Anal. Chem.

2000, 72, 3218-3222

Bismuth-Coated Carbon Electrodes for Anodic Stripping Voltammetry

Joseph Wang,* Jianmin Lu, Samo B. Hocevar, and Percio A. M. Farias

Department of Chemistry and Biochemistry, New Mexico State University, Las Cruces, New Mexico, 88003
Bozidar Ogorevc

Laboratory for Analytical Chemistry, National Institute of Chemistry, Hajdrihova 19, SI-1000, Ljubljana, Slovenia

Bismuth-coated carbon electrodes display an attractive stripping voltammetric performance which compares favorably with that of common mercury-film electrodes. These bismuth-film electrodes are prepared by adding 400 g/L (ppb) bismuth(III) directly to the sample solution and simultanously depositing the bismuth and target metals on the glassy-carbon or carbon-fiber substrate. Stripping voltammetric measurements of microgram per liter levels of cadmium, lead, thallium, and zinc in nondeaerated solutions yielded well-defined peaks, along with a low background, following short deposition periods. Detection limit of 1.1 and 0.3 ppb lead are obtained following 2- and 10-min deposition, respectively. Changes in the peak potentials (compared to those observed at mercury electrodes) offer new selectivity dimensions. Scanning electron microscopy sheds useful insights into the different morphologies of the bismuth deposits on the carbon substrates. The in situ bismuthplated electrodes exhibit a wide accessible potential window (-1.2 to -0.2 V) that permits quantitation of most metals measured at mercury electrodes (except of copper, antimony, and bismuth itself). Numerous key experimental variables have been characterized and optimized. High reproducibility was indicated from the relative standard deviations (2.4 and 4.4%) for 22 repetitive measurements of 80 g/L cadmium and lead, respectively. Such an attractive use of mercury-free, environmetally friendly electrodes (with a performance equivalent to that of mercury ones) offers great promise to centralized and decentralized testing of trace metals.
Electrochemical stripping analysis has always been recognized as a powerful tool for measuring trace metals.1-3 Its remarkable sensitivity is attributed to the combination of an effective preconcentration step with advanced measurement procedures that generates an extremely favorable signal-to-background ratio. Four to six metals can be measured simultaneously in various matrixes
On leave from National Institute of Chemistry, Ljubljana, Slovenia. On leave from Pontificia Universidad Catolica do Rio, Rio de Janeiro, Brazil. (1) Wang, J. Stripping Analysis; VCH Publishers: Deerfield Beach, 1985. (2) Achterberg, E. P.; Braungardt, C. Anal. Chim. Acta 1999, 400, 381. (3) Batley, G. E. Mar. Chem. 1983, 12, 107.

at concentration levels down to 10-10 M, utilizing relatively inexpensive and portable instrumentation. Such characteristics have prompted the adaptation of stripping voltammetry for numerous centralized and decentralized metal testing applications.4,5 Two basic electrode systems, the mercury-film electrode (MFE) and hanging mercury drop electrode (HMDE), have gained wide acceptance in the development of stripping voltammetry.1,6 In most cases, a glassy-carbon electrode or a carbonfiber one are used to support the mercury film. Although preplated mercury films were used initially, the best results are achieved with the in situ plated mercury film introduced by Florence.6 Such small-volume mercury electrodes offer an attractive stripping performance.1,6-8 However, because of the toxicity of mercury, future regulations and occupational health considerations may severely restrict (or even ban) the use of mercury as an electrode material. New alternative electrode materialsswith a similar performancesare highly desired, particularly for meeting the growing demands for on-site environmental monitoring of trace metals and decentralized clinical testing of toxic metals. Different bare carbon, gold, or iridium electrodes have been used as possible alternative to mercury.1,2,9,10 While offering useful signals for several metals, the overall performance of these non-mercury electrodes has not approached that of mercury ones, due to a low cathodic potential limit, multiple peaks, or large background contributions. Preconcentrating modified electrodes, based on surface-confined ligands or ion exchangers, have also been proposed,11 but their overall sensitivity and reproducibility has not been satisfactory for routine measurements of trace metals. Despite these intensive efforts, an alternative stripping electrode, with a performance approaching that of mercury, has not emerged. This article describes the attractive stripping voltammetric behavior of bismuth-coated carbon electrodes. Bismuth is an environmentally friendly element, with very low toxicity and a
(4) Tercier, M.; Buffle, J. Electroanalysis 1993, 5, 187. (5) Wang, J. Analyst 1994, 119, 763. (6) Florence, J. Electroanal. Chem. 1970, 27, 273. (7) Golas, J.; Osteryoung, J. Anal. Chim. Acta 1986, 181, 211. (8) De Vitre, R. R.; Tercier, M. L.; Tsacopoulos, M.; Buffle, J. Anal. Chim. Acta 1991, 249, 419. (9) Wang, J.; Tian, B. Anal. Chem. 1993, 65, 1529. (10) Nolan, M. A.; Kounaves, S. P. Anal. Chem. 1999, 71, 3567. (11) Arrigan, D. Analyst 1994, 119, 1953. 10.1021/ac000108x CCC: $19.00 2000 American Chemical Society Published on Web 06/08/2000

