Chelation Preconcentration with Resin Analysis by Direct Sample Insertion Inductively Coupled Plasma Spectrometry ROBIN RATTRAY AND ERIC D. SALIN* Department of Chemistry McGill University 801 Sherbrooke St. West Montreal Quebec H3A 2K6 Canada Batch preconcentration with Chelex-100 followed by direct analysis of the analyte-laden resin by direct sample insertion inductively coupled plasma atomic emission spectrometry (DSI-ICP-AES) is described. The performance of the technique is element specific. Quantitative retention of Cu Zn Cd and Pb on the resin is achieved but only for Cu and Zn does the ratio of the signal before and after preconcentration approach the theoretical preconcentration factor. This observation is mainly caused by the adverse effect of the remnants of the resin after ashing on the excitation properties of the plasma.This is clearly shown by monitoring the ratio of the intensity of a Pb ionic line to a Pb atomic emission line. If the ashing temperature is increased Cd and Pb are prematurely vaporized in the ashing stage which is performed with inductive heating in the graphite DSI probe. Increasing the radiofrequency power sustaining the ICP improves the performance of the technique. Keywords Chelation; preconcentration; resin analysis; direct sample insertion; inductively coupled plasma spectrometry The use of chelating resins for preconcentration and matrix elimination is now popular for lowering detection limits and avoiding interferences in analytical atomic spectrometry. Most applications involve selective retention of the analyte on a resin column followed by elution with an appropriate solution.However the elution step contributes to dilution of the analyte the degree of dilution being dependent on the volume in which the analyte is removed. In addition some resins such as those based on poly(dithiocarbamate) are reluctant to release the bound analyte and the resin is usually dissolved prior to Dissolution is undesirable as it is labour intensive time consuming and a potential source of contamination. Milley and Chatt' directly analysed Chelex-100 resin using instrumental neutron activation analysis to determine the concentration of 15 trace elements in acid rain samples. They mentioned several of the disadvantages associated with analysis involving column elution and suggested that the approach should be generally applicable to other techniques amenable to the analysis of solid samples.Van Berkel and Maessen6 later reported the direct analysis of analyte-laden poly(dithi0- carbamate) resin using electrothermal vaporization (ETV) sample introduction for inductively coupled plasma atomic emission spectrometry (ICP-AES). They found that compro- mise conditions for ashing had to be chosen to avoid vaporiz- ation losses or matrix effects from the pyrolysis products of the resin. Good results were nevertheless obtained for Cu Zn Cd and Pb in a urine certified reference material. However the signal improvement factor achieved was not reported. A commercially available system based on analysis of the resin-bound analyte has been manufactured by CETAC * To whom correspondence should be addressed.I Journal of 1 Analytical 1 Atomic Spectrometry Technologie~.~ In their implementation of the concept sub- micrometre sized resin beads are mixed with the sample to retain the analyte. The mixture is then pumped through a hollow fibre bundle which traps the resin. Finally the resin beads are washed from the fibre bundle to a direct injection nebulizer (DIN) for introduction into the ICP. The system can be operated in a batch mode wherein about 1 ml of mixture can be collected off-line and later analysed. This mode permits the generation of a steady-state signal which may be more convenient to process with most spectrometer systems than a transient signal. Again however the preconcentration factors and therefore detection limits obtained are not optimum because of the dilution that occurs while removing the resin from the fibre bundle.There are other advantages to performing the analysis of the resin-bound analyte. Unlike column-based systems each sample is presented with a fresh portion of resin which makes resin regeneration unnecessary and avoids memory effects from the resin. Also by eliminating the elution step it may be easier to design more selective resins since only the chemistry of retention as compared with retention and elution is of concern. Direct sample insertion (DS18v9) sample introduction for ICP spectrometry seemed to be a suitable technology for direct analysis of the resin from a chelation preconcentration." The graphite probes commonly used can easily accommodate the milligram amounts of resin that would be used and the resin can be conveniently dried and ashed inductively