Microwave-assisted Dilute Acid Extraction of Trace Metals From Biological Samples for Atomic Absorption Spectrometric Determination CHAO YAN ZHOU MING KEONG WONG* AND L I P LIN KOH Department of Chemistry National University of Singapore Lower Kent Ridge Road Singapore 11 9260 YEOW CHIN WEE Department of Botany National University of Singapore Lower Kent Ridge Road Singapore 119260 A microwave extraction method with dilute nitric acid for biological samples was developed using a pressure feedback microwave extraction system. The concentration of nitric acid pressure setting and extraction time were optimized by a mixed level orthogonal array design. The interactions among the selected parameters for the extraction were investigated. The effect of sample mass was also studied. Six replicate analyses of NIST SRM 1515 Apple Leaves were performed under the optimized extraction conditions (5 ml of 14% v/v nitric acid; pressure setting 1104 kPa; extraction time 30 min; sample mass 0.3 g).Eight elements oiz. Ca Cu Fe K Mg Mn Ni and Zn were determined by FAAS or ETA AS. Recoveries of 96-103% were achieved for all the elements. The RSDs of the test elements were less than 3.8% except for Ni. The extraction method was also employed to extract four other biological SRMs and CRMs. With the exception of Ni the recoveries of the eight test elements were 91-107% in all the reference materials. Keywords Microwave-assisted acid extraction; trace metal; biological sample ; atomic absorption spectrometry The successful dissolution of a biological sample is a crucial first stage in many spectrochemical and electrochemical analy- ses for trace metal constituents.There are two types of decomposition methods commonly employed wet decompo- sition and dry ashing.' Contamination and the loss of volatile elements are serious problems for the dry-ashing method. In wet decomposition various acid mixtures have been proposed for different sample matrices and different analytes. In most instances it is essential to employ an oxidizing agent to obtain complete decomposition. Both methods are tedious and time consuming but their careful execution is the backbone of a successful laboratory analysis.' Two recent developments in the sample preparation pro- cedure are the use of closed vessels to accelerate sample extraction or digestion and minimize contamination and losses of volatile elements and the use of microwave radiation to assist in extraction or dige~tion.~ With closed-vessel sample preparation systems various acid mixtures have been employed to destroy rapidly the organic matrix of the samples at elevated temperature or pre~sure.~-'~ The oxidizing power of acids is significantly increased with elevated temperature. The contami- nation loss of volatile elements and dissolution time are red~ced.~." However most of the established extraction or digestion methods employ concentrated acids.Extraction with dilute nitric acid is a simple and safe technique that has been applied with some success in the * To whom correspondence should be addressed. I Journal of I Analytical 1 Atomic 1 Spectrometry 1 determination of a range of elements in animal tissue CRMs at room temperat~re'~.'