Effect of pH and Surface Functionalities on the Cyclic Voltammetric

Justin A. JohnsonCaddy N. HobbsR. Mark Wightman. Analytical Chemistry ... Adam K. Dengler , R. Mark Wightman , and Gregory S. McCarty. Analytical Chem...
0 downloads 0 Views 87KB Size
Download PDF
Anal. Chem. 1999, 71, 2782-2789

Effect of pH and Surface Functionalities on the Cyclic Voltammetric Responses of Carbon-Fiber Microelectrodes Petrise L. Runnels, Joshua D. Joseph, Michael J. Logman, and R. Mark Wightman*

Department of Chemistry, University of North Carolina, Chapel Hill, North Carolina 27599-3290

Carbon electrodes are useful for the detection of oxidizable species with cyclic voltammetry. In particular, carbonfiber microelectrodes have been employed for the measurement of several neurotransmitters in brain tissue. However, during cyclic voltammetry with carbon-fiber electrodes the current varies with changes in concentration of some inorganic cations as a result of their interaction with surface functional groups. The electrode’s response to the hydronium ion is a particular concern because its voltammetric response occurs over a broad range of potentials that overlap those of neurotransmitters of interest such as dopamine. This is especially a problem in vivo because simultaneous changes of dopamine and pH frequently occur in brain tissue. In this work, voltammetric current changes are shown to arise from pH dependent shifts in the peak potentials of background voltammetric waves that arise from species confined to the carbon-fiber electrode surface. Polishing the electrode with alumina suspended in cyclohexane in an environment containing lowered oxygen, a method previously demonstrated to remove oxides from the carbon surface, leads to a substantial reduction in the sensitivity to pH changes. However, this is accompanied by a loss in signal amplitude for dopamine. The dopamine response can be restored using the cation exchanger Nafion without significantly increasing the pH response. To investigate which oxide functional groups play a direct role in the electrode’s current responses to changes in pH, surfaceconfined carbonyl and alcohol functionalities were chemically modified. In both cases, the modification did not affect the carbon-fiber electrode’s responsiveness to changes in pH. Nonetheless, the polishing technique proved to be effective in reducing pH interferences in in vivo applications. Electrochemistry at carbon tends to be markedly different from that at metal electrodes such as mercury, platinum, and gold. Considerable evidence suggests that these differences have their origin in the surface chemistry of carbon electrodes. This chemistry can affect the kinetics of various classes of redox systems in different ways.1 Indeed, cyclic voltammetric results at * To whom correspondence should be addressed. Tel.: (919) 962-2439. Fax: (919) 962-2388. E-mail: [email protected]. (1) McCreery, R. L. In Electroanalytical Chemistry; Bard, A. J., Ed,; Dekker: New York, 1991; Vol. 17, pp 221-374.

