Preparation of electrodeless discharge lamps for atomic fluorescence

Anal. Chem. 1983, 55, 1817-1819. 1817. Table I. Determination of Palladium(Il) in Synthetic. Mixtures Corresponding to Jewelry Alloy amt of Pd(II) pre...
0 downloads 0 Views 392KB Size
Anal. Chem. 1983, 55, 1817-1819

Application. Determination of Palladium in Jewelry Alloy. A typical jewelry alloy contains 95.5% palladium and 4.5% ruthenium. With the increasing use of palladium for jewelry alloy, a need has arisen for a simple, rapid, and accurate method for determining palladium in its alloys. Synthetic mixtures corresponding to jewelry alloy were prepared and their palladium content was determined by the recommended procedure. The results are presented in Table I.

Table I. Determination of Palladium(I11)in Synthetic Mixtures Corresponding to Jewelry Alloy amt of Pd(I1) amt of Ru(III), amt of Pd(I1) PPm found,a ppm present, ppm 2.00 4.00

a

0.090 0.19 6.00 0.26 8.00 0.38 0.480 10.00 12.00 0.520 Average of five determinations.

1817

2.01 4.00

6.10 7.98 9.96 12.03 -

following amounts (Kg/mL) of foreign ioins were found to give less than 2% error in the determinatioin of 6 pg/mL of palladium(I1): Cu(II), 8210; Ni(II), 650; Co(II), 720; Fe(III), 3.0; Ru(III), 6.0; Pt(IV), 10; Os(VIII), 22; Rh([II), 26.0; Ir(III), 32.0; Hg(II), 850; U(VI), 1100; Zr(IV), 960; Zn(III), 1100; Cr(IlI), 280; fluoride, 2600, chloride, 4800; bromide, 3600; iodide, 10.0; nitrate, 8000; sulfate, 10000; phosphate, 2100; acetate, 1200; oxalate, 850; citrate, 820. ETDA, thiosulfate, Ag(I), Au(IXI), iodate, permanganate, dichromate, and vanadate interferred even in small amounts. Also, alkali metal and alkali earth metal ions caused no interference. Efforts to increase the tolerance limit of cations by the addition of masking agents were unsuccessful. Composition and Stability Constaint of the Complex. The composition of palladium-PPC complex was studied by continuous variations (10, II), mole ratio (12),and slope ratio (13)methods. These methods showed the formation of a 1:1 complex between the metal ion and the reagent. The apparent stability constant of the complex evaluated by the mole ratio method was log K = 4.78 f 0.1 at 27 V. Nature of the Complex. The nature of the complex was studied by passing am aliquot of solution of the complex through cation exchange resin, Dowex 50W-X8, and anion exchange resin, Dowex 1-X8. The orange-red complex was retained by Dowex 50W-X8 and not by Dowex 1-X8. This indicated that the complex was cationic in nature.

ACKNOWLEDGMENT The authors are grateful to L. Julou and J. Molle, Rhone-Poulenc-Centre, Nicolas Grillet, Paris, for supplying pure PPC. Registry No. PPC, 2622-26-6;palladium, 7440-05-3;palladium base, ruthenium alloy, 12727-67-2. LITERATURE CITED (1) Borisova, R.; Mosheva, P.; Ivanova, 7.; Topalova, E. Z . Anal. Chem. 1975. 2 7 4 . 31-34. (2) Sanke Gwda, H.; Thimmaiah, K. N. Z . Anal. Chem. 1976, 208, 279. (3) Sanke Gowda, H.; Thimmaiah, K. N. Indian J . Chem., Sect. A 1976, 74A, 821. (4) Sanke Gowda, H.; Thimmalah, K. N. Rev. Roum. Chim. 1977, 22 (5), 745. (5) Sandell, E. B. "Colorimetrlc Determination of Traces"; Interscience, New York, 1944; Pp 358-360. (6) Beamish, F. E.; Van Loon, J. C. "Recent Advances in Analytical Chemistry of Noble Metals"; Pergamon Press: Oxford, 1972; pp 306-345. (7) Vogei, A. I."A Text Book of Quantitative Inorganic Analysis"; The ELBS and Longmans: London, 1968; p 512. (8) Britton, H. T. S. "Hydrogen Ions"; Chapman and Hall: 1955; Vol. I , p 353. (9) Rlngbom, A. 2.Anal. Chem. 1939, 775, 332-343. (10) Irving, H.; Pierce, T. B. J . Chem. SOC. 1959, 2565-2574. (11) Job, P. C . R. Hebd. Seances 1925, 780, 928-930. (12) Yoe, J. H.; Jones, A. L. Ind. Eng. Chem., Anal. Ed. 1944, 111-115. (13) Harvey, A. E.; Manning, D. L. J . Am. Chem. SOC. 1950, 72, 4488-4493.

