Determination of tellurium in indium antimonide by slurry sampling electrothermal atomic absorption spectrometry M. Y. Shiue, Y. C. Chan, J. Mierzwa and M. H. Yang* Department of Nuclear Science, National Tsing-Hua University, 30043 Hsinchu, Taiwan Received 17th August 1998, Accepted 2nd November 1998 A method for the determination of tellurium dopant concentration in indium antimonide (InSb) by Zeeman eVect electrothermal atomic absorption spectrometry using the slurry sampling technique was developed.The eVects of chemical modifier type and mass on the absorbance-peak characteristics of tellurium in InSb slurried samples were studied. The atomization behavior of tellurium in InSb slurries could be greatly improved by the use of palladium nitrate as a chemical modifier. The detection limit of the optimized procedure was 0.4 mg g-1. In the determination of tellurium at the concentration level of 16 mg g-1, a relative standard deviation of 7% was obtained.Good agreement of the results obtained by the slurry sampling technique with those obtained by solution electrothermal atomic absorption spectrometry and inductively coupled plasma mass spectrometry was found. dissolution. Unfortunately, a significant decrease in sensitivity 1. Introduction was obtained owing to the interference caused by the parent The III–V semiconductor compounds, e.g., gallium arsenide matrix. Moreover, this method also has the disadvantage of a (GaAs), indium phosphide (InP) and indium antimonide higher acid concentration of sample solution contacting the (InSb), are very important materials in the microelectronics graphite tubes, thus decreasing their lifetime.The third techand optoelectronics industries.1 Indium antimonide has nique is based on a prior preconcentration and/or separation become the material of choice in the fabrication of photocond- procedure.5,12,20 A typical example based on reductive copreuctors, magnetoresistors and infrared detectors since it has the cipitation with palladium using ascorbic acid for the determihighest electron mobility and maximum drift velocity.2 It is nation of trace amounts of tellurium in nickel-based known that the performance of the final devices is markedly superalloys and several other metals was reported.5 A detection influenced by the presence of defects on the substrates.3 limit as low as 0.01 mg g-1 was found to be achievable.Another Tellurium-doped InSb single crystals are grown by the possibility based on in situ preconcentration via volatilization Czochralski method, by adding elemental Te to the undoped of tellurium hydride compounds from the sample matrix polycrystalline InSb in the starting charge.The concentration followed by trapping and atomization in a graphite atomizer, of Te through the ingot depends on its concentration in the coated with silver or palladium, has also been reported.21,22 liquid phase in front of the growing crystal face.This concen- The sensitivity of determination reported in terms of tration is not constant if the distribution coeYcient of Te in characteristic mass was about 20 pg.21,22 the crystal is not unity. Generally, only the chemical analysis The purpose of this study was to develop a relatively simple of the wafers is conclusive in determining the distribution and rapid method for the determination of tellurium at mg g-1 coeYcients. concentration levels in tellurium-doped InSb with a slurry- Electrothermal atomic absorption spectrometry (ETAAS) sampling (SS) ETAAS method.The important features of this has long been used for the determination of trace impurities technique are its simplicity even compared with direct solid in Group III–V semiconductors.4 Methods for the determi- sample analysis and that it is less prone to the contamination nation of Te in a wide variety of real matrices including heat- that is frequently encountered in sample dissolution and resisting alloys,5–8 environmental,9,10 biological11–13 and geo- mineralization procedures.23,24 This technique also oVers adequate sensitivity.In this paper, a method of slurry prep- logical samples have been reported.14 The determination of Te aration and direct