33615-81-C-2065.
Optimization of Multielement lnstrumental Neutron Activation Analysis Donald D. Burgess Department of Chemistry, McMaster University, Hamilton, Ontario, Canada L8S 4K1
A method for obtaining optimum conditions in muitielement Instrumental neutron actlvatlon analysis Is described. The technique of simplex optlmlratlon is employed through the prediction, by calculatlon, of y-ray spectra for samples of typical composttion. Response functions sultable for optlmization of muitieiement anaiysls are discussed and the performance of the proposed method Is evaluated.
Neutron activation analysis (NAA) has often been used for the simultaneous determination of several elements in a single sample. This capability is especially useful where multivariate data are required. The determination of the sources of environmental pollution and geological studies are examples. The procedures adopted for multielement analysis must be carefully designed if comprehensive and accurate analytical results suited to the task in hand are to be obtained. The design of procedures in NAA requires that the analyst take into account not only the widely varying properties of the analyte elements but also those of matrix elements. Effects such as instrumental dead time that occur during measurements must also be considered. Although other authors (1-4) have proposed computational methods that assist in this task, method design in multielement NAA has traditionally been carried out on the basis of experience and informed intuition. This paper proposes a new approach to the determination of 0003-2700/85/0357-1433$01.50/0
optimum conditions and procedures in multielement NAA.
THEORY The approach adopted in this work employs the method of simplex optimization by advance prediction of y-ray spectra reported earlier ( 5 ) for the optimization of single-element NAA. The process of optimization is separated into a number of interdependent modules: task definition, spectrum prediction, spectrum evaluation, and simplex optimization. Figure 1 illustrates the organization of these modules and the following paragraphs describe their operation. The optimization process is most easily understood if the adjustable analytical parameters such as irradiation, decay, and counting times are considered to define the several dimensions of a geometrical space. Each point of this space can be associated with a value of a quantity (response) that represents the performance of a procedure that uses the corresponding parameter values. Optimization is then the task of discovering the point in parameter space that possesses the most favorable response. The first module defines the optimization task. The typical composition of samples of the kind under consideration is entered by the analyst and necessary nuclear data are retrieved from a data base. Escape peaks are then added and the saturation activity for unit neutron flux and unit sample size is computed for each y-ray peak obtained from the data base. The analytes and their respective analytical y-ray lines are 0 1985 American Chemical Society
