Anal. Chem. 2000, 72, 1576-1584
Nanostructured Carbon Fiber Disk Electrodes for Sensitive Determinations of Adenosine and Uric Acid Anna Brajter-Toth,* Kholoud Abou El-Nour,† Eder T. Cavalheiro,‡ and Roberto Bravo§
Department of Chemistry, University of Florida, Gainesville, Florida 32611-7200
Nanostructured carbon fiber microdisk electrodes were prepared by a combination of mechanical polishing and electrolytic treatment, where the latter involved moderate oxidation of the surface, followed by a reduction. A high density of surface defects contributed to a high capacitance of the nanostructured electrodes. Facilitated proton transfer was observed at the nanostructured surface and was associated with cation-exchanged oxide defects. The nanostructured surfaces intercalated uric acid and adenosine and engaged in fast electron/proton transfer in the oxidation of both analytes. As a result, electrolytic treatment followed by fast-scan voltammetry determinations led to a sensitive response to both analytes in physiological buffers. The nanostructured electrodes showed remarkable stability and could be easily regenerated and reused. With long use, electrode activity decreased. Kinetic discrimination of the surface-mediated reaction of ascorbate was achieved at high scan rates. We report here novel surface properties of carbon fiber microdisk electrodes, obtained by polishing and electrolytic activation of the fiber surface,1 that were prepared to optimize the response of uric acid and the pentose sugar-containing purine nucleoside adenosine, which participate in in vivo energy metabolism.2 The electrodes were prepared by moderate electrochemical oxidation of the surface, followed by reduction, using physiological buffers as electrolytes.1 To characterize surface properties of the electrodes, the electrode response of structurally different electrochemical probes, Fe(CN)63-/4-, Ru(NH)63+/2+ and ascorbate, was tested, and the surface was further characterized by measuring the electrode capacitance.1 Electrochemical treatment of carbon fiber electrodes has led to practical applications of the treated electrodes in bioanalysis.3-5 A successful example is the development of carbon fiber elec* Author to whom correspondence should be addressed: (e-mail) atoth@ chem.ufl.edu. † Permanent address: Department of Chemistry, University of Suez Canal, Ismailia, Egypt. ‡ Permanent address: Department of Chemistry, University of Sao Carlos, Sao Carlos, Brazil. § Center for Disease Control and Prevention, Building 17, Mail Stop F-17, 4770 Buford Highway, NE, Atlanta, GA 30341-3724. (1) Bravo, R.; Brajter-Toth, A. Chem. Anal. (Warsaw) 1999, 44, 423-436. (2) Chen, W.; Hoerter, J.; Gueron, M. J. Mol. Cell. Cardiol. 1996, 28, 21632174. (3) Wightman, R. M.; May, L. J.; Michael, A. C. Anal. Chem. 1988, 60, 769A779A.
1576 Analytical Chemistry, Vol. 72, No. 7, April 1, 2000
trodes that are sensitive to neurotransmitters such as dopamine.3-5 Applications of carbon fiber electrodes to in vivo detection of neurotransmitter release in the brain, and from single cells such as chromaffin cells, have led to significant advances in the understanding of the process of neurotransmission. Because of an incomplete understanding of structure-activity relations at graphite, the methods of electrolytic treatment for in situ activation of carbon electrodes have remained empirical.5 There are problems with a more systematic approach to electrolytic activation of carbon electrodes because carbon materials are structurally complex, as are products of carbon oxidation. High activity of carbon electrodes has been associated in part with electrolytic oxidation of the surface.5 Using this approach, methods of treatment of carbon fiber electrodes developed for detection of neurotransmitters were designed for high sensitivity, mostly in single-shot determinations,3,4 and the treatment also aimed at limiting the response of electroactive interferences such as ascorbate. To facilitate and simplify characterization of electrolytically treated carbon surfaces, focus has been placed on the characterization of highly ordered pyrolytic graphite (HOPG) electrodes, which are structurally well-defined.5,6 With combined use of surface techniques, including atomic force microscopy (AFM), X-ray microprobe, Auger surface analysis, and Raman spectroscopy, HOPG surfaces have been characterized during and after electrochemical oxidation.6 The results reveal a significant agreement with those obtained after similar treatment of structurally less well-defined glassy carbon (GC) electrodes.7 This is important in view of the structural similarity of carbon fiber to HOPG and to GC.5 However, a method of surface activation that is generally accepted remains elusive.5 The present investigation, of a new approach to the activation of carbon fiber electrodes, was initiated to optimize determinations of biological purines such as adenosine, which are much less reactive than many previously studied biological analytes, such as neurotransmitters.8-10 Low reactivity to electrooxidation requires more extreme positive potentials for the determinations (4) Michael, D.; Travis, E. R.; Wightman, R. M. Anal. Chem., 1998, 70, 587A592A. (5) McCreery, R. L. In Electroanalytical Chemistry; Bard, A. J., Ed.; Dekker: New York, 1991; pp 271-374. (6) Goss, C. A.; Brumfield, J. C.; Irene, E. A.; Murray, R. W. Anal. Chem. 1993, 65, 1378-1389. (7) Freund, M. S.; Brajter-Toth, A.; Cotton, T. M.; Henderson, E. R. Anal. Chem. 1991, 63, 1047-1049. (8) Chen, T.; Strein, T.; Ewing, A. Electroanalysis 1994, 6, 746-750. 10.1021/ac9906680 CCC: $19.00
© 2000 American Chemical Society Published on Web 02/26/2000
