Automatic Preparation of Milk Dessert Slurries for the Determination of Trace Amounts of Aluminium by Electrothermal Atomic Absorption Spectrometry MARCO A. z. ARRUDA MERCEDES GALLEGO AND MIGUEL VALCARCEL* Department of Analytical Chemistry Faculty of Sciences University of Cbrdoba E- 14004 Cdrdoba Spain A method for the direct determination of aluminium in milk desserts is proposed. The method uses a flow injection system connected semi-on-line to a graphite furnace atomic absorption spectrometer through an autosampler for preparation of the slurries addition of a chemical modifier and dilutiow homogenization in a mixing chamber. Calibration is performed with aqueous standards. The flow system can be used to implement the standard additions method and investigate the effect of potential interferents in parallel by using two injection loops and the merging-zones mode.The slurries prepared contained up to 10% m/v. Aluminium was determined in various milk desserts at concentrations between 0.007 and 0.121 mg per 100 g. Keywords Aluminium determination; milk dessert analyses; electrothermal atomic absorption spectrometry; slurry sample introduction; flow injection Interest in the potential link between high aluminium contents in tissues and various neurodegenerative disorders such as Alzheimer's disease has drawn attention to the intake of aluminium from food drinking water parenteral nutrition or dialysis fluids in individuals with chronic renal disease.' There are several compilations of aluminium contents in foods and the most comprehensive of which was published by Pennington.2 Based on a total dietary study (TDS) on foods commonly consumed in the USA the main contributor to the aluminium content of the American diet is seemingly the grain and cereal product group followed by the dairy product group.The aluminium content increases from fluid cow milk to dairy products (e.g. yoghurt butter cheese); the mean aluminium content in yoghurt ranges from less than 0.1 to 0.112mg per 1OOg on a wet basis.4 Delves et ~ 1 . ~ recently studied the aluminium content in UK foods and beverages and found the relative bioavailability of the element from these foods to vary widely as does uptake by adult individuals. Because of its typically low content the most commonly used techniques for measuring aluminium in foods are electrothermal (graphite furnace) atomic absorption spectrometry (ETAAS) and induc- tively coupled plasma atomic emission spectrometry (ICP- AES).1,2 Direct analysis by ETAAS has previously been used with liquid foods such as milk fruit juices drinks and infusions; many solid foods require acid digestion (ideally with concen- trated nitric acid).However acid digestion of foods with high fat contents (e.g. dairy products) is potentially hazardous. One alternative to solid food pre-treatment is direct introduction of the sample as a slurry viz. a suspension of finely powdered sample in water or another solvent. This technique minimizes or avoids some typical problems encountered in sample pre- * To whom correspondence should be addressed. Journal of Analytical Atomic Spectrometry treatment such as contamination or analyte However running calibrations or using the standard additions method with solid or slurry samples still poses serious problem^.^ Aluminium has been widely determined in foods by ETAAS,1-3 but very rarely by using a The main interferences in the determination of aluminium in foodstuff by ETAAS arise from volatilization losses of Al,Cl from chloride-rich media and variable enhancements of the aluminium atomic signal by some ions present in foods (e.g.Po43- SO4,- Ca2+) and carbonaceous residues formed during electrothermal decomposition.' These interferences can be magnified by the slurry technique. A variety of chemical reagents [e.g. Mg(N03)2 Pd(N03) K2Cr207 NH4N03] have been used to decrease matrix interferences as far as possible prior to the atomization step in food analyse~;'~-'~ in any event the palladium salt is seemingly gaining wide accept- ance as a modifier for many elements including aluminium.'6-'8 The flow injection (FI) technique has been widely used in connection with flame atomic absorption spectrometry ( FAAS)19-22 and ICP-AES23 for the determination of metals in food slurries but comparatively rarely with ETA AS owing to the difficulties involved in their coupling (in fact only semi- on-line FI systems have so far been combined with ETAAS in~trumentationl~.