Talanta 61 (2003) 603 /610 www.elsevier.

com/locate/talanta

A study of bismuth-lm electrodes for the detection of trace metals by anodic stripping voltammetry and their application to the determination of Pb and Zn in tapwater and human hair
G. Kefala, A. Economou *, A. Voulgaropoulos, M. Sofoniou
Laboratory of Analytical Chemistry, Department of Chemistry, Aristotle University of Thessaloniki, 541 24 Thessaloni, Greece Received 13 February 2003; received in revised form 7 May 2003; accepted 15 May 2003

Abstract This work reports the simultaneous determination of Cd(II), Pb(II) and Zn(II) at the low mg l  1 concentration levels by square wave anodic stripping voltammetry (SWASV) on a bismuth-film electrode (BFE) plated in situ. The metal ions and bismuth were simultaneously deposited by reduction at /1.4 V on a rotating glassy carbon disk electrode. Then, the preconcentrated metals were oxidised by scanning the potential of the electrode from /1.4 to 0 V using a square-wave waveform. The stripping current arising from the oxidation of each metal was related to the concentration of each metal in the sample. The parameters for the simultaneous determination of the three metals were investigated with the view to apply this type of voltammetric sensor to real samples containing low concentrations of metals. Using the selected conditions, the limits of detection were 0.2 mg l  1 for Cd and for Pb and 0.7 mg l 1 for Zn at a preconcentration time of 10 min. Finally, BFEs were successfully applied to the determination of Pb and Zn in tapwater and human hair and the results were in satisfactory statistical agreement with atomic absorption spectroscopy (AAS). # 2003 Elsevier B.V. All rights reserved.
Keywords: Square wave anodic stripping voltammetry; Bismuth lm electrodes

1. Introduction Stripping analysis has proved a powerful technique for the determination of trace metals in samples of environmental, clinical and industrial origin [1]. Mercury film electrodes (MFEs) and the hanging mercury drop electrode (HMDE) have

* Corresponding author. Tel.: /30-231-099-7728; fax: /30231-099-7719. E-mail address: [email protected] (A. Economou).

been traditionally used for ASV owing to the advantageous analytical properties of mercury in the negative potential range. However, the increased risks associated with the use, manipulation and disposal of metallic mercury or mercury salts have led some countries to completely ban mercury [2]. Following this general trend for more environmentally-friendly analytical methods, alternative electrode materials are continuously been sought. Among other materials, chemically modified electrodes [3], boron-doped electrodes [4]

0039-9140/03/$ - see front matter # 2003 Elsevier B.V. All rights reserved. doi:10.1016/S0039-9140(03)00350-3

604

G. Kefala et al. / Talanta 61 (2003) 603 /610

and screen-printed electrodes [5] have been developed in order to substitute mercury electrodes in stripping analysis. Recently, BFEs, consisting of a thin bismuth film deposited on a carbon substrate have been shown to offer comparable performance to MFEs in ASV [6 /10]. While it is well known that the utility of mercury as an electrode material in ASV is based on its ability to form amalgams with many heavy metals, the performance of bismuth can be attributed to its property of forming alloys with different metals [11]. The excellent analytical characteristics of BFEs combined with the low toxicity of bismuth in comparison to that of mercury can lead to more widespread use of these devices. However, the limited number of published work on BFEs deal with fundamental studies and insights into the behaviour of BFEs in ASV at relatively higher concentrations (typically greater than 10 mg l 1). To our knowledge, no analytical applications of BFEs in real samples, and especially samples containing low concentrations of metals, have been reported so far. In this case, the selection of the instrumental and chemical conditions of the analysis becomes critical for the successful application of the technique. In this work, we performed a thorough study of the parameters affecting the performance of BFEs (i.e. the choice of the working electrode, the Bi plating conditions, the deposition time, the conditions of mass-transport, the parameters of the voltammetric waveform and the tolerance to surfactants) in order to achieve high sensitivity in the simultaneous determination of Zn, Cd and Pb at low concentrations. Finally, the most appropriate conditions were used for the first analytical application of ASV on a BFE, namely the determination of Pb and Zn in tap water and human hair.