of analytes such as adenosine, where significant oxidation of graphite can occur, and where water oxidation starts to interfere with the determinations.5,6 Generally, the more positive potentials have been avoided in bioanalysis11 to prevent empirically observed instability and inactivation of graphite, which has been associated with irreversible oxidation of the graphite surface, occurring simultaneously with the oxidation of water. Our second goal was to integrate surface treatment/activation with the acquisition of analytical signals to simplify analysis.12 Third, we wished to characterize the analytically useful surfaces, using a combination of diagnostic measurements. Our aim was to provide a more systematic approach to electrolytic treatment, through a correlation of the analytical response with surface properties, such as the density of surface defects and amorphous surface character, which have been determined previously for similarly treated HOPG5,6 and for GC7,13 electrodes by surface methods. The results show that highly active carbon fiber electrodes can be fabricated for the targeted determinations1,14 by methods that minimize irreversible oxidation of the carbon fiber surface6 and that produce a nanostructured surface.6,15 The results reveal a remarkable degree of chemical stability of the nanostructured electrodes. The surface intercalates analytes,15 and the surface groups that are formed appear to engage in efficient proton transfer.16,17 EXPERIMENTAL SECTION Chemicals. Ru(NH3)6Cl3 was obtained from Johnson Matthey (Alfa Products, Ward Hill, MA). All other chemicals were obtained from Fisher (Fisher, Pittsburgh, PA) and Sigma (Sigma, St. Louis, MO). Analyte solutions were prepared before each experiment. Compositions of electrolytes and buffers (in mM) that were used in electrolytic treatment and in analytical determinations of uric acid and adenosine were as follows: 100 KCl; 12.5 tris(hydroxymethyl)aminomethane hydrochloride (Tris-HCl), pKa ) 8.08; Krebs-Henseleit, 117.9 NaCl, 4.8 KCl, 2.5 CaCl2‚2H2O, 1.18 MgSO4‚7H2O, 1.2 KHPO4, 0.5 Na2EDTA‚2H2O, 0.14 ascorbic acid, 5.5 glucose, 25.0 NaHCO3, 2.0 pyruvic acid; 70 NaH2PO4/Na2HPO4; Krebs, 126.0 NaCl, 2.5 KCl, 2.4 CaCl2‚2H2O, 20.0 HEPES, 1.2 NaH2PO4, 0.5 Na2EDTA‚2H2O, 0.14 ascorbic acid, 5.5 glucose, 25.0 NaHCO3. Unless otherwise specified, buffer pH was 7.4. Buffer pH was adjusted with HCl or NaOH. The determinations were performed at room temperature. Electrodes. A saturated calomel electrode (SCE) was used as a reference electrode. Carbon fiber of ∼7 µm diameter (Textron Specialty Materials, Lowell, MA) was the working electrode. Fabrication of carbon fiber microdisk electrodes has been de(9) Spurlock, L. D.; Jaramillo, A.; Prserthdam, A.; Lewis, J.; Brajter-Toth, A. Anal. Chim. Acta 1996, 336, 37-46. (10) Singhal, P.; Kuhr, W. G. Anal. Chem. 1997, 69, 4828-4832. (11) Pihel, K.; Hsieh, S.; Jorgensen, J. W.; Wightman, R. M. Anal. Chem. 1995, 67, 4514-4521. (12) Hsueh, C. C.; Bravo, R.; Jaramillo, A. J.; Brajter-Toth, A. Anal. Chim. Acta 1997, 349, 67-76. (13) Bodalbhai, L.; Brajter-Toth, A. Anal. Chim. Acta 1990, 231, 191-201. (14) Bravo, R.; Hsueh, C. C.; Jaramillo, A.; Brajter-Toth, A. Analyst 1998, 123, 1625-1630. (15) Besenhard, J. O.; Fritz, H. P. Angew. Chem., Int. Ed. Engl. 1983, 22, 950975. (16) Thorp, H. M. J. Chem. Educ. 1992, 69, 250-252. (17) Snider, K. E.; Merzbacher, C. I.; Hagaus, P. L.; Rolison, D. R. Chem. Mater. 1997, 9, 1248-1255.
scribed.1,14 Briefly, a single carbon fiber was connected to a copper wire with silver epoxy (EPO-TEK 410E, Epoxy Technology, Billerica, MA). After the epoxy dried (∼24 h at room temperature), the fiber was sealed in a micropipet tip with Shell epoxy (Shell Epon 828, Miller-Stephenson Chemical, Danbury, CT), with methylenediamine (Miller-Stephenson Chemical) as a hardener (12 wt %). The epoxy was first dried overnight at room temperature and was next cured in an oven for 1 h at 150 °C. After curing, the excess epoxy was removed from the disk surface by polishing on 600-grit SiC paper (Mark V Laboratory, East Granby, CT) for 1 min, using a polishing wheel (Ecomet I, Buehler Laboratory, Evanston, IL). The response of polished electrodes was tested by voltammetry in 10 mM Fe(CN)63- in 0.5 M KCl (pH 3) at a scan rate of 50 mV/s. Electrodes that did not respond were repolished; electrodes that did not respond after repolishing were discarded. In addition, electrodes with high initial capacitance were discarded. Disk electrodes were next polished with a γ-Al2O3 suspension of 0.1-µm particle size (Gamal, Fisher Scientific, Pittsburgh, PA) for 5-10 s on a polishing cloth (Mark V Laboratory, East Granby, CT), using a polishing wheel (Ecomet I, Beuhler Laboratory). Polished electrodes were rinsed with water, dipped in 2-propanol18 for 1-15 min, and were finally sonicated in doubly distilled water. Finally, the polished electrodes were treated electrolytically, by continuous potential cycling from -1.0 to +1.5 V, except in HCl and Krebs, -1.0 to +1.2 V, for 30 min at 10 V/s.14 A continuous triangular waveform was used for the treatment. Buffer that was used in analytical determinations was used in electrolytic treatment. Determination of the apparent radius of microdisk electrodes has been described.14 Microdisk electrode radii were determined from cyclic voltammetry (50 mV/s), from a limiting current of 10 mM Fe(CN)63- in 0.5 M KCl (pH 3). A disk equivalent area was used to calculate electrode radius, using a diffusion coefficient of Fe(CN)63-, Do ) 7.7 × 10 -6 cm2 s-1.19 Instrumentation. Steady-state voltammograms were obtained with a BAS-100, equipped with a home-built preamplifier,20 interfaced to a PC. Fast-scan voltammetry has been described previously.12,20-22 Fast-scan voltammetry experiments were performed at 500 V/s unless specified. A continuous triangular waveform was applied to the SCE reference electrode, using a function generator (Universal Programmer, model 175, EG& G Princeton, NJ), in a two-electrode system.20 A home-built current transducer20 measured the current at the disk electrode in fastscan voltammetry. A copper mesh Faraday cage was used to minimize noise. The current was converted to voltage, amplified, and recorded and stored by a digital oscilloscope (LeCroy model 9310, Chestnut Ridge, NY). The waveform applied to the SCE was recorded at a separate channel of the oscilloscope. Stored data were transferred from the oscilloscope to a PC for display and analysis. (18) Cahill, S. P.; Walker, W. D.; Finnegan, M. J.; Mickelson, E. G.; Travis, R. E.; Wightman, R. M. Anal. Chem. 1996, 68, 3180-3186. (19) Stackelberg, M.; Pilgram, M.; Toome, V. Z. Electrochim. 1953, 57, 342347. (20) Hsueh, C. C.; Brajter-Toth, A. Anal. Chim. Acta 1996, 321, 209-214. (21) Millar, J.; Armstrong-James, M.; Kruk, Z. L. Brain Res. 1981, 205, 418424. (22) Baur, J. E.; Kristensen, E. W.; May, L. J.; Wiedemann, D. J.; Wightman, R. M. Anal. Chem. 1988, 60, 1268-1272.
