Environ. Sci. Technol. 2004, 38, 3674-3682
Electro-oxidation and Amperometric Detection of Chlorinated Phenols at Boron-Doped Diamond Electrodes: A Comparison of Microcrystalline and Nanocrystalline Thin Films GRACE W. MUNA, NATASHA TASHEVA, AND GREG M. SWAIN* Department of Chemistry, Michigan State University, East Lansing, Michigan 48824-1322
We report on the electro-oxidation and amperometric detection of phenol and chlorinated phenols, the latter coupled with flow injection analysis (FIA) and high performance liquid chromatography (HPLC), using borondoped microcrystalline and nanocrystalline diamond thinfilm electrodes. The low background current, good response without extensive pretreatment, and low susceptibility to fouling are properties that make diamond an attractive new electrode for monitoring this class of pollutants. Cyclic voltammetric studies were performed to evaluate the redox response of phenol, 2-chlorophenol, 3-chlorophenol, 4-chlorophenol, and pentachlorophenol (PCP) in phosphate buffer, pH 3.5, as a function of the potential scan rate and cycle number. The diamond electrode performance for the amperometric detection of these contaminants in FIA-EC and HPLC-EC was evaluated in terms of the linear dynamic range, limit of quantitation, sensitivity, response precision, and response stability. Both diamond types yielded low mass limits of quantitation of 1001000 pg for all the phenolic compounds in FIA-EC, except PCP which was 3 ng, and 100-600 pg for all the compounds in HPLC-EC. In all cases, the S/N was 3 or greater. Both electrode types also exhibited good sensitivity, excellent response reproducibility (av 2.7% for FIA-EC and av 4.2% for HPLC-EC), and superb response stability for all the analytes. The electrodes could be used from days to weeks in the measurement with only a periodic soak in distilled 2-propanol required to maintain optimum performance. Both types of diamond outperformed glassy carbon, which exhibited short-lived responsiveness as a consequence of fouling by reaction products and potentialdependent changes in the electrode’s physiochemical properties. The use of the HPLC-EC assay for the determination of 2-chlorophenol in a contaminated soil sample is also demonstrated.
Introduction Chlorinated phenols (CPs) are environmental pollutants that need to be monitored due to their adverse effects on human health. They can enter the environment as byproducts of * Corresponding author phone: (517)355-9715, ext 229; fax: (517)353-1793; e-mail:
[email protected]. 3674
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 13, 2004
industrial processes, such as the production of antioxidants, dyes, and drugs (1); the chlorination of drinking water; and the chlorinated bleaching of paper (2). Several chlorinated phenols, such as 2-chlorophenol, 2,4-dichlorophenol, 2,4,6trichlorophenol, and pentachlorophenol, have been classified as priority pollutants by the Environmental Protection Agency (3). A maximum concentration level of 40 µg/L (40 ppb) has been established for 2-chlorophenol in drinking water (socalled “lifetime exposure”). This is defined as the maximum concentration not expected to produce adverse health effects during a lifetime of exposure (4). Unfortunately, large-scale use of chlorinated phenols over the years has led to the contamination of terrestrial and aquatic ecosystems. The past practice of chemical waste disposal in ordinary landfills has led to these areas becoming sources of CP contamination in groundwater (i.e., drinking water). The EPA-recommended protocols (Methods 604, 625, and 1625) for phenol and chlorophenol speciation and detection involve liquid-liquid extraction followed by either gas chromatography with electron capture or mass spectrometric (MS) detection of derivatized analytes. Although routinely employed, these protocols have the disadvantages of being relatively expensive, involving significant sample preparation, and requiring somewhat lengthy analysis times. Separation science coupled with electrochemical detection is a practical approach for the speciation and detection of some environmental contaminants in water supplies. The instrumentation required is relatively inexpensive, easy to use, and field deployable. The limiting factor in implementing electrochemical technologies for water quality monitoring is often the electrode materialsthe stability of its structure and properties and its electrochemical responsiveness. This is particularly true for the detection of phenol and chlorinated phenols. Phenols are electro-oxidizable at modest potentials in aqueous media, making electrochemistry a viable detection method. Unfortunately, electrochemical detection of this class of pollutants is often plagued by the strong, irreversible adsorption of oxidation reaction intermediates and products (i.e., fouling). Fouling by oligomeric and or polymeric (e.g., polyphenol) reaction products leads to poor electrode response reproducibility and stability and sometimes complete response attenuation (6-9). In such cases, the electrode must be pretreated (e.g., mechanical polishing) to reactivate the surface, and the assay recalibrated and revalidated. The first step in the electro-oxidation of phenol or chlorinated phenols is phenoxy radical formation; a species that can be further oxidized to form soluble products or can undergo radical-radical or radical-substrate coupling to form polymeric products (5-13). The polymeric products tend to have low solubility in aqueous media and adsorb on electrode surfaces, passivating them from further reaction. Electropolymerization reactions and fouling have been have been reported for many electrodes, including Fe, Cu, Ni, Cr, Ti, Zn, Au, Pt, and Ag (9-11, 13-16). The extent of passivation/ fouling depends on several of parameters: the phenolic compound concentration, surface properties of the electrode, solution pH, solvent additives, electrode potential, and current density (8, 9). Various approaches have been used to minimize the electrode fouling problem. One is to optimize the electro-oxidation conditions, as fouling can be minimized at low phenol concentrations, elevated temperatures, and large values of applied overpotential, which increases the rate of anodic discharge of H2O to generate OH• (6, 11). The OH• oxidatively degrades the adsorbed oligomeric and or polymeric reaction products, thereby removing them from 10.1021/es034656e CCC: $27.50
2004 American Chemical Society Published on Web 05/20/2004