in situ using methodology and instrumentation developed previously." The sample introduction efficiency for DSI is 100% so any precon- centration advantage procured with the resin should be pre- served.Based on these premises the investigations described in this paper were carried out. EXPERIMENTAL The Plasma Therm ICP and the modified Thermo Jarrell Ash spectrometer that were used have been described previously.'2 Briefly the modifications to the spectrometer included the installation of a galvanically driven quartz refractor plate behind the entrance slit for rapid background correction and a diode laser at the zero order for experimental alignment.I3 The stepper motor-driven DSI device that was used was built and operated with essentially the same characteristics as that described previ0us1y.l~ The temperature of the graphite DSI probe during the ashing step was measured with a Pyro Micro- Optical Pyrometer (Pyrometer Instrument Northvale NJ USA).Table 1 lists some of the important DSI ICP and data acquisition parameters that were used in this study. The data acquisition software used (SF20) was written by G. Leghe; an earlier version of this program with many of the salient capabilities is commercially available (Trulogic Systems Oakville Ontario Canada). The elements monitored were Cu Zn Cd and Pb at 324.7,213.8,228.8 and 220.3 nm respectively. Data processing was carried out with user-written programs Journal of Analytical Atomic Spectrometry December 1995 Vol.10 1053Table 1 DSI ICP and data acquisition parameters Insertion distance Insertion time Radiofrequency (r.f.) power Reflected power Viewing height Plasma gas flow rate Auxiliary gas flow rate Integration time per point Background measurement position 3 mm above TOLC* 20 s 1.0-1.75 kW <low 17 mm above TOI,C* 12-16 1 min-l 1.8-2.0 1 min-' 20 ms -0.1 nm below peak * TOLC = top of load coil. Insertion distance measured relative to top of DSI probe. and Lab Calc (Galactic Industries Salem NH USA). The peak area of the background corrected signal was used exclusively for quantification. Multi-element solutions approximately 1 YO in nitric acid were prepared by dilution of 1000ppm atomic absorption standards (Fisher Scientific) with 18 M a deionized distilled water obtained from a Milli-Q water purification system ( Millipore Bedford MA USA).Working solutions were freshly prepared each day. The preconcentration procedure used was as follows 1.0 ml of the multi-element solution contained in a 1.5 ml polypropylene flip-top microcentrifuge tube was adjusted to pH 9 by the addition of 75 pl of 2.0 mol 1-1 aqueous ammonia. The Chelex-100 resin (100-200 mesh) (Bio- Rad Laboratories Richmond CA USA) was converted into the ammonium form as described by Marino and 1ngle.l" A 100 pl aliquot of the resin slurry (about 7.5 mg of dry resin) was taken with an Eppendorf micropipette the tip of which had been cut off to enlarge the orifice and allow the resin to be able to pass through.This aliquot of resin slurry was added to the standard solution and mechanically agitated for 30 min. This time was found to be insufficient for complete extraction of Pb and was later increased to 3 h. The supernatant (100 pl) was analysed by DSI-ICP-AES to determine the extent of completion of the extraction process. A nitric acid blank M YO) was treated in an identical fashion. Resin samples were dried and ashed in the graphite DSI probe using the induction field in the load coil of the IlCP according to the protocol in Table 2. Two consecutive drying steps were necessary to prevent the slurry from boiling violently and being ejected from the probe. Two arshing schemes as described in Table2 termed 'low temperature' and 'high temperature' were initially studied.During the 'low- temperature' ashing step the graphite DSI probe barely glowed to a dull red colour suggesting a temperature of approximately 650 "C. This temperature is only approximate since the lowest temperature that could be measured with the optical pyrometer used was 700°C. However a good correlation was obtained between the net rf power applied (forward minus reflected) and the measured probe temperature; interpolation of that curve also indicated that the lowest ashing temperature used was about 650°C. If milder ashing conditions were used the plasma would be extinguished on insertion of the probe. Under the 'high-temperature' conditions the temperature was meas- ured as 815 "C. Two higher ashing temperatures (950 and Table 2 Drying and ashing protocol Forward Reflected Stage power/W power/W Duration/s Drying 1 25 8 30 Drying 2 50 16 60 Ashing (815 "C) 140 50 60 Ashing ( % 650 "C) 70 25 90 1080 "C) were also investigated later. The aqueous samples did not require ashing and were only dried as described in Table 