Analytical Chemistry, Vol. 72, No. 7, April 1, 2000
1577
A syringe was used to inject all solutions into a ∼80-µL electrochemical cell12 which was used in fast-scan voltammetric experiments, and the injections allowed the solutions to be pumped into the cell to maintain permanent contact between the solution and the electrode. First background current was recorded in buffer and next a solution of analyte, in the same buffer, was injected into the cell, and the analytical signal was recorded. This sequential procedure facilitated recording of accurate background currents and facilitated accurate background subtraction.12 To minimize noise, signal averaging with digital processing of the data was used.12,14 A total of 250 scans were recorded in buffer only, and then in the analyte solution, and were averaged and stored for digital background subtraction.12,14 The potential windows and scan rates were the same in background measurements in buffer and analytical measurements. The number of scans (cycles) recorded was a compromise which aimed to achieve a high signal-to-noise ratio in a short time, with limited number of cycles.12 Determination of Electrode Capacitance. Apparent electrode capacitance, Cobs (F cm-2) was determined from background current measured by voltammetry at a scan rate of 10 V/s at 0.75 V vs SCE. Background current was recorded in buffer in the absence of analyte in solution, in a potential window of -1.0 to +1.5 V, except in Tris-HCl and Krebs, -1.0 to +1.2 V.1 It was assumed that at 0.75 V voltammetric background current was due to double-layer charging only, with minimum contributions from surface faradaic reactions. The apparent capacitance was normalized by electrode area, using the disk radius determined from voltammetry. Analytical Determinations. Sensitivity values that are reported represent an average slope value of at least 10 independent determinations, at 10 different electrodes. The slopes were obtained from calibration plots obtained at at least five concentrations, with at least four determinations at each concentration. Error analysis was based on propagation of error, and the error is reported as the final uncertainty of the slope (sensitivity). Limit of detection (LOD) for uric acid was determined for a signal-tonoise ratio of 3. Noise was defined as a peak-to-peak noise, rather than peak-to-peak noise over 4 (95% confidence level). RESULTS AND DISCUSSION Oxidation/Reduction Voltammetry of the Carbon Fiber Electrode Surface. The primary goal of this work was to produce stable and active carbon fiber electrodes for the determination of physiological purines, under physiological conditions.9 Electrolytic treatment/activation of carbon fiber electrodes in physiological buffers, at pH 7.4, was therefore evaluated to avoid the need for reequilibration of electrodes during voltammetric determinations in physiological media, at pH 7.4.1 Generally, electrolytically treated carbon fiber electrodes showed a response that was independent of the composition of buffers that were used in electrode treatment; this is discussed further below. AFM analysis has shown that surface features of graphite that has been electrolytically treated in phosphate buffer (pH 7) are similar to (23) Kinoshita, K. Carbon: Electrochemical and Physicochemical Properties; Wiley: New York, 1988; Chapter 6, pp 210-211. (24) Beck. F.; Junge, H.; Krohn, H. Electrochim. Acta 1981, 26, 799-809. (25) Maeda, Y.; Okemoto, Y.; Inagaki, M. J. Electrochem. Soc. 1985, 132, 23692372.
1578 Analytical Chemistry, Vol. 72, No. 7, April 1, 2000
Figure 1. Background current at carbon fiber microdisk electrode before (solid line) and after (small dash) electrolytic treatment: ∼7-µm electrode diameter, scan rate 10 V/s, and 70 mM pH 7.4 phosphate buffer. Background current after long use is shown by a large dash.
