JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1991 VOL. 6 159 Correction of Mineral Acid Interferences in Optical Emission Spectrometry on Copper Standardization Louise M. Garden* ICI Advanced Materials Engineering Compounds and Polymers 8JE UK Inductively Coupled Plasma and Manganese Using Internal Group Wilton Middlesbrough Cleveland TS6 John Marshall ICI plc Wilton Materials Research Centre Wilton Middlesbrough Cleveland TS6 8JE UK David Littlejohn Department of Pure and Applied Chemistry University of Strathclyde Cathedral Street Glasgow G 1 IXL UK A study has been made of mineral acid matrix interferences in trace element determinations by inductively coupled plasma optical emission spectrometry. It has been shown that signal suppressions of up to 40% can be observed under normal operating conditions.The interferences can largely be compensated for by using an internal stan- dardization procedure. Precision is also improved several fold when scandium is used as a real-time internal stan- dard. This is attributed to the reduction of noise by simultaneous measurement of emission signals. Keywords Inductively coupled plasma optical emission spectrometry; Myers-Tracy signal compensation; internal standardization; acid interference effect Although interferences found in inductively coupled plasma optical emission spectrometry (ICP-OES) are in general less severe than in competitive techniques such as flame atomic ab- sorption or electrothermal atomic absorption spectrometry significant matrix effects have been reported. l 4 Inorganic and organic acids are commonly used for sample preparation e.g.extraction digestion or dissolution and consequently are major components of the sample matrix in many routine ana- lytical situations. One of the first reports of interferences caused by acids was published by Greenfield et The inter- ferences found were largely attributed to differences in the transport of the acid-containing solutions compared with those of aqueous solutions into the plasma. Farino et d.,6 in a more extensive study of the effects of mineral acids suggested that the observed suppression of the analyte emission was caused by differences in the mass of analyte entering the plasma. The change in analyte mass transport into the plasma was attri- buted to differences in the viscosity between the acid and the aqueous solutions.Viscosity is known to affect directly primary droplet formation by the nebulizer as indicated by the viscosity term which appears in the Nukiyama and Tanasawa equation.' Viscosity also affects droplet formation by changing the solution uptake rate of the nebulizer. The effects of organic acids have also been and have generally been found to cause an increase in the analyte emission intensity. This effect has been attributed to increased nebulization efficiency and increased plasma temperature.' There are several methods of compensating for or removing the effects of acids during an analysis. The most successful and most widely utilized approaches for routine analysis involve the use of matrix matched calibration standards and the standard additions method.I0 Generic protocols of this type can be applied to most problems associated with analyses in acid matrices.However alternatives such as measurement of the hydrogen line (HP) at 486.133 nm to compensate for the effects of varying acid concentrations on the introduction of the sample into the plasma" and the use of mathematical models based on regression analysis have also been explored." Internal standardization has been used to improve the per- formance of ICP-OES by (indirectly) reducing plasma noise." This method has also been used to reduce matrix interferences. Two specific developments which have been used successfully for this purpose are the generalized internal reference methodr4 and the parameter related internal standard method.l5.lh Both of these methods use two or more internal standards to correlate the analyte emission with various plasma fluctuations.However other workers have shown that the use of a single in- ternal standard might be applicable in certain situations. Dal- quist and Knoll" used cobalt as an internal standard to compensate for acid interferences. Uchida et al." used yttrium as an internal standard with a micro-sampling technique to compensate for changes in the amount of sample fed to the plasma. The choice of the internal standard for use during analysis has been the subject of extensive study. Barnett et ~ l . l ' . ~ ~ ' pro- posed guidelines for matching the physical properties of the analyte and reference element so that the ratio of the emission intensity from the analyte to the emission intensity