2.RESULTS AND DISCUSSION Calibration at 1 kW The system was calibrated at 1 kW with 50 pl aliquots of aqueous Standards. Linear calibration graphs with log-log slopes of unity were obtained from 10 ppb (1 ng) to 1 ppm (100 ng). The signal produced by Cu at 1 ppm (100 ng) just saturated the detector; this upper limit was later extended by modifying the timing parameters in the SF20 data acquisition software. Detection limits (3s) were calculated from the slopes of these graphs and an estimate of the blank noise.These limits obtained at 1 kW were 0.2 0.2 0.6 and 0.6 ppb for Cu Zn Cd and Pb respectively. The ICP was initially operated at 1.0kW to generate the following data. However it was later realized that it was more prudent to use higher powers to avoid any problems with memory effects in the probe especially for Cu the least volatile of the four elements. The probe was also 'burned' in the plasma between runs to detect and remove any residual analyte. Analysis of the supernatant from an extraction of 1 ml of a 100 ppb multielement solution revealed that quantitative extraction was being achieved for all elements except for Pb. As mentioned before these initial experiments used a relatively short extraction time (30 min).By increasing the extraction time to 3 h all the elements were quantitatively removed from solution suggesting that the kinetics of the extraction for Pb are slower than for the other elements. However when the resin was transferred to the DSI probe and analysed lower values than expected (1 m1/100 pl= 10) for the signal improve- ment factor were observed. In fact for Cd and Pb the signal was suppressed to or even below the level seen with a 100 p1 aliquot of the same solution (Table 3). Further experiments were done to determine the cause of these apparent discrepancies. Possible sources of the losses observed included premature volatilization and/or physical ejection of the resin from the probe on insertion. However when aqueous standards were subjected to the 'low- temperature' ashing protocol applied to the resin analysis no significant losses were observed even for Cd the most volatile element studied.When ashed at 815°C about 70% of the Cd and 20% of the Pb were lost but Cu and Zn were not appreciably affected. Fig. 1 summarizes the effect of ashing temperature on the signal obtained from a 100 pl aliquot of a solution 50 ppb in Cu and Zn and 100 ppb in Cd and Pb. When the resin ashed at 650 "C was inserted into the plasma qualitative changes in the ICP were seen. An intense green flash from the top of the probe of about 1 s duration was Table 3 Preconcentration with Chelex-100-DSI-ICP-AES ( 1 kW) average peak area (n = 3) (arbitrary units) CU Zn Cd Pb Solution analysis- 100 p1 100 ppb 20 214 3185 3638 568 100 pl supernatant nd* nd 36 130 from 100 ppb extraction Extraction (%) 100 100 99 75 Analysis of resin- 1 ml 100 ppb 136 100 12 758 3893 306 Blank 450 206 nd nd Signal improvement 6.5 3.7 1 0.6 factor * nd =No peak detected.1054 Journal of Analytical Atomic Spectrometry December 1995 Vol. 101.2 1 1 600 700 800 900 I000 1100 Ashing temperaturd'c Fig. 1 Variation of ashing temperature (aqueous standards). A Cu; B Pb; C Cd; D Zn followed by a pale orange glow which lasted several seconds further up in the 'tail' of the plasma. These emissions were due to products of resin pyrolysis; they were also observed when the blank resin but not the aqueous standards were analysed. For example the green flash was almost certainly due to emission from the C2 Swan bands arising from components of the hydrocarbon skeleton of the resin.At higher ashing tem- peratures this green emission was not observed suggesting that the concentration of these species was dramatically reduced or eliminated. The following experiments were done to test the hypothesis that these species were adversely affecting the excitation properties of the plasma. The pure resin was first ashed (65OOC) in the probe as described before. A 100 pl spike of the 100 ppb standard multi- element solution was then dried on top of the resin ash and the probe was inserted into the ICP. The response obtained was compared with the response to the standard solution alone and the response from the pure resin. These experiments were performed in triplicate; the averaged results are presented in Table 4. The results in Table4 and the previous experiments imply that the presence of the resin ash was suppressing the analyte signal and that the losses were not due to premature volatiliz- ation during the ashing stage.In addition at the Cd I Pb I1 and Zn I channels the background (off-line) signal intensity from the ashed resin was essentially the same as from the aqueous standards so the decrease in sensitivity was not due to a change in spectral characteristics. However at longer wavelengths (e.g. in the Cu I channel at 324.8 nm) a 'dip' in the background was seen which suggests that the plasma temperature has probably changed. This is similar to the well known 'pressure pulse' phenomenon observed after vaporization in many ETV-ICP-AES system^.'