JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY FEBRUARY 1993 VOL. 8 35 Optimization and Use of Flow Injection Vapour Generation Inductively Coupled Plasma Mass Spectrometry for the Determination of Arsenic Antimony and Mercury in Water and Sea-water at Ultratrace Levels Andreas Stroh and Uwe Vollkopf Bodenseewerk Perkin-Elmer GmbH Postfach I0 I I 64 W-7770 Oberlingen Germany Vapour generation flow injection inductively coupled plasma mass spectrometry (FI-ICP-MS) was used for the determination of As Sb and Hg in four international water reference materials (National Research Council of Canada). A commercially available FI device was easily connected to the ICP-MS system. Flow injection parameters such as sample volume purge gas flow rate and concentration of reductant were investigated and optimized (univariate) as were different sample pre-reduction techniques.The precision and accuracy of the results obtained show the applicability of this method to the determination of vapour-forming elements at ultratrace levels in environmental samples. Detection limits are in the range 0.5-7 ng I-' for Bi Sb %e Te Hg and As (with pre-reduction). Keywords Flow injection inductively coupled plasma mass spectrometry; on-line vapour generation; optimization of flow injection parameters; matrix effects; certified reference materials for water Inductively coupled plasma mass spectrometry (ICP-MS) has proved to be a very powerful technique for trace multi- element and isotopic determinations in many types of samples. 1-4 In this technique predominantly singly-charged positive ions are generated in an argon plasma source and then transferred into and analysed with a quadrupole mass analyser.For the majority of the elements listed in the Periodic Table the degree of ionization in an Ar plasma is >9Ooh. However for As Se and Hg the number of ions produced in the plasma source drops dramatically to 48.87 30.53 and 32.31% respecti~ely.~ The reason for this is the high ionization potential of these elements As 9.8 1 eV; Se 9.752 eV; and Hg 10.437 eV,6 which are much closer to the Ar ionization potential (Ar 15.759 eV) than most other elements. This leads to a decrease of detection power for these elements however in most cases ICP-MS detection limits are still sufficient for their determination. In samples that contain very low concentrations of As/Se and Hg elements (e.g.unpolluted surface water biological or clinical materials) or samples that have to be diluted because they have a harsh sample matrix detection capabil- ities are restricted making analysis difficult or even impossible. As ICP-MS is also considered to be a very flexible detection technique alternate sample introduction systems can be easily used in conjunction with ICP-MS such as laser sampling electrothermal vaporization (ETV) ultra- sonic nebulization high-performance liquid chromato- graphy and flow injection (FI).'-13 Flow injection was first described by Ruzicka and HansenI4 in 1975 and since then has emerged as a versatile sampling tool capable of enhancing considerably the analytical capabilities of atomic absorption spectrometry (AAS) ICP atomic emission spec- trometry and ICP-MS.15-z1 Hydride generation is frequently used in analytical chemistry to improve the sensitivity of measurements of volatile vapour-forming elements such as As Sb Bi Se Te Ge Sn and Pb which are thought to be harmful at much lower levels than previously believed. Vapour generation is commonly performed using organiczz or inorganicz3 phase vapour generation. Heitkemper and CarusoZ4 have used sample introduction by continuous flow hydride generation for the determination of As Cd Pb and Cu in National Institute of Standards and Technology (NIST) Standard Reference Material (SRM) 1566a Oyster Tissue with ICP-MS. Sample introduction by FI with a sample loop of 500 pl was used.One of the advantages of the FI technique over continuous flow