- Email: [email protected]
AC impedance studies of anodically treated polycrystalline and homoepitaxial boron-doped diamond electrodes
Electrochimica Acta 48 (2003) 2739 /2748 www.elsevier.com/locate/electacta
AC impedance studies of anodically treated polycrystalline and homoepitaxial boron-doped diamond electrodes Takeshi Kondo, Kensuke Honda, Donald A. Tryk, Akira Fujishima 1,* Department of Applied Chemistry, School of Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan Received 10 January 2003; received in revised form 5 May 2003; accepted 13 May 2003
Abstract The electrochemical properties of several types of diamond electrodes, including polycrystalline and homoepitaxial films, that underwent anodic treatment were examined with the electrochemical impedance spectroscopic (EIS) technique, as well as with capacitance /potential measurements. From an analysis of the impedance behavior, it was found that an additional capacitance element, which is apparent in the relatively high-frequency range (100 /1000 Hz), was generated on the polycrystalline and (1 0 0) homoepitaxial diamond electrodes after anodic treatment. This capacitive element can be characterized as being non-Faradaic, because it has negligible dependence on the applied potential. Acceptor densities and depth profiles were calculated from the Mott / Schottky plots, and the acceptor densities in the near-surface region of the anodically treated surfaces were found to be extremely low. These results indicate that passive layers were generated on the diamond surfaces by the anodic treatment. The capacitance / potential behavior was also consistent with a model consisting of a semiconductor with a passive surface film. The passive film is proposed to arise as a result of the removal of hydrogen acting as an acceptor in the subsurface region, leaving hydrogen that is paired essentially quantitatively with the boron dopant, effectively neutralizing it. # 2003 Elsevier Science Ltd. All rights reserved. Keywords: Boron-doped diamond electrodes; Single-crystal homoepitaxial diamonds; Electrochemical impedance spectroscopy; Anodic treatment; Passivation layer
1. Introduction Conductive boron-doped diamond (BDD) thin films have several unique properties, including an extremely wide potential working range, in which there is a negligibly small background current, as well as extreme physical and chemical stability [1 /9]. Thus, BDD films have recently come to be considered as being among the most ideal electrode materials for electroanalysis [10 / 14] and electrolysis [15,16]. For BDD films prepared by use of the microwave plasma-assisted chemical vapor deposition (MPCVD) method, the as-deposited surfaces are hydrogen-terminated. These surfaces, however, can
* Corresponding author. Present address: Department of Chemistry, University of Puerto Rico, Rio Piedras Campus, PO Box 23346, San Juan, Puerto Rico, 00931-3346.Tel.: /81-44-818-2020; fax: /81-44-819-2038. E-mail address: [email protected] (A. Fujishima). 1 ISE member.
be oxidized while being used as anodes, i.e., at relatively high potentials, and in the process, the physical and electrochemical properties can be modified. One of the consequences of the electrochemical oxidation is that the as-deposited hydrogen termination converts to oxygen termination [17,22,23]. The physical properties are quite different for these two types of surfaces. For example, hydrogen-terminated diamond surfaces exhibit hydrophobicity [17] and high electrical conductivity [18 /21], whereas oxygen-terminated diamond surfaces, which can be obtained from anodic treatment [22,23], oxygen-plasma treatment [17,23], boiling in strong acid [18] or simple long-term storage in air, exhibit hydrophilicity [3,17] and very low conductivity [19 /21]. Furthermore, the electrochemical properties, e.g., the electron-transfer behavior for some compounds, have also been reported to be greatly affected by anodic and oxygen-plasma treatment of BDD electrode surfaces [10,17,22].
0013-4686/03/$ - see front matter # 2003 Elsevier Science Ltd. All rights reserved. doi:10.1016/S0013-4686(03)00391-8
2740
T. Kondo et al. / Electrochimica Acta 48 (2003) 2739 /2748
As one of the possible causes of the changes in the physical properties of diamond surfaces upon oxidation, the effect of oxygen-containing surface functional groups could be proposed [23,24]. Because the direction and magnitude of surface dipoles are different for hydrogen- and oxygen-terminated surfaces [5], the hydrophilic /hydrophobic properties would be expected to be quite different [17]. In addition, the electrontransfer properties for ions and polar molecules can change significantly [17]. It has also been argued that changes in surface functional groups can cause great changes in surface conductivity [21], but it has also been argued that the latter are caused by changes in the amounts of hydrogen present in a shallow surface film region [18,25 /29]. The mechanisms underlying the differences in the surface conductivities for the hydrogen- and oxygenterminated BDD surfaces remain controversial, even though the experimental phenomena are well known [18,25/29]. Understanding these mechanisms should be important for the further development of applications that utilize the surface properties of diamond, including electronic devices, as well as electrochemical ones such as electroanalytical techniques and electrolytic processes. In order to gain insight into the effect of surface oxidation on the electrical properties of diamond surfaces, the electrochemical properties of homoepitaxial single-crystal-like BDD films were particularly focused upon in the present study. Because such films have extremely flat surfaces, no grain boundaries or extraneous crystal facets, they should be eminently suitable for examining the basic properties of diamond.
2. Experimental The diamond films were prepared with an MPCVD system (Seki Technotron Corp., formerly ASTeX Corp., Model AX3120, 1.5 kW). Polycrystalline diamond thin films were deposited on conducting n -Si (1 1 1) substrates. Homoepitaxial diamond thin films were deposited on (1 0 0) and (1 1 1) single-crystal diamonds synthesized by the high-temperature high-pressure method (De Beers). These diamond substrates were polished by use of a proprietary method (Namiki Precision Jewel Co., Ltd.) finishing with a slight offaxis angle (3 /48), which can promote step-flow crystal growth [30,31]. The conditions of deposition, including the carbon and boron sources, were essentially the same as those described in a previous report [12]. Before the surface oxidation or the electrochemical measurement, each sample was heated to 900 8C in hydrogen ambient to ensure that the surface was hydrogen-terminated, although it should be noted that the as-deposited surfaces are already hydrogen-terminated.
The electrochemical measurements were carried out with a conventional three-electrode setup, which included a Pt wire and an Ag/AgCl electrode as counter and reference electrodes, respectively, connected to a potentiostat (HZ-3000, Hokuto Denko). Working and counter electrode surface areas were 0.03 and 4.7 cm 2, respectively. The homoepitaxial (single-crystal substrate) electrode samples were coated with silver paste from the edge of the conductive surface of the electrode, around the edges of the crystals, to the bottom, where the electrical contact was made. This procedure was found to be necessary in order to ensure adequate ohmic contact, and thus to avoid artifacts in the highly sensitive electrochemical impedance measurements. The latter were carried out with a conventional system, including a potentiostat and a frequency response analyzer [32].
3. Result and discussion 3.1. Impedance properties of anodically treated polycrystalline and homoepitaxial diamond electrodes Fig. 1 shows impedance Bode plots (phase angle vs. frequency, modulus vs. frequency) obtained in 0.1 M sulfuric acid at hydrogen-terminated polycrystalline and (1 0 0), (1 1 1) homoepitaxial single-crystal-like diamond electrodes. The impedance behavior for these hydrogenterminated diamond electrodes can be well expressed with the electrical equivalent circuit represented in Fig. 3a, which includes a constant phase element (CPE), over a wide frequency region (10000 /0.01 Hz). The fitting parameters are listed in Table 1. This equivalent circuit has previously been proposed in an impedance study of the semiconducting diamond electrode/electrolyte interface [33]. The origin of the CPE circuit element for the diamond is still not understood. In general, one of the possible origins of a CPE is the presence of a particular type of surface morphology, e.g., a fractal-type surface structure [34]. However, in the present case, the possibility that such a surface morphology gives rise to the CPE characteristics can be dismissed, because the singlecrystal-like electrodes, which have very flat surfaces, also exhibit such characteristics. Grain boundaries are not necessary for the CPE characteristics, either. The physical interpretation of the CPE at diamond electrodes is still being discussed, but it may be considered to be due to a relaxation process on the crystal surface and/ or in the space charge region [33]. Fig. 2 shows Bode plots for the corresponding anodically treated diamond electrodes. Comparing with the hydrogen-terminated surfaces (Fig. 1), a peak at ca. 300 /1000 Hz is seen in the phase angle vs. frequency plot for the anodically treated polycrystalline
T. Kondo et al. / Electrochimica Acta 48 (2003) 2739 /2748
2741
Fig. 1. Bode plots for hydrogen-terminated diamond electrodes: (a) polycrystalline, (b) (1 1 1) homoepitaxial, and (c) (1 0 0) homoepitaxial. Measured for 0.1 M H2SO4. Circles: experimental, line: calculated.
