Voltammetric studies on the electrochemical determination of methylmercury in chloride medium at carbon microelectrodes

Analytica Chimica Acta 579 (2006) 227–234

Voltammetric studies on the electrochemical determination of methylmercury in chloride medium at carbon microelectrodes F. Ribeiro a , M.M.M. Neto a,b,∗ , M.M. Rocha a , I.T.E. Fonseca a a

Centro de Electroqu´ımica e Cin´etica da Universidade de Lisboa, Departamento de Qu´ımica e Bioqu´ımica, Faculdade de Ciˆencias, Universidade de Lisboa, Campo Grande, Ed. C8, 1749-016 Lisboa, Portugal b Departamento de Qu´ımica Agr´ıcola e Ambiental, Instituto Superior de Agronomia, Tapada da Ajuda, 1349-017 Lisboa, Portugal Received 20 March 2006; received in revised form 6 July 2006; accepted 10 July 2006 Available online 14 July 2006

Abstract Electroanalytical techniques have been used to determine methylmercury at low levels in environmental matrices. The electrochemical behaviour of methylmercury at carbon microelectrodes in a hydrochloric acid medium using cyclic, square wave and fast-scan linear-sweep voltammetric techniques has been investigated. The analytical utility of the methylmercury reoxidation peak has been explored, but the recorded peak currents were found to be poorly reproducible. This is ascribed to two factors: the adsorption of insoluble chloromercury compounds on the electrode surface, which appears to be an important contribution to hinder the voltammetric signal of methylmercury; and the competition between the reoxidation of the methylmercury radical and its dimerization reaction, which limits the reproducibility of the methylmercury peak. These problems were successfully overcome by adopting the appropriate experimental conditions. Fast-scan rates were employed and an efficient electrochemical regeneration procedure of the electrode surface was achieved, under potentiostatic conditions in a mercury-free solution containing potassium thiocyanate—a strong complexing agent. The influence of chloride ion concentration was analysed. Interference by metals, such as lead and cadmium, was considered. Calibration plots were obtained in the micromolar and submicromolar concentration ranges, allowing the electrochemical determination of methylmercury in trace amounts. An estuarine water sample was analysed using the new method with a glassy carbon microelectrode. © 2006 Elsevier B.V. All rights reserved. Keywords: Methylmercury determination; Carbon microelectrodes; Square wave voltammetry; Fast-scan linear-sweep voltammetry; Water analysis

1. Introduction Monitoring of toxic chemical species in environmental samples has become an important analytical problem, prompting much research globally. Electroanalytical approaches are in many cases the techniques of choice because of their great precision, accuracy and relatively cheap cost [1]. Detection and quantification at trace level of different species of a chemical by these techniques in aquatic media usually does not require pretreatment—a significant advantage for monitoring programmes. Methylmercury is an extremely toxic chemical species that is found in the environment [2,3]. It usually appears as a result of the conversion of inorganic mercury by microorganisms into much more toxic organomercury forms, particularly the water soluble monomethylmercury cation. Its solubility in lipids leads

∗

Corresponding author. Tel.: +35 1962763055. E-mail address: [email protected] (M.M.M. Neto).

0003-2670/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.aca.2006.07.025

to elevated concentration in the biological tissues. The inclusion of methylmercury in the food chain is very harmful to humans because of its cumulative effects. Consequently the analytical determination of methylated forms of mercury in dilute samples is of primary importance for food safety. Although non-electrochemical methods for quantifying organomercury species have been widely applied, they are somewhat complicated and time-consuming, and usually require expensive instrumentation [4–9]. High pressure liquid chromatography (HPLC) with electrochemical detection has also been employed [10]. Procedural and economic advantages make voltammetric techniques very attractive for analyses. However, there are very few reports in the literature describing the voltammetric determination of methylmercury [11–15], most probably because of the complexity of the particular electrochemical reduction process [11,16]. Several papers have reported mechanistic studies on the electrochemical reduction of methylmercury, some of them presenting controversial results [17–19], but over the last 30 years the mechanism proposed by Heaton