the surface. Another approach is to chemically modify the electrode surface. For example, electrode surfaces have been modified with enzymes to enhance the oxidative reactivity. However, this approach has the drawbacks of complexity and maintenance of the enzyme activity over time (17). Boron-doped, hydrogen-terminated diamond electrodes possess a number of important and practical electrochemical properties and exhibit much versatility as an electrode (1827). Some of the important diamond electrode properties for water quality monitoring are (i) low and stable background current, (ii) wide working potential window, (iii) an active response for many redox analytes without extensive electrode pretreatment, (iv) stable surface microstructure and morphology over a wide potential range, and (v) weak adsorption of polar molecules on the hydrogen-terminated surface. This last property is particularly relevant to chlorophenol detection because both microcrystalline (26) and nanocrystalline (27) diamond exhibit less tendency to become fouled by adsorbed reaction products than do other common electrodes. This is due, at least in part, to the fact that the diamond surface is relatively nonpolar when hydrogen-terminated and contains no extended π-electron system, which could promote π-π interactions with an adsorbate. Boron-doped microcrystalline diamond thin-film electrodes are attractive new materials for water quality monitoring and have only recently been applied for both the oxidative detection and remediation of chlorinated phenols (28-34). Fujishima and co-workers reported on the electrochemical oxidation and amperometric detection of chlorinated phenols at anodically pretreated (i.e., oxidized) diamond (28). The electrodes exhibited good response stability and reproducibility with low detection limits for several chlorinated phenols in the amperometric detection mode coupled with FIA or HPLC. They reported some fouling of diamond at high phenolic compound concentrations (5 mM) and that the electrodes could be effectively reactivated by anodic treatment at 2.6 V vs SCE for 4 min. The authors postulated that the treatment generates OH•, which oxidatively degrades the passivating polymer layer. Anodically pretreated diamond was superior to the hydrogen-terminated material in terms of resistance to fouling and its ability to be reactivated. In another report, Comninellis and co-workers investigated the electrochemical oxidation of 4-chlorophenol at boron-doped diamond thin-film electrodes (29). Under their conditions, the formation of phenoxy radicals, 1,4benzoquinone, and polymeric products were observed. O’Grady and co-workers demonstrated that phenol can be oxidatively remediated at boron-doped diamond electrodes stably over time without being fouled (30). Several very recent papers have addressed the electro-oxidation reaction mechanism of phenol and chlorinated phenols in acidic media at diamond electrodes (31-34). Of particular note is paper by Zhi et al., which contains evidence for a direct electrochemical oxidation pathway for phenol at diamond anodes (34). Presently, we report on a comparison of the electrochemical performance of boron-doped, hydrogen-terminated microcrystalline and nanocrystalline diamond thin-film electrodes for the electro-oxidation and amperometric detection of phenol and chlorinated phenols in phosphate buffer, pH 3.5. Cyclic voltammetry and amperometric detection coupled with both flow injection analysis (FIA-EC) and reversed-phase liquid chromatography (HPLC-EC) were used to study the electro-oxidation reaction and to determine the analytical detection figures of merit, respectively. The HPLCEC assay was also used to quantify the 2-chlorophenol concentration in a contaminated soil sample. The results demonstrate (i) the usefulness of diamond electrodes for the sensitive, reproducible, and stable detection of these important pollutants, consistent with the published literature and (ii) that the phenol and chlorophenol oxidation reaction
proceeds in a similar fashion at both diamond types, regardless of their morphology and microstructure.
Experimental Section Diamond Growth Conditions. The microcrystalline diamond thin films were deposited on boron-doped p-Si (100) (e10-3 Ω-cm, Virginia Semiconductor Inc., Fredricksburg, VA) substrates using a commercial microwave-assisted chemical vapor deposition (CVD) system. The substrates (0.1 cm thick × 2 cm2 in area) were first rinsed with ultrapure water, methanol, and acetone. They were then polished with 0.1 µm diameter diamond powder (GE Superabrasives, Worthington, OH) for 5 min on a felt pad. The substrates were then ultrasonicated for 1 min in clean acetone to remove polishing debris. The clean scratches and residual powder particles serve as the initial nucleation sites for diamond growth. The microcrystalline films were deposited from a 0.5% methane/hydrogen (C/H) source gas mixture, a total gas flow of 200 sccm, 1 kW of microwave power, a system pressure of 45 Torr, an estimated substrate temperature of 825 °C, and a growth time ca. 10 h. Ultrahigh purity (99.999%) methane and hydrogen were used as the source gases. The boron doping was performed using either a solid-state or gas-phase boron source. Solid-state doping was accomplished using a BoronPlus (GS 126, Techneglas, Inc. Perrysburg, OH) ceramic disk diffusion source that was placed under the substrates. B2O3 diffuses from the disk and serves as the source for boron dopant atoms. B2H6 diluted in H2 was also employed as gas-phase boron source. A B2H6 concentration of 10 ppm was used during the growth. After deposition, the methane flow was stopped, and the films were slowly cooled in the presence of atomic hydrogen to an estimated temperature of less than 300 °C by slowly reducing the power and pressure over 4 min period. The film thickness was nominally 5 µm, and the boron dopant concentration was in the low to mid 1020 cm-3 range, as determined from boron nuclear reaction analysis measurements of other films deposited using similar conditions. The film resistivity was ∼0.01 Ω cm, or less, as measured with a tungsten four-point probe. The boron-doped nanocrystalline diamond thin films were deposited on p-type Si (100) substrates (e10-3 Ω-cm, Virginia Semiconductor Inc., Fredricksburg, VA), using the same microwave CVD system. The surface of the Si substrate was mechanically scratched with 0.1 µm diameter diamond powder (GE Superabrasives, Worthington, OH). The scratched substrate was then washed with ultrapure water, isopropyl alcohol (IPA), acetone, IPA, and ultrapure water. Ultrahigh purity, CH4, Ar, and H2 (99.999%) were used as the source gases. The gas flow rates were 1, 94, and 5 sccm, respectively, for CH4, Ar, and H2. The microwave power and system pressure were maintained at 800 W and 140 Torr, respectively. The substrate temperature was estimated by an optical pyrometer to be about 800 °C. The deposition time was 2 h giving approximately a 4 µm-thick film, as estimated from sample weight change. B2H6 was used as the source gas for doping at a concentration of 10 ppm. All gases flowed into the reactor simultaneously to initiate the diamond deposition. At the end of the deposition period, the CH4 flow was stopped and the Ar and H2 flows continued. This film was exposed to an H2/Ar plasma for approximately 10 min. The substrate was then cooled in the presence of atomic hydrogen to an estimated temperature of less than 300 °C by slowly reducing the power and pressure over a period of 4 min. Both microcrystalline and nanocrystalline diamond were characterized for their material properties, as previously reported (24, 27, 35, 36). The microcrystalline diamond films were well faceted, with a predominance of octahedral and cubo-octahedral VOL. 38, NO. 13, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3675