vapour generation is the significantly lower reagent and sample consumption. Together with the hydride-forming elements As and Sb on- line detection of Hg cold vapour was performed. The aim of the study was the development of an FI vapour generation ICP-MS method for the determination of As Sb and Hg in waters at ultratrace levels. Experimental Instrumentation The ICP mass spectrometer used for this work was an Elan 5000 (Perkin-Elmer SCIEX Norwalk CT USA). The system was equipped with a four-channel mass flow controller to ensure gas flqw stability. A FIAS-200 (Boden- seewerk Perkin-Elmer Uberlingen Germany) equipped with a random access AS-90 autosampler (Perkin-Elmer) for on-line FI vapour generation was used.Operating conditions used for the ICP mass spectrometer and the FIAS-200 are summarized in Tables 1 and 2 respectively. All operating parameters of the FI accessory are controlled from the transient signal software application which is incorporated in the Elan user software. Data acquisition was performed using multichannel analysis with only one channel per mass unit. No time is wasted using multiple points per mass unit so the speed of analysis is increased and time resolution for observing the transients is guaran- teed. This means that it is possible to determine more than 75 elements using a single transient signal. The timing and Table 1 Operating conditions of the Elan 5000 ICP mass spectrometer Forward power/W 1100 15 0.8 Purge gas flow rate/l min-I 1 .oo Outer gas flow rate/l min-I Intermediate gas flow rate/l min-l Sampler cone Platinum Skimmer cone Platinum Operating pressure InterfaceIPa Quadrupole/Pa Data acquisition Dwell time/ms No.of readings 266.6-400 800 x Multichannel analysis one point per mass (peak hopping) 20 6036 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY FEBRUARY 1993 VOL. 8 Table 2 Instrument parameters of the FIAS-200 FI device Pump 11 Pump 21 Step Time/s rev min-I rev min-I 1 10 100 0 2 8 I00 0 3 2 100 0 4 10 0 120 5 1 75 0 6 60 50 0 Valve position Fill Fill Fill Inject Fill Fill Remarks Rinse tubing Fill sample loop Start measurement Start hydride generation Back to fill position Rinse tubing and loop (only between samples) duration of the quadrupole scan cycles are also computer controlled when working with the FI device. Integrated transient signal software allows calculation of analytical results based on both transient signal peak height and peak area.The optimization of the ICP-MS operating parameters such as purge (carrier) gas flow rate and ion optic settings was simply performed by continuously aspirating a test solution typically a mixed standard solution containing 10 ng m1-I of Mg Rh and Pb. The FI manifold is shown in Fig. 1. The gas-liquid phase separator has been described e l s e ~ h e r e ~ ~ - ~ ~ and was used to extract the volatile vapour-forming elements from the liquid sample. For vapour generation FI-ICP-MS the normal spray chamber including the nebulizer was re- moved from the torch adapter. A 1000 mm length of poly(tetrafluoroethy1ene) (PTFE) tubing (1.75 mm i.d.) connected the gas-liquid separator with the torch adapter of the Ar plasma. Argon purge gas supply was taken from the nebulizer gas mass flow controller which was directly connected with the gas inlet of the FI device.Thus the change from conventional solution aspiration to vapour generation FI can be performed in less than 3 min. Because the generator-interface design of the mass spectrometer used in this study ensures fixed ion energies,27 the optimum operating conditions are identical for a wet plasma (contin- uous solution nebulization) and dry plasma (vapour genera- tion laser sampling ETV). Because of this behaviour no special optimization of the ICP mass spectrometer was necessary for the FI vapour generation study.Reagents and Sample Pre-treatment As sample reductant NaBH (96% m/v Merck No. 6371 Darmstadt Germany) was used. The