~~).The aim of this work was to avoid sample weighing dilution and homogenization in preparing milk dessert slurries by using an automated preparation module that allows aqueous stan- dards to be automatically inserted in such a way that Cali- bration (whether conventional or by use of the standard additions method) can be implemented automatically with no sample weighing.The module was coupled to the instrument autosampler for determining aluminium in milk desserts. EXPERIMENTAL Apparatus A Perkin-Elmer (Uberlingen Germany) Model 1100 B atomic absorption spectrometer fitted with an HGA-700 graphite furnace and an AS-70 furnace autosampler was used. Background absorption was corrected by using a deuterium lamp in all experiments. A Perkin-Elmer aluminium hollow- cathode lamp operated at 25 mA and pyrolytic graphite-coated tubes (Perkin-Elmer part No. B-013-5653) with L'vov plat- forms (Perkin-Elmer part No. B-012-1091) were also used.Aluminium atomic absorption was measured at 309.4 nm by using a 0.7 nm spectral bandpass. Atomization signals (in the peak-area mode) were printed on a Epson (Wembley UK) FX-850 printer. The temperature programmes used are described in Table 1. The FI system was constructed from a Gilson (Villiers-Le-Bel France) Minipuls-2 peristaltic pump fitted with poly(viny1 chloride) tubing a laboratory-made three-piece injector commutator12 and a customized 1 ml Journal of Analytical Atomic Spectrometry January 1995 Vol. 10 55Table 1 Furnace conditions used to obtain the pyrolysis-atomization curves in the analysis of yoghurt slurry samples. The injected volume was 20 pl of sample + 10 p1 of modifier for the conventional determination For optimization Final choice Step Temperature/”C Hold/s 1 100 35 2 200 15 3 800 5 4 Variable 10 5 Variable 4 6 2650 1 Temperature/”C Hold/s 100 35 200 15 800 5 1700 15 26QO 4 26.50 1 Ar flow rate/ Ramp/s ml min-’ 10 300 15 300 10 300 15 300 0 0 1 300 PTFE mixing chamber described el~ewhere.’~ A Heidolph magnetic stirrer was used to homogenize samples.Reagents All reagents used were of analytical-reagent grade (Merck Darmstadt Germany) and high-purity water obtained with a Milli-Q water-purification system (Millipore) was employed throughout. A lOOOmgl-’ A1 stock standard solution was prepared by dissolving 1.OOO g of aluminium wire in 20 ml of concentrated H2S04 plus 50ml of concentrated HN03 and diluting to 1 1 with water. Working standard solutions contain- ing 40-400 pg 1-’ of A1 were prepared from the stock standard solution by serial dilution with 0.2% v/v HNO prior to use.The standards were diluted tenfold in the FI manifold. Several modifiers including aqueous magnesium nitrate palladium nitrate and lanthanium nitrate dissolved at concen- trations between and 0.1 mol 1-’ were tested. Cleaning and Storage Material All PTFE vessels were cleaned by soaking in 10% v/v HNO for 48 h rinsing five times with water and filling with water until use.12 No glass vessels were used in order to minimize aluminium release and adsorption. Yoghurt samples were stored refrigerated (4 “C) in their plastic packages until analysis. Sample Preparation Milk dessert samples were preliminarily homogenized manu- ally with a spatula for 2 min. Samples containing fruit pieces were ground to complete homogeneity in a blender. Three different sample preparation procedures were used as follows.