Working metal ion solutions were prepared from 1000 mg l1 atomic absorption standard solutions after appropriate dilution with water. A 1 mol l1 acetate buffer (pH 4.5) stock solution, prepared by mixing the appropriate amounts of glacial CH3COOH and NaOH, was used to prepare solutions of the supporting electrolyte. A 50 mg l1 Bi(III) stock solution was prepared in 0.1 mol l1 acetate buffer. A 1000 mg l 1 solution of Triton X-100 (BDH, Poole, England) was prepared in water.

2.2. Apparatus Voltammetric measurements were performed with a home-made potentiostat interfaced to a PC though a 6025E PCI multi-purpose interface card (National Instruments, Austin, TX). The experimental sequence was fully automated and controlled by the PC using a control application developed in LABVIEW 5.1 (National Instruments) as reported previously [12]. The voltammetric cell was a standard 50 ml glass vial (Metrohm, Switzerland) equipped with a Ag/AgCl reference electrode and a Pt counter electrode. An electrode rotator (Metrohm 62810) or a magnetic stirrer were used during the preconcentration and cleaning steps. The glassy carbon working electrode (3 mm in diameter) was from Metrohm. Before use, the electrode was polished with a water slurry of 0.3 mm Al2O3 and rinsed with ethanol and water. A carbon paste electrode was prepared by mixing thoroughly 1.5 g of graphite powder with 1.2 ml of Nujol and filling a syringe-type holder with the resulting paste; the diameter of the active surface was 5 mm. A smooth surface was formed by gently rubbing the electrode on a clean filter paper. The impregnated graphite electrode (6 mm in diameter) was donated by M. Halama (University of Cosice, Slovakia) and was polished by rubbing on a clean filter paper before use. For the AAS measurements of Pb and Zn, a Perkin /Elmer 5100 AAS spectrometer (CT, USA) was used. Pb was determined by electrothermal AAS in a Perkin /Elmer 5100 ZL furnace module while Zn was determined by flame AAS.

2. Experimental 2.1. Chemicals and reagents All the chemicals were of analytical grade and purchased from Merck (Darmstadt, Germany). Doubly-distilled water was used throughout.

G. Kefala et al. / Talanta 61 (2003) 603 /610

605

2.3. Experimental procedure 2.3.1. Sample preparation Tap water was collected from taps in our laboratory. For the determination of Pb(II), 16 ml of the tap water, 2 ml of the 1 mol l 1 acetate buffer and 2 ml of deionised water were placed in the cell and the analysis was carried out as described below (except that the deposition potential was /1.2 V). For the determination of Zn(II), 4 ml of tap water, 2 ml of the 1 mol l 1 acetate buffer and 14 ml of deionised water were placed in the cell and the analysis was carried out as described below with a preconcentration time of 60 s. Hair was supplied from a volunteer in our laboratory. The hair sample was washed with acetone, rinsed thoroughly with deionised water and dried at 60 8C. An accurately weighted amount of hair (approximately 0.5 g) was treated with 20 ml of 65% HNO3 and 2 ml of 30% H2O2 and heated almost to dryness. The residue was further heated with 1 ml of 65% HNO3 until decolorisation, the solution was brought to pH 4.5 with NaOH and diluted to 100 ml with water. 2.5 ml of this solution, 2 ml of the 1 mol l 1 acetate buffer and 15.5 ml of deionised water were placed in the cell and the analysis was carried out as described below. Urine was supplied from a volunteer in our laboratory. 5 ml of urine, 2 ml of the 1 mol l 1 acetate buffer and 13 ml of water were placed into the cell and the analysis was carried out as described below. 2.3.2. Measurement procedure For preplating the substrate with a bismuth film, the electrodes were immersed in the bismuth plating solution and the Bi film was formed by holding the working electrode potential at /1.2 V under rotation or stirring. Then, the electrodes were immersed into the sample solution and deposition of the metals was carried out at /1.4 V under rotation for a carefully defined period of time followed by a 10 s rest period. The voltammogram was recorded between /1.4 and /0.4 V by applying a square wave waveform and the