those observed after electrolytic treatment in KNO3.7 Generally, anodic behavior of carbon depends on the composition, concentration, and pH of the electrolyte.15,23-25 Because of differences in graphite structure and electrolyte composition, it is difficult to predict conditions that minimize irreversible side reactions during electrolytic activation of carbon fiber electrodes in different electrolytes. Irreversible side reactions that occur during electrolytic activation of carbon can contribute to irreversible behavior of carbon electrodes because these reactions lead to the formation of nonconducting surface oxides, to electrolysis of water, and to production of carbon dioxide and oxygen.6,15 To minimize irreversible side reactions that can occur during electrolytic treatment of carbon, and to integrate treatment and reactivation of the electrode surface with the acquisition of analytical signals, carbon fiber electrodes were treated by continuous cycling of the electrode potential, at a rate of 10 V/s, in a potential window of +1.5 to -1.0 V vs SCE, which involved oxidation of the electrode surface, followed by a reduction, until a steady-state response was achieved.1,12 Average background currents of electrodes, before and after electrolytic treatment, are shown in Figure 1. After electrolytic treatment, several features can be identified in the background response of the carbon fiber surface. First, as indicated by an increase in the background current at ∼1.5 V, the fiber surface becomes oxidized after voltammetric treatment in the potential window of +1.5 to -1.0 V (small dash).6,15 Second, the mostly amorphous character of the surface is indicated by the featureless shape of the background current at positive potentials, which lacks peaks that are typically associated with oxidation of more ordered crystalline graphite.6,15 The method of treatment of the surface,
Table 1. Ru(NH3)63+ and Fe(CN)63- at the Carbon Fiber Electrode untreated pH 1.11 3.11 6.08 2.87 5.59
analytea,b
E,0′c mV
Fe(CN)63289 ( 1 Fe(CN)63160 ( 5 Fe(CN)63137 ( 8 Ru(NH3)63+ -246 ( 1 Ru(NH3)63+ -242 ( 1
slope,c
treated
E vs log(iI - i)/i
E,0′c mV
slope,c E vs log(iI - i)/i
69 ( 5 79 ( 3 100 ( 2 69 ( 1 64 ( 1
290 ( 3 171 ( 3 166 ( 4 -244 ( 3 -251 ( 1
68 ( 5 66 ( 2 66 ( 5 71 ( 2 72 ( 1
a 1.97 mM Ru(NH ) 3+, 0.5 M KCl, pH adjusted with HCl, 50 mV 7 3 6 s-1. b 1.675 mM Fe(CN)63-, 0.5 M KCl, pH adjusted with HCl, 50 mV s-1. c Three electrodes and four determinations for each electrode were used for calculations (n ) 12).
Figure 2. Voltammetry of ferricyanide at electrolytically treated (small dash) and untreated (solid line) carbon fiber electrode: electrode radius r ) 3.23 ( 0.11 µm, 2.2 mM Fe(CN)63-, 0.5 M KCl pH 6, and voltammetric scan rate 50 mV/s.
which includes polishing,1 roughens and may oxidize the surface, contributing to the amorphous surface character. Although oxidation of the carbon fiber electrode surface after electrolytic treatment is apparent from the increase in the background current at ∼1.5 V, shown in Figure 1, the relatively small oxidation current at positive potentials indicates a lack of extensive oxidation of the surface6,15 as a result of treatment by potential cycling, which involves relatively fast surface oxidation and reduction. An increase in background current after electrolytic treatment, in the entire window of potential cycling of +1.5 to -1.0 V, indicates an increase in electrode capacitance; this is discussed further below. Fe(CN)63-/4- and Ru(NH3)63+/2+ as Electrochemical Probes of Activity and Structure of Carbon Fiber Surface. Figure 2 shows voltammetric curves of Fe(CN)63-/4- that were obtained at a carbon fiber electrode before (solid line) and after (small dash) electrolytic treatment of the electrode surface. Several features of the steady-state voltammograms of ferricyanide in Figure 2 are of interest. First, there is a visible change in the shape of the voltammetric curve of ferricyanide: a gradual rise in current and a poorly defined plateau, observed before treatment (solid line), are replaced by a steep current increase and a flat well-defined plateau (small dash). The changes in the shape of the i-E curve of ferricyanide in the rising part reflect changes at the carbon fiber surface from an initially rough surface,26 formed by polishing, to a more uniform and more active surface formed by electrolytic treatment. High activity of carbon fiber electrodes after electrolytic treatment is confirmed by analysis of the i-E curves of ferricya(26) Zoski, C. G.; Bond, A. M.; Allison, E. T.; Oldham, K. B. Anal. Chem. 1990, 62, 37-45.
nide in Figure 2. The results of this analysis are summarized in Table 1 and show that at pH 6.08 after treatment of the electrode surface the slope of E vs log il - i/i plots27 of i-E curves obtained for ferricyanide is ∼60 mV as expected for a fast electrode reaction.27 This slope value has been obtained for ferricyanide at active carbon electrodes.5 In addition, after electrolytic treatment of the carbon fiber electrode surface, separation between the forward and reverse current traces in i-E curves of ferricyanide in Figure 2 increases. This can clearly be seen in the plateau region and at the foothill of the i-E curve in Figure 2 where double-layer charging controls the current. Increased separation of the forward and reverse current traces after electrolytic treatment indicates that the electrode capacitance of the treated electrode increases.28 The increase in capacitance, shown by the results in Figure 2 obtained by slow-scan voltammetry, is in agreement with results in Figure 1, where a significant increase in capacitance was detected by fast-scan voltammetry from an increase in background current in the entire window of potential cycling. Since electrolytic treatment of carbon fiber can lead to the formation of surface defects, during oxidation and reduction of the surface during treatment,1 high capacitance of surface defects15 can account for the observed increase in electrode capacitance. In addition, defects formed by electrolytic treatment can contribute to improved electrode kinetics, which are apparent from lower slope values of semilogarithmic plots of i-E curves of ferricyanide after electrolytic treatment (Table 1). The essentially unchanged magnitude of the limiting current of ferricyanide before and after electrolytic treatment of the fiber surface, measured at the plateau of i-E curves in Figure 2 at -0.2 V, verifies an unchanged electrode area (approximately the radius), as seen by the negatively charged ferricyanide.29-31 Thus, formation of surface defects during electrolytic treatment, under the conditions used here, limits oxidation of the surface and formation of negatively charged15 passivating nonconducting surface oxides which can decrease the current of anions such as ferricyanide.29-31 The low density of nonconducting oxides at a treated electrode surface is also evident from the relatively small (27) Bard, A. J.; Faulkner, L. R. Electrochemical Methods Fundamental and Applications; Wiley: New York, 1980; pp 222-223. (28) Green, S. J.; Stokes, J. J.; Hostetler, M. J.; Pietron, J.; Murray, R. W. J. Phys. Chem. 1997, 2663-2668. (29) Cheng, Q.; Brajter-Toth, A. Anal. Chem. 1992, 64, 1998-2000. (30) Cheng, Q.; Brajter-Toth, A. Anal. Chem. 1995, 67, 2767-2775. (31) Cheng, Q.; Brajter-Toth, A. Anal. Chem. 1996, 68, 4180-4185.
