386
Anal. Chem. 1990, 62,386-389
(11) Berridge. J. C. J. Chromatogr. 1982,244,1-14. (12) Baiconi, M. L.; Sigon, F. Anal. C h h . Acta 1986, 797, 299-307. (13) Dose, E. V. Anal. Chem. 1987,5 9 , 2420-2423. (14) Morgan, S. L.; Deming, S. N. J. Chromatogr. 1975, 772, 267-285. (15) Morgan, S. L.; Jacques, C. A. J . Chromatogr. Sci. 1978, 76, 500-504. (16) Bartu, V.; Wicar, S.;Scherpenzeel, G. J.; Leciercq, P. A. J. Chromatogr. 1986,370, 219-234. (17) Jurs, P. C. Computer Software Applications in Chemistry; Wiley: New York, 1986. (18) Deming, S. N.; Parker, L. R., Jr. CRC Crit. Rev. Chem. 1978 (September), 187-202. (19) Shavers, C. L.; Parsons, M. L.; Deming, S. N. J . Chem. Educ. 1979, 56 (5),307-309. (20) Wegscheider, W.; Lankmayr, E. P.; Budna, K. W. Chromatographia 1982. 15 (E), 498-504. (21) Debets. H. J. G.:Baiema, 8. L.; Doornbos, D. A. Anal. Chim. Acta 1983,757, i31-14i. (22) Christophe, A. 8. Chromatographie 1971,4, 455-458. (23) Smits. R.; Vanroelen, C.; Massart, D. L. 2.Anal. Chem. 1975,273, 1-5 (24) Schoenmakers, P. J.; Naish, P. J.; Hunt, R. J. Chromatographia 1987, 24, 579-587. (25) Crow, J. A.; Foley, J. P. M C CC,J. High Resolut. Chromatogr. Chromatogr. Commun. 1989, 12, 467-470. (26) Snyder, L. R. I n Hlgh-Performance Liquid Chromatography: Ad-
vances and Perspectives; Hwvath, C.. Ed.; Academic Press: New York, 1980; VoI. 1, pp 208-316. (27) Nelder, J. A.; Mead, R. Compufer J. 1965, 7 , 308-313. (28) Fjeldsted, J. C.; Jackson, W. P.; Peaden, P. A.; Lee, M. L. J. Chromatogr. Sci. 1983,27, 222-225. (29) Later, D. W.; Campbell, E. R.; Richter, B. E. W C CC, J . High Resolut. Chromatogr . Chromatogr . Commun. 1988, 7 7 , 65-69. (30) Fischer, N. H.; Olivier, E. J.; Fischer, H. D. Fortschr. Chem. Org. Netu&. 1970338, 47-390. (31) Ober, S. S. I n Gas Chromatography; Coates, V. J., Noebeis, H. J., Fagerson, I. S., Eds.; Academic Press: New York, 1958; pp 41-50. (32) Snyder, L. R.; Doian, J. W.; Gant, J. R. J. Chromafogr. 1979, 165, 3-30. (33) Svoboda, V. J. Chromatogr. 1980,207,241-252. (34) Kong, R. C.; Sachok. B.; Deming, S. N. J . Chromatogr. 1980, 799, 307-321. (35) Laub, R. J.; Purnell, J. H. J. Chromatogr. 1975, 172, 71-79.
RECEIVEDfor review August 14, 1989. Accepted November 3, 1989. J.P.F. gratefully acknowledges support provided by grants from the Exxon Education Foundation and the LSU Center for Energy Studies. J.A.C. is the recipient of an LSU Alumni Federation Fellowship.
Determination of Iodine in Oyster Tissue by Isotope Dilution Laser Resonance Ionization Mass Spectrometry J. D. Fassett* and T.J. Murphy Center for Analytical Chemistry, National Institute of Standards and Technology, Gaithersburg, Maryland 20899
The technique d laser resonance ionization mass spectrometry has been comMned wHh Isotope dllutlon analysis to determine lodlne In oyster tissue. The long-lived radioisotope, lMI, was used to splke the samples. Samples were equiibrated witti the wet ashed uwkr controlled c o d t b m and Iodine separated by copreclpttation with sllver chloride. The analyte was ckled as SHVer ammonium Iodide upon a tantalum filament from whlch lodlne was thermally desorbed In the resonance ionlratlon mass spectrometry instrument. A single-color, two-photon resonant plus one-photon Ionization scheme was used to form positive Iodine ions. Long-lived iodine signals were achieved from 100 ng of iodine. The precislon of 1271/1201 measurement has been evaluated by replicate determlnatlons of the spike, the splke callbration samples, and the oyster tissue samples and was 1.0%. Measurement precision among samples was 1.9% for the spike calibration and 1.4% for the oyster tissue. The concentration of iodine determined In SRM 1566a, Oyster Tissue, was 4.44 pg/g wHh an estimate of the overall uncertainty for the analysis of f0.12 pg/g.