2782 Analytical Chemistry, Vol. 71, No. 14, July 15, 1999

carbon electrodes can be greatly altered by surface pretreatments such as chemical derivatization,2-5 polishing,2-8 laser activation, 7,9 heat treatment,2,3,6,10,11 and electrochemical pretreatment.6,12-17 The actual chemical and physical processes that are affected by these treatments are still unclear, partially because the mechanisms resulting in surface activation depend on the initial state of the carbon. For example, carbon surfaces with the basal plane predominantly exposed to solution exhibit slower electrochemical kinetics while pretreatments generate more edge plane.6,7,18,19 Processes such as vacuum heat treatment and laser activation at glassy carbon or other carbon surfaces with a significant amount of edge-plane orientation are believed to remove surface contaminants and reveal more active sites.7,9,10 Heat treatment at atmospheric pressure improves kinetics presumably by adding oxide functionalities to the surface.11 Electrochemical pretreatment appears to free the surface from polishing debris and add a porous oxide film layer.15 Removal of surface oxides through vacuum heat treatment or by polishing in an inert solvent also has a dramatic effect on electrode kinetics.3 McCreery has proposed three classes of redox processes at carbon electrodes.3 First are the outer sphere systems such as Ru(NH3)63+/2+ for which electron-transfer kinetics are relatively insensitive to surface chemistry. For such systems, neither reduction of surface oxygen/carbon ratios nor modification of the carbon surface affect electron-transfer rates beyond what would be expected as a result of electron tunneling. A second class, (2) Deakin, M. R.; Stutts, K. J.; Wightman, R. M. J. Electroanal. Chem. 1985, 182, 113. (3) Chen, P.; McCreery, R. L. Anal. Chem. 1996, 68, 3958. (4) Chen, P.; Fryling, M. A.; McCreery, R. L. Anal. Chem. 1995, 67, 3115. (5) Tse, D. C.; Kuwana, T. Anal. Chem. 1978, 50, 1315. (6) Wightman, R. M.; Deakin, M. R.; Kovach, P. M.; Kuhr, W. G.; Stutts, K. J. J. Electrochem. Soc. 1984, 131, 1578. (7) Rice, R. J.; Pontikos, N. M.; McCreery, R. L. J. Am. Chem. Soc. 1990, 112, 4617. (8) Hu, I. F.; Karweik, D. H.; Kuwana, T. J. Electroanal. Chem. 1985, 188, 59. (9) Poon, M.; McCreery, R. L. Anal. Chem. 1986, 58, 2745. (10) Fagan, D. T.; Hu, I. F.; Kuwana, T. Anal. Chem. 1985, 57, 2759. (11) Stutts, K. J.; Kovach, P. M.; Kuhr, W. G.; Wightman, R. M. Anal. Chem. 1983, 54, 1632. (12) Gonon, F. G.; Fombarlet, C. M.; Buda, M. J.; Pujol, J. F. Anal. Chem. 1981, 53, 1386. (13) Falat, L.; Cheng, H. Y. Anal. Chem. 1982, 54, 2108. (14) Engstrom, R. C., Strasser, V. A. Anal. Chem. 1984, 56, 136. (15) Kepley, L. J.; Bard, A. J. Anal. Chem. 1988, 60, 1459. (16) Engstrom, R. C. Anal. Chem. 1982, 54, 2310. (17) Cabaniss, G. E.; Diamantis, A. A.; Murphy, W. R., Jr.; Linton, L. W.; Meyer, T. J. J. Am. Chem. Soc. 1985, 107, 1845. (18) McDermott, C. A.; Kneten, K. R.; McCreery, R. L. J. Phys. Chem. 1992, 96, 3124. (19) Kneten, K. R.; McCreery, R. L. Anal. Chem. 1992, 64, 2518. 10.1021/ac981279t CCC: $18.00

© 1999 American Chemical Society Published on Web 05/29/1999

which includes inner-sphere systems such as Fe3+/2+ and Eu2+/3+, is found to be catalyzed by surface oxides. The third class includes inner-sphere systems that are surface-sensitive but not oxidesensitive. In this category are compounds such as ascorbic acid and Fe(CN)63-/4-. Such molecules experience a decrease in electron-transfer rates when a molecular monolayer coats the carbon electrode surface, but no change in rate is apparent when particular oxide functionalities are blocked. Recent work has shown that carbon electrode surfaces are responsive to pH as well as to redox-active molecules.20 Changes in solution pH lead to shifts in the voltammetric features that arise from surface functional groups.21 Therefore, when voltammograms obtained before and after a pH change are subtracted, a difference current is observed that has broad peaks that resemble a cyclic voltammogram of a solution-phase redox species. Subtraction of cyclic voltammograms is frequently done to remove the background current when rapid scan rates (300 V/s) are employed to monitor the concentration dynamics of neurotransmitters.22 However, interpretation of such data is difficult when pH changes occur. Such changes occur frequently in biological tissue as a result of the coupling of respiration and energy production to pH. In this work, we show that the pH response and the voltammetric features arising from surface functional groups can be diminished with a polishing procedure previously demonstrated to lower surface oxides.3 This technique has been applied at the surface of carbon-fiber microelectrodes, the style of electrode most commonly used for in vivo applications. However, this procedure leads to a decrease in signal for various classes of electroactive molecules, including dopamine. To restore sensitivity to this important neurotransmitter, the electrodes were coated with Nafion. Electrodes treated in this manner were shown to be effective in removing pH interferences in in vivo applications. To investigate the specific type of oxide group responsible for the surface wave dependence on pH, the carbon electrode surfaces were covalently modified, through methods previously described,3 with 2,4-dinitrophenylhydrazine (DNPH) and 3,5-dinitrobenzoyl chloride (DNBCl) that selectively block carbonyl and alcohol groups, respectively. These modifying agents contain nitro groups that can report a successful surface modification by their voltammetric reduction. Neither treatment removed the pH sensitivity. EXPERIMENTAL SECTION Apparatus and Solutions. The potentiostat employed was an EI-400 (Cypress Systems, Lawrence, KS). Cyclic voltammetry experiments were run with either CV6.EXE (Cypress Systems) or VA.EXE (program written in-house). For flow cell experiments, a -0.4-1.0 V, computer-generated, triangular wave scanned at 300 V/s was used with 100 ms between each voltammogram. The reference electrode was a sodium-saturated calomel reference electrode (SSCE), and the counter electrode was a platinum wire. Cyclic voltammograms were recorded with the carbon-fiber electrode in the outlet of a flow injection analysis system.23 (20) Jones, S. R.; Mickelson, G. E.; Collins, L. B.; Kawagoe, K. T.; Wightman, R. M. J. Neurosci. Methods 1994, 52, 1. (21) Kawagoe, K. T.; Garris, P. A.; Wightman, R. M. J. Electroanal. Chem. 1993, 359, 193. (22) Cahill, P. S.; Walker, Q. D.; Finnegan, J. M.; Mickelson, G. E.; Travis, E. R.; Wightman, R. M. Anal. Chem. 1996, 68, 3180. (23) Kristensen, E. W.; Wilson, R. L.; Wightman, R. M. Anal. Chem. 1986, 54, 986.