crystallites. The nominal crystallite size was 1-3 µm. The nanocrystalline diamond was much smoother with the films consisting of 50-100 nm aggregates of ∼15 nm diameter grains of diamond. The grains were randomly oriented, phase-pure diamond with well-defined grain boundaries consisting of π-bonded carbon atoms. The diamond electrodes were hydrogen-terminated, at least at the beginning of the measurements, with a nominal XPS oxygen-to-carbon atomic ratio (O/C) of e0.02. Undoubtedly, oxygen was incorporated onto the electrode surface during use given the positive potentials employed for oxidation and detection. These levels likely reached a maximum of 0.15 (O/C), based on other research in our laboratory with electrochemically oxidized diamond. Glassy carbon (2 cm2, GC-30, Tokai Ltd.) was prepared by polishing with successively smaller grades of alumina powder slurried in ultrapure water (1.0, 0.3, and 0.05 µm) on separate felt pads. After each polishing step, the electrode was rinsed with copious amounts of ultrapure water and ultrasonicated in this medium for 5 min prior to use. Cyclic Voltammetry. All electrochemical measurements were made with a CYSY-2000 computerized potentiostat (Cypress System Inc. Lawrence, KS) using a single compartment glass cell. The diamond or glassy carbon electrode was pressed against a viton O-ring and clamped to the bottom of the glass cell (24). Ohmic contact was made on the backside of the Si substrate with a Cu or Al plate, after scratching the substrate, cleaning the surface, and applying a bead of Ag paste. A graphite rod was used as the counter electrode and a commercial Ag/AgCl electrode (saturated KCl) served as the reference. The geometric area of the working electrode was ca. 0.2 cm2. All measurements were made at room temperature, and all solutions were deoxygenated with nitrogen for at least 5 min prior to any measurement and remained blanketed with the gas during the measurement. Flow Injection Analysis (FIA) and HPLC Systems. The FIA system has been described elsewhere (37-39). The HPLC system consisted of a consta-Metric III (Milton Roy) metering pump for regulating the carrier solution; a model 7125 (Rheodyne) syringe-loaded sample injector valve with a 20 µL loop; a model LP-21 (Scientific System Inc.) pulse dampener to reduce pump noise; and a homemade thinlayer cell. The entire system set up was electrically grounded, and the flow cell was housed in a Faraday cage to reduce the electrical noise. Electrical contact was made by pressing a piece of copper foil against the backside of the cleaned and conducting Si substrate, after it was scratched and cleaned and Ag paste applied. A 0.1-cm-thick neoprene rubber gasket separated the surface of the working electrode from the top piece of the cell. A rectangular groove was cut in the gasket that defined the flow channel. An Omni-90 potentiostat was used to control the applied potential. The gradient separation of phenol and chlorophenols was carried out using a C18 reversed-phase column (X-Terra, Waters) with a 5 µm particle size. Gradient elution was carried out by changing the 50 mM phosphate buffer (pH 3.5)/acetonitrile mobile phase from 65:35 to 20:80 (v/v). The flow rate was 0.6 mL/min. The carrier solution was continuously sparged with nitrogen (99%). Soil Sample Analysis. A 10 g soil sample containing at least 60 different organic pollutants was obtained from Absolute Standards, Inc. (Hamden, CT). The sample was prepared according to principles outlined in the “National Standard for Water Proficiency Testing Studies, Criteria Document” - USEPA 12/30/98. The sample came with a list of concentration ranges for each of the pollutants added. No information was provided about the mineralogy of the soil or how the sample was prepared. 2-Chlorophenol (2-CP) was one contaminant present and was assayed for. The sample was prepared for analysis in the following manner. 3676
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 13, 2004
First, all 10 g of the soil sample was added to 40 mL of acetonitrile and ultrasonically mixed for 30 min. This served to extract the 2-CP into the liquid phase. Second, the sonicated sample was then centrifuged for 30 min, and the liquid phase was carefully decanted into a 250 mL roundbottom flask. Third, the solvent was evaporated on a rotary flash evaporator down to a volume of 8-10 mL. The sample was again centrifuged, and the extract is designated as the crude sample. Fourth, solid-phase extraction was carried out on the crude sample using 3 mL SupelClean LC-18 cartridges (Supelco, 491 m2/g). The C-18 cartridges were conditioned with 6 mL of methanol followed by 6 mL of ultrapure water. After conditioning, 2 mL of the crude sample was mixed with 4 mL of ultrapure water in a clean vial. The sample was eluted through the solid phase (vacuum-assisted)sa process that took approximately 10 min. Fifth, the 2-CP was eluted off the solid phase with 6 mL of a solvent mixture containing dichloromethane, hexane, and acetonitrile (50:47:3, v/v)sa process that took approximately 10 min. Sixth, solvent from this solution was evaporated by passing N2 down to a final volume of ca. 1 mL. The concentrated sample was then adjusted to 4.0 mL with the addition of the mobile phase solution (65:35 50 mM phosphate buffer, pH 35./acetonitrile). This sample was injected onto the column for separation and detection. Chemicals. Reagent grade quality chemicals and ultrapure deionized water (18 MΩ, Barnstead E-pure) were used to prepare all solutions. High purity (98-99%) phenol, 2-chlorophenol (2-CP), 3-chlorophenol (3-CP), 4-chlorophenol (4CP), and pentachlorophenol (PCP) were obtained from Sigma and used without additional purification. Standard solutions were prepared in the carrier solution except for PCP, which was prepared in pure acetonitrile. HPLC-grade acetonitrile was used as organic modifiers (EM Science). Phosphoric acid (Aldrich) was used to adjust the pH of the buffer. All glassware was cleaned in a KOH-ethanol bath followed by rinsing with ultrapure water.