electrode was cleaned from residual metals for 30 s at /0.4 V under rotation. In-situ bismuth films were prepared by spiking the sample with 500 mg l 1 Bi(III) and simultaneously depositing Bi and the metals on the surface of the electrode at /1.4 V under rotation or stirring for a carefully defined period of time followed by a 10 s rest period. The voltammogram was recorded between /1.4 and 0 V by applying a square wave waveform and the electrode was cleaned from residual metals and the bismuth film for 30 s at 0.0 V under rotation. In-situ mercury films were prepared by spiking the sample with 1000 mg l 1 Hg(II) and simultaneously depositing Hg and the metals on the surface of the electrode at /1.4 V under rotation for a carefully defined period of time followed by a 10 s rest period. The voltammogram was recorded between /1.4 and 0.6 V by applying a square wave waveform and the electrode was cleaned from residual metals and the mercury film for 30 s at 0.6 V under rotation or stirring.

3. Results and discussion 3.1. Choice of the BFE substrate In this work three carbon substrates were studied (glassy carbon, carbon paste and impregnated graphite). Each substrate was plated in-situ with bismuth and stripping voltammograms were obtained under identical chemical and instrumental conditions for a solution containing 20 mg l1 each of Zn(II), Cd(II) and Pb(II) in 0.1 mol l1 acetate buffer (pH 4.5). Typical stripping voltammograms on the three substrates are illustrated in Fig. 1. Taking under consideration the surface area of each electrode, the glassy carbon substrate produced the highest current densities for all of the metals and the sharpest peak for Zn than the two other electrodes. The GC electrode also exhibited the flattest baseline and the lowest noise levels and was, therefore, selected as the most suitable substrate for the bismuth coating. A comparison between an MFE and a BFE, both plated in-situ on glassy carbon, in a solution containing 20 mg l1 each of Zn(II), Cd(II) and Pb(II) in 0.1 mol

606

G. Kefala et al. / Talanta 61 (2003) 603 /610

Fig. 1. Comparative voltammograms of a solution containing 20 mg l  1 each of Zn(II), Cd(II) and Pb(II) on in-situ plated BFE with carbon paste (CP), impregnated graphite (IG) and glassy carbon (GC) as the substrate. Supporting electrolyte: 0.1 mol l  1 acetate buffer (pH 4.5); deposition potential: /1.4 V; stirring: 500 rpm; deposition time: 80 s; frequency 25 Hz; pulse height: 40 mV; step increment: 2 mV; Bi(III) concentration: 500 mg l  1.

Fig. 2. Voltammograms of a solution containing 20 mg l  1 each of Zn(II), Cd(II) and Pb(II) on in-situ plated BFE using different concentrations of Bi(III): (a) 100; (b) 200; (c) 500; (d) 1000, and; (e) 2000 mg l  1. Conditions as in Fig. 1 with rotation at 500 rpm instead of stirring.

l 1 acetate buffer (pH 4.5) suggested that the sensitivity for Cd was comparable on the two electrodes, the sensitivity for Pb was better on the BFE than on the MFE while the MFE produced a higher signal than the BFE for Zn. The positions of the Pb and Zn peaks were approximately the same on the two electrodes but the Cd peak was shifted to more negative values on the BFE; this effect improved the resolution between the Cd and Pb peaks on the BFE [13]. These results suggest that the performance of the BFE is comparable to, or even better than, that of the MFE for the determination of Pb and Cd but inferior for the determination of Zn. 3.2. Conditions for plating the bismuth lm Representative stripping voltammograms for a solution containing 20 mg l 1 each of Zn(II), Cd(II) and Pb(II) in 0.1 mol l 1 acetate buffer (pH 4.5) on in-situ plated BFEs with different Bi(III) concentrations are shown in Fig. 2. In this case, the concentration of the Bi(III) solution controls the thickness of the Bi film. The thickness of the film did not affect the peak position of any metal and, in addition, the height of the Cd peak was largely unaffected by the thickness of the

