ch9 Diamond Electrodes with Functional

9. Diamond Electrodes with Functional Structures and SurfacesYasuaki Einaga, Tribidasari A. Ivandini, and Akira FujishimaThe outstanding properties of diamond make it a very attractive material for use in many potential applications. In particular, the superior electrochemical properties of highly boron-doped conductive diamond films, prepared by the CVD process, have received attention from electrochemists. This article reports the fabrication of boron-doped diamond (BDD) electrodes, creating various functional structures or functional surfaces such as microdisk array (MDA) electrodes, ion-implanted BDD electrodes, and electrodes with ultrasmooth surfaces. Studies have been made of the electrochemical properties of each system and their applications in electroanalysis are discussed.9.1. Boron-Doped Diamond Microdisk Array ElectrodesMicrodisk array (MDA) electrodes of boron-doped diamond (BDD) were fabricated on structured silicon substrates. The BDD-MDA electrodes exhibited sigmoidal voltammetric curves, which show that they function as assemblies of single microelectrodes. The microelectrode behavior was also confirmed with biologicallyimportant species such as ascorbic acid and 3,4-dihydroxyphenylacetic acid, indicating its applicability in electroanalysis [1].Microelectrodes have attracted much attention in electroanalysis, due to their small size. They possess many advantages, such as steady-state response, small IR drop and small capacitance [2-4]. The small size of microelectrodes gives them unique properties, for example, an increased rate of mass transport, which results in improved signal–to-background current ratio in comparison to their planar counterparts. Generally, arrays of microelectrodes are often used in order to increase the current signal, because the current measured at a single microelectrode is very small. Micro-array electrodes such as microdisk array (MDA) electrodes and interdigitated array electrodes can be obtained by use of various micro-fabrication technologies [5,6]. Combining the advantages of diamond and microelectrode arrays makes them more attractive for electroanalytical applications. In this work, microelectrode arrays of BDD were fabricated on silicon substrates.A schematic diagram of the procedure for fabricating the BDD-MDA electrode is shown in Figure 9.1. A Si(100) surface was masked with patterned photoresist and etched with a mixture of HF, HNO3 and H2O. The structured silicon surface was seeded with 10-nm diamond powder. BDD was deposited using a microwave plasma-assisted chemical vapor deposition system. The details of the diamond deposition have been reported elsewhere [7]. After the deposition of diamond, polyimide varnish was spin-coated on the diamond surface. The polyimide layer wasmechanically polished until the diamond tips were just exposed. Electrochemical measurements were carried out at room temperature using a potentiostat and an X-Y recorder. A three-electrode configuration was used for the electrochemical measurements, with Ag/AgCl (sat. KCl) electrode as the reference electrode and platinum wire as the counter electrode. Electrical contact from the microelectrode array was taken from the silicon substrate at the backside.-MDA electrodes: (A) photoresist pattern formed on a silicon substrate; (B) isotropic etching of the substrate;(C) deposition of BDD; and (D) spin-coating and mechanical etching of the polyimide film.First, the distance between each microelectrode in the array and the size of each microelectrode must be considered so that the array may realize the properties of a single microelectrode. That is, if the packing density is low, the diffusion layers will overlap and the array will behave as a macro-sized electrode [4]. In the method used in this work, the distances between the microdisk electrodes were controlled by the mask pattern, and the electrode size was controlled by the sharpness of the etched silicon substratetips, which depends on the etching conditions. A diamond array with a tip size of 25 to 30 µm in diameter was used for the electrochemical measurements. The distance between tips was 250 µm.Fig. 9.2 Cyclic voltammogram at BDD-MDA electrodes for the oxidation of 1 mM K4Fe(CN)6 in 0.1 M Na2SO4; potential sweep rate, 10 mV s-1. The inset is a laser microscope image of the BDD-MDA electrodes.Figure 9.2 shows cyclic voltammograms for the oxidation of ferrocyanide at the BDD-MDA electrode. The voltammogram exhibited a sigmoidal curve, indicating that the array functions as a microelectrode. The half-wave potential was +0.23 V vs. Ag/AgCl, which agrees well with