co-reactants system at boron-doped diamond electrodes

Electrochimica Acta 51 (2005) 588–597

Hydroxyl radical-related electrogenerated chemiluminescence reaction for a ruthenium tris(2,2
)bipyridyl/co-reactants system at boron-doped diamond electrodes K. Honda a,∗ , Y. Yamaguchi a , Y. Yamanaka a , M. Yoshimatsu a , Y. Fukuda a , A. Fujishima b,1 a

Department of Chemistry and Earth Sciences, Faculty of Science Yamaguchi University, 1677-1 Yoshida, Yamaguchi-shi, Yamaguchi 753-8512, Japan b Kanagawa Academy of Science and Technology (KAST), 3-2-1, Sakato, Takastu-ku, Kawasaki-shi, Kanagawa 213-0012, Japan

Received 9 November 2004; received in revised form 23 March 2005; accepted 7 May 2005 Available online 12 July 2005

Abstract An electrogenerated chemiluminescence (ECL) reaction of the Ru(bpy)3 2+ (2,2
-bipyridyl, bpy)/co-reactant system in the extremely highpotential region (over 2.6 V versus Ag/AgCl) was probed using a boron-doped diamond (BDD) electrode. At the BDD electrode, three ECL waves (1.25, 2.30 and 3.72 V) were observed in cyclic voltammograms for 20 mM ascorbic acid (AA). For the ECL peaks observed at 1.25 V corresponding to the oxidation potential for Ru(bpy)3 2+ (1.15 V), the light intensities and current densities were found to depend on the square root of the AA concentration. This suggests that AA oxidation, followed by the formation of the reducing radical that is necessary for generating the excited state of Ru(bpy)3 2+* occurred through homogeneous electron-transfer between Ru(bpy)3 3+ and the AA species. However, for the ECL peaks at 2.30 V, the current densities and light intensities linearly increased with increasing AA concentration, suggesting that the reducing radical was formed through the direct oxidation at the electrode surface. The ECL reaction at 3.72 V was observed only at the BDD electrode and not at other electrodes. The onset potentials for the light intensity were approximately 2.6 V, independently of the type of the co-reactants (e.g. 2-propanol and AA). The peak potentials exhibited linear relation with the co-reactant concentration. In the analysis of the ECL intensity for various co-reactants (alcohols) that show different reactivity for the hydrogen abstraction reaction, the order of the light intensities at the peaks for alcohols was found to be consistent with that for the rate constants of the hydrogen abstraction reaction. These results indicate that the co-reactant radical was formed through the hydrogen abstraction reaction with the hydroxyl radical (HO• ) generated during the oxygen evolution reaction. © 2005 Elsevier Ltd. All rights reserved. Keywords: Boron-doped diamond; Electrogenerated chemiluminescence; Ru(bpy)3 2+ ; Ascorbic acid; Hydroxyl radical

1. Introduction The boron-doped diamond (BDD) thin films have several superior properties, including an extremely wide potential working range and a small background current in aqueous media, compared to those for other carbon electrodes (e.g. ∗ 1

Corresponding author. Tel.: +81 3 3933 5735; fax: +81 8 3933 5273. E-mail address: [email protected] (K. Honda). ISE member.

0013-4686/$ – see front matter © 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.electacta.2005.05.018

glassy carbon) and metal electrodes [1]. Thus, BDD thin films have been considered as an ideal electrode material for electroanalysis [2] and many potential applications for the highly sensitive electrochemical detection of biological species have been studied [3]. Moreover, as diamond possesses low adsorption property for organic compounds, an application study for in vivo detection of ascorbic acid (AA) or glucose in a biological sample has been reported [4]. However, the BDD electrode does not exhibit the electrocatalytic activity such as methanol oxidation at the