RECEIVED for review March 28,1983. Accepted June 8, 1983. A.T.G. thanks the University Grants Commission, New Delhi, India, for the a.ward of a Teacher Fellowship under the Faculty Improvement Program.

Preparation of Electrodeless Discharge Lamps for Atomic Fluorescence Spectrometry M. D. Seltzer and EL. G. Michel*

Department of Chemistry, University of Connecticut, Storrs, Connecticut 06268 Electrodeless discharge lamps (EDLts) have been studied for many years as potential sources foy atomic fluorescence spectroscopy (AFS).The early work was reviewed by Haarsma et al. (1).Michel et al. (2-4) have more recently concentrated on improving the reproducibility of preparation of these lamps. A continuation of the latter work is reported here. Michel et al. (2-4) showed that the reproducibility of the manufacture of EDLs depends primarily on carefully identifying and controlling the variablles which are inherent in the preparation of the lamps. This approach was successful for cadmium (2, 3) and selenium ( 4 ) when ten variable$ were identified and rigorously optimized by using the Simplex algorithm (5, 6). The classical method of preparation of EDLs (1)was very simple. Usually the lamp blank was cleaned by flame heating the quartz to white heat under vacuum. Then the appropriate metal or metal halide 'was put in the lamp and sublimed while in the lamp by flame heating to drive off volatile impurities. Then a few torr of an inert gas (usually argon) was added and the lamp sealed. Some variables were obvious and almost

always carefully optimized, for example, the weight of material and pressue of the inert gas put in the lamp. The method of Michel et d. (2-4) demonstrated, however, that the sublimation process was critical. In that work, the required amount of material was put in the lamp and the sublimation was then brought about in a controlled manner by initiating a microwave discharge in the lamp blank while it was on the vacuum system. The variables that were controlled were the duration and applied power of the discharge and the fill pressure of argon during the discharge. This procedure worked fine for the relatively volatile cadmium and selenium EDLs but here it did not work for manganese EDLs. This was because the heat provided by the microwave field was not sufficient to sublime the manganese iodide that was introduced into the lamp. Two approaches are described here which were designed to facilitate the sublimation of less volatile materials. First, the EDL was thermostated (7) while it was on the vacuum system prior to and during the sublimation stage and, second, ground silica chips of approximately 0.5 mm average

0003-2700/83/0355-1817$01.50/00 1983 American Chemical Soclety

1818

ANALYTICAL CHEMISTRY, VOL. 55, NO. 11, SEPTEMBER 1983

Table I. Instrumentation and Apparatus component

model no.

monochromator focal length, 0.32 m grating, 1800 g/mm aperture, fl4.2 dispersion, 1.67 nm/mm slit width, 1.1mm photomultiplier tube PMT housing photon counter mechanical chopper microwave generator Broida cavity temperature controller heating element air heating assembly premix chamber capillary burner and flame separator vacuum system quartz for EDL blanks silica chips

manufacturer

HR 320

Instruments SA, Metuchen, N J

9789QB PR-1400RF 1112

EMI, NY Products for Research, Danvers, MA Princeton Applied Rsch, Princeton, NJ laboratory constructed Electro-Medical Supplies, Wantage, UK

Microtron 200 210 L (3/4h) TC-1000 CHE29767 EHA 129763 '

see ref 2-4 Vi treosil CFX 3600-04

Theall Engineering Co., Oxford, PA GTE-Sylvania, Exeter, NH Perkin-Elmer, Norwalk, CT laboratory constructed University of Connecticut Technical Serivces Thermal American Fused Quartz, Montville, N J G&M Associates, Oakland, CA

Table 11. Optimized Variables variables

description of each variable

optimum level

boundariesa

Q

weight of silica chips weight of Mn (introduced as MnI, solution) time under vacuum after water removal argon pressure for discharge thermostated hot air temperature during sublimation preheating period prior to sublimation applied microwave power after onset of sublimation duration of sublimation discharge time for cooling after discharge time under vacuum after cooling argon fill pressure before sealing microwave power for operation thermostated hot air temperature for operation