^.^^ These observations reinforce the hypothesis that the decrease in signal is at least partially due to the effect of the products of pyrolysis of the resin on the excitation properties of the ICP.This hypothesis is supported further by the fact that the severity of the interference observed increases in the same order as the excitation or ionization potential for the analytical lines that were used (Table 4). Similar conclusions were drawn by the previous study using electrothermal vaporization and the poly(dithi0carbamate) resin6 During these experiments (at 1 kW) it was noted that the background-corrected signal for Cu did not return to the baseline and there was a slight memory effect. Also the peak shapes for Zn were ragged although they were much sharper with the resin ash present.Because of these observations no further experiments were carried out at 1 kW. Variation of RF Power In an attempt to confirm and overcome the matrix interference a series of experiments were performed at various rf power levels. At each level the response to 1OOpl of a 100 ppb solution (10 ng of analyte) was determined. As expected an increase in sensitivity was observed at elevated power (Fig. 2). The ratio of the response of these aqueous solutions (100 pl) to the response of the resin extract of 1 ml of the same solution (treated as described under Experimental) was calculated. While improvements in this ratio were generally observed as the rf power level was raised (Fig. 3) the signal improvement was still well below the theoretical preconcentration factor of 10 particularly for Cd and Pb.These graphs represent only one experiment per data point which may explain the erratic trend of the Pb curve. The trend of higher ratios at high power is however applicable to all elements. Variation of Plasma Observation Height A short study of the effect of varying the position in the plasma that was viewed by the spectrometer was carried out to find out if the interferences could be spatially resolved. The ICP was operated at 1.25 kW and the results are summarized in Fig. 4. The emission intensities decreased as the viewing height increased (Fig. 4) and there was no appreciable change in the ratio of the signals from aqueous standard and analyte-loaded resin. Qualitatively however the signals seem to be smoother and more reproducible when measured at 25mm above the top of the load coil.As a result of the previous experiments the calibration with aqueous standards was later repeated at 1.5 kW in order to enhance the removal of the analyte from the probe and to reduce the matrix effects which appeared to be less severe at higher power. Again the graphs were linear with log-log slopes near unity. Detection limits were not calculated but the slopes of these graphs are four times greater for Cd and Zn and two times greater for Cu and Pb than at 1.0 kW. Variation of Ashing Temperature A high ashing temperature is desirable to remove as much of the interfering resin pyrolysis products from the probe as possible. The effect of ashing at a higher temperature was therefore examined.A multi-element solution (1 ml) 5 ppb in Cu and Zn and 10 ppb in Cd and Pb was extracted with Chelex-100 as described under Experimental and the resin was ashed at both 650 and 815°C. A 1% nitric acid blank was also taken but the resin from extraction of this blank was only ashed at the lower temperature to minimize loss of Cd or Pb. The experiments were carried out in triplicate at 1.5 kW and the results (averaged) are presented in Table 5. Table 4 Effect of resin ash on analyte signal (1 kW). Signal (peak area) is the average of three measurements (arbitrary units) Pure 100 p1 Resin ash+ Suppression Excitation Species resin 100 ppb 100 p1 100 ppb ("/) potentiallev CU' 200 20 214 16 159 20 3.82 Cd' nd* 3638 1594 56 5.42 Zn' nd 3185 459 86 5.79 Pb" nd 568 nd 100 7.42 * nd =No peak detected.Journal of Analytical Atomic Spectrometry December 1995 Vol. 10 1055v) 0 0 0 3 25- u * - . - - 1 1.2 1.4 1.6 1.8 v) 0 0 51 1100 1000 s 900 I 800 700 a a 600 Power/W . - . . . . . . . 