from the in- ternal standard was insensitive to slight fluctuations in the plasma parameters.Sedcole et al." found that no one element was suitable for correcting interferences on all analytes. The importance of the purity and solubility of the internal standard was highlighted in a study by Wallace,'! who also studied pos- sible spectral interferences by the internal standard on the analyte wavelengths. In investigating real-time internal standardization (i-e. si- multaneous measurement of the analyte and internal standard intensity) Belchamber and Horlick23 found that it was possible to correlate the emission from the internal standard with the emission from the analyte element. Scandium was shown to be a promising internal standard (i.e. there was good correlation between the scandium emission and the emission from several of the analytes) compared with the other elements studied.It was found that measurement precision was generally improved by a factor of two using real-time internal standardization. In a similar correlation study Myers and Tracy" used manganese as an internal standard while studying the noise behaviour of 20 analytically important elements. Carrier gas flow-rate and observation height were found to change significantly the degree of correlation between the emission from the analyte element and the internal standard. Hence by altering plasma conditions a single internal standard could be used to improve analytical performance. The plasma background emission in the range 200-700 nm was also studied.It was found that the160 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1991 VOL. 6 plasma background showed sensitivity to the fluctuations in the r.f. power and the carrier gas flow-rate. However fluctuations in the analyte emission and the background emis- sion were generally uncorrelated. Schmidt and SlavinI3 pursued the reaktime internal stan- dardization approach using scandium as an internal standard. In addition to sampling the analyte and internal standard emis- sion in real time the background intensity was also continous- ly measured. Prior to calculating the internal standard ratio the background emission intensity was subtracted from both the analyte and internal standard emission intensities.The net in- tensities were then used to determine the internal standard ratio. This procedure has been referred to as Myers-Tracy signal compensation (MTSC)zS-z6 and has been used to com- pensate successfully for the interference effects caused when solutions of NaCII3 and KCIzs with a high content of dissolved solids have been nebulized. It has also been shownzs that by using MTSC there can be considerable improvements in ana- lytical precision for elements which exhibit different behavi- ours in the ICP. It would appear from a survey of the literature that the in- ternal standardization approach may offer the option of an aqueous standard calibration strategy for the examination of samples containing acids in conjunction with improved meas- urement precision.The application of the MTSC procedure to the ICP-OES determination of copper and manganese in acidic media is described in the present paper. Experimental Instrumentation A Perkin-Elmer Plasma I1 sequential ICP-OES system was employed for all measurements. The spectrometer has a focal length of 1 .O m and spectral resolution of 0.018 nm. The stan- dard operating parameters for the spectrometer can be found elsewhere.2S The plasma conditions used throughout unless otherwise stated are given in Table 1. The de-mountable plasma torch unitz7 was fitted with a 2 mm i.d. acid resistant alumina injector. The sample introduction system consisted of a Ryton crossflow nebulizer and associated double-pass spray chamber. Samples were introduced into the nebulizer by means of the manufacturer’s peristaltic pump.An integration time of 1 s was employed for all measurements. Precision data were obtained from ten replicate integrations. The mode of op- eration of the MTSC assembly has been described in detail previ~usly.