NaBH solutions were stabilized with 0.05% m/v NaOH (Merck No. 6495 solid). As acid carrier for the samples 3% v/v nitric acid purified by sub-boiling distillation (BSP 929 Berghof Eningen mI min-' Fig. 1 FI manifold for vapour generation FI-ICP-MS P 1 and P2 pumps; C chemifold; W waste; L sample loop; G gas-liquid separator; V injection valve; and AS autosampler Germany) was used. All reagents were diluted/dissolved with ultra pure de-ionized water (Barnstead Nanopure Boston MA USA) and prepared fresh daily. For the initial basic studies multi-element solutions were prepared with 2 pg 1-i of Bi and Sb and 5 pg 1-i of As Se Te and Hg without any further sample pre-treatment.For quantitative mea- surements it is important to pre-reduce the samples to ensure the same valence state of the elements. This is necessary because the sensitivity of the hydride forming process is different depending on the valence of the element e.g. AsItt and AsV. The pre-reduction procedure is different for the vapour-forming elements. Arsenic and Sb have to be mixed with a solution containing 5% m/v ascorbic acid (Merck No. 500074) and 5% m/v KI (Merck No. 5044). Unfortunately this reductant mixture is strong enough to reduce Se and Te to the elemental state (SeO TeO) hence reducing the advantage of ICP-MS as a fast multi- elemental technique. As an alternative pre-reductant 5% m/v KBr (Merck No. 4904) which has been tested in the past,28 was examined.The sample was heated at 50 "C for 50 min after the addition of 1 ml Of 5% m/v KBr solution or with an excess of 6 mol 1-i HC1 at 90 "C for 15 min. It was found that neither of these two methods reduces As and Sb to the + 3 oxidation state thus it was not possible to use identical sample preparation procedures for all the analyte elements of interest. If all vapour-forming elements (As Sb Se Te Bi Hg Sn Ge and Hg) must be determined in the same sample(s) it is necessary to split the sample into two aliquots and to prepare these as required. It may be possible to overcome this problem by adding KI after mixing the sample with the NaBH r e d ~ c t a n t ~ ~ or by mixing the pre- reductant with the NaBH4,23 although these approaches were not examined in this study Bi and Hg were found to need no special sample pre-treatment steps.Aliquots ( 10 ml) of thoroughly mixed water samples were taken and acidified with 3 ml of concentrated HCl. A mixture of 1 ml of 5% m/v u-5% m/v ascorbic acid was added to ensure reduction of As and Sb. After 15 min reaction time 500 pl of a 100 pg 1-i Bi solution were added and the sample solution diluted to a final volume of 25 ml. This solution contained 2 pg 1-l of Bi added as internal standard. Results and Discussion Acid Carrier Hydrochloric acid is typically used as the acid carrier in hydride generation AAS. In this study the use of HCl as acid carrier for vapour generation ICP-MS was avoided because chlorine vapour generated in the phase separator was expected to enter the plasma ion source with the a n a l y t e ~ .~ ~ This could then result in an ArCl interference at m/z 75 which is the only stable isotope of As. The influence of increasing C1 content in the sample solution on the signals of m/z 75 and 77 is shown in Fig. 2. The C1 matrix was simulated by various HCl concentrations from 0.3 to 3.0% v/v. Both masses that can be interfered with by ArCl species (35Cl and 37Cl) were analysed in order to examine whetherJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY FEBRUARY 1993 VOL. 8 37 Concentration HCI (% vhr) Fig. 2 Influence of C1 content in the measurement solution on signal intensities of A relative atomic mass 75 and B relative atomic mass 77 1000000 'VJ VJ B = I - II Y d 1 .o 2.0 0.2 0.02 0.002 0.0002 0.00002 Concentration NaBH (% m/v) Fig.3 Dependence of A As hydride and B Hg cold vapour generation signal intensity on NaBH concentration the signals were coming from impurities of the HCl or really