and (1 0 0) diamond electrodes (Fig. 2a, c). Generation of this peak indicates that an additional capacitive element is produced after the anodic surface treatment. The impedance properties of the anodically treated polycrystalline and (1 0 0) electrodes can be well fitted
by a modified equivalent circuit model (Fig. 3b), which includes that shown in Fig. 3a in series with the parallel circuit R2C2. The fitting parameters are also listed in Table 1. In the phase angle plot for the anodically treated (1 1 1) electrode, no such capacitive element was
T. Kondo et al. / Electrochimica Acta 48 (2003) 2739 /2748
2742 Table 1 Fitting parameters in Figs. 1 and 2
s (Fa Va 1 cm 2) a Polycrystalline
b
HT AT b HT c AT b HT c AT c
(1 1 1) (1 0 0)
5 /10 5 5 /10 5 5 /10 6 1 /10 5 1.4 /10 5 5 /10 6
Rs (V cm2) Csc (mF cm 2) Cdl (mF cm 2) R2 (V cm2 C2 (F cm 2)
0.7 16 0.8 18 0.7 50 0.7 200 0.9 150 0.8 150
10 3 2 2.5 1 2.5
4 3 2.2 1 2.8 0.75
/ 200 / / / 1700
/ 4 / / / 0.75
a
Average relative error (%)
1.77 2.51 1.50 0.91 1.42 1.32
HT, hydrogen-terminated; AT, anodically treated surfaces. a Calculated for the frequency region of B/1 kHz of the Bode plots (Figs. 1 and 2). b Equivalent circuit models: Fig. 3a. c Equivalent circuit models: Fig. 3b.
observed in the present study (Fig. 2b). Possible reasons for this behavior will be discussed later. In general, a proposed electrical equivalent circuit model based on the impedance behavior can only express the total electrical properties of the electrochemical system. Therefore, a single element in the circuit model does not always represent a real physical property of the system. Nevertheless, from the comparison of the impedance behavior for the hydrogen-terminated and anodically treated diamond electrode surfaces, it seems reasonable to assume that the addition of a relatively simple circuit element due to the anodic treatment would be explainable by a modification of the physical nature of the interface. Fig. 4 shows the dependence of the capacitance and impedance phase angle vs. frequency upon the severity of the anodic treatment on (1 0 0) homoepitaxial diamond electrode surfaces. The differential capacitance was estimated from the impedance spectra with the assumption of the simplest possible equivalent circuit, i.e., a series RC circuit. The capacitance was found to be ca. 3 mF cm 2 for a hydrogen-terminated surface at 0.1 /100 Hz at a potential of /0.4 V vs. Ag/AgCl (opencircuit potential). However, after only a single potential sweep from 0 to /3 V vs. Ag/AgCl in 0.1 M sulfuric acid, the capacitance decreased to half or less (1.0 /1.5 mF cm 2) (Fig. 4a). Also, in the phase angle plot (Fig. 4b), a peak in the high-frequency region was observed at the same time. This means that only a single anodic potential sweep can give rise to the observation of the same electrochemical behavior as that found for anodically treated BDD surfaces. With subsequent anodic sweep treatments, the capacitance and phase angle behavior changed step by step, and after a total of three cycles, became almost the same as those of the anodically treated surface obtained by constant potential polarization (3 V vs. Ag/AgCl, 30 min). In order to investigate the properties of the smalltime-constant RC sub-circuit generated by anodic treatment in more detail, the potential dependence of the impedance behavior for an anodically treated (1 0 0)
homoepitaxial BDD film was examined (Fig. 5). From /0.2 to 0.0 V vs. Ag/AgCl, the Bode plots exhibit similar behavior. Also, a similar situation was observed up to 1.0 V vs. Ag/AgCl (data not shown). However, for potentials more negative than /0.2 V vs. Ag/AgCl, both the phase angle and modulus in the low-frequency region (B/10 Hz) decreased, as the potential became more negative. This behavior is proposed to be due to the hydrogen evolution reaction on the electrode surface [35]. In contrast, concerning the peak in the highfrequency region, almost no change was observed over a wide potential region ( /0.4 V to /1.0 V vs. Ag/ AgCl). Therefore, it can be supposed that the corresponding small-time-constant RC sub-circuit also has a negligible dependence on potential. A capacitance element of this type, found as a peak in the high-frequency region of the Bode plots, has been reported for polycrystalline diamond electrodes in several papers [36 /38], and this has been ascribed to a film capacitance of the electrode. It has also been reported for a homoepitaxial film [39]; however, detailed discussion concerning the origin or the characteristics of the capacitance has thus far been lacking. Based on the results of the present investigation, it can be concluded that the capacitance element is associated with the oxidized BDD surface. An important aspect is that the capacitance element was observed for single-crystal-type as well as polycrystalline diamond electrodes. Thus, it should not be associated with imperfections that polycrystalline films possess, such as grain boundaries and carbon impurities. Considering that the small-time-constant RC subcircuit was generated after anodic treatment of the diamond, surface functional groups may be considered to play a role in the generation of the element. For sp2 carbon electrode materials (e.g., graphite), it has long been known that redox-active groups, such as /OH and C /O, can give rise to pseudo-capacitive properties associated with an Faradaic process [40]. In such cases, if the time constants are different, the double-layer capacitance and the pseudo-capacitance can be char-
T. Kondo et al. / Electrochimica Acta 48 (2003) 2739 /2748
2743
Fig. 2. Bode plots for anodically treated diamond electrodes: (a) polycrystalline, (b) (1 1 1) homoepitaxial, and (c) (1 0 0) homoepitaxial. Measured for 0.1 M H2SO4. Circles: experimental, lines: calculated.
acterized separately with impedance measurements, as in the present case (Fig. 2). For as-deposited diamond, C /H is the dominant group on the hydrogen-terminated surfaces, and it has been found that /OH [23] and C /O groups [24] are in fact generated on oxidized diamond surfaces. However, in the case of the voltammetry for
conventional electrolytes at high-quality polycrystalline and single-crystal diamond electrodes, no peaks based on redox reactions of the surface functional groups have been observed, even though the surfaces are oxidized. The /OH and C /O groups on sp2 carbon materials can exhibit phenol- and quinone-like properties, respec-
2744
T. Kondo et al. / Electrochimica Acta 48 (2003) 2739 /2748
Fig. 3. Electrical equivalent circuit models. Rs, series resistance; Cdl, double layer capacitance; Csc, space charge capacitance; R2, C2, additional resistance and capacitance, respectively, and ZCPE, impedance corresponding to the constant phase element (CPE).
tively. In contrast, single crystalline diamond, with no grain boundaries, consists only of sp3 carbon, and thus the same types of oxygen-containing surface groups would not exhibit such redox activity. Therefore, the origin of the RC sub-circuit properties described here should not be associated with a pseudo-capacitance of Fig. 5. Bode plots for anodically treated (1 0 0) homoepitaxial diamond electrodes with applying various potentials. Measured for 0.1 M H2SO4. Potentials: 0 V (2), /0.1 V (I), /0.2 V (^), /0.3 V ( /), /0.4 V (k) vs. Ag/AgCl.
surface functional groups, particularly because the latter would depend on potential. Another case in which multiple capacitive elements with different time constants can be observed in the impedance measurements is that of an electrode covered by a surface passivation layer. Metikosˇ-Hukovic et al. [41] have expressed the impedance behavior of a passivation layer on an SnO2 electrode surface with an additional circuit element, the same as that shown in Fig. 3b (R2C2). The electrical properties of surface passivation layers on other materials, such as Fe [42], steels [43] and Si [44,45], have also been expressed by the same type of circuit element. Moreover, a capacitive element due to a passivation layer does not depend on the potential and is thus basically due to a non-Faradaic process. This is consistent with the present results; however, the detailed characteristics of such a layer need to be examined in greater detail. 3.2. Estimation of depth profiles of acceptor densities in the diamond electrode surfaces Fig. 4. (a) Capacitance /log frequency and (b) phase angle /log frequency plots for (1 0 0) homoepitaxial diamond electrodes. Hydrogenterminated (2) after 1 (I), 2 (^) 3 ( /) anodic potential cycles (from 0 to /3 V vs. Ag/AgCl, 100 mV s 1) and anodic potential holding (k) (/3 V vs. Ag/AgCl, 30 min).
In order to investigate the acceptor properties of hydrogen-terminated and anodically treated diamond electrode surfaces, Mott /Schottky (M /S) plots were prepared (Fig. 6). For all of the hydrogen-terminated surfaces, the plots exhibit non-linear behavior, i.e., the
T. Kondo et al. / Electrochimica Acta 48 (2003) 2739 /2748
2745
films do not conform to the model of a simple, homogeneously doped semiconductor. Tentatively, the total capacitance of the electrode/electrolyte system (Ctot) can be expressed as follows: 1 1 1 ; Ctot Cdl Csc
(1)
where Cdl and Csc are the double layer and the space charge capacitance, respectively. Generally for semiconductor electrodes, the value of Csc is much smaller than that of Cdl, and thus the value of Ctot is approximately the same as that of Csc. This is not true in the present case, as found from the M /S plots. The values for Csc are large enough that significant contributions to Ctot are made by Cdl; the large values of Csc indicate that the acceptor densities in these samples are larger than those that would exhibit simple semiconductor properties. Furthermore, as described later, the M /S behavior for these hydrogen-terminated surfaces is consistent with a non-homogeneous distribution of acceptors. In contrast, for the anodically treated surfaces, the M /S plots for all samples exhibited linear behavior, and thus these electrodes were found to have p-type semiconductor properties. Other workers have also reported linear behavior for M/S plots of oxidized semiconducting diamond surfaces [33,39]. Assuming a p-type semiconductor/electrolyte interface, the acceptor densities estimated from the slopes of the M /S plot were smallest at the (1 0 0) (5.5 /1018 cm 3) surface, larger at the (1 1 1) (2 /1019 cm 3) surface and largest at the polycrystalline (7.5 /1019 cm 3) surface. These values are consistent with the former report [12]. Also, these values were much smaller than the boron concentration expected from the concentration of the boron source in the gas feed during the film preparation (ca. 1021 cm 3). As mentioned above, the non-linear M /S behavior is consistent with a non-homogeneous acceptor concentration. In fact, the slope at any point in the M/S curve can be used to calculate an acceptor concentration at a given depth within the sample [46]. Fig. 7 shows estimated depth profiles of the acceptor densities from the data of the M /S plots (Fig. 6). The procedure to calculate values is the same as that in the former report [47]. Regarding the total capacitance as that of a space charge layer, the depth was estimated from the simple model of a parallel-plate capacitor, and the acceptor density was estimated based on the M /S theory: @C 2 2 ; oqNA (w) @V Fig. 6. Mott /Schottky plots for hydrogen-terminated and anodically treated diamond electrodes at 100 Hz. (a) Polycrystalline, (b) (1 1 1) homoepitaxial, and (c) (1 0 0) homoepitaxial. Measured for 0.1 M H2SO4. Circles, hydrogen-terminated surface; squares, anodically treated surface.