K. Honda et al. / Electrochimica Acta 51 (2005) 588–597

polycrystalline Pt electrode [5]. Therefore, the selectivity for specific compounds is relativity low. To improve the selectivity of the diamond electrode, modification of the diamond surface with metal catalysts or enzymes has been attempted [6]. By these treatments, the ideal characteristics of diamond, i.e. wide electrochemical potential window and low background current were partially lost. Electrogenerated chemiluminescence of the Ru(bpy)3 2+ / co-reactant system has been reported as a promising method to improve the selectivity of the BDD electrode without losing its superior electrochemical properties [7]. At the BDD electrode, the ECL for primary amines with high oxidation potentials could be observed in the high-potential region (ca. 2.5 V versus Ag/AgCl), due to its wide potential window. In addition, the diamond electrode exhibits high reproducibility for the Ru(bpy)3 2+ /amine system [7], thus demonstrating that BDD is a promising electrode material for the ECL detector. The reaction process of the ECL of the Ru(bpy)3 2+ system has been studied in detail [8]. According to the reaction scheme reported in the literature [8], the generation of the excited state of Ru(bpy)3 2+* has been explained by the reaction between the highly reducing species (Co–R• ) and Ru(bpy)3 3+ (Eq. (2)) [8]. Ru(bpy)3 2+ − e− → Ru(bpy)3 3+

(1)

Ru(bpy)3 3+ + Co–R• → Ru(bpy)3 2+∗ + product

(2)

Ru(bpy)3 2+∗ → Ru(bpy)3 2+ + hν

(3)

The ECL light is the emission of the excited state of Ru(bpy)3 2+* (Eq. (3)). In the spectrum of the ECL signal, the emission peak is centered at ca. 610 nm. In this scheme, the highly reducing species (Co–R• ) is necessary to give an electron to Ru(bpy)3 3+ to form Ru(bpy)3 2+* . Tertiaryamines (e.g. tripropylamine) and oxalic acid are well known as a co-reactant. The reducing radicals are formed through the catalytic Co–R oxidation by electrogenerated Ru(bpy)3 3+ or the direct Co–R oxidation at the electrode, followed by deprotonation of the co-reactant radical cations. For the purpose of wastewater treatment, numerous studies have been carried out on the oxidation of organics mediated by a hydroxyl radical (HO• ) that is generated at BDD electrode during water discharge [9]. The role of the hydroxyl radical as electrogenerated intermediates has also been confirmed by electron spin resonance spectroscopy with the spin-trap method [10]. Organic compounds (i.e. phenol) can be fully oxidized to CO2 via intermediation of electrogenerated HO radicals at the BDD electrode [11]. The objective of this study is to explore the novel ECL reaction at BDD electrodes including the formation of the reducing radical mediated by the electrogenerated hydroxyl radicals. The potential for the OH radical generation is much higher (>2.6 V) than the potential of the typical ECL reaction (ca. 1.2 V) [12]. Therefore, it might be possible to separately detect the compounds from which hydrogen can be easily abstracted by the OH radical. In addition, as those

589

co-reactants are thought to be different from the typical coreactants for the ECL system with Ru(bpy)3 2+ (e.g. TPrA and oxalic acid), there is a possibility to detect different types of co-reactants by using this ECL reaction. Ascorbic acid is a well-known molecule that can easily form the radical through the hydrogen abstraction reaction by HO• . By use of AA as a co-reactant, we could first demonstrate a hydroxyl radicalrelated ECL reaction of the Ru(bpy)3 2+ /AA system at the BDD electrode and could make clear the mechanism of this ECL reaction.

