Direct Determination of Trace Metals in Sea-water Using Electrothermal Vaporization Inductively Coupled Plasma Mass Spectrometry Journal of Analytical 1 Atomic GRAEME CHAPPLE Perkin-Elmer Australia P.O. Box 309 North Ryde Sydney 21 13 Australia JOHN P. BYRNE* Department of Chemistry University of Technology Sydney P.O. Box 123 Broadway New South Wales 2007 Australia A method is described for the direct analysis of five transition elements (Co Cu Mn Ni V) in sea-water using ETV- ICP-MS. Interferences from the sea-water matrix were eliminated by a combination of in situ separation of analyte and matrix components in the ETV and use of nitric acid as a chemical modifier. The nitric acid facilitates the removal of chloride from the sea-water matrix during the sample drying stage whilst optimization of the ETV heating programme allows for effective separation of residual matrix and analyte species thus reducing ionization suppression and space charge effects in the ICP-MS.Calibration standards prepared by spiking a sea-water matrix stripped of trace metals by ion exchange gave excellent linearity and allowed for direct determination of selected analytes without the use of standard additions. Detection limits ranged from 0.003 pg I-' for V to 0.14 pg I-' for Cu. The precision and accuracy of the method were checked by analysis of two certified reference sea-waters. Keywords Electrothermal vaporization; inductively coupled plasma mass spectrometry; sea-water analysis; trace metals; transition elements The high salt content of sea-water presents many difficulties for direct nebulization analysis of trace metals by ICP-MS. Matrix ions generated in the plasma deposit on the interface cones cause changes in the ionization potential of the plasma and reduce the transport efficiency of the analyte ions in the ion lens region due to space charge effects. In each case signal suppression results.They also combine with other matrix or plasma ions to form polyatomic interferences. These effects can be lessened by simply diluting the sample. However this can significantly degrade the detection limit of the method. Some of the more successful approaches toward direct analysis involve matrix separation and preconcentration using alternative sample introduction devices. Flow injection vapour generation'l2 has been used for direct determination of As Sb and Hg in sea-water by ICP-MS with detection limits in the range of 0.5-7 ng I-'.Matrix separation and preconcentration using both ~ f f - l i n e ~ - ~ and on-line6-" ion exchange have been used in conjunction with ICP-MS for sea-water analysis. Ion exchange resins used include silica immobilized 8-hydroxy- q ~ i n o l i n e ~ - ~ . ~ Metpac CC-16,8,'o and Chelex Elements determined by this technique include the rare e a r t h ~ ~ . ~ * " Ga and In,4 and Cd and Pb,6,7,9 as well as a number of transition metal^.^.'.^." Electrothermal atomic absorption spectrometry (ETAAS) has also been used for the direct determination of some trace metals in sea-water. Recently methods of determi- nation for Mn Cr and CU,'~ Zn and Mn Zn,I3 Mo and CrI4 * To whom correspondence should be addressed.I Spectrometry have been reported. Matrix interferences can be reduced and good limits of detection can be achieved with ETAAS if isothermal atomization either in a transversely heated furnace or from a L'vov platform is used in conjunction with a variety of chemical However this technique is usually limited to single element determination and can require cali- bration by standard additions. For example Huang et a1.I4 could determine Mo in an open ocean reference sea-water sample using aqueous standard calibration but required stan- dard additions for the analysis of Mn and Cr. Electrothermal vaporization (ETV) offers an alternate method of sample introduction for ICP-MS which couples the graphite furnace's capacity for in situ matrix separation with the high sensitivity and multielement detection capability of ICP-MS.19,20 Recently Hastings et aL2' have reported a method for the determination of vanadium in sea-water using isotope dilution ICP-MS with sample introduction by ETV.The present paper describes the application of ETV-ICP-MS