~~ All calculations were performed on a Perkin- Elmer 7500 Series computer. Reagents Manganese and copper standard solutions were freshly prepared as required from stock solutions containing 1000 pg ml-1 of the appropriate element (SpectrosoL BDH Poole Dorset UK). A 1000 pg ml-I stock solution was prepared by dissolving the ap- propriate amount of scandium oxide (BDH laboratory-reagent grade 99.9%) in hydrochloric acid. Aristar grade acids were used in the preparation of all ‘interferent’ solutions. Deionized distilled water was used as a diluent and blanks of all the rea- gents were monitored during the study for contamination.Table 1 Standard plasma conditions Plasma parameter Value 1 .oo 1 .oo Plasma gas flow-ratefi min-’ 15.0 Observation height (above induction coil)/mm 15 Carrier gas flow-rate/l min-’ Intermediate gas flow-rate/l min-l Applied power/kW 1 .o Table 2 Optimum plasma conditions for manganese and copper Plasma parameter Mn (257.6 nm) Cu (324.8 nm) Carrier gas flow-ratel1 min-’ 1 S O 1.30 Plasma gas flow-ratefl min-’ 15.0 15.0 Observation height (above induc- tion coil)/mrn 12 18 Applied power/kW 0.80 0.80 Intermediate gas flow-rate/l min-’ 1 .oo 1 .oo Internal Standard A concentration of 20 pg ml-l of scandium was used in all solu- tions because this level of internal standard had been previously shown to be the minimum required to achieve adequate preci- sion and to avoid spectral interferences at analyte wavelengths.” Results and Discussion Aqueous Solutions It was of interest initially to determine whether the MTSC pro- cedure improved the performance of the Perkin-Elmer Plasma I1 instrument for determinations in aqueous solutions. Calibra- tion graphs for manganese and copper were obtained while using the optimum conditions for each element both with and without signal compensation.The optimum conditions for both the manganese ion line at 257.6 nm and the copper atom line at 324.8 nm were obtained by univariate optimization. They are given in Table 2. Four solutions were used to obtain the calibration graphs for manganese.These contained 0 1.0 5.0 and 10.0 pg ml-’ of manganese and 20 pg ml-I of scandium (internal standard). The solutions were used for both calibration graphs i.e. with and without MTSC. This ensured that any differences ob- served in the calibration graphs were not caused by using dif- ferent solution concentrations but resulted from the different signal processing procedures. Background correction was em- ployed. Calibration graphs were constructed for manganese from data obtained with and without the use of MTSC. The results of the calibration for manganese including values of the relative standard deviation (RSD) of the manga- nese intensities and the manganese ratios are shown in Table 3. Linear regression was used to fit the best line through the points on the calibration graphs.The values of the regression coefficient (Y) standard error of the slope and standard error of the intercept for each line are given in Table 3. It was noticeable that when MTSC was used the scatter of data points about the regression line was significantly reduced in comparison with conventional measurement. It was also noted that the measurement precision improved by more than a factor of two when MTSC was used. It was important to verify that the observed improvement in instrumental performance using the MTSC procedure for man- ganese was not only true of this particular element but would also be observed for analytes with different excitation charac- teristics. Copper I (324.8 nm) is considered a typical ‘soft’ line and as such requires different excitation conditions from man- ganese.’H Cali bration graphs were therefore obtained for copper using a similar procedure to that described for manganese. The emission intensities emission ratios and RSDs for each meas- urement are presented in Table 4.The precision of the measure- ments was again found to be improved generally by a factor of two when MTSC was employed. Acid Solutions Manganese and copper (wavelengths 257.6 and 324.8 nm re- spectively) were chosen as representative analyte lines for the study of acid interferences. The effects of hydrochloric nitricJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1991 VOL. 6 161 Table 3 Calibration data for manganese with and without internal stan- Table 5 Effect of hydrochloric acid on recovery of manganese with and without internal standardization dardizat ion Manganese Emission concentration/ intensity pg ml-' 0 0 I .00 1 4 902.37 5.00 70204.77 10.00 127488.35 Line of best fit- A.From intensities Intercept Slope of line Standard error (slope) Standard error (intercept) I' RSD Emission RSD (intensity) ratio (ratio) (%I (%) 0 2.5 12.34 1 . 