the value for macro-sized BDD electrodes. The calibration curve for the limiting current for oxidation of ferrocyanide was linear over a wide range of concentrations, from 1 µM to 1 mM. The area of the electrode exposed to the electrolyte was 0.13 cm2, in which approximately200 microelectrodes exist. Assuming that each of these electrodes functions as an independent microdisk electrode, the radius of each microelectrode can be estimated from the equation, I lim = 4nFDCr, where I lim is the limiting current, n is the number of electrons, F is the Faraday constant, D is the diffusion coefficient, C is the concentration, and r is the radius of the electrode. By using the value 6.5 ×10-6cm2s-1for the diffusion coefficient of ferrocyanide, the radius was calculated to be 14 µm, which is consistent with the tip size of the electrode observed by SEM.The limiting current at MDA electrodes is known to be independent of potential sweep rate for pure spherical diffusion without cross-talk between neighboring microelectrodes. The limiting current for the oxidation of ferrocyanide measured at the potential of +0.5 V vs Ag/AgCl was constant up to 200 mV s-1. Above this sweep rate, an increase in the limiting current was observed and the voltammogram began to acquire a peak-shape. The microelectrode behavior in this study is similar to that reported for single BDD microelectrodes [8], except that the observed currents were two orders of magnitude higher. This indicates our success to increase the detection current, while the properties as microelectrodes are retained.BDD-MDA electrodes are useful for electrochemical detection of various biologically and environmentally important chemical species with high sensitivity and stability. Sigmoidal voltammograms could be obtained for (A) ascorbic acid and (B) 3,4-dihydroxyphenylacetic acid (DOPAC), as shown in Figure 9.3. Steady-state current was observed even at potentials as high as +1.2 V vs Ag/AgCl, due to the wide potential window of BDD.Electrochemical sensing of chemical species with high oxidation potentials and trace metal analysis are major applications for both microelectrodes and BDD electrodes. These electrodes are also expected to exhibit high stability, similar to their planar counterparts [9]. Further work on electroanalysis using BDD-MDAs is in progress in our laboratory.Fig. 9.3. Cyclic voltammograms at BDD-MDA electrodes for the oxidation of (A) 1 mM ascorbic acid in 0.1 M Na 2SO 4 and (B) 1 mM DOPAC in 0.1 M Na 2SO 4; potential sweep rate, 10 mV s -1).9.2 Ion-ImplantedBoron-Doped DiamondElectrodesNickel-implanted boron-doped diamond electrodes (Ni-DIA) were fabricated in view of their application for carbohydratedetection.This electrode produced well-defined and reproducible voltammograms for 1 mM glucose in alkaline media. The electrode exhibited excellent electrochemical stability with low background current, even after ultrasonic treatment, indicating the strong bonding of nickel with carbon. These results indicate the promising use of Ni-DIA for the detection of carbohydrates and amino acids, and thus, an application of ion implantation, as this surface modification method is effective for controlling the electrochemical properties of polycrystalline diamond thin films [10].Electrochemical detection of carbohydrates is also very attractive due to the possibility of high sensitivity and wide dynamic range. Conventionally, metal electrodes such as nickel and copper are known to oxidize carbohydrates in alkaline solution [11]. The advantage of nickel and copper electrodes is that they produce quite stable responses. These electrodes have been widely used in liquid chromatography and capillary electrophoresis. However, dispersion of metallic particles within an organic polymer or simply on an inert surface results in a drastic increase in the catalyic activity and sensitivity of the electrode [12-14]. A stable, inert electrode with low background current would be the best choice for the deposition of metal electrocatalysts. Metal-modified diamond electrodes appear to be well suited to overcome these problems. As one example, we recently reported on nickel-modified BDD electrodes, demonstrating their use for the detection of carbohydrates [15]. The electrodes were prepared by depositing Ni(NO3)2solution on the surface of the boron-doped diamond. Although the electrochemical stability was relatively good (at least one week), higher stability is required whenstringent conditions, such as sonication or high flow rates are applied, for example, to clean the electrode.Ion implantation has been successfully used in doping semiconductors such