228

F. Ribeiro et al. / Analytica Chimica Acta 579 (2006) 227–234

and Laitinen [11] has become accepted. Two single-electron steps are involved in this mechanism: one being the reversible generation of the methylmercury radical, the other being an electron-transfer reaction associated with the irreversible reduction of the methylmercury radical to mercury and methane. A linear calibration plot for the determination of methylmercury at a mercury drop electrode was obtained over the concentration range from 10−4 to 10−7 M [11]. Other analytical studies using mercury film [10,14] or gold film [12] electrodes have been reported. Lower detection limits down to the nanomolar concentration range were achieved using polymer-coated carbon electrodes [13,15]. The voltammetric determination of methylmercury has been usually based on the reversible wave, where the reduction to methylmercury radical takes place. However, the subsequent reactions involving the production of elemental mercury and other reaction products need to be investigated and taken into account for the analysis, especially when solid electrodes are employed. The shape and size of the reversible wave, as well as its poor reproducibility, appear to be influenced significantly by the modification of the solid electrode surface, which is due to the deposition of reaction products derived in subsequent steps, including free mercury. Therefore, the hypothesis of formation of a mercury-modified electrode during the electrochemical process has to be considered. Complexing matrices will inevitably influence this process [16]. Recently, the application of microelectrodes to electroanalysis has received considerable attention, as these offer many advantages over macroelectrodes [20–22]. High mass transport with low ohmic drop and charging currents make them very convenient for direct speciation measurements in resistive solutions in the presence of dissolved oxygen, when fast electrochemical techniques are employed. On the other hand, the goal is now to develop environmental friendly sensors as alternatives to the highly toxic mercury electrodes, which clearly are prone to unavoidable contamination of the environment. The trend is to avoid mercury as an electrode material thereby protecting the environment. We demonstrated in an earlier paper that carbon microelectrodes are promising tools for in situ determination of methylmercury in natural waters [16]. Firstly, they allow direct measurements in environmental samples taking advantage of the microelectrodes features, and secondly they support a clean methodology by using environmentally friendly electrodes. A method for the experimental determination of methylmercury in aqueous matrices is developed and applied to the analysis of a natural water sample. In the present work, the application of voltammetric techniques for monitoring low levels of methylmercury in natural aquatic systems using carbon microelectrodes is explored and investigated. The implications of the presence of chloride ions and the deposition of mercury compounds on the electrode surface are considered. The influence of metal ions, such as lead and cadmium, is also studied. 2. Experimental Supporting electrolyte solutions, hydrochloric acid and potassium thiocyanate, were prepared from reagents of Suprapur

quality (Merck) in Millipore Milli-Q ultrapure water (conductivity < 0.1 ␮S cm−1 ). Pb(II) and Cd(II) solutions were prepared from lead acetate and cadmium chloride (Merck) of analytical grade without further purification. The solutions were stored in Pyrex glass flasks for a few days. Transfer of methylmercury(II) chloride (Aldrich) require special Viton gloves (Sigma-Aldrich) and masks. Milimolar stock solutions of methylmercury(II) chloride were prepared by introducing weighed amounts of the salt into Pyrex volumetric flasks and then filling them to the marks with Milli-Q water. The working solutions were obtained by adding the stock solution to the appropriate supporting electrolyte. The disposable pipette tips employed were rinsed with a 1% HNO3 aqueous solution before being discarded. The experiments were performed in a homemade glass cell (25 mL) equipped with a platinum wire counter electrode, an Ag | AgCl | 3 M NaCl reference electrode, from Bioanalytical Systems, Inc. (BAS) and a carbon microelectrode as the working electrode. Carbon fiber microelectrodes (∅11 ± 2 ␮m), from BAS, and glassy carbon microelectrodes (∅10 ± 2 ␮m), from Princeton Applied Research (PAR), were employed. A platinum microelectrode (∅10 ± 2 ␮m) and a glassy carbon rotating disc electrode (area: 0.385 cm2 ) were also used. Electrodes were hand-polished with 0.05 ␮m alumina slurry (Buehler), rinsed thoroughly with Milli-Q water and allowed to dry. After running experiments, the microelectrodes were immersed in nitric acid in an ultrasound bath. Whenever found necessary they were also electrochemically cleaned to ensure a more efficient removal of the deposited mercury. A computer-controlled Autolab PGSTAT12 potentiostat (Eco Chemie, Utrecht, The Netherlands) equipped with a low current amplifier module (ECD) and a dual channel fast analogue-todigital converter (ADC750) with the GPES software (Version 4.9) was used for running the electrochemical experiments. A Bacharach Coleman Model 50B atomic absorption spectrometer equipped with a peristaltic pump for cold vapour analysis was employed for the photometric determination of mercury. Voltammograms were recorded keeping the cell in a Faraday cage, in order to eliminate electronic noise. All experiments were conducted at temperatures of (25 ± 1) ◦ C. 3. Results and discussion 3.1. Voltammetric behaviour of methylmercury at carbon microelectrodes The study of the electrochemical reduction of the methylmercury cation at a carbon electrode is an interesting but not a straightforward task. The experimental results are poorly reproducible and denote unusual electrochemical behaviour. Cyclic voltammetry of methylmercury in HCl was carried out at carbon microelectrodes. Similar cyclic voltammograms (CVs) were obtained using either carbon fiber or glassy carbon as electrode materials. Typical voltammograms run over different potential ranges are shown in Figs. 1 and 2. Fig. 1 illustrates the reduction of methylmercury, starting at −0.3 V, and the corresponding oxidation peak, A1 (Ep = −0.33 V). In progressing to more positive potentials, the presence of elemental mercury as