(2)
where V is the potential, o is the dielectric constant of diamond, q is the unit charge, and NA(w ) is the acceptor density at a particular depth w . This procedure, although very simple, seems to be reasonable. The acceptor densities for every one of the hydrogen-
2746
T. Kondo et al. / Electrochimica Acta 48 (2003) 2739 /2748
Fig. 7. Acceptor densities vs. depth plot for hydrogen-terminated and anodically treated diamond electrodes. Circles, Polycrystalline; triangles, (1 1 1) homoepitaxial; squares, (1 0 0) homoepitaxial diamond electrodes. Open and closed symbols represent hydrogen-terminated and anodically treated surface, respectively. Calculated from the data of Fig. 6. Lines: schematic profiles of the acceptor densities.
terminated samples were estimated to be up to more than 1020 cm 3, and this value is one to two orders of magnitude greater than the 11B concentration measured with SIMS in the previous report [12]. The distribution of the acceptor was found to be mainly in the nearsurface region ( B/5 /10 nm depth). The acceptor in the anodically treated diamond surface was estimated to distribute in the region deeper than 10 /30 nm from the surface. Also, it is important to note that the estimated values of acceptor densities were almost the same as the boron concentrations in the bulk measured with SIMS for both the (1 0 0) and (1 1 1) electrode samples [12]. It should be noted that this estimation with M /S theory is very approximate, due to the fact that we are stretching the theory to the case of high-acceptor density (/1020 cm 3) and simplifying the idea of space charge capacitance, as well as that of the frequency dependence of the capacitance. However, in spite of roughness of the approximation, the values and the tendency are somewhat reasonable, especially for the anodically treated surfaces, as mentioned above. The depth dependences of the acceptor densities, found in the present study, are similar to that given in the former report of our group [47]. The origin of this dependence may be explained in terms of the role of hydrogen atoms in the diamond surface. In a number of reports, it has been stated that hydrogen-terminated surfaces exhibit surface conductivity, even for undoped polycrystalline or single-crystal diamond thin films [19,20,25,48]. However, the mechanism of the conductivity has not absolutely been elucidated, although several models have been proposed. One possible model is that hydrogen atoms in the near-surface region of diamond act as acceptors, causing the conductivity [49,50]. Chevallier et al. [51] reported that hydrogen
(deuterium) atoms could diffuse into the diamond bulk from the surface during annealing in a hydrogen (deuterium) atmosphere. In the present case, the hydrogen-terminated surface was ensured with a hydrogen annealing treatment; however, the possible presence of hydrogen in the near-surface region is more likely to have resulted from the microwave plasma treatment [52]. It is reasonable to assume that hydrogen atoms may exist to some extent in the near-surface region of the diamond, as well as on the surface. Therefore, if the hydrogen acts as an acceptor, the resulting additional acceptor density can result in a conductivity of the diamond surface higher than that expected from the boron doping level. Another possible role of hydrogen in the diamond involves the reverse process, i.e., that of a donor, passivating boron acceptors by forming H/B dimers. Zeisel et al. [49] have reported that H/B dimer formation causes a passivation layer to form on semiconductor diamond, thus modifying the electrical properties. It is necessary to note that the hydrogen (deuterium) plasma treatment that Zeisel et al. [49] and Chevallier et al. [51] used produced the opposite results in terms of conductivity compared with our results; in those studies, the plasma treatment produced passivation, whereas in ours, it produced an increase in conductivity. In addition, with a microwave plasma, the hydrogen (deuterium) treatment has been shown to lead to decreased conductivity [52]. We propose that this difference is based on the two different types of behavior of hydrogen: HB dimers (hydrogen as donor) and H h pairs (hydrogen as acceptor). In the present work (see below), the hydrogen concentration in the hydrogenterminated diamond electrodes is found to be higher than that of boron, similar to results obtained with SIMS [50 /52]. In such a case, hydrogen atoms could produce H h pairs in the film, although some would also be expected to produce H B dimers, which is more to be expected with the microwave plasma [52]. This could be the mechanism for the acceptor density being greater than the boron concentration near the surface for the hydrogen-terminated diamond surfaces (Fig. 7), but then dropping to values below the boron concentration further into the bulk. This result is consistent with those of Chevallier et al. [52], showed that for RF plasma treated surfaces the hydrogen (deuterium) concentration can be quite high close to the surface and then matches the boron concentration closely at greater depths. Considering the direction of the electric field during the anodic treatment, hydrogen in the diamond would be expected to migrate from the bulk to the electrode surface as H . However, the diffusion energy barrier [53] is lowest for H (0.1 eV), compared with those of H (1.9 eV) and H (2.4 eV). Thus, migration may not be as important as diffusion over the small distances
T. Kondo et al. / Electrochimica Acta 48 (2003) 2739 /2748
involved in the surface-enriched layer (in the order of 10 nm). During oxidation, the concentration of H and/or H would be depleted at the electrode surface, providing a driving force for diffusion. This could selectively eliminate the acceptor hydrogen and consequently create a passivation layer if trapped hydrogen in the form of H B dimers remained even after the anodic treatment. Formation of such a passivation layer is consistent with the present results, i.e., the depth profile of acceptors in the diamond electrode surface from the M /S plots and the generation of an additional smalltime-constant RC sub-circuit properties found from the impedance behavior. In the present study, the additional capacitance element could not be seen in the impedance behavior for the anodically treated (1 1 1) diamond electrode. It can be suggested that the time constant of the element is simply too large to be seen definitely, because no significant difference was found in the acceptor depth profile for this surface compared with those of the other two surfaces. A further test of this model, i.e., a semiconductor with a surface passivation layer is provided by the work of Zeisel et al. [49,54,55], who developed a mathematical treatment of the capacitance /potential behavior of Schottky junctions involving diamond. According to the model, the space charge capacitance of a semiconductor with a passivation layer C can be represented as a function of the bias potential V (in the electrochemical situation, the bias potential can be converted to a deviation from the open-circuit potential, /0.4 V vs. Ag/AgCl): o ffi; C(V ) sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi p 2o N (VFB V ) w2 1 A qNA NA
(3)
where VFB is the flat-band potential, and NAp and NA are the acceptor densities in the passivation layer and in the adjacent region of the diamond surface, respectively. The value of VFB (/0.6 V vs. SCE, corresponding to / 0.25 V vs. open circuit) is taken from work on polycrystalline BDD films (1017 /3 /1018 cm 3) [33]. As shown by Pleskov et al. [33], the values cannot be directly estimated from the M /S plots (Fig. 6); these are certainly too large, due to the additional capacitance being discussed. The values of NA are estimated from the slopes of the M /S plots (Figs. 6 and 7). For NAp ; values about two orders of magnitude smaller than those of NA were tentatively used. Calculated curves were then fitted to the experimental data by varying the value of w (Fig. 8), and the fits were reasonably good. The values of w for each sample were close to those obtained from the M /S plots (Table 2). This agreement indicates that a simple passivation layer model can
2747
Fig. 8. Capacitance /overpotential plots for anodically treated diamond electrodes and the simulated curves (lines). Circle, polycrystalline; triangle, (1 1 1) homoepitaxial; and square, (1 0 0) homoepitaxial diamond electrodes.
Table 2 Fitting parameters in Fig. 8 NAp (cm 3)
NA (cm 3) Polycrystalline (1 1 1) (1 0 0)
/
19
7.5 /10 2 /1019 5.5 /1018
17
10 1017 1016
w (nm) 7.7 15 28
adequately represent the anodically treated diamond electrode/electrolyte system. Thus, the changes in the electrical properties before and after the anodic treatment of the hydrogen-terminated diamond surface may be explained with this passivation layer model. In order to fully understand the physical basis and mechanism of generation of such a passivation layer on the diamond surface, detailed information on the diffusion and/or migration of hydrogen in diamond via electrochemical treatment must be obtained, however.