1 1.2 1.4 1.6 1.8 Fig. 2 Effect of rf power on signal from liquid standards (a) Cu; (b) Zn; (c) Cd; and ( d ) Pb 0.6 0.5 0.4 0.3 0.2 3.5 3 2.5 2 1.5 1.4 1.3 1,2 1.1 1 1.2 1.4 1.6 1.8 1 1.2 1.4 1.6 1.8 PowerkW Fig. 3 Effect of variation of rf power on intensity ratio (a) Cu; (b) Zn; (c) Cd; and (d) Pb 300 i I *-*--I- v) 0 2 50 1 1 " I t\ I % 500 &-- 16 18 20 22 24 26 16 18 20 22 24 26 Viewing heighvmm above TOLC Fig. 4 Viewing height variation. A Standard and B resin 1056 Journal of Analytical Atomic Spectrometry December 1995 Vol.10Table 5 Effect of ashing temperature (1.5 kW). Values are averages of three measurements peak area in arbitrary units Table 7 Characteristics of species used and experimental conditions in intensity ratio study 650°C ash 815 "C ash Blank (650 "C ash) 100 pl 50 ppb (Cu Zn) 100 pl 100 ppb (Cd Pb) Signal ratio (650 "C) Signal ratio (8 15 "C)t RSD (Yo) RSD (Yo) c u 21 011 15 22 073 2 2674 2457 7.5 7.9 Zn 7450 17 8751 6 1505 808 7.4 9.0 Cd 936 7 nd* nd nd 1499 0.62 0 Pb 211 23 243 3 nd 137 1.5 1.8 * nd = No peak detected. t Blank-extracted resin ashed at 650 "C used for blank correction. Qualitative differences were observed in the response to the two ashing methods. For Cu when the resin was ashed at the lower temperature a 'spike' in the signal was always observed 0.8 s after probe insertion and 2.0 s before the main peak.After the higher temperature ash this feature was either greatly diminished or absent from the signal profile (Fig. 5). The spike is probably due to the physical ejection of some analyte with the sudden vaporization of resin pyrolysis prod- ucts that still remain in the probe after a low-temperature ash. It is likely that all elements are being affected but the effect is obvious for Cu despite the short residence time in the plasma owing to its low excitation potential (Table 4). The low vola- tility of this element also allows good temporal resolution of this effect from the main analyte peak. A comparison of the peak areas suggests that the loss due to physical ejection is about 5%.It can be seen from Table 5 that quantitatively Cu Zn and Pb showed only slight improvements after ashing at a higher temperature suggesting that the interfering species were still not completely removed in the ashing process. For Cu under low-temperature ashing conditions the total peak area (i.e. spike plus main analyte signal) was used. Cd was completely lost when ashed at 815°C but was seen under the milder conditions. Also noteworthy is the fact that the precision of the measurements was much better under higher temperature ashing conditions possibly because of the absence of the disruptive effect of the rapid volatilization of resin pyrolysis products as the probe entered the plasma. Several strategies were adopted in order to monitor the entry of the resin pyrolysis products into the plasma.The emission signal from the C I line at 193.1 nm was monitored during both ashing methods to determine the extent of the hydrocarbon load (in the form of C) in the plasma. However neither this line nor the H I line at 486.1 nm proved to be particularly useful as the signals tended to plateau after a low resin load (2 mg) in the DSI probe. The ratio of the intensity of emission lines arising from ionic and atomic species of the same element has been used as a diagnostic tool in AES.'*-'' Previous workers with resin analy- sis by ETV-ICP-AES used the Cu I1 224.7 nm:Cu I 324.7 nm ratio to verify that the excitation conditions were in fact changing. The Cu I1 line at 224.7 nm was not available with our spectrometer but another Cu I1 line at 213.6nm was C u I CuII P b I P b I I Wavelength/nm 324.8 213.6 405.8 220.4 Ionization po tential/eV na* 7.7 na 7.4 Excitation potential/eV 3.8 8.5 4.4 7.3 Energy sumt/eV 3.8 16.2 4.4 14.7 Ashing conditions- Forward power/W 70 140 200 260 Temperature/" C - 650 815 9 50 1080 Reflected power/W 25 50 75 100 * na =Not applicable. t Ionization potential plus excitation potential?' Table 6 Effect of ashing temperature on Cu atomic line ionic line intensity ratio accessible in the existing Zn I channel by adjusting the position of the refractor plate at the exit slit assembly.The 'energy sums' (ionization energy plus excitation energy") of these two lines are comparable (224.7 nm= 15.9 eV; 213.6 nm= 16.2 eV22) and similar results to the previous work with ETV- ICP-AES were expected.However when the measurements were made with DSI-ICP-AES there was no difference between the Cu 1:Cu I1 intensity ratio calculated from aqueous stan- dards (100 ply 50 ppb) alone in the presence of the resin ashed at 650 or 815 "C (Table 6). This apparent discrepancy can be explained by considering the vaporization characteristics of the species of interest. The appearance time of the peak maximum is a measure of the volatilization rate and is 0.1 1.5 1.9 and 2.8 s after probe insertion for Cd Pb Zn and Cu respectively. However the pyrolysis products are rapidly volatilized approximately 1 s after insertion based on visual inspection and the 'spike' observed in the Cu signal profile (Fig. 5). This suggests that the elements of higher volatility that co-vaporize with the