Anal. Chem. 2006, 78, 5052-5060
Voltammetric Detection of Mercury and Copper in Seawater Using a Gold Microwire Electrode Pascal Salau 1 n and Constant M. G. van den Berg*
Department of Earth and Ocean Sciences, University of Liverpool, Liverpool, L69 3GP, UK
A procedure is presented by which mercury and copper are determined simultaneously in seawater and dilute acid (0.01 M HCl) by anodic stripping voltammetry using gold microwire electrodes. It was found that anion (halide) adsorption is the cause for a gradual decrease in the height and potential of the mercury peak. The effect is eliminated by including an anion desorption step in the analysis at -0.8 V prior to each scan. This step was found to greatly improve the stability of the scans and enabled the use of background subtraction. Advantages of the microwire electrodes were a low roughness of the surface, without a need for pretreatment, and a very small diffusion layer (2 µm with stirring). Under the optimized voltammetric conditions, the detection limits were 6 pM mercury and 25 pM copper using 300-s deposition. These values are well below those reported previously for other electrodes including rotating disk electrodes. Measurements of the influence of the major anions I-, Br-, Cl-, SO42-, F-, HCO3-, and B(OH)4 on the response for mercury showed that bromide and chloride are predominantly responsible for the underpotential deposition mechanism of mercury in seawater. The method was applied to coastal water samples from Liverpool Bay. The determination of mercury in seawater is a challenge because of the very low levels encountered from subpicomolar levels in the surface of the Pacific1 to low-picomolar levels in the North Atlantic and increasing near the Celtic Sea.2 Mercury can be harmful to the environment and human health because of processes of methylation, bioaccumulation, and biomagnification.3 Mercury is known to be released from contaminated sediments and is available to organisms for periods lasting many years.4 There is therefore a need for long-term monitoring of mercury and if possible including its speciation. Such monitoring would be facilitated by the availability of a voltammetric method based on a microelectrode, as this minimizes the requirement for stirring, convenient for in situ detection. * Corresponding author: (e-mail)
[email protected]. (1) Laurier, F. J. G.; Mason, R. P.; Whalin, L.; Gill, G. A. Mar. Chem. 2004, 90, 3. (2) Cossa, D.; Bretaudeau-Sanjuan, J.; Cotte´-Krief, M. H.; Mason, R. P. Mar. Chem. 2004, 90, 21. (3) Morel, F. M. M.; Kraepiel, A. M. L.; Amyot, M. Annu. Rev. Ecol. Syst. 1998, 29, 543-566. (4) Sunderland, E. M.; Gobas, F. A. P. C.; Heyes, A.; Branfireun, B. A.; Bayer, A. K.; Cranston, R. E.; Parsons, M. B. Mar. Chem. 2004, 90, 91.
5052 Analytical Chemistry, Vol. 78, No. 14, July 15, 2006
Much work has been published to optimize voltammetric methods for mercury detection. Gold electrodes are suitable if mercury coverage at the end of the deposition step is less than a monolayer.5 In this case, a single symmetrical peak is apparent at a much more positive potential than for bulk deposition of mercury on gold. This is known as the underpotential deposition (UPD) mechanism. Studies of the UPD mechanism of mercury on polycrystalline6 or single-crystal gold7,8 have shown that, at submonolayer coverage, the adsorbed mercury layer is stable and its structure is strongly dependent on the crystallographic plane of the gold substrate and on the interactions with the anions present in the solution. Reproducible cathodic and anodic peaks obtained by cyclic voltammetry on Au(111) electrodes in chloride solutions are thought to be due to the deposition and dissolution of calomel (Hg2Cl2).7,8 Once the coverage exceeds one monolayer, at higher mercury concentrations, or after a long deposition time, the mercury becomes fully reduced to Hg(0) and begins to diffuse into the gold:9 this causes gradual changes in the gold surface and leads to a shift in the peak potential and a lowering of the peak height. However, the deposition from uncontaminated seawater usually occurs in the UPD regime, due to the low mercury levels. Voltammetric detection of mercury in seawater using gold electrodes is normally by anodic stripping voltammetry (ASV) combined with medium exchange. Thus, using rotating disk electrodes, detection limits (DL) have been achieved of between 15 (15-min plating)10 and 10 pM (60-min plating).11 Using chronopotentiometric stripping analysis (CSA) and a 10-µm gold microwire,12 the DL was 225 pM Hg, also with medium exchange. Direct detection without medium exchange has been demonstrated with a 100-µm gold wire electrode with detection by CSA13,14 with a DL of 300 pM Hg and using a gold rotating disk electrode (AuRDE) with a DL of 25 pM.15 In low chloride solutions (5) Andrews, R. W.; Larochelle, J. H.; Johnson, D. C. Anal. Chem. 1976, 48, 212-214. (6) Salie, G. J. Electroanal. Chem. 1988, 245, 1-20. (7) Li, J.; Herrero, E.; Abruna, H. D. Colloids Surf., A 1998, 134, 113-131. (8) Fabregas, J.; Herrero, C. J. Ind.l Microbiol. 