4. Conclusions In the present study, the electrochemical properties of anodically treated polycrystalline and homoepitaxial single-crystal-type diamond electrodes were investigated. From the impedance behavior, it was found that an additional film capacitance element, which can be expressed, by an RC parallel circuit element, was generated by the anodic treatment. The characteristics of this RC sub-circuit exhibit a very slight dependence on the applied potential, suggesting that it is due to a passivation layer on the diamond surface. In contrast, the capacitive character of the hydrogen-terminated surface is greatly dependent upon the potential and indicates the presence of a high concentration of
2748
T. Kondo et al. / Electrochimica Acta 48 (2003) 2739 /2748
acceptors, higher than that expected from the boron doping level alone, suggestive of a surface region containing hydrogen acting as an acceptor. A possible generation mechanism of the oxidatively generated passivation layer involves the presence of boron dopant. Both the M /S behavior (C 2 vs. E ) and the capacitance vs. potential plots for the anodically treated diamond electrodes are consistent with the surface passivation layer model, with the thickness of this layer varying over the range 8/28 nm. It can be concluded that the anodic treatment of BDD electrodes can cause significant changes in the electrochemical properties of these electrodes, including both polycrystalline and homoepitaxial diamond, due to changes in both the doping of the near-surface region, as well the termination of the surface itself.
Acknowledgements This research was supported by the New Energy and Industrial Technology Development Organization (NEDO) International Joint Research Grant.
References [1] N. Vinokur, B. Miller, Y. Avyigal, R. Kalish, J. Electrochem. Soc. 143 (1996) L238. [2] F. Bouamrane, A. Tadjeddine, J.E. Butler, R. Tenne, C. Le´vyCle´ment, J. Electroanal. Chem. 405 (1996) 95. [3] H.B. Martin, A. Argoitia, U. Landau, A.B. Anderson, J.C. Angus, J. Electrochem. Soc. 143 (1996) L133. [4] G.M. Swain, A.B. Anderson, J.C. Angus, MRS Bull. 23 (1998) 56. [5] J. Xu, Q. Chen, G.M. Swain, Anal. Chem. 70 (1998) 3146. [6] R. Tenne, C. Le´vy-Cle´ment, Israel J. Chem. 38 (1998) 57. [7] G.M. Swain, J. Electrochem. Soc. 141 (1994) 3382. [8] R. DeClements, G.M. Swain, J. Electrochem. Soc. 144 (1997) 856. [9] D. Gandini, P.-A. Michaud, I. Duo, E. Mahe, W. Haenni, A. Perret, C. Comninellis, New Diamond Front. Carbon Technol. 9 (1999) 303. [10] E. Popa, H. Notsu, T. Miwa, D.A. Tryk, A. Fujishima, Electrochem. Solid State Lett. 2 (1999) 49. [11] B.V. Sarada, T.N. Rao, D.A. Tryk, A. Fujishima, Anal. Chem. 72 (2000) 1632. [12] T. Kondo, Y. Einaga, B.V. Sarada, T.N. Rao, D.A. Tryk, A. Fujishima, J. Electrochem. Soc. 149 (2002) E179. [13] C. Terashima, T.N. Rao, B.V. Sarada, D.A. Tryk, A. Fujishima, Anal. Chem. 74 (2002) 895. [14] T.A. Ivandini, B.V. Sarada, C. Terashima, T.N. Rao, D.A. Tryk, H. Ishiguro, Y. Kubota, A. Fujishima, J. Electroanal. Chem. 521 (2002) 117. [15] S. Ferro, A. De Battisti, I. Duo, C. Comninellis, W. Haenni, A. Perret, J. Electrochem. Soc. 147 (2000) 2614. [16] M.A. Rodrigo, P.A. Michaud, I. Duo, M. Panizza, G. Cerisola, C. Comninellis, J. Electrochem. Soc. 148 (2001) D60. [17] I. Yagi, H. Notsu, T. Kondo, D.A. Tryk, A. Fujishima, J. Electroanal. Chem. 473 (1999) 173. [18] K. Hayashi, S. Yamanaka, H. Watanabe, T. Sekiguchi, H. Okushi, K. Kajimura, J. Appl. Phys. 81 (1997) 744.
[19] M. Tachiki, T. Fukuda, K. Sugata, H. Seo, H. Umezawa, H. Kawarada, Jpn. J. Appl. Phys. 39 (2000) 4631. [20] M. Tachiki, T. Fukuda, K. Sugata, H. Seo, H. Umezawa, H. Kawarada, Appl. Surf. Sci. 159 (2000) 578. [21] T. Kondo, M. Yanagisawa, L. Jiang, D.A. Tryk, A. Fujishima, Diamond Relat. Mater. 11 (2002) 1788. [22] H. Notsu, I. Yagi, T. Tatsuma, D.A. Tryk, A. Fujishima, Electrochem. Solid State Lett. 2 (1999) 522. [23] H. Notsu, T. Fukazawa, T. Tatsuma, D.A. Tryk, A. Fujishima, Electrochem. Solid State Lett. 4 (2001) H1. [24] H. Notsu, I. Yagi, T. Tatsuma, D.A. Tryk, A. Fujishima, J. Electroanal. Chem. 492 (2000) 31. [25] H.J. Looi, M.D. Whitfield, J.S. Foord, R.B. Jackmann, Thin Solid Films 344 (1999) 623. [26] K. Tsugawa, H. Noda, A. Hokazono, K. Kitatani, K. Morita, H. Kawarada, Electron. Commun. Jpn, Part 2 81 (1998) 172. [27] A. Denisenko, A. Aleksov, A. Pribil, P. Gluche, W. Ebert, E. Kohn, Diamond Relat. Mater. 9 (2000) 1138. [28] F. Maier, M. Riedel, B. Mantel, J. Ristein, L. Ley, Phys. Rev. Lett. 85 (2000) 3472. [29] S. Yamanaka, D. Takeuchi, H. Watanabe, H. Okushi, K. Kajimura, Appl. Surf. Sci. 159 (2000) 567. [30] D. Takeuchi, H. Watanabe, S. Yamanaka, H. Okushi, K. Kajimura, Diamond Relat. Mater. 9 (2000) 231. [31] M. Yanagisawa, L. Jiang, D.A. Tryk, K. Hashimoto, A. Fujishima, Diamond Relat. Mater. 8 (1999) 2059. [32] K. Honda, T.N. Rao, D.A. Tryk, A. Fujishima, M. Watanabe, K. Yasui, H. Masuda, J. Electrochem. Soc. 147 (2000) 659. [33] L.Y. Pleskov, V.V. Elkin, M.A. Abaturov, M.D. Krotova, V.Y. Mishuk, V.P. Varnun, I.G. Teremetskaya, J. Electroanal. Chem. 413 (1996) 105. [34] W.H. Mulder, J.H. Sluyters, T. Pajkossy, L. Nyikos, J. Electroanal. Chem. 285 (1990) 103. [35] K. Honda, T.N. Rao, D.A. Tryk, A. Fujishima, M. Watanabe, K. Yasui, H. Masuda, J. Electrochem. Soc. 148 (2001) A668. [36] R. Ramesham, M.F. Rose, Thin Solid Films 300 (1997) 144. [37] R. Ramesham, M.F. Rose, Corros. Sci. 39 (1997) 2019. [38] I. Garcı´a, A. Conde, J.J. de Damborenea, A.J. Va´zquez, Thin Solid Films 310 (1997) 217. [39] J. van de Lagemaat, D. Vanmaekelbergh, J.J. Kelly, J. Electroanal. Chem. 475 (1999) 139. [40] D. Qu, J. Power Sources 109 (2002) 403. [41] M. Metikosˇ-Hukovic, S. Omanovic, A. Jukic, Electrochim. Acta 45 (1999) 977. [42] B.D. Cahan, C.T. Chen, J. Electrochem. Soc. 129 (1982) 700. [43] V.A. Alves, C.M.A. Brett, Electrochim. Acta 47 (2002) 2081. [44] P. Schmuki, H. Bo¨hni, J.A. Bardwell, J. Electrochem. Soc. 142 (1995) 1705. [45] B.K. Patel, S.N. Sahu, Appl. Phys. A 71 (2000) 695. [46] E.H. Rhoderick, R.H. Williams, Metal-Semiconductor Contacts, Clarendon Press, Oxford, 1988. [47] D.A. Tryk, K. Tsunozaki, T.N. Rao, A. Fujishima, Diamond Relat. Mater. 10 (2001) 1804. [48] R. Ramesham, Thin Solid Films 315 (1998) 222. [49] R. Zeisel, C.E. Nebel, M. Stutzmann, Diamond Relat. Mater. 9 (2000) 413. [50] C. Uzan-Saguy, A. Reznik, C. Cytermann, R. Brener, R. Kalish, E. Bustarret, M. Bernard, A. Deneuville, E. Gheeraert, J. Chevallier, Diamond Relat. Mater. 10 (2001) 453. [51] J. Chevallier, A. Lusson, D. Ballutaud, B. Theys, F. Jomard, A. Deneuville, M. Bernard, E. Gheeraert, E. Bustarret, Diamond Relat. Mater. 10 (2001) 399. [52] J. Chevallier, D. Ballutaud, B. Theys, F. Jomard, A. Deneuville, E. Gheeraert, F. Pruvost, Phys. Stat. Sol. A 174 (1999) 73. [53] S.P. Mehandru, A.B. Anderson, J. Mater. Res. 9 (1994) 383.