injection into the electrothermal atomizer in GaAs and InP has also been described.15,16 Although some for the determination of tellurium in InSb without sample fundamental studies on the tellurium atomization eYciency in dissolution is reported.The eVects of palladium nitrate and ETAAS from aqueous solution have been reported,17,18 there palladium nitrate–magnesium nitrate chemical modifiers on appear to have been no studies on the determination of Te the atomization of Te from slurried samples were also investi- in InSb. gated. The quality of the analyte peak shape, precision, Up to now, three main techniques for tellurium accuracy and limit of detection achievable by the proposed determination by ETAAS have generally been used.The first method were evaluated and are discussed. is based on tellurium determination by direct atomization of the analyte from solid samples. Headridge and Nicholson19 analyzed a nickel-base alloy using solid sampling ETAAS. However, this method needs calibration with standard alloys 2. Experimental and it is diYcult to obtain a series of well characterized 2.1. Apparatus standard alloys.The second technique is based on the determination of tellurium by ETAAS after acid dissolution of solid AAS measurements were made on a Perkin-Elmer (Norwalk, samples.6,7,16 Taddia et al.16 determined tellurium in indium CT, USA) Zeeman/5100 PC atomic absorption spectrometer equipped with an HGA-600 graphite furnace atomizer, an phosphide by ETAAS after hydrochloric acid and nitric acid J. Anal. At. Spectrom., 1999, 14, 69–74 69Table 1 Experimental conditions for the determination of tellurium 2.4.ETAAS procedure by slurry sampling ETAAS A 20mL aliquot of the slurry sample solution followed by Temperature/ Ramp Hold Argon flow 20 mL of chemical modifier solution was injected into the Step °C time/s time/s rate/mL min-1 furnace. Two modifiers including Pd and Pd–Mg (with a mass ratio of 251) were used. The thermal cycle in Table 1 was Temperature program— applied and the peak absorbance was read. The method of Drying 120 10 40 300 standard additions based on spiking the slurries with an Pyrolysis 1100a 10 30 300 aqueous standard solution was used to determine the sample Cooling 20 1 15 300 Atomization 2200 or 2600 0 5 0 concentration.Clean out 2650 1 5 300 Sample volume 20 mL 3. Results and discussion Instrumental parameters— Wavelength 214.3 nm 3.1. Optimization of experimental conditions Slit width 0.2 nm Radiation Te electrodeless discharge lamp Pyrolysis temperature. The optimization of the furnace Read time 5 s pyrolysis temperature was carried out for both 5 mg mL-1 Signal mode Peak area InSb slurry and 2 ng of Te standard aqueous samples with USS 100 device power 40% for 25 s and without the presence of a chemical modifier.The results output setting of these experiments are shown in Fig. 1. As can be seen in a800 °C in the absence of Pd. Fig. 1(a) and (d), in the absence of modifier, for both the slurry and aqueous standard volatilization losses occur if the samples are pyrolyzed above 800 and 600 °C, respectively. The AS-60 autosampler and a USS-100 slurry sampler.A tellurium fact that the maximum pyrolysis temperature can be increased electrodeless discharge lamp (EDL) and pyrolytic graphite- by 200 °C in the presence of a slurry sample (as compared coated tubes with integrated platforms (Perkin-Elmer, with its absence) may imply that the matrix of slurries, U� berlingen, Germany) were used throughout. The optimized probably indium, can work as a chemical modifier. A similar heating program and instrumental parameters are given in result was reported by Taddia et al.16 in their study of the Table 1.For the preparation of the powdered InSb sample, a determination of Te in the sample solution containing indium Retsch Mixer Mill MM 2000 (F. Kurt Retsch, Germany) trichloride. In the presence of palladium modifier as shown in equipped with tungsten carbide grinding jars and bas Fig. 1(b) and (e), tellurium is thermally stabilized without was used.Micro-weighing was performed on a Mettler significant losses up to about 1100 and 1300 °C, respectively. (Hightstown, NJ, USA) AT 