Analytical Chemistry, Vol. 72, No. 7, April 1, 2000
1579
Table 2. Electrode Parameters capacitance, µF cm-2 supporting electrolytea
beforeb ECP
afterb ECP
ib,c nA
∆Ep mV
UAd sensitivity, pA µM-1
LOD,e µM
KCl Tris-HCL Krebs-Henseleit sodium phosphate Krebs
1304 550 1412 772
1797 1975 2227 2091
25 25 18 10
258 ( 32 628 ( 16 583 ( 21 307 ( 35 680 ( 45
78 ( 5 38 ( 2 346 ( 2 440 ( 2 57 ( 9
25 20 10 5 15
a For composition, see Experimental Section. b ECP, electrolytic treatment. c Background current at 0.25 V, 10 V s-1, 50 cycles, after ECP electrolytic treatment. d UA, uric acid. From the slope of the calibration curve, 500 V s-1. At least five points for each calibration curve and four determinations for each point. Number of electrodes (n) used g4 for sodium phosphate and Krebs-Henseleit, n e 3 for KCL, Tris-HCL, and Krebs. e Limit of detection (LOD) when S/N ) 3.
background current in Figure 1, at positive potentials around 1.5 V, which is a measure of surface oxidation after treatment. Effect of Solution pH on Electrode Response. Results that show the effect of solution pH on electrode kinetics of ferricyanide are summarized in Table 1, and these results provide additional insights into the surface structure of carbon fiber electrodes. Before electrolytic treatment of carbon fiber electrodes, electrode kinetics of ferricyanide depend on solution pH and are slow at high pH as indicated by a 100 mV slope of E vs log il - i/i plot of i-E curves at pH 6.08. Electrode kinetics of ferricyanide are faster (at untreated carbon fiber electrodes) at lower solution pH, as shown by a lower slope of E vs log il - i/i of 69 mV at pH 1.11 (Table 1). This observation is consistent with an inner-sphere electrode reaction of ferricyanide.5 Faster kinetics of ferricyanide at low pH point to the presence of surface functional groups at the untreated surface which can be protonated at low solution pH, which creates additional surface sites for electrode reaction of the ferricyanide anion and improves the apparent electrode kinetics.5,29-31 After electrolytic treatment of the carbon fiber electrode surface, electrode kinetics of ferricyanide are fast at high and low solution pH, as indicated by an approximately 60-mV slope of E vs log il - i/i plot of i-E curves (Table 1). After the electrolytic treatment, electrode kinetics of ferricyanide become independent of solution pH. The results at different pH indicate that after electrolytic treatment new surface sites form at the electrode surface, which can facilitate the electrode reaction of ferricyanide at high as well as low solution pH and are consistent with the formation and reported stoichiometry of oxide defects (C10O1.4K0.66) which have been identified at moderately oxidized carbon.6 These defects differ from nonconducting surface oxides, which are present at heavily oxidized graphite surfaces and may be present at an untreated carbon fiber electrode surface;15 the latter can slow electrode reactions of anions at high solution pH.5,12 Results of analysis of Ru(NH3)63+/2+ voltammetric curves obtained before and after electrolytic treatment of carbon fiber electrodes are also summarized in Table 1. These results support the results obtained for ferricyanide. Before electrolytic treatment, electrode kinetics of Ru(NH3)63+/2+ cation are a little slower at low than high solution pH as indicated by higher slope values at lower pH (Table 1). This is expected if surface functional groups are protonated at low pH, which can interfere with the electrode reaction of the Ru(NH3)63+/2+ cation.5,29-31 1580 Analytical Chemistry, Vol. 72, No. 7, April 1, 2000
Before electrolytic treatment, the sensitivity of the electrode reaction of Ru(NH3)63+/2+ to large changes in solution pH is lower than that of ferricyanide as indicated by smaller changes in slopes of semilog plots of Ru(NH3)63+/2+ in Table 1. This difference in behavior of Ru(NH3)63+/2+ and ferricyanide is consistent with faster electrode kinetics of outer-sphere Ru(NH3)63+/2+.5 After electrolytic treatment of the carbon fiber electrode, electrode kinetics of ferricyanide as well as of Ru(NH3)63+/2+ become independent of the solution pH. However, at high pH, electrode kinetics of Ru(NH3)63+/2+ are a little slower than before electrolytic treatment. This indicates that after treatment surface groups that are formed are cation-exchanged at both high and low solution pH, slowing down the electrode reaction of Ru(NH3)63+/2+ cation both at high and low solution pH.29-31 The stoichiometry of oxide defects (C10O1.4K0.66) that form at moderately oxidized carbon6 supports this conclusion, which is also supported by the already discussed small increase in background current after electrolytic treatment which is associated with surface electrooxidation and formation of surface oxide groups. Relatively fast apparent kinetics of both probes at carbon fiber electrodes after electrolytic treatment are consistent with a fraction of carbon fiber electrode area covered by oxide defects, of less than 1.32,33 Electrode Capacitance. An increase in the electrode capacitance is expected after electrolytic treatment of polished carbon fiber electrodes if defects, which have high capacitance, form as a result of treatment.15 Electrodes polished on 600-grit SiC paper show a relatively high capacitance before treatment, as expected for rough electrodes; electrode capacitance of polished electrodes increases after electrolytic treatment (Table 2). Table 2 summarizes values of electrode capacitance that were determined in different buffers, after electrolytic treatment in these buffers. Capacitance values that were determined in different buffers at 0.75 V at pH 7.4 are similar (Table 2). This is consistent with an earlier observation of structural similarity of graphite surfaces produced by electrolytic treatment in different buffers. Capacitance values that were determined at carbon fiber electrodes after electrolytic treatment in different buffers depend on the potential window used in electrolytic treatment. Capacitance values increase when the positive potential limit of electrolytic treatment increases,1 as expected,5 but the values also increase when the negative potential limit of the treatment increases from (32) Amatore, C.; Saveant, J. M.; Tessier, D. J. Electroanal. Chem. 1983, 147, 39-53. (33) Freund, M. S.; Brajter-Toth, A. J. Electroanal. Chem. 1991, 300, 347-363.