2. Experimental Boron-doped diamond films were deposited on nSi(1 1 1) substrates using a high-pressure plasma-assisted chemical vapor deposition (CVD) system (ASTeX Corp., Woburn, MA). The resistivity of the Si substrate was ca. 5 × 10−3  cm. The details of the deposition process have been described previously [13]. A mixture of acetone and methanol in the ratio of 9/1 (v/v) was used as the carbon source. B2 O3 (Extra Pure, Wako Chemical Co. Ltd.) was used as the boron source. B2 O3 was dissolved in the acetone–methanol solution so that the nominal B/C ratio (mol/mol) was 104 ppm in the gas phase. Carbon source was introduced in H2 plasma by bubbling with high purity hydrogen gas (99.999%). The deposition of the film was carried out at a microwave power of 5 kW. A film thickness of ca. 20 ␮m was achieved after 10 h of deposition. The typical boron concentration obtained under these conditions is ca. 1.5 × 1021 cm−3 [14]. The BDD electrode was used as the working electrode without removal from the Si substrates. The resistivity of the BDD films (ca. 10−3  cm) was in the same range as that of the Si substrate. Therefore, there was no difference in the electrochemical behavior between the diamond film with which the ohmic contact was made from the backside of the Si substrate and that for the front side of the diamond films. The oxide layer of the back of the highly conductive Si substrate was removed with sandpaper. The BDD electrode was rinsed thoroughly and ultrasonicated in isopropanol (Wako Chemical Co. Ltd.). The glassy carbon (GC, GC-20, Tokai Ltd.) electrode was prepared by polishing with successively finer grade of diamond powder (2.0 and 0.5 ␮m; FDC W-2 and 0.5, Fujimi Corp.). The GC electrode was rinsed thoroughly and ultrasonicated in ultrapure water (Milli-Q water) after each polishing step. The polycrystalline Pt electrode was preactivated by scanning the potential in 0.1 M H2 SO4 solution between −0.2 and 1.2 V versus Ag/AgCl at 50 mV s−1 for 60 min. By this pretreatment, organic compounds that were adsorbed on the Pt surface were removed by oxygen and hydrogen evolution and the surface of the Pt electrode was cleaned and activated. The electrochemical measurements were carried out in a single-compartment, three-electrode glass cell with an Ag/AgCl reference electrode (3 M KCl) and platinum gauze counter electrode. All experiments were done at room

590

K. Honda et al. / Electrochimica Acta 51 (2005) 588–597

temperature (23 ◦ C). The electrode area was 0.126 cm2 . The electrogenerated chemiluminescenee (ECL) signal was measured with a photomultiplier tube (PMT, Hamamatsu R928) installed under the electrochemical cell. A voltage of 1000 V was supplied to the PMT with a home-built highvoltage supply. The current density and the normalized ECL intensity (PMT output IECL ) were calculated according to the geometric area of the electrodes. Electrochemical measurement was carried out using potentiostat/galvanostat (Hokuto Denko Research, model HZ-3000). Cyclic voltammograms were obtained at a sweep rate of 50 mV s−1 . All chemicals were used without further purification. Ru(bpy)3 Cl2 ·6H2 O (minimum 98%) was obtained from Aldrich (Milwaukee, WI). Ascorbic acid, methanol, ethanol, 1-propanol, 2-propanol and tetrahydrofurane (THF) were reagent quality (Wako Chemical Co. Ltd.). Those chemicals were dissolved in 0.1 M phosphate buffer solution (PBS). The pH value of the solution was adjusted to 8.0 with concentrated NaOH. The solutions were prepared in Milli-Q water (Millipore). The solutions were deoxygenated with nitrogen for 15 min prior to analysis.

3. Results and discussion 3.1. Comparison of the ECL behavior at different electrodes The ECL behavior of the Ru(bpy)3 2+ /co-reactant system was investigated using cyclic voltammetry (CV). Fig. 1 shows voltage–ECL curves of 0.2 mM Ru(bpy)3 2+ in 0.1 M PBS in the presence of 20 mM ascorbic acid at BDD, GC and polycrystalline Pt electrodes. At the BDD electrode (Fig. 1A), the light intensity started at approximately 1.1 V versus Ag/AgCl and three ECL waves appeared during the positive scan. The first ECL peak was observed at 1.25 V. Beyond 1.25 V, voltage–ECL curve and voltammetric curve showed a plateau from 1.25 to 1.7 V. The current density increased again at 1.7 V and overlapped with the current for oxygen evolution at 2.6 V. The light intensity also re-increased at approximately 1.7 V and the second ECL peak appeared at 2.30 V. The re-increasing potential of the light intensity corresponds to that of the current density. After the second ECL peak, the ECL intensity rapidly decreased and reached approximately zero at 2.6 V. Interestingly, in the highest potential region (>2.6 V), the light intensity increased and the third ECL peak was obtained at 3.72 V in spite of the potential for oxygen evolution reaction [8]. This third ECL reaction was observed only at the BDD electrode and the ECL wave at such a highpotential region has not been reported in earlier studies. This is one unique characteristic of the ECL reaction at the BDD electrode. At the GC electrode (Fig. 1B), two ECL waves similar to those at BDD were obtained for the forward sweep up to 4.0 V at the first cycle. At GC, the ECL emission was not observed at the second cycle of CV (the steady state for CV) in the region