AC impedance studies of anodically treated polycrystalline and homoepitaxial boron-doped diamond electrodes Takeshi Kondo, Kensuke Honda, Donald A. Tryk, Akira Fujishima 1,* Department of Applied Chemistry, School of Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan Received 10 January 2003; received in revised form 5 May 2003; accepted 13 May 2003
Abstract The electrochemical properties of several types of diamond electrodes, including polycrystalline and homoepitaxial films, that underwent anodic treatment were examined with the electrochemical impedance spectroscopic (EIS) technique, as well as with capacitance /potential measurements. From an analysis of the impedance behavior, it was found that an additional capacitance element, which is apparent in the relatively high-frequency range (100 /1000 Hz), was generated on the polycrystalline and (1 0 0) homoepitaxial diamond electrodes after anodic treatment. This capacitive element can be characterized as being non-Faradaic, because it has negligible dependence on the applied potential. Acceptor densities and depth profiles were calculated from the Mott / Schottky plots, and the acceptor densities in the near-surface region of the anodically treated surfaces were found to be extremely low. These results indicate that passive layers were generated on the diamond surfaces by the anodic treatment. The capacitance / potential behavior was also consistent with a model consisting of a semiconductor with a passive surface film. The passive film is proposed to arise as a result of the removal of hydrogen acting as an acceptor in the subsurface region, leaving hydrogen that is paired essentially quantitatively with the boron dopant, effectively neutralizing it. # 2003 Elsevier Science Ltd. All rights reserved. Keywords: Boron-doped diamond electrodes; Single-crystal homoepitaxial diamonds; Electrochemical impedance spectroscopy; Anodic treatment; Passivation layer
1. Introduction Conductive boron-doped diamond (BDD) thin films have several unique properties, including an extremely wide potential working range, in which there is a negligibly small background current, as well as extreme physical and chemical stability [1 /9]. Thus, BDD films have recently come to be considered as being among the most ideal electrode materials for electroanalysis [10 / 14] and electrolysis [15,16]. For BDD films prepared by use of the microwave plasma-assisted chemical vapor deposition (MPCVD) method, the as-deposited surfaces are hydrogen-terminated. These surfaces, however, can
* Corresponding author. Present address: Department of Chemistry, University of Puerto Rico, Rio Piedras Campus, PO Box 23346, San Juan, Puerto Rico, 00931-3346.Tel.: /81-44-818-2020; fax: /81-44-819-2038. E-mail address: [email protected] (A. Fujishima). 1 ISE member.
be oxidized while being used as anodes, i.e., at relatively high potentials, and in the process, the physical and electrochemical properties can be modified. One of the consequences of the electrochemical oxidation is that the as-deposited hydrogen termination converts to oxygen termination [17,22,23]. The physical properties are quite different for these two types of surfaces. For example, hydrogen-terminated diamond surfaces exhibit hydrophobicity [17] and high electrical conductivity [18 /21], whereas oxygen-terminated diamond surfaces, which can be obtained from anodic treatment [22,23], oxygen-plasma treatment [17,23], boiling in strong acid [18] or simple long-term storage in air, exhibit hydrophilicity [3,17] and very low conductivity [19 /21]. Furthermore, the electrochemical properties, e.g., the electron-transfer behavior for some compounds, have also been reported to be greatly affected by anodic and oxygen-plasma treatment of BDD electrode surfaces [10,17,22].
0013-4686/03/$ - see front matter # 2003 Elsevier Science Ltd. All rights reserved. doi:10.1016/S0013-4686(03)00391-8
2740
T. Kondo et al. / Electrochimica Acta 48 (2003) 2739 /2748
As one of the possible causes of the changes in the physical properties of diamond surfaces upon oxidation, the effect of oxygen-containing surface functional groups could be proposed [23,24]. Because the direction and magnitude of surface dipoles are different for hydrogen- and oxygen-terminated surfaces [5], the hydrophilic /hydrophobic properties would be expected to be quite different [17]. In addition, the electrontransfer properties for ions and polar molecules can change significantly [17]. It has also been argued that changes in surface functional groups can cause great changes in surface conductivity [21], but it has also been argued that the latter are caused by changes in the amounts of hydrogen present in a shallow surface film region [18,25 /29]. The mechanisms underlying the differences in the surface conductivities for the hydrogen- and oxygenterminated BDD surfaces remain controversial, even though the experimental phenomena are well known [18,25/29]. Understanding these mechanisms should be important for the further development of applications that utilize the surface properties of diamond, including electronic devices, as well as electrochemical ones such as electroanalytical techniques and electrolytic processes. In order to gain insight into the effect of surface oxidation on the electrical properties of diamond surfaces, the electrochemical properties of homoepitaxial single-crystal-like BDD films were particularly focused upon in the present study. Because such films have extremely flat surfaces, no grain boundaries or extraneous crystal facets, they should be eminently suitable for examining the basic properties of diamond.
2. Experimental The diamond films were prepared with an MPCVD system (Seki Technotron Corp., formerly ASTeX Corp., Model AX3120, 1.5 kW). Polycrystalline diamond thin films were deposited on conducting n -Si (1 1 1) substrates. Homoepitaxial diamond thin films were deposited on (1 0 0) and (1 1 1) single-crystal diamonds synthesized by the high-temperature high-pressure method (De Beers). These diamond substrates were polished by use of a proprietary method (Namiki Precision Jewel Co., Ltd.) finishing with a slight offaxis angle (3 /48), which can promote step-flow crystal growth [30,31]. The conditions of deposition, including the carbon and boron sources, were essentially the same as those described in a previous report [12]. Before the surface oxidation or the electrochemical measurement, each sample was heated to 900 8C in hydrogen ambient to ensure that the surface was hydrogen-terminated, although it should be noted that the as-deposited surfaces are already hydrogen-terminated.
The electrochemical measurements were carried out with a conventional three-electrode setup, which included a Pt wire and an Ag/AgCl electrode as counter and reference electrodes, respectively, connected to a potentiostat (HZ-3000, Hokuto Denko). Working and counter electrode surface areas were 0.03 and 4.7 cm 2, respectively. The homoepitaxial (single-crystal substrate) electrode samples were coated with silver paste from the edge of the conductive surface of the electrode, around the edges of the crystals, to the bottom, where the electrical contact was made. This procedure was found to be necessary in order to ensure adequate ohmic contact, and thus to avoid artifacts in the highly sensitive electrochemical impedance measurements. The latter were carried out with a conventional system, including a potentiostat and a frequency response analyzer [32].
3. Result and discussion 3.1. Impedance properties of anodically treated polycrystalline and homoepitaxial diamond electrodes Fig. 1 shows impedance Bode plots (phase angle vs. frequency, modulus vs. frequency) obtained in 0.1 M sulfuric acid at hydrogen-terminated polycrystalline and (1 0 0), (1 1 1) homoepitaxial single-crystal-like diamond electrodes. The impedance behavior for these hydrogenterminated diamond electrodes can be well expressed with the electrical equivalent circuit represented in Fig. 3a, which includes a constant phase element (CPE), over a wide frequency region (10000 /0.01 Hz). The fitting parameters are listed in Table 1. This equivalent circuit has previously been proposed in an impedance study of the semiconducting diamond electrode/electrolyte interface [33]. The origin of the CPE circuit element for the diamond is still not understood. In general, one of the possible origins of a CPE is the presence of a particular type of surface morphology, e.g., a fractal-type surface structure [34]. However, in the present case, the possibility that such a surface morphology gives rise to the CPE characteristics can be dismissed, because the singlecrystal-like electrodes, which have very flat surfaces, also exhibit such characteristics. Grain boundaries are not necessary for the CPE characteristics, either. The physical interpretation of the CPE at diamond electrodes is still being discussed, but it may be considered to be due to a relaxation process on the crystal surface and/ or in the space charge region [33]. Fig. 2 shows Bode plots for the corresponding anodically treated diamond electrodes. Comparing with the hydrogen-terminated surfaces (Fig. 1), a peak at ca. 300 /1000 Hz is seen in the phase angle vs. frequency plot for the anodically treated polycrystalline
T. Kondo et al. / Electrochimica Acta 48 (2003) 2739 /2748
2741
Fig. 1. Bode plots for hydrogen-terminated diamond electrodes: (a) polycrystalline, (b) (1 1 1) homoepitaxial, and (c) (1 0 0) homoepitaxial. Measured for 0.1 M H2SO4. Circles: experimental, line: calculated.