F. Ribeiro et al. / Analytica Chimica Acta 579 (2006) 227–234

Fig. 1. CVs of 500 ␮M CH3 Hg+ in 0.1 M HCl at a carbon fiber microelectrode, at various sweep rates. Ei = 0 V (vs. Ag | AgCl).

a reaction product of the reduction of methylmercury is seen in Fig. 2a, where a reduction peak at −0.064 V, C2 , and an oxidation peak at + 0.14 V, A2 , indicate the presence of mercury. At slower scan rates (v < 0.2 V s−1 ) an irreversible cathodic wave at ca. −1.2 V, C3 , was recorded (Fig. 2b). These voltammetric data corroborate the following mechanism proposed by Heaton and Laitinen [11] for the electrochemical reduction of methylmercury at a DME, in acidic solutions: CH3 Hg+ + e−  CH3 Hg•

(1)

2CH3 Hg•  (CH3 Hg)2

(2a)

(CH3 Hg)2  (CH3 )2 Hg + Hg +

−

(2b)

CH3 Hg• + H + e → CH4 + Hg

0

(3)

It should be pointed out that the reoxidation peak of the radical species CH3 Hg• was not noticeable before metallic mercury was formed on the electrode surface, which denotes a somewhat irreversible behaviour of the redox couple at the carbon electrodes. A parallel study performed for comparison with a

229

platinum microelectrode was found quite unsuccessful, probably due to a weak adhesion of mercury on platinum. It is well known that carbon is a much more suitable substrate for mercury deposition [23,24]. From these results we conclude that the electrode process does occur preferably when mercury is present on the electrode surface. The influence of scan rate is also illustrated in Figs. 1 and 2. At fast sweep rates, the oxidation peak current of the radical species (reaction (1)) is enhanced. The analysis of this dependence is reported in an earlier paper [16], where an explanation is given based on the occurrence of an EC mechanism for high sweep rates (v > 50 V s−1 ) in which the fast dimerization of the radical takes place (reaction (2)). In contrast, the irreversible cathodic peak (reaction (3)) which appears at −1.2 V, is hidden at high sweep rates (Fig. 2b), due to the hydrogen evolution interference. Square wave voltammetry (SWV) was then applied to more dilute solutions of methylmercury cation. Considering that the electrochemical process (reaction (1)) is followed by a chemical reaction that is second order in the radical (reaction (2)), for low concentrations (≤10−6 M), the monomer should be favoured when the equilibrium is reached, according to other authors [11]. However, the electrochemical signal of methylmercury was only found to be detectable for relatively high concentration values (of the milimolar order of magnitude). Well-defined peak currents were recorded for lower methylmercury concentrations when the electrode potential was hold at −1.0 V for a period of time before running the square wave scan. This experimental evidence demonstrates that the electrochemical oxidation of methylmercury radicals at carbon electrodes only takes place when metallic mercury is previously formed on the electrode surface. In fact, the production of mercury is in agreement with the proposed mechanism. Linear dependence of peak current on accumulation time was observed for moderate concentrations. For a 0.5 mM methylmercury solution, a regression equation given by Ip (nA) = 0.011 ta + 0.346 was obtained, yielding a regression coefficient of 0.9985 (n = 6). These unexpected results reflect the influence of the amount of mercury accumulated on the elec-

Fig. 2. CVs of 500 ␮M CH3 Hg+ in 0.1 M HCl at a carbon fiber microelectrode. (a) At 1 V s−1 , Ei = −1.0 V; (b) at 0.1 V s−1 , Ei = 0 V (vs. Ag | AgCl).