201 electronic microbalance. Moreover, in the presence of Pd–Mg modifier (both as nitrates), as shown in Fig. 1(c), tellurium is also stable up to 2.2. Reagents and sample about 1100 °C in a slurry sample. The results presented above show that in the presence of All reagents were of the analytical reagent grade, unless stated either Pd or Pd–Mg as a chemical modifier a pyrolysis tempera- otherwise.High-purity water, which was purified by deminture of 1100 °C is equally applicable without significant loss eralization and two-stage quartz distillation, was used throughof Te and the sensitivities are almost the same in both cases. out. Nitric acid and hydrochloric acid were prepared by sub- However, the background absorbance in the presence of boiling distillation in quartz stills.A stock standard solution Pd–Mg modifier is slightly higher than that in the presence of of Te (1000 mg L-1) from Aldrich (Milwaukee, WI, USA) Pd modifier. On the basis of the above results, palladium as a was diluted to the desired concentrations with high-purity chemical modifier and a pyrolysis temperature of 1100 °C were water containing 0.2% nitric acid. Palladium (10 000 mg L-1, employed throughout the following study. as nitrate) and magnesium nitrate (10 000 mg L-1) solutions were purchased from Inorganic Ventures (Lakewood, NJ, Atomization temperature.The influence of atomization USA). Triton X-100 was obtained from Merck (Darmstadt, temperature, as shown in Fig. 2, on the tellurium signal from Germany). Tellurium-doped indium antimonide slices were a 5 mg mL-1 indium antimonide slurry was compared with supplied by the MCP Wafer Technology (UK). 2.3. Slurry preparation procedure A slice of InSb was washed with acetone to remove traces of grease, dipped in 1 M hydrochloric acid for a few minutes and then rinsed with high-purity water and air-dried in a class 100 clean bench.For the preparation of slurries, approximately 1 g of InSb slice was ground in Retsch Mixer Mill MM2000 at a 50% power output setting for 20 min. The particle diameter of the InSb powder so obtained did not exceed 3 mm as examined on several scanning electron microscope micrographs. A portion of sample (from 2.5 to 40 mg) was weighed into a 2.5 mL polyethylene vial and 2 mL of 0.2% HNO3 containing 0.005% of Triton X-100 surfactant were added. Shortly before analysis, the suspensions were pre-treated for 2 min in an ultrasonic bath to disintegrate larger particle agglomerates.Fig. 1 Pyrolysis curves of tellurium obtained for (a) slurry sample The vials containing the slurry were then directly transferred without Pd modifier, (b) slurry sample with addition of 200 mg of Pd, into the autosampler tray. Prior to taking each slurry aliquot (c) slurry sample with addition of 200 mg of Pd+100 mg ofMg(NO3)2, by the sampling capillary, it was homogenized by ultrasonic (d) aqueous tellurium standard (2 ng) without Pd modifier and agitation using a USS-100 slurry sampler at a 40% power (e) aqueous tellurium standard (2 ng) with addition of 20 mg of Pd modifier.Atomization temperature, 2200 °C. output setting for 25 s. 70 J. Anal. At. Spectrom., 1999, 14, 69–74palladium is an eVective chemical modifier for the ETAAS determination of tellurium in slurried InSb.EVect of amount of palladium. A few papers have reported that some of the analyte signal is often reduced by the presence of larger masses of Pd modifier.26,27 Qiao and Jackson28 suggested a physical mechanism of the eVect of Pd on analyte modification. Thus, if the mass of Pd is increased too much, larger droplets of Pd are formed and the analyte diVuses more slowly out of these larger droplets and consequently result in lower peak heights and greater signal tailing, i.e., the analyte signal is reduced.Moreover, Frech et al.27 reported that there Fig. 2 Atomization curves of tellurium obtained (a) for slurry sample existed not only the eVects of analyte release from the palwith addition of 200 mg of Pd and (b) aqueous tellurium standard ladium modifier, but also the eVects of analyte adsorption and (2 ng) with addition of 20 mg of Pd modifier. Pyrolysis temperature, desorption