to the analysis of trace metals in sea-water samples. The origin and control of matrix interferences are discussed and an automated method for the determination of five transition elements is presented. The method requires minimal sample handling or treatment and uses direct calibration without the use of standard additions. EXPERIMENTAL A Perkin-Elmer Sciex ELAN 5100 ICP-MS equipped with an HGA-600 MS electrothermal vaporizer and an AS-60 model autosampler was used.Tube wall vaporization from pyrolyt- ically coated graphite tubes was used for all experiments. The HGA-600 MS was connected to the plasma torch using a PTFE tube of 85 cm length and 6mm id. The operating conditions for the ICP-MS are given in Table 1; the heating Table 1 Instrumental operating conditions ICP mass spectrometer - RF power/W 1 loo Intermediate Ar flow rate/l min-' Outer Ar flow rate/l min-' Carrier Ar flow rate/l min-' Sampler and skimmer cone Ni 15.0 0.8 0.9 Data acquisition - Dwell time/ms 10 Points/spectral peak 1 m/z monitored/measurement cycle 5 Scan mode Peak hopping Signal measurement Peak area MS resolution 0.7 u at 10% peak height Journal of Analytical Atomic Spectrometry August 1996 V02.11 (549-553) 549programme used for the ETV is discussed and details given in the next section.During the drying and pyrolysis steps of this temperature programme opposing flows of argon gas (300 ml min-l) directed from both ends of the graphite tube removed water and matrix vapours via the dosing hole. This dosing hole was automatically sealed 3 s before the onset of the high temperature vaporization step when analyte and residual matrix were transported to the plasma by the argon carrier flow. The mass spectrometer was optimized for fast transient signal response by using 10 ms dwell times for each analyte isotope and by measuring 100 readings per replicate for each transient signal. In this way five analyte species may be monitored. The ion optics were optimized by direct nebuliz- ation of aqueous standards prior to changeover to ETV.The optimum carrier gas flow was then determined by maximizing the ETV-ICP-MS signals using aqueous standards. Reagents and Standards Sea-water blanks were prepared from coastal sea-water samples (Bronte Beach Sydney Australia) by removal of the trace transition elements using ion exchange. Portions (100 ml) of the samples were eluted at pH 7.5 and at a flow rate of 1 .O ml min-' through an ion exchange column containing about 0.45 g of 8-hydroxyquinoline bonded to 100-150 mesh Porasil (Waters Milford MA USA). The column material was synthesized by the method of Deionized distilled water was prepared by passage of doubly distilled water through a Milli-Q PLUS (Millipore SA Molsheim France) water purification system.Concentrated nitric acid (analytical-reagent grade Ajax Chemicals Australia) was purified by sub-boiling distillation in a quartz still. Multi- element standard solutions were prepared by mixing and serial dilution of 1000 mg 1-1 single element standards (Merck Poole UK). Certified reference sea-water materials (NASS-3 and CASS-4) were obtained from the National Research Council of Canada. RESULTS AND DISCUSSION Sea-water Matrix Interferences The presence of high concentrations of matrix ions causes two major problems with the direct ICP-MS analysis for trace elements in sea-waters (a) isobaric overlap from polyatomic background ions; and (b) signal suppression from ion-ion interactions or space charge effects in the plasma and interface region.The potential for isobaric interferences from a sea- water matrix for the first row transition elements is illustrated in Table2. This table lists the integrated signal background counts in ETV-ICP-MS when 10 pl of a stripped sea-water blank is vaporized using a low ashing temperature of 600°C. Background spectral features are observed at virtually every mass number from 47 to 67 Da. These can be attributed to polyatomic ions formed by atom combination between the major ionic species from the sea-water matrix (Na' Mg2+ Ca2+ C1- SO4'-) along with 0 H from the water Ar from the plasma and in some cases C released from the ETV.23 Sample introduction via ETV however allows for a reduction in this spectral background if strategies normally employed for interference reduction in ETAAS are used i.e.matrix separation by a suitable thermal pre-treatment programme and use of chemical modifiers. For example HNO has been used to reduce chloride interferences in the ETAAS analysis of Zn Mn Fe and Cu in sea-water sample^.'