(1) Dry ashing a 2.5 g portion of natural yoghurt was placed in a platinum crucible and evaporated to dryness in a sand- bath followed by ashing at 600 “C for 40 min. The ash was then extracted to 25 ml with 0.2% v/v HN03. (2) Wet ashing a 2.5 g portion of natural yoghurt was placed in a platinum crucible and mixed with 5 ml of concentrated H2S04 plus 5ml of concentrated HN03 and evaporated to dryness in a sand-bath; after evaporation a fresh 5 ml portion of concentrated H2S04 plus 5 ml of concentrated HNO was added followed by evaporation to dryness twice more. Finally the resulting residue was dissolved to 25ml with 0.2% v/v HNO .(3) Direct preparation of yoghurt slurries by diluting 2.5 g of yoghurt to 25 ml with 0.2% v/v HNO and homogenizing in an ultrasonic bath for 10 min. For aluminium measurements diluted samples were placed in autosampler cups. A reagent blank was also prepared simultaneously in procedures 1 and 2. The National Institute of Standards and Technology (NIST) Standard Reference Material (SRM) SRMl549 Non-fat Milk Powder was dried to constant mass by freeze-drying at 6 Pa for 24 h after which an accurately weighed amount of approxi- mately 1.25 g was mixed with 25 ml of 0.2% v/v HN03. The slurry thus formed was shaken before analysis in order to homogenize the solid particles in the solution. Conventional Procedure In order to obtain a calibration graph 2Opl of standard solution containing 5 10 20 30 or 40 pg I-’ of A1 (or diluted yoghurt sample) and 10 pl of 0.1 moll-’ magnesium nitrate plus 0.01 moll-’ palladium nitrate as chemical modifier were introduced sequentially into the graphite furnace. The blank solution consisted of 20 pl of 0.2% v/v HNO and 10 pl of the chemical modifier; both this and a standard solution containing 15 pg 1-1 of A1 (for re-sloping) were run after every five samples.All measurements were made in triplicate. Accuracy and Precision of Slurry Deposition The accuracy and precision of slurry sample deposition into a graphite tube were determined from recovery studies according to Lynch and Littlejohn.I8 The recovery studies give an estimate of the accuracy of the autosampler deposition if the programmed volume is taken as a true volume.The AS-70 autosampler was programmed to inject 20 pl of slurry. The mass of slurry actually introduced into the graphite furnace was determined by weighing the autosampler cups before and after sample deposition. The volume of slurry deposited was then calculated by dividing the mass into the slurry density. The mean value and precision (sr YO) were calculated in several slurry volume determinations (n = 8 ) ; the accuracy was calcu- lated by dividing the mean slurry volume deposited into the programmed volume. Flow Injection Procedure The flow injection system used is depicted in Fig. 1. Initially the manually homogenized milk dessert or the standard was aspirated through loop L at a flow rate of 1.5 ml min-’ while the chemical modifier C0.l mol 1-1 Mg(N03),+0.01 moll-’ Pd(N03)2] and the carrier (0.2% v/v HN03) were recycled outside the measurement line of the injector commutator.As the commutator was switched to its alternative position 200 pl of standard or sample were inserted into the carrier stream at a flow rate of 0.7 ml min-’ and then merged with the chemical modifier at 0.3 ml min-’ at merging point X; the mixed solution was passed through the mixing chamber for dilution and homogenization of the slurry. After mixing the solution was loaded into an autosampler cup for 120 s (2 ml). The ratio of the flow rate of the carrier to that of the chemical modifier used allowed the required dilution (1 +9) of the slurry sample for the determination of aluminium. In order to implement the standard additions method and investigate potential interferences with the proposed method two loops (one for the sample and the other containing different standard or interferent ions concentrations) were included in the FI system; also the original carrier stream (0.2% v/v 56 Journal of Analytical Atomic Spectrometry January 1995 Vol.10c T L W - 1 I g I @ IC Yoghurt d MC ETAAS v Fig. 1 FI manifold used for the determination of A1 in milky desserts M chemical modifier; C carrier stream; L sample loop (200 p.1); IC injector-commutator; MC mixing chamber; P peristaltic pump; W