at the cooler ends of the tube in the presence of 1100 °C.increasing modifier mass. In this study, a similar eVect of Pd modifier on Te was also observed. A detailed study using electrothermal vaporization inductively coupled plasma that from a 2 ng of Te aqueous solution in the presence of mass spectrometric (ETV-ICPMS), laser ablation-ICP-MS palladium modifier. It can be seen clearly that no plateau appears with either atomization curve. The low atomization eYciency of Te at lower atomization temperatures resulted in broader tellurium peaks and a lower integrated absorbance.At higher temperatures, owing to the sudden atomization of Te, the diVusion losses were significantly higher, hence a lower integrated absorbance and narrower peak profile were observed. For the curve obtained with aqueous standard solution, the maximum sensitivity was obtained at an atomization temperature of around 2000 °C, but, a relatively broader peak was also observed.In order to obtain adequate sensitivity and a better profile, an optimum atomization temperature of 2200 °C was selected for both slurry and aqueous solution. 3.2. EVect of palladium modifier on tellurium absorbance The tellurium atomization profiles with and without palladium chemical modifier were investigated. The results indicate that when tellurium is atomized from an InSb slurry without the addition of palladium modifier, the background absorbance is too high (&2) and seriously overlaps the Te signal.The irregular shape of the Te peak shown could be attributed to the extremely high background absorbance that is beyond the capacity of the Zeeman-eVect background corrector. With the addition of palladium modifier, the background absorbance signal suddenly decreases from &2 to around 1.1 and shifts to a later appearance time, leading to only partial overlap with the Te signal. Therefore, a better tellurium peak shape is obtained.The pronounced eVect of palladium modifier on tellurium absorbance mentioned above may be explained on the basis of co-expulsion of the analyte element with the volatilized matrix.25 As described previously, in the absence of chemical modifier only a relatively lower pyrolysis temperature, i.e., 800 °C, can be used in order to prevent volatilization loss of Te. This may consequently result in the retention of a substantial amount of InSb matrix in the graphite tube at this lower temperature.In the subsequent atomization stage Te is expected to be carried into the absorption volume by co-explusion with the rapidly expanding vapors of the InSb matrix that remained in the graphite tube. The large background absorbance and the irregular shape of the tellurium peak as previously described can be attributed to a gas-phase interference caused by the matrix substance. On the other hand, in the presence of palladium modifier a higher pyrolysis temperature (1100 °C) can be used to expel the sample matrix more eYciently and can therefore result in a less significant Fig. 3 Absorption signals of tellurium in 5 mg mL-1 InSb slurries influence of the matrix on the atomization stage. The lower with diVerent additions of Pd modifier: (a) 100 mg, atomic absorption background absorbance and better peak shape of tellurium (AA)=0.119, background (BG)=1.37; (b) 200 mg, AA=0.239, BG= obtained in the presence of palladium modifier can be 1.10; (c) 300 mg, AA=0.122, BG=0.78; and (d) 400 mg, AA=0.129, BG=0.73.explained on this basis. It can therefore be concluded that J. Anal. At. Spectrom., 1999, 14, 69–74 71Table 2 Tellurium partitioning in 5 mg mL-1 InSb slurry Concentration of Concentration of liquid fraction/ HNO3 (% v/v) concentration of slurry (%) 0.2 0 1 0 5 0 20 50 Table 2 gives the Te partitioning data for four diVerent concentrations of HNO3, 0.2, 1, 5 and 20% v/v. The results show that the amount of Te present in the liquid phase was below 1% (close to 0%) when the liquid phase contained 0.2, 1 and Fig. 4 EVect of amount of palladium on the absorbance of tellurium 5% HNO3. However, the amount of Te present in the liquid obtained for (a) 0, (b) 5, (c) 10 and (d) 20 mg mL-1 InSb slurries. phase increased to 50% at 