Background-subtracted voltammograms were obtained by averaging several scans obtained in the presence of buffer and subtracting them from the average of several scans obtained during the injection. To investigate the effect of pH on the cyclic voltammograms, buffer containing 10 mM dimethyl glutaric acid (DMGA, ICN Biomedicals Inc., Aurora, OH) and 100 mM NaCl (Mallinckrodt, Paris, KY) was used. Buffer solutions of pH 3.1, 3.6, 4.1, 4.6, 5.1, 5.6, 6.1, 6.6, 7.1, and 7.6 were used as “run” buffers. Injections were made with DMGA solutions that differed by -0.1 pH unit. Other measurements were made in a buffer containing 20 mM HEPES (Sigma, St. Louis, MO) and 150 mM NaCl adjusted to pH 7.4. To determine the responsiveness to pH, HEPES buffer of pH 7.0 was injected and the response compared with that for 10 µM dopamine (DA, Sigma) in the HEPES buffer. In addition to DA, redox systems used in these experiments were 100 µM 4-methyl catechol (4-MC, Sigma), 200 µM dihydroxyphenylacetic acid (DOPAC, Sigma), 500 µM Fe(CN)63-/4(synthesized in-house), 500 µM Ru(NH3)63+/2+ (Aldrich, Milwaukee, WI), and 10 µM 4-acetamidophenol (acetaminophen, Sigma). Stock solutions for DA, 4-MC, acetaminophen, and DOPAC were prepared in 0.1 N HClO4 (diluted from 60% HClO4, Mallinckrodt), and stock solutions for Fe(CN)63-/4- and Ru(NH3)63+/2+ were prepared in doubly distilled deionized water. The sample solutions of DA, DOPAC, and Ru(NH3)63+/2+ were prepared in HEPES buffer, and the sample solution of acetaminophen was prepared in HEPES buffer containing 1.2 mM CaCl2‚2H2O. The sample solution of Fe(CN)63-/4- was prepared in pH 3.0 K2SO4 (Fisher Scientific, Fair Lawn, NJ) buffer. All buffer solutions were prepared in doubly distilled deionized water. Electrode Preparation. Carbon-fiber microelectrodes were constructed from single 5-µm radius Thornel P55 carbon fibers (Amoco, Greenville, SC) as previously described.24 The tips were polished on an imbedded diamond dust polishing wheel (Sutter Instruments, Novoto, CA) at 25° to give an elliptical surface. Electrodes were soaked in 2-propanol for at least 15 min before use. Electrode Surface Modification. Covalent modification procedures were adapted from Chen et al.3 A 10 mM ethanol (1% HCl by volume) solution of DNPH (Sigma, St. Louis, MO) was deoxygenated with nitrogen prior to use. Carbon-fiber electrodes were immersed into the solution, and it was heated to its boiling point (about 75 °C). The solution was then removed from heat, and the electrodes were kept in solution for 2 h. Then, they were removed, thoroughly rinsed with ethanol, and soaked in isopropyl alcohol for 20 min prior to use. A 0.1 M solution of DNBCl (Sigma, St. Louis, MO) was prepared in pyridine. A polished electrode was immersed in the solution, which was then heated. After 5 min, the solution was removed from heat, and the electrode continued to soak for 1 h. The electrode was then removed from the solution and rinsed, first with pyridine and then with isopropyl alcohol. The electrode was soaked in isopropyl alcohol prior to use. The electrodes were tested by flow injection for changes to the voltammetric response to DA and pH changes before and after modification. After use, the modified electrode was submerged (24) Kawagoe, K. T.; Zimmerman, J. B.; Wightman, R. M. J. Neurosci. Methods 1993, 48, 225.