1990, 5, 259-263. (9) Salie, G.; Bartels, K. J. Electroanal. Chem. 1988, 245, 21-38. (10) Sipos, L.; Nurnberg, H. W.; Valenta, P.; Branica, M. Anal. Chim. Acta 1980, 115, 25-42. (11) Gustavsson, I. J. Electroanal. Chem. 1986, 214, 31-36. (12) Huang, H. L.; Jagner, D.; Renman, L. Anal. Chim. Acta 1987, 201, 1-9. (13) Wang, J.; Tian, B. M.; Lu, J. M.; Wang, J. Y.; Luo, D. B.; MacDonald, D. Electroanalysis 1998, 10, 399-402. (14) Daniele, S.; Bragato, C.; Baldo, M. A.; Wang, J.; Lu, J. Analyst 2000, 125, 731. (15) Riso, R. D.; Waeles, M.; Monbet, P.; Chaumery, C. J. Anal. Chim. Acta 2000, 410, 97-105. 10.1021/ac060231+ CCC: $33.50
© 2006 American Chemical Society Published on Web 05/27/2006
(10 mM), low levels of mercury can also be determined by ASV and an AuRDE without medium exchange with a DL of 50 pM for 2-min deposition.16 An alternative approach for the detection of mercury is the use of chemically modified electrodes.17,18 They are based either on the chemical interaction between Hg(II) and a functional group attached to the surface of the electrode or the attached modified layer prevents the adherence of interfering substances. Although these electrodes have potential in terms of speciation studies and measurements of trace metals in the presence of surfactants or organic compounds, they can suffer from a high DL, or have stability problems, and are not yet ready for routine analysis. Microelectrodes have been used only occasionally for the detection of mercury. A heated 25-µm gold microwire had a DL of 0.4 nM Hg after 10-min deposition,19 and a 10-µm wire was used for higher levels of mercury in urine by flow analysis.20 The DL was 5 nM Hg using a gold microelectrode array21 and 0.5 nM Hg with an iridium microelectrode array,22 both much higher than expected. The previous work appears to indicate that low mercury levels in seawater cannot be determined using a gold electrode by ASV without medium exchange or surface modification, but for some reason, this is possible when detection is by CSA. Only one paper demonstrates the possibility of direct determination of mercury in natural waters,14 but this had a DL of 300 pM Hg. The aim of this work was to improve our understanding of the electrode processes underlying the mercury determination and to develop a voltammetric procedure for the detection of mercury in chloride media including seawater, which is ultimately suitable for in situ detection in the marine environment. Gold microwire electrodes of between 5- and 100-µm diameter were used and the analytical parameters were optimized. The influence of the major anions (chloride, bromide, iodide, fluoride, bicarbonate) was investigated and demonstrated the importance of chloride and bromide on the underpotential deposition of mercury in seawater. Although the voltammetric procedure was developed for mercury, the same procedure can be used for the detection of copper in seawater, which is therefore detected simultaneously. The ASV new procedure is relatively simple but more sensitive than previous voltammetric methods, allowing determination of picomolar levels of mercury and copper in coastal waters. EXPERIMENTAL SECTION Reagents. Solutions were prepared using Milli-Q water (18.2 MΩ‚cm resistivity). Standard solutions of Cu and Hg were prepared by appropriate dilution of the stock solutions (1000 mg‚L-1 in nitric acid, atomic absorption standard solutions, Aldrich) and acidified to pH 2 with HCl. HCl and HNO3 were purified by sub-boiling distillation. The lowest concentrated (16) Bonfil, Y.; Brand, M.; Kirowa-Eisner, E. Anal. Chim. Acta 2000, 424, 6576. (17) Widmann, A.; Van Den Berg, C. M. G. Electroanalysis 2005, 17, 825-831. (18) Ugo, P.; Moretto, L. M.; Bertoncello, P.; Wang, J. Electroanalysis 1998, 10, 1017-1021. (19) Wang, J.; Grundler, P.; Flechsig, G. U.; Jasinski, M.; Lu, J. M.; Wang, J. Y.; Zhao, Z. Q.; Tian, B. M. Anal. Chim. Acta 1999, 396, 33-37. (20) Huiliang, H.; Jagner, D.; Renman, L. Anal. Chim. Acta 1987, 202, 117122. (21) Uhlig, A.; Schnakenberg, U.; Hintsche, R. Electroanalysis 1997, 9, 125129. (22) Nolan, M. A.; Kounaves, S. P. Anal. Chem. 1999, 71, 3567-3573.