from the formation of ArCl. To correct for the differences in the natural abundance of 35Cl and 37Cl (35C1:37C1=3.08) the intensities for m/z 77 are multiplied by 3.08 to correct for this difference. While the intensities at m/z 75 show a significant increase with increasing concentration of C1 the corrected count rate at m/z 77 is only slightly increased. If there were significant formation of ArCl the intensities of both masses should behave in a similar manner. However the observations indicate that the contribution to the As signal results more from impurities of As in the HC1 used for this study rather than from formation of ArCl.The equivalent As concentration found in 3% v/v HCl is 0.029 ,ug 1-I. In general HC1 also gives higher blank values for Hg. Vollkopf et aL3* and G i i n ~ e l ~ ~ have shown that HN03 can be used instead of HCl as carrier solution. A concentration of 3% v/v HN03 results in good sensitivity during the hydride-forming process. Nitrates in the measurement solution can cause some interferences in the hydride-forming process for Se. If the samples have to be pre-reduced by boiling with 6 mol 1 - I hydrochloric acid nitrites can be formed which can significantly interfere with the determination of Se.33934 In this case the samples must be treated with sulfanilamide or sulfamic acid. 0.6 0.7 0.8 0.9 1.0 1.05 1.1 1.2 Purge gas flow rate/l min-' Fig.4 Dependence of vapour generation signal intensities on purge gas flow A Te; B As; C Sb; D Hg and E Bi. Note the scale has been expanded between 1.0 and 1.1 1 min-' in order to show the real maximum more clearly Sodium Tetrahydroborate The rate of formation of volatile vapour greatly depends on the concentration of the NaBH solution. An experiment was carried out during which the concentration of NaBH was increased stepwise from 0.00002 to 2.0% m/v while the concentration of HN03 carrier was held constant. The signal intensities of As and Hg were monitored during this study. For As the best signal to noise ratio (S/N) was achieved using an NaBH concentration of 2.0% m/v. A further increase of the concentration of NaBH produced higher signal intensities but also higher background.There was also a greater risk of the generation of foam in the gas-liquid separator. Foam bubbles can enter the PTFE tubing leading to the gas torch and cause problems when entering the plasma without passing through a spray chamber. With decreasing concentrations of the NaBH solution As sensitivity falls and reaches background levels at 0.02% m/v (see Fig. 3). However Hg shows exactly the opposite behaviour. With a more dilute solution of NaBH signal intensity is improved for this element reaching a maximum intensity at concentrations of 0.002-0.0002% m/v NaBH (see Fig. 3). As compromise conditions an NaBH concentration of 0.2% m/v was employed for all subsequent measurements which gave in general good signal intensities for all elements investigated.For single element determinations of Hg where maximum detection power is required a 0.0002% m/v NaBH reductant solution should be used as this results in maximum sensitivity and the lowest contamination risk from the NaBH solution. Purge Gas Flow Rate The optimization of the purge gas flow rate was performed with a multi-element solution containing 2 pg 1-1 of Sb and Bi and 5 ,ug 1-' of As Te and Hg. The flow rate of the purge gas was increased in 0.1 1 min-l increments from 0.6 up to 1. I 1 min-l. The maximum sensitivity was achieved for all elements at a flow rate of between 1 .O and 1.1 1 min-I as Table 3 Influence of purge gas flow on S/N Element Flow rate/l min-' Element ng ml-I 0.6 0.7 0.8 0.9 1 1.05 1.1 As 5 3.7 8 8.3 31 57 52 42 Sb 2 1 1 21 8.6 37 69 67 53 Bi 2 45 1 1 1 193 714 784 672 359 Hg 5 35 69 69 148 252 232 137 concent rat ion/ .38 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY FEBRUARY 1993 VOL.8 c c - I i c) - 4 ) - - 1 - 1 I 1 0.95 ' 1 I I I I I 0 0.5 1 .o 0.5 2.0 2.5 3.0 Concentration NaCl (?h mlm) Fig. 6 Matrix effects of NaCl on the hydride-forming process for A Bi and B As and vapour generation for C Hg 150 I 1 C .