1 2.8 6 1.45 0.7 1.5 125.83 0.7 - - B. From ratios 2482.13 Intercept -0.77 12748.84 Slope of line 12.62 0.9986 I' 0.9999 473.7 I Standard error (slope) 0. I70 Standard error 2 658.7 1 (intercept) 1.102 Table 4 Calibration data for copper with and without internal standardi- zation Copper Emission RSD Emission RSD concentration/ intensity (intensity) ratio (ratio) pg ml-' (%) (%) - - 0 0 0 I .00 7 339.69 1.8 6.36 0.7 5 .OO 36091.13 2.5 3 1.90 0.4 10.00 7 1 894.25 3.6 62.64 0.4 50.00 352403.43 3.0 3 18.76 1.6 Line of best fit- A.From intensities B. From ratios Intercept 2 132.56 Intercept -0.35 Slope of line 7039.49 Slope of line 6.248 I' 0.99999 r 0.99994 Standard error (slope) 16.14 Standard error (slope) 0.0171 Standard error Standard error (intercept) 369.86 (intercept) 0.4384 and sulphuric acids were investigated. Standard plasma condi- tions were used. Hydrochloric acid A number of solutions were used containing 10 pg ml-' of manganese 10 pg ml-' of copper and 20 pg ml-' of scandium with hydrochloric acid concentrations of from 0 to 20% (v/v). Analyte signal recovery was calculated as a percentage of the response obtained for aqueous analyte solution.The results of the study for manganese are presented in Table 5. From these results it can be seen that when MTSC was not used the recovery of manganese from hydrochloric acid was about 90% of the total manganese content. If MTSC was used then the manganese recovery was increased to ap- proximately 98% of the total an improvement in recovery of 8%. It was also apparent that without MTSC the precision of the measurements was approximately 2% whereas with MTSC the precision was improved by a factor of five to give values of 0.4%. It is clear that the procedure involving inter- nal standardization resulted in a considerable improvement in both manganese signal recovery and in measurement preci- sion. However it was noted that although values of recovery increased to 98% of the expected value there was still a deficit of 2%.This deficit may be caused by some additional factor which cannot be compensated for by using the internal standardization method or alternatively may be a reflection of Without MTSC With MTSC Acid concentration Recovery RSD* Recovery RSD* (% v/v) (%) (%) (%) 0 100.0 2.2 100.0 0.5 5 91.0 2 .o 97.3 0.4 10 89.2 2.3 97.7 0.3 15 90.4 2.3 97.8 0.4 20 90.5 2.4 98.6 0.5 * The RSD quoted is that of the manganese emission intensity not the RSD of the recovery. ~ ~~~~ the measurement error in the internal standardization proce- dure. A similar series of measurements was also made for copper. The measurement procedure and the solutions used were the same as those in the previous section. The results of the inves- tigation are tabulated in Table 6.As observed for manganese without MTSC there was a considerable decrease in the copper signal recovery when the acid solutions were nebulized. With copper however it was observed that as the acid concentra- tion increased the percentage recovery decreased unlike man- ganese where there was no noticeable trend in recovery with increasing acid concentration. It was observed that the copper signal recovery was significantly decreased when even small acid concentrations were added to the aqueous solution and that the addition of larger acid concentrations did not cause further significant decreases in copper recovery. When signal compensation was used the copper recovery was approximate- ly 100%. Thus MTSC compensated fully for the interference effect of hydrochloric acid on copper.The measurement preci- sion for copper was improved by about a factor of four when using MTSC. Nitric acid The effect of nitric acid on manganese emission was evaluated using a similar procedure to that described above. The nitric acid was added in concentrations ranging from 0 to 20% (v/v). Manganese copper and scandium concentrations were 10 10 and 20 pg ml-I respectively. The results obtained with and without signal compensation are shown in Table 7. The interference effect caused by nitric acid on manganese emission was of approximately the same order of magnitude as that of hydrochloric acid. When internal standardization was used the recovery of manganese from nitric acid was increased by 8% relative to the recovery obtained without MTSC.This improvement was similar to that obtained when MTSC was applied to the measurement of signal recovery for manganese in hydrochloric acid solutions. Precision of measurement was im- proved by about a factor of six using internal standardization. ~~~ Table 6 without internal standardization Effect of hydrochloric acid on recovery of copper with and Without MTSC With MTSC Acid concentration Recovery RSD* Recovery RSD* (% v/v) ( % I (%I (%) (% 1 0 100.0 I .s 100.0 0.4 5 92.6 I .4 100.4 0.4 10 89.9 1.2 101.3 0.4 IS 89.4 1.4 100.9 0.3 20 89.2 I .0 101.3 0.3 * The RSD quoted is that of the copper emission intensity. not the RSD of the recovery.162 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 199 1 VOL. 6 Table 7 internal standardization Effect of nitric acid on recovery of manganese with and without Without MTSC With MTSC Acid concentration Recovery RSD* Recovery RSD* (9% v/v) (%I (%I (%) 0 100.0 1.8 100.0 0.7 5 90.8 1.4 98.8 0.3 10 90.1 2.5 97.9 0.3 15 89.5 2.1 97.2 0.4 20 87.7 1.8 96.4 0.3 * The RSD quoted is