-0.8 to -1.0 V.1 The increase in electrode capacitance with an increase in the negative potential limit of electrolytic treatment may be a result of more efficient reduction of surface groups, such as oxides, during electrolytic treatment at more negative potentials. Oxides form initially as a result of surface oxidation during electrode treatment at positive potentials, and the reduction of graphite oxides is a slow process.15 The increase in carbon fiber electrode capacitance with an increase in the negative potential limit of electrolytic treatment may also be a consequence of increased exposure of surface defects at the electrode surface. This may result from reactions of aqueous buffers at the electrode, which lead to the evolution of hydrogen and to the generation of hydroxide and which have been shown to accompany reduction of graphite surfaces at negative potentials.34 Electrode Stability. Stability of carbon fiber electrodes treated by potential cycling at a rate of 10 V/s in a potential window from +1.5 to -1.0 V was verified by measuring the background current of the electrode in buffer after the treated electrode was exposed to air for 5-15 min. The background current of the treated electrodes remains unchanged after an electrode is exposed to air for different periods of time up to 15 min.1 After long use, which involves many determinations in a potential window of -1.0 to +1.5 V, the electrode capacitance of the electrolytically treated carbon fiber electrodes increases in the entire potential window of cycling, as shown in Figure 1 (large dash). This increase in capacitance is accompanied by a decrease in the apparent radius of the disk surface of the electrode that is detected as a lower limiting current of ferricyanide (at pH 7.4, not shown). The increase in capacitance and decrease in limiting current of ferricyanide, after many determinations, must be a result of surface changes. These changes can result from irreversible reactions at the electrode surface during use. Since the limiting current of ferricyanide is lower (at high pH), this suggests formation of nonconducting oxides, which are not protonated at high pH and can block electrode reaction of negatively charged ferricyanide, because of their negative charge density. The apparent surface changes indicate that irreversible oxidation of the electrolytically treated carbon fiber electrode surfaces can occur after extensive potential cycling in the potential window of +1.5 to -1.0 V. The deactivation is only observed after long use, after many such cycles. We have, for example, determined that after continuous use, during 1 day, after more than 4000 cycles (at 500 V/s, in the potential window of +1.5 to -1.0 V), the background current of electrolytically treated carbon fiber electrodes remains unchanged.1 The stability of all electrodes was tested on the day of electrolytic treatment to identify the effects of continuous use, which involved potential cycling. Electrolytically treated carbon fiber electrodes could be stored under nitrogen for months without loss of activity, as determined by constant sensitivity in determinations of uric acid. Determinations of Uric Acid. The method of treatment of carbon fiber electrodes by potential cycling at 10 V/s in pH 7.4 buffer in a potential window of +1.5 to -1.0 V, and the method of (34) O’Brien, C. J.; Shumaker-Parry, J.; Engstrom, C. R. Anal. Chem. 1998, 70, 1307-1311.
Figure 3. Voltammetry of uric acid at carbon fiber electrodes at pH 7.4 in phosphate buffer. (A) Slow-scan voltammetry at a scan rate of 50 mV/s before (solid line) and after (small dash) electrolytic treatment. (B) Fast-scan voltammetry at 500 V/s at electrolytically treated carbon fiber electrode. Concentrations of uric acid are shown in the figure. Electrode diameter ∼7 µm. Fast-scan voltammogram in (B) is background subtracted, 250 cycles signal averaged.
acquisition of background current between each determination of analyte, produce a microdisk carbon fiber electrode surface that allows sensitive and reproducible determinations of uric acid,12,14 and adenosine. Uric acid anion (urate, pKa ) 5.435) is not detected under physiological conditions (pH 7.4) at electrodes prepared using past procedures (of electrolytic treatment of carbon fiber electrodes), which were developed to suppress the response of negatively charged analytes.12 Modifications of these procedures to optimize determinations of purines, such as adenosine and urate, have not been straightforward.8,12,36 As illustrated by results in Figure 3, carbon fiber electrode surfaces that are produced by the methods of surface treatment developed here, which include cycling of the electrode potential, show excellent sensitivity in determinations of uric acid. Figure 3A illustrates slow-scan voltammetry of uric acid before (solid line) and after (small dash) electrolytic treatment of the carbon fiber electrode surface by potential cycling at a rate of 10 V/s in a potential window of +1.5 to -1.0 V. After electrolytic treatment of the electrode surface, an i-E curve of uric acid with a steep rise and well-developed plateau is observed, in contrast to the response obtained for uric acid before electrolytic treatment of the electrode surface. The steep rise and well-developed plateau of the i-E curve of uric acid both point to fast electrode kinetics of uric acid at an electrolytically treated carbon fiber electrode surface.14 Similar behavior is observed for ferricyanide (Figure 1). A small peak is observed on top of the steady-state voltammogram of uric acid in Figure 3A after electrolytic treatment, (35) Dryhurst, G. Electrochemistry of Biological Molecules; Academic Press: New York, 1997; p 137. (36) Nowall, W. B.; Kuhr, W. G. Anal. Chem. 1995, 67, 3583-3588.
