Fullerene a Sensitive and Selective Sorbent for the Continuous Preconcentration and Atomic Absorption Determination of Cadmium YANEIRA PETIT DE PEN� Aa MERCEDES GALLEGOb AND MIGUEL VALCA� RCEL*b aDepartment of Chemistry. University of L os Andes Me� rida Venezuela bDepartment of Analytical Chemistry Faculty of Sciences University of Co� rdoba E-14004 Co� rdoba Spain preconcentration is attracting much interest although there seems to be an unnecessary proliferation of reagents used; the most recent advances in this technique as regards the preconcentration of cadmium and other metals in biological and environmental samples by flow injection (FI) for subsequent determination by atomic spectrometry have been reviewed14 and were compiled in a recent monograph.15 A silica gel sorbent loaded with sodium diethyldithiocarbamate (Na-DDC) was used for the preconcentration of cadmium prior to its determination by FAAS; under dynamic conditions large sample volumes (about 200 ml) provided recoveries of 96.2%.16 Fang et al.17 developed an on-line FI system for the preconcentration of cadmium; the metal was precipitated from digested hair and rice with Na-DDC sorbed on a reactor eluted with isobutyl methyl ketone (IBMK) and determined by FAAS.From the foregoing it is obvious that the use of an on-line incorporated sorbent in an analytical system considerably improves the analytical sensitivity and results in a high sampling frequency which justifies current trends towards an increasing use of this preconcentration technique.On the other hand fullerenes have a high analytical potential for metal preconcentration even though only two papers have so far dealt with their performance in this respect. In this work a previously reported simple FI system11 was used to assess the potential of C60 fullerene for preconcentrating cadmium. For this purpose ammonium pyrrolidinedithiocarbamate (APDC) and 8-hydroxyquinoline were used to form neutral chelates of the metal. The effect of various ions on the determination was investigated in order to evaluate the selectivity of the sorbent. A complementary comparative study on the adsorption of the chelates on fullerene and silica RP-C18 phases was also conducted. Finally three biological reference materials were analysed as samples in order to test the performance of the proposed method.Ultratrace levels of cadmium were quantitatively sorbed on a C60 fullerene mini-column to form neutral chelates which were eluted with 200 ml of isobutyl methyl ketone and transferred to an atomic absorption spectrometer. Two chelating reagents viz. ammonium pyrrolidinedithiocarbamate (APDC) and 8-hydroxyquinoline were tested in a simple flow system. The adsorption constant was dramatically increased for the APDC reagent and C60 fullerene and cadmium was selectively separated from co-existing copper lead zinc and iron among other metals. Similar experiments were performed in parallel achieved was 110 and the detection limit was 0.3 ng ml-1 Cd. by using C18-bonded silica as sorbent.The concentration factor For analytical validation cadmium was determined in certified reference biological samples; only the APDC method with C60 fullerene as sorbent provided accurate results. Keywords Fullerene sorbent; preconcentration; chelates ; cadmium determination; atomic absorption spectrometry 60 60 18 Ever since two physicists Huffman and Kra�tschmer devised a method for producing macroscopic amounts of fullerenes in 1990 the properties of new products from these forms of carbon which include new types of polymers improved batteries superconductors catalysts etc. have been predicted but scarcely demonstrated.1 Numerous reviews on the discovery synthesis characterization reactivity and physico-chemical properties of fullerenes and related materials have been published; 2–5 there is even an on-line database of fullerene knowledge, 6 and a recent monograph about this topic has also been published.7 The main problem with fullerenes continues to be the separation of their fraction and purification of its components from the starting soot.Jinno and co-workers8,9 have published about 20 papers on the separation of fullerenes on monomeric and polymeric phases by HPLC. C fullerenebonded trimethylsilylsilica as a stationary phase for the separation of polycyclic aromatic hydrocarbons has been investigated by Jinno’s group,10 who demonstrated one of the possible uses of these new