3218 Analytical Chemistry, Vol. 72, No. 14, July 15, 2000

widespread pharmaceutical use.12,13 The new bismuth electrodes are prepared by simultaneous deposition of bismuth and the target heavy metals, in a manner analogous to in situ plated mercuryfilm electrodes.6 Experiments conducted under identical conditions indicate that the stripping voltammetric performance of the resulting bismuth-coated electrodes compares favorably with that of common mercury films. For example, such electrodes display well-defined, sharp and highly reproducible stripping peaks for low (ppb) concentrations of lead, cadmium, zinc, or thallium over a low background current. Such use of mercury-free electrodes is particularly attractive for the development of disposable metal sensors. We report these observations in the following sections. EXPERIMENTAL SECTION Apparatus. Stripping voltammetry was performed with a BAS CV-50W voltammetric analyzer (Bioanalytical Systems (BAS)), in connection with a personal computer. A coated glassy-carbon (GC) disk (3 mm diameter, BAS) or carbon-fiber microelectrode (CFME) served as the working electrode, with the Ag/AgCl (3 M NaCl) and platinum wire acting as the reference and counter electrodes, respectively. Scanning electron micrographs were obtained with a Hitachi model S-3200N microscope, using an accelerating voltage of 20 kV. Reagents. All solutions were prepared with double-distilled water. The bismuth, lead, cadmium, thallium, zinc, gold, and mercury standard stock solutions (1000 mg/L in 1 (Cd, Zn, Pb) or 5 (Bi) wt % nitric acid) were obtained from Aldrich and diluted as required. An acetate buffer solution (0.10 M, pH 4.5) served as the supporting electrolyte. Preparation of CFME. The preparation of the CFME was described earlier.14,15 Briefly, a cleaned single carbon fiber (7 m in diameter, 2-3 cm in length, Goodfellow Co., Oxford, U.K.) was attached to a copper wire using silver paint (SPI Supplies, West Chester, PA) and inserted into a glass capillary tube (7 cm in length, Euroglass, Ljubliana, Slovenia). Then, employing a fine pulling technique,11 the carbon fiber was directly sealed by pulling the glass capillary tube using a microelectrode puller (PP-830, Narishige, Tokyo, Japan). Prior to the pulling, the copper wire was fixed at the stem end of the capillary tube by casting a drop of nonconducting epoxy resin. Finally, the exposed carbon fiber was cut using a microsurgical scalpel blade to a length of 1 mm. The resulting CFMEs were inspected electrochemically (in 1 mM ferricyanide solution) and optically (using an inverted microscope; Eclipse, Nikon, Tokyo, Japan). The selected CFMEs, were stored in sealed boxes, prior to their stripping application. Procedure. Stripping voltammetric measurements were performed with an in situ deposition of the bismuth film and target metals (e.g., Pb, Cd, and Zn) in the presence of dissolved oxygen. Prior to its use, the glassy-carbon electrode was polished with a 0.05-m alumina slurry. The three electrodes were immersed into a 15-mL electrochemical cell, containing 0.1 M acetate buffer (pH 4.5) and 400 g/L bismuth. The deposition potential (usually -1.2 V; -1.4 V for measurements of Zn) was applied to the carbon
(12) Rodilla, V.; Miles, A. T.; Jenner, W.; Hawksworth, G. Chem.-Biol. Interact. 1998, 115, 71. (13) Sun, H. Z.; Li, H. Y.; Sadler, P. J. J. Biol. Chem. 1999, 274, 29094. (14) Zhang, X.; Ogorevc, B.; Rupnik, M.; Kreft, M.; Zorec, R. Anal. Chim. Acta 1999, 378, 135. (15) Zhang, X.; Zhang, W.; Zhou, X.; Ogorevc, B. Anal. Chem. 1996, 68, 3338.

