Analytica Chimica Acta 359 (1998) 205±212 Determination of Cu, Zn, Cd and Pb in sh samples by slurry sampling electrothermal vaporization inductively coupled plasma mass spectrometry Yi-Ching Li, Shiuh-Jen Jiang * Department of Chemistry, National Sun Yat-Sen University, Kaohsiung 80424, Taiwan Received 9 July 1997; received in revised form 30 October 1997; accepted 1 November 1997 Abstract Ultrasonic slurry sampling electrothermal vaporization isotope dilution inductively coupled plasma mass spectrometry (USS- ETV-ID-ICP-MS) has been applied to the determination of Cu, Zn, Cd and Pb in several sh reference samples. NH 4 NO 3 was added to the sample solution to work as a modi er. Since the sensitivity of the elements studied in sh slurry and aqueous solution was quite different, isotope dilution was used for the determination of Cu, Zn, Cd and Pb in reference samples. The in uences of instrument operating conditions, slurry preparation, non-spectroscopic and spectroscopic interferences on the ion signals and accuracy and precision of isotope ratio determination were reported. The isotope ratios of each element were calculated from the peak areas of each injection peak. Precision of isotope ratio determination was better than 6%. This method has been applied to the determination of Cu, Zn, Cd and Pb in dog sh muscle reference material (DORM-2), dog sh liver reference material (DOLT-1) and oyster tissue (NIST SRM 1566a). Accuracy was better than 19% and precision was better than 14% with ultrasonic slurry sampling and electrothermal vaporization isotope dilution inductively coupled plasma mass spectrometry. Detection limits estimated from standard addition curves were in the range of 5±50, 200±500, 8±20 and 20±50 ng g 1 for Cu, Zn, Cd and Pb, respectively, in different samples. # 1998 Elsevier Science B.V. Keywords: Ultrasonic slurry sampling; Electrothermal vaporization; Isotope dilution; Inductively coupled plasma mass spectrometry; Copper; Zinc; Cadmium; Lead; Fish samples 1. Introduction The majority of analysis by inductively coupled plasma mass spectrometry (ICP-MS) are carried out on solutions using a conventional pneumatic nebulizer. However, the type of analytical tasks that can be solved by ICP-MS can be extended using a number of *Corresponding author. other sample introduction techniques that can be easily adapted to ICP-MS. Electrothermal vaporization (ETV) is one of the sample introduction techniques, that is currently employed in ICP-MS [1±4]. This alternative technique to solution nebulization presents several advantages, including improved sensitivity, small sample size requirements and the capability for solids analysis. Perhaps the most notable bene t of ETV-ICP-MS is the possibility to perform direct solids analysis [2±4]. 0003-2670/98/$19.00 # 1998 Elsevier Science B.V. All rights reserved. P I I S 0 0 0 3-2 6 7 0 ( 9 7 ) 0 0 6 5 4-5
206 Y.-C. Li, S.-J. Jiang / Analytica Chimica Acta 359 (1998) 205±212 Ultrasonic slurry sampling (USS) is one of the methods for direct solid sample introduction, that has been successfully used in electrothermal atomic absorption spectrometry (ETAAS) [5±7]. More recently, this approach has been extended to ETV- ICP-MS [2±4]. Compared to traditional sample preparation methods such as acid digestion and dry ashing, slurry sampling offers several bene ts including reduced sample preparation time, reduced possibility of sample contamination, and decreased possibility of analyte loss before analysis. Furthermore, slurry sampling combines the bene ts of solid and liquid sampling and permits the use of conventional liquid sample handling apparatus such as autosampler [2±7]. Isotope dilution (ID) techniques were applied in several previous ICP-MS applications [8±11], and well recognized as a de nitive analytical technique for the determination of trace elements. Since another isotope of the same element represents the ideal internal standard for that element, isotope dilution results are expected to be highly accurate even when the sample contains high concentrations of concomitant elements and/or loss in sample preparation or transport through sample introduction device and ICP. In this study, USS-ID-ICP-MS was used to determine the concentrations of Cu, Zn, Cd and Pb in several sh reference samples. The in uences of instrument operating conditions, slurry preparation, non-spectroscopic and spectroscopic interferences due to the matrix on the ion signals, and the precision and accuracy of isotope ratio determination, were investigated. This method has been applied to the determination of Cu, Zn, Cd and Pb in dog sh muscle reference material (DORM-2), dog sh liver reference material (DOLT-1) and oyster tissue reference material (NIST SRM 1566a). 