waste. Flow rate of carrier and chemical modifier 0.7 and 0.3 ml min- ' respectively HN03) at 0.7 ml min-l was split into two streams circulating at 0.35 ml min-l. Therefore as the commutator was switched the contents of both loops were injected into the carrier stream and the sample and standard/interferent were mixed in the merging-zones mode.The operational sequence was then continued as stated above. RESULTS AND DISCUSSION Optimization of the Conventional Procedure Errors in volumetric sampling of slurries relative to aqueous solutions can be attributed to several sources; some arise from the slurry (e.g. viscosity density) and others from the sample injector (e.g. wettability material composition). These physico- chemical properties affect the accuracy and precision of sample deposition and hence must be controlled for complete optimiz- ation of the operational procedure method. The slurry sample introduction process was studied by manually shaking directly prepared slurries prior to manual injection in order to prevent settling out of material. The accuracy and precision with which natural yoghurt slurry was deposited into a graphite furnace up to a concen- tration of 50% m/v is shown in Fig. 2.The accuracy was higher than 100% in all instances so the slurry volumes deposited by the autosampler injector were always higher than the programmed volume (20 pl); however above a 10% slurry concentration the mean volume remained constant at 20.4 pl. This increased volume may be the result of a higher density 110 90 95 [ 0 5 10 20 50 Slurry concentration (% m/v) Fig. 2 Accuracy and precision of slurry introduction into a graphite furnace. Error bars indicate 95% confidence intervals (for precision as sr) of the slurries relative to distilled water.The precision of sample deposition was acceptable (s < 8%); however it became worse with decreasing slurry concentration probably as a result of a steeper concentration gradient in the bulk slurry. Based on the results the injector can be used for introduction of slurry samples containing up to 50% m/v with acceptable accuracy and precision. The furnace programme used is shown in Table 1; it included two temperature ramps in the drying step in order to prevent splattering of the sample on the L'vov platform owing to the high boiling-point of the matrix. The pyrolysis step also included two temperature ramps in order to overcome the potential effect of the carbonaceous residue by exploiting the low volatility of aluminium.Maximum sensitivity was achieved by stopping the internal argon flow through the furnace during the atomization step. Under these conditions the signal-to- noise ratio was optimum and the background absorbance was very small and corrected by the deuterium lamp. Because direct measurements of aluminium in the yoghurt samples provided high analyte concentrations and background signals (approximately 0.3 absorbance per second) both of which are limiting factors in considering the extent of dilution factor first the yoghurt dilution was studied. Different dilution factors were tested (1 + 1 1 +4 1 + 9 and 1 + 19 m/v) for a commercially available natural yoghurt. A dilution factor of 1 +9 m/v was chosen as optimum because it resulted in a lower background signal than 1 + 1 and 1 +4 dilutions and provided an aluminium concentration in the central region of the calibration graph.Aluminium oxides carbides cyanamides and cyanides have been detected in the furnace near the atomization temperature of aluminium; these compounds may result in surface losses of aluminium and account for the relatively low atomization efficiency for aluminium.25 If these aluminium-containing species are dissociated at temperatures other than the atomiz- ation temperature they may give rise to broadened or distorted analyte signals. Chemical modifiers are an excellent choice for circumventing these problems. We used a natural yoghurt sample diluted 1 + 9 and a standard solution of 20 pg I-' of A1 plus various chemical modifiers (Fig. 3) at a fixed