Figure 1. Stripping voltammograms of lead, cadmium, and zinc at glassy-carbon (A) and carbon-fiber (B) electrodes coated with bismuth (a) and mercury (b) films. Solutions, 0.1 M acetate buffer (pH 4.5) containing 50 g/L lead(II), cadmium(II), and zinc(II), along with 400 g/L bismuth (a) or 10 mg/L mercury (b). Deposition for 120 s at -1.4 V; cleaning for 30 s at +0.3 V. Square-wave voltammetric stripping scan with a frequency of 20 Hz, potential step of 5 mV, and amplitude of 25 mV.

working electrode, while the solution was stirred. Following the preconcentration step (usually 120 s), the stirring was stopped, and after 10 s, the voltammogram was recorded by applying a positive-going square-wave voltammetric potential scan (with a frequency of 20 Hz, amplitude of 25 mV, and potential step of 5 mV). The scan was terminated at 0.0 V. Aliquots of the target metal standard solution were introduced after recording the background voltammograms. A 30-s conditioning step at +0.3 V (with solution stirring) was used to remove the target metals and bismuth, prior to the next cycle. In situ plated mercury-coated carbon electrodes were employed for comparison, with measurement procedures similar to those employed with the bismuthcoated electrodes (with the exception of using 10 mg/L mercury). All experiments were carried out at room temperature. RESULTS AND DISCUSSION Stripping and Background Signals at Bismuth-Coated Electrodes. Figure 1 illustrates typical anodic stripping voltammograms for some common ions, present at the 50 ppb (g/L) level, obtained at the bismuth (a)- and mercury (b)-coated glassycarbon (A) and carbon-fiber (B) electrodes. Both electrodes display well-defined, sharp (nearly symmetrical) and separated peaks following the 2-min deposition. While the zinc and lead peak potentials are nearly identical for the bismuth and mercury electrodes, the cadmium peak (of the bismuth electrode) appears at a more negative potential (-0.79 vs -0.70 V at the mercury surface). As will be illustrated below, such a potential shift adds new dimensions of selectivity. The peak sharpness is also not compromised upon using the bismuth electrode (e.g., peak half widths of 26 and 35 mV for Cd and Zn at the bismuth-coated glassy-carbon electrode, vs 30 and 34 mV, respectively, at the corresponding mercury film). Note also that the sensitivity of the bismuth-plated carbon-fiber electrode toward lead and cadmium compares favorably with that of the mercury-coated one; yet, the bismuth-based microelectrode exhibits a lower zinc response. The square-wave voltammetric scan results also in a low (nearly flat)
Analytical Chemistry, Vol. 72, No. 14, July 15, 2000
3219

Figure 2. Stripping voltammograms at glassy-carbon (A) and carbon-fiber (B) electrodes: (a) 0.1 M acetate buffer (pH 4.5); (b) as in (a) but after the addition of 50 g/L Cd(II), Pb(II), and Zn(II); (c) as in (b) but after the addition of 400 g/L Bi(III). Deposition and cleaning, as in Figure 1.

background current in the presence of dissolved oxygen. With the exception of a slightly higher hydrogen evolution background contribution (at -1.2V), the bismuth electrode displays a similar background current. Overall, the data of Figure 1 indicate that the signal-to-background characteristics are not compromised by the use of the bismuth coating (instead of the mercury film). Analogous stripping potentiometric experiments yielded similar results (not shown). In general, the bismuth-coated glassy-carbon and carbon-fiber electrodes display relatively similar signal-tobackground characteristics, although zinc yields the highest response at the GC disk, compared to cadmium at the fiber substrate. The attractive stripping performance of the new bismuth-film electrode is indicated also from the stripping voltammograms of Figure 2, which also displays the corresponding control (bare electrode) experiments. No stripping signals are observed at the bare glassy-carbon (A(b)) and carbon-fiber (B(b)) electrodes for a sample containing 50 g/L lead and cadmium. In contrast, adding 400 g/L bismuth to the sample, and simultaneously depositing it along with the target metals, resulted in the appearance of sharp and undistorted stripping peaks for both analytes, as well as bismuth (c). Note that both in situ bismuthplated electrodes display such well-defined peaks, over a low background, despite the use of a short (2-min) deposition time and a nondeaerated solution. Clearly, the co-deposition of bismuth is responsible for the attractive stripping behavior observed in the Figures 1 and 2. Scanning electron microscopy (SEM) can shed useful insights into the growth patterns of the bismuth film. Figure 3 shows scanning electron micrographs (obtained with 1000 and 5000 magnifications) of typical regions of carbon-fiber (A) and glassycarbon (B) electrodes, respectively, before (a) and after (b) the bismuth deposition. Different surface morphologies are observed at these carbon substrates. The SEM image of the bismuth-coated glassy-carbon electrode indicates a highly porous, three-dimensional fibril-like network. A thick, quite uniform, nonporous bismuth deposit is observed on the carbon-fiber substrate. Note the substantial increase of the fiber diameter (from 7 to 35 m) that indicates a significant amount of deposited metal. Apparently, the carbon substrate has a profound effect upon the nucleation
3220