20% HNO3. These results indicate that Te in the InSb slurry is not easily extracted into the liquid phase by dilute HNO3. However, as the concentration of (LA-ICP-MS) and scanning electron microscopic (SEM) techniques to investigate the mechanism of tellurium atomization HNO3 increases to 20%, some chemical reactions may occur during the ultrasonic mixing.This may be clearly indicated by in the presence of palladium nitrate modifier in ETAAS is currently in progress. the color change of the slurry sample from black to gray–white and finally to a white precipitate, most probably in the form The eVect of palladium modifier mass on the tellurium signal was first investigated.Fig. 3 shows the eVect of various of Sb2O3. This shows that InSb can be partially dissolved by 20% HNO3 and about half of the Te goes into a readily soluble palladium masses on the Te absorption profiles for a 5 mgmL-1 InSb slurried sample with the furnace heating form. The total dissolution of the sample is achieved in concentrated nitric acid. program in Table 1. Fig. 3(a)–(d) represent the absorbance profiles for Te in the presence of 100, 200, 300 and 400 mg of Replicate aliquot precisions in the 3–4% range were obtained for 5 mg mL-1 InSb slurries prepared in 0.2, 1, 5 and 20% v/v palladium, respectively.Comparison of these figures indicates a trend of decreasing peak height and increasing signal tailing HNO3. The good precision of measurements obtained for both 0 and 50% of Te extracted into the liquid phase indicates that with increasing addition of Pd to the samples. However, when the amount of Pd added is too small, as in Fig. 3(a), a higher the slurry is very homogeneous if the proposed method of slurry preparation is used. background absorbance is observed, presumably due to co-expulsion of analyte with the volatile matrix as described previously. In the present study, the maximum sensitivity was 3.4. Standardization and sample analysis achieved for this specific sample (5 mg mL-1 InSb) with In general, the use of solid standards with certified addition of 200 mg of Pd as modifier.concentrations of the elements of interest and matrices corre- In order to find a general guideline to be followed for the sponding to those of the samples is the most accurate stan- optimum amount of Pd to be added to an InSb slurried sample dardization method in solid sampling ETAAS, including the of specific concentration, a series of experiments were carried slurry sampling technique.31 Unfortunately, such standard out as follows. To slurry samples containing 0, 5, 10 and materials are rare and are not available for all matrices.The 20 mg mL-1 InSb, various additions of Pd modifier were made next choice, considering the accuracy achievable, is calibration and the eVect on the absorbance was investigated. The results using a calibration graph based on aqueous standard solutions. obtained are shown in Fig. 4. As can be seen, in the absence However, in this standardization technique, a similar behavior of InSb, i.e., the Te standard solution, the maximum of the analyte element in the standard and in the slurry during absorbance was not appreciably changed in the presence of the pyrolysis and atomization stages is prerequisite for good Pd amounts in the range 10–80 mg, whereas for the InSb accuracy.32 slurried samples, a certain range of maximum absorbance was In this study, the pyrolysis was studied for a 5 mg mL-1 obtained upon specific addition of palladium modifier.These InSb slurry and an aqueous Te standard using 200 and 20 mg results clearly indicate that the choice of the optimum mass of palladium modifier, respectively.The results shown in of palladium modifier is dependent on the slurry concentration. Fig. 1(b) and (e) clearly indicate a similar trend between these From the practical point of view, it means that a higher mass two pyrolysis curves. This may serve as an evidence for a of palladium modifier must be used to obtain the best sensisimilar behavior of Te atoms in both the aqueous standard tivity when a more concentrated InSb slurry is to be analysed.and InSb slurry during the pyrolysis step. The absorption