2. Experimental 2.1. Apparatus and conditions A Perkin-Elmer Sciex (Thornhill, Ontario, Canada) ELAN 5000 ICP-MS spectrometer equipped with a HGA-600MS electrothermal vaporizer was used. Pyrolytic graphite coated graphite tubes with platforms of the same material were used throughout. The transfer line consisted of an 80 cm long, 6 mm i.d. PTFE tubing. The sample introduction system included a Model AS-60 autosampler equipped with an USS-100 ultrasonic slurry sampler. PTFE autosampler cups were used. The USS-100 was set at 12 W (30% power), and a 30 s mix time was used to mix slurries before injection of 20 ml sample aliquots for analysis. ICP operating conditions were selected to maximize sensitivity for the isotopes of interest and to get the best precision and accuracy for isotope ratio determination. The ICP conditions were selected to maximize ion signals while a solution containing 10 ng ml 1 of Cu, Zn, Cd and Pb in 1% HNO 3 was continuously introduced with a conventional nebulizer. The sensitivity of the instrument might vary slightly from day-to-day. The ICP operating conditions used throughout this work are summarized in Table 1. Mass spectrometer parameters used for isotope ratio measurements are also listed in Table 1. The measurements were made by peak hopping. For the best accuracy and precision for isotope ratio determination, 10 ms dwell time was used in this study. Ion lenses voltages were set to get the best ion signals for the Table 1 Equipment and operating conditions ICP Mass spectrometer Outer gas flow rate/l min 1 14 Intermediate gas flow rate/l min 1 0.74 Carrier gas flow rate/l min 1 1.1 RF power/kw 1.1 Sampler/skimmer Nickel HGA-6000MS electrothermal vaporizer Sample volume/ml 20 Drying stage (20 s ramp) 1208C for 30 s Internal gas flow rate/l min 1 0.3 Charring stage (15 s ramp) 4008C for 25 s Internal gas flow rate/l min 1 0.3 Vaporization stage (0 s ramp) 23008C for 5 s. Cooling stage (1 s ramp) 208C for 5 s Clean-up stage (1 s ramp) 24008C for 5 s Cooling stage (1 s ramp) 208C for 10 s Data acquisition Dwell time/ms 10 Scan mode Peak-hopping Sweeps per reading 1 Readings per replicate 50 Signal measurement mode Integrated
Y.-C. Li, S.-J. Jiang / Analytica Chimica Acta 359 (1998) 205±212 207 elements studied simultaneously. Since the concentrations of Cu and Zn in several samples were quite high, for the measurement of Cu and Zn an offset voltage was applied to one of the ion lenses of the mass spectrometer through the OmniRange facility. This was done to effectively reduce the sensitivity of the ICP-MS. 2.2. Reagents Trace metal grade HNO 3 (70% m/m) was obtained from Fisher (Fair Lawn, NJ, USA). NH 4 NO 3, NaCl, Mg(NO 3 ) 2, (NH 4 ) 2 HPO 4 and (NH 4 ) 2 SO 4 were obtained from Merck (Darmstadt, Germany). Ti element standard solution was obtained from SPEX (Edison, NJ, USA). Triton X-100 was obtained from Sigma (St. Louis, MO, USA). Enriched isotopes of 65 CuO, 111 CdO and 204 Pb(NO 3 ) 2 were purchased from the Oak Ridge National Laboratory (Oak Ridge, TN, USA) and 67 Zn metal was obtained from Cambridge Isotope Laboratories (Andover, MA, USA). Stock solutions of approximately 500 mg l 1 of each compound were prepared by dissolution of an accurately weighed quantity of the material in nitric acid and dilution to volume. The concentrations of the spike solutions were veri ed by reversed spike isotope dilution ICP-MS. 2.3. Preparation of slurries The applicability of the method to real samples was demonstrated by the analysis of dog sh muscle reference material DORM-2, dog sh liver reference material DOLT-1 (National Research Council of Canada, Ottawa, Canada) and oyster tissue reference material NIST SRM 1566a (National Institute for Standards and Technology). The slurry was prepared by the following procedure. A 0.1 g portion of the reference material was transferred into a 10 ml ask. Suitable amounts of NH 4 NO 3, HNO 3 and Triton X-100 were added to make the nal solution containing 0.4% (m/ v) NH 4 NO 3, 2% (v/v) HNO 3 and 0.1% (v/v) Triton X- 100. After a suitable amount of enriched isotope was added, this slurry was diluted to the mark with pure water. The slurry was then sonicated for 30 min in an ultrasonic bath, and 1 ml aliquots were removed as needed for analysis with the use of a pipette while the slurry was being mixed with a vortex mixer. These aliquots were then deposited in the PTFE autosampler cups for analysis. A blank was subjected to the procedure, as outlined above, to correct for any analyte in the reagent used for sample preparation. For the studies the effect of ETV conditions and slurry preparations on the ion signals, a DORM-2 slurry sample was prepared as the procedure described above and spiked with 10 ng ml 1 Cd and Pb. 2.4. Isotope dilution calculation The analyte concentration in the sample was calculated by the following formula: C s ˆ C t R m R t XB0 M R s R m X B M 0 where C s is the concentration of analyte, C t is the concentration of the spike. R t is the isotope ratio of the spike, R s is the natural isotope ratio, and R m is the experimentally determined isotope ratio. X B is the 0 natural abundance of isotope B and X B is the abundance of isotope B in the enriched sample. M is atomic mass of the analyte element and M 0 is the atomic mass of the spike. Due to mass bias effects, intensities obtained during isotope ratio determinations were used to calculate the isotopic abundance of each element. Since the mass bias effect could be eliminated during isotope dilution calculation, in this study, the measured isotope ratio did not correct for the mass bias effect. 3. Results and discussion 3.1. NH 4 NO 3 as modifier Modi ers are commonly used in ETV-ICP-MS in order to reduce losses of analyte due to condensation on different parts of the ETV cell or the transfer line that connects the ETV to the ICP-MS [1±3,12±15]. Fig. 1 shows the effect of charring temperature on the ion signals of Cu, Zn, Cd and Pb. Since no modi er was used in this ETV-ICP-MS analysis, Zn and Cd was evaporated and ion signals of Zn and Cd decreased rapidly when the charring temperature was higher than 4008C. Due to the volatility of Zn and Cd, matrix modi er is needed for the simultaneous determination of Cu, Zn, Cd and Pb by the ETV-ICP-MS method. In
208 Y.-C. Li, S.-J. Jiang / Analytica Chimica Acta 359 (1998) 205±212 Fig. 1. Effect of charring temperature on ion signals of Cu, Zn, Cd and Pb. Slurry solution contained 1/100 (m/v) DORM-2 dogfish muscle reference material, 0.1% Triton X-100 and 2% HNO 3. No modifier was used in this experiment. All data were relative to the first point. Fig. 2. Effect of charring temperature on ion signals with NH 4 NO 3 as the modifier. Slurry solution contained 1/100 (m/v) DORM-2 dogfish muscle reference material, 0.8% m/v NH 4 NO 3, 0.1% Triton X-100 and 2% HNO 3. All data were relative to the first point. this study, several modi ers, including Pd, NH 4 NO 3, Mg(NO 3 ) 2 and a mixture of Pd and ascorbic acid, were tested for best signals of Zn and Cd. Effect of charring temperature on the ion signal when NH 4 NO 3 was used as the modi er is shown in Fig. 2. Although NH 4 NO 3 could not delay the vaporization of Zn, NH 4 NO 3 could delay the vaporization of Cd to about 6008C. Meanwhile the signal of Cd and Pb could increase slightly when NH 4 NO 3 was used as modi er. Fig. 3 shows the effect of the amount of the NH 4 NO 3 on the ion signals of Cu, Zn, Cd and Pb. As shown, the amount of NH 4 NO 3 did not affect ion signals signi cantly. Palladium has been used as the chemical modi er to improve the signals of certain volatile elements in many ETV-ICP-MS applications [12±15]. In another experiment, we found that Pd could not delay the vaporization of Cd and Zn in sh slurry, furthermore