materials. The analytical potential of C60 fullerene as a sorbent material for the preconcentration of metals was first demonstrated by Gallego et al.;11 subsequent experiments with C and C70 fullerenes showed that both sorbents have a high analytical potential for metal preconcentration probably because of their large molecular surface area and volume.Higher sensitivity and selectivity are obtained with neutral chelates than by formation of ion pairs. Fullerenes perform better in metal preconcentration than do conventional solid materials (e.g. C -bonded silica activated carbon and resins).12 Several approaches have been devised for separatinganalytes from matrix elements and preconcentrating the former using a variety of techniques; however ion-exchange and chelating resins are routinely used for this purpose.13 On-line sorbent Journal of Analytical Atomic Spectrometry April 1997 Vol.12 (453–457) Apparatus A Varian (Palo Alto CA USA) 1475 atomic absorption spectrometer equipped with a bead impact system in the burner chamber and deuterium arc background correction was used throughout. The spectrometer output was connected to a Varian 9176 recorder. The hollow cathode lamp for cadmium was operated at 4 mA and the spectrometer was set at228.8 nm with a spectral band-width of 0.7 nm. The acetylene flow rate was 2.0 l min-1 and an air flow rate of 21.5 l min-1 was employed to ensure a clean blue flame. The flow manifold consisted of a Gilson (Villiers-le-Bel France) Minipuls-2 peristaltic pump furnished with poly(vinyl chloride) tubes two Rheodyne (Cotati CA USA) 5041 injection valves and labora- EXPERIMENTAL 453 tory-made sorption mini-columns packed with 80 mg of C60 fullerene or silica RP-C18.The mini-columns were made from PTFE capillaries of 3 mm id and sealed at both ends with small glass-wool plugs to prevent material losses. The minicolumns were initially flushed with 0.1 mol l-1 nitric acid; subsequent use of IBMK as eluent in each operating cycle was sufficient to make them ready for re-use. Reagents and Standard Solutions A 1000 mg l-1 cadmium stock solution was prepared by dissolving 1.000 g of the metal in a small volume of concentrated nitric acid and diluting to 1 l with 1% v/v nitric acid. A 0.1% m/v aqueous solution of APDC (Aldrich Madrid Spain) was prepared; the solution remained stable for at least 3 d.A 0.05% m/v 8-hydroxyquinoline solution (Aldrich) in 5% v/v ethanol was also prepared. IBMK (Probus Madrid Spain) was also used. C60 fullerene (>99.4% Hoechst Frankfurt-am- Main Germany) and polygosyl-bonded silica reversed-phase sorbent with octadecyl functional groups (RP-C18) 60–100 mm particle size (Millipore Madrid Spain) were employed as sorbent materials. Standard solutions (100 ml) containing 0.5–100 ng ml-1 cadmium were all freshly prepared by appropriate dilution of the stock standard solution (1000 mg l-1) in 0.1 mol l-1 nitric acid or at pH 4.5 (adjusted with dilute nitric acid) for the APDC-Cd and 8-hydroxyquinoline-Cd methods respectively. Sample Preparation The reference materials analysed were as follows NIST SRMs 1566a Oyster Tissue and 1577a Bovine Liver and Community Bureau of Reference (BCR) CRM 186 Pig Kidney.All were dried to constant mass by freeze-drying at 6 Pa (0.04 mmHg) for 24 h after which an accurately weighed amount of 1–2 g was digested with 15 ml of 65% nitric acid and 1 ml of 24.5% sulfuric acid in a glass beaker. The mixture was heated at about 200°C on a hot-plate until the sample was completely dissolved and nitrogen dioxide fumes were evolver was allowed to cool for about 2 min and the digestion procedure was repeated (about five times) with multiple additions of 5 ml of nitric acid until a clear solution was obtained and nitrogen dioxide fumes ceased to be evolved. Once cool the solution was transferred quantitatively into a calibrated flask of 50–250 ml capacity and made up to volume with ultrapure (Milli-Q) water.A reagent blank was prepared in parallel. Sub-samples (diluted 2-fold if required) at pH 1 or 4.5 (adjusted with nitric acid) were analysed immediately after preparation by introducing them into the manifold depicted in Fig. 1. On-line Preconcentration–Elution Procedure 2 The FI manifold used for on-line preconcentration and elution is shown in Fig. 1. First [Fig. 1(a)] 6 ml of standard or sample solution containing 0.5–50 ng ml-1 CdII in 0.1 mol l-1 nitric acid (for APDC) or 3–100 ng ml-1 CdII