and (1 0 0) diamond electrodes (Fig. 2a, c). Generation of this peak indicates that an additional capacitive element is produced after the anodic surface treatment. The impedance properties of the anodically treated polycrystalline and (1 0 0) electrodes can be well fitted
by a modified equivalent circuit model (Fig. 3b), which includes that shown in Fig. 3a in series with the parallel circuit R2C2. The fitting parameters are also listed in Table 1. In the phase angle plot for the anodically treated (1 1 1) electrode, no such capacitive element was
T. Kondo et al. / Electrochimica Acta 48 (2003) 2739 /2748
2742 Table 1 Fitting parameters in Figs. 1 and 2
s (Fa Va 1 cm 2) a Polycrystalline
b
HT AT b HT c AT b HT c AT c
(1 1 1) (1 0 0)
5 /10 5 5 /10 5 5 /10 6 1 /10 5 1.4 /10 5 5 /10 6
Rs (V cm2) Csc (mF cm 2) Cdl (mF cm 2) R2 (V cm2 C2 (F cm 2)
0.7 16 0.8 18 0.7 50 0.7 200 0.9 150 0.8 150
10 3 2 2.5 1 2.5
4 3 2.2 1 2.8 0.75
/ 200 / / / 1700
/ 4 / / / 0.75
a
Average relative error (%)
1.77 2.51 1.50 0.91 1.42 1.32
HT, hydrogen-terminated; AT, anodically treated surfaces. a Calculated for the frequency region of B/1 kHz of the Bode plots (Figs. 1 and 2). b Equivalent circuit models: Fig. 3a. c Equivalent circuit models: Fig. 3b.
observed in the present study (Fig. 2b). Possible reasons for this behavior will be discussed later. In general, a proposed electrical equivalent circuit model based on the impedance behavior can only express the total electrical properties of the electrochemical system. Therefore, a single element in the circuit model does not always represent a real physical property of the system. Nevertheless, from the comparison of the impedance behavior for the hydrogen-terminated and anodically treated diamond electrode surfaces, it seems reasonable to assume that the addition of a relatively simple circuit element due to the anodic treatment would be explainable by a modification of the physical nature of the interface. Fig. 4 shows the dependence of the capacitance and impedance phase angle vs. frequency upon the severity of the anodic treatment on (1 0 0) homoepitaxial diamond electrode surfaces. The differential capacitance was estimated from the impedance spectra with the assumption of the simplest possible equivalent circuit, i.e., a series RC circuit. The capacitance was found to be ca. 3 mF cm 2 for a hydrogen-terminated surface at 0.1 /100 Hz at a potential of /0.4 V vs. Ag/AgCl (opencircuit potential). However, after only a single potential sweep from 0 to /3 V vs. Ag/AgCl in 0.1 M sulfuric acid, the capacitance decreased to half or less (1.0 /1.5 mF cm 2) (Fig. 4a). Also, in the phase angle plot (Fig. 4b), a peak in the high-frequency region was observed at the same time. This means that only a single anodic potential sweep can give rise to the observation of the same electrochemical behavior as that found for anodically treated BDD surfaces. With subsequent anodic sweep treatments, the capacitance and phase angle behavior changed step by step, and after a total of three cycles, became almost the same as those of the anodically treated surface obtained by constant potential polarization (3 V vs. Ag/AgCl, 30 min). In order to investigate the properties of the smalltime-constant RC sub-circuit generated by anodic treatment in more detail, the potential dependence of the impedance behavior for an anodically treated (1 0 0)
homoepitaxial BDD film was examined (Fig. 5). From /0.2 to 0.0 V vs. Ag/AgCl, the Bode plots exhibit similar behavior. Also, a similar situation was observed up to 1.0 V vs. Ag/AgCl (data not shown). However, for potentials more negative than /0.2 V vs. Ag/AgCl, both the phase angle and modulus in the low-frequency region (B/10 Hz) decreased, as the potential became more negative. This behavior is proposed to be due to the hydrogen evolution reaction on the electrode surface [35]. In contrast, concerning the peak in the highfrequency region, almost no change was observed over a wide potential region ( /0.4 V to /1.0 V vs. Ag/ AgCl). Therefore, it can be supposed that the corresponding small-time-constant RC sub-circuit also has a negligible dependence on potential. A capacitance element of this type, found as a peak in the high-frequency region of the Bode plots, has been reported for polycrystalline diamond electrodes in several papers [36 /38], and this has been ascribed to a film capacitance of the electrode. It has also been reported for a homoepitaxial film [39]; however, detailed discussion concerning the origin or the characteristics of the capacitance has thus far been lacking. Based on the results of the present investigation, it can be concluded that the capacitance element is associated with the oxidized BDD surface. An important aspect is that the capacitance element was observed for single-crystal-type as well as polycrystalline diamond electrodes. Thus, it should not be associated with imperfections that polycrystalline films possess, such as grain boundaries and carbon impurities. Considering that the small-time-constant RC subcircuit was generated after anodic treatment of the diamond, surface functional groups may be considered to play a role in the generation of the element. For sp2 carbon electrode materials (e.g., graphite), it has long been known that redox-active groups, such as /OH and C /O, can give rise to pseudo-capacitive properties associated with an Faradaic process [40]. In such cases, if the time constants are different, the double-layer capacitance and the pseudo-capacitance can be char-
T. Kondo et al. / Electrochimica Acta 48 (2003) 2739 /2748
2743
Fig. 2. Bode plots for anodically treated diamond electrodes: (a) polycrystalline, (b) (1 1 1) homoepitaxial, and (c) (1 0 0) homoepitaxial. Measured for 0.1 M H2SO4. Circles: experimental, lines: calculated.
acterized separately with impedance measurements, as in the present case (Fig. 2). For as-deposited diamond, C /H is the dominant group on the hydrogen-terminated surfaces, and it has been found that /OH [23] and C /O groups [24] are in fact generated on oxidized diamond surfaces. However, in the case of the voltammetry for
conventional electrolytes at high-quality polycrystalline and single-crystal diamond electrodes, no peaks based on redox reactions of the surface functional groups have been observed, even though the surfaces are oxidized. The /OH and C /O groups on sp2 carbon materials can exhibit phenol- and quinone-like properties, respec-
2744
T. Kondo et al. / Electrochimica Acta 48 (2003) 2739 /2748
Fig. 3. Electrical equivalent circuit models. Rs, series resistance; Cdl, double layer capacitance; Csc, space charge capacitance; R2, C2, additional resistance and capacitance, respectively, and ZCPE, impedance corresponding to the constant phase element (CPE).
tively. In contrast, single crystalline diamond, with no grain boundaries, consists only of sp3 carbon, and thus the same types of oxygen-containing surface groups would not exhibit such redox activity. Therefore, the origin of the RC sub-circuit properties described here should not be associated with a pseudo-capacitance of Fig. 5. Bode plots for anodically treated (1 0 0) homoepitaxial diamond electrodes with applying various potentials. Measured for 0.1 M H2SO4. Potentials: 0 V (2), /0.1 V (I), /0.2 V (^), /0.3 V ( /), /0.4 V (k) vs. Ag/AgCl.
surface functional groups, particularly because the latter would depend on potential. Another case in which multiple capacitive elements with different time constants can be observed in the impedance measurements is that of an electrode covered by a surface passivation layer. Metikosˇ-Hukovic et al. [41] have expressed the impedance behavior of a passivation layer on an SnO2 electrode surface with an additional circuit element, the same as that shown in Fig. 3b (R2C2). The electrical properties of surface passivation layers on other materials, such as Fe [42], steels [43] and Si [44,45], have also been expressed by the same type of circuit element. Moreover, a capacitive element due to a passivation layer does not depend on the potential and is thus basically due to a non-Faradaic process. This is consistent with the present results; however, the detailed characteristics of such a layer need to be examined in greater detail. 3.2. Estimation of depth profiles of acceptor densities in the diamond electrode surfaces Fig. 4. (a) Capacitance /log frequency and (b) phase angle /log frequency plots for (1 0 0) homoepitaxial diamond electrodes. Hydrogenterminated (2) after 1 (I), 2 (^) 3 ( /) anodic potential cycles (from 0 to /3 V vs. Ag/AgCl, 100 mV s 1) and anodic potential holding (k) (/3 V vs. Ag/AgCl, 30 min).
In order to investigate the acceptor properties of hydrogen-terminated and anodically treated diamond electrode surfaces, Mott /Schottky (M /S) plots were prepared (Fig. 6). For all of the hydrogen-terminated surfaces, the plots exhibit non-linear behavior, i.e., the
T. Kondo et al. / Electrochimica Acta 48 (2003) 2739 /2748
2745
films do not conform to the model of a simple, homogeneously doped semiconductor. Tentatively, the total capacitance of the electrode/electrolyte system (Ctot) can be expressed as follows: 1 1 1 ; Ctot Cdl Csc
(1)
where Cdl and Csc are the double layer and the space charge capacitance, respectively. Generally for semiconductor electrodes, the value of Csc is much smaller than that of Cdl, and thus the value of Ctot is approximately the same as that of Csc. This is not true in the present case, as found from the M /S plots. The values for Csc are large enough that significant contributions to Ctot are made by Cdl; the large values of Csc indicate that the acceptor densities in these samples are larger than those that would exhibit simple semiconductor properties. Furthermore, as described later, the M /S behavior for these hydrogen-terminated surfaces is consistent with a non-homogeneous distribution of acceptors. In contrast, for the anodically treated surfaces, the M /S plots for all samples exhibited linear behavior, and thus these electrodes were found to have p-type semiconductor properties. Other workers have also reported linear behavior for M/S plots of oxidized semiconducting diamond surfaces [33,39]. Assuming a p-type semiconductor/electrolyte interface, the acceptor densities estimated from the slopes of the M /S plot were smallest at the (1 0 0) (5.5 /1018 cm 3) surface, larger at the (1 1 1) (2 /1019 cm 3) surface and largest at the polycrystalline (7.5 /1019 cm 3) surface. These values are consistent with the former report [12]. Also, these values were much smaller than the boron concentration expected from the concentration of the boron source in the gas feed during the film preparation (ca. 1021 cm 3). As mentioned above, the non-linear M /S behavior is consistent with a non-homogeneous acceptor concentration. In fact, the slope at any point in the M/S curve can be used to calculate an acceptor concentration at a given depth within the sample [46]. Fig. 7 shows estimated depth profiles of the acceptor densities from the data of the M /S plots (Fig. 6). The procedure to calculate values is the same as that in the former report [47]. Regarding the total capacitance as that of a space charge layer, the depth was estimated from the simple model of a parallel-plate capacitor, and the acceptor density was estimated based on the M /S theory: @C 2 2 ; oqNA (w) @V Fig. 6. Mott /Schottky plots for hydrogen-terminated and anodically treated diamond electrodes at 100 Hz. (a) Polycrystalline, (b) (1 1 1) homoepitaxial, and (c) (1 0 0) homoepitaxial. Measured for 0.1 M H2SO4. Circles, hydrogen-terminated surface; squares, anodically treated surface.