the addition of Pd modi er suppressed the ion signal of Zn signi cantly. The other modi ers also did not work properly. After careful evaluation, NH 4 NO 3 was chosen as the best modi er to be used for the elimination of some of the possible interferences and to improve signals of some selective elements in previous ETAAS and ETV-ICP-MS applications [16±18]. In the following experiments, an NH 4 NO 3 solution of 0.4% m/v was added to the sample solution to work as modi er in the ETV-ICP-MS analysis. Fig. 3. Effect of the concentration of NH 4 NO 3 on ion signals. Slurry solution contained 1/100 (m/v) DORM-2 dogfish muscle reference material, 0.1% Triton X-100 and 2% HNO 3 and various concentrations of NH 4 NO 3. All data were relative to the first point. 3.2. Selection of charring and vaporization temperature As shown in Fig. 2, although NH 4 NO 3 was used as the modi er, the signal of Zn decreased rapidly when the charring temperature was larger than 4008C. For the simultaneous determination of the elements stu-
died, the charring temperature was set at 4008C in the following experiments. In order to evaporate the elements studied completely and simultaneously, vaporization temperature was set at 23008C. The summary of the ETV operating conditions are listed in Table 1. 3.3. Effect of acid and surfactant concentrations in the prepared slurry on ion signal Y.-C. Li, S.-J. Jiang / Analytica Chimica Acta 359 (1998) 205±212 209 Fig. 4. Effect of HNO 3 concentration in the prepared slurry on ion signals. Slurry solution contained 1/100 (m/v) DORM-2 dogfish muscle reference material, 0.1% Triton X-100, 0.4% NH 4 NO 3 and various concentrations of HNO 3. All data were relative to the first point. Electrothermal atomic absorption spectrometry (ETAAS) has been successfully applied to the analysis of slurries [5±7]. Certain factors such as particle size, analyte partitioning, maximum slurry concentration and homogeneous slurry were important for the success of the analysis of slurries by ETAAS [5,6]. Effects of several parameters of the slurry preparation on the ion signals are investigated in the following experiments. An important factor in the slurry technique is the slurry concentration. However, dilution of the slurry can only be carried out within a limited range. In order to balance sample homogeneity, analyte signals and the complete vaporization of introduced sample, a dilution factor of 100 was used in the following experiments. The concentration of acid in the slurry solution could affect the rate of the extraction of the metal ions and the precision of signal measurement [7]. If a large percentage of the analyte is extracted into the liquid phase, the precision will approach that obtainable with a conventional liquid digest. In addition, the analysis will be more representative for the analyte concentration in the original solid sample. As described by Gregoire et al. [1], the presence of mineral acid in ETV-ICP-MS analysis could affect ion signals to some extent. They found that analyte signals were enhanced by as much as a factor of two in the presence of 1% (v/v) HNO 3. Since the use of HCl and H 2 SO 4 could produce extra molecular ion interferences in ICP-MS analysis, HNO 3 was used in this study. The effect of HNO 3 concentration in the slurry sample on ion signals was studied carefully in this work. Results are shown in Fig. 4. As shown, HNO 3 concentration did not affect ion signal of Cu while the signals of other elements increased 20±50% when various amounts of HNO 3 were added. For the simultaneous determination of these elements, in the following experiments, 2% (v/v) HNO 3 was used in all slurry preparations. Fig. 5 shows the dependence of the ion signal on the concentration of Triton X-100 in the slurry sample. As shown, concentration of surfactant did not affect ion signal signi cantly. For a better precision (sample Fig. 5. Effect of Triton X-100 concentration in the prepared slurry on ion signals. Slurry solution contained 1/100 (m/v) DORM-2 dogfish muscle reference material, 0.4% NH 4 NO 3, 2% HNO 3 and various concentrations of Triton X-100. All data were relative to the first point.