signals of tellurium in the slurried samples, spiked slurry and 3.3. Tellurium partitioning in slurries aqueous solution were further tested and the results are shown in Fig. 5(a)–(c). As can be seen, the atomization behavior of The distribution of analyte in the slurry is of particular interest in the characterization of the precision of the slurry sampling Te is also very similar in all three cases, despite the appearance of peak maxima that slightly deviate from each other.approach. When no analyte is found in the liquid phase, the limiting source of measurement variability from the replicate Furthermore, the characteristic masses of tellurium for the aqueous solution and indium antimonide slurry using a pal- aliquots of a single slurry will be related to slurry mixing and the heterogeneity of the analyte in the insoluble solid fraction.ladium modifier were calculated and the results show that as little as 19 and 34 pg, respectively, can be achieved. The slopes When large percentages of analyte are extracted into the liquid phase, replicate aliquot precision may approach those of pure of the calibration curve method and standard addition method are 0.004 and 0.0026, respectively, indicating a significant liquid digests.29,30 To test Te partitioning, slurries were agitated and then left undisturbed for 90 min to ensure that settling suppression of the tellurium signal due to the presence of InSb matrix.Hence the method of standard additions based on would occur. After this time interval, the top portion of the slurry was carefully sampled and the concentration of the spiking the slurries with aqueous standard solutions is needed for quantification purposes. dissolved Te was determined by ETAAS. The slurry was re-suspended and the Te concentration was determined again.The applicability of the method was tested for the analysis 72 J. Anal. At. Spectrom., 1999, 14, 69–74Fig. 6 Dependence of Te integrated absorbance on InSb slurry concentration (mg mL-1). method, measurements of Te absorbance with respect to diVerent slurry concentrations were conducted, and for each slurry concentration the amount of palladium modifier was adjusted accordingly. As is evident from Fig. 6, for InSb slurry concentrations up to 20 mg mL-1, there is a linear relationship between the slurry concentration and absorbance.It was observed for the tested sample that the contribution of the sample inhomogeneity starts to be significant at InSb slurry concentrations below about 1.2 mg mL-1, indicating that slurry concentrations lower than this level could not be recommended with this analytical technique. 3.5. Method detection limit The method detection limit is defined as the analyte concentration that gives a signal which is three times the standard deviation of the procedure blank (n=7).The method detection limits were found to be 0.4 and 0.8 mg g-1 for slurry ETAAS and solution ETAAS, respectively. In this study, the evaluation of the blank and limits of detection were based on the analysis of a 5 mg mL-1 undoped InSb. As can be seen, Fig. 5 Absorption signals of tellurium for (a) 5 mg mL-1 InSb slurry the limit of detection obtained by SS-ETAAS is two-fold better (with about 1.74 ng of Te) with the addition of 200 mg of Pd, than that achievable by solution ETAAS.The detection limit (b) 5 mg mL-1 InSb slurry spiked with 3.75 ng of Te with the addition obtained by SS-ETAAS is good enough to determine typical of 200 mg of Pd and (c) aqueous tellurium standard (2 ng) with the addition of 20 mg of Pd. dopant concentrations of tellurium in InSb. of a tellurium-doped InSb and the results are presented in 4. Conclusion Table 3. Since no standard samples with known reference A method for the direct determination of tellurium in InSb by values are available, the evaluation of the reliability of the SS-ETAAS was developed.Palladium nitrate was used as an data can alternatively be conducted by using diVerent analyte Vective chemical modifier to overcome otherwise strong ical methods. In Table 3 the results of two independent matrix interferences. The amount of modifier must be carefully methods, i.e., solution ETAAS