Analytical Chemistry, Vol. 72, No. 7, April 1, 2000
1581
indicating some accumulation of uric acid at the treated carbon fiber electrode surface. Accumulation of uric acid at the electrode surface must result in part from the large surface area of the electrolytically treated carbon fiber electrode, which is apparent from the large electrode capacitance (Table 2). A large electrode capacitance is also indicated by the large current at the foothill of uric acid wave in i-E curves in Figure 3. Panel B in Figure 3 shows the response of 12 µM uric acid at an electrolytically treated carbon fiber electrode obtained by fastscan voltammetry after background subtraction.12,14 Compared to the response of 300 µM uric acid in panel A, obtained by slowscan voltammetry, the fast-scan voltammetric peak of uric acid, at a much lower solution concentration of uric acid (12 µM), is well-developed, and the signal-to- noise ratio12,14 is high in this measurement. Table 2 summarizes sensitivity values for uric acid that were determined from linear working curves37 obtained from measurements of fast-scan voltammetric peak currents at carbon fiber electrodes that were electrolytically treated in different buffers.1,14 Because of the high stability of electrolytically treated electrodes, background current measured in buffer only, in a separate experiment, was easily subtracted from current measured for uric acid in the same buffer by fast-scan voltammetry.12,14 A peak in the background response of electrolytically treated carbon fiber electrodes observed in buffer at ∼0.5 V (Figure 1) has been associated with quinone/hydroquinone groups at the electrode surface38 and was also subtracted. An important observation from determinations of the sensitivity of uric acid in different buffers by fast-scan voltammetry is that the high sensitivity is largely independent of the buffer used in electrode treatment and in determinations of uric acid. This is in agreement with previously discussed results which have shown that the surface structure of graphite produced by electrolytic treatment is largely independent of the buffer used in treatment. Differences in sensitivity that are observed in different buffers (Table 2) are mostly a result of lower potentials (less than 1.2 V vs SCE) that were used to activate electrodes in buffers containing tertiary amines (Tris-HCl and Krebs), which are electroactive at potentials of ∼1.4 V.39 The high sensitivity of uric acid is also due to an apparent efficient accumulation of uric acid at the electrode surface, which is facilitated by the large surface area of the electrolytically treated carbon fiber electrodes. The large surface area is due to defects resulting from electrolytic treatment of the electrode surface. Diffusion can be quite rapid in defect-rich graphite,6,15 and at electrolytically treated carbon fiber electrodes voltammetric current of uric acid is diffusion-controlled.12,14 Using sensitivity values determined for uric acid at electrolytically treated carbon fiber electrodes, an accumulation factor, or a sensitivity enhancement factor, fa UA,, was determined for uric acid (7.6 µM) from peak current ratio, fa UA ) ip expt/ip theor ) 106, from fast-scan voltammetry (500 V/s). To calculate ip theor, the electrode radius determined at pH 7.4 at electrolytically treated carbon fiber electrodes using ferricyanide was used; the calculation of ip theor
assumed a purely diffusion-controlled irreversible reaction of urate at an electrolytically treated carbon fiber electrode.40 Electrolytically treated carbon fiber electrode surfaces can also incorporate significant amounts of solvent and electrolyte. After prolonged (∼24 h) exposure of the treated electrodes to phosphate buffer, at open circuit potentials, a considerable increase in background current is observed. Ascorbate as a Probe of the Electrode Surface. The high sensitivity of uric acid in fast-scan voltammetry (Table 2) is due in part to accumulation of uric acid at surface defects at the electrolytically treated carbon fiber electrode surface, but the high sensitivity is also a result of fast apparent electrode kinetics of uric acid accumulated at the surface. Electrode kinetics of uric acid oxidation have been shown to be fast at carbon fiber electrodes, at scan rates as high as 10 000 V/s.40 For ascorbate (2 mM), an accumulation factor, fa AA ) ip expt/ip theor ) 0.62, was determined at electrolytically treated carbon fiber electrodes at pH 7.4 at 100 V/s. The lower (sensitivity) factor for ascorbate than for uric acid indicates that ascorbate is not accumulated as well as urate at a surface of electrolytically treated carbon fiber electrode at pH 7.4. This results in a selectivity factor fa UA/fa AA ) 170 at 100 V/s. Nevertheless, apparent electrode kinetics of ascorbate are fast at electrolytically treated carbon fiber electrodes at pH 7.4, and a well-defined ascorbate peak is detected at low positive potentials at scan rates as high as 100 V/s; the peak position does not shift significantly with scan rate.14 At scan rates of 500 V/s and above, the voltammetric oxidation peak of ascorbate is not detected at electrolytically treated carbon fiber electrodes, and uric acid, which is oxidized at the same potential as ascorbate, can be detected by fast-scan voltammetry without interference from an excess of ascorbate even when the concentration of ascorbate exceeds the concentration of uric acid by more than 100-fold.12,14 The apparent fast electrode kinetics of ascorbate, and eventual suppression of ascorbate voltammetric response at scan rates of 500 V/s, are likely a result of kinetic effects associated with proton transfer, which follows electron transfer in the electrooxidation of ascorbate. Proton transfer can be mediated by surface groups at carbon fiber electrode surfaces, and this step can become limiting41 in the electrooxidation of ascorbate at electrolytically treated carbon fiber electrodes at scan rates of 500 V/s. The kinetic effects responsible for suppression of voltammetric response of ascorbate at electrolytically treated carbon fiber electrodes at high scan rates differ from those observed previously at carbon electrodes prepared by different methods of surface treatment.12 The previous method of carbon fiber surface treatment has led to a positive shift of the ascorbate voltammetric oxidation peak potential with scan rate, as expected for a nonmediated electrode reaction.12 Determination of Adenosine. Figure 4 shows the voltammetric response of adenosine (10 µM) that was obtained at an electrolytically treated carbon fiber electrode after background subtraction. Large noise spikes in the voltammogram in Figure 4
(37) Cavalheiro, E. T. G.; Brajter-Toth, A. J. Pharmacol. Biomed. Anal. 1999, 19, 217- 230. (38) Kepley, L. J.; Bard, A. J. Anal. Chem. 1988, 60, 1459-1467. (39) Pihel, K.; Schroeder, J. T.; Wightman, R. M. Anal. Chem. 1994, 66, 6, 45324537.
(40) Hsueh, C. C.; Brajter-Toth, A. Anal. Chem. 1993, 65, 1570-1575. (41) Hirst, J.; Duff, J. L. C.; Jameson, G. N. L.; Kemper, M. A.; Burgess, B. K.; Armstrong, F. A. J. Am. Chem. Soc. 1998, 120, 7085-7094. (42) DeFelice, L. J. Introduction to Membrane Noise; Plenum Press: New York, 1981.