^^^^^^^ Table 2 shows that some of these background features from sea-water are attributed to polyatomic ions which contain chlorine. These interferences can be reduced by the addition of HNO chemical SW+Nitric Acid I m n 0 47 51 53 58 59 60 67 Mass Number Fig. 1 Integrated ion counts for a sea-water blank (SW) and one containing 5% HNO modifier (SW + nitric acid) measured at masses 47( 12C3'C1+ ) 51 (35C1'60+) 53(37C1'60 + ) 58(23Na35Cl+) 59(24Mg35C1+) 60(23Na37C1+) and 67(35C11602+) Background polya- tomic ions are shown in brackets Table 2 ETV-ICP-MS background ion counts and potential interferences from a sea-water matrix on first row transition elements Mass number 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 Integrated counts 25 900 4 060 OOO 8 000 9 400 34 600 382 000 16 800 104 000 1 900 124 000 11 500 18 300 13 800 8 100 7 500 2 200 50 500 38 800 3 230 2 600 4 570 Background ions 12c35clf 48ca+ 32 16 48ca~+ 12 37 34~160 + 35c1160 + 4 0 ~ ~ 1 2 ~ + 1 2 ~ 4 0 ~ ~ + 1 3 ~ 4 0 ~ ~ + 37c160+ 4 0 ~ ~ 1 4 ~ + 4 0 c a l 4 ~ + 4 0 ~ ~ 1 5 ~ + 4 0 c a 1 5 ~ + 4 0 ~ ~ 1 6 0 + 4 0 ~ ~ 1 6 0 + 2 3 ~ ~ 3 5 ~ 1 + 4 0 ~ ~ 1 8 0 + 2 3 ~ ~ 3 7 ~ 1 + 25 2 5 ~ ~ 3 7 ~ 1 + 44ca180+ 2 4 ~ ~ 4 0 ~ ~ + 32sl6o2+ s o+ c c1+ Y 7 9 40Ca'60H+ 40Ar'60H+ 24Mg35C1 + Mg35C1+ 24Mg37C1+ 26Mg35C1+ 23Na40Ar+ 26Mg37C1+ 25Mg40Ar + 26Mg40Ar + 35c11602 + Analyte (%abundance) Ti (7.3) Ti (73.8) Ti (5.5) Ti (5.4) Cr (4.3) V (99.8) Cr (83.8) Cr (9.5) Cr (2.4) Fe (5.8) Mn (100) Fe (91.7) Fe (2.2) Ni (68.3) Ni (26.1) Ni (1.1) Ni (3.6) Cu (69.2) Zn (48.6) Cu (30.8) Zn (27.9) Zn (4.1) c o (100) 550 Journal of Analytical Atomic Spectrometry August 1996 Vol.111.OE-147 n 5-41 I It @ 1 0 2 0 3 0 4 0 ~ 6 i 96 Nitric Acid Fig. 2 Reduction in integrated counts for 37Cl+ as increasing amounts of nitric acid modifier are added to a sea-water blank. Ash 600°C vaporization 2400 "C modifier. Fig. 1 shows the reduction in background signal obtained when 10% v/v HNO is added to the sea-water blank. The seven masses shown correspond to polyatomic ions which contain chlorine (see Table2).The amount of C1- removed from the sea-water blank by addition of HNO modifier is shown in Fig. 2. In these experiments the stripped sea-water blank was dosed with increasing amounts of HNO ashed for 10 s at 600 "C prior to vaporization at 2400 "C; the integrated counts for 37Cl+ isotope were measured using Omnirange signal attenuation. These results show that the HNO modifier will remove over 95% of the chloride ion during the drying and ashing stage provided the concentration of added HNO is above about 5% v/v. However no concomi- tant reduction in the 23Na+ or 26Mg' signal was observed at this low ashing temperature. Hence HNO modifier removes C1- from the furnace but major cations such as Na+ and Mg2+ are not removed; these instead would be expected to form nitrate salts during the drying step which should then decompose to form oxides in the thermal pyrolysis These oxides of Na Mg and Ca would then be released along with any analytes during the high temperature vaporization step.If this residual matrix is not removed by an effective ashing programme it will cause analyte signal depression by ionization suppression and space charge effects in the ICP-MS. This point is illustrated by the results presented in Table 3. In these experiments the recoveries for 10 pg I-' solutions of Mn Co Ni and Cu spiked into a stripped sea-water matrix were compared with those for a 1% HNO solution (expressed as 100%). Portions (10 pl) of samples were ashed for 30 s at 800°C and then vaporized at 2400°C using maximum power heating (0 s ramp).Nitric acid (5% v/v) modifier was added to the sea-water samples. Table 3 shows that for an ashing temperature of 800"C there is sufficient residual matrix to cause around a four- to five-fold suppression of the analyte signal. Table 3 Recoveries for 