resin breakdown products would be more prone to detrimental effects of varying plasma excitation conditions.Cu on the other hand vaporizes much later and is therefore at least partially temporally resolved from this interference. A channel for a PbI emission line (405.8 nm Table 7) was therefore installed in the spectrometer and the experiment repeated to compare the behaviours of these two elements of differing volatilization properties. A series of ashing temperatures were used to note more carefully any trends in performance. Fig. 6 clearly shows that as the ashing temperature was increased the Pb line intensity ratio increased dramatically (approximately six-fold) whereas the Cu line intensity ratio remained essentially constant.These data confirm and explain the experimental data presented earlier. CONCLUSIONS Direct analysis of the resin from a Chelex-100 type preconcen- tration by DSI-ICP-AES does not seem universally applicable to trace analysis because of the matrix effects of the resin ash on the plasma excitation conditions. Elements of lower vola- tility are affected less because they are temporally separated from the interfering matrix which volatilizes shortly after insertion. Some compensation can be achieved by working at high rf power but variation of the viewing height does not appear to be beneficial. Using a high ashing temperature Cu I peak area* Cu I1 peak area* Ratio Cu I Cu I1 Aqueous standard (dried only) 25 450 2120 12.0 Aqueous standard + resin (650 "C ash) 22 630 1880 12.0 Aqueous standard + resin (8 15 "C ash) 26 010 2170 12.0 * Peak areas in arbitrary units.Journal of Analytical Atomic Spectrometry December 1995 Vol. 10 10576000 4000 2000 0 0 .- c E 20 - ----On-peak sign& 0 2 4 6 8 10 12 14 16 Time/s / Fig. 5 Effect of ashing temperature on Cu signal from loaded resin (1.5 kW) (a) approximately 650 "C and (b) 815 "C results in losses of all but the least volatile elements. The pyrolysis products of other resins or solid phases used in preconcentration may not have as detrimental an effect and may be more suitable to this application. Although not investi- gated here the beneficial aspects of the elimination of the sample matrix may still make the technique feasible for the analysis of certain sample types.Also worthy of investigation may be the use of larger sample volumes to provide higher preconcentration factors since the resin capacity is high enough for the sample volume to be increased by at least an order of magnitude with an expected signal improvement of approxi- mately two orders of magnitude to be expected for less vola.tile elements such as Fe Mn Co and Ni. This work was supported by the Ontario Ministry of the Environment and Energy (Project 5746) and was also facili- tated by the assistance of Fisons Instruments. R.R. gratefully acknowledges financial support from the University of the West Indies and the Canadian International Development Agency. 25 30 k c .- ---a *!15 j 10 fi 3 ,/ f l r ; 600 700 800 900 1000 1100 Ashing temperat urd0C Fig.6 Pb I1:Pb I; and B Cu 1:Cu I1 Variation of intensity ratios with ashing temperature A REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 Murthy R. S. S. Horvath Z. and Barnes R. M. J. Anal. At. Spectrom. 1986 1 269. Barnes R. M. and Genna J. S. Anal. Chem. 1979,51 1065. Barnes R. M. Fodor P. Inagaki K. and Fodor M. Spectrochim. Acta Part B 1983 38 245. Fodor P. and Barnes R. M. Spectrochim. Acta Part B 1983 38 229. Milley J. E. and Chatt A. J. Radioanal. Nucl. Chem. Articles 1987 110 345. Van Berkel W. W. and Maessen F. J. M. J. Spectrochim. Acta Part B 1988 43 1337. Smith F. Wiedierin D. and Gjerde D. Winter Conference on Plasma Spectrochemistry Sun Diego January 1994 Paper 131 0. Salin E. D. and Horlick G. Anal. Chem. 1979 51 2284. Karanassios V. and Horlick G. Spectrochim. Acta Rev. 1990 13 89. RbiiEka .I. personal communication. Rattrdy R. Minoso J. and Salin E. D. J. Anal. At. Spectrom. 1993 8 1031. Legere G. and Burgener P. ZCP ZnJ Newsl. 1988 13 521. Ren J. M. Legere G. and Salin E. D. Appl. Spectrosc. 1993 47 1953. Sing R. L. A. and Salin E. D. Anal. Chem. 1989 61 163. Marino D. F. and Ingle J. Anal. Chem. 1981 53 292. Gunn A. M. Millard D. L. and Kirkbright G. F. Analyst 1978 103 1066. Ng K. C. and Caruso J. A. Anal. Chim. Acta 1982 143 209. Boumans P. W. J. M. Theory of Spectrochemical Excitation Plenum Press New York 1966. Mermet J. M. Spectrochim. Acta Part B 1989 44 1109. Carre M. Poussel E. and Mermet J. M. J. Anal. At. Spectrom. 1992 7 791. Murillo M. and Mermet J. M. Spectrochim. Acta Part B 1987 42 1151. Robinson J. W. Handbook of Spectroscopy CRC Press Cleveland OH 1974. Paper 51008285 Received February 2 1995 Accepted July 24 1995 1 058 Journal of Analytical Atomic Spectrometry December 1995 Vol. I0