as silicon and gallium arsenide. In particular, applications of ion-implanted diamond have recently come to light. In these studies, the electrical conductivity and other physical properties could be controlled by ion-implanting diamond. However, only a few applications for electrochemical uses by preparing conductive electrodes have been reported [16,17].Here, the ion implantation method was applied to prepare electrodes with better electrochemical stability. As a result, we have shown the promising use of nickel-implanted BDD electrodes (Ni-DIA) for glucose detection.The detailed preparation of boron-doped polycrystalline diamond thin films has been described elsewhere [7]. These films were implanted with 750 keV Ni2+with a dose of 5 ×1014cm-2 (Tandetron 4117-HC, HVEE). The annealing was performed at 850o C for 10 min in an H2 ambient (80 Torr). Raman spectroscopy was carried out with Ar+laser illumination (wavelength = 514.5 nm) in a Renishaw Raman imaging microscope system (Renishaw System 2000). A scanning electron microscope (SEM, model JMS-6100, JEOL) was used for imaging the surface morphology. Electrochemical measurements were carried out in 0.2 M NaOH aqueous solution with a single-compartment cell. An Ag/AgCl (saturated KCl) electrode was used as the reference electrode (+0.199 V vs. SHE) and a Pt wire was used as the counter electrode. Current-potential curves were recorded using a potentiostat (Hokuto Denko, Hz-3000). The flow-injection analysis (FIA) systemfor amperometric measurement in the present study consisted of a binary pump (GL Sciences, Inc., PU611), an auto-sampler (Spark-Holland, Triathlon) for constant 10-µL injections, a thin-layer flow cell (GL Sciences, Inc.), an amperometric detector (Bioana]yiical Systems, LC-4C), and a data acquisition system (EZChrom Elite, Scientific Software, Inc.). The wall-jet-type flow cell consisted of the Ag/AgCl reference electrode and a stainless steel (type 316) counter electrode. The geometric area of the electrode in the cell was estimated to be 0.1 cm2. The mobile phase for FIA consisted of a 0.2 M NaOH aqueous solution, and the flow rate was set at 0.2 mL min-1.Figure 9.4 shows the Raman spectra of (a) as-deposited, (b) as-implanted, and (c) annealed samples. The spectrum in Figure 9.4a exhibits a sharp first-order peak (single-phonon scattering) at 1332 cm-1, which provides strong evidence for well-crystallized diamond. After implantation, not only the peak at 1332 cm-1but also a peak at 1550 cm-1, related to sp2-carbon, were observed (Figure 9.4b).Although no changes in surface morphology or color were observed after implantation in the microscope, an SEM image of the surface after implantation indicates the presence of small holes.After 10 min. of annealing at 850o C, the scattering intensity of the characteristic peak is comparable to that of the as-deposited samples (Figure 9.4c). This implies that the strain in the diamond film generated by irradiation has been partially relieved by annealing. Usually, this may be achieved by the recombination of excess interstitials and vacancies. The presence of nickel on the diamond surface was further confirmed by X-ray photoelectron spectroscopy (XPS) spectra, which showed a clear peak at a binding energy of 855 eV, corresponding to Ni 2p l/2.Fig. 9.4. Raman spectra of (a) as-deposited, (b) as-implanted and (c) annealed diamond.Fig. 9.5. (a) Cyclic voltammogram for 1 mM glucose at as-deposited diamond electrode in 0.2 M NaOH; (b) background voltammogram at Ni-DIA in 0.2 M NaOH; (c) cyclic voltammogram for 1 mM glucose at Ni-DIA in 0.2 M NaOH. The potential sweep rate was 100 mV s -1.Figure 9.5a shows a cyclic voltammogram (CV) obtained for an as-deposited diamond electrode in 0.2 M NaOH solution containing l mM glucose. No Faradaic response was observed within the potential window. Furthermore, the background current was very low, as mentioned previously [18-20]. Ni-DIA produced a peak-shaped voltammogram, which shows very low background current at less than +0.7 V vs. Ag/AgCl, in the absence of glucose (Figure 5b). A large increase in the current at about +0.7 is due to the catalytic evolution of oxygen. However, in the presence of 1 mM glucose, a significant increase in the anodic peak currentat+0.70 V vs. Ag/AgCl was observed, which is attributable to redox mediation by the Ni(ll)/Ni(III) couple (Figure 5c). In previous studies with Ni-modified electrodes, anodic and cathodic peaks were observed at +0.48 and +0.36 V vs. SCE, respectively, which were attributed to the Ni(II)/Ni(lll) couple [11,12,15]. The fact that we do not observe the peaks corresponding to Ni(II)/Ni(lll) in