that of the manganese emission intensity not the RSD of the recovery.Table 8 Effect of nitric acid on recovery of copper with and without in- ternal standardization Without MTSC With MTSC Acid concentration Recovery RSD* Recovery RSD* (% v/v) (S) (%) (%) (%I 0 100.0 1.3 100.0 0.2 5 90.4 1.3 99.0 0.3 10 88.9 1.4 98.6 0.4 15 90.2 2.0 98.6 0.4 20 87.4 3.0 97.7 0.3 * The RSD quoted is that of the copper emission intensity not the RSD of the recovery. The interference effect of nitric acid on copper emission was investigated using a similar experimental procedure to that de- scribed above.The results of this study are summarized in Table 8 and show that the recovery of copper from nitric acid solutions was reduced compared with the copper recovery from hydrochloric acid. When internal standardization was not used the recovery of the copper signal for copper in 20% (v/v) nitric acid was 87.4% whereas under similar conditions the copper recovery from hydrochloric acid [20% (v/v)] was 89.2%. When MTSC was used the copper recovery from nitric acid [20% (v/v)] was 97.7% and the recovery from 20% hydrochloric acid was 101.3%. These results suggest that the interference effect caused by nitric acid is somewhat greater than that produced by hydrochloric acid. As found previously the precision of measurement was improved considerably (up to a factor of six) by using the internal standardization method.This is attributed to removal of the variation in rate of sample uptake with the less viscous acid solution. Thus it would appear that source flicker noises in the plasma derived from sample transport effects may be eliminated by correlating emission intensities in real time using an intensity ratio meas- urement. Sulphuric acid A similar but extended study of the interference effects of sul- phuric acid was carried out. The concentrations used were 0 0.5 1.0,2.0,5.0 10.0 15.0 and 20.0% (v/v). The experimental procedure used was identical to that described previously. The results are shown in Tables 9 and 10.Considering initially the results obtained for manganese without using signal compensation it can be seen that the in- terference effects of sulphuric acid are considerably more severe than those of hydrochloric or nitric acids. The manga- nese emission recorded from the 20% (v/v) acid solution was only 60% of the expected value. When MTSC was applied the manganese recovery was increased by up to 33% and the minimum manganese recovery was 94.2%. There was a trend Table 9 Effect of sulphuric acid on recovery of manganese with and without internal standardization Without MTSC With MTSC Acid concentration Recovery RSD* Recovery RSD* (% v/v) (%Jo) (%I (%) (%b) 0.0 I 0.5 1 .o 2.0 5 .O 10.0 15.0 20.0 100.0 95.7 95.7 92.9 82.4 73.6 66.0 61.0 I .4 2.2 1.8 2.8 1.8 1.4 1.6 2.3 100.0 98.9 98.0 98.7 96.9 95.6 95.0 94.2 0.1 0.2 0.2 0.2 0.3 0.4 0.6 0.4 * The RSD quoted is that of the manganese emission intensity not the RSD of the recovery.Table 10 without internal standardization Effect of sulphuric acid on recovery of copper with and Without MTSC With MTSC Acid concentration Recovery RSD* Recovery RSD* (% VIV) (%o) (%'o) 0.0 0.5 1 .o 2.0 5.0 10.0 15.0 20.0 100.0 96.8 95.8 93.7 82.3 72.9 66.7 61.6 1.8 1.4 2.4 1.2 1 .o 0.7 0.8 1.5 100.0 98.1 96.2 97.8 95.4 94.4 93.9 92.3 0.3 0.3 0.2 0.4 0.4 0.5 0.5 0.6 * The RSD quoted is that of the copper emission intensity not the RSD of the recovery. of decreasing signal recovery with increasing acid concentra- tion which was most noticeable when MTSC was not used. The precision was improved by up to ten times when internal standardization was used.However the improvement in meas- urement precision was not as consistent as that observed for hydrochloric and nitric acids (when an improvement in preci- sion by a factor of 5-6 was observed at each acid concentra- tion). Nevertheless the benefits of MTSC for analysis of a sulphuric acid matrix are obvious. The recovery of copper in the presence of sulphuric acid so- lutions was also substantially lower than for the other acids (see Table 10). The magnitude of the interference effect of the sulphuric acid on copper was similar to that observed for man- ganese. There was again a significant improvement in signal re- covery when the signal compensation procedure was used [from 61.6 to 92.3% at 20% (v/v) sulphuric acid].As expected measurement precision was dramatically improved using MTSC but the values of RSD obtained were less consistent than those for similar solutions of hydrochloric or nitric acid. There is a trend towards poorer precision at higher acid concen- trations using MTSC but even using 20% (v/v) acid the RSD is still better than that achieved without signal