atomiz- ation temperature of 2600 "C and various pyrolysis tempera- tures.In the absence of a chemical modifier the aluminium present in the natural yoghurt started to volatilize at about 1100°C and the standard solution at 1500°C. In the presence of a chemical modifier however the pyrolysis temperature for the yoghurt was similar to that for the aqueous aluminium standard solution. Although differences in the background signals obtained in the presence of the different chemical modifiers were very small 0.1 moll- Mg( NO& plus 0.01 moll-' Pd(N03)2 was chosen as the optimum modifier as it resulted in a slightly higher sensitivity for the sample. A 0.01 mol 1-1 La(NO,) solution also tested as a chemical modifier was discarded because the aluminium sensitivity started to decrease above about 1500 "C.Flow Injection System After manual homogenization the yoghurt sample was aspir- ated into a 200 pl sample loop. The sample aspiration flow rate was critical; the optimum pump tubing diameter-rotation speed combination was 2.3 mm id. and 200 rev min-l which allowed the loop to be filled with the yoghurt sample at 1.5 ml min-'. In order also to use the optimum dilution factor ( 1 + 9 m/v) for the conventional procedure the sample volume injected into the FI system (200~1) was diluted to a final volume of 2 ml in the autosampler cup. The modifier flow rate (0.3mlmin-') should be roughly half that of the carrier solution (0.7 ml min-I) in order that the ratio of diluted sample volume introduced into the furnace (20pl) to the chemical Journal of Analytical Atomic Spectrometry January 1995 VoE.10 570.15 la’ I I b 900 1300 1700 2100 Pyrolysis temperature/”(= Fig. 3 Influence of pyrolysis temperature of (a) natural yoghurt manually diluted tenfold and (b) 20 pg 1-1 of A1 standard A in the absence of chemical modifier and B in the presence of 0.1 mol I-’ Mg(N03)2; C 0.01 moll-’ Pd(N03)2; and D 0.1 mol 1-1 Mg(N03)2 plus 0.01 mol 1-1 Pd(N03)2. Injected volumes 20 pl (sample) and 10 p1 (chemical modifier). Furnace conditions as in Table 1 modifier volume (10 pl) for the conventional procedure be identical with that used in the FI procedure. These flow rate values ensure negligible carry-over of sample into the FI system. Finally the potential advantages of the automatic slurry preparation method over its manual counterpart were checked by determining the precision achieved with both.Thus ten slurries of the same yoghurt were prepared by manual dilution (1 + 9 m/v) and homogenized in the ultrasonic bath for 10min or in the proposed FI system where ten sequential injections of 200 pl of the same yoghurt were performed. The results obtained showed the manual procedure to be poorly reproducible (the repeatability as s was 15.8%) and the automatic dilution procedure to be acceptably precise (the repeatability as s was 6.5%) because of its reduced human participation (no weighing dilution or homogenization of the sample is needed). For comparison it should be noted that 200 pl of yoghurt weighed about 205 mg as the yoghurt density was about 1.025 g m1-I.Determination of Aluminium By using the proposed method aluminium could be determined over the linear range from 4 ,to 40 pg 1-1 (r>0.998; n=6); however the FI system depicted in Fig. 1 increased the linear range from 40 to 400 pg I-’ owing to the dilution factor (1 + 9) used. The detection limit calculated according to IUPAC recommendations,26 was found to be 6 pg 1-1 (0.58 pg per 100 g) for the FI system. The results obtained for commercially available natural yoghurt were precise [s = 6.2% as repeat- ability (n = 15) and s = 10.6% as reproducibility (n= lo)]. A calibration graph for slurries was constructed by using the standard additions method; for this purpose a second 200 pl loop for injection of aqueous standards containing 25-250 pg I-’ of A1 was incorporated into the FI system as described under Flow Injection Procedure.Both the natural yoghurt and the standards were diluted tenfold inside the manifold. The high consistency between the aluminium concen- trations