230

F. Ribeiro et al. / Analytica Chimica Acta 579 (2006) 227–234

trode surface, and suggest the occurrence of a catalytic effect of mercury on the reoxidation of methylmercury. Although these results may sound satisfactory and reasonably encouraging to go on with the development of an electroanalytical methodology to determine methylmercury, it should be emphasized that problems related to poor reproducibility of the methylmercury reoxidation peak remain to be solved. On the one hand, the results seem to be largely affected by reaction products of the side reactions that may occur depending on experimental conditions. Those products are prone to be adsorbed on the electrode surface, degrading its performance. The presence of hydrochloric acid should undoubtedly contribute to those unwanted effects, since Cl− will form complexes with CH3 Hg+ , as well as with mercury. In addition, mercury will not completely dissolve during anodic polarization in solutions containing chloride ions, where the formation of calomel and maybe also a mixed compound is likely [25–27]. On the other hand, the poor reproducibility of the methylmercury reoxidation peak, which is very often too small or nearly non-existent in many of the recorded voltammograms, may be attributed to the fact that the reoxidation of the methylmercury radical is competing with not only its fast dimerization and the subsequent slow formation of dimethylmercury and elemental mercury (reaction (2)), but also with the irreversible reduction of the radical in acidic medium giving free mercury and methane (reaction (3)). Even when fast sweep rates are employed, sometimes the peak appears, sometimes it does not . . . and consequently the analytical utility of the methylmercury reoxidation peak may become questionable. Actually, under the present experimental conditions, one cannot rely on outrunning reaction (2) to quantitatively determine methylmercury based on reaction (1), even when high scan rate values are applied [14]. Since mercury was not added to the solutions, cyclic voltammetry was carried out in an attempt to better understand the formation of mercury on the electrode surface as a result of the electrochemical reduction of methylmercury. The reoxidation reaction is difficult to occur at carbon electrodes [28] but, after the electrode undergoes a couple of cycles, mercury droplets are supposed to start growing on the carbon surface due to the deposition of metallic mercury produced in reactions (2) and (3), and the electrode should be functioning as a mercury-modified electrode. In fact, the anodic scan that succeeds to the first cathodic scan always presents the oxidation peak of mercury at a slightly positive potential, even when the methylmercury reoxidation peak is absent in the cyclic voltammograms. This is a clear indication that the reduction of methylmercury is definitely occurring, since the mercury oxidation peak grows with continuous cycling. In the first cycles, at low scan rates, typical voltammograms exhibit non-steady curves showing oscillations in current, due to the production and growth of Hg(0) nuclei on the carbon surface [29]. This is almost overcome with the application of a few cycles (two or three); the formation of a more permanent deposit of mercury on the electrode surface appeared to be a favourable factor, leading to smoother voltammogram recordings. Unfortunately, the mercury reoxidation peak current is yet non reproducible, evolving erratically.