- a .- c g z 30 60 90 Time/min 50 o Fig. 7 Long-term precision of vapour generation for 200 replicate injections of 5 pg 1-I Hg shown in Fig. 4. A calculation of S/N carried out for each of the elements studied during the experiments on purge gas flow rates showed that the best ratios were obtained using a flow rate of 1.0 1 min-l.This flow rate was therefore used for all further measurements on all elements. The results of the S/N calculations are summarized in Table 3. Injection Volume The dependence of element signal intensities on the sample loop volume is shown in Fig. 5. The aim of this experiment was to minimize reagent and sample consumption without losing too much sensitivity for the analysis of the vapour- forming elements at ultratrace levels. Flow rates of carrier acid and NaBH solution were kept constant. Maximum peak height signal intensities were achieved with a sample loop volume of 800 pl. A further increase of the sample loop volume does not result in higher signals. A slight increase of the peak area intensities is indicated by a small amount of peak broadening.No steady-state signal was observed with increasing loop volumes so it seems probable that chemical saturation during the vapour-formation process is reached when a certain volume of sample is mixed with the adjusted NaBH and acid carrier flow rates. Since approximately 90% of the maximum possible signal intensity was reached with a sample loop volume of 500 pl this volume of sample loop was used for all further measurements leading to sufficient sensitivity and low sample consumption. Influence of Sample Matrix on Vapour Generation Signal and Internal Standardization The possibility of C1 vapour generation leading to a contribution to the As intensities caused by the formation of ArCl was discussed earlier. For the analysis of sea-water samples it is also very important to estimate the influence of increasing sample matrix on the vapour generation process for the different elements.Synthetic NaCl solutions from 0.1 to 3.0% m/v have been prepared and spiked with a 1 pg 1-l of Bi and a 10 pg 1-l of Hg and As standard solution. The recoveries found for these solutions expressed as relative concentrations compared with the matrix free calibration solution are shown in Fig. 6. All three elements show a similar pattern in these solutions and the inaccuracy is as low as 1-7%. This implies that the use of Bi as internal standard will lead to better precision and accuracy for the analysis of these types of samples. For different sample types e.g. biological and clinical materials Bi is not useful because it is part of the sample composition.Even in small amounts it will result in incorrect concentration values because of its tremendous sensitivity for the vapour formation process. There is no other suitable candidate as an internal standard for the analysis of such samples. Therefore use of an internal standard is not recommended. For the investigation of the stability during extended operation without the use of internal standardization a standard solution was prepared containing 5 pg 1-l of Hg. Mercury was used for this test with regard to two different analytical questions namely stability of signals and drift for Hg caused by contamination or carryover when analys- ing this element over long time periods. Two hundred replicate injections (loop volume 500 pl) were performed for 31 s each leading to a total measurement time of approximately 90 min.The precision of the particular injection without internal standardization is presented normalized to the first injection in Fig. 7. The calculated relative standard deviation (RSD O/o) for the 200 replicate injections is 1.67%. Precision and Accuracy The precision of the vapour generation