10 pg 1-' solutions of analyte in a sea-water matrix after thermal pre-treatment at 800 "C Analyte Recovery in sea-water (YO) 27 19 23 22 Optimum Ashing Conditions For optimum analytical sensitivity the thermal pre-treatment temperature must be high enough to remove the bulk of the matrix thus minimizing both isobaric interferences from polya- tomic ions and signal suppression from ion-ion interactions; however it cannot be too high or analyte loss will occur. Fig. 3 shows the effect of increased ashing temperature on the removal of background spectral features from a stripped sea-water blank treated with 5% HNO at masses 55 58 59 and 63.These masses correspond to the analyte isotope masses for Mn Ni Co and Cu. The background counts for polyatomic ions at 55 (Mn) and 59 (Co) are low. However for ashing temperatures below 1000 "C high polyatomic background counts are obtained at 58 (23Na35C1+ and 40Ca180+) and 63 (23Na40Ar+ and 26Mg37Clf). Fig. 3 shows that the ashing temperature must be raised to about 1200°C if these polya- tomic interferences are to be reduced to an acceptable level. In order to determine if this temperature is high enough to eliminate ionization suppression effects on the analyte signal and still retain the analyte spiked recoveries for 10 pg 1-' analyte solutions in a sea-water matrix were measured at various ashing temperatures. The results for "Mn are shown in Fig.4. In these experiments 10 pl samples containing 10 pg 1-' of the Mn analyte in A 1% HNO B stripped sea- water and C stripped sea-water with 5% HNO modifier were ashed for 30 s and then vaporized using maximum power \ -.+- 5 8 ; ...o... 5 9 8 . 8 s'..... *...e .......... -0 .......- I 0 - . I I I Fig. 3 Effect of increased ashing temperature on the background ion counts for a sea-water blank at mass numbers 55 58 59 and 63. Vaporization temperature 2400 "C 2-( Temperature PC Fig.4 Integrated signal counts for 55Mn+ as a function of ashing temperature for 10 pl of 10 pg 1-' manganese in A; 1 % HNO,; B; a stripped sea-water matrix; C and D a stripped sea-water matrix with 5% HNO modifier.Curves A B and C measured for a 0 s vaporization heating ramp; curve D is for a 1 s ramp Journal of Analytical Atomic Spectrometry August 1996 Vol. 1 1 551heating (0 s ramp). The ashing curve for Mn in 1% HNO shows the expected analyte loss at temperatures above 1000 "C. In the presence of the sea-water matrix an increase in inte- grated counts for 55Mn+ is observed as the ashing temperature is increased from 800 to 1200°C. This is explained by the reduction in suppression caused by ion-ion interactions as increasing amounts of the matrix are removed at higher ashing temperatures. In the presence of sea-water the maximum ashing temperature is raised to 1200 "C compared with 1000 "C for the 1% HNO solution.An explanation for this observation is that sea-water which contains about 0.5% m/v MgC1 is 'self modifying' i.e. it contains its own Mg modifier which permits slightly higher ashing temperatures.12 A similar ashing curve is obtained when 5% HNO chemical modifier is used. The optimum ashing temperature is again 12OO0C but the analyte recovery is improved presumably because the removal of C1- ions by the HNO reduces the ionization suppression and space charge effects. Further improvement in the "Mn' recovery in sea-water was obtained when the heating rate during the vaporization step was slowed. Curve D (Fig. 4) shows the ashing curve for a stripped sea-water matrix with 5% HN03 modifier when a 1 s ramp is used in place of the 0 s ramp of curve C.The slower heating rate would be expected to enhance the temporal separation of matrix and analyte thus reducing ion-ion interaction between matrix and analyte. No improvement in analyte recovery was obtained when the ashing time was extended beyond 35 s. Similar ashing curves were obtained for 58Ni 59C0 60Ni and 63Cu. These results suggest that ETV-ICP-MS can be used for the simultaneous determination of up to five transition elements in an undiluted sea-water provided optimum thermal pre- treatment conditions are used in conjunction with a 5% HNO chemical modifier. The details of this optimum ETV heating programme used in the