Iodine is an essential trace element for man. Its accurate measurement in foods is vital to understanding human dietary intake and verifying that minimum daily allowances are observed, especially in restricted diets such as infant formulations. Isotope dilution mass spectrometry (IDMS) is an inherently accurate technique ( I ) , a “definitive method” for which systematic errors can be thoroughly evaluated. In general, the quantity of an element is determined by IDMS by measurement of the change in isotopic ratio that is produced by adding a calibrated amount of an enriched isotope
of the element to the sample. As such, the technique can only be used for elements with more than one isotope. Iodine has only one stable isotope in nature, lZ7I. However, the radioisotope 1291is long-lived (half-life, 1.59 X lo7 years) and available. Thus, the accurate measurement of I in biological and botanical matrices by using IDMS is made possible. Mass spectrometric methods have been developed both to measure isotope ratios of iodine for IDMS and to measure ultratrace amounts of lZ9Iin the environment, which is typically a t levels IO4 to of stable lZ7I,itself a t part-permillion levels in botanical and biological material. A negative thermal ionization (TIMS) technique has been developed by Heumann et al. They have used IDMS to determine I in salt, chemicals, food, and water (2-4). A second negative thermal ionization technique has been published by Delmore in which lanthanum hexaboride is cataphoretically deposited onto a rhenium ionization filament. This filament treatment lowers the work function of the rhenium and results in high ionization efficiency and measurement sensitivity for iodine (5). Delmore’s technique was adapted in our laboratory and used to determine iodine in SRMs 1572 (Citrus Leaves) and 1549 (Powdered Milk) by IDMS (6). Secondary ionization mass spectrometry (SIMS) was initially investigated to measure 1291(7). Ion sputtering ionization, as is done in SIMS, is also used in accelerator mass spectrometry (AMs), which has achieved the lowest detection limits for lZgI(8). AMS effectively eliminates the limiting background of isobaric molecular interferences which are observed in both secondary ionization and thermal ionization mass spectrometry. The detection of lBI in the environment after the Chernobyl reactor accident is an example of AMS capabilities (9). Laser resonance ionization mass spectrometry (RIMS) has been studied in our laboratory for possible application to lB1
This article not subject to US. Copyright. Published 1990 by the American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 62, NO. 4, FEBRUARY 15, 1990
measurement (IO). A relatively simple, single-color resonance ionization scheme (two-photon resonance + one-photon ionization) was shown to be both efficient and moderately selective. The practical application of RIMS to the problem of lZsImeasurement is impeded by the difficulties in achieving high sample utilization efficiencies while using a combination of pulsed lasers and continuous thermal atomization. These experiments did serve to demonstrate that long-lived ion beams of moderate intensity could be generated from nanogram quantities of iodine. Thus, RIMS appeared capable of being used straightforwardly with IDMS to determine part per million levels of iodine in biological and botanical materials. We report here the results of an IDMS analytical project for determination of iodine in SRM 1566a, Oyster Tissue, using ID-RIMS. This project included the calibration of the lZgIspike for concentration and isotopic composition; the determination of iodine analytical blank; the analysis of previously certified material, SRM 1549 (Powdered Milk); and the analysis of the new SRM material. EXPERIMENTAL SECTION Reagents. High-purity reagents, produced by subboiling distillation (11) at NIST and stored in Teflon bottles, were used in this study. Ammoniacal cyanide solution was prepared by passing 50 mL of a solution containing 1 g of KCN followed by 10 mL of high-purity water through an acid-cleaned cation exchange column and collecting the eluate, containing dilute HCN, in 40 mL of 2 mol/L NH40H. Isotopic Spike. The isotopic spike was SRM 4949A, Iodine-129 Radioactivity Standard, obtained from the Office of Standard Reference Materials, NIST, Gaithersburg, MD. A solution of this material was prepared in a Teflon bottle and made alkaline by the addition of sodium carbonate. The solution was calibrated by ID-RIMS against high-purity potassium iodide and contained (4.181 f 0.081) X lo* mol/g (Is, n = 6). The measured spike ratio, 1271/12BI, was 0.1579 0.0013 (Is, n = 5) or atom fraction content of 0.1364 for lZ7Iand 0.8636 for lmI. Chemistry. Duplicate samples of about 1 g, accurately weighed, were taken from each of three different bottles of SRM 1566a, Oyster Tissue. A third sample was taken from each bottle and dried for 48 h in a vacuum desiccator (160 Torr, 2.1 X lo4 Pa) at room temperature