Analytical Chemistry, Vol. 71, No. 14, July 15, 1999

2783

into a stationary disodium citrate buffer (pH 2.5). The potential was scanned from 0 to -1.0 to 0 V at a rate of 0.1 V/s, and the current arising from the reduction of the nitrophenyl group was plotted on an X-Y recorder. Two consecutive scans were collected. Cyclohexane Polishing. The polishing procedure to remove surface oxides was also adapted from Chen et al.3 Cyclohexane was deoxygenated by saturation with nitrogen for 20-30 min. Nine and one-half micrometer Al2O3 (Buehler, Lake Bluff, IL) was dried in a 500° C oven for at least 24 h. The electrode which was previously polished as described above was polished again at the same angle for 10 min in the cyclohexane/Al2O3 slurry on a nylon polishing cloth (Buehler) attached to a glass polishing wheel (World Precision Instruments, Sarasota, FL). After polishing, the electrode was sonicated for 10 min in deoxygenated cyclohexane to remove polishing debris. Some electrodes were coated with Nafion by dipping and holding them in a 2.5% solution of Nafion (suspended in 2-propanol from a 5% solution, Aldrich, Milwaukee, WI). The coated electrodes were then dried with a heat gun for 10 min. In Vivo Experiments. Surgical procedures for in vivo voltammetry have been described elsewhere.25 Briefly, male SpragueDawley rats (Charles River Inc., Raleigh, NC) weighing 300-400 g were deeply anesthetized with urethane (1.5 g/kg; Sigma). The rats were immobilized in a stereotaxic apparatus (Kopf), and holes were made in the skull for placement of reference (Ag/AgCl, Bioanalytical Systems Inc., West Lafayette, IN) and carbon-fiber working electrodes. The placement of working electrodes was based on flat-skull coordinates obtained from a brain atlas.26 Working electrodes were placed in the caudate-putamen (CP). Coordinates (in mm) anteroposterior (AP) and mediolateral (ML) from bregma and dorsoventral (DV) from dura were +1.2 AP, +2.2 ML, -4.8 to -5.0 DV. Carbon-fiber microelectrodes used in vivo were polished at 20°. They were calibrated with 2.5 µM dopamine and 10 µM acetaminophen in buffer containing 150 mM NaCl, 20 mM Hepes, and 1.2 mM CaCl2‚2H2O before and after cyclohexane treatment and Nafion coating. A triangle wave (-0.4 to 1.0 V vs a Ag/AgCl reference electrode, at a scan rate of 300 V/s) was applied every 500 ms to working electrodes upon placement in the rat brain. The carbonfiber microelectrodes were lowered slowly to the CP. Background current from the electrode was monitored from 45 to 90 min prior to drug injection, after which a 75 mg/kg injection of acetaminophen was administered. Electrodes were typically in the brain for 4-6 h for these experiments. Simulations. Simulations were conducted using DigiSim 2.1. For the simulation of a solution-phase species at pH 7.0, a planar electrode geometry was assumed with semi-infinite linear diffusion. Solution resistance and double layer capacitance were assumed to be zero. Simulations were made of the H+e-H+eproton/electron-transfer mechanism,27 with the previously reported pKa values28 and the following parameters: E°1 ) 0.02V, E°2 ) 0.13 V, R1 ) R2 ) 0.5, k°1 ) 0.02, k°2 ) 0.06 cm/s. Because (25) Garris, P. A.; Collins, L. B.; Jones, S. R.; Wightman, R. M. J. Neurochem. 1993, 61, 637-647. (26) Paxinos, G.; Watson, C. The Rat Brain in Stereotaxic Coordinates; Academic Press: New York, 1986. (27) Deakin, M. R.; Wightman, R. M. J. Electroanal. Chem. 1986, 206, 167.