Fig. 1. Cyclic voltammograms and ECL curves at: (A) BDD; (B) GC; (C) polycrystalline Pt electrodes in 0.1 M phosphate buffer solution containing 0.2 mM Ru(bpy)3 2+ and 20 mM AA. The dotted line for (A) represents the data in the absence of AA. The dotted line for (B) represents the data of the first cycle and the solid line represents the second cycle; sweep rate, 50 mV s−1 ; geometric area, 0.1 cm2 ; “×60” in (A) refers to the number of the magnification (60 times) of the current density in the absence of AA.

from 0.0 to 4.0 V versus Ag/AgCl. The onset potential for the ECL light is ca. 1.1 V and the first ECL peak was observed at 1.25 V. The potential and the light intensity at the first ECL peak were identical with those at the second cycle of CV up to 1.4 V, indicating that the reproducibility and the stability of the first ECL reaction at GC were relatively high. Therefore, the peak potential of the first ECL in Fig. 1B was thought to be close to that at the second cycle. As shown in Fig. 1, the peak potential of the first ECL at GC is close to that at BDD and the onset for these first ECL peaks corresponds to the oxidation potential for Ru(bpy)3 2+ . Therefore, at the first peak, the ECL process at GC is supposed to be the same as that at BDD and the Ru(bpy)3 2+ may play an important role in this ECL process. Beyond 1.25 V, the light intensity re-increased from 1.55 V and the second ECL peak was observed at 2.10 V. In Fig. 1B, the current density formed a plateau from ca. 1.2 to 1.55 V and the second voltammetric wave was obtained at the potential of ca. 1.9 V, the same region as the second ECL wave. This potential correspondence of the second voltammetric wave to the second ECL peak was also observed at the BDD electrode. Hence, the ECL mechanism for the second peak at GC and BDD is supposed to be closely related with the direct oxidation of ascorbic acid. The reproducibility of the second ECL peak at the first cycle was relatively