solutions were prepared daily. NaF, KI, KBr, NaCl, Na2SO4, NaHCO3, H2SO4, K3Fe(CN)6, and KCl were all Analar grade from BDH. Apparatus. Voltammetric experiments were performed with a PGSTAT 20 (Ecochemie) with associated GPES software, and a 663 VA stand (Metrohm) was used to hold the electrodes. Microelectrodes made from gold microwires of 5, 25, and 100 µm (GoodFellow), as described below, were used as working electrodes. The reference electrode was double-junction Ag/AgCl// KCl (3 M) with NaNO3 (0.1 M) in the salt bridge. For measurements of low mercury concentration in seawater, the bridge was filled with 0.1 M HCl. The auxiliary electrode was a 2-mm-long iridium wire of 100-µm diameter fabricated in the same manner as the working electrodes. All electrical connections and cables were carefully wrapped in aluminum foil and grounded to minimize electrical noise as much as possible. To minimize contamination problems, the PTFE cap of the Metrohm VA663 stand was cleaned by soaking in 0.1% KMnO4 in 4% H2SO4 to remove any adsorbed mercury10 followed by 30 min in 0.1 M oxalic acid to remove brown coloring due to deposits of MnO2. For measurements of low levels of mercury in seawater samples, the inner cell compartment was blanketed continuously with nitrogen to prevent the ingression of mercury from the stand or the laboratory atmosphere: for this reason, nitrogen was passed at a rate of 100 mL‚min-1 through a length of Teflon tubing through the cap above the solution. Experiments were carried out either in 50-mL polyethylene beakers or in a quartz voltammetric cell. The quartz cell and beakers were stored in 10-2 M HCl. The scanning electrode microscope was a Philips XL30. The oven used to fabricate the microwire electrodes was homemade: 23 the actual oven was a vertical quartz tube of 1-cm diameter surrounded by a heating coil; it could be set to temperatures up to 800 °C but was set to 400 °C to melt the pipet tip holding the microwire electrode. The UV digestor was home-built and contained a 100-W high-pressure mercury vapor lamp surrounded by four 30-mL PTFE-capped, quartz, tubes. Method To Fabricate Microwire Electrodes. The procedure to fabricate the microwire electrodes (5 or 25 µm) was adapted from that described before.23,24 A 10-cm length of 100-µm copper wire was passed through a 200-µL pipet tip (polyethylene, Corning Inc.). Its end was dipped in a conductive silver solution (Leitsilber L100, Maplin, UK), which had been freshly agitated and acted as conductive adhesive. The copper wire was then attached to a 10mm length of gold microwire (5 or 25 µm) by gently touching it. The copper wire was then carefully pulled through the pipet tip until the microwire passed halfway through. The microwire was sealed in the tip by holding it in the top of the oven, which had been set to 400 °C, during 8 s to melt it uniformly. The pipet was then held vertically in the air to cool for ∼30 s. The 100-µm wire was detached from the 25-µm wire by gently rotating the pipet tip or pushed inside the pipet to ensure a better connection when a 5-µm wire was used. A robust electrical contact was made by back-filling the tip with silver, conducting, epoxy resin (SL65, Rite Lok) and inserting an electrical wire (∼2-mm diameter). The microwire was cut to the desired length with scissors. The (23) Billon, G.; van den Berg, C. M. G. Electroanalysis 2004, 16, 1583-1591. (24) Nyholm, L.; Wikmark, G. Anal. Chim. Acta 1992, 257, 7-13.