in the diluted natural yoghurt obtained by using the standard additions (0.01 10 + 0.0003 mg per 100 g) and FI (0.0109 +0.0002 mg per 100 g) methods indicates that the proposed method can be calibrated with liquid standards. A t-test was made to compare the concentrations of aluminium provided by the FI and standard addition procedures (95% confidence interval) and no significant difference was observed. Selectivity Potential interferences with the determination of aluminium were investigated by studying most major ions usually present in the mineral fraction of natural y o g h ~ r t ; ~ ~ for this we used an FI system similar to that for implementation of the standard additions method and injected interferent ions at concen- trations ten times higher than in natural yoghurt uia the second 200 pl loop both the sample and interferent ions finally being diluted to a volume of 2ml.The ion concentrations assayed were 10 mg ml-I Na+ (nitrate) 15 mg ml-1 Ca2+ (nitrate) 20 mg ml-1 K+ (nitrate) 15 mg ml-I C1- (ammonium salt) 5 mg ml-1 S042- (CuSO,) and 50 mg ml-1 (KH2P04). No interferences were observed; the differences in the analytical response relative to diluted natural yoghurt in the absence of interferent ions were small (- 3.0 to + 10.90/,). Therefore none of these ions interfere with the determination of aluminium at concentrations ten times higher than those typically present in natural yoghurt. Analysis of Milk Desserts In order to check the applicability of the proposed method to the direct determination of aluminium in milk desserts the same natural yoghurt was analysed after different pre- treatments (dry and wet digestion see Sample Preparation).The results obtained in ten consecutive determinations were 0.01 1 +0.0003 0.01 1 +0.0004 and 0.035 & 0.001 mg per 100 g (on a wet basis) for the direct FI method and the dry and the wet process respectively. The blank signals obtained by dry ashing were negligible and similar to those provided by the direct process; on the other hand the blank signal obtained by wet digestion was about twice that corresponding to the real content in the natural yoghurt sample which can be ascribed to aluminium contamination during acid sample pre- treatment and also to the glassware (pipettes) employed. Therefore wet digestion was discarded for sample preparation as the aluminium concentrations obtained exceed the actual contents in natural yoghurt.In order to validate the accuracy of the proposed method the analytical procedure was applied to a milk reference standard supplied by NIST SRM 1549 Non-fat Milk Powder (uncertified aluminium concentration 0.2mg per 100 g). For this purpose an accurately weighed amount of approximately 1.25g was mixed with 25ml of 0.2 v/v HN03. The average content found by ten consecutive determinations on independent test portions of the reference material was 0.197+0.012 mg per 100 g.The results obtained demonstrate the accuracy of the proposed method. Hence the proposed slurry method is a good alternative to the determi- nation of aluminium in this type of sample by using ashing digestion a widely accepted procedure. The proposed method was applied to the determination of aluminium in commercially available milk desserts. The results obtained in five individual determinations of aluminium by using a dilution factor of 1 +9 and their standard deviations are given in Table 2. A dilution factor of 1 + 19 or 1 + 39 was used in the FI system for samples with high aluminium 58 Journal of Analytical Atomic Spectrometry January 1995 Vol. 10Table 2 Aluminium contents in various milky desserts (dilution factor 1 + 9) as determined by using the proposed FI method (n = 5) Yoghurt Natural Pineapple (flavoured) Banana (flavoured) Macedbnia (flavoured) Raspberry (flavoured) Wood fruits* Pear (flavoured) Bifidus plum Peach (with fruit)* Lemon (flavoured) Coconut (flavoured) Strawberry (flavoured) Skim strawberry (flavoured) Skim apple (with fruit) Chocolatet Vanilla custard? Egg custard? A1 concentration/mg per 100g 0.01 1 k 