bismuth coating. However, the height of the Pb peak displayed a clear dependence on the film thickness as its decreased with increasing thickness of the bismuth film while the Zn peak slightly increased with increasing thickness of the bismuth film. The results also indicated that the background current at negative potentials, where the Zn peak appeared, increased with the amount of Bi deposited; this resulted in a more sloping baseline for the Zn peak. This effect was rather surprising, as it was initially expected that higher surface coverage with Bi would cause a negative shift of the hydrogen evolution overpotential and would, therefore, lower the background contribution close to the Zn peak. The increase in the background at the anodic side of the Pb at high Bi concentrations was due to the onset of the large Bi oxidation peak. The Cd and Pb peaks became wider at higher film thickness but the width of the Zn peak did not show a clear dependence on the bismuth film thickness. The same trends were also observed with preplated bismuth films, prepared as described in Section 2.3.2 (voltammograms not shown). In general, the two plating methods offered comparable performance in terms of sensitivity and reproducibility, yet the in-situ method was selected as it greatly simplified the experimental procedure.

G. Kefala et al. / Talanta 61 (2003) 603 /610

607

3.3. Effect of the preconcentration time and mass transport The metal preconcentration time was studied in the range 20 /600 s for a solution containing 20 mg l 1 each of Zn(II), Cd(II) and Pb(II) in 0.1 mol l 1 acetate buffer (pH 4.5). The peak heights increased linearly up to 400 s while at higher deposition time the peak heights versus time plots started to slightly deviate from linearity, as shown in Fig. 3(a). This clear dependence of the stripping current on the deposition time could be used for increasing the sensitivity in ASV. The mass transport conditions were studied by varying the electrode rotation speed between 500 and 2000 rpm for a solution containing 20 mg l 1 each of Zn(II), Cd(II) and Pb(II) in 0.1 mol l 1 acetate buffer (pH 4.5). The Levich equation

predicts that the diffusion layer, and thus the stripping current, on the RDE is linearly related to the square root of the electrode rotation speed [13]. The experimental data supported this prediction as the plot of the metal peak heights versus the square root of the electrode rotation speed produced straight lines for the three metals (Fig. 3(b)). Then non-linear relationship between the rotation speed and stripping current means that there was only a small sensitivity gain by increasing the rotation speed of the electrode. At the same time, very high rotation speeds increase the risk of the bismuth film being mechanically damaged. 3.4. Effect of the parameters of the stripping waveform Among the stripping waveforms, the square wave modulation combines high sensitivity with high speed [14]. The square wave waveform was found to be preferable to either the differential pulse or the linear scan stripping waveforms for ASV on BFEs. One additional advantages of the SW waveform is its insensitivity to dissolved oxygen which allowed the analysis to be carried out in the presence of oxygen and avoided the time-consuming deoxygenation step. The SW parameters affecting the response were the SW frequency, the SW step increment and the SW pulse height and were investigated univariantly using a solution containing 20 mg l 1 each of Zn(II), Cd(II) and Pb(II) in 0.1 mol l 1 acetate buffer (pH 4.5). The effect of the frequency was studied in the range 12.5 /200 Hz and the effect of the step increment was investigated between 1 and 16 mV. The peak potentials shifted to the anodic direction with increasing frequency or step increment (presumably as the oxidation of the metals became less reversible at higher frequencies) while the peak heights for Pb and Cd increased at higher frequencies and step increments. In contrast, the Zn peak increased only marginally with increasing frequency and step increment and, at the same time, became wider. The effect of the SW pulse height was studied in the range from 10 to 80 mV. The peak potentials were shifted to the cathodic direction and the peak heights increased upon

Fig. 3. (a) Effect of the deposition time on the peak heights for a solution containing 20 mg l  1 each of Zn(II), Cd(II) and Pb(II) on in-situ plated BFE. Conditions as in Fig. 2. (b) Effect of the electrode rotation speed on the peak heights for a solution containing 20 mg l  1 each of Zn(II), Cd(II) and Pb(II) on in-situ plated BFE. Conditions as in Fig. 2.

608

G. Kefala et al. / Talanta 61 (2003) 603 /610

increase of the pulse height. However, the background deteriorated and the peak widths increased at higher pulse heights.