and growth of the bismuth deposit and, hence, upon its structural features. Mercury films on glassy-carbon substrates commonly consist of finely divided mercury droplets (whose size and distribution depends on the deposition potential).16 The potential window of the new bismuth-coated electrode has a profound effect upon the scope of its stripping operation. Figure 4 compares the accessible potential window of several working electrode materials relevant to stripping voltammetric work (using an acetate buffer solution). As expected, the bare carbon (a) and gold (b) electrodes display a wide anodic potential window (>0.80 V) and a limited cathodic range (up to -1.0 and -0.70 V, respectively). The mercury-coated carbon electrode (d) exhibits a high hydrogen overvoltage (in the vicinity of -1.30 V) and a limited anodic range due to the oxidation of mercury (that starts around +0.30 V). The bismuth-coated electrode (c) is also characterized with a limited anodic region (due to the bismuth oxidation) to the vicinity of -0.2 V and an extended cathodic potential range (at -1.20 V). With such potential window of 1.0 V, five to six stripping peaks might be observed simultaneously. The bismuth-coated electrode is readily applicable for measurements of electrolytically deposited elements with standard potentials more negative than bismuth. While this would preclude the quantitation of copper, antimony, and bismuth itself (among the metals commonly measured at mercury electrodes), preliminary results indicate that it is possible to detect trace copper (on the anodic side of the bismuth peak). Note also, from Figure 2, the low background current observed within the potential window of the bismuth-coated electrode (despite the use of a nondeaerated solution). We found that the bismuth-coated electrodes are less susceptible to oxygen interference (i.e., background reactions). This was indicated not only from square-wave stripping voltammetric experiments (commonly used to minimize the oxygen interference) but also from linear-scan stripping studies. Conducting such experiments in nondeaerated media yielded a flat baseline using the bismuth electrodes, and a large background slope at the mercury ones (not shown). We are examining in detail the reactivity of oxygen at bismuth-plated electrodes. Effect of Experimental Variables. Figure 5 examines the influence of the bismuth ion concentration upon the resulting stripping response. Both the lead (a) and cadmium (b) peaks increase rapidly upon raising the bismuth concentration from 50 to 100 g/L and level off above 200 g/L. The different trends observed for cadmium and lead below 50 g/L bismuth are attributed to their different standard potentials. Apparently, a low bismuth coverage is sufficient for depositing the more easily reduced lead ion. A bismuth concentration of 400 g/L was used for most subsequent analytical work. Mercury levels higher than 400 g/L (2 10-6 M) are required for the successful use of in situ plating of mercury film electrodes.6 The response of the bismuth-film electrode to 120 g/L lead increased linearly with the preconcentration time at first (up to 4 min), then more slowly, and started to level off above 5 min (not shown; deposition at -1.2 V; other conditions, as in Figure 1B(a)). Very short accumulation periods (of 30-60 s) are thus sufficient for attaining very favorable signal-to-background characteristics for microgram per liter concentrations of heavy
(16) Stulikova, M. J. Electroanal. Chem. 1973, 48, 33.

Analytical Chemistry, Vol. 72, No. 14, July 15, 2000

Figure 3. SEM images of carbon-fiber (A) and glassy-carbon (B) electrodes before (a) and after (b) the bismuth deposition. Deposition, 10 min at -1.2 V, from an acetate buffer solution (pH 4.5) containing 50 mg/L bismuth. Accelerating voltage, 20 kV.