over magnesium perchlorate. The percentage weight loss of each sample was used to correct the sample weight to dry weight for each sample from the corresponding bottle. (Weight losses ranged from 1.9 to 2.3%.) Each sample was transferred to a 250-mL bottle made of Teflon-FEP and spiked with a weighed quantity of lZsIsolution. The spiked samples were wet ashed by heating with 15 mL of fuming nitric acid (90% HN03),ACS reagent grade. This procedure retains iodine by oxidation to iodate while destroying the organic matrix. Once red fumes of nitric oxide were no longer produced, 5 mL of perchloric acid was added to each solution. The solutions were heated to complete the oxidation. Each solution was then diluted with about 10 mL of high-purity water and transferred to a 40-mL centrifuge tube. The insoluble matter (mainly KCIOJ was removed by centrifugation followed by decantation. Iodate was then reduced with 5 mL of hydrazine sulfate solution (2 g of hydrazine sulfate/lOO mL of water). The resulting iodide was coprecipitated with chloride by the addition of 1 mL of 0.01 mol/L HCl and 1mL of 0.005 mol/L silver nitrate solution. The mixed AgCl-AgI precipitate was allowed to stand overnight in the dark and separated by centrifugation,washed, and dissolved in ammoniacal cyanide solution. Ten milliliters of 2.5 mol/L HN03 was added to reprecipitate AgCl-AgI. The precipitate was again separated by centrifugation and dissolved in ammoniacal cyanide for mass spectrometric analysis at about 20 pg of I/mL. Instrumentation. The RIMS instrument has been previously described in detail (12,131.The system consists of a pulsed, 10-Hz laser system capable of producing 3 mJ of UV radiation ranging from 260 to 306 nm, a 60' radius-of-curvature magnetic sector thermal source mass spectrometer with an ion multiplier detection system, and a transient digitizer to accumulate the pulsed ion flux. The mms spectrometer is fully automated with respect to magnet
*
9
387
switching, laser wavelength scanning, and data acquisition from the transient digitizer. For iodine, a 30-cm quartz lens is used to focus the laser to a 200-pm-diameter beam waist in front of the sample filament. Sample Filaments and Sample Loading. Tantalum filaments, 6.2 mm X 1.3 mm X 0.025 mm, in which a shallow groove had been formed, were used to dry the samples. A 5-mL aliquot of the sample solution was dried on the sample filament (100 ng of I) for each individual mass spectrometric determination. RESULTS A N D DISCUSSION The ultimate purpose of this work was to determine the concentration of iodine reliably in oyster tissue by using isotope dilution mass spectrometry. There were two main components to this work: chemical and instrumental. Chemical procedures were developed to ensure equilibration of the spike with the sample and to separate the iodine in a form suitable for measurement. Laser mass spectrometric procedures were investigated for accurate and precise isotopic ratio measurement. Chemistry. The accuracy of IDMS is totally reliant upon the isotopic equilibration of the spike isotope with the natural element in the sample. Thus, loss of analyte from the sample (or the spike added) before equilibration is achieved cannot be tolerated. Once equilibration occurs between the isotopes, loss of the element can be tolerated, since it is the isotopic ratio that defines the original concentration. The key step is dissolving the sample and bringing the spike and natural isotopes to the same oxidation state with no elemental loss. The equilibration of the spike isotope and natural 12'1was studied by using radioactive '%I (half-life,60.1 days) as a tracer isotope and by monitoring the radioactivity during the chemical processing steps. A quantity of lZ5Iwas added to each sample so that sufficient y-ray counts over background were observed in reasonable counting periods (a few minutes). Various methods of wet ashing the organic matrix and oxidizing iodide to iodate were investigated. Fuming nitric acid solubilized the sample while retaining and oxidizing the iodide. Dissolution could be accomplished overnight on a hot plate; addition of perchloric acid and further heating completed the process. Once the samples were oxidized, a slow loss of iodine was observed. However, samples that suffered variable losses of iodine a t this stage were analyzed and the precision of the measured results for concentration indicated that equilibration had been achieved. The 1251tracer experiments indicate that after the final recovery steps, reduction of iodate to iodide and coprecipitation with chloride, 7045% of the iodine is recovered in the experiment. Resonance Ionization. The ionization potential of iodine is 84 340 cm-' (10.25 eV) and the first electron excited state is 54633 cm-'. These parameters dictate the choice of a resonance ionization scheme that uses readily generated UV radiation: a three-photon process, where the absorption of two photons is required to reach the discrete