3.6. Tolerance to surface-active compounds BFEs in common with MFEs are prone to interference form surface-active compounds that adsorb on the electrode and cause deactivation of its surface [15]. Triton X-100 was selected as a model compound to study the effect of surfactants on the stripping response of BFEs at low concentrations of metals. The effect of this compound at different concentrations on the stripping peak heights for Zn, Cd and Pb in a solution containing 10 mg l 1 each of Cd(II) and Pb(II) and 20 mg l1 of Zn in 0.1 mol l 1 acetate buffer (pH 4.5) is illustrated in Fig. 5(a). It is clear that the Zn peak was more sensitive to the presence of surfactants as

3.5. Calibration data Calibration was performed on BFEs for the simultaneous determination of Pb(II), Cd(II) and Zn(II) by ASV using the following conditions: deposition potential: /1.4 V; deposition time: 120 or 180 s; frequency: 50 Hz; pulse height: 40 mV; step increment: 4 mV; rotation speed: 1000 rpm. Four different concentration ranges were investigated: 1 /10, 2 /20, 5 /50 and 10 /100 mg l 1. A series of stripping voltammograms at low concentrations (2 /18 mg l1) is shown in Fig. 4. The relative standard deviation was 3.2% for Pb, 3% for Cd and 2.6% for Zn at the 10 mg l1 level (n / 10) with a preconcentration time of 120 s. The limit of detection was calculated as 0.2 mg l 1 for Cd and for Pb and 0.7 mg l 1 for Zn at a preconcentration time of 10 min.

Fig. 4. A series of voltammograms for increasing concentrations of Zn(II), Cd(II) and Pb(II) on an in-situ plated BFE (Zn: -k-; Pb: -j-; Cd: '). From below: blank and nine successive additions of 2 mg l  1 of Zn(II), Cd(II) and Pb(II). Supporting electrolyte: 0.1 mol l  1 acetate buffer (pH 4.5); deposition potential: /1.4 V; rotation speed: 1000 rpm; deposition time: 180 s; frequency 50 Hz; pulse height: 40 mV; step increment: 4 mV.

Fig. 5. (a) Effect of the presence of TritonX-100 on the peak heights for a solution containing 10 mg l  1 each of Cd(II) and Pb(II) and 20 mg l  1 of Zn(II) on in-situ plated BFE. Conditions as in Fig. 2. (b) Voltammograms after a series of standard additions for a diluted urine sample spiked with 4 mg l  1 of each Cd(II), Pb(II) and Zn(II) on in-situ plated BFE. From below: unspiked sample, spiked sample and three successive additions of 4 mg l  1 each of Cd(II), Pb(II) and Zn(II). Conditions as in Fig. 4 with a Bi(III) concentration of 1000 mg l  1.

G. Kefala et al. / Talanta 61 (2003) 603 /610

609

4 mg l 1 Pb(II), Cd(II) and Zn(II) as illustrated in Fig. 5(b). The results in the urine sample confirmed the earlier observation that the Cd peak was more resistant to the presence of surfactants than the Pb peak (for equal mass concentrations of Pb(II) and Cd(II) in the urine sample, the peak current of Cd was approximately three times higher than the Pb peak current). Although many compounds are present in urine that are potential ligands for metal ions, it has been previously shown that Pb and Cd measured in urine by ASV in the pH range 4 /6 reflect total metal [16]. Indeed, the recovery was estimated as 102% for Pb and 104% for Cd. Finally, the high susceptibility of Zn to surfactants is clearly demonstrated in the analysis of urine as the Zn peak height is so low that does not allow quantitative evaluation in this medium. 3.7. Application to tap water and human hair BFEs were finally applied to the analysis of tapwater and human hair. Representative stripping voltammograms for the determination of Pb and Zn in tapwater are illustrated in Fig. 6(a and b), respectively. In the case of the determination of Pb, when the preconcentration potential was set to /1.4 V, a large Zn peak appeared that obscured the small Pb peak; this large peak was due to the

Fig. 6. (a) Voltammograms for the determination of Pb(II) in tapwater on in-situ plated BFE. From below: sample and four standard additions of 0.8 mg l  1 Pb(II) each. Conditions as in Fig. 4 with /1.2 V deposition potential. (b) Voltammograms for the determination of Zn(II) in tapwater on in-situ plated BFE. From below: sample and four standard additions of 25 mg l  1 Zn(II) each. Conditions as in Fig. 4 with 60 s deposition time.