~ and of botanical CRMs with heat- ing.16 However generally it takes several hours or a day to obtain the digest.The problem of the dilute nitric acid extrac- tion can be overcome by the use of closed-vessel microwave extraction. The use of dilute nitric acid removes the danger of rapid pressure build-up as well as the possibility of systematic errors of using strong oxidizing acids. In this work orthogonal array design was used to evaluate the parameters of the microwave oven and acid concentration. The effect of sample mass was studied after other parameters had been optimized. Orthogonal array design is a chemometric approach combining the advantages of both the simplex method and factorial design.It provides an efficient and effective testing strategy.17-19 From the work a biological sample preparation method using a closed-vessel microwave extraction technique and dilute nitric acid was developed. The NIST SRM 1515 Apple Leaves was used to evaluate the selected parameters. EXPERIMENTAL Instrumentation All biological samples were extracted in an MDS-2000 Microwave Sample Preparation System (CEM Matthews NC USA). This system features a 630 W magnetron programmable in 1% power increments. An internal pressure control system allows for the control of pressure from 0 to 1380 kPa (200 psi) in five separate stages. A 12 vessel 360" revolving turntable operates via an optical sensor at 6 rev min-'. The extraction vessels have rupture membranes for safe operation under 1725 kPa (250 psi).The high concentration elements in the biological samples were determined with a Shimadzu (Tokyo Japan) flame atomic absorption spectrometer (AA-670) equipped with a gas control- ler and PR-4 graphic printer (FAAS). The low concentration elements were determined using a Perkin-Elmer (Norwalk CT USA) Model 4100ZL atomic absorption spectrometer equipped with an HGA-600 graphite furnace and Zeeman background correction (ETAAS). Specific instrumental para- meters for the various elements are given in Table 1. In all instances the determinations were carried out using a calibration graph. Chemicals The concentrated nitric acid used as well as all the other chemicals were of analytical-reagent grade (Merck Darmstadt Germany).All glass and poly(propy1ene) apparatus was washed with 5% v/v nitric acid and de-ionized distilled water. Journal of Analytical Atomic Spectrometry August 1996 Vol. 11 (585-590) 585Table 1 Instrumental conditions for the determination of the elements FAAS- Gas flow rate/l min-' Element Acetylene Mn Fe Mg 1.9 2.0 1.6 ETAAS- Lamp Element current Wavelength c u 15 mA 324.8 nm Furnace programme for Cu- Furnace settings Dry 1 Temperature/"C 120 Ramp time/s 1 Hold time/s 20 Ar flow-rate/ml min-' 250 Air 8.0 8.0 8.0 Slit-width 0.7 nm Dry 2 140 5 40 250 Waveleng t h/nm 279.5 248.3 285.2 Aliquot Absorbance Tube volume Peak area THGA 20 pl Ash Atomize 1100 2050 10 0 20 5 250 0 Slit-width/nm 0.4 0.2 0.5 Modifier 0.03 mg Mg(NO3 )2 Clean 2400 1 2 250 The samples and acid were all diluted with de-ionized distilled water.Each stock standard solution of the test element containing 1000 mg 1-1 of the element in the nitrate form for AAS was obtained from BDH (now Merck Poole Dorset UK). All working standard solutions were prepared by immediate serial dilution with 0.5% v/v nitric acid solution. Microwave Extraction About 0.3g of NIST SRM 1515 Apple Leaves was weighed into a PTFE extraction vessel. After adding 5ml of dilute nitric acid the vessel was shaken for 3 min to allow for thorough mixing. The vessel was then capped and placed on the turntable. The microwave oven was operated according to the settings listed in Tables 2 and 3. Each trial had its own blank with the same amount of dilute nitric acid and microwave programme.After extraction the vessel was vented when the inner pressure was under 138 kPa (20 psi). The extract was filtered and then diluted to 15 ml with de-ionized distilled water. Assignment of Experiments For the extraction of biological samples using the closed-vessel microwave extraction procedure the most important variables are the extraction programme (power and time pressure and time) the power output the number of samples extracted the type and the concentration of acids used and the sample mass.'l In the pressure feedback microwave extraction pro- cedure the sample dissolution procedure was controlled by the pressure and power setting pr~gramme.