at pH 4.5 (for 8-hydroxyquinoline) were continuously pumped into the system and mixed on-line with the reagent (0.1% APDC or 0.05% 8-hydroxyquinoline in 5% ethanol).The preconcentration time was 2 min. The chelate was adsorbed on the sorbent mini-column (located in the loop of IV1) and the sample matrix sent to waste. During preconcentration an aqueous carrier was pumped to the instrument in order to record the baseline and the loop of IV was filled with eluent (IBMK). In the elution step [Fig. 1(b)] both injection valves were switched simultaneously to pass 200 ml of eluent (injected into the aqueous carrier) through the adsorbed chelate to desorb it and sweep the cadmium to the detector. Peak heights were used as analytical measurements and a blank consisting 454 Journal of Analytical Atomic Spectrometry April 1997 Vol. 12 Fig. 1 FI manifold for the preconcentration/determination of Cd.(a) Preconcentration and (b) elution. Bold lines denote lines relevant in the individual step. IV Injection valve; W waste; IBMK isobutyl methylketone (eluent); C fullerene or RP-C packedcolumn. Reagent 0.1% APDC or 0.05% 8-hydroxyquinoline18(5% ethanol–water). of 200 ml of IBMK injected prior to sample preconcentration was also used (about 0.040 A).To avoid any carry-over between samples the sample tube was filled with the next sample during the elution step. RESULTS AND DISCUSSION Aprevious study showed that the best sensitivityand selectivity of fullerenes as sorbent materials were obtained with neutral chelates.12 In order to test fullerenes for the preconcentration of cadmium two chelating agents were assessed in an FI system similar to that described elsewhere.11 Thus 8-hydroxyquinoline (a classical reagent that precipitates as a lemon-yellow complex with cadmium from dilute acetic acid neutral or ammoniacal solutions and is extractable into nonpolar organic solvents)and APDC (the most common chelating reagent for enrichment of metals in the FAAS technique) were assessed.Dithizone has been used as a chelating reagent for collecting trace amounts of cadmium on an activated carbon column by the FI technique;18 although the sensitivity achieved with this reagent was 20 and 60% higher than that obtained with APDC and 8-hydroxyquinoline respectively the selectivity was not optimum because of the non-specific character of the complexant dithizone. A comparative study of chelate retention (with APDC and 8-hydroxyquinoline) on C60 fullerene and RP-C18 was carried out here in order to select the best conditions for the determination of cadmium.range 0.01–0.3%. Both sorbents provided similar results the absorbance remaining constant above a 0.03% concentration; a 0.1% concentration of APDC in water was chosen. The influence of the 8-hydroxyquinoline concentration was examined over the range 0.01–0.05% (in 5% ethanol–water); the best results were obtained at the highest concentration tested. Because of the insolubility of the reagent in the ethanol–water mixture concentrations above 0.05% required a higher ethanol content which decreased the chelate adsorption and hence the analyticalsignal. A0.05% concentration of 8-hydroxyquinoline was selected for the two sorbent columns.Under the selected conditions the blank had a negligible effect; hence the IBMK injected before the sample served as a suitable blank. FI Preconcentration–Elution Conditions Optimization of the Chelates of APDC and 8-Hydroxyquinoline with Cadmium The effect of the APDC concentration was studied over the Initially an attempt was made to dissolve 8-hydroxyquinoline in dilute ammonia and ethanol; the best results were obtained by dissolving 50 mg of 8-hydroxyquinoline in 100 ml of a 5+95 v/v ethanol–water mixture. The effect of sample pH was examined by introducing 6 ml of a solution containing 10 ng ml-1 (for APDC) or 20 ng ml-1 cadmium (for 8-hydroxyquinoline) at a flow rate of 3.0 ml min-1 into the system for 2 min (Fig.1) the pH being adjusted with dilute nitric acid or ammonia and the reagents (0.1% APDC or 0.05% 8-hydroxyquinoline) being used as carriers. Several organic solvents were tested for elution of the adsorbed chelate from the column viz. IBMK ethanol,acetone and chloroform. The best results (difference between the sample and blank signals) were provided by IBMK where the chelate was most readily soluble and desorbed; in addition the blank The influence of the sample and reagent flow rates the length