K. Honda et al. / Electrochimica Acta 51 (2005) 588–597

high, but the second peak was not observed at the second cycle of CV (the steady state). Therefore, the observed second ECL peak was treated as the pseudo-peak in the voltage–ECL curves. The peak potential of the second ECL at GC is slightly lower than that for BDD (2.3 V). As shown in Fig. 1, the voltammetric peak for AA oxidation at BDD (ca. 0.9 V) was observed at higher potential than that at GC (0.76 V) [15]. The oxidation potential (Eo ) of AA has been reported to be +1.13 V versus SCE [16]. The oxidation reaction of AA has been reported to be the inner sphere charge-transfer reaction that is a highly irreversible and overall two-electron process [17–19]. The inner sphere redox species have been reported to be relatively oxidized at higher potential at the BDD electrode as in the case of oxygen evolution. The higher potential values obtained for the second ECL peak at BDD might be due to the difference in reactivity for AA oxidation between BDD and GC. Detailed reaction mechanism at the second ECL peak will be discussed in a following section (Section 3.3). Beyond the second ECL peak, the light intensity for GC steeply decreased and reached the anodic limit of the ECL at 2.6 V. At the GC electrode, the ECL reaction in the highest potential region (>2.6 V) was not observed. The light intensity at the first cycle of CV at GC was in the same range as that for BDD, as shown in Fig. 1. However, at the second cycle, both light intensity and current density were drastically decreased, indicating that the ECL reaction at GC is influenced a great deal by the previous cycle (e.g. scanning potential range). It has been reported that by applying the higher potential over 2.0 V on GC, the surface of GC was significantly oxidized and the surface morphology was drastically changed [20,21]. After the measurement of the voltage–ECL behavior (Fig. 1B), it was confirmed the increase of the roughness on the GC surface. As a result, the ECL reaction for GC in the higher potential region was found to be unstable due to the low resistance to the surface oxidation. At the polycrystalline Pt electrode (Fig. 1C), only one ECL peak was obtained at 1.3 V during the positive scan and the second ECL peak (which was observed at BDD or GC electrode) was not observed. The peak intensity at the Pt electrode was a factor of ca. 10 lower than that at BDD and GC. This low reactivity for the ECL reaction at polycrystalline Pt has been explained by the surface oxide layer on Pt generated from ∼0.2 V. This film inhibits the direct oxidation of Ru(bpy)3 2+ and AA, resulting in the low ECL intensity [8]. The anodic limit of the ECL reaction is 1.8 V. As the potential for oxygen evolution at Pt is relatively lower than those at BDD and GC electrodes, the excited state of Ru(bpy)3 2+* is quenched by electrogenerated oxygen in the potential region around 1.8 V [8]. In the highest potential region (>2.6 V), the rate for oxygen evolution reaction at Pt was quite high. Therefore, it was difficult to observe the ECL due to oxygen bubbles attached on the electrode surface, indicating that the polycrystalline Pt electrode is not suitable for ECL in the high-potential region. Even at the BDD electrode, in the CV measurements up to 4.0 V, the ECL signal gradually decreased with increas-

591

ing number of cycles. After several cycles, the ECL signal approached the stable value of approximately 0.6 of the initial. The possible explanations for the decrease of the ECL intensity are the formation of oxygen bubbles and the formation of insulating materials on the electrode surface. However, after removing the oxygen bubbles from the surface, the ECL intensity returned to the initial value. Therefore, the decrease of the ECL intensity was mainly due to the formation of oxygen bubbles on the electrode surface. At BDD, the stable ECL value after several cycles indicates the higher reproducibility for ECL in the high-potential region due to the wider potential working range (compared to that at GC). At the BDD electrode, the potential difference for the three ECL peaks suggests that the ECL process at each potential might include a different route to generate the excited state of Ru(bpy)3 2+* . In the following section, we will discuss the ECL reaction route for each peak potential at the BDD electrode. 3.2. ECL route for the first peak at the BDD electrode The light intensity for the first ECL peak (1.25 V) started soon after the oxidation reaction of Ru(bpy)3 2+ , suggesting a close relation of the ECL process with Ru(bpy)3 2+ oxidation. In the voltammetric curve of Fig. 1A, the oxidation peak for Ru(bpy)3 2+ was observed at 1.15 V in 0.1 M PBS + 0.2 mM Ru(bpy)3 2+ . The light intensities for the Ru(bpy)3 2+ /AA system were lower compared to that for TPrA such that the spectra of the ECL for AA could not be measured with photonic multichannel analyzer (PMA-10, HAMAMATSU). However, the fact that the ECL signal was not observed in the absence of Ru(bpy)3 2+ , suggests that the excited species for ECL can be attributed to Ru(bpy)3 2+* . In addition, in the absence of ascorbic acid, as the ECL was not observed at BDD, AA was found to be related to the ECL reaction for the formation of the reducing radical. The ECL reaction in the Ru(bpy)3 2+ system has been well investigated by numerous researchers using the polycrystalline Pt electrode and it was clarified that the ECL process includes two routes for the formation of the highly reducing species (Co–R• ) from the co-reactant [7,8,22]. A highly reducing radical plays an important role to reduce Ru(bpy)3 3+ to the excited state of Ru(bpy)3 2+* . One of those routes is called the catalytic route. In this route, the oxidation of the co-reactant occurs through the homogeneous electron-transfer reaction with electrogenerated Ru(bpy)3 3+ . The other route to produce Co–R• is the direct Co–R oxidation at the electrode surface. The reducing radical (Co–R• ) is formed by the deprotonation of the co-reactant radical cation (Co–R•+ ) that is the oxidation form of the co-reactant. By use of the BDD electrode, the separation of two routes for the Ru(bpy)3 2+ /tripropylamine system have been reported [7]. The ECL reaction that was observed in the potential close to the oxidation potential of Ru(bpy)3 2+ and formed a plateau after the peak, has been attributed to the catalytic route [7]. These characteristic behaviors for the catalytic route were observed for AA, as