210 Y.-C. Li, S.-J. Jiang / Analytica Chimica Acta 359 (1998) 205±212 homogeneity), 0.1% (v/v) Triton X-100 was used in all slurry preparations. 3.4. Non-spectroscopic and spectroscopic interferences For samples with high salt content, the major component of the sample could produce several molecular ions which interfere the determination of several elements. For example, 63 Cu is interfered with the 47 Ti 16 O, 40 Ar 23 Na and 31 P 16 O 16 O, 65 Cu is interfered with 49 Ti 16 O and 40 Ar 25 Mg, 66 Zn is interfered with 34 S 16 O 16 O, and 67 Zn is interfered with 34 S 16 O 16 OH and 35 Cl 16 O 16 O. In order to evaluate the signi cance of these interferences, experiments were performed to check the interferences caused by the major components in sh samples. A stock slurry sample was prepared by the method described previously. Then the prepared DORM-2 slurry sample was spiked with 40 and 80 mg ml 1 of Na, 6 and 12 mg ml 1 of Mg and 30 and 60 mg ml 1 of P. These concentrations are about half of and equivalent to the concentration of Na, Mg and P in the prepared slurry sample, respectively. As shown in Table 2, although the ion signals changed slightly when additional amounts of Na, Mg and P were added, the isotope ratio of Cu was not affected by Na, Mg and P at these concentrations. The effect of ClO 2 ion interference on the 66 Zn/ 67 Zn isotope ratio determination is shown in Table 3. The spiked chloride concentrations were about half of the concentration and the equivalent concentration of chloride in the prepared slurry solution. As shown, the isotope ratio of Zn was not affected by the chloride at these concentrations. Also listed in Table 4 are the interferences of 34 S 16 O 16 O and 34 S 16 O 16 OH ions on 66 Zn and 67 Zn. The spiked sulfur concentration was about the equivalent concentration of sulfur in the prepared slurry solution. As shown, the isotope ratio of Zn in the slurry sample was similar to the ratio determined by element standard solution and was not affected by the sulfur at these concentrations. As shown in Table 5, in another experiment we found that the Ti contents in the prepared slurry samples did not affect 63 Cu/ 65 Cu isotope ratio determinations. The alleviation of the oxide ion interferences could be due to the introduction of much more dry aerosol with ETV sample introduction device. These experiments demonstrated that the concentrations of Cu and Zn in the sh samples can be determined directly by USS-ETV- ID-ICP-MS without signi cant interferences. 3.5. Determination of Cu, Zn, Cd and Pb in fish samples by USS-ETV-ID-ICP-MS In order to validate the USS-ETV-ID-ICP-MS method, the concentrations of Cu, Zn, Cd and Pb were determined in the dog sh muscle reference material DORM-2, dog sh liver reference material DOLT-1 and oyster tissue reference material NIST SRM 1633a. As shown in Table 6, the sensitivities of the elements studied in sh slurries and aqueous solution were quite different, and external calibration methods could not be used for the quanti cation of these elements in this study. The isotope dilution method was used for the determination of Cu, Zn, Cd and Pb in the reference materials. Analysis results are shown in Table 7. The determined con- Table 2 Effect of Na, Mg and P matrix on Cu isotope ratio determination, (nˆ7) Sample 63 Cu 65 Cu 63 Cu/ 65 Cu DORM-2 slurry 6250810 3020410 2.070.03 DORM-2 slurry 40 mg ml 1 Na 7190180 347080 2.070.05 DORM-2 slurry 80 mg ml 1 Na 7980270 3970130 2.010.03 DORM-2 slurry 6 mg ml 1 Mg 7650170 3730190 2.050.08 DORM-2 slurry 12 mg ml 1 Mg 7750210 384060 2.020.03 DORM-2 slurry 30 mg ml 1 P 6470350 3140180 2.060.08 DORM-2 slurry 60 mg ml 1 P 6440170 3140110 2.050.06 a The DORM-2 slurry solution contained 1/100 (m/v) DORM-2 dogfish muscle reference material, 0.4% NH 4 NO 3, 2% HNO 3 and 0.1% Triton X-100.