3.2. Regeneration of the electrode surface The adsorption of mercury compounds on the electrode surface is reflected by the difficulty in regenerating the electrodes. After running a few voltammograms in solutions containing methylmercury at both either carbon fiber or glassy carbon microelectrodes, attempts to clean the electrodes by removing the deposits using electrochemical oxidation, holding them at a fixed positive potential value (E = + 0.5 V versus Ag | AgCl) for a period of time (30 min or more), were found to be unsuccessful. Wiping with a wet filter paper did not remove completely the deposits either. Moreover, even after carrying out a polishing with alumina powder followed by immersion in concentrated nitric acid, the electrodes still responded to traces of mercury when they underwent potentiodynamic cycling in a mercury-free background electrolyte solution. The following experiment done with a glassy carbon macroelectrode confirmed these observations. Voltammetry of methylmercury in chloride medium was performed at a rotating glassy carbon disc electrode. The aforementioned cleaning process was then used in an attempt to regenerate the electrode surface. Instead of the shiny “mirror-like” black surface characteristic of polished glassy carbon, a persisting grey colour was visible at naked eye. Only a time-consuming procedure, consisting of a 2–3 h treatment with nitric acid in an ultrasound bath was found successful to restore the electrode surface. Thus, one can conclude that some insoluble mercury compounds remain in fact strongly attached to the electrode surface as a result of the oxidation of the methylmercury reduction reaction products in chloride solution. The oxidation of mercury may be accompanied by the formation of calomel and/or a mixed compound [27,30] that will not be totally reduced to mercury at negative potentials. In order to ensure a proper regeneration of the carbon microelectrode surface, an electrochemical treatment was carried out holding the electrode at −0.8 V, for 300 s, in a chloride-free solution containing a strong complexing agent, 0.1 M KSCN, to remove all traces of mercury species. The efficiency of this cleaning process was evidenced by the electrode response obtained in blank solutions, which showed no traces of mercury. Carbon fiber electrodes were found to be much more problematic to regenerate than the glassy carbon ones. 3.3. Influence of chloride concentration on the voltammetric response of methylmercury Published work concerning the electrochemical behaviour of mercury-modified carbon electrodes in chloride medium [27,30–33] indicates that the formation of calomel is detrimental to the mercury film. According to Nolan and Kounaves [31] and taking into account the results reported by other authors [27,32], two forms of the mercury(I) chloride are produced, depending on the mole ratio of Cl− /Hg2+ in solution: an electrochemically generated compound, which is reversibly reduced to mercury in the cathodic polarization, and a nonelectrochemically generated calomel, which is formed under open-circuit conditions according to: Hg(0) + Hg(II) → Hg2 2+ + 2Cl− → Hg2 Cl2 (s). The latter is not reduced during the cathodic polarization, but it is reduced during the anodic scan producing a “reverse peak”. In solutions

F. Ribeiro et al. / Analytica Chimica Acta 579 (2006) 227–234

Fig. 3. CV from a glassy carbon microelectrode in 500 ␮M CH3 Hg+ + 0.05 M HCl. v = 20 V s−1 ; Ei = −0.8 V (vs. Ag | AgCl). The CH3 Hg• oxidation peak, A1 , is shown in the inset.

5 × 10−6 M

Hg2+

containing less than in this anomalous phenomenon is not observed [27]. We investigated the influence of chloride concentration on the electrochemical reaction of methylmercury at a glassy carbon microelectrode, using seven different chloride concentrations ranging from 0.001 to 1.0 M. In contrast to previous reports, no “reverse” peaks were observed in our study. Consequently we conclude that no significant amount of free Hg2+ was present in our methylmercury working solutions. The cyclic voltammogram pattern shown in Fig. 3 exhibits differences in the shape of the oxidation and reduction peaks for mercury, that are indicative of the formation of calomel during the anodic scan, with mercury being oxidized to Hg(I) which promptly reacts with chloride, followed by the reduction to mercury in the cathodic scan. The oxidation peak, A2 , shows the tailing typical of a diffusion-controlled wave, while the corresponding reduction peak is sharper, suggesting the destruction of an insoluble film. Square wave anodic stripping voltammetry was also performed. A typical voltammogram showed two peaks (Fig. 4) characteristic of the reoxidation of the methylmercury radical (A1 ) and mercury (A2 ). Linear dependence of peak currents (A1 and A2 ) on accumulation time was obtained in media of lower chloride concentrations (≤0.1 M HCl). For higher chloride concentrations (≥0.5 M), this dependence started to deviate from linearity. The oxidation peak currents always increased with increasing chloride concentration, the corresponding peak potentials being shifted to more negative values. These experimental results clearly indicate the formation of calomel and/or other insoluble chloromercury compounds on the electrode surface, which are not completely removed at negative potentials. The fit to linear relationships observed between the square wave peak potentials, Ep (A1 ) and Ep (A2 ), and the logarithm of chloride concentration, given by Ep (A1 ) = −0.026 ln[Cl− ] − 0.048 and Ep (A2 ) = − 0.023 ln[Cl− ] − 0.0083, and having the following regression coefficients, 0.994 (n = 7) and 0.997 (n = 7), respec-