Results and Discussion Electro-oxidation of Phenol and Chlorophenols. Cyclic voltammetric i-E curves for phenol, 2-chlorophenol (2-CP), 3-chlorophenol (3-CP), 4-chlorophenol (4-CP), and pentachlorophenol (PCP), all in 50 mM phosphate buffer, pH 3.5, were recorded at microcrystalline and nanocrystalline diamond thin-film electrodes. The oxidation of phenol and chlorinated phenols can involve direct and or indirect (e.g., OH• mediated) pathways, which result in the formation of quinones, carboxylic acids, and polymeric reaction products (31-34). All measurements were made with analyte concentrations of 1 mM, or less, and a low number of potential cycles (5-10). The same electrode, without any conditioning between measurements, was used to measure the response for all five compounds. The electrode surfaces were hydrogenterminated to begin with but progressively became oxygenterminated during exposure to the positive potentials. Although fouling was generally not a problem with any of the diamond electrodes (either hydrogen- or presumably oxygen-terminated), as is discussed below, a combination of extended potential cycling (positive limit of 1.2 V vs Ag/ AgCl) and soaking in distilled 2-propanol was effective, when necessary, restoring the electro-oxidation response by removing adsorbed reaction products. This is somewhat different from the observation of Rodrigo et al., who reported that washing a fouled electrode with organic solvents was ineffective at reactivating the surface after oxidizing 4-CP (29). Our observation is also different from what was reported Terashima et al., as we found no need to reactivate the electrodes using severe anodic polarization, at least under our analysis conditions (28). Background cyclic voltammetric i-E curves were first recorded in the supporting electrolyte.
FIGURE 1. Cyclic voltammetric i-E curves for 0.1 mM 2-chlorophenol (2-CP) in 50 mM phosphate buffer, pH 3.5, at (A) microcrystalline and (B) nanocrystalline diamond thin-film electrodes. Scan rate ) 0.1 V/s. Geometric area ) 0.2 cm2. The background current for both diamond types was lower than that for glassy carbon, featureless up to the oxygen evolution onset potential of ca. 1.4 V and quite stable during extended potential cycling. Cyclic voltammetric i-E curves for phenol, 2-CP, and 4-CP were similar to each other at both diamond types. As an example, Figure 1 shows five consecutively recorded voltammograms (0.1 V/s) for 0.1 mM 2-CP at a (A) microcrystalline and (B) nanocrystalline diamond thin-film electrode. The first forward scan has a single, well-defined oxidation peak at ca. 1.1 V (peak a). The relatively stable peak charge after five scans ranged from 23 to 30 µC for all three analytes. This oxidation peak occurs near the onset potential for the water discharge reaction so that the oxidant, OH•, is also being simultaneously generated. This oxidant is involved in the indirect oxidation of phenol molecules, forming phenoxy radicals at the surface via an anodic oxygen transfer reaction (i.e., the addition of OH groups to the aromatic ring) (2834). Oxidation products, such as hydroquinone and catechol, are initially formed (28). However, at these anodic potentials, the hydroxylated products can undergo further oxidation via electron and proton loss to form quinone-like species. It is these quinone-like oxidation products that are reduced during the first negative-going sweep, as two reduction peaks are present at 0.55 and 0.48 V (peaks b and c). There is also a broad reduction wave at 0.05 V (peak d). The reduced species at 0.55 and 0.48 V can be reoxidized at 0.55 and 0.60 V (peaks e and f) on the subsequent positive-going scan. The peak currents for these two redox processes increase with cycle number, stabilizing after the fifth scan. Reduction peaks b and c are associated with oxidation peaks e and f, as all were present when the forward scan was reversed at 0.8 V. In other words, these peaks are associated with oxidation products formed during the initial positive-going scan. For phenol at diamond, these two sets of peaks have been reported to be associated with adsorbed hydroquinone/pbenzoquinone and catechol/o-benzoquinone redox couples (28). For 2-CP and 4-CP, the corresponding chlorinated or non-chlorinated hydroquinone/benzoquinone is expected (33). The current for peaks a, ipox, and d, ipred, at both electrode types varied linearly with ν1/2 between 0.05 and 0.25 V/s (r2 > 0.990) indicating that the reaction rates for the initial oxidation of 2-CP and the reduction of the product(s) formed are controlled by semi-infinite linear diffusion of the reactant to the electrode-solution interface. This suggests that soluble, rather than surface-confined, reactants and products are involved in the redox reactions. Rodrigo et al. observed similar voltammetric features for 4-CP and reported that the peak currents, a and d, increased with the addition of hydroquinone to the solution (29). We suppose that peak d is associated with the reduction of a dissolved quinone-like product formed at 1.1 V. The peak currents, b and c and e and f, on the other hand, increased linearly with the scan rate (r2 > 0.990), consistent with surface-confined redox processes. Further evidence that these redox reactions involve adsorbed species is the fact
that the peak currents were, for the most part, unaffected by electrolyte mixing. In these measurements, five cyclic voltammetric scans were recorded between -0.5 and 1.5 V to form the adsorbed oxidation reaction products. The analytecontaining solution was then vigorously agitated for several minutes by bubbling nitrogen gas. This was followed by recording another voltammetric scan. Oxidation peaks, e and f, and reduction peaks, b and c, were still present after the bubbling, although with some minor current attenuation. However, if the analyte-containing solution was removed from the cell, the electrode copiously rinsed with ultrapure water, and the cell filled with fresh electrolyte, then the background voltammetric i-E curves were devoid of these redox peaks. This indicates that these oxidation reaction products are weakly adsorbed to the diamond surface. We suppose that these surface-confined oxidation reaction products are either individual molecules adsorbed on the diamond surface, perhaps at localized sites where some polyphenol has formed, or they may be redox-active units within a polyphenol film on the surface. In other words, the adsorption is promoted by either the formation of surface oxides or the deposition of the polar polymer. The mechanistic issues of these reactions need to be further investigated, but preliminary microscopic surveys of the diamond electrode after extensive phenol electrolysis revealed the presence of isolated polymeric deposits. The fact that the primary oxidation peak current at 1.1 V is not significantly attenuated with prolonged potential cycling indicates that a blocking polyphenol layer does not form on the surface. In other words, some polymer may form on the electrode surface, but large uncoated areas of the electrode exist for the primary oxidation reaction to occur at. We suppose that the oxidation current at 1.1 V is due to the direct one-electron, one-proton oxidation of the phenol to form 2-chlorophenoxy radical (8, 9). It has been reported that Epox for the oxidation of another chlorinated phenol, 2,4-dichlorophenol, at “as deposited” (i.e., presumably hydrogen-terminated) diamond shifts toward less positive potentials with increasing pH up to the pKa value of ca. 8. A change of -59 mV/dec was found (28). This trend is consistent with an equal number of protons and electrons being transferred during the redox reaction. Rodrigo et al. have reported that this oxidation peak results from the oneelectron/one-proton oxidation to phenoxy radical or from an additional one-electron oxidation to form the phenoxy cation (29). The seminal mechanistic work by Gattrell and Kirk indicates that the phenol oxidation reaction (and presumably for the chlorinated phenol oxidation as well) proceeds initially through the formation of phenoxy radical species (6-9). Once formed, the radicals can react via two pathways: (i) radical-radical coupling to form oligomeric and polymeric species or (ii) subsequent direct or indirect oxidation to produce soluble or surface-confined oxidation products, like hydroquinone and catechol. The voltammetric responses for 3-CP and PCP were similar to each other at both electrode types but quite different from VOL. 38, NO. 13, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3677
FIGURE 2. Cyclic voltammetric i-E curves for 0.1 mM pentachlorophenol (PCP) in 50 mM phosphate buffer, pH 3.5, at (A) microcrystalline and (B) nanocrystalline diamond thin-film electrodes. Scan rate ) 0.1 V/s. Geometric area ) 0.2 cm2.