1582
Analytical Chemistry, Vol. 72, No. 7, April 1, 2000
reactions that have been achieved previously have been slow at common electrodes.9 Methods of carbon fiber electrode surface treatment/reactivation developed here, which likely produce defects and cation-exchanged surface oxides, ultimately lead to apparent fast electrode kinetics of adenosine which in turn result in high sensitivity in adenosine determinations by fast-scan voltammetry. The sensitivity value determined for adenosine from fast-scan voltammetry, from the adenosine oxidation peak current, is 0.3 nA/µM at 500 V/s, with a linear dynamic range from 1 to 15 µM, and a LOD of ∼2 µM. This sensitivity value is in agreement with values obtained from fast-scan voltammetry of uric acid at similarly treated carbon fiber electrodes (Table 2).14
Figure 4. Adenosine (10 µM) at electrolytically treated carbon fiber electrode: electrode diameter ∼7 µm, 70 mM phosphate buffer pH 7.4, voltammetric scan rate 10 V/s, and 100 cycles with background subtraction and no additional smoothing.
may be due to ion transport at surface defects42 formed by electrolytic treatment of the carbon fiber electrode surface. Noise spikes can be minimized by smoothing,12 but the noise spikes have not been removed here to illustrate the sensitivity and signalto-noise ratio that can be achieved in measurements of low concentrations of adenosine at electrolytically treated carbon fiber electrodes. Fast kinetics of adenosine electrode reaction at electrolytically treated carbon fiber electrode are indicated in the voltammogram in Figure 4 by well-developed oxidation peaks.27 Fast electrode kinetics of adenosine have not been reported previously at carbon electrodes but are not totally unexpected in view of the known fast electrode kinetics in electrooxidation of adenine at active carbon electrodes.35 Adenine is the purine part of adenosine, which is accessible to oxidation in adenosine. A tentative assignment of adenosine oxidation peaks that are observed by voltammetry at electrolytically treated carbon fiber electrodes can be made.35 On the basis of this assignment, the oxidation peak of adenosine at positive potentials of 1.3 V is likely due to electrooxidation of adenine moiety of adenosine, which is likely accompanied by oxygen transfer.35 An oxidation product that is formed in electrooxidation at 1.3 V is likely responsible for a reversible couple observed by voltammetry in Figure 4 at less positive potentials.35 The accumulation factor determined for adenosine, from the adenosine oxidation peak current in fast-scan voltammetry at 500 V/s, is fa ADO ) 32. The high value of the accumulation factor obtained for adenosine indicates that high sensitivity can be achieved in adenosine determinations by fast-scan voltammetry. Similar sensitive amperometric determinations of adenosine have been difficult8,9 previously because kinetics of adenosine electrode
CONCLUSIONS We have demonstrated that microdisk carbon fiber electrodes, which are highly active in electrode reactions of ferricyanide, urate, and adenosine, and which are stable during repeated determinations of uric acid and adenosine, can be fabricated by electrolytic treatment of the carbon fiber electrode surface, which limits irreversible oxidation of the electrode surface. Successful fabrication of active and stable defects at carbon fiber electrode surfaces, by methods of electrolytic treatment that were developed here, appears to be a main reason for the excellent surface properties of these electrodes, as demonstrated by fast electrode kinetics and good stability in repeated determinations. This appears to result from a high density of surface defects at the electrode surface and from a limited density of nonconducting surface oxides.6 Previous results6 are consistent with the formation of defects at the carbon electrode surface and of some defect oxides of high capacitance.15 Nanostructuring of carbon fiber electrode surfaces by the electrolytic treatment described here, through formation of surface defects and through formation of some cation-exchanged defect oxides, is verified by different evidence, which includes a significant increase in electrode capacitance after treatment. At moderately oxidized carbon fiber electrodes, defects have been associated with high electrode capacitance and facilitated electron transfer, as observed. Cation-exchanged oxide defects have been shown to promote a pH-independent electrode response,16,17 as is observed here in the electrode reactions of ferricyanide and Ru(NH3)63+/2+. Only a small surface coverage by oxide-containing groups is evident at the nanostructured carbon fiber electrodes that were prepared here. The electrode treatment described here likely produces a moderately oxidized carbon fiber electrode. The electrode is highly stable, and this appears to be related to reduction of surface oxides which form initially at positive potentials.38 Reduction of surface oxides at graphite is a slow process,15 and oxides are known to contribute to instability of carbon electrode surfaces.38 Fast-scan voltammetry reduces the time spent at oxidizing potentials during signal acquisition, and this may further limit irreversible surface changes and may help maintain electrode stability. Additionally, background signal averaging12 before each analytical determination allows accurate subtraction of background signal from an analytical signal, which improves sensitivity of measurements and minimizes noise. Nanostructured surfaces that have been fabricated, because of a large density of surface defects and a large area, can Analytical Chemistry, Vol. 72, No. 7, April 1, 2000
1583
selectively and rapidly preconcentrate some analytes.37 Accumulation of analytes at surface defects, together with high activity of those defects, result in high sensitivity in analytical determinations. This is illustrated here in determinations of adenosine, which normally shows poor sensitivity,9 and of uric acid.1,12,14 A benefit of fast kinetics of adenosine is the ability to identify adenosine from its voltammetric peaks and potentials. Voltammetry is selective when the peaks appear at or near thermodynamically predicted potentials.40 Selectivity in analytical determinations is further enhanced in fast-scan voltammetry through kinetic discrimination, which is
1584
Analytical Chemistry, Vol. 72, No. 7, April 1, 2000
achieved as a result of differences in electrode kinetics between different analytes.12,14 This selectivity is a result of kinetically slower processes, including processes limited by proton-transfer kinetics, being too slow to generate a response when the rate of signal acquisition is increased. As a result, the slower processes do not interfere with the faster processes.
Received for review June 18, 1999. Accepted January 3, 2000. AC9906680