compensation. Conclusions During this study it was found that when MTSC was applied to the measurement of copper and manganese emis- sion intensities from aqueous solutions the measurement precision increased by a factor of two. It was also apparent that in constructing calibration graphs the MTSC procedure may reduce the scatter of data points about the regression line. In determining the analyte signal recovery from hydro-JOURNAL O F ANALYTICAL ATOMIC SPECTROMETRY MARCH 1991 VOL.6 163 chloric nitric and sulphuric acids it was shown that the MTSC procedure enhanced the analyte signal recoveries thereby reducing or removing totally (for nitric acid) the suppressive effects of the acids on analyte emission. In general an improvement in measurement precision was also observed. In view of the consistently high signal recoveries obtained using MTSC which were largely independent of acid concen- tration it is clear from this work that it should be possible to use aqueous calibration standards for ICP-OES analysis of acid solutions. The advantage of this approach is that the inter- nal standard can be monitored simultaneously to check for bias and drift problems and thereby offers an in-built quality check for the analysis.Although signal recoveries were not always 100% the use of an internal standard ensures consistency of response which is not necessarily true when matrix matching of standards are used. Obviously the two procedures are not mutually exclusive and there may still be an advantage in using internal standardization in conjunction with the matrix matching. Internal standardization was not found to compensate totally for the effects of the mineral acids used. However it may be possible to remove residual interferences by optimizing the plasma parameters. In a multi-element situation it may be necessary to optimize the plasma conditions sequentially in order to give improved correlation of the analyte emission with that of the internal standard.This possibility is presently being examined. The authors gratefully acknowledge the support of the SERC (for L.M.G.) and ICI Wilton Materials Research Centre (Dr. W. C. Campbell) for providing funding for this work. They also thank ICI plc for permission to submit this paper for publication. References 1 Blades M. W. and Horlick G. SpecnmAim. Acta. Part B. 198 I 36. 881. 2 Koirtyohann S. R. Jones J. S.. Jester C. P.. and Yates D. A.. Spec- trochim. Acta Part B I98 I 36,49. 3 Moore G. L. Humphries-Cuff. P. J. and Watson A. E.. Specfro- chim. Acta Part B 1984,39. 9 15. 4 Bamiro F. O. Littlejohn D. and Marshall J.. J. Anal. At. Spec~trnni.. 1988.3.279. 8 9 10 I I 12 13 14 IS 16 17 18 19 20 21 22 23 24 25 26 27 28 Greenfield S.McGeachin H. McD. and Smith P. B.. Anal. Chim. Acta I976.84,67. Farino J. Miller J. R.. Smith D. D. and Browner R. F. Anal. Chem. 1987 59,2303. Nukiyama S.. and Tanasawa Y.. in Experiments on the Atomisation of Liquids in an Air Stream. translator Hope E.. Defence Research Board Department of National Defence Ottawa Ontario Canada 1950. Xu. J. Kawaguchi H.. and Mizuike A. Anal. Chim. Acts 1983 152 133. Hettipathirana T. D. Wade A. P. and Blades M. W. Spactroc.him. Acta Part B. 1990.45 27 I . Kalivas J . H. and Kowalski B. R.. Anal. Chem. 1982,54 560. Botto R. I. Spectrochim. Acta. Part B 1985 40 397. Delijska A.. and Vouchkov M. Fresenius Z. Anal. Chem.. 1985 321,448. Schmidt. G. J. and Slavin W. Anal. Chem.. 1982,54 2491. Lorber A. Goldbart 2.. Harel A.. Sharvit E. and Eldan. M. Spec- rroc.him. A m Purr B . 1986,41 105. Ramsey. M. H. and Thompson M. Analyst 1985 110. 5 19. Thompson M.. and Ramsey M. H.. AnalJst. 1985. 110. 1413. Dahlquist. R. L. and Knoll I. W.. Appl. Spe(,trosc. 1978 32. I . Uchida H. Nojiri Y. Haraguchi H.. and Fuwa K. Anal. Chim. Acta 198 1. 123 57. Bamett W. B. Fassel V. A. Kniseley R. N. Spectrochim. Acta Part B 1968,23,643. Barnett W. 9.. Fassel V. A.. Kniseley R. N.. Spertrochim. Acta Part B 1970,25 139. Sedcole J. R. Lee J.. and Pritchard M. W.. Spertr-ochim. Actu Part B 1986.41.227. Wallace G. F. At. Specrrosc. 1984 5 5. Belchamber R. M. and Horlick. G.. Specri.ochin1. Acta Part B. 1982 37. 1037. Myers S. A. and Tracy D. H.. Spectrochim. A m Part B. 1983 38. 1227. Marshall J.. Rodgers. G.. and Campbell W. C.. J. A~zul. At. Spec- rrom. 1988 3 241. Salit M. L.. Pruszkowski. E.. Yates. D. A.. and Collins J. 9.. paper presented at the Pittsburgh Conference and Exposition New Orleans. LA. USA 1988 Paper 1320. Wallace G. F.. Pirc. V. V.. and Angel A. A. Appl. SpectImc.. 1985. 5. 195. Boumans. P. W. J. M.. and Lux-Steiner. M. Ch. Spwrrochim Acta Part B . 1982.37,97. Paper- 0/01962C Receiited May 2nd. I990 Accepted December l o t h 1990