FIGURE 3. Cyclic voltammetric i-E curves for 0.1 mM 3-chlorophenol (3-CP) in 50 mM phosphate buffer, pH 3.5, at (A) microcrystalline and (B) nanocrystalline diamond thin-film electrodes. Scan rate ) 0.1 V/s. Geometric area ) 0.2 cm2. the responses for the other three phenols. As an example, Figure 2A,B shows five consecutively recorded cyclic voltammograms, at 0.1 V/s, for 0.1 mM PCP at a (A) microcrystalline and (B) nanocrystalline diamond thin-film electrode. The first forward scan for both electrode types has a welldefined oxidation peak at ca. 0.90 V, with a corresponding weak reduction peak at ca. -0.20 V during the reverse scan. The oxidation peak is less positive that the peak for phenol, 2-CP, and 4-CP by approximately 200 mV. This is because the pKa for PCP is 4.74, as compared to values from 8 to 10 for phenol and the other chlorinated phenols. The lower value reflects a greater ease of deprotonating the phenolic hydrogen. The oxidation peak charge, after stabilization, was nominally 18 µC. The 0.90 V peak is presumed to be associated with the formation of the dimer, 2,3,4,6-pentachlorophenoxy2,5-cyclohexadienone (13). Gatrell and MacDougall studied the electrochemical oxidation of PCP and concluded that the electro-oxidation involves a one-electron, one-proton oxidation reaction to form pentachlorophenoxy radical over a wide potential range. The radical then undergoes radicalradical coupling to form the dimmer (13). A slight decrease in the oxidation peak current and a slight increase in the reduction peak current are observed for the second and all subsequent scans. It is unclear at this point what reaction is associated with the reduction current. The pentachlorophenoxy radical most likely dimerizes to form a water insoluble species through C-O rather than C-C coupling due to steric limitations. The dimers formed could be 2,3,4,5,6-pentachloro-4-pentachlorophenoxy-2,5-cyclohexadienone and/or 2,3,4,5,6-pentachloro-2-pentachlorophenoxy-3,5-cyclohexadienone (13). Preliminary AFM surveys of a microcrystalline diamond electrode after 25 voltammetric scans in 0.1 mM PCP revealed the presence of isolated polymeric deposits on the surface. Figure 3A,B shows five consecutive cyclic voltammograms for 0.1 mM 3-CP at a microcrystalline and nanocrystalline diamond thin-film electrode. An oxidation peak is observed for all scans at ca. 1.2 V, with only a weak reduction peak at ca. -0.20 V. The anodic peak charge, after stabilization, was nominally 26 µC. This oxidation peak is believed to be due to the one-electron, one-proton oxidation reaction to form 3-chlorophenoxy radical. The oxidation peak current decreases with cycle number, more so for the nanocrystalline 3678
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 13, 2004
film, before reaching a pseudo-steady-state value. Contrary to the behavior of phenol, 2-CP, and 4-CP, no new oxidation or reduction peaks develop at lower potentials. Apparently, the surface-confined electroactive quinones are not byproducts of the oxidation of this analyte. For both 3-CP and PCP at microcrystalline and nanocrystalline diamond, the oxidation peak current (1.2 and 0.9 V, respectively) varied linearly with ν from 0.05 to 0.25 V/s (r2 > 0.990). This is consistent with a surface-confined redox reaction. These two reactants and or their products adsorb on the electrode more strongly than do the other phenolic compounds as a featureless background voltammogram was only observed if the passivated electrode was first potential cycled (ca. 10 scans) well into the oxygen evolution regime and then exposed to a fresh electrolyte solution. This suggests that the adsorbing products need to be oxidatively degraded by OH• in order to be removed from the surface. The same cannot be said for GC, as similar agitation did not restore the original current. The trends in the shapes of the cyclic voltammograms for each phenolic compound with cycle number were the same for GC and both diamond types. A difference was the ca. 100 mV less positive phenolic oxidation peak for GC. The oxidation reaction products very strongly adsorb to the GC surface resulting in significant fouling. For example, GC was irreversibly fouled after cycling in 2-CP and 3-CP. The oxidation peak current at 1.1 V decreased with cycle number and the original value could not be regained even with vigorous solution agitation and or replacement with fresh electrolyte. Polishing was the only means to remove all traces of the phenols. These cyclic voltammetric studies reveal that (i) all five phenolic compounds can be electro-oxidized at both types of diamond without significant electrode fouling, as is the case for many other electrodes such as GC, (ii) the electrooxidation reaction mechanism appears to proceed in a similar manner at both diamond types, regardless of the differences in film morphology and microstructure, with phenol, 2-CP, and 4-CP yielding similar responses, (iii) the oxidation of phenol, 2-CP, and 4-CP leads to the formation of surfaceconfined hydroquinone/p-benzoquinone and catechol/obenzoquinone redox couples (28), whereas the oxidation of 3-CP and PCP do not, and (iv) any adsorbed reaction products were easily removed from the diamond surface by copiously