and ICP-MS, are also preoptimized and calibration must be performed with the standard sented.As in slurry ETAAS, a significant suppression of the additions method. The good analytical reliability achievable tellurium signal due to the presence of the InSb matrix was with this technique was confirmed by comparison with solution also observed in solution ETAAS and therefore the method ETAAS and ICP-MS. The detection limit is suYcient to of standard additions was used for quantification purposes. determine typical dopant concentrations of tellurium in InSb.As can be seen, the concentrations of Te in Te-doped InSb The proposed method has been proved to be relatively simple determined by the present method, solution ETAAS and and rapid. Another advantage is the possibility of avoiding ICP-MS are 16.2, 16.4 and 17.0 mg g-1, respectively. From the sample decomposition, which reduces time-consuming good agreement with results obtained by solution ETAAS and sample preparation procedures and the use of hazardous ICP-MS and its good reproducibility (RSD 7.5%), the concentrated acids.analytical reliability of the established method is confirmed. To establish the linear working range for the proposed Acknowledgement Table 3 Comparison of the results of tellurium determination by The authors gratefully acknowledge the financial support of SS-ETAAS, solution ETAAS (Sol-ETAAS) and solution ICP-MS the National Science Council of Taiwan. Method Te concentration/mg g-1 RSD (%) References SS-ETAAS 16.2±1.2 (n=5) 7 Sol-ETAASa 16.4±0.7 (n=4) 4 1 S.M. Sze, Semiconductor Devices Physics and Technology, Wiley, New York, 1985. ICP-MSa 17.0±0.8 (n=3) 5 2 O. Sugiura and M. Matsumura, J. Appl. Phys., 1985, 24, 25. aDetermination after acid dissolution of the sample. 3 K. Tada, M. Tatsumi, M. Morioka, T. Araki and T. Kawase, in J. Anal. At. Spectrom., 1999, 14, 69–74 73Semiconductor and Semimetals, ed. R. K. Willardson and 18 K.Matsumoto, Anal. Sci., 1993, 9, 447. 19 J. B. Headridge and R. A. Nicholson, Analyst, 1982, 107, 1200. A. C. Beer, Academic Press, Boston, 1990, vol. 31, ch. 5, 20 E. M. Donaldson and M. E. Leaver, Talanta, 1990, 37, 173. pp. 175–238. 21 Z.-M. Ni, B. He and H.-B. Han, J. Anal. At. Spectrom., 1993, 4 I. G. Judelevich, N. F. Beisel, T. S. Papina and K. Dittrich, 8, 995. Spectrochim. Acta, Part B, 1984, 39, 467. 22 Z. Li, Z. M. Ni and X. Q. Shan, Spectrochim. Acta, Part B, 1989, 5 T. Ashino and K. Takada, Anal. Chim. Acta, 1995, 312, 157. 44, 751. 6 B. L. Gong, Y. M. Liu, Z. H. Li and T. Z. Lin, Anal. Chim. Acta, 23 C. Bendicho and M. T. C. de Loos-Vollebregt, J. Anal. At. 1995, 304, 115. Spectrom., 1991, 6, 353. 7 S.-J. J. Tsai and C.-C. Jan, Analyst, 1993, 118, 1183. 24 N. J. Miller-Ihli, Anal. Chem., 1992, 64, 964A. 8 O. Kujirai, K. Ide, T. Kobayashi and E. Sudo, Talanta, 1982, 25 B. Welz, G. Schlemmer and J. R. Mudakavi, J. Anal. At. 29, 27. Spectrom., 1992, 7, 1257. 9 S. Muangnoicharoen, K. Y. Chiou and O. K. Manuel, Talanta, 26 M. W. Hinds and K. W. Jackson, J. Anal. At. Spectrom., 1990, 1988, 35, 679. 5, 199. 10 H. Klinkenberg, T. Beeren and W. V. Borm, Spectrochim. Acta, 27 W. Frech, K. Li, M. Berglund and D. C. Baxter, J. Anal. Atom. Part B, 1993, 48, 649. Spectrom., 1992, 7, 141. 11 Z. H. Siddik and R. A. Newman, Anal. Biochem., 1988, 172, 190. 28 H. Qiao and K. W. Jackson, Spectrochim. Acta, Part B, 1991, 12 R. Kobayashi and K. Imaizumi, Anal. Sci., 1990, 6, 83. 46, 1841. 13 G.Weibust, F. J. Langmyhr and Y. Thomassen, Anal. Chim. Acta, 29 N. J. Miller-Ihli, Fresenius’ J. Anal. Chem., 1993, 345, 482. 1981, 128, 23. 30 N. J. Miller-Ihli, Fresenius’ J. Anal. Chem., 1990, 337, 271. 14 H. Niskavaara and E. Kontas, Anal. Chim. Acta, 1990, 231, 273. 31 F. J. Langmyhr and G. Wibetoe, Prog. Anal. At. Spectrosc., 1985, 15 S. Nakayama, M. Shibata, H. Mizusuna and S. Harada, Bunseki 8, 193. Kagaku, 1987, 36, 499. 32 S. Hauptkorn and V. Krivan, Spectrochim. Acta, Part B, 1994, 16 M. Taddia, A. Bellini and R. Fornari, J. Anal. At. Spectrom., 1995, 49, 221. 10, 433. 17 W. Frech and D. C. Baxter, Spectrochim. Acta, Part B, 1990, 45, 867. Paper 8/06462H 74 J. Anal. At. Spectrom., 1999, 14, 69–74