(2)
where V is the potential, o is the dielectric constant of diamond, q is the unit charge, and NA(w ) is the acceptor density at a particular depth w . This procedure, although very simple, seems to be reasonable. The acceptor densities for every one of the hydrogen-
2746
T. Kondo et al. / Electrochimica Acta 48 (2003) 2739 /2748
Fig. 7. Acceptor densities vs. depth plot for hydrogen-terminated and anodically treated diamond electrodes. Circles, Polycrystalline; triangles, (1 1 1) homoepitaxial; squares, (1 0 0) homoepitaxial diamond electrodes. Open and closed symbols represent hydrogen-terminated and anodically treated surface, respectively. Calculated from the data of Fig. 6. Lines: schematic profiles of the acceptor densities.
terminated samples were estimated to be up to more than 1020 cm 3, and this value is one to two orders of magnitude greater than the 11B concentration measured with SIMS in the previous report [12]. The distribution of the acceptor was found to be mainly in the nearsurface region ( B/5 /10 nm depth). The acceptor in the anodically treated diamond surface was estimated to distribute in the region deeper than 10 /30 nm from the surface. Also, it is important to note that the estimated values of acceptor densities were almost the same as the boron concentrations in the bulk measured with SIMS for both the (1 0 0) and (1 1 1) electrode samples [12]. It should be noted that this estimation with M /S theory is very approximate, due to the fact that we are stretching the theory to the case of high-acceptor density (/1020 cm 3) and simplifying the idea of space charge capacitance, as well as that of the frequency dependence of the capacitance. However, in spite of roughness of the approximation, the values and the tendency are somewhat reasonable, especially for the anodically treated surfaces, as mentioned above. The depth dependences of the acceptor densities, found in the present study, are similar to that given in the former report of our group [47]. The origin of this dependence may be explained in terms of the role of hydrogen atoms in the diamond surface. In a number of reports, it has been stated that hydrogen-terminated surfaces exhibit surface conductivity, even for undoped polycrystalline or single-crystal diamond thin films [19,20,25,48]. However, the mechanism of the conductivity has not absolutely been elucidated, although several models have been proposed. One possible model is that hydrogen atoms in the near-surface region of diamond act as acceptors, causing the conductivity [49,50]. Chevallier et al. [51] reported that hydrogen
(deuterium) atoms could diffuse into the diamond bulk from the surface during annealing in a hydrogen (deuterium) atmosphere. In the present case, the hydrogen-terminated surface was ensured with a hydrogen annealing treatment; however, the possible presence of hydrogen in the near-surface region is more likely to have resulted from the microwave plasma treatment [52]. It is reasonable to assume that hydrogen atoms may exist to some extent in the near-surface region of the diamond, as well as on the surface. Therefore, if the hydrogen acts as an acceptor, the resulting additional acceptor density can result in a conductivity of the diamond surface higher than that expected from the boron doping level. Another possible role of hydrogen in the diamond involves the reverse process, i.e., that of a donor, passivating boron acceptors by forming H/B dimers. Zeisel et al. [49] have reported that H/B dimer formation causes a passivation layer to form on semiconductor diamond, thus modifying the electrical properties. It is necessary to note that the hydrogen (deuterium) plasma treatment that Zeisel et al. [49] and Chevallier et al. [51] used produced the opposite results in terms of conductivity compared with our results; in those studies, the plasma treatment produced passivation, whereas in ours, it produced an increase in conductivity. In addition, with a microwave plasma, the hydrogen (deuterium) treatment has been shown to lead to decreased conductivity [52]. We propose that this difference is based on the two different types of behavior of hydrogen: HB dimers (hydrogen as donor) and H h pairs (hydrogen as acceptor). In the present work (see below), the hydrogen concentration in the hydrogenterminated diamond electrodes is found to be higher than that of boron, similar to results obtained with SIMS [50 /52]. In such a case, hydrogen atoms could produce H h pairs in the film, although some would also be expected to produce H B dimers, which is more to be expected with the microwave plasma [52]. This could be the mechanism for the acceptor density being greater than the boron concentration near the surface for the hydrogen-terminated diamond surfaces (Fig. 7), but then dropping to values below the boron concentration further into the bulk. This result is consistent with those of Chevallier et al. [52], showed that for RF plasma treated surfaces the hydrogen (deuterium) concentration can be quite high close to the surface and then matches the boron concentration closely at greater depths. Considering the direction of the electric field during the anodic treatment, hydrogen in the diamond would be expected to migrate from the bulk to the electrode surface as H . However, the diffusion energy barrier [53] is lowest for H (0.1 eV), compared with those of H (1.9 eV) and H (2.4 eV). Thus, migration may not be as important as diffusion over the small distances
T. Kondo et al. / Electrochimica Acta 48 (2003) 2739 /2748
involved in the surface-enriched layer (in the order of 10 nm). During oxidation, the concentration of H and/or H would be depleted at the electrode surface, providing a driving force for diffusion. This could selectively eliminate the acceptor hydrogen and consequently create a passivation layer if trapped hydrogen in the form of H B dimers remained even after the anodic treatment. Formation of such a passivation layer is consistent with the present results, i.e., the depth profile of acceptors in the diamond electrode surface from the M /S plots and the generation of an additional smalltime-constant RC sub-circuit properties found from the impedance behavior. In the present study, the additional capacitance element could not be seen in the impedance behavior for the anodically treated (1 1 1) diamond electrode. It can be suggested that the time constant of the element is simply too large to be seen definitely, because no significant difference was found in the acceptor depth profile for this surface compared with those of the other two surfaces. A further test of this model, i.e., a semiconductor with a surface passivation layer is provided by the work of Zeisel et al. [49,54,55], who developed a mathematical treatment of the capacitance /potential behavior of Schottky junctions involving diamond. According to the model, the space charge capacitance of a semiconductor with a passivation layer C can be represented as a function of the bias potential V (in the electrochemical situation, the bias potential can be converted to a deviation from the open-circuit potential, /0.4 V vs. Ag/AgCl): o ffi; C(V ) sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi p 2o N (VFB V ) w2 1 A qNA NA
(3)
where VFB is the flat-band potential, and NAp and NA are the acceptor densities in the passivation layer and in the adjacent region of the diamond surface, respectively. The value of VFB (/0.6 V vs. SCE, corresponding to / 0.25 V vs. open circuit) is taken from work on polycrystalline BDD films (1017 /3 /1018 cm 3) [33]. As shown by Pleskov et al. [33], the values cannot be directly estimated from the M /S plots (Fig. 6); these are certainly too large, due to the additional capacitance being discussed. The values of NA are estimated from the slopes of the M /S plots (Figs. 6 and 7). For NAp ; values about two orders of magnitude smaller than those of NA were tentatively used. Calculated curves were then fitted to the experimental data by varying the value of w (Fig. 8), and the fits were reasonably good. The values of w for each sample were close to those obtained from the M /S plots (Table 2). This agreement indicates that a simple passivation layer model can
2747
Fig. 8. Capacitance /overpotential plots for anodically treated diamond electrodes and the simulated curves (lines). Circle, polycrystalline; triangle, (1 1 1) homoepitaxial; and square, (1 0 0) homoepitaxial diamond electrodes.
Table 2 Fitting parameters in Fig. 8 NAp (cm 3)
NA (cm 3) Polycrystalline (1 1 1) (1 0 0)
/
19
7.5 /10 2 /1019 5.5 /1018
17
10 1017 1016
w (nm) 7.7 15 28
adequately represent the anodically treated diamond electrode/electrolyte system. Thus, the changes in the electrical properties before and after the anodic treatment of the hydrogen-terminated diamond surface may be explained with this passivation layer model. In order to fully understand the physical basis and mechanism of generation of such a passivation layer on the diamond surface, detailed information on the diffusion and/or migration of hydrogen in diamond via electrochemical treatment must be obtained, however.