signal was lower and no dispersion occurred during transfer of the preconcentration coil and the volume of IBMK was to the detector because IBMK is immiscible with water. The studied by using a sample of 10 or 20 ng ml-1 cadmium in the two proposed methods were optimized by using a column APDC and 8-hydroxyquinoline methods respectively.The packed with 80 mg of C60 (1.1 cm×3 mm id) or 80 mg of elution steps was also investigated. Changes in the flow rate effect of the IBMK flow rate in the preconcentration and maximum chelate adsorption was achieved at pH 0.5–5.0 RP-C18 (1.6 cm×3mm id). As can be seen in Fig. 2 the of the sample (6.0 ml) between 1.0 and 3.0 ml min-1 resulted (for C60) and 0.7–2.0 (for RP-C18) for the APDC reagent flow rates through a decreased residence time. Increasing the in very small variations in the signal which decreased at higher and 3.2–6.0 (for C60) and 4.0–5.5 (for RP-C18) for reagent flow rate resulted in concomitant sample dilution and 8-hydroxyquinoline.The optimum pH range was wider for C60 fullerene probably as a consequence of its adsorption hence in decreased atomic signals. Reagent (APDC or constant being greater than that for the RP-C18 sorbent 8-hydroxyquinoline) and sample flow rates of 0.3 and consistent with experimental results obtained for the 3.0 ml min-1 respectively were chosen for the two methods and both columns (C lead–APDC chelate (adsorption constants were 575 and 155 fullerene and RP-C18). The optimum for C60 and C18 respectively).11 On the other hand the 300 cm (0.5 mm id) in all instances; hence a length of 250 cm sorption of the metal chelates is effected through the comlength of the preconcentration 60 coil ranged between 200 and plexing ligands and the results show that APDC forms was used throughout.a cadmium chelate that has a higher affinity for C60 than The volume of eluent (IBMK) played a major role in the for RP-C18 while the 8-hydroxyquinoline chelate exhibits The desorption process was found not to depend on the type chelate elution its effect being studied between 50 and 300 ml. smaller differences with both sorbents. In the APDC method samples were prepared in 0.1 mol l-1 nitric acid. In the of sorbent or chelating reagent used; elution was complete (no 8-hydroxyquinoline method samples were prepared in 0.1 mol carry over) for volumes above 150 ml. At volumes higher than l-1 acetic acid–sodium acetate buffer (pH 4.75); however the 200 ml the atomic signal difference decreased through disperanalytical signal of cadmium decreased by about 30% relative sion of cadmium in the organic solvent; the volume of eluent to samples adjusted to the same pH with dilute nitric acid finally selected was 200 ml.The aqueous flow rate (eluent probably because of the acetate anion at high concentrations carrier) was found to affect peak height which increased up complexing cadmium and hence decreasing the chelate adsorp- to 4.0 ml min-1 owing to the increasing nebulizer efficiency. tion. Consequently samples were adjusted to pH 4.5 with Based on the above results flow variables affect the dilute nitric acid in the 8-hydroxyquinoline method. determination of cadmium similarly with APDC and 8-hydroxyquinoline reagents and with either column; therefore the same flow system can be used in both methods.60 Fig. 2 Effect of pH on Cd absorbance as measured after on-line preconcentration with APDC (#) or 8-hydroxyquinoline ($) on a C fullerene or RP-C18 column. Sample 10 or 20 ng ml-1 Cd for the APDC and 8-hydroxyquinoline methods respectively. Comparison of Preconcentration Methods 18 60 fullerene was used in the present study. Two columns were A comparative study of the chelate retention of APDC and 8-hydroxyquinoline on C60 fullerene and a conventional RP-C sorbent was carried out. In the above-described experiments, 12 we compared the adsorption capacity of C60 and C70 fullerenes; the larger molecular surface area and volume of C70 make this material slightly more effective for metal preconcentration; however because C70 fullerene is very expensive only C packed with 80 mg of C60 fullerene (1.1 cm×3 mm id) or 80 mg of RP-C18 (1.6 cm×3 mm id).Both columns were intercalated sequentially into the FI system depicted in Fig. 1. By using this manifold and optimized chemical and flow variables several calibration graphs for cadmium were constructed using the two methods studied. The figures of merit of the calibration graphs (r=0.998 or 0.999 in all instances) obtained for cadmium (sample volume 6.0 ml) using the APDC and 8-hydroxyquinoline chelates with the two sorbents are summarized in Table 1. The detection limits were calculated as three times the standard deviation of the absorbance obtained for Journal of Analytical Atomic Spectrometry April 1997 Vol. 12 455 Table 1 Figures of merit for the atomic absorption determination of cadmium Range/ng ml-1 0.5–20 Regression equation* A=0.004+0.015x Method APDC–CdII‡ 2–50 3–80 4–100 A=0.002+0.006x A=0.003+0.004x A=0.002+0.003x APDC–CdII§ 8-Hydroxyquinoline–CdII‡ 8-Hydroxyquinoline–CdII§ * A absorbance; x cadmium concentration (ng ml-1 ).Sample volume 6 ml. † Compared with conventional sample introduction of an aqueous solution (A=0.005+1.35×10-4 x). ‡ With C60 fullerene. § With RP-C18. 60 15 injections of 200 ml of IBMK (blank). The precision of the method (expressed as the RSD) was evaluated on 11 samples containing 10 or 20 ng ml-1 cadmium in the APDC and 8-hydroxyquinoline methods respectively. The overall time required for preconcentration (2 min) and elution (a few seconds) of a sample was about 2.5 min; hence the throughput was about 25 samples h-1 (blank processing included).The most interesting conclusions that can be drawn from Table 1 are as follows first the sensitivity (slope of the calibration graph) is higher with APDC than with 8-hydroxyquinoline; second the sensitivity is higher for C fullerene columns in both methods probably because of the higher adsorption capacity of fullerenes.11 Therefore APDC is the better choice for the preconcentration of cadmium on C60 fullerene because it provides a preconcentration factor of 110 for a sample volume of 6 ml which can be increased by using a larger volume. The precision was similar in all instances. 18 Tolerated [metal]5[Cd] ratio 8-Hydroxyquinoline method Dithizone method* Activated carbon 18 60 18 RP-C 200 C 400 RP-C 600 1000 200 500 1000 300 60 500 500 300 500 250 200 800 800 600 800 400 400 600 600 500 400 350 300 150 200 100 150 300 300 400 600 800 500 200 400 400 Interferences The influence of metals that might react with APDC or 8-hydroxyquinoline and replace cadmium in the original chelate was investigated in order to identify potential interferences.Table 2 lists the tolerated ratios of foreign cations to cadmium; the maximum concentration tested was 1000 times that of the analyte. Interferents decreased the cadmium signal in all instances by competition for and consumption of the reagent; hence uncomplexed cadmium was not retained on the column; otherwise if all chelates were formed the volume of IBMK used was inadequate to elute them.As can be seen in Table 2 the selectivity was higher for C60 fullerene as a consequence of its large surface area relative to RP-C18 in addition to its high interstitial volume (which ensured more uniform distribution of the chelate throughout the column and hence readier elution). With the RP-C sorbent (white in colour) the chelate (at high metal concen- Table 2 Tolerated concentrations of foreign cations in the determination of 10 (with APDC) or 20 (with 8-hydroxyquinoline) ng ml-1 cadmium APDC method 60 C 1000 Ion Al3+ Zn2+ 1000 Pb2+ 1000 Mn2+ 1000 1000 Co2+ Cu2+ Fe3+ Ni2+ Sn2+ Hg2+ 1000 1000 1000 800 600 * Data from ref.18. 456 Journal of Analytical Atomic Spectrometry April 1997 Vol. 12 CONCLUSIONS The sensitivity of the proposed methods depends on both the reagent and sorbent (APDC and C sorbent are the best option) while the selectivity is more 60 markedly dependent on the sorbent (C60 is the best alternative). The APDC method with C fullerene as sorbent is clearly superior to existing continuous-flow 60 alternatives using other sorbents or resins in terms of sensitivity and selectivity.15,18 Furthermore most available preconcentration methods have been applied to water samples and few to biological materials owing to the interferences posed by the latter.Thus as shown in this work one should rigorously check for potential interferences in order to avoid unexpected results such as those found here for the Bovine Liver sample. Therefore the APDC reagent in combination with C fullerene as sorbent can probably be applied to highly complex 60 matrices to determine trace levels of cadmium. Enhancement factor† Detection limit/ng ml-1 RSD (%) 1.9 110 45 30 20 0.3 1.1 2.0 2.6 2.1 2.3 2.5 trations) was not adsorbed uniformly on the column; hence