this work (Figure 9.5a) is probably due to the very small concentrations of Ni on the diamond surface (Ni/C = 0.1 %) as determined by XPS. However, the large catalytic current for glucose indicates the high catalytic activity of the oxidized form of Ni on the diamond. No peaks for glucose were observed at neutral pH. The voltammograms obtained in the presence of glucose were very reproducible. The observed higher peak voltage (+0.70 V vs. Ag/AgCl) in comparison to electrochemically modified Ni-diamond [15] was due to the lower electrical conductivity (15 Ω cm). That is, Ni-DIA contains a low boron concentration (100 ppm). The presence of a metal oxide/hydroxide film with two different oxidation states at the metal surface appears to be a prerequisite for the electro-oxidation of glucose [21]. In Ni-DIA, Ni(III) acts as a strong oxidant, reacting with the organic compound in a rate-limiting step by abstraction of a hydrogen atom to yield a radical. Further reaction of the radical with additional surface sites results in product formation. Thus, it has been domonstrated that the implanted metal shows promising characteristics for electrochemical sensors, while the properties of the BDD electrode, with chemical stability and low background current, etc., were also demonstrated.Figure 9.6 shows the amperometric response of Ni-DIA for a 10-µL injection of 1 mM glucose in 0.2 mM NaOH solution, with 0.2 mM NaOH as the mobile phase. The operational potential of +0.54 V vs. Ag/AgC1 was selected from the hydrodynamic voltammogram for these measurements. A highly reproducible response, with peak variability less than 9% was observed. The background current for Ni-DIA in Figure 9.6 is as low as 80 nA. This value is lower than that for the bulk nickel electrode, with the response for glucose also being higher for the Ni-DIA electrode. The lowest experimental detection limit was estimated to be 500 nM. The Ni-DIA electrode showed excellent stability, at least for five months with regular use, even with sonication.We have presented the advantages of the ion implantation technique to prepare highly stable metal-modified diamond electrodes. Although only the electrochemical application for glucose detection was shown here, the present work offers new perspectives into functional materials derived from ion-implanted diamond. The most important advantage of ion implantation is that we can design highly stable metal-modified materials by choosing the individual target elements. We have also succeeded in controlling the electrochemical properties of nitrogen-BDD electrodes [22]. In that case, implanted nitrogen made the conductive diamond insulating. Recently there has been an increasing interest in studying the potential application of diamond films, for example, in the electronics field; p- and n-type diamond films are required for these technologies [23]. For these purposes also, the ion implantation technique is thought to have great potential. Thus, further efforts to apply the ion implantationmethod for preparing stable composite materials will be likely to open up many possibilities in the development of new superior functional materials.Smoothed SurfacesWe have focused on the surface modification of diamond electrodes in order to improve their electrochemical properties. Surface modification at the atomic level is a well-known phenomenon, in that the electrochemical properties of the electrodes are found to be quite sensitive to the chemical termination on the surface. Forexample, the electrochemical responses to several different redox systems for oxygen-terminated diamond electrodes and hydrogen-terminated diamond electrodes are remarkably different [25]. As described above, hybrid electrodes, such as metal-modified diamond electrodes, have been prepared by electrochemical deposition methods or ion-implantation methods [10] to realize novel multi-functional electrodes.Next, we focus on the effects of surface morphology. We have reported that the initial rough surface of polycrystalline BDD could be smoothed very easily by use of a radio-frequency glow discharge optical emission spectroscopy (rf-GDOES) technique [26]. Here, we examine the differences in the electrochemical properties between the rough, as-deposited surface and the smoothed diamond surface and discuss the electrochemical properties of the ultrasmooth diamond electrodes from the point of view of a novel electrode material [24].The initial rough, faceted, as-deposited BDD surfaces were smoothed by Ar+ ion sputtering at very low energy (50 eV). A lower background current was measured at these mirror-like modified electrodes than