2784 Analytical Chemistry, Vol. 71, No. 14, July 15, 1999

proton transfers are considered to be fast, both forward protontransfer rates were set at 1 × 1010 M-1s-1. Electrode radii were experimentally determined from steady-state amperometry in solutions of 10 µM DA (D ) 6 × 10-6 cm2/s) yielding electrode areas of approximately 5 × 10-6 cm2. To simulate the peak potentials of the cyclic voltammetric response of a surface-confined species, a finite diffusion thickness of 1 × 10-6 cm was assumed. RESULTS AND DISCUSSION Voltammetric Surface Waves at Carbon-Fiber Electrodes. When the potential of a carbon-fiber electrode was scanned in aqueous electrolyte between -0.4 and 1.0 V at 300 V/s, the observed charging current corresponded to a double layer capacitance of 17 µF/cm2. Superimposed on the charging current were broad current peaks in each scan direction. The position of these peaks was pH dependent, (Figure 1) and the current amplitude tended to decrease with higher pH. Because of the pH dependence of the position and shape of the peaks, subtraction of cyclic voltammograms recorded in buffers differing in pH by 0.1 unit resulted in the peaks in the subtracted cyclic voltammograms (Figure 1). Over the pH range from 3.0 to 7.5, the peak current for the reductive peak of the subtracted voltammogram decreased from approximately 7 to 1% of the charging current and the oxidative peak decreased from 3 to 1% of the charging current. In general, the cathodic peak was sharp and distinct, whereas the anodic peak was low and broad, especially at pH 5.5. The potentials of the peaks of the anodic and cathodic background waves are shown as a function of pH in Figure 2. The peak potential of the cathodic feature (Epc) exhibited a greater dependence on pH than that of the oxidative peak (Epa), especially at higher pH values. (For the anodic peak, the potential was taken from the subtracted voltammograms where it is better defined than in the unsubtracted cyclic voltammogram.) Over the pH range from 3.0 to 5.0, the value of Epc changed with a slope of -66 mV/pH. This slope increased to -134 mV/pH over the pH range from 5.5 to 7.0. For Epa, a slope of only -30.8 mV/pH was obtained over the pH range from 3.0 to 5.0, and there was virtually no change in position of the anodic peak potential over the higher pH values tested. Background waves such as are shown in Figure 1 can have several origins. One process involves the protonation and deprotonation of surface-confined acids.29 The potential distribution across the interface and its accompanying double-layer charge can interact with acidic groups at the interface. For example, as the potential of the electrode is made positive with respect to the point of zero charge, positive charge accumulates at the electrode surface. If this occurs in a pH range near the pKa of the surfaceconfined acid, the accumulation of positive charge is accompanied by a deprotonation of the acid requiring an increased flow of positive charge at potentials near the point of zero charge of the electrode. The opposite process will occur when the electrode potential is swept negative of the point of zero charge. Since the degree of protonation depends on the solution pH and the number of functional groups, current peaks that arise as a function of this (28) Deakin, M. R.; Kovach, P. M.; Stutts, K. J.; Wightman, R. M. Anal. Chem. 1986, 58, 1474. (29) Cui, Q.; Stevenson, K. J.; White, H. S. Proceedings of the 191st Mtg. of the Electrochem. Soc., Montreal, 1997.

Figure 1. pH dependence of voltammetric surface waves: Unsubtracted (left) and background-subtracted (right) cyclic voltammograms of acidic pH changes (∆pH ) -0.1) recorded at 300 V/s.

Figure 2. Peak potential vs pH. (B) Cathodic; backgroundsubtracted cyclic voltammograms. (2) Anodic; unsubtracted cyclic voltammograms. Insets: Representative unsubtracted cyclic voltammograms. Scan rate for all: 300 V/s.

phenomenon should be maximal at a solution pH near the pKa of the surface-confined acid. This mechanism does not seem to describe the pH dependent surface currents observed at carbonfiber electrodes because they increase in magnitude with decreasing pH over the entire range investigated. In addition, the pH dependence of surface waves is different on the anodic and cathodic scanssa prediction not made by this model.29

Figure 3. Ratio of postmodification to premodification oxidation current (ipost/ipre × 100) for 5 µM DA and an acidic pH change (∆pH ) -0.1). (a) DNPH; (b) DNBCl. Insets: Voltammograms demonstrating successful modification recorded at 0.1 V/s.