remainder of this work are given in Table 4. Calibration Curves and Detection Limits The linearity of the analyte response in a sea-water matrix was established for five transition elements (V Mn Co Cu Ni).Five multielement standards in the concentration range 0.5-10 pg 1-' were prepared by spiking 100 pg I-' standards into a stripped sea-water matrix; 5% HNO chemical modifier was added to each standard. Portions (10 pl) of samples were introduced into the ETV dried and pyrolysed using the ETV heating programme given in Table 4 and the transient ICP-MS signals for the five analyte isotopes were monitored simul- taneously. In addition to the major 63Cu and 58Ni isotopes calibration data was measured for 65Cu and 60Ni. Good linear calibrations were obtained for all five elements. The linear regression coefficient (r2) for each analyte isotope is given in Table 5. The good linearity of these data suggest that use of Table 4 Optimum heating programme for HGA-600MS electrother- mal vaporizer Stage TemperaturePC Ramp/s Hold/s Dry 90- 140 60 10 Pyrolysis 1150 20 35 Clean-up 2650 1 4 Vaporize 2400 1 6 Table 5 in a sea-water matrix Linear regression coefficients for five-point calibration curves Analyte "V 55Mn "Co "Ni 60Ni 63Cu 65Cu rz 0.9996 0.9998 0.9991 0.9990 0.9987 0.9975 0.9980 Table6 Detection limits ( 3 0 blank n = 5 ) measured in a sea-water matrix Analyte slv 5 9 c ~ "Mn 58Ni 60Ni 63cu W U Detection limit/pg 1- ' 0.003 0.006 0.016 0.12 0.13 0.14 0.18 two matrix matched standards in this concentration range would give sufficient analytical accuracy.Detection limits for each element in the sea-water matrix were determined from the stripped sea-water blanks (3a,- blank n=5) and the slopes of the calibration graphs. The detection limits are given in Table 6.The higher detection limits obtained for both isotopes of Ni and Cu result from the higher blank values for both of these elements. Despite optimiz- ation of the ETV heating programme some low levels of background polyatomic ion counts mainly from chlorides and argides of Na and Mg were still obtained. For example the integrated counts for the 58"i blank were around 1500 com- pared with 250 for 59C0. Determination of Certified Reference Materials This method of direct determination of transition metals using ETV-ICP-MS was checked for four elements (Cu Ni Co and Mn) using certified reference sea-water materials (CASS-3 and NASS-4). Calibration standards (0.5 and 2.0 pg 1-I) were prepared in a stripped sea-water matrix using 5% HNO modifier as described in the previous section.Modifier was added to the reference sea-waters and 10 pl volumes of blanks standards and samples were dried ashed and vaporized using the ETV heating programme presented in Table 4. Copper and nickel were determined at m/z = 63 and 60 respectively. The 63Cu isotope was chosen in preference to "Cu because the detection limit for this isotope is marginally lower (Table 6). Both copper isotopes have some isobaric interference from argides (23Na40Ar+ and 25Mg40Arf) however the S/B at m/z=63 is marginally better because Na is more effectively removed from the furnace during the ashing step. The compari- Table 7 Comparison between measured and certified concentrations (pg I-') for sea-water reference materials CASS-3 NASS-4 Analyte Measured Certified Measured Certified Mn 2.56k0.02 2.51 k0.36 0.351 kO.018 0.380+0.023 Co 0.038 -t 0.003 0.041 -t 0.009 0.007 _+ 0.003 0.009 _+ 0.002 Ni 0.428 k 0.068 0.386 +_ 0.062 0.17 f 0.07 0.228 & 0.009 Cu 0.52k0.17 0.517k0.062 0.48k0.18 0.228+0.011 V 1.50 & 0.1 5 * * - 1.49 & 0.03 - * Certified value not available Table 8 Recoveries for 1 pg 1-' concentration additions of analyte to CASS-3 Analyte Recovery( %) c o 99+4 Ni 99f5 63cu 95 + 18 T U 106 + 20 Mn 108k4 V 101 & 14 552 Journal of Analytical Atomic Spectrometry August 1996 Vol.11son between the values obtained by ETV-ICP-MS and the certified values is given in Table 7. The analyte recoveries obtained from a sea-water matrix using this method of analysis were also measured. Table 8 gives the recoveries for a 1 pgl-I standard spiked into the CASS-3 reference sea-water.For these five elements recoveries in the range 95-108% were