30 mg 69 /.@' 385 s 8.0 mbar 464 "C 394 s 125 W 112 s 292 s 586 s 13.0 mbar 125 W 480 "C

5-50 mg 10-300 pg 60-800 s 1-25 mbar 400-520 "C 100-900 s 50-220 W 30-210 s 100-540 s

wl tl A1 T1 t5 P1 t2 t3 t4 A2 P2 T2 a

100-1080 s

1-32 mbar 50-220 W 400-540 "C

Range over which each variable was tested during the Simplex optimization.

diameter were put into the lamp blank. Thermostating the EDL with hot air raised the temperature to encourage sublimation. The silica chips probably increased the rate of sublimation by increasing the surface area from which the material could sublime. EXPERIMENTAL SECTION Except for the modifications described here, manganese EDLs were prepared in the manner described in ref 2-4 and the output of each EDL was measured by using it to excite atomic fluorescence of manganese in a stoichiometric nitrogen-separated airacetylene flame. The performance of the instrumentation used (Table I) was comparable to the photon counting instrumentation used in ref 2-4. Atomic fluorescence signals were detected by using a slit width which encompassed all lines of the manganese triplet at 280 nm. The variables used to control the preparation of the EDLs were those described in ref 3 and 4 plus three more variables which were related to the modifications, making a total of 13. The new variables were defined as follows: Q, weight of ground silica chips (mg); T1, thermostated air temperature ("C) used prior to and during the sublimation stage of the preparation; t5, preheating time (s) prior to sublimation (this variable was designed to encourage rapid sublimation as soon as the microwave power was switched on during the preparation). Manganese(I1) iodide, which was used for EDL preparation, and a stock lo00 ppm manganese chloride solution which was used for preparing dilute standards, were obtained from Alfa Products, Danvers, MA. RESULTS AND DISCUSSION A Simplex optimization of the 13 variables was carried out and resulted in EDLs that gave significantly better detection

limits than those prepared at the beginning of the optimization process. The optimum values for these variables, as determined by the simplex procedure, are given in Table 11. In order to evaluate the reproducibility of preparation, ten EDLs were made by using the optimized levels of variables in Table 11. The average detection limit (10 s integration time) was 0.2 pg/L (signal to noise ratio equal to 2, where the noise was measured by taking the square root of the background as measured on a photon counter. The detection limit was found to be determined by shot noise). A 2 pg/L manganese solution was used to measure the detection limit. The average detection limit (0.2 pg/L) was better than previously reported detection limits for manganese EDLs (typically 6.0 pg/L (I)). The fluorescence signals obtained from each of the ten EDLs, when using a 100 pg/L manganese solution, gave an average of 1.1 X lo4 counts/s with a relative standard deviation of 15%. This is comparable to the previously reported reproducibilities (2-4). The 100 pg/L fluorescence signal was within the linear part of the calibration curve. The confidence interval based on the t distribution and a 95% confidence level was between 1.0 X lo4 and 1.2 X lo4 counts/s. The stability of all EDLs was such that over about 4 h of operating time their output would drift by about 3%/h. Lifetimes, to half output, were at least 50 h but were not measured beyond that. The modifications to the method were aimed at providing for the same rapid sublimation of the rather involatile manganese iodide as was achieved previously for the more volatile cadmium and selenium compounds (2-4). The attempt to make manganese EDLs with the unmodified method resulted in a slow sublimation (up to 20 min) and demanded high

Anal. Chem. 1983, 55, 1819-1821

r

10

20

30

LO

50

60

70

80

WEIGHT OF SILICA CHIPS IN EDL (mgi

Flgure 1. Influence of the weight of the silica chips on the fluorescence signal. Error bars represent the 15 YO relative standard deviation of the fluorescence signals of ten EDLs (see text).

microwave power. The addition of silica chips to the lamp blank and heating the lamp blank with ibermostated hot air made rapid sublimatic~npossible (less than 2 min). A univariate search (Figure 1) revealed that ai, least 20 mg of silica chips in the EDL was necessary to achieve fluorescence signals comparable to those obtained with the Simplex optimized level of 30 mg of silica chips. Increasing the weight of silica chips beyond 30 mg did not result in appreciable increases in fluorescence signal (Figure 1). We are carrying out further experiments to determiine the effect of the particle size of the chips and of materials other than silica. It may be that the improved detection limit over published work was a result of optimal instrumentation (a high light throughput monochromator and photon counting) and careful

1819

control of all variables which led to facile and hence more accurate optimization. However, there is a possibility that the presence of the silica chips during the operation of the lamp affects its radiant output, for example, by promoting more rapid vaporization-condensation processes in the lamp. This could lead to more rapid regeneration of excited state species in the lamp and more rapid purging of quenching species by condensation. We are currently designing experiments to investigate these postulates. An additional benefit resulting from the modifications appeared to be a reduction in warmup time for the lamp. Full intensity and stability for all lamps were reached within 5 min.