An alternative cause of background features is the presence of surface-confined species that undergo oxidation and reduction during cycling of the electrode. In previous work it was suggested that oxidation of a catechol moiety could be responsible.30 Indeed, a variety of oxygen containing functional groups are found on Analytical Chemistry, Vol. 71, No. 14, July 15, 1999

2785

Figure 4. Cyclic voltammograms showing effects of cyclohexane polishing at pH 7.4. Before cyclohexane polishing (left). After cyclohexane polishing (right). (a,b) Background; (c,d) [Dotted line in c represents simulated voltammogram.]10 µM DA; (e,f.) Acidic pH change (∆pH ) -0.4 units). Scan rate for all: 300 V/s.

carbon surfaces including hydroxyls, carboxylic acids, carbonyls, quinones, and lactones. If the background voltammetric features are due to the presence of such surface functional groups, the associated redox processes are under kinetic control as evidenced by the broad separation between the anodic and cathodic waves. In such circumstances, the pH dependence of these processes indicates the order of sequential proton- and electron-transfer events.21 It was previously determined that the oxidation of catechols in aqueous solution follows the following sequence:27

Reactions that follow this sequence have a greater pH dependence for the anodic process than for the cathodic one.21 However, as demonstrated in Figures 1 and 2, the opposite trend is observed suggesting that the oxidation of surface-bound species at carbonfiber electrodes follows the alternate sequence:

Consistent with the experimental data, digital simulation of the e-H+e-H+ reaction mechanism for a surface-confined species (thin-layer cell assumption) demonstrated a 120 mV/pH dependency for the cathodic peak and no dependence on pH for the anodic peak. This is consistent with results shown in Figures 1 (30) Evans, J. F.; Kuwana, T. J. Electroanal. Chem. 1977, 80, 409.

2786 Analytical Chemistry, Vol. 71, No. 14, July 15, 1999

and 2 over the pH range 5.0 to 7.2. If a surface-confined species is the origin of the surface wave, it must be one that is less acidic than a solution catechol, such as an alcohol or a carbonyl group. Surface Modification. The evidence presented above suggests that an oxide group covalently bonded to the carbon-fiber surface is responsible for the electrode’s sensitivity to changes in pH. The method normally used in this laboratory to polish carbonfiber disk electrodes employs alumina or an embedded diamond dust polishing wheel with water as a lubricant in both cases. During polishing, carbon-carbon bonds are cleaved, and atmospheric oxygen or hydroxide groups from water can react with the dangling carbon bonds. Thus, in these procedures, surface oxides are an anticipated result. In an attempt to identify the chemical nature of the surface group giving rise to the pH effect, specific surface modifications were used. Carbonyl groups were targeted with DNPH and alcohols with DNBCl. Both of these groups would be less acidic than catechols, a feature suggested from the pH dependence of the surface wave. Both nitrocontaining compounds were attached as evident by their reduction at ∼-700 mV. This is apparent in the first forward scan in Figure 3. No oxidation peak is observed on the reverse scan. The second cycle resulted in only background current because the nitro groups have been permanently reduced to hydroxylamines. This detection method proved valid even after voltammetry with dopamine and pH changes. However, neither surface modification resulted in a change in dopamine or in pH responsiveness. Thus, neither hydroxyl nor carbonyl oxide groups appear to be responsible for the pH response or dopamine adsorption.

Figure 5. Background-subtracted cyclic voltammograms showing effect of Nafion on cyclohexane-polished electrodes. (a)10 µM DA and (b) 10 µM acetaminophen at an electrode polished in water (s) and then polished in cyclohexane and Nafion-coated (- - -). Scan rate: 300 V/s.

Polishing in Anaerobic Cyclohexane Removes the pH Sensitivity. It has previously been shown by surface analytical techniques that polishing glassy carbon electrodes in an oxygenfree environment with cyclohexane on dried alumina powder lowers the presence of oxygen on the electrode surface.3 In our laboratory, dramatic changes were found in the cyclic voltammetric responses of carbon-fiber electrodes polished in this manner. As shown in Figure 4, the amplitude of the background current was decreased by approximately 60% on average. Subtracted cyclic voltammograms of 10 µM dopamine and of an acidic pH shift (∆pH ) -0.4 units), obtained both before and after polishing the disk electrode in cyclohexane on dried alumina powder, are also shown in Figure 4. The responses to the pH change and dopamine are reduced by about 80 and 65%, respectively. This polishing procedure also greatly diminished the background features (Figure 4), and this lowers the sensitivity to pH changes. How the removal of oxides from the carbon surface contributes to the reduction in charging current is less clear. When carbon is polished, carbon-carbon bonds are cleaved and reformed. However, because the polishing was done in a reducedoxygen environment, fewer oxides were available to react with the surface. Because hydrophobic surfaces have smaller charging currents than more hydrophilic ones, it may be that removing the hydrophilic surface oxides lowers the charging current. Studies using steady-state amperometry indicate that a reduction