process was evalu- ated using a mixed standard solution containing 5 pg 1-* of Hg As Se and Te and 2 pg 1-I of Bi and Sb. The SDs and RSDs were calculated for each element from five replicate sample injections with a loop volume of 500 pl. The results of the calculations are summarized in Table 4. Bismuth gave the best precision 0.5% RSD while the RSDs for the other elements determined at the same time were between 0.9 and 1.6%. To establish the accuracy of the method four interna- tional water standards (National Research Council of Canada NRCC) were analysed two Riverine Waters SLRS-1 and SLRS-2 Coastal Seawater CASS-2 and Open Ocean Seawater NASS-3.Five replicates of each sample were prepared using the sample preparation procedure Table 4 Precision of FI-ICP-MS vapour generation for five sample injections. Loop volume 500 pl Mean value Concentration/ of intensity/ SD/ RSD Element 1-I counts s-l counts s-I (Oh) As 2 17 350 319 1.6 Sb 1 76 384 951 1.1 2916 0.5 Bi 1 477 697 Hg 2 214 544 3057 1.3 82Se 2 27 369 484 1.6 78Se 2 65 157 678 0.9 Te 2 287 785 5096 1.6JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY FEBRUARY 1993 VOL. 8 39 Table 5 Results for the determination of As Sb and Hg in reference water samples.Recovery (O/O) was estimated by spiking samples with 1 pg g-' of Sb and 2 pg 1-I of As and Hg Certified value/ FI-ICP-MS Element Pg I-' SD /pg I-' SD Recovery (O/O) SLRS-I- As 0.55 0.08 0.50 0.03 98.5 Sb 0.63 0.05 0.69 0.04 113 Hg NC* - 0.5 0.1 108.5 SLRS-2- As 0.77 0.09 0.8 0.06 112 Sb 0.26 0.05 0.33 0.02 1 1 1 Hg NC - (0.04) - 101.5 CASS-2- As 1.01 0.07 1.08 0.04 102 Sb NC - 0.3 0.01 96 Hg NC - 0.5 0.07 95 NASS-3- As 1.65 0.19 1.74 0.04 105 Sb NC - 0.189 0.008 97 Hg NC - 1.36 0.2 93 *NC = Not certified. Table 6 Detection limits of FI-ICP-MS vapour generation with and without sample pre-reduction Detection limiting 1-I Element Without pre-reduction Pre-reduction with KI As 26 78Se 16 82Se 7 Sb 4 Te 1 Hg 4 Bi 0.5 6 - - 1.3 43.I* 0.6 - * Hg detection limit degraded due to reagent contamination. outlined previously. External calibration was performed using a blank and two aqueous standard solutions contain- ing 1 and 2.5 pg 1-' for Sb and 2 and 5 pg 1-' for As and Hg. All solutions were prepared in the same way as the samples and 2 pg 1-l of Bi were added as internal standard. Because Se and Te are reduced to the elemental state during the reduction step they could not be detected together with the other elements. The results for As Sb and Hg are presented in Table 5. Mean values are shown for three replicate measurements of the five individual sample preparations of the reference material. The calibration graphs for As Sb and Hg were all linear with correlation coefficients varying between 0.997 and 0.999.The results for As and Sb show excellent agreement with the certified values. The concen- trations of Hg are not certified in these reference water samples. To confirm the results for this element recovery tests were performed. The samples were spiked with 1 pg 1-l of Sb and 2 pg 1-I of As and Hg and re-analysed. The results are shown in Table 5 and satisfactory recoveries of the order of 93- 1 13% were found. The reproducibility of the quantitative results was calcu- lated based on three replicate measurements of five indivi- dual sample preparations of the water reference samples. The RSDs of the measurements varied between 2.8 and 7.5% for Sb and As and is somewhat poorer (14-20%) for the very low concentrations of Hg in all four samples (see Table 5).Detection Limits The detection limits summarized in Table 6 were estimated for multi-element determinations with the standard solu- tions used for the determination of precision. Detection limits were calculated based on 30 using blank solutions ( 18 Mi2 de-ionized water) which were pre-reduced in exactly the same way as the samples and without pre-reduction prior to analysis. Lower detection limits were