at the initial polycrystalline electrodes. The electrochemical responses to several redox systems also showed a morphological dependence in some cases.Polycrystalline BDD electrodes were deposited onto Si substrates using a microwave plasma-assisted chemical vapor deposition system. The detailed procedures for the preparation have been described elsewhere [7]. After the diamond was deposited, it was sputtered using a GDOES instrument at an Ar pressure of 0.51 Torr by applying an rf power of 40 W at 13.56MHz. The values of the gas pressure and the applied power relate to the plasma per se. The surface of the diamond became mirror-like in appearance. The smoothed surfaces of the polycrystalline diamond films were characterized by Raman spectroscopy, X-ray photoelectron spectroscopy (XPS) and atomic force microscopy (AFM). Electrochemical measurements were carried out in a single-compartment cell. An Ag/AgCl electrode was used as the reference electrode, and a Pt wire was used as the counter electrode. Current-potential curves were recorded using a potentiostat. The electrochemical properties were studied for both hydrogen-terminated and oxygen-terminated electrodes. Although the as-deposited diamond electrodes were terminated with hydrogen, we were able to oxidize the electrodes so that they became oxygen-terminated by employing anodic oxidation, i.e., +3.0 V for 30 min.Figure 9.7 shows AFM images and their corresponding Raman spectra for samples before and after sputtering. Over the area that we examined, the maximum peak-to-valley height was 1.49 µm and 267 nm, respectively for the two types of surfaces, while the average surface roughnesses R a were 238 nm and 30 nm, respectively. The Raman spectra of both samples, i.e., before and after sputtering, exhibited sharp peaks for the sp3 carbon-related band at 1331 cm-1. This shows that the diamond retained the sp3-carbon structure, even after Ar+ ion-sputtering.The XPS spectrum for the sputtered BDD shows a clear Ar 2p 1/2 peak at 250.6 eV, which indicates the presence of argon on the diamond surface (not shown). Because argon atoms were physically adsorbed at the surface, it showed less surface conductivity. In order to remove the argon atoms from the surface and to increase the surface conductivity, the sample was annealed at 800 °C in an H 2 ambient. The Ar 2p 1/2 peak disappeared, and the surface conductivity was recovered after annealing, which indicates that the surface was H-terminated. After the anodic oxidation, a sharp O 1s peak and an O KLL Auger peak clearly appeared in the XPS spectrum (not shown); the calculated O/Cratio of the O-terminated diamond was 0.23, while the calculatedFig. 9.7. AFM images and Raman spectra of a BDD electrode surface (a) before and (b) after Ar+ sputtering.O/C ratio of the H-terminated diamond was 0.021. Therefore, it can be confirmed that the surface became O-terminated.First, we measured the CV for a 0.1 M H 2SO 4 solution at both of the BDD electrodes before and after sputtering. The background current was lower for the smoothed electrode than it was for the initial polycrystalline electrode (Figure 9.8). Determination by AFM gave a surface area of 1.21 cm 2 for the polycrystalline electrode and 1.04 cm 2 for the smoothed electrode per unit apparent area.Next, we studied the electrochemical responses for several redox couples. The morphological changes in the surface did not appear to cause a notable change, within experimental error, in the electrochemical behavior when the H-terminated surfaces were used.On the other hand, the electrochemical responses for several redox couples changed when the electrodes wereO-terminated.Fig. 9.8. CVs for 0.1 M H 2SO 4 at BDD electrodes (a) before and (b) after Ar + sputtering; potential sweep rate, 100 mV s -1.Figure 9.9 shows CVs in a 0.1 M Na2SO4 solution containing 1 mM K3Fe(CN)6 before and after sputtering. A 540-mV anodic-cathodic peak separation was observed in the CV for the electrode before sputtering, and a smaller peak separation (320 mV) was obtained for the electrode after sputtering. This fact indicates an increase in the heterogeneous electron transfer rate constant at the electrode with the smoothed surface compared to the electrode with the rough surface. An increase in the apparent electron transfer rate constant due to the sputtering was also observed for the IrCl62-/3-redox couple. The results of the electrochemical measurements are summarized in Table 9.1.However, we never observed any changes for the Ru(NH3)62+/3+ and Fe3+/2+ couples at the smoothed surfaces. As described above, it is known that the electrochemical properties of diamond