Analytical Chemistry, Vol. 78, No. 14, July 15, 2006
5053
Table 1. Optimized Voltammetric Procedures for Detection of Mercury in 0.01 M HCl and Seawater conditioning step
analytical scan
background scan
HCl (0.01 M)
(1) Econd ) 0.65 V for 15s (2) SW scan 0 f 0.65 V
(1) Edes ) -0.8 V for 15s (2) equilibration 15 s (3) SW scan 0 f 0.65 V
seawater
(1) open circuit for 15 s (2) Econd ) 0.5 V for 15 s (3) Edep ) -0.8 V for 2 s (4) CSA scan to E ) 0.5 V
(1) Edep ) 0 V for 120 s (2) Edes ) -0.8 V for 15 s (3) equilibration 15 s (4) SW scan 0 f 0.65 V (1) Edep ) -0.2 V for 120 s (2) Edes ) -0.8 V for 15 s (3) equilibration 15 s (4) SW scan 0 f 0.5 V
procedure to fabricate 10 5-µm microwire electrodes typically took Br- > Cl- > SO42-).37 Au and Hg are classified as B-type metal cations, with a low electronegativity and a high polarizability, and therefore, their affinity toward anions increases in the same order (I- > Br- > Cl- > F-).43 The peak potential for mercury in UV-digested seawater was found to be ∼50 mV more negative than in 0.57 M (NaCl), but addition of 0.86 mM bromide (the same concentration as in seawater) to the chloride solution eliminated this difference, confirming that chloride and bromide are the two main inorganic ligands for mercury in seawater. In seawater, the inorganic mercury speciation is thought to be dominated by the following species:44 HgCl42- (65.8%), HgCl3Br2- (12.3%), HgCl3- (12.0%), HgCl2Br- (4.3%), HgCl2 (3.0%), and HgClBr (1.1%). In addition, the peak of Hg was narrowed (from 70 to 52 mV) and more symmetric after the addition of 0.85 mM bromide in 0.6 NaCl solution. The presence of bromide in seawater therefore facilitates the detection of mercury. Finally, the shape of the residual current in the mixed bromide/chloride solution was similar to that in the UV-digested seawater. Further additions of 200 nM I-, 0.4 mM B(OH)4, 28 mM SO42-, and 2.4 mM HCO3- did not influence the mercury peak or the residual current. Bromide and chloride are therefore the two main anions involved both in the complexation of mercury and in the mercury adsorption on gold. Influence of the Length of the Electrode. The effect of varying the length l of the electrode was evaluated with a 25-µm microwire electrode in 10-2 M HCl containing 10 nM Hg(II). The same electrode was shortened progressively with scissors, the length measured with a calliper and 10 mercury measurements were carried out at each length (24.1, 17.3, 9.8, 5.7, and 2.0 mm). As expected, the sensitivity was found to be directly proportional with the electrode length (S ) 4.49 nA‚mm-1; r2 ) 0.9994), so a larger mercury peak is obtained with a larger electrode. However, the electrical noise and the relative standard deviation were found (43) Stumm, W.; Morgan, J. J. Aquatic Chemistry: Chemical equilibria and rates in natural waters, 3rd ed.; Wiley-Interscience: New York, 1995. (44) Lindquist, O.; Rhode, H. Tellus 1985, 37B, 136-159.
to decrease with l (from 6% down to 2.6%). The higher signal/ noise ratio favors the use of short electrodes even though the peak half-width increased from 48 to 58 mV when l decreased from 24 to 2 mm. Long-Term Stability of the Electrode. Stability of the electrode is important when it is to be used for long-term in situ monitoring in natural waters. Continuous measurements over periods of several days showed the gradual appearance of a broad peak at ∼0.45 V, which became larger with further use of the electrode. Although this peak was present in both the analytical and the background scan and was therefore mostly removed by the background correction, it had a detrimental effect on measurements close to the detection limit by affecting the slope of the baseline. Once an electrode produced this residual current, it produced it in dilute HCl and in seawater. After two weeks of intensive use with one electrode, the detection limits were found to have doubled due to higher noise and the residual current, although the sensitivity remained almost the same. The appearance of this peak was unpredictable, although always slow, and varied between electrodes. Apparently the peak’s appearance is related to the history of the electrode. This peak has been observed before11,17,45 and was attributed to the possible formation of insoluble Hg2Cl2 at the surface of the electrode. Usually a new electrode was used when the peak started to affect the measurements. When not changing the electrode, the peak could be removed by three different procedures: (1) by strong anodic polarization of the gold; (2) by ultrasonication in ethanol; and (3) by decreasing the desorption potential or increasing the desorption time (for the analytical and the background scan). The first method was avoided to minimize changes in the roughness of the electrode. Ultrasonication was effective but not convenient as the peak reappeared quickly and as it cannot be done in situ. A decrease of the desorption potential from -0.8 to -1 V was found to strongly decrease the peak in 0.01 M HCl but had no effect in pH 2 seawater even when decreased further to -1.2 V. Limit of Detection of Mercury and Copper. The detection limits for mercury and copper were determined in dilute HCl and (45) Sahlin, E.; Jagner, D. Anal. Chim. Acta 1996, 333, 233-240.