231

Fig. 4. SWV of a 500 ␮M CH3 HgCl in 0.1 M HCl at a glassy carbon microelectrode, after pretreatment at −1.0 V for 300 s. Ei = −1.0 V to Ef = −1.0 V (vs. Ag | AgCl). Square wave parameters: amplitude = 70 mV, Es = 3 mV, frequency = 50 Hz.

tively, also are consistent with this hypothesis [33]. In addition, the recovery of the electrode surface was more difficult as the concentration of chloride in solution increased above 0.5 M. This problem could be overcome replacing bare carbon electrodes by glassy carbon electrodes coated with anionexchangers. According to Moretto et al., the ion-exchange sites of the polymer coating interact with Cl− providing a locally elevated chloride concentration, so that the oxidation process of mercury to calomel will be followed by its disproportionation reaction: Hg2 Cl2 (s) + 2Cl− → Hg0 + HgCl4 2− , thereby effectively regenerating the electrode surface [34]. Voltammetric studies of methylmercury at polymer-coated electrodes in media with chloride concentration ≥0.5 M are now in progress. 3.4. Interferences Possible interferences include electroactive trace elements, such as lead and cadmium, and also inorganic mercury. Lead is undoubtedly a serious interferent because of the likely overlap of its stripping peak and the methylmercury oxidation peak. In a chloride medium, these peak potentials are particularly close, being about −0.45 and −0.40 V, for lead and methylmercury, respectively. As a result of the addition of lead to a methylmercury solution in the presence of chloride ions, the peak current of methylmercury at −0.385 V increased (Fig. 5), which means a slightly positive shift of the lead stripping peak potential. This suggests the presence of methyllead species in solution, such as (CH3 )3 Pb+ , as a consequence of the methylation of lead [35]. In contrast, extending the anodic excursion of the potential to the positive range, the anodic stripping peaks characteristic of mercury oxidation, A1 , do not show any increment, as depicted in Fig. 5. Consecutive additions of methylmercury chloride and lead acetate solutions were performed in a blank solution. The corresponding peak currents, recorded before and after each addition of lead, Ip,b and Ip,a , respectively, produced the fol-

232

F. Ribeiro et al. / Analytica Chimica Acta 579 (2006) 227–234

Fig. 5. Linear-sweep voltammograms of 10 ␮M CH3 HgCl in 0.1 M HCl at a glassy carbon microelectrode, in the absence (broken line) and presence (solid line) of 10 ␮M Pb(II). v = 50 V s−1 . A potential of −1.0 V was applied for 300 s before running the voltammogram. Ei = −0.8 V to Ef = + 0.5 V (vs. Ag | AgCl).

lowing ratios, Ip,b /Ip,a , 1.07, 0.89 and 0.80. From these results, it seems to us that the presence of lead and its presumed methylation partly inhibited the reduction of methylmercury. Cadmium present at a concentration 104 -fold more than that of methylmercury chloride was found not to interfere with the methylmercury voltammetric signal. The interfering effect of inorganic mercury, HgII , was also examined in 0.1 M HCl medium, when present at a 102 -fold excess over the methylmercury cation. Linear-sweep voltammetry, at 20 V s−1 , showed that the A1 peak current, which corresponded with mercury oxidation, increased significantly after the addition of HgII to the methylmercury solution. However, limiting the anodic excursion to 0.1 V, it was observed that the A2 peak, which is characteristic of the oxidation of the

Fig. 6. Dependence of peak current (at −0.35 V) on CH3 HgCl concentration. Ei = −0.6 V to Ef = −0.1 V (vs. Ag | AgCl). Other conditions as in Fig. 5.

methylmercury radical, was not affected. The presence of inorganic mercury is tolerated probably due to the large difference between the oxidation potential of methylmercury and that of mercury. 3.5. Analytical performance and application Although relatively fast-scan rates can be achieved using square wave voltammetry, it appeared not to be fast enough to reversibly oxidize the methylmercury radicals at carbon microelectrodes. Hence, fast-scan linear-sweep voltammetry at 50 V s−1 with a previous 300 s accumulation time at −1.0 V was adopted as a compromise between sensitivity and speed of analysis. A calibration plot of anodic peak current, A1 , against methylmercury concentration was constructed using standard additions (Fig. 6). There was a close fit to linearity from 0.3 to

Fig. 7. Fast-scan linear-sweep voltammetry, at a glassy carbon microelectrode, for increasing levels of CH3 HgCl in 0.1 M HCl. Ei = −1.0 V to Ef = + 1.0 V (vs. Ag | AgCl). Other conditions as in Fig. 5. Also shown on the right is the resulting calibration plot.