~*~'.~' When a biological sample is extracted with strong oxidizing acids the power setting programme can be used to reduce the volatile reaction. Dilute nitric acid is not a strong oxidizing acid hence the maximum power output (100%) and rapid extraction programme were used.The number of samples extracted was four. Hence the following four variables were chosen in order to obtain the optimum trace element recoveries (i) the concen- tration of nitric acid (variable A); (ii) the pressure setting (variable B); (iii) the extraction time (variable C); and (iv) the mass of sample. The effect of sample mass was studied separ- ately as its interaction with the other variables. The other variables were evaluated by a mixed level orthogonal array design analysis. Orthogonal arrays are a classical design of experirnenk2' In forms orthogonal arrays are different from the way these arrays are usually displayed in the statistical literature.Orthogonal array design is in fact a saturated fractional factorial design. The precursor of orthogonal array design was Latin and Graceo-Latin squares,17 the theory of which was described in the 1940s by Ra022923 and B o ~ e . ~ ~ It was successfully applied in the engineering area for quality control by T a g ~ c h i . ~ ~ The theory and methodology of ortho- gonal array design as a chemometric method for optimization of analytical procedures have been discussed in previous p a p e r ~ . ' ~ - ~ ~ In orthogonal array designs orthogonal arrays are used to assign factors to a series of experiment combinations the results of which can then be analysed by using a common mathematical procedure.The main effects of the factors and preselected interactions are independently extracted. In an orthogonal array different combinations of numerals of any two columns have an equal appearance frequency. Here ortho- gonal means balanced. By arranging experiments orthogonally different effects can be separated. This method combines the advantages of both the simplex method and factorial design. Much more information can be obtained from a limited number of experirnent~.l~-'~ The method has been applied in liquid chromatography,26 gas ~hromatography~~ and micro- wave digestion of sediment samp1es.l' In this work the concen- tration of nitric acid (variable A ) was the most important variable considered. It was evaluated in a four-level design. The pressure setting (variable B) and extraction time Table 2 Assignment of factors and level settings for evaluating dilute nitric acid extraction of biological samples in the OA,,(4I x 212) matrix* Column No.Level 1 2 3 4 5 15 7 8 9 10 11 12 13 A B ( A x B) (A+B)z ( A X B)3 4 ( A X C) ( A X C) ( A x C)3 B x C 1 2% 1104 30 2 14% 552 15 3 10% 4 6 Yo * A Concentration of dilute nitric acid (v/v); B pressure setting (kPa); C extraction time (min). 586 Journal of Analytical Atomic Spectrometry August 19!36 Vol. 11Table 3 OA1,(4' x 212) matrix and experimental results ~- Column No. Trial 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 K1 K2 K3 K4 1 1 1 1 1 2 2 2 2 3 3 3 3 4 4 4 4 242.8 382.2 364.8 291.6 2 1 1 2 2 1 1 2 2 1 1 2 2 1 1 2 2 649.1 632.3 3 1 1 2 2 1 1 2 2 2 2 1 1 2 2 1 1 646.5 634.9 4 1 1 2 2 2 2 1 1 1 1 2 2 2 2 1 1 649.9 631.5 5 1 1 2 2 2 2 1 1 2 2 1 1 1 1 2 2 651.7 629.7 6 1 2 1 2 1 2 1 2 1 2 1 2 1 2 1 2 661.9 619.5 7 1 2 1 2 1 2 1 2 2 1 2 1 2 1 2 1 643.3 638.1 8 1 2 1 2 2 1 2 1 1 2 1 2 2 1 2 1 648.7 632.7 9 1 2 1 2 2 1 2 1 2 1 2 1 1 2 1 2 654.5 626.9 10 1 2 2 1 1 2 2 1 1 2 2 1 1 2 2 1 646.4 635.0 11 1 2 2 1 1 2 