electronic level for iodine and absorption of the third photon results in ionization, a so-called 2 1process. The utility of the generalized 2 + 1process in resonance ionization has been addressed and experimental results shown for Au, Bi, Ca, Cu, Mg, Y, Ta, and Zr (14) using visible wavelengths. We have previously used this scheme to demonstrate RIMS for both Be (15)and C (16) using UV wavelengths. Roughly half of the elements in the periodic table are suitable for resonance ionization using the 1 + 1single-color scheme with wavelengths between 260 and 355 nm. The addition of the 2 + 1 resonance ionization scheme expands significantly the number of elements that can be resonantly ionized using a single tunable laser system. The efficiency of the 2 + 1 process has been examined for C and for other elements (14,16). Both theory and experiment have demonstrated that the sensitivity of the process can be
+
388
ANALYTICAL CHEMISTRY, VOL. 62, NO. 4, FEBRUARY 15, 1990
high for selected transitions. In general, the 2 + 1 scheme will require higher laser intensities than resonance ionization schemes in which all excitation steps are resonantly enhanced. In our experiment, the higher laser intensity is achieved by the insertion of a 30-cm focal length lens that increases the laser fluence in front of the sample filament by a factor of 400, at a cost of decreasing the geometrical overlap with the atom cloud thermally desorbed from the filament. Although we have not measured the efficiency of this arrangement experimentally, ionization efficiency is not the limit to sensitivity in the IDMS determinations described here. The analytical blank provides the sensitivity limit in our measurements. We have previously reported on the spectroscopy for iodine using our experimental system, examining the wavelength range 278-310 nm (10). The most intense signals were observed at 279.7 and 298.2 nm. These wavelengths correspond to two-photon transitions from the ground state of iodine to levels at 71 501.5 and 67 062.1 cm-l. The 298.2-nm wavelength is analytically more useful, being both more sensitive and selective than the 279.7-nm wavelength. Isotope effects must be considered in any RIMS measurement process (17). Even when isotope shifts and hyperfine splitting of resonant levels are expected to be small relative to the laser bandwidth, accurate measurement of isotope ratios cannot be assumed. Fairbank et al. have observed anomalous ratios for Sn and Mo (18). They attribute their observations to the redistribution of population between ground- and excited-state magnetic sublevels by the resonance radiation. We have examined wavelength-dependent effects on iodine isotopes ratios in two ways, using the enriched iodine spike. First, we scanned the laser wavelength for both isotopes and ratioed the results. Second, we measured isotope ratios at wavelengths on both sides of the resonance maximum. Neither experiment showed a significant isotope effect. Our previous investigations have shown that a number of iodine-containing molecular species are formed in addition to iodine atoms during heating in the mass spectrometer. The formation of these species affects the sample utilization efficiency, but the generation of adequate signals was not a problem here. There exists a much larger nonresonant background a t 279.1 nm than a t 298.2 nm. This nonresonant background has been ascribed to dissociation and ionization of these iodine-containing species, which occurs more efficiently for the lower wavelength, more energetic photons. The typical ratio of resonance to nonresonance signal at 298.2 nm was 100. Memory. One of the problems previously noted for iodine analysis using thermal ionization mass spectrometry is instrumental memory (6). This memory is caused by the relatively high vapor pressures of iodine-containing species that are desorbed into the source of the mass spectrometer. The presence of gas-phase iodine species is readily observed using RIMS since iodine signals can be detected without heating of the sample filament. In thermal ionization, the need for having a hot, ionizing filament (1000 "C)increases the likelihood of desorbing iodine which has been deposited on source components. Practically, the problem is avoided by more frequent cleaning of the source of the mass spectrometer, especially between samples of disparate isotopic ratios. The temperature used in heating the sample filament in RIMS is not measurable with an optical pyrometer and is probably less than 550 "C, the melting point of AgI. Thus, memory is expected to be less of a problem for RIMS relative to TIMS. The lower precision for ratio measurement in RIMS makes the observation of memory less obvious. However, when the spike (1271/1291 = 0.1579) was run immediately after a set of samples (1271/12917), the measured ratio was about 4% high. Baking the source with a blank tantalum filament
-
Table I. Analysis of Oyster Tissue, SRM 1566a bottle/ sample
weight, g
run no.