F. Ribeiro et al. / Analytica Chimica Acta 579 (2006) 227–234

3.0 ␮M (r = 0.996, intercept equals 1.10 nA), but above 3.0 ␮M the plotted values deviated from linearity. The reproducibility of measurements was evaluated from the relative standard deviations (5–12%) obtained for five consecutive runs. The results shown in Fig. 6 lead us to conclude that, at the applied scan rate (50 V s−1 ), for concentrations higher than 3.0 ␮M, the dimerization rate of the methylmercury radicals increases (reaction (2)), hindering the reversible reoxidation to methylmercury cations (reaction (1)). Furthermore, the accumulation of calomel and/or other chloromercury compounds on the electrode surface likely contribute to the decline in the methylmercury voltammetric signal. For more dilute solutions (c ≤ 0.3 ␮M) the oxidation peak currents of methylmercury were barely visible, so a similar analysis was not possible. However, by extending the scan of potential to more positive values, a series of well-defined mercury oxidation peaks (peak A2 , in Fig. 5) were recorded for methylmercury concentrations between 1 and 100 nM (Fig. 7). The peak current measured at +0.25 V increased proportionally with logarithm of methylmercury concentration, yielding a highly linear calibration plot (Fig. 7, right) described by I(nA) = 5.589 log CCH3 HgCl + 60.45, with a correlation coefficient of 0.998. The reproducibility of these measurements is evident from the relative standard deviations (3–10%) obtained in four consecutive runs. The method was applied to a water sample collected from the Aveiro Ria, an estuary on the western coast of Portugal. It was filtered through a 44 ␮m filter and analysed within 48 h. The sample was tested for total mercury by cold vapour AAS (atomic absorption spectroscopy) with no preconcentration, and showed no appreciable mercury content (detection limit: 5.0 pM Hg). Application of the voltammetric method described above did not show any traces of mercury or methylmercury either. Spiking the water sample with five standard additions of methylmercury chloride resulted in recoveries of 86–92%, from comparison with the data of the calibration plots. After each measurement the electrode was refreshed under potentiostatic conditions at −0.8 V, for 300 s, in a mercury-free potassium thyocianate solution to get a reproducible electrode surface. 4. Conclusions The presented data clearly indicate that carbon microelectrodes are promising tools for monitoring methylmercury in environmental samples by fast-scan linear-sweep voltammetry. The obtained results strongly suggest that, in chloride medium, insoluble chloromercury compounds are formed and remain attached on the electrode surface, hindering its response. The removal of the deposit is problematic, but an efficient regeneration procedure was achieved by potentiostating the electrode at −0.8 V for 300 s, in a mercury-free potassium thiocyanate solution. Glassy carbon was elected the preferred electrode material, since it permits an easier and more efficient regeneration of the electrode surface than carbon fiber. Also the competition between the dimerization of the methylmercury radical and its reoxidation can seriously affect the reproducibility of the methylmercury peak current. Fast-scan linear-sweep voltam-