obtained. CONCLUSIONS The results show that ETV-ICP-MS can be used for the direct determination of up to five transition elements in sea-water samples. Sample introduction by ETV achieves in situ separa- tion of the analyte and matrix components provided HNO chemical modifier and an optimum ETV heating programme are used. Excellent linear calibrations and good analyte recoveries were achieved provided the standards were matrix matched using a sea-water stripped of trace metals by ion exchange. Detection limits ranging from 0.003 pg 1-' for V to around 0.14 pg 1-' €or Cu were obtained.The method has a number of advantages over some alterna- tive methods. These include the use of minimal sample volumes (10-50 pl per replicate); this could be useful in the analysis of pore water samples from estuarine or ocean sediments. No sample dilution or off-line separation of analyte and matrix components is required which minimizes sample handling procedures and contamination risks. The HN03 chemical modifier used is readily obtainable at ultra-high purity and can be separately dispensed into the graphite furnace using the autosampler programme. Of the ten transition elements Cr would present a particular problem for this technique. High background ion counts are obtained from 12C4'Ar+ and 13C4'Ar+ at 52 and 53.Since these ions both contain carbon which originates in the ETV during the high temperature vaporization step use of modifiers or altered ashing programs will not reduce this background. As a result the detection limits for this element are severely degraded. Similar problems occur with titanium. Its major isotope 48Ti is overlapped by Ca (see Table2) which occurs at high concentrations in sea- water. The minor 47Ti isotope (7.3% abundance) has an isobaric interference from 12C3'Cl+ which can be reduced by the addition of HNO modifier (see Fig. 1). However this residual background plus the low isotopic abundance results in a higher detection limit of around 0.5-0.7 pg I-'. The authors wish to thank Jim Keegan University of Technology Sydney for the preparation of the silica immobil- ized 8-hydroxyquinoline.REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 Stroh A. and Vollkopf U. J. Anal. At. Spectrom. 1993 8 35. Huang M.-F. Jiang S.-J. and Hwang C.-J. J. Anal. At. Spectrom. 1995 10 31. Moller P. Dulski P. and Luck J. Spectrochim. Acta Part B 1992 47 1379. Orians K. J. and Boyle E. A. Anal. Chim. Acta 1993 282 63. Esser B. K. Volpe A. Kenneally M. and Smith D. K. Anal. Chem. 1994 66 1736. McLaren J. W. Lam J. W. H. Berman S. S. Akatsuka K. and Azeredo M. A. J. Anal. At. Spectrom. 1993 8 279. Beauchemin D. McLaren J. W. Mykytiuk A. P. and Berman S. S. J. Anal. At. Spectrom. 1988 3 305. Ebdon L. Fisher A. Handley H. and Jones P. J. Anal. At. Spectrom. 1993 8 979.Halicz L. Lam J. W. H. and McLaren J. W. Spectrochim. Acta Part B 1994 49 637. Bettinelli M. and Spezia S. At. Spectrosc. 1995 16 133. Bloxham M. J. Hill S. J. and Worsfold P. J. J. Anal. At. Spectrom. 1994 9 935. Cabon J. Y. and Le Bihan A. Spectrochim. Acta Part B 1995 50 1703. Huang S-D. and Shih K. Y. Spectrochim. Acta Part B 1995 50 837. Huang S.-D. Lai W.-R. and Shih K.-Y. Spectrochim. Acta Part B 1995 50 1237. Carnrick G. R. Slavin W. and Manning D. C. Anal. Chem. 1981 53 1866. Hydes D. J. Anal. Chem. 1980 52 959. Welz B. Schlemmer G. and Mudakavi J. R. Anal. Chem. 1988 60 2567. Huang S-D. and Shih K-D. Spectrochim. Acta Part B 1993 48 1451. GrCgoire D. C. Lamoureux M. Chakrabarti C. L. Al-Maawali S. and Byrne J. P. J. Anal. At. Spectrom. 1992 7 579. Beres S. Thomas R. Denoyer E. and Bruckner P. Spectroscopy (Eugene Oreg.) 1994 9 20. Hastings D. W. Emerson S. R. and Nelson B. K. Anal. Chem. 1996 68 371. Hill J. M. J. Chromatogr. 1973 76 455. Gregoire D. C. and Sturgeon R. E. Spectrochim. Acta Part B 1993 48 1347. Sturgeon R. E. Berman S. S. Desaulniers A. and Russell D. S. Anal. Chem. 1979 51 2364. Cabon J. Y. and Le Bihan A. Anal. Chim. Acta 1987 198 87. Byrne J. P. Chakrabarti C. L. Gregoire D. C. Lamoureux M. and Ly T. J. Anal. At. Spectrom. 1992 7 371. Paper 6 f02029A Received March 25 1996 Accepted May 30 1996 Journal of Analytical Atomic Spectrometry August 1996 Vol. 11 553