Analytical Chemistry, Vol. 78, No. 14, July 15, 2006
5059
Figure 8. Measurement of mercury in Liverpool Bay water. (A) Voltammetric scans (background corrected) for mercury and copper. The copper peak goes off-scale. The initial mercury concentration was 70 pM Hg; subsequent scans are for additions of 65, 50, and 50 pM Hg. (B) Peak intensity as a function of the mercury concentration and the 95% confidence intervals. Microwire: 5-µm diameter/1.3 mm. Scan conditions: deposition at -0.2 V for 270 s, desorption at -0.8 for 45 s. Background scan: desorption at -0.8 V for 15 s. Other conditions as in Table 1.
UV-digested seawater using a new electrode. A short, 5-µm gold wire electrode was used of ∼0.3-mm length, the deposition potential was -0.2 (seawater) or 0 V (0.01 M HCl), the desorption potential was -0.8 V for 20% of the deposition time (1-min desorption at 5-min deposition), and background correction was used. Although the sensitivity in seawater was 40% smaller than in dilute HCl, the detection limit was similar because of smoother voltammograms in the seawater, possibly because of the much greater ionic strength. The detection limits were calculated from 3× the standard deviation of the derivative peak height of at least six consecutive measurements. The detection limits for mercury and copper were 12 and 50 pM, respectively, after a 2-min deposition time (20-s desorption) in dilute HCl containing 128 pM Hg and 580 pM Cu, and 16 pM for Hg in seawater containing 122 pM Hg. Using a longer deposition time of 5 min, the DLs were 6 pM Hg and 25 pM Cu in dilute HCl (containing 60 pM Hg and 137 pM Cu) and 9 pM Hg in seawater (containing 70 pM Hg). The peak height of mercury and copper was found to increase linearly with the deposition time up to at least 30 min (0 < tdep < 30 min, r2 > 0.994) in dilute HCl (for 300 pM Hg and 440 pM Cu), so in principle, these detection limits can be lowered further by using a longer deposition time. Measurements in Coastal Seawater. Water samples were collected from Liverpool Bay (Irish Sea) using a modified Niskin bottle and cartridge-filtered (0.22 µm) on-board ship. The samples were then stored in 1-L polyethylene bottles. Analysis was by the ASV method after UV digestion and acidification to pH 2 to liberate mercury from organic species. Background-corrected scans for the response in the seawater and after standard mercury additions are shown in Figure 8. The initial mercury concentration in this sample was 70 ( 15 pM (95% confidence interval). Although this concentration is high compared to uncontaminated seawater, it is in line with expectation for these waters, which are known to be contaminated with mercury; it is actually nearly 50% less than the lowest concentration reported (125 pM) for this region 25 years ago.46 The large peak at ∼ -0.25 V is due to copper, and its concentration was determined to be 8 nM, i.e., 100 times higher
than the mercury levels. If the copper were found to interfere with the mercury detection due to a much higher concentration, it would be easy to use a more positive deposition potential to lower the apparent copper peak. Although SWASV scans are presented in this work, similar results were obtained using CSA in terms of detection limits and stability of measurements with the difference that the copper and mercury peaks were better separated in CSA.
(46) Gardner, D. Nature 1978, 272, 49-51.
AC060231+
5060
Analytical Chemistry, Vol. 78, No. 14, July 15, 2006
CONCLUSIONS Instability of the mercury peak on a gold electrode is demonstrated to be due to slow adsorption of halide anions, which causes a gradual decrease of the peak height and peak potential over periods of several hours. A constant peak signal is obtained by applying a desorption step as part of the measurement prior to each scan. In addition, the stripping scans are more reproducible and therefore suitable for using background subtraction. Together with the low roughness value of the gold wire and the enhanced spherical diffusion at 5-µm diameter wire, very low limits of detection are achieved. The method is sufficiently sensitive to detect copper in any environment and mercury at levels of >9 or 16 pM in seawater using 5- or 2-min deposition time, respectively. Mercury levels (