ACKNOWLEDGMENT We thank the University of Connecticut glass technicians A1 Brown and Ward Cornel1 for assistance with silica lamp blanks and the vacuum system. Registry No. Mn, 7439-96-5;manganese(I1) iodide, 7790-33-2; silica, 7631-86-9. LITERATURE CITED (1) Haarsma. J. P. S.; De Jong, G. J.; Agterdenbos, J. Spectrochim. Acta, Part U 1974, 298, 1-18. (2) Michel, R. G.; Coleman, Julia; Winefordner, J. D. Spectrochim. Acta, Part 8 1978, 338, 195-215. (3) Michel, R. G.; Ottaway, J. M.; Sneddon, J.; Fell, G. S. Analyst (London) 1978, 703, 1204-1209. (4) Mlchel, R. G.; Ottaway, J. M.; Sneddon, J.; Fell, G. S. Analyst (London) 1979, 104, 687-691. (5) Deming, S. N.; Morgan, S. L. Anal. Chem. 1974, 45, 278A-28314. (6) Yarbro, L. PI.; Demlng, S. N. Anal. Chin?. Acta 1974, 73, 391-398. (7) Browner, R. F.; Wlnefordner, J. D. Spectrochim. Acta, Part 8 1975, 288, 263-288.

RECEIVED for review February 24, 1983. Accepted June 10, 1983. Acknowledgment is made to the donors of the Petroleum Research Fund, administered by the American Chemical Society, for the support of this research.

Internal Standardization in Energy-Dispersive X-ray Fluorescence Spectrometric Determination of Trace Elements In Urine after Preconcentration with a Chelating Filter Liu Ping, Kazuko M[atsumoto,* and. Keiichiro Fuwa Department of Chemistry, Faculty of Science, University of Tokyo, Bunkyo-ku, Tokyo 113, J a p a n We have previously reported a completely new internal standardization method in energy-dispersive X-ray fluorescence spectrometry (XRF) for trace and major elements in geological and biological powder samples ( I ) . The method is based on the fact that, in energy-dispersive XRF, a linear relation is observed between log (X-ray fluorescent intensity/concentration) and log (energy of the characteristic X-ray line) for the elements of atomic number from 13 (Al) to 35 (Br) in a sample. If two internal standard elements are added to the sample and their X-ray intensities are measured, the linear line can be determined for the sample. The concentrations of the analyte elements can be calculated from the linear line and the X-ray intensity of each analyte element. In the previous report, we applied the method to thick powder samples and found that the addition of three internal standard elements is necessary for obtaining satisfactory results. This is probably due to the existence of enhancement effect that is not included in the theoretical basis of the linear relation ( I ) . In the present study, the method is applied to the determination of trace elements in urine collected on a chelating

filter. Such a sample consists mainly of light elements and no severe absorption or enhancement effect exists (2). Therefore, these samples would be more suitable than a powder sample for the observation of the theoretical relation. As internal standard elements for urine samples, Cr and Ga are selected, since the original concentrationsof these elements are negligibly small compared with the added amounts. The concentrations of Mn, Ni, Cu, and Zn are determined, which are compared with those obtained with the usual calibration method using synthetic standard solutions.

EXPERIMENTAL SECTION Apparatus and Measurement Conditions. The apparatus used for the measurement was a TEFA 6110 system, Ortec Inc., with a Mo anode and a Mo filter. The X-ray tube was operated at 40 kV and the anode current was 200 MA. The measurement was done under vacuum for 2000 s. The characteristics of the Si(Li) detector and the treatment of the X-ray spectra are the same as previously reported (I). Acid Decomposition of Urine Samples. Forty milliliters of urine was placed in a Teflon beaker and 10 mL of HN03 was

0003-2700/83/0355-1819$01.50/00 1983 Amerlcan Chemical Society