its subsequent elution was more difficult. The selectivity of the fullerene column was at least twice that of the RP-C18 column in both methods. The higher selectivity of the APDC method relative to the 8-hydroxyquinoline method can be ascribed to the low sample pH (1.0).In the 8-hydroxyquinoline method some ions (viz. Al3+ Fe3+ Sn2+ Hg2+ and Pb2+) were hydrolysed at pH 4.5 and their hydroxide or basic salts precipitated as a result; however when the sample was mixed with the 8-hydroxyquinoline reagent their corresponding 8-hydroxyquinolinolates were probably formed; hence the ions were tolerated at relatively high levels. For comparative purposes the selectivity of another automated method for the preconcentration of cadmium (pH 2.0) with dithizone (chelating reagent) on an activated carbon column (70 mg sorbent; 2.5 cm×3 mm id) is also included in Table 2.18 This method is scarcely selective owing to the low selectivity of dithizone.Determination of Cadmium in Certified Reference Materials sorbent owing to its low selectivity. Cadmium was determined in three reference materials Oyster Tissue Pig Kidney and Bovine Liver. Although the APDC method was the better alternative for this determination the selectivity achieved with 8-hydroxyquinoline also permitted the determination of cadmium in some samples; hence both methods were employed for comparison. The blank absorbances corresponded to a cadmium concentration of less than 0.5 ng ml-1 (this blank allowed the contribution of cadmium ion in the reagents to the digested sample to be assessed). The analytical results listed in Table 3 are in good agreement with the certified values (within the 95% confidence intervals) for Oyster Tissue and Pig Kidney.However Bovine Liver can only be analysed by the APDC method with C fullerene as sorbent because it is subject to interferences from60iron (content 194 mg g-1) and copper (content 158 mg g-1). As a consequence the concentrations in Bovine Liver obtained with the 8-hydroxyquinoline method were lower particularly with the RP-C18 Table 3 Determination of cadmium in certified reference materials. All values are given in mg g-1 cadmium APDC (C60 ) Certified value 4.15±0.38 Reference material NIST SRM 1566a Oyster Tissue 4.05±0.30 2.82±0.17 0.45±0.05 2.71±0.15 0.44±0.06 BCR CRM 186 Pig Kidney NIST SRM 1577A Bovine Liver The Spanish DGICyT is gratefully acknowledged for financial support (Grant No.PB95–0977). REFERENCES 1 Baum R. M. Chem. Eng. News 1993 November 22 8. 2 Kroto H. W. Allaf A. W. and Balm S. P. Chem. Rev. 1991 91 1213. 3 Special Issue on Buckminsterfullerenes Acc. Chem. Res. 1992 25 (3). 4 Manteca-Diego C. and Moran E. Ann. Quim. 1994 90 143. 5 Lieber Ch. M. and Chia-Chun C. in Preparation of Fullerenes and Fullerene-Based Materials eds. Ehrenreich H. and Spaepen F. Solid State Physics of Fullerenes Academic Press Boston MA 1994. 6 Smalley R. E. T he Almost (but never quite) Complete Buckminsterfullerene Bibliography. (Database available from the author via E-mail BUCKY@-SOL1.LRSM.UPENN.EDU.) 7 Dresselhaus M. S. Dresselhaus G. and Eklund P. C. Science of Fullerenes and Carbon Nanotubes Academic Press San Diego CA 1996.8 Saito Y. Ohta H. Nagashima H. Itoh K. Jinno K. Okamoto Journal of Analytical Atomic Spectrometry April 1997 Vol. 12 Found 60) 8-Hydroxyquinoline (RP-C18) APDC 8-Hydroxyquinoline (RP-C18) (C 4.10±0.39 3.92±0.37 2.78±0.21 0.22±0.13 2.80±0.17 0.38±0.09 3.96±0.34 2.70±0.20 0.32±0.09 M. Chen Y. L. Luehr G. and Archer J. J. L iq. Chromatogr. 1995 18 1987. 9 Kimata K. Hirose T. Moriuchi K. Hosoya K. Araki T. and Tanaka N. Anal. Chem. 1995 67 2556. Jinno K. and Itoh K. J. High Resolut. Chromatogr. 1995 18 569. 1994 66 4074. 1995 67 2524. Spectrom. 1993 8 979. 10 Saito Y. Ohta H. Terasaki H. Katoh Y. Nagashima H. 11 Gallego M. Petit de Pen�a Y. and Valca�rcel M. Anal. Chem. 12 Petit de Pen�a Y. Gallego M. and Valca�rcel M. Anal. Chem. 13 Ebdon L. Fisher A. Handley H. and Jones P. J. Anal. At. 14 Taylor A. Branch S. Crews H. M. Halls D. J. and White M. J. Anal. At. Spectrom. 1996 11 103R. 15 Fang Z. Flow Injection Atomic Absorption Spectrometry Wiley Chichester 1995. 16 Rio-Segade S. Pe�rez-Cid B. and Bendicho C. Fresenius’ J. Anal. Chem. 1995 351 798. 17 Fang Z. Xu S. Dong L. and Li W. T alanta 1994 41 2165. 18 Petit de Pen�a Y. Gallego M. and Valca�rcel M. J. Anal. At. Spectrom. 1994 9 691. Paper 6/06990H Received October 14 1996 Accepted Janua