ratio
I, wglg
1-1
1.200 73
1-2
1.03865
2-1
1.18093
2-2 3-1
1.008 84 0.993 26
3-2
0.907 45
29 31 27 33 25 34 24 28 32 30
3.001 2.990 2.404 2.373 3.187 3.132 2.853 2.462 2.422 2.785
4.486 4.468 4.455 4.393 4.420 4.340 4.376 4.518 4.439 4.529 av
std dev RSD
av 4.477 4.424 4.380 4.376 4.478 4.529 4.444 0.061 1.4%
(2000 "C) was effective in eliminating any observable cross contamination among samples, and this precaution was taken between sets of samples with disparate ratios. Fractionation. Isotopic fractionation in the thermal vaporization process is often the limiting source of measurement precision in thermal ionization. We have assessed the extent of isotopic fractionation by dividing the ratio measurements made for each mass spectrometric run into two equal time periods. The average relative difference between the first and second halves of measurement was 0.00% (95% confidence limit, *0.54%). Thus, isotopic fractionation is not significant relative to the other sources of measurement uncertainty. Isotope Dilution Results. The isotope dilution experiment required characterization of the lZsIspike, both isotopic analysis and concentration measurements, as well as the measurement of iodine in the oyster tissue samples and the chemical blank for the procedure. In addition, the previously certified material Powdered Milk, SRM 1549, was analyzed as a check sample to verify the accuracy of the technique. Replicate isotopic measurements were made in the experiment to assess the variability of mass spectrometric measurement. A total of 40 mass spectrometric runs were made (excluding set up runs) consisting of 5 measurements of the isotopic composition of the spike, 10 measurements made on 6 spike calibration samples, 13 measurements made on 6 Oyster Tissue and 3 Powdered Milk samples, and 6 blanks. Typical internal measurement imprecision among ratios was 1% (2 standard deviations of the mean of 50 ratios); the range was 0.55-2.1 %. Ratio measurement precisions for replicates were 1.1% relative standard deviation with 4 degrees of freedom for the spike calibration samples and 1.0% relative standard deviation with 4 degrees of freedom for the oyster tissue samples. The relative standard deviation was 0.9% for measurement of the spike (five determinations). Thus, ratio measurement precision is consistently about 1.0%. Six blanks were determined. These were deliberately overspiked but gave results ranging from 8 to 14 ng. An average blank of 11 ng was subtracted from the isotope dilution results, representing an adjustment of 0.2-0.25'70. The results for sample analysis are summarized in Table I. Since the spike calilbration, spike, and samples were mass spectrometrically determined in an identical fashion, any systematic error due to instrumental discrimination will be cancelled. The uncertainty is dominated by the instrumental measurement precision. The between-bottle variance is larger than the within-bottle variance, but the small number of degrees of freedom makes it difficult to attach significance to this observation. An uncertainty which combines the imprecision of the spike calibration and sample measurement by summing twice their standard errors has been assigned. The result is 4.44 i 0.12 pg/g (2.7% relative).
ANALYTICAL CHEMISTRY, VOL. 62, NO. 4, FEBRUARY 15, 1990
Three samples of Powdered Milk, SRM 1549, with a certified iodine concentration of 3.38 f 0.02 pg/g were also analyzed. The results, 3.44 f 0.10 pg/g (Is, n = 3), were in good agreement with the certified value.
CONCLUSIONS Any new analytical technique requires extensive validation before it will be accepted by the analytical community. IDRIMS was used previously to determine iron in water (SRM 1643b) and serum (SRM 909) as part of the certification programs for these materials (19). Furthermore, a considerable number of osmium and rhenium determinations in geological materials, elemental concentrations by isotope dilution as well as naturally varying Os isotopic compositions, have been made (20-22). Here we have expanded the demonstrated capabilities of the RIMS instrument to another element, iodine, while at the same time providing further evidence for precision and accuracy of the generalized ratio measurement process. Thermal ionization measurements of iodine can be done with both high precision (6) and with high sample utilization efficiency, i.e., ions generated per atoms in the sample (5). The ratio measurement imprecision of RIMS of about 1% that has been demonstrated here is greater than that of TIMS, but still very useful for IDMS measurements. Since the samples are prepared essentially in the same way for RIMS and TIMS, it is probable that the variability is due to the pulsed ionization/measurement process. The advent of new lasers and detection schemes for RIMS promises to reduce this variability. Although the ionization efficiency for RIMS may be high (ions generated per atom in the laser beam), the sample utilization efficiency will remain low when using a pulsed laser with a continuous source of atoms such as a thermal filament. Since a nanogram of iodine represents a 5 X 10l2atoms, this duty cycle loss (a factor of about can be tolerated. The measurement limit for iodine by isotope dilution is presently the 10-ng chemical blank. The single-filament, sample loading procedure used in RIMS is simpler than the triple filament procedure, which requires preparation of a LaB, treated ionization filament. The preparation and maintenance of this ionization filament
389
are critical to both the sensitivity and precision of the TIMS measurement (6). And, as discussed above, source memory should be able to more easily controlled using a single, lowtemperature vaporization procedure. Thus, there are practical advantages for RIMS relative to TIMS. Registry No. I, 7553-56-2.