K. Honda et al. / Electrochimica Acta 51 (2005) 588–597

592

described previously. The ECL reaction by the catalytic route can be expressed by the following equations [7,8]: Ru(bpy)3 2+ − e− → Ru(bpy)3 3+

(4)

Ru(bpy)3 3+ + Co–R → Co–R•+ + Ru(bpy)3 2+

(5)

Co–R•+ → Co–R•

(6)

Co–R• + Ru(bpy)3

3+

→ product + Ru(bpy)3

Ru(bpy)3 2+∗ → Ru(bpy)3 2+ + hν

2+∗

(7) (3)

For an irreversible catalytic reaction following a reversible charge-transfer reaction like Eqs. (8) and (9) (Eqs. (4) and (5)), the theory of stationary electrode porarography for CV has been treated by Nicholson and Shain [23]. R ⇔ O + n e− kf

O + Z −→ R

(8) (9)

According to this theory, for the large value of kf /a (kf is the rate constant of reaction 9 and a is the parameter related to the scan rate ν, a = nFv/RT), the relationship of the current–voltage–concentration is: √ nFA Dkf CR∗
nF  (10) i= 1 + exp RT (E − E1/2 ) where A is the electrode area and D and CR∗ the diffusion coefficient and bulk concentration of Ru(bpy)3 2+ , respectively. E1/2 is the polarographic half-wave potential. For the very anodic potentials, Eq. (10) reduces to Eq. (11): i = nFACR × D1/2 kf 1/2

(11)

This relationship suggests the characteristic flat limiting current region [23]. Since kf = k[AA], the current density should be proportional to [AA]1/2 and a plot of the current as a function of [AA]1/2 should be linear. This flat limiting current arises when the rate of removal of Ru(bpy)3 2+ by the electrolysis is exactly compensated by the rate of production of Ru(bpy)3 2+ by Eqs. (5) or (9), so that the concentration of Ru(bpy)3 2+ at the electrode surface (the reaction plane) attains a value independent of time (or ν). In Fig. 2, the light intensity and the current density of the first ECL peak exhibit the linear relation with the square root of the AA concentration. From these results, we were able conclude that the ECL process at the first peak is attributable to the catalytic route. In Fig. 1A, the current density for the ECL reaction is ca. 30 times higher than that for the current density of the oxidation reaction for Ru(bpy)3 2+ . In this case, Ru(bpy)3 3+ was thought to be trapped at the reaction plane and the cyclic reaction of Eqs. (4) and (5) might occur, resulting that the catalytic AA oxidation was thought to be described by mass transport-controlled kinetics of AA. The catalytic route has been reported to play the dominant role at higher Ru(bpy)3 2+ concentration and at lower AA concentration [7]. However, as the ECL intensity for AA was extremely lower compared to those for TPrA and oxalic acid

Fig. 2. The dependence of the light intensities () and the current densities ( ) on ascorbic acid concentrations at the first ECL peak potential in voltage–ECL curves of Fig. 1. The values of the slop and the correlation coefficient (r2 ) for the linear regression are (5.53 ␮A cm−2 mM−1/2 , 0.989) and (0.150 mA cm−2 mM−2 , 0.991), respectively.