Figure 5. Effect of the bismuth concentration upon the stripping voltammetric response of 100 g/L lead(II) (a) and cadmium(II) (b). Deposition for 120 s at -1.2 V. Other conditions, as Figure 1B(a). Figure 4. Accessible potential windows of carbon-fiber electrodes coated with different films: (b) gold, (c), bismuth, and (d) mercury. Also shown (a) is the window of the bare carbon-fiber electrode. Film deposition for 120 s at -1.0 V (b) and at -1.4 V (c, d) in 0.1 M acetate buffer (pH 4.5). Other conditions, as Figure 1B(a).

metals. While the leveling off observed at longer periods is attributed to surface saturation, no multiple or broader peaks are observed under these conditions. Analytical Performance. The different stripping potentials observed at the bismuth-coated electrode pave the way to new dimensions of selectivity compared to mercury electrodes. For example, the quantitation of thallium in the presence of cadmium and lead represents a common problem in stripping voltammetry due to overlapping stripping signals.1 Such a resolution problem is indicated from Figure 6, where the thallium peak is obscured

by an overlapping cadmium peak (b). In contrast, the bismuthfilm electrode (a) results in separated peaks and offers convenient quantitation of the three metal ions. A well-defined concentration dependence serves as the basis for the analytical utility of bismuth-based stripping electrodes. Figure 7 displays stripping voltammograms obtained upon increasing the lead concentration in 20 g/L steps. Well-defined sharp peaks, over a flat baseline, are observed following the 2-min deposition. The 5 measurements shown are part of a series of 11 concentration increments up to 150 g/L lead. The resulting calibration plot (also shown) is linear over the entire range (slope, 0.0023 AL/g; intercept, -0.009 A; correlation coefficient, 0.995). Highly linear calibration plots were obtained in another calibration experiment involving simultaneous measurements of cadmium and lead over the 40-200 g/L range (not shown). The
Analytical Chemistry, Vol. 72, No. 14, July 15, 2000
3221

solution containing 80 g/L lead and cadmium resulted in highly reproducible stripping peaks, with relative standard deviations of 4.4 and 2.4%, respectively (not shown; 2-min deposition at -1.2 V). Such good precision is attributed to the reproducible film renewal accrued from the in situ bismuth plating. Compared to the stabilization period required for in situ plating of mercury film electrodes,6 bismuth-coated electrodes display an highly stable response starting with the first run. CONCLUSIONS We have demonstrated that the stripping voltammetric performance of bismuth-coated carbon electrodes compares favorably with that of common mercury-based electrodes. Among the proposed alternative (non-mercury) electrodes, the bismuth one offers the closest behavior to mercury. The large cathodic working range of the bismuth-film electrodes makes it suitable for detecting metals more electronegative than bismuth. Such use of mercuryfree stripping electrodes, with similar response characteristics, should facilitate the decentralized testing of heavy metals and the development of single-use metal sensors and should address possible restrictions on the use of mercury electrode materials, in general. Work is in progress for further assessing the scope and power of bismuth-based electrodes and for understanding the role of the bismuth deposition on the stripping behavior. We are thus exploring the use of preplated bismuth films for adsorptive stripping measurements of additional metals (that cannot be deposited) and for remote sensing of heavy metals in harsh environments. Preliminary results with the nickel-dimethylglyoxime system are very encouraging. We are also evaluating the use of the bismuth peak as a built-in internal standard and the ability to detect metals with standard potentials more positive than bismuth (primarily copper). Preliminary results in these directions are very encouraging. ACKNOWLEDGMENT This project was supported by grants from the U.S. DOE (Grant FG07-96ER62306) and DOE-WERC Program, and the Ministry of Science and Technology of Republic of Slovenia (Contract 3411-99-71-0022). Received for review February 3, 2000. Accepted May 1, 2000.
AC000108X

Figure 6. Stripping voltammograms of a mixture of 50 g/L lead(II), cadmium(II), and thallium(I) at bismuth (a) and mercury (b) thin-film electrodes. Solutions, 0.1 M acetate buffer (pH 4.5) containing 400 g/L bismuth (a) and 10 mg/L mercury (b). Deposition for 120 s at -1.2 V. Other conditions, as Figure 1B.

Figure 7. Stripping voltammograms for increasing levels of lead(II) in 20 g/L steps (b-f), along with the background response (a). Also shown (on the right) is the resulting calibration plot. Deposition for 120 s at -1.2 V. Other conditions, as Figure 1B(a).

signal-to-noise (S/N) characteristics of the 10 g/L data point were used for estimating the detection limit (1.1 g/L lead; S/N ) 3; not shown). As expected, a longer deposition period of 10 min further enhanced the S/N characteristics and offered a detection limit of 0.3 g/L (1.4 10-9 M) lead (based on measurements of a 5 g/L lead solution; not shown). Good precision is another attractive feature of bismuth-coated stripping electrodes. A series of 22 repetitive measurements of a

3222

Analytical Chemistry, Vol. 72, No. 14, July 15, 2000