2 1 2 1 1 2 2 1 1 2 638.0 643.4 12 1 2 2 1 2 1 1 2 1 2 2 1 2 1 1 2 641.0 640.4 13 1 2 2 1 2 1 1 2 2 1 1 2 1 2 2 1 643.8 637.6 CO-R 71.5 58.5 61.3 51.5 95.0 94.6 96.6 96.0 93.6 89.0 92.0 90.2 77.9 69.0 74.0 70.7 (variable C) were evaluated in a two-level design.Because one four-level variable and two two-level variables were to be optimized and their interactions to be considered an 0A16(4' x 212) matrix was employed to assign variables and their interactions.Following a randomization process for vari- ables the level settings of the variables studied are listed in Table 2. The randomization process for level setting was important in reducing the errors from personal inclination on level assignment. The interactions were presented as A x B A x C and B x C . The positions of the variables chosen and level settings used in the mixed level orthogonal array design are displayed in Table 3. The column assignment was from Table 2. Column trials show the 16 experiments to be carried out. The variable settings are represented by 1 2 3 and 4. The composition response (Co-R) is listed in the last column. The Krn (rn = 1-4) values in the last four rows are the sum of Co- R with m level of variable assigned in the column.Krn is used to evaluate the effect of the variable. The analysis of variance (ANOVA) technique was employed to estimate the variable effects. For evaluating the effects of the variables at different levels from 16 experimental trials the recoveries of Mn Fe Mg and Cu in extracts were used as response functions. Their concen- trations in the extracts were determined by FAAS or ETAAS. For each test element the output response of one test element (Relement) is defined as Relement = accuracy of test element (%) element concentration from experiment x 100% certified concentration given by SRM 1515 The Co-R of the four elements was also used to judge the quality of the microwave extraction. It was calculated as - RESULTS AND DISCUSSION Considering the interactions of the selected variables three of them were evaluated by a mixed level orthogonal array design analysis OAl6(4l x 212).The results of Relement of Mn Fe Mg and Cu in the 16 experimental trials are listed in Table 4. The results of Co-R are listed in Table 3. The significance of different variables was quantitatively evaluated by the ANOVA tech- nique for Co-R. The ANOVA analysis results and F-test are Table4 Recoveries of the four elements used for evaluating the extraction efficiency in the 16 OA16(4' x 2") trials Trial 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Mn 88.0 85.6 83.7 82.0 95.0 96.3 93.9 88.7 91.9 90.7 92.6 91.1 85.6 83.3 89.4 90.2 Fe 87.4 45.2 51.9 26.9 92.2 88.6 93.6 90.2 80.4 76.1 83.1 71.2 64.7 73.0 67.0 100 Mg 93.3 90.0 87.1 86.6 96.5 93.3 99.0 96.1 92.4 100 100 95.3 97.7 91.2 95.8 88.9 c u 17.3 13.2 22.3 10.4 96.1 100 100 100 99.2 84.8 99.1 91.1 62.1 36.7 37.7 36.8 shown in Table 5.The sum of the squares are the statistical results from the Co-R for different variables or interactions. The equations of the sum of the squares and other items in the ANOVA table for testing the magnitude of the different variables in orthogonal array design were given in a previous paper.17 From the analysis it is seen that the concentration of nitric acid (A) and the extraction time (C) are statistically significant at p < 0.01. The pressure setting (B) and the inter- actions of A x B and A x C are significant at p < 0.05. The interaction of B x C is significant at p < 0.25. The effect of sample mass was studied later after the other variables had been optimized.Effect of the Concentration of Nitric Acid Most methods for the microwave extraction or digestion of biological samples have employed