FIGURE 4. Hydrodynamic voltammetric i-E curves and S/B vs potential plots for 40 µM concentrations of each phenolic compound in 50 mM phosphate buffer, pH 3.5, at (A and C) microcrystalline and (B and D) nanocrystalline diamond thin-film electrodes. Flow rate ) 0.6 mL/min. Each datum represents the average of 5 injections. rinsing with ultrapure water and exposure to fresh electrolyte. In some cases, such as that for PCP, limited potential cycling into the oxygen evolution regime was necessary to remove all traces of the reaction products. Flow Injection Analysis with Amperometric Detection. In FIA-EC, the background current for both the microcrystalline and nanocrystalline films stabilized rapidly after detector turn-on (Eappl ) 1.1 V), reaching a constant value within 5-10 min. In contrast, the background current for a GC electrode often required some 45-60 min to stabilize. A similar observation was reported by Terashima et al. (28). The rapid stabilization time is an advantageous property of diamond in this assay because it leads to a shorter overall analysis time. Figure 4A,B shows hydrodynamic i-E curves for one microcrystalline and one nanocrystalline diamond thin-film electrode during 20-µL injections of 40 µM phenolic solutions in 50 mM phosphate buffer, pH 3.5. The mobile phase flow rate was 0.6 mL/min. Each datum corresponds to the average signal for 5 injections. The error bars are within the size of the marker. The responses for phenol, 2-CP, 3-CP, and 4-CP are not sigmoidal in shapesa shape that is expected if the currents were limited exclusively by mass transport. The only sigmoidially shaped curve is that for PCP. These observations are different from those reported for diamond by Terashima et al. (28). The authors presented sigmoidally shaped hydrodynamic voltammograms for several di- and trichlorophenols. It is supposed that the maximum oxidation current for phenol, 2-CP, 3-CP, and 4-CP in the potential region positive of 800 mV is under mixed control by direct and indirect electro-oxidation reactions and mass transport. As discussed previously, the electro-oxidation involves the initial formation of phenoxy radical species. This is followed by either radical-radical coupling to form polymeric products or direct and indirect (OH• generation) oxidation reactions to form quinone-like products (8, 9, 28-34). At these two particular electrodes, the most negative onset potential for the oxidation current is seen for 2-CP. The maximum current varied, depending on the particular compound, with the largest value seen for 2-CP at both electrodes. Figure 4C,D shows plots of the signal-to-background (S/B)(Itot - Ibkg/Ibkg) ratio as a function of the applied potential for phenol, 2-CP, 3-CP, and 4-CP. Such plots are useful for determining the optimum detection potential when sigmoi-
dal hydrodynamic voltammograms are not obtained. The optimum detection potential for the analytes, including PCP, is in the 1100-1200 mV range for both diamond types. A sharp decrease in S/B for the nanocrystalline diamond is observed at 1300 mV. This is because of the onset of oxygen evolution, which occurs at a slightly less positive potential than for microcrystalline diamond. The most negative onset potential for the oxidation current and the largest response is observed for 2-CP. Irrespective of the analyte injection order, the largest S/B is seen for 2-CP at both electrode types. The lowest S/B (not shown) was observed for PCP. Table 1 contains a summary of the FIA-EC detection figures of merit observed for both microcrystalline and nanocrystalline diamond. The data presented are for at least three electrodes of each type. All measurements were made at 1.1 V for microcrystalline and 1.2 V for nanocrystalline diamond. The linear dynamic range for all the analytes at both diamond types is 3-4 orders of magnitude (r2 g 0.990) from the nM to µM range. Overall, slightly higher sensitivities are seen for nanocrystalline diamond, by a factor of 1.3-1.5 for some analytes. The sensitivities for all the analytes, except PCP, are approximately the same for a given diamond type with the value for PCP being about a factor of 3 lower. The lower sensitivity for PCP, as compared to the other analytes, could be due to a lower number of electrons transferred in the overall electrochemical reaction. The data for only one electrode of each type, shown in Figure 4C,D, reveal a sizable difference in the maximum response for each of the compounds. However, we observed this to be the exception rather than the rule. The more common observation is reflected in the sensitivity data in Table 1. The response precision (2-6%), based upon 20 consecutive measurements, is approximately the same for both electrode types. The concentration (i.e., the lowest concentration actually injected) and mass limits of quantitation for each analyte are also similar for both electrode types. The mass limits of quantitation are in the high pg to low ng range (S/N g 3). By way of comparison, reduced linear dynamic ranges, higher limits of quantitation, and greater response variabilities were observed for GC. More importantly, the response stability was poor for GC due to fouling by reaction intermediates and products. HPLC-EC with Amperometric Detection. All of the phenol and chlorinated phenols were effectively separated (gradient VOL. 38, NO. 13, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3679
TABLE 1. Summary of the Analytical Figures of Merit for the FIA-EC (and HPLC-EC) Detection of Phenol and Chlorinated Phenols Using Microcrystalline and Nanocrystalline Diamond Electrodes
a
compd
linear dynamic range r 2 > 0.99 (µM)
sensitivity (nA/µM)
phenol 2-CP 3-CP 4-CP PCP
0.3-100 0.05-200 0.1-100 0.5-100 0.6-1200
Microcrystalline 9.85 ( 1.24 14.19 ( 0.51 10.03 ( 2.16 11.03 ( 1.25 3.39 ( 0.12
phenol 2-CP 3-CP 4-CP PCP
0.3-100 0.05-200 0.1-100 0.5-100 0.6-1200
Nanocrystalline 18.41 ( 1.29 17.85 ( 1.82 18.82 ( 1.37 16.35 ( 1.48 6.85 ( 1.86
concn LOQa (µM) S/N g 3
mass LOQa (ng)
4.12 1.91 5.76 1.66 1.58
0.30 0.050 0.10 0.50 0.60
0.77 0.13 0.26 1.28 3.19
3.89 1.43 2.18 2.23 2.57
0.30 0.050 0.10 0.50 0.60
0.77 0.13 0.26 1.28 3.19
response precision (RSD%)
LOQ, limit of quantitation.
TABLE 2. Summary of the Analytical Figures of Merit for the FIA-EC (and HPLC-EC) Detection of Phenol and Chlorinated Phenols Using Microcrystalline and Nanocrystalline Diamond Electrodes
a
compd
linear dynamic range r 2 > 0.99 (µM)
sensitivity (nA/µM)
phenol 2-CP 3-CP 4-CP PCP
0.1-80 0.1-60 0.1-60 0.3-60 0.1-80
Microcrystalline 1.93 ( 0.14 2.03 ( 0.37 1.04 ( 0.22 1.40 ( 0.31 0.90 ( 0.12
phenol 2-CP 3-CP 4-CP PCP
0.1-80 0.1-60 0.1-60 0.1-60 0.1-80
Nanocrystalline 1.96 ( 0.48 2.33 ( 0.38 0.98 ( 0.22 1.39 ( 0.29 1.03 ( 0.09
concn LOQa (µM) S/N g 3
mass LOQa (ng)
3.76 3.18 2.85 4.39 3.82
0.10 0.10 0.10 0.10 0.10
0.19 0.26 0.26 0.26 0.53
5.16 4.09 3.19 5.12 5.62
0.10 0.10 0.10 0.10 0.10
0.19 0.26 0.26 0.26 0.53
response precision (RSD%)
LOQ, limit of quantitation.