233

metry preceded by potentiostatic control at −1.0 V for 300 s, produced satisfactory results for the development of an electroanalytical method to determine trace amounts of methylmercury at glassy carbon microelectrodes. However, we emphasize that the detection limit attained was still quite high for real analysis of natural waters. Lower concentration detection limits are expected to be achieved using more sophisticated equipment, which will enable to operate at faster scan rates. Also the use of ion-exchanger polymeric coatings on the glassy carbon electrode surface looks very promising as a means of improving the signal of the methylmercury sensor. Attempts will be made to optimise the proposed methodology in the near future. Acknowledgements Financial support from the Fundac¸a˜ o para a Ciˆencia e a Tecnologia (FCT) to Project POCTI/QUI/37903/2001, under the Operational Program of Science, Technology and Innovation (POCTI), is gratefully acknowledged. F. Ribeiro acknowledges FCT for Ph.D. Grant SFRH/BD/8840/2002. We thank Mr. J. Ferreira for assistance with AAS measurements. References [1] J. Wang, Analytical Electrochemistry, VCH, New York, 1994, p. 1. [2] F. Zahir, S.J. Rizwi, S.K. Haq, R.H. Khan, Environ. Toxicol. Pharmacol. 20 (2005) 351. [3] N.S. Bloom, L.M. Moretto, P. Scopece, P. Ugo, Mar. Chem. 91 (2004) 85. [4] W. Holak, Analyst 107 (1982) 1457. [5] D.S. Byshee, Analyst 113 (1988) 1167. [6] M. Horvat, L. Liang, N.S. Blowm, Anal. Chim. Acta 282 (1993) 153. [7] C.J. Capon, J.C. Smith, Anal. Chem. 49 (1976) 365. [8] N.S. Bloom, E. Preus, J. Katon, M. Hiltner, Anal. Chim. Acta 479 (2003) 233. [9] L. Liang, M. Horvat, N.S. Bloom, Talanta 41 (1994) 371. [10] W.A. MacCrehan, R.A. Durst, Anal. Chem. 50 (1978) 2108. [11] R.C. Heaton, H.A. Laitinen, Anal. Chem. 46 (1974) 547. [12] J. Ireland-Ripert, A. Bermond, C. Ducauze, Anal. Chim. Acta 143 (1982) 249. [13] R. Agraz, M.T. Sevilla, L. Hernandez, J. Electroanal. Chem. 390 (1995) 47. [14] R. Lai, E.L. Huang, F. Zhou, D.O. Wipf, Electroanalysis 10 (1998) 926. [15] L.M. Moretto, P. Ugo, R. Lacasse, G.Y. Champagne, J. Chevalet, J. Electroanal. Chem. 467 (1999) 193. [16] F. Afonso, F. Ribeiro, L. Proenc¸a, M.I.S. Lopes, M.M. Rocha, M.M.M. Neto, I.T.E. Fonseca, Electroanalysis 17 (2005) 127. [17] V. Vojir, Collect. Czech. Chem. Commun. 16 (1951) 489. [18] R. Benesch, R.E. Benesch, J. Phys. Chem. 56 (1952) 648. [19] B. Fleet, R.D. Jee, J. Electroanal. Chem. 25 (1970) 397. [20] M.I. Montenegro, in: R.G. Compton, R.G. Hancock (Eds.), Research Chemical Kinetics, vol. 2, Elsevier, New York, 1994. [21] P.R.M. Silva, M.A. El Khakani, M. Chaker, G.Y. Champagne, J. Chevalet, L. Gastonguay, R. Lacasse, M. Ladouceur, Anal. Chim. Acta 385 (1999) 249. [22] A. Jaworski, Z. Stojek, J.G. Osteryoung, J. Electroanal. Chem. 558 (2003) 141. [23] T.M. Florence, J. Electroanal. Chem. 27 (1970) 273. [24] M. Stulikova, Electroanal. Chem. Interf. Electrochem. 48 (1973) 33. [25] L. Luong, F. Vydra, J. Electroanal. Chem. 50 (1974) 379. [26] T.M. Florence, Anal. Chim. Acta 119 (1980) 217. [27] N.F. Zakharchuk, Kh.Z. Brainina, Electroanalysis 10 (1998) 379.

234

F. Ribeiro et al. / Analytica Chimica Acta 579 (2006) 227–234

[28] D. Durst, SEAC Communications—The Newsletter On-Line, November 1988, vol. 5. No. 4 (http://electroanalytical.org/SEACcom/SEACcomnov88.pdf). [29] E. Sahlin, D. Jagner, R. Ratana-Ohpas, Anal. Chim. Acta 346 (1997) 157. [30] H. Gunasingham, K.P. Ang, C.C. Ngo, Analyst 113 (1988) 1533.

[31] [32] [33] [34]

M.A. Nolan, S.P. Kounaves, Electroanalysis 12 (2000) 96. W. Frenzel, Anal. Chim. Acta 273 (1993) 123. T.M. Florence, Anal. Chim. Acta 119 (1980) 217. L.M. Moretto, G.A. Mazzocchin, P. Ugo, J. Electroanal. Chem. 427 (1997) 113. [35] N. Mikac, Y. Wang, R.M. Harrison, Anal. Chim. Acta 326 (1996) 57.