achieved for As and Sb when the samples were pre-reduced prior to measurement since higher signal intensities are found for these two elements when they are in oxidation state +3. The pre-reduction procedure has no beneficial effect on the detection limits for Bi and the detection limit for Hg is worse after the pre-reduction procedure. This is due to slight contamination of the reagents used for pre-reduction.The detection limits for Hg could be improved by using reagents of higher purity. Conclusions Flow injection vapour generation in conjunction with ICP- MS is a very powerful technique that enhances considerably the analytical capabilities of ICP-MS for volatile vapour- forming elements thus allowing multi-elemental determi- nations of those analytes at ultratrace levels in environmen- tal samples with very good accuracy and precision. Vapour generation offers several advantages over continuous solu- tion aspiration for several vapour-forming elements. The advantages of increased sensitivity and avoidance of spec- tral interferences caused by a high salt matrix are high- lighted. The sensitivity precision and accuracy of the technique for fresh and ocean waters are demonstrated by the analysis of certified reference materials. References 1 Doherty W.and Van der Voet A. Can. J. Spectrosc. 1985 30 135. 2 Gregoire D. C. Anal. Chem. 1987 59 2479. 3 Beauchemin D. McLaren J . W. Mykytink A. P. and Berman S. S. Anal. Chem. 1987 59 778.40 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY FEBRUARY 1993 VOL. 8 4 Boomer D. W. and Powell M. J. Anal. Chem. 1987 59 2810. 5 Inductively Coupled Plasmas in Analytical Atomic Spectrome- try eds. Montaser A. and Golightly D. W. VCH Weinheim 1987 p. 361. 6 Applications of Inductively Coupled Plasma Mass Spectroscopy eds. Date A. R. and Gray A. L. Blackie Glasgow 1989. 7 Arrowsmith P. Anal. Chem. 1987 59 1437. 8 Park C. J.Van Loon J. C. Arrowsmith P. and French J. B. Anal. Chem. 1987 59 2191. 9 Beauchemin D. Siu K. W. M. and Berman S. S. Anal. Chem. 1988 60 2587. 10 Vollkopf U. Paul M. and Denoyer E. R. Fresenius’Z. Anal. Chem. 1992,342 9 17. 11 Vollkopf U. Giinsel A. Paul M. and Wiesmann H. Applicalions of Plasma Source Mass Spectrometry eds. Hol- land G. and Eaton A. Royal Society of Chemistry Cam- bridge 1991 p. 162. 12 Denoyer E. R. Fredeen K. J. and Hager J. W. Anal. Chem. 1991,63 445A. 13 Flow Injection Analysis eds. Ruzicka J. and Hansen E. A. Wiley New York 198 1. 14 Ruzicka J. and Hansen E. A Anal Chim. Acta 1975 78 145. 15 Ruzicka J. and Hansen E. A Anal. Chim. Acta 1988 180 1. 16 Tyson J. F. Analyst 1985 110 419. 17 Fang Z. Welz B. and Schlemmer G. J. Anal. At. Spectrom. 1989 4 91. 18 Israel Y. Lhsztity A. and Barnes R. M. Analyst 1989 114 1259. 19 Wang X. Viczian M. Lasztity A. and Barnes R. M. J. Anal. At. Spectrom. 1988 3 821. 20 Fang Z. and Welz B. J. Anal. At. Spectrom. 1989 4 543. 21 Ruzicka J. and Arndal A. Anal. Chim. Acta 1989 216 243. 22 Menkndez Garcia A. M. Sanchez Uria J. E. S. and Sanz- Medel A. J. Anal. At. Spectrom. 1989 4 581. 23 Schramel P. and Xu L.-q. Fresenius’ 2. Anal. Chem. 1991 340 41. 24 Heitkemper D. T. and Caruso J. A. Appl. Spectrosc. 1990 44 228. 25 Schrader W. and Schulze H. LaborPraxis 1989 5 406. 26 Welz B. and Schubert-Jacobs M. At. Spectrosc. 1991,12,91. 27 Douglas D. J. and French J. B. Spectrochim. Acta Part B 1986 41 197. 28 de Oliveira E. McLaren J. W. and Berman S. S. Anal. Chem. 1983,55 2047. 29 Nygaard D. D. and Lowry J. H. Anal. Chem. 1982,54,803. 30 Branch S. Corns W. T. Ebdon L. Hill S. and O’Neill P. J. Anal. At. Spectrom. 1991 6 155. 31 Vollkopf U. Gunsel A. and Janssen A. At. Spectrosc. 1990 11 135. 32 Giinsel A. M.Sc. Thesis Fachhochschule Steinfurt 1990. 33 Welz B. Proceedings of the Third Colloquium Atomic Spectros- copy ed. Welz B. Verlag Chemie Weinheim 1985 p. 571. 34 Cutter G. A. Anal. Chim. Acta 1983 149 391. Paper2/02550G Received May 18. 1992 Accepted August 10 1992