Y.-C. Li, S.-J. Jiang / Analytica Chimica Acta 359 (1998) 205±212 211 Table 3 Effect of Cl matrix on Zn isotope ratio determination, (nˆ7) Sample 66 Zn 67 Zn 66 Zn/ 67 Zn DORM-2 slurry 185002000 2920400 6.340.38 DORM-2 slurry 55 mg ml 1 Cl 205001200 3160350 6.490.46 DORM-2 slurry 110 mg ml 1 Cl 194001600 2980220 6.510.28 a The DORM-2 slurry solution contained 1/100 (m/v) DORM-2 dogfish muscle reference material, 0.4% NH 4 NO 3, 2% HNO 3 and 0.1% Triton X-100. Table 4 Effect of S matrix on Zn isotope ratio determination, (nˆ7) Sample 66 Zn 67 Zn 66 Zn/ 67 Zn 200 ng ml 1 Zn 482004100 7600580 6.340.09 Standard solution DORM-2 slurry 371004000 5700620 6.510.20 DORM-2 slurry 100 mg ml 1 S 379001800 5890220 6.430.21 a The DORM-2 slurry solution contained 1/100 (m/v) DORM-2 dogfish muscle reference material, 0.4% NH 4 NO 3, 2% HNO 3 and 0.1% Triton X-100. Table 5 Effect of Ti component on Cu isotope ratio determination, (nˆ7) Sample 63 Cu 65 Cu 63 Cu/ 65 Cu DORM-2 slurry 5980460 2940230 2.030.04 DORM-2 slurry 1 mg ml 1 Ti 7530190 3660100 2.060.02 DORM-2 slurry 2 mg ml 1 Ti 7530270 366090 2.060.05 a The DORM-2 slurry solution contained 1/100 (m/v) DORM-2 dogfish muscle reference material, 0.4% NH 4 NO 3, 2% HNO 3 and 0.1% Triton X-100. Table 6 Sensitivities of the elements studied in various matrices as measured by external calibration method and standard addition method Calibration method and sample matrix Sensitivity/counts ng 1 ml Cu Zn Cd Pb External calibration 3470 1200 2510 13600 Aqueous solution Standard addition DORM-2 slurry 2380 310 a 2190 6570 DOLT-1 slurry 370 a 25 a 1630 6540 NIST SRM 1633a slurry 290 a 26 a 2240 12200 a Various OminiRange settings were used. centrations are in good agreement with the certi ed values. This experiment indicated that Cu, Zn, Cd and Pb could be readily quanti ed by isotope dilution inductively coupled plasma mass spectrometry with ultrasonic slurry sampling electrothermal vaporization. Detection limits of the elements in various samples (Table 7) were determined from the standard addition curves of each element in different samples. It was based on the usual de nition as the concentration of the analyte yielding a signal equivalent to three times the standard deviation of the blank signal. Detection limits estimated from standard addition curves were in the range of 5±50, 200±500, 8±20 and 20±50 ng g 1 for Cu, Zn, Cd and Pb, respectively, in different samples. Better detection limits are to be expected with better puri ed reagents.