electrodes are quite sensitive to the surface termination [25]. That is, a negative surface charge density due to oxygen termination will affect the potential at the reaction plane. As a result, the negative charge of the ionized carboxyl group can act as an electrostatically repulsive site with respect to a redox species with a negative charge.When we measured the electrochemical responses for as-grown electrodes under the same conditions, including the surface terminated species, such differences (bold in Table 9.1) were not observed. This indicates that the observed differences of electrochemical response must be explained in terms of the morphological dependences, as follows. When the surface is rough and has an O-terminated surface, there are more repulsive carboxyl sites for redox species with a negative charge because ofthe three-dimensional roughness. The roughness of the surface was decreased by sputtering, so the amount of surface oxygen decreased in parallel. Indeed, the surface roughness of the electrode before sputtering (Ra = 238 nm) was 8 times greater than it was after sputtering (Ra = 30 nm). This is consistent with the results of the electrochemical measurements.We conclude that an O-terminated smoothed surface only accelerates the apparent electron transfer rate constant for redox species with a negative charge. The present work offers new insights into how the surface morphology of polycrystalline diamond electrodes can affect the electrochemical properties. Also, diamond electrodes with a smoothed surface may not only be useful for electrochemical applications, but also for the study ofbasic electrochemical properties.Figure 9.9. CVs for 1 mM K 3Fe(CN)6 at BDD electrodes (a) before and (b) after Ar + sputtering.Table 9.1. Comparison of anodic-cathodic peak potential separations for several redox species at diamond electodes before and after Ar+ sputtering.9.4. ConclusionsSeveral functional BDD electrodes have been investigated. Each system exhibited superior electrochemical properties. Efforts to improve the electrochemical properties of BDD electrodes will open up many possibilities in the development, not only of sensing applications using electrolysis and electroanalysis, but also in the design of novel electrode materials.References1.K. Tsunozaki, Y. Einaga, T. N. Rao, A. Fujishima, Chem. Lett.(2002) 502.2.M. Bond, Analyst (Cambridge, U.K.), 119 (1994) R1.3.M. Wightman, C. Amatore, R. C. Engstrom, P. D. Hale, E. W.Christensen, W. G. Kuhr, and L. J. May, Neuroscience (Oxford, U.K.) 25 (1988) 513.4.“Microelectrodes: Theory and Applications”, ed. by M. I.Montenegro, M. A. Queiros, and J. L. Daschbach, Kluwer Academic Publishers, The Netherlands (1991).5.Wittkampf, K. Cammann, M. Anrein, and R. Reichelt, Sens.Actuators B, 40 (1997) 79.6. C. Fiaccabrino, X. M. Tang, N. F. de Rooij, and M.Koudelka-Hep, Sens. Actuators B, 35-36 (1996) 247.7.T. Yano, E. Popa, D. A. Tryk, K. Hashimoto, and A. Fujishima,J. Electrochem. Soc., 146 (1999) 1081.8. B. V. Sarada, T. N. Rao, D. A. Tryk, and A. Fujishima, J.Electrochem. Soc., 146 (1999) 1469.9. A. Fujishima, T. N. Rao, E. Popa, B. V. Sarada, I. Yagi, and D.A. Tryk, J. Electroanal. Chem., 473 (1999) 179.10.K. Ohnishi, Y. Einaga, H. Notsu, C. Terashima, T. N. Rao, S.-G.Park, and A. Fujishima, Electrochem. Solid-State Lett., 5 (2002) D1.11.W. Buchberger, Fresenius J. Anal. Chem., 354 (1996) 797.12.G. Casella, E. Desimoni, and T. R. I. Cataldi, Anal. Chim. Acta.,248 (1991) 117.13.J. M. Zadeii, J. Marioli, and T. Kuwana, Anal. Chem., 63 (1991)649.14.F. Luo and T. Kuwana, Anal. Chem., 66 (1994) 2775.15.Uchikado, T. N. Rao, D. A. Tryk, and A. Fujishima, Chem,Lett.(2001) 144.16.K. Takahashi, M. Iwaki, and H. Watanabe, J. Electroanal.Chem., 396 (1995) 541.17.B. Miller, R. Kalish, L. C. Feldman, A. Katz, N. Moriya, K.Short, and A. E. White, J. Electrochem. Soc., 141 (1994) L41. 18.T. N. Rao and A. Fujishima, Diamond Relat. Mater., 9 (2000)84.19.G. M. Swain and R. Ramesham, Anal. Chem., 65 (1993) 345.20.S. Alehashem, F. Chambers, J. W. Strojek, G. M. Swain, and R.Ramesham, Anal. Chem., 67 (1995) 2812.21.M. Fleischmann, K. Korinek, and D. Pletcher, J. Electroanal.Chem., 31 (1971) 39.22.Y. Einaga, K. Ohnishi, and A. Fujishima, unpublished results.23.J. F. Prins, Phys. Rev. B, 61 (2000) 7191.24.R. Sato, T. Kondo, K. Shimizu, K. Honda, Y. Shibayama, K.Shirahama, A. Fujishima, Y. Einaga, Chem. Lett., 32 (2003) 972.25.I. Yagi, H. Notsu, T. Kondo, D. A. Tryk, and A. Fujishima, J.Electroanal. Chem., 473 (1999) 173.26.K. Shimizu, Y. Einaga, K. Ohnishi, A. Fujishima, H. Habazaki,P. Skeldon, and G. E. Thompson, Surf. Interface Anal., 33 (2002)。