in microscopic surface area is not the likely origin of the decrease in charging current. Adsorption to surface oxides causes the voltammetric signal of cationic catechols to be amplified at high scan rates;31 thus, the decrease in dopamine response following anaerobic polishing is explained by a decrease in available adsorption sites. This view is supported by simulation of a solution-phase HeHe mechanism shown in Figure 4c. The peak amplitudes for the oxidation and reduction in the simulation are only 59.2 and 51.5%, respectively, of the peaks observed for DA at a bare, untreated carbon-fiber electrode; however, the oxidative and reductive peaks for DA at the same electrode after cyclohexane polishing are 39.2 and 43.5%, respectively, of those at the untreated electrode. Nafion Coating of the Oxide-Free Surface. Coating the electrode with a cation-exchange membrane protects the electrode and restores the current response lost as a result of cyclohexane polishing. Previously, Nafion, a cation-exchange polymer, has been coated onto carbon microelectrodes to screen out or block endogenous anionic compounds such as ascorbic acid and DOPAC, a dopamine metabolite. Nafion also increases the amplitude of the dopamine oxidation wave by preconcentrating it near the electrode surface. With Nafion, the anodic peak current for dopamine was increased to 70% of its level before polishing in anaerobic cyclohexane (Figure 5a). The Nafion treatment also increased the electrode’s response to the pH change. Nonetheless, the desired effects were maintained, as the anodic current for the pH response was only 28% of the original signal, an increase of only 8% from the value obtained after cyclohexane polishing. The ratio of the background-subtracted current responses for 10 µM dopamine to that for ∆pH ) -0.4 units at the peak potential for dopamine is 4.81 ( 0.43 at untreated electrodes (n ) 4) and is 10.00 ( 0.29 at electrodes polished in cyclohexane and Nafioncoated (n ) 3). Acetaminophen, a compound often used as an analgesic, is also readily oxidized at carbon electrodes.32 Its local in vivo concentration can be monitored with electrochemical techniques. Although it is neutral at physiological pH, it can also penetrate through Nafion. Like dopamine, the amplitude of its oxidation wave is diminished by the combined anaerobic-polishing and Nafioncoating procedure. It can also be observed in Figure 5 that the presence of Nafion on the electrode surface causes a shift in the position of the peaks for oxidation and reduction processes. As reported previously,33 the pH inside a Nafion film immersed in a solution of pH 7.4 is 6.52. Thus, the more acidic nature inside the Nafion film causes a shift in redox potentials.21 Effect of Cyclohexane Polishing on Other Compounds. The effects of polishing carbon-fiber electrodes in cyclohexane were investigated for the oxidation of 4-methyl catechol and DOPAC and for the reduction of ruthenium hexaamine and ferricyanide. The background-subtracted cyclic voltammograms for each of these compounds both before and after a microelectrode was cyclohexane-polished are shown in Figure 6. The cyclohexane polishing has the effect of reducing the electrode’s response to all four of the electroactive species, and for all (31) Allred, C. P.; McCreery, R. L. Anal. Chem. 1992, 64, 444. (32) Miner, D. J.; Rice, J. R.; Riggin, R. M.; Kissinger, P. T. Anal. Chem. 1981, 53, 2258. (33) Ciolkowski, E. L.; Cooper, B. R.; Jankowski, J. A.; Jorgenson, J. W.; Wightman, R. M. J. Am. Chem. Soc. 1992, 114, 2815.

Analytical Chemistry, Vol. 71, No. 14, July 15, 1999

2787

Figure 6. Background-subtracted cyclic voltammograms of other redox systems before (s) and after (- - -) polishing in cyclohexane. (a) 100 µM 4MC; (b) 200 µM DOPAC; (c) 500 µM Ru(NH3)63+/2+; (d) 500 µM Fe(CN)63-/4-.