4. Conclusions In the present study, the electrochemical properties of anodically treated polycrystalline and homoepitaxial single-crystal-type diamond electrodes were investigated. From the impedance behavior, it was found that an additional film capacitance element, which can be expressed, by an RC parallel circuit element, was generated by the anodic treatment. The characteristics of this RC sub-circuit exhibit a very slight dependence on the applied potential, suggesting that it is due to a passivation layer on the diamond surface. In contrast, the capacitive character of the hydrogen-terminated surface is greatly dependent upon the potential and indicates the presence of a high concentration of
2748
T. Kondo et al. / Electrochimica Acta 48 (2003) 2739 /2748
acceptors, higher than that expected from the boron doping level alone, suggestive of a surface region containing hydrogen acting as an acceptor. A possible generation mechanism of the oxidatively generated passivation layer involves the presence of boron dopant. Both the M /S behavior (C 2 vs. E ) and the capacitance vs. potential plots for the anodically treated diamond electrodes are consistent with the surface passivation layer model, with the thickness of this layer varying over the range 8/28 nm. It can be concluded that the anodic treatment of BDD electrodes can cause significant changes in the electrochemical properties of these electrodes, including both polycrystalline and homoepitaxial diamond, due to changes in both the doping of the near-surface region, as well the termination of the surface itself.
Acknowledgements This research was supported by the New Energy and Industrial Technology Development Organization (NEDO) International Joint Research Grant.
References [1] N. Vinokur, B. Miller, Y. Avyigal, R. Kalish, J. Electrochem. Soc. 143 (1996) L238. [2] F. Bouamrane, A. Tadjeddine, J.E. Butler, R. Tenne, C. Le´vyCle´ment, J. Electroanal. Chem. 405 (1996) 95. [3] H.B. Martin, A. Argoitia, U. Landau, A.B. Anderson, J.C. Angus, J. Electrochem. Soc. 143 (1996) L133. [4] G.M. Swain, A.B. Anderson, J.C. Angus, MRS Bull. 23 (1998) 56. [5] J. Xu, Q. Chen, G.M. Swain, Anal. Chem. 70 (1998) 3146. [6] R. Tenne, C. Le´vy-Cle´ment, Israel J. Chem. 38 (1998) 57. [7] G.M. Swain, J. Electrochem. Soc. 141 (1994) 3382. [8] R. DeClements, G.M. Swain, J. Electrochem. Soc. 144 (1997) 856. [9] D. Gandini, P.-A. Michaud, I. Duo, E. Mahe, W. Haenni, A. Perret, C. Comninellis, New Diamond Front. Carbon Technol. 9 (1999) 303. [10] E. Popa, H. Notsu, T. Miwa, D.A. Tryk, A. Fujishima, Electrochem. Solid State Lett. 2 (1999) 49. [11] B.V. Sarada, T.N. Rao, D.A. Tryk, A. Fujishima, Anal. Chem. 72 (2000) 1632. [12] T. Kondo, Y. Einaga, B.V. Sarada, T.N. Rao, D.A. Tryk, A. Fujishima, J. Electrochem. Soc. 149 (2002) E179. [13] C. Terashima, T.N. Rao, B.V. Sarada, D.A. Tryk, A. Fujishima, Anal. Chem. 74 (2002) 895. [14] T.A. Ivandini, B.V. Sarada, C. Terashima, T.N. Rao, D.A. Tryk, H. Ishiguro, Y. Kubota, A. Fujishima, J. Electroanal. Chem. 521 (2002) 117. [15] S. Ferro, A. De Battisti, I. Duo, C. Comninellis, W. Haenni, A. Perret, J. Electrochem. Soc. 147 (2000) 2614. [16] M.A. Rodrigo, P.A. Michaud, I. Duo, M. Panizza, G. Cerisola, C. Comninellis, J. Electrochem. Soc. 148 (2001) D60. [17] I. Yagi, H. Notsu, T. Kondo, D.A. Tryk, A. Fujishima, J. Electroanal. Chem. 473 (1999) 173. [18] K. Hayashi, S. Yamanaka, H. Watanabe, T. Sekiguchi, H. Okushi, K. Kajimura, J. Appl. Phys. 81 (1997) 744.
[19] M. Tachiki, T. Fukuda, K. Sugata, H. Seo, H. Umezawa, H. Kawarada, Jpn. J. Appl. Phys. 39 (2000) 4631. [20] M. Tachiki, T. Fukuda, K. Sugata, H. Seo, H. Umezawa, H. Kawarada, Appl. Surf. Sci. 159 (2000) 578. [21] T. Kondo, M. Yanagisawa, L. Jiang, D.A. Tryk, A. Fujishima, Diamond Relat. Mater. 11 (2002) 1788. [22] H. Notsu, I. Yagi, T. Tatsuma, D.A. Tryk, A. Fujishima, Electrochem. Solid State Lett. 2 (1999) 522. [23] H. Notsu, T. Fukazawa, T. Tatsuma, D.A. Tryk, A. Fujishima, Electrochem. Solid State Lett. 4 (2001) H1. [24] H. Notsu, I. Yagi, T. Tatsuma, D.A. Tryk, A. Fujishima, J. Electroanal. Chem. 492 (2000) 31. [25] H.J. Looi, M.D. Whitfield, J.S. Foord, R.B. Jackmann, Thin Solid Films 344 (1999) 623. [26] K. Tsugawa, H. Noda, A. Hokazono, K. Kitatani, K. Morita, H. Kawarada, Electron. Commun. Jpn, Part 2 81 (1998) 172. [27] A. Denisenko, A. Aleksov, A. Pribil, P. Gluche, W. Ebert, E. Kohn, Diamond Relat. Mater. 9 (2000) 1138. [28] F. Maier, M. Riedel, B. Mantel, J. Ristein, L. Ley, Phys. Rev. Lett. 85 (2000) 3472. [29] S. Yamanaka, D. Takeuchi, H. Watanabe, H. Okushi, K. Kajimura, Appl. Surf. Sci. 159 (2000) 567. [30] D. Takeuchi, H. Watanabe, S. Yamanaka, H. Okushi, K. Kajimura, Diamond Relat. Mater. 9 (2000) 231. [31] M. Yanagisawa, L. Jiang, D.A. Tryk, K. Hashimoto, A. Fujishima, Diamond Relat. Mater. 8 (1999) 2059. [32] K. Honda, T.N. Rao, D.A. Tryk, A. Fujishima, M. Watanabe, K. Yasui, H. Masuda, J. Electrochem. Soc. 147 (2000) 659. [33] L.Y. Pleskov, V.V. Elkin, M.A. Abaturov, M.D. Krotova, V.Y. Mishuk, V.P. Varnun, I.G. Teremetskaya, J. Electroanal. Chem. 413 (1996) 105. [34] W.H. Mulder, J.H. Sluyters, T. Pajkossy, L. Nyikos, J. Electroanal. Chem. 285 (1990) 103. [35] K. Honda, T.N. Rao, D.A. Tryk, A. Fujishima, M. Watanabe, K. Yasui, H. Masuda, J. Electrochem. Soc. 148 (2001) A668. [36] R. Ramesham, M.F. Rose, Thin Solid Films 300 (1997) 144. [37] R. Ramesham, M.F. Rose, Corros. Sci. 39 (1997) 2019. [38] I. Garcı´a, A. Conde, J.J. de Damborenea, A.J. Va´zquez, Thin Solid Films 310 (1997) 217. [39] J. van de Lagemaat, D. Vanmaekelbergh, J.J. Kelly, J. Electroanal. Chem. 475 (1999) 139. [40] D. Qu, J. Power Sources 109 (2002) 403. [41] M. Metikosˇ-Hukovic, S. Omanovic, A. Jukic, Electrochim. Acta 45 (1999) 977. [42] B.D. Cahan, C.T. Chen, J. Electrochem. Soc. 129 (1982) 700. [43] V.A. Alves, C.M.A. Brett, Electrochim. Acta 47 (2002) 2081. [44] P. Schmuki, H. Bo¨hni, J.A. Bardwell, J. Electrochem. Soc. 142 (1995) 1705. [45] B.K. Patel, S.N. Sahu, Appl. Phys. A 71 (2000) 695. [46] E.H. Rhoderick, R.H. Williams, Metal-Semiconductor Contacts, Clarendon Press, Oxford, 1988. [47] D.A. Tryk, K. Tsunozaki, T.N. Rao, A. Fujishima, Diamond Relat. Mater. 10 (2001) 1804. [48] R. Ramesham, Thin Solid Films 315 (1998) 222. [49] R. Zeisel, C.E. Nebel, M. Stutzmann, Diamond Relat. Mater. 9 (2000) 413. [50] C. Uzan-Saguy, A. Reznik, C. Cytermann, R. Brener, R. Kalish, E. Bustarret, M. Bernard, A. Deneuville, E. Gheeraert, J. Chevallier, Diamond Relat. Mater. 10 (2001) 453. [51] J. Chevallier, A. Lusson, D. Ballutaud, B. Theys, F. Jomard, A. Deneuville, M. Bernard, E. Gheeraert, E. Bustarret, Diamond Relat. Mater. 10 (2001) 399. [52] J. Chevallier, D. Ballutaud, B. Theys, F. Jomard, A. Deneuville, E. Gheeraert, F. Pruvost, Phys. Stat. Sol. A 174 (1999) 73. [53] S.P. Mehandru, A.B. Anderson, J. Mater. Res. 9 (1994) 383.