its peak decreased drastically even at modest concentrations of the surfactant while the Cd peak was the least affected by the presence of the surfactant. Generally, the susceptibility of the metal peaks to the presence of surfactants increased in the order Zn /Pb /Cd. However, insitu plated BFEs exhibited a remarkable stability in terms of their response in the presence of surfactants. To demonstrate the ability of in-situ plated BFEs to function properly in the presence of interfering surface-active compounds in a real sample, a series of standard additions was performed in a diluted urine sample spiked with

Fig. 7. Voltammograms for the determination of Pb(II) in human hair on an in situ plated BFE. From below: sample and three standard additions of 5 mg l  1 Pb(II) each. Conditions as in Fig. 4.

610

G. Kefala et al. / Talanta 61 (2003) 603 /610

Table 1 The results of the determination of Pb and Zn in tapwater and hair with ASV and AAS ASV Tap water (mg l ) Zn 2049/6 Pb 2.09/0.05 Human hair (mg g  1 ) Pb 9.69/0.3
1

there is satisfactory agreement between the two techniques.

AAS

References
2119/10 1.99/0.1 9.19/0.5 [1] J. Wang, Stripping Analysis, VCH, Deereld Beach, FL, 1985. [2] E. Gustafsson, Water Air Soil Pollut. 80 (1995) 99. [3] D.W.M. Arrigan, Analyst 119 (1994) 1953. [4] Y.C. Tsai, B.A. Coles, K. Holt, J.S. Foord, F. Marken, R.G. Compton, Electroanalysis 13 (2001) 831. [5] J. Wang, B.M. Tian, Anal. Chem. 65 (1993) 1529. [6] J. Wang, J.M. Lu, S.B. Hocevar, P.A.M. Farias, B. Ogorevc, Anal. Chem. 72 (2000) 3218. [7] J. Wang, J.M. Lu, U.A. Kirgoz, S.B. Hocevar, B. Ogorevc, Anal. Chim. Acta 434 (2001) 29. [8] A. Krolicka, R. Pauliukaite, I. Svancara, R. Metelka, A. Bobrowski, E. Norkus, K. Kalcher, K. Vytras, Electrochem. Commun. 4 (2002) 193. [9] S.B. Hocevar, B. Ogorevc, J. Wang, B. Pihlar, Electoanalysis 14 (2002) 1707. [10] K. Vytras, I. Svancara, R. Metelka, Electroanalysis 14 (2002) 1359. [11] G.G. Long, L.D. Freedman, G.O. Doak, Bismuth and bismuth alloys, in: Encyclopedia of Chemical Technology, Wiley, New York, 1978, pp. 912 /937. [12] A. Economou, S.D. Bolis, C.E. Efstathiou, G. Volikakis, Anal. Chim. Acta 467 (2002) 179. [13] J. Wang, Stripping Analysis, VCH, Deereld Beach, FL, 1985, p. 19. [14] J. Osteryoung, J.J. ODea, Square wave voltammetry, in: A.J. Bard (Ed.), Electroanalytical Chemistry, Marcel Dekker, New York, 1986, pp. 209 /308. [15] J. Wang, R.P. Deo, S. Thongngamdee, B. Ogorevc, Electroanalysis 14 (2001) 1153. [16] W. Lund, R. Eriksen, Anal. Chim. Acta 107 (1979) 37.

The results are means of five determinations.

fact that Zn exists in large excess over Pb in tapwater (over 1:200 in terms of mass concentration). If the preconcentration potential was set to /1.2 V, a prominent Zn peak was also apparent but did not interfere with the Pb peak (Fig. 6(a)). In order to determine Zn in tapwater, dilution of the sample and use of shorter preconcentration time were necessary (under these conditions the Pb peak was just discernible; Fig. 6(b)). These experiments demonstrated that it was possible to use ASV on BFEs in order to determine metals coexisting at greatly varying concentrations by a judicious choice of the experimental conditions. Stripping voltammograms for the determination of Pb in human hair are shown in Fig. 7. The small peak at /1.05 V was due to the presence of Zn. The same determinations were also carried out by AAS which was used as a reference method. The comparative results, shown in Table 1, suggest that