0.0006 0.016It.:0.0015 0.010 _+ 0.0007 0.013 k0.0012 0.007 f O.OOO1 0.042 f 0.0035 0.026 f0.0018 0.029 f 0.0030 0.054 f0.0054 0.007 & 0.0007 0.010 k 0.0003 0.030 & 0.003 0.021 & 0.0004 0.019 f 0.0017 0.068 f 0.0043 0.121 k0.006 0.073 t- 0.0072 * 1 + 19 dilution factor.t 1 + 39 dilution factor. contents in conjunction with a sample loop volume of 100 or 50 pl respectively; the final volume was 2 ml.Taking into account that the density of this type of dessert is approximately 1.03&0.01 g ml-’ the amount of A1 con- tained in the 200 pl of injected sample is about 205 mg which allows the aluminium content per 100 g of milky dessert to be readily calculated. As can be seen from the results (Table 2 ) (a) the presence of fruit pieces increases the A1 content of the yoghurt because fruit contains more aluminium than does cow milk2 (b) the higher A1 contents of custard and chocolate yoghurt may be contributed in part by eggs or chocolate and/or additives and (c) one yoghurt (approximately 125 g) contributes to aluminium intake but at a non-toxic level (the allowed average dietary intake of A1 is about 6 mg d-I).’ Finally it is interesting to note that the proposed method can be used for the direct determination of aluminium (and probably also other metals at low levels) in milk desserts by using aqueous standards for calibration with no need for sample (slurry) weighing dilution or homogenization; the results thus obtained are comparable to those achieved by dry digestion of the sample.The Spanish CICyT is acknowledged for financial support (Project PB93-0717). M.A.Z. Arruda is also grateful to the Conselho Nacional de Desenvolvimento Cientifico e Tecnol6gico - CNPq (Brasilia Brazil) and to the Programa de Cooperacih Cientifica con Iberoamkrica (Madrid Spain) for additional financial support. 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 27 Aluminium in Food and the Environment ed.Massey R. C. and Taylor D. Royal Society of Chemistry London 1991. Pennington J. A. T. Food Addit. Contam. 1987 5 161. Vaessen H. A. M G. van de Kamp C. G. and Szteke B. Z . Lebenm.-Unters.-Forsch. 1992 194 456. Pennington J. A. T. and Jones J. W. in Aluminum in Health a Critical Review ed. Gitelman H. J. Marcel Dekker New York 1988. Delves H. T. Sieniawaska C. E. and Suchak B. Anal. Proc. 1993 30 358 Bendicho C. and de Loss-Vollebregt M. T. C. J. Anal. At. Spectrom. 1991 6 353. Caroli S. Microchem. J. 1992 45 257. de Benzo Z. A. Velosa M. Ceccarelli C. de la Guardia M. and Salvador A. Fresenius’ J. Anal. Chem. 1991 339 235. Minami H. Zhang Q. Itoh H. and Atsuya I. Microchem. J. 1994 49 126. Frech W. and Baxter D. C. Fresenius’ J. Anal. Chem. 1987 328,400. Gervais L. S. and Salin E. D. J. Anal. At. Spectrom. 1991,6,41. Arruda M. A. Z. Gallego M. and Valcarcel M. Anal. Chem. 1993 65 3331. Arruda M. A. Z. Gallego M. and Valcarcel M. Analyst 1994 119 779. Tsalev D. Z. At. Spectrosc. 1991 12 169. de Benzo Z. A. Velosa M. Ceccarelli C. and de la Guardia M. Fresenius’ J. Anal. Chem. 1991 339 235. Littlejohn D. Egila J. N. Gosland R. M. Kunwas U. K. Smith C. and Shan X. Q. Anal. Chim. Actu. 1991 250 71. Welz B. Schlemmer G. and Mudakavi J. R. J. And. At. Spectrom. 1992 7 1257. Lynch S. and Littlejohn D. J. Anal. At. Spectrom. 1989 4 157. Andrade J. C. Strong F. C. 111 and Martin N. J. Talanta 1990 37 711. Morales-Rubio A. Salvador A. and de la Guardia M. Fresenius’ J. Anal. Chem. 1992 342 452. Haswell S. J. and Barclay D. Analyst 1992 117 117. Gluodenis T. J. Jr. and Tyson J. F. J. Anal. At. Spectrom. 1993 8 697. Fagioli F. Landi S. Locatelli C. Righini F. Settimo R. and Magarini R. J. Anal. At. Spectrom. 1990 5 519. Arruda M. A. Z. Gallego M. and Valcarcel M. J. Anal. At. Spectrom. 1994 9 657. Redfield D. A. and Frech W. J. Anal. At. Spectrom. 1989,4,685. Analytical Methods Committee Analyst 1987 112 199. Tamime A. Y. and Robinson R. K. Yogur Cienciu y Tecnologia Acribia Zaragoza 1991. Paper 4104524F Received July 25 1994 Accepted September 19 1994 Journal of Analytical Atomic Spectrometry January 1995 VoE. 10 59