elution), using a 50 mM phosphate buffer/acetonitrile, pH 3.5 (65:35 (v/v)) mobile phase for the first four solutes (phenol, 2-CP, 3-CP, and 4-CP), and a 20:80 (v/v) composition for the fifth solute (PCP), and detected amperometrically at both diamond electrode types. Both diamond types had low and stable background currents in the mixed mobile phase, even at the positive detection potentials. Isocratic elution is usually the separation method of choice with electrochemical detection because gradient elution often causes drifting baseline currents and long stabilization times for many electrodes. The background current for diamond at the detection potential (1.1 or 1.2 V) was low at about 8 nA and increased only slightly with an increase in the organic modifier content. The peak-to-peak noise was ca. 200 pA. Rapid stabilization times were observed for both diamond types after detector turn-on in this mobile phase. Figure 5 shows a typical chromatogram for the five phenols detected amperometrically at a microcrystalline diamond electrode. The separation is complete in about 16 min, and the peaks are symmetric with minimal peak tailing. The analytical detection figures of merit are summarized in Table 2. The linear dynamic range for all the analytes at both diamond types is 2-3 orders of magnitude (r2 > 0.990). It should be emphasized that the maximum of 60 or 80 µM does not represent the upper limit of the assay, rather just the maximum concentration that was used in these particular measurements. The analyte sensitivities are approximately the same for both electrode types and are a little lower in the mixed aqueous/organic solution than they are in the pure aqueous medium (see Table 1). The response precisions (26%) are the same for both electrode types and were 3680
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 13, 2004
FIGURE 5. Reversed-phase chromatogram for the separation and amperometric detection of the phenol and chlorinated phenols at a microcrystalline diamond thin-film electrode. The concentration of each compound was 20 µM. Injection volume ) 20 µL. Detection potential ) +1.15 V vs Ag/AgCl. The compounds are (1) phenol, (2) 2-CP, (3) 3-CP, (4) 4-CP, and (5) PCP. A two-step gradient elution was used. Initially, a 65:35 50 mM phosphate buffer, pH 3.5/ acetonitrile mobile phase was used and was switched to a 20:80 mixture at the 10 min mark. determined from repetitive injections of 20 µM analyte concentrations. Both diamond types exhibited good runto-run response precision with no evidence of fouling or deactivation. The concentration limit of quantitation is the same for all the analytes, 0.1 µM (low to mid ppb range), which corresponds to a mass limit of quantitation ranging from 0.1 to 0.5 ng (S/N g 3). The detection figures of merit for the microcrystalline and nanocrystalline diamond compare favorably with those previously reported for oxygenterminated diamond electrodes (28, 40) and are somewhat improved over those reported for some other electrodes (40-
chemical detection of phenol and chlorinated phenols is possible using boron-doped diamond thin-film electrodes. Both microcrystalline and nanocrystalline diamond are responsive for these pollutants without extensive pretreatment and provide a stable response over time with minimal electrode fouling. The electro-oxidation of phenol and the chlorinated phenols and the amperometric detection of these pollutants occurs in a similar fashion at both diamond electrode types regardless of their differences in morphology and microstructure. The preliminary result for the soil sample illustrates the potential for using this assay to monitor phenols and chlorinated phenols in complex “real world” samples.
Acknowledgments
FIGURE 6. Reversed-phase chromatogram for the separation and amperometric detection of 2-CP in a contaminated soil sample at a microcrystalline diamond thin-film electrode. Injection volume ) 20 µL. Detection potential ) +1.15 V vs Ag/AgCl. A two-step gradient elution was used. Initially, a 65:35 50 mM phosphate buffer, pH 3.5/acetonitrile mobile phase was used and was switched to a 20:80 mixture at the 10 min mark.
43). In addition, the limits of quantitation for the phenolic analytes in our HPLC-EC method are comparable to those reported for EPA Method 1625 (44). Soil Analysis. As mentioned above, a soil sample containing some 60 different organic contaminants was procured from a commercial source. The organic pollutants, many of which are electroactive, were extracted and analyzed by HPLC-EC. Figure 6 shows a chromatogram for the sample extract. Amperometric detection was accomplished at 1.1 V using a microcrystalline diamond electrode. Twenty microliters of the sample was injected onto the reversed-phase column, and elution was performed using a two-step gradient. A 65:35 50 mM phosphate buffer, pH 3.5/acetonitrile mobile phase was used initially to elute the more polar analytes, and this composition was adjusted to 20:80 at ca. the 10 min mark to elute the more nonpolar solutes. The chromatogram reveals the presence of 21 different peaks, each of which presumably corresponds to an electroactive solute. The peak for 2-CP occurs at 6.2 min and is labeled on the chromatogram. This signal was determined to be associated with 2-CP based on results for the soil sample spiked with the compound. However, the peak in the chromatogram represents the raw signal for the unknown 2-CP in the soil sample. A standard calibration curve was generated between 0.1 and 80 µM for quantitation of 2-CP. A linear response was observed with a linear regression correlation coefficient of 0.9963. Standard addition is the method of choice for quantifying 2-CP in this complex sample, but we did not have enough of the soil to conduct this type of calibration. Based on the calibration curve generated from standard solutions, the concentration of 2-CP in the soil was determined to be 2650 µg (2-CP)/kg (soil). The supplier of the contaminated soil sample specified a low and high acceptance range of 2643 and 5983 µg/kg and an assigned value of 4313 µg/kg for 2-CP. The experimentally determined value is lower than the assigned value and is on the low end of the acceptance range. However, given that specific details regarding the sample preparation and the length of time the sample was stored are not readily available and that we did not conduct any tests to determine the percent recovery using our extraction procedure, we feel that our experimentally determined value reflects the true level of 2-CP present. The FIA-EC and HPLC-EC results confirm that efficient separation, and sensitive, reproducible and stable electro-
The research was generously supported by the National Research Initiative Competitive Grants Program, administered by the U.S. Department of Agriculture (2001-3510210045).