concentrated nitric acid or acid mixtures with strong oxidizing On the other hand to avoid the violent reaction of acids with the organic samples in closed vessels it is necessary to use the time and power or time and pressure pr~gramming.',~~ The violent reaction would be reduced if dilute nitric acid was employed because of its low oxidizing power. One of the purposes of this work was to identify the concentration of dilute nitric acid that could decompose the organic matrix within a short time Journal of Analytical Atomic Spectrometry August 1996 Vol.1 1 587Table 5 Results of ANOVA analysis of Co-R in the OA16(4' x 212) matrix 100 - 2 sf. 0 5 0 5 0 - E! E Source A B C A x B A x C B x C Error h M n - Fe - cu - Mg Sum of squares 3 160.4475 17.6400 112.3600 58.8200 65.3000 8.1225 4.247 5 - CO-R Degrees of freedom 3 1 1 3 3 1 3 Mean square F 1057.4825 774.09* 17.6400 12.467 112.3600 79.365 19.9400 14.089 21.7667 15.379 8.1225 5.747 1.4158 Best level 2 1 1 A B and A B1 A,C and A,C B1 c * p = 0.01 (99% confidence level) F(3,3) = 29.46. t p = 0.05 (95% confidence level) F(3.1) = 9.28. 5 p = 0.01 (99?'0 confidence level) F(3.1) = 34.12. 9 p = 0.05 (95% confidence level) F(3,3) = 10.17. 7 p = 0.25 (75% confidence level) F(3.1) = 2.02. and keep the violent reaction to a minimum. Four different concentrations of nitric acid were evaluated in the OA16(4' x 212) matrix. The ANOVA results of Co-R indicated that the concentration of nitric acid strongly influenced the sample decomposition.Fig. 1 shows the relationship between the output response of four test elements (Relement) and Co-R and the concentrations of nitric acid. Dilute nitric acid at 2 and 6% v/v did not completely destroy the organic sample (some solid substances were still present in the extract) and gave relatively low recoveries of the test elements. Relatively good Relement values were obtained with concentrations of nitric acid of 10 and 14% v/v. Considering the complicated matrix of the organic samples 14% v/v nitric acid was selected. Effect of Pressure Pressure is a very important parameter in pressure feedback microwave extraction.It is used to control the decomposition reaction in a closed ve~sel.'~ Fig 2 shows the pressure curves of trials 5 and 9. There was no overshooting of the pressure control setting point. The decomposition reaction was not violent and the pressure control was efficient in controlling the reaction. The effect of pressure setting was found to be statistically significant at p < 0.05 from ANOVA results of Co-R. The best element recoveries were obtained at a pressure setting of 1104 kPa (160 psi). On the other hand the effect of the interaction between the concentration of nitric acid (A) and the pressure setting (B) was also important. Table 6 shows the interaction of variables A and B which was obtained from the sum of Co-R values containing the different level settings of variables A and B in 16 trials.The high pressure setting was useful in improving the recoveries of test elements when samples were extracted with nitric acid at very low concen- tration. When the concentrations of nitric acid were at 10 and 14% v/v the effect of high pressure settings was insignificant. Fig. 1 Effect of the concentration of nitric acid on sample extraction 1200 s *O0 2 v) 400 * Trial9 0 = I I 1 0 10 20 30 Time/min Fig. 2 nitric acid Relation between pressure and extraction time using 14% (v/v) Table 6 Four-by-two table of the interaction between the nitric acid concentration (variable A) and the pressure setting (variable B) Concentration of nitric acid level Pressure level 1 2 3 4 1 130.0 189.6 182.6 146.9 2 112.8 192.6 182.2 144.7 Considering the fact that some organic samples would