LITERATURE CITED (1) Fassett, J. D.; Paulsen, P. J. Anal. Chem. 1989, 61, 643A-649A. (2) Heumann, K. G.; Schindlmeier, W. Fresenlus' 2.And. Chem. 1982, 312, 595-599. (3) Heumann, K. G.; Seewald, H. Fresenius' 2.Anal. Chem. 1985, 320, 493-497. (4) Schindlmeier, W.; Heumann, K. G. Fresenius' Z. Anal. Chem. 1985, 320, 745. ( 5 ) Delmore, J. E. Int. J. Mass Spectrom. Ion fhys., 1982, 4 3 , 273-281. (6) Gramlich, J. W.; Murphy, T. J. J. Res. NIST 1989, 9 4 , 215-220. (7) McHugh, J. A.; Stevens, J. F. Anal. Chem. 1972, 4 4 , 2187. (8) Elmore, D.; Phillips, F. M. Science 1987, 236, 543. (9) Paul, M.; Fink, D.; Hollos, G.; Kaufman, A.; Kutschera, W.; Magarltz, M. Nucl. Instrum. Methods Phys. Res. 1987, 6 2 9 , 341-345. (IO) Fassett. J. D.; Walker, R. J.; Travis, J. C.; Ruegg, F. C. Anal. Instrum. 1988, 17, 69-86. (11) Kuehner, E. C.; Akarez, R.; Paulsen, P. J.; Murphy, T. J. And. Chem. 1972, 44, 2050-2056. (12) Fassett, J. D.; Moore, L. J.; Travis, J. C.; Lytle, F. E. Anal. Chem. 1983, 55, 765-770. (13) Fassett, J. D.; Walker, R. J.; Travis, J. C.; Ruegg, F. C. Int. J. Mess Spectrom. Ion Processes, 1987, 75, 111-126. 14) Apel, E. C.; Anderson, J. E.; Estler, R. C.; Nogar, N. S; Miller, C. E. Appi. Opt., 1987, 26, 1045-1050. 15) Clark, C. W.; Fassett, J. D.; Lucatorto, T. B.; Moore, L. J.; Smith, W. W. J. Opt. SOC.Am. B 1985, 2 . 891-896. 16) Moore, L. J.; Fassett, J. D.; Travis, J. C.; Lucatorto, T. B.; Clark, C. W. J. Opt. SOC.Am. 81985, 2 , 1561-1565. 17) Miller, C. M.; Fearey, B. L.; Palmer, B. A.; Nogar, N. S. I n Resonance Ionkatlon Spectroscopy 1988; Lucatorto, T. B., Parks, J. E., Eds.; Insitute of Physics: Bristol, 1989; pp 297-300. (18) Fairbank, W. M. Jr.; Spaar, M. T.; Parks, J. E.; Hutchinson, J. M. R. I n Resonance Ioniretion Spectroscopy 7986; Lucatorto, T. B., Parks, J. E., Eds.; Institute of Physics: Bristol, 1989; pp 293-296. (19) Fassett, J. D.; Powell, L. J.; Moore, L. J. Anal. Chem. 1984, 5 6 , 2228-2233. (20) Walker, R. J.; Fassett, J. D. And. Chem. 1988, 5 8 , 2923-2927. (21) Walker, R. J.; Morgan, J. W. Science 1989, 243. 519-522. (22) Lambert, D. D.; Morgan, J. W.; Walker, R. J.; Shirey, S. B.; Carlson, R. W.; Zientak, M. L.; Koski, M. S. Science 1989. 244, 1169-1174.
RECEIVED for review September 15,1989. Accepted November 20, 1989.