Literature Cited (1) Gilman, A.; Douglas, U.; Arbuckle-Sholtz, T.; Jamieson, J. Chlorophenols and Their Impurities: A Health Hazard Evaluation; Bureau of Chemical Hazards, Environmental Health Directorate, National Health and Welfare: Ottawa, Ontario, Canada 982. (2) Kristiansen, N.; Froshaug, M.; Aune, K.; Becher, G. Environ. Sci. Technol. 1994, 28, 166. (3) Toxic Substance Control Act; U.S. Environmental Protection Agency: Washington, DC 1979. (4) Drinking Water Standards and Health Advisories, Summer, 2000; U.S. Environmental Protection Agency: 2000. (5) Sharifian, H.; Kirk, D. W. J. Electrochem. Soc: Electrochem. Sci. Technol. 1986, 133, 921. (6) Gattrell, M.; Kirk, D. W. Can. J. Chem. Eng. 1990, 68, 997. (7) Gattrell, M.; Kirk, D. W. J. Electrochem. Soc., 1992, 139, 2736. (8) Gattrell, M.; Kirk, D. W. J. Electrochem. Soc. 1993, 140, 903. (9) Gattrell, M.; Kirk, D. W. J. Electrochem. Soc. 1993, 140, 1534. (10) Belhadj Tahar, N.; Savall, A. J. Electrochem. Soc. 1998, 145, 3427. (11) Rodgers, J. D.; Jedral, W.; Bunce, N. J. Environ. Sci. Technol. 1999, 33, 1453. (12) Johnson, S. K.; Houk, L. L.; Feng, J.; Houk, R. S.; Johnson, D. C. Environ. Sci. Technol. 1999, 33, 2638. (13) Gattrell, M.; MacDougall, B. J. Electrochem. Soc. 1999, 146, 3335. (14) Bruno, F.; Pham, M. C.; Dubois, J. E. Electrochim. Acta 1997, 22, 451. (15) Mengoli, G.; Musian, M. J. Electrochem. Soc. 1987, 143, 643C. (16) Fleischmann, M.; Hill, I. R.; Mengoli, G.; Musian, M. M. Electrochim. Acta 1983, 28, 1545. (17) Lu, W.; Wallace, G. G.; Imisides, M. D. Electroanalysis 2002, 14, 325. (18) Swain, G. M.; Ramesham, R. Anal. Chem. 1993, 65, 345. (19) Swain, G. M. Adv. Mater. 1994, 6, 388. (20) Swain, G. M. J. Electrochem. Soc. 1994, 141, 3382. (21) Chen, Q.; Granger, M. C.; Lister, T. E.; Swain, G. M. J. Electrochem. Soc. 1997, 144, 3086. (22) Xu, J.; Granger, M. C.; Chen, Q.; Strojek, J. W.; Lister, T. E.; Swain, G. M. Anal. Chem. 1997, 69, 591A. (23) Swain, G. M.; Anderson, A. B.; Angus, J. C. MRS Bull. 1998, 23, 56. (24) Granger, M. C.; Witek, M.; Xu, J.; Wang, J.; Hupert, M.; Hanks, A.; Koppang, M. D.; Butler, J. E.; Lucazeau, G.; Mermoux, M.; Strojek, J. W.; Swain, G. M. Anal. Chem. 2000, 72, 3793. (25) Chen, Q.; Gruen, D. M.; Krauss, A. R.; Corrigan, T. D.; Witek, M.; Swain, G. M. J. Electrochem. Soc. 2001, 148, E44. (26) Xu, J.; Chen, Q.; Swain, G. M. Anal. Chem. 1998, 70, 3146. (27) Show, Y.; Witek, M. A.; Sonthalia, P.; Swain, G. M. Chem. Mater. 2003, 15, 879. (28) Terashima, C.; Rao, T. N.; Sarada, B. V.; Tryk, D. A.; Fujishima, A. Anal. Chem. 2002, 74, 895. (29) Rondrigo, M. A.; Michaud, P. A.; Duo, I.; Panizza, M.; Cerisola, G.; Comninellis, Ch. J. Electrochem. Soc. 2001, 148, D60. (30) Hagans, P. L.; Natishan, P. M.; Stoner, B. R.; O’Grady, W. E. J. Electrochem. Soc. 2001, 148, E298. (31) Polcaro, A. M.; Vacca, A.; Palmas, S.; Mascia, M. J. Appl. Electrochem. 2003, 33, 885. (32) Codognoto, L.; Machado, S. A. S.; Avaca, L. A. J. Appl. Electrochem. 2003, 33, 951. VOL. 38, NO. 13, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3681
(33) Canizares, P.; Garcia-Gomez, J.; Saez, C.; Rodrigo, M. A. J. Appl. Electrochem. 2003, 33, 917. (34) Zhi, J.-F.; Wang, H.-B.; Nakashima, T.; Rao, T. N.; Fujishima, A. J. Phys. Chem. B 2003, 107, 13389. (35) Fischer, A.; Show, Y.; Swain, G. M. Anal. Chem. 2004, 76, 2553. (36) Swain, G. M. In Electroanalytical Chemistry; Bard, A. J., Rubinstein, I., Eds.; Marcel Dekker: New York, 2003; Vol. 22, p 181. (37) Koppang, M.; Witek, M.; Blau, J.; Swain, G. M. Anal. Chem. 1999, 71, 1188. (38) Granger, M. C.; Xu, J.; Strojek, J. W.; Swain, G. M. Anal. Chim. Acta 1999, 397, 145. (39) Jolley, S.; Koppang, M.; Jackson, T.; Swain, G. M. Anal. Chem. 1997, 69, 4099.
3682
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 13, 2004
(40) Prado, C.; Murcott, G. G.; Marken, F.; Foord, J. S.; Compton, R. Electroanalysis 2002, 14, 975. (41) Agui, L.; Serra, B.; Yanez-Sedeno, P.; Reviejo, A.; Pingarron, J. M. Electroanal. 2001, 13, 1231. (42) Liu, X.; Frank, H. J. High Resol. Chromatogr. 1998, 21, 309. (43) Dressman, S. F.; Simeone, A. M.; Michael, A. C. Anal. Chem. 1996, 68, 3121. (44) Standard Methods for the Examination of Water and Wastewater; 18th ed.; U.S. Environmental Protection Agency: 1992.
Received for review June 25, 2003. Revised manuscript received April 5, 2004. Accepted April 6, 2004. ES034656E