be more difficult to decompose than SRM 15 15 Apple Leaves a pressure setting of 1104 kPa (160 psi) was also used in subsequent sample extractions. Effect of Extraction Time For most element determinations the sample dissolution step is time consuming.In the normal sample extraction by the dilute nitric acid method it takes several hours to a day to obtain the homogenate ~ o l u t i o n . ~ ~ * ~ ~ One of the advantages of closed-vessel microwave extraction is the reduction of the extraction time. In this work the extraction time was evaluated at two levels of 30 and 15 min respectively. From the results of ANOVA analysis of Co-R the effect of the extraction time was statistically significant at p < 0.01.The best recoveries of the test elements were obtained with an extraction time of 30 min. The interactions of A x C and B x C were also signifi- cant at p < 0.05 and 0.25 respectively. Table 7 shows that the use of a 30min extraction improved the element recoveries when very low concentrations of nitric acid were employed. When 14% v/v nitric acid was used there was no difference in the element recoveries between the two levels. The best combi- 588 Journal of Analytical Atomic Spectrometry August 1996 Vol. 11Table 7 Four-by-two table of the interaction between the concen- tration of nitric acid (variable A ) and the digestion time (variable C ) Concentration of nitric acid level Time level 1 2 3 4 1 132.8 191.6 185.6 151.9 2 110.0 190.6 179.2 139.7 nation of variables B and C was B C1 (Table 8).Tables 7 and 8 were obtained in a similar way to Table 6. An extraction time of 30 min was used in subsequent experiments. Effect of Sample Mass According to the mixed level orthogonal array design analysis OA,,(4' x 212) the concentration of dilute nitric acid the pressure setting and the extraction time were optimized. Under the optimized conditions different sample masses were extracted in several groups. The recoveries of the four test elements are listed in Table 9. Using a mass of 0.4 g or less resulted in complete extraction by 5 ml of 14% v/v nitric acid as shown by the good recoveries obtained. When the sample mass was over 0.5 g the recoveries were reduced. Also some solid substances were present in the extract solutions.A sample mass of 0.3 g is thus recommended. Table 8 Two-by-two table of the interaction between the pressure setting (variable B) and the extraction time (variable C) Extraction time level Pressure level 1 2 1 332.6 324.0 2 306.2 308.4 Table 9 Recoveries (YO) of the four test elements using different masses of NIST SRM 1515 Apple Leaves (n = 3) Mass of sample/g Mn Fe Mg c u 0.2 9 7 f 2 101f3 102f3 99 * 2 0.3 98 f 5 9 6 f 4 9 9 f 4 102f3 0.4 9 6 f 5 100f4 101f6 96 f 5 0.5 98 f 5 8 6 f 9 9 2 f 4 8 1 f 7 0.6 8 9 f 6 83+9 9 2 k 6 63 f 9 Application Six replicate analyses of NIST SRM 1515 Apple Leaves were carried out under the optimized conditions. Ca Cu K Fe Mg Mn Ni and Zn were determined by FAAS or ETAAS. The results are listed in Table 10.Recoveries of 96-103% were obtained. With the exception of Ni the RSDs of the test elements were less than 3.8%. The optimized parameters were applied to the extraction of different biological samples namely NIST SRM 1557b Bovine Liver and CRMs No. 1 Pepperbush No. 3 Chlorella and No. 6 Mussel from the National Institute for Environmental Study (NIES) of Japan. The results obtained are summarized in Table 10. With the exception of Ni recover- ies of 91-107% were obtained for the eight test elements. CONCLUSION The microwave extraction technique can be used to improve the dilute nitric acid extraction of biological samples. By using a rapid extraction programme viz. 100% power output 1104 kPa (160 psi) pressure and a 30 min extraction time biological samples of 0.3 g were completely extracted with 5 ml of 14% v/v nitric acid.This extraction method is also suitable for other biological samples. C.Y.Z. thanks the National University of Singapore for a research scholarship. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 Van Loon J. C. Selected Methods of Trace Metal Analysis Biological and Environmental Samples Wiley-Interscience New York 1985 pp. 77-141. Hoke F. Scientist 1992 6( 14) 19. Matusiewicz H. Sturgeon R. E. and Berman S . S. J. Anal. At. Spectrom. 1991 6 283. Bettinelli M. Baroni U. and Pastorelli N. Anal. Chim. Acta 1989 225 159. de la Fuente M. A. and Juhrez M. Analyst 1995 120 107. Yang Q. Penninckx W. and Smeyers-Verbeke J. J. Agric. Food Chem. 1994 42 1948. Krushevska A. Barnes R. M. and Amarasiriwaradena C.Analyst 1993 118 1175. Lajunen L. H. J. Piispanen J. and Saari E. At. Spectrosc. 1992 13 127. Baldwin S. Deaker M. and Maher W. Analyst 1994 119 1701. Tahan J. E. Granadillo V. A. Sanchez J. M. Cubillan H. S. and Romero R. A. J. Anal. At. Spectrom. 1993 8 1005. Kingston H. M. and Jassie L. B. Anal. Chem. 1986 58 2534. Mohd A. A. Dean J. R. and Tomlinson W. R. Analyst 1992 117 1743. Temminghoff E. J. M. and Novozamsky I. Analyst 1992,117,23. Table 10 Analytical results (pg g-') for NIST SRM and NIES CRM biological samples (n = 4) Element Ca Found Cu Found Fe Found K Found Mg Found Mn Found Ni Found Zn Found Certified Certified Certified Certified Certified Certified Certified Certified NIST SRM* 1515 15 100 & 200 15 260 f 150 5.5 f 0.1 5.6 f 0.2 8 0 k 3 8 3 f 5 16 600 f 300 16 100 f 200 2680 f 60 2710 & 80 5 2 k 2 54rfr 3 0.9 k 0.2 0.91 f 0.12 12.7 -t 0.3 12.5 f 0.3 NIST SRM 1577b 107 f 3 116f4 170 f 4 160 f 8 192 f 5 184 f 15 10 200 f 200 9940 f 20 600 f 40 601 f 28 11.0 f 0.8 10.5 f 1.7 - 118f4 127 f 16 NIES CRM No.1 14 200 f 500 13 800 f 700 12.0 k 0.7 1 2 k 1 219 f 4 205 f 17 14 800 f 100 15 100 & 600 4030 & 100 4080 f 200 2000 f 60 2030 f 170 8.5 f 0.5 8.7 k 0.6 343 f 11 340 f 20 NEIS CRM No. 3 5010 f 170 4900 f 300 3.6 0.2 3.5 f 0.3 1680 50 1860 f 100 11 800 f 100 12 400 _+ 60 3380 f 90 3300 k 200 6 3 f 2 69 & 5 0.6 f 0.2 20.4 f 0.3 20.5 f 1.0 - NIES CRM No. 6 1190 f 70 1300 f 100 4.7 & 0.2 4.9 f 0.3 158 f 7 158 f 8 5160 f 80 5400 f 200 2020 f 30 2100 f 100 16.0 k 0.8 16.3 rfr 1.2 1.1 & 0.1 0.93 f 0.06 9 9 k 2 106 f 6 * Six replicate analyses. Journal of Analytical Atomic Spectrometry August 1996 Vol. 1 1 58914 15 16 17 18 19 20 21 22 Asp T. N. and Lund W. Talanta 1992 39 563. Niazi S. B. Littlejohn D. and Halls D. J. Analyst 1993 118,821. Chow P. Y. T. Chua T. H. Tang K. F. and Ow B. Y. Analyst 1995 120 1221. Lan W. G. Wong M. K. Chen N. and Sin Y. M. Analyst 1994 119 1659. Lan W. G. Wong M. K. Chen N. and Sin Y. M. Analyst 1994 119 1669. Zhou C. Y. Wong M. K. Koh L. L. and Wee Y. C. Anal. Chim. Acta 1995 314 121. Hasty E. T. Littau S. E. and Revesz R. paper presented at the Pittsburgh Conference and Exhibition 1991. Kacker R. N. Lagergren E. S. and Filliben J. J. J. Res. Natl. Inst. Stand. Technol. 1991 96 577. Rao C. R. Bull. Calcutta Math. SOC. 1946 38 67. 23 Rao C. R. Proc. Edinburgh Math. SOC. 1947 8 119. 24 Bose R. C. Sankhya 1947 8 107. 25 Taguchi G. System of Experimental Design Kraus International Publishers New York 1987 vols. 1 and 2. 26 Wan H. B. Lan W. G. Wong M. K. and Mok C. Y. Anal. Chim. Acta 1994 289 371. 27 Oles P. J. J . Assoc. Off Anal. Chem. 1990 73 724. 28 Lamble K. and Hill S . J. Analyst 1995 120 413. 29 de Bore J. L. M. and Maessen F. J. M. J. Spectrochim. Acta Part B 1983 38 739. Paper 61001 49A Received January 8 1996 Accepted May 15 1996 590 Journal of Analytical Atomic Spectrometry August 1996 Vol. 11