compounds except ruthenium hexaamine, increasing the separation of the peaks. Like dopamine at physiological pH, Ru(NH3)63+/2+ is cationic, and its adsorption to the electrode’s surfacesapparent at high scan ratessappears to be due to the oxide functionalities. Hence, removal of these electron-rich groups reduces the amount of Ru(NH3)63+/2+ adsorbed at the surface, leading to lower current values. However, none of the other species studied are cationic at pH 7.4, and therefore, the observed decrease in current responses is more difficult to explain. In fact, we anticipated that removal of oxide groups would lead to a more favorable approach to the electrode’s surface for anionic compounds such as Fe(CN)63-/4- and DOPAC, leading to an increase in redox current. However, the opposite is observed. After cyclohexane polishing, a surface has been created which, not only lacks fixed sites of negative charge, but may have fewer extra electron-exchange sites.7,19 This could explain why redox currents for anionic and uncharged species are decreased as well as why the current amplitudes of the cationic species are decreased to the extent that they are.7,19 Furthermore, the peak separations have increased after cyclohexane polishing, consistent with slower electron transfer for all of the compounds examined except for Ru(NH3)63+/2+. In agreement with our results, the electron-transfer kinetics of Ru(NH3)63+/2+ have been previously shown to have no dependence on the presence of surface oxides, a feature attributed to its outer-sphere mechanism of electron transfer.3 However, though shown to be directly independent of carbonyl and hydroxyl functionalities, electron transfer for Fe(CN)63-/4- does exhibit sensitivity to changes in the surface chemistry of the electrode.3 Our observations further support these findings. Use of Carbon-Fiber Microelectrodes In Vivo. To test the in vivo performance of the cyclohexane-polished electrodes, we chose to monitor electrochemical changes following intraperito2788 Analytical Chemistry, Vol. 71, No. 14, July 15, 1999

neal injections of acetaminophen. In preliminary studies, we found that with repetitive cycling of the electrode, the decreased sensitivity to pH was maintained, and the response to acetaminophen was also unaltered. Acetaminophen is of interest in this application for several reasons. First, because it is electroactive and permeates Nafion as demonstrated above, its concentration in the extracellular space can be monitored with carbon-fiber electrodes.32,34 Second, because it inhibits prostaglandin synthesis, it alters blood flow.35 Because of the change in blood flow, large pH excursions are anticipated in the extracellular space. Initial experiments were with Nafion-coated electrodes polished in aqueous media in the normal fashion. Acetaminophen (75 mg/ kg) was administered to an anesthetized rat, and cyclic voltammetry was used to monitor changes in the caudate nucleus of its brain. Approximately 60 min after its administration, maximal changes in the current signal were observed. However, subtraction of a voltammogram obtained before drug administration from one obtained at the time of maximal response did not yield a simple voltammogram representative of acetaminophen. Rather, the cyclic voltammogram appeared to indicate an increase in acetaminophen accompanied by an acidic pH change. When the changes were measured with a carbon-fiber electrode polished in cyclohexane and coated with Nafion, the subtracted voltammogram obtained at the maximal response time closely resembled that obtained for acetaminophen during the calibration after the in vivo experiment. To quantify these different responses, the difference current at the potential where a maximal pH response was obtained was measured and compared with that obtained at the peak potential for acetaminophen oxidation. When expressed as a ratio of pH to acetaminophen currents, the values were 0.78 ( 0.14 for (34) Sabol, K. E.; Freed, C. R. J. Neurosci. Methods 1988, 24, 163. (35) Flower, R. J.; Vane, J. R. Nature (London) 1972, 240, 410.

electrodes polished in aqueous solution and 0.33 ( 0.06 for electrodes polished in cyclohexane (results from three animals for each condition). Thus, the electrode treatment clearly provides a less distorted signal, and importantly, provides information on the molecule of interest. The procedure outlined earlier36 for removing the effect of voltammetric interferences on the temporal changes can be used more easily to evaluate the acetaminophen concentration. CONCLUSIONS This report has shown that the shifting of voltammetric background features of carbon-fiber microelectrodes with variations in pH leads to changes in background-subtracted voltammetric responses. These voltammetric shifts occur in a manner that is consistent with an e-H+e-H+ transfer reaction mechanism. This suggests that the response to changes in pH is due to surface(36) Michael, D.; Travis, E. R.; Wightman, R. M.; Anal. Chem. 1998, 70, 586A.

bound oxide functionalities that are less acidic than solution catechols. Results obtained after surface modifications with DNBCl and DNPH eliminate hydroxyl and carbonyl groups, respectively, as contributors to the pH sensitivity. Nonetheless, removing surface oxides by polishing the electrode in a reduced-oxygen environment significantly reduces the pH sensitivity of the electrode. Losses in dopamine responsiveness caused by this procedure are restored with the use of Nafion without jeopardizing the desired reduction in pH sensitivity. Carbon-fiber microelectrodes treated in this manner have proven useful for in vivo studies where fluctuations in pH are common. ACKNOWLEDGMENT This research was supported by NIH (NINDS and NIDA). Received for review November 18, 1998. Accepted April 14, 1999. AC981279T

Analytical Chemistry, Vol. 71, No. 14, July 15, 1999

2789