212 Y.-C. Li, S.-J. Jiang / Analytica Chimica Acta 359 (1998) 205±212 Table 7 Determination of Cu, Zn, Cd and Pb in selected reference materials by USS-ETV-ID-ICP-MS, (nˆ7) Sample and elements Acknowledgements Concentration (mg g 1 ) Found a Certified b DOLT-1 dogfish liver Cu 18.00.5 20.81.2 0.05 Zn 92.33.0 92.52.3 0.3 Cd 4.170.36 4.180.28 0.02 Pb 1.340.07 1.360.29 0.05 DORM-2 dogfish muscle Cu 2.520.32 2.340.16 0.005 Zn 24.31.4 25.62.3 0.2 Cd 0.0510.004 0.0430.008 0.008 Pb 0.0760.008 0.0650.007 0.02 NIST 1633a oyster tissue Cu 60.93.3 66.34.3 0.03 Zn 78336 83057 0.5 Cd 4.120.16 4.150.38 0.008 Pb 0.3540.051 0.3710.014 0.05 Detection limit c (mg g 1 ) a Values are mean of seven measurementsstandard deviation. b NRCC and NIST certified value. Values are given in 95% confidence limits. c Detection limit was estimated from the standard addition calibration curve as the signal equivalent to three times the standard deviation of the blank. This research was supported by a grant from the National Science Council of the Republic of China under Contract NSC 86-2113-M-110-021. References [1] D.C. Gregoire, D.M. Goltz, M.M. Lamoureux, C.L. Chakrabarti, J. Anal. At. Spectrom. 9 (1994) 919. [2] D.C. Gregoire, N.J. Miller-Ihli, R.E. Sturgeon, J. Anal. At. Spectrom. 9 (1994) 605. [3] R.W. Fonseca, N.J. Miller-Ihli, Appl. Spectrosc. 10 (1995) 1403. [4] M.-J. Liaw, S.-J. Jiang, J. Anal. At. Spectrom. 11 (1996) 555. [5] C. Bendicho, M.T.C. de Loos-Vollebregt, J. Anal. At. Spectrom. 6 (1991) 353. [6] N.J. Miller-Ihli, Anal. Chem. 64 (1992) 964. [7] N.J. Miller-Ihli, J. Anal. At. Spectrom. 9 (1994) 1129. [8] J.R. Dean, L. Ebdon, J. Anal. At. Spectrom. 2 (1987) 369. [9] D.C. Gregoire, J. Lee, J. Anal. At. Spectrom. 9 (1994) 393. [10] J.W. McLaren, A.P. Mykytiuk, S.N. Willie, S.S. Berman, Anal. Chem. 57 (1985) 2907. [11] T.-J. Hwang, S.-J. Jiang, J. Anal. At. Spectrom. 11 (1996) 353. [12] D.C. Gregoire, R.E. Sturgeon, Spectrochim. Acta, Part B 48 (1993) 1347. [13] R.D. Ediger, S.A. Beres, Spectrochim. Acta, Part B 47 (1992) 907. [14] D.C. Gregoire, M. Lamoureux, C.L. Chakrabati, S. Al- Maawali, J.P. Byrne, J. Anal. At. Spectrom. 7 (1992) 579. [15] D.C. Gregoire, S. Al-Maawali, C.L. Chakrabati, Spectrochim. Acta, Part B 47 (1992) 1123. [16] I. Game, L. Balabanoff, R. Valdebenito, L. Vivaldi, Analyst 111 (1986) 1139. [17] Y. Tan, W.D. Marshall, J.-S. Blais, Analyst 121 (1996) 483. [18] C.J. Park, J.C. Van Loon, P. Arrowsmith, J.B. French, Anal. Chem. 59 (1987) 2191.