at the optimized condition (0.1 mM Ru(bpy)3 2+ and 1000 V of PMT voltage), the minimum AA concentration observable for ECL was ca. 1 mM. The behavior of the light intensities at lower AA concentration than 1 mM will be measured after the improvement in the sensitivity of the ECL detection system. The first ECL peak was observed at higher potential than that for the irreversible AA oxidation at 0.85 V. The oxidation peak potential of the irreversible AA oxidation has been reported to be +1.13 V versus SCE [16]. If the reducing radical of AA was formed thorough this irreversible oxidation reaction, the direct oxidation route should mainly contribute to the first ECL reaction, resulting in the linear increase of the first ECL intensities with the AA concentration. Therefore, the oxidation reaction for the formation of the reducing radical might be different from the irreversible oxidation reaction at 0.85 V and the necessity of the deprotonation reaction for the formation of the reducing radical (Eq. (6)) was also not clarified. The clarification of the detailed AA oxidation scheme at the catalytic route is now in progress. 3.3. ECL route for the second peak at the BDD electrode The second ECL process for BDD at approximately 2.3 V in the Ru(bpy)3 2+ /AA system will be discussed in this section. For the second ECL at the BDD electrode in Fig. 1A, the correspondence of the potential for the re-increase in the current density to the peak potential is also quite similar to that for the Ru(bpy)3 2+ /TPrA system [7]. Previous study has made clear that the reaction process in the Ru(bpy)3 2+ /TPrA system at the second ECL involves the direct oxidation of the co-reactant to the radical cation at the electrode surface (direct oxidation route). The reactions proposed for the direct oxidation route are [8]: Ru(bpy)3 2+ − e− → Ru(bpy)3 3+ Co–R − e− → Co–R•+

(1) (12)

K. Honda et al. / Electrochimica Acta 51 (2005) 588–597

Fig. 3. The dependence of the light intensities (
and ) and the current densities ( ) on ascorbic acid concentrations at the second ECL peak potential in voltage–ECL curves of Fig. 1. (
) represents the observed ECL intensity and () represents the value after subtracting the first ECL intensity from the second ECL. The values of the slop and the correlation coefficient (r2 ) for the linear regression are (3.30 ␮A cm−2 mM−1 , 0.998) for the observed intensity (2.49 ␮A cm−2 mM−1 , 0.998) for the subtracting intensity and (0.0477 cm−2 mM−1 , 0.996) for the current density.

Co–R•+ → Co–R•

(6)

Ru(bpy)3 3+ + Co–R• → Ru(bpy)3 2+∗ + product

(2)

Ru(bpy)3

2+∗

→ Ru(bpy)3

2+

+ hν

(3)

The current density for direct oxidation can be expressed by [24] ip = (2.69 × 105 )n1/2 DR 1/2 v1/2 CR∗

(13)

DR and CR∗ are the diffusion coefficient and bulk concentration of ascorbic acid, respectively, n and ν are the number of electrons involved in the electrochemical reaction and potential sweep rate, respectively. This equation indicates that the current densities for the direct oxidation route should be linearly proportional to the AA concentration. Fig. 3 shows the dependence of the second ECL on the AA concentrations. In this potential region (2.3 V), the generation of the reducing radical through the catalytic route also might occur, then, the light intensities at the second ECL peaks are thought to include the ECL for the catalytic route. Hence, the observed ECL intensities and the values after subtracting the first ECL intensity from the second ECL are shown in Fig. 3. The current density and the subtracting intensities exhibit the linear relation with the AA concentration. Hence, it could be concluded that the ECL process at the second peak includes AA radical formation through the direct oxidation reaction at the electrode surface (direct oxidation route). The direct AA oxidation reaction to form the reducing radical (from 1.7 V) might have a close relation to the catalytic AA oxidation by Ru(bpy)3 3+ (ca. 1.15 V). However, the potential difference between the first and second ECL is relatively large. The possible explanation for this difference might be due to the difference of the reactivity between outer and inner sphere redox species at the diamond electrode. At the BDD electrode, it has been reported that inner sphere redox species were relatively oxidized higher over potential

593

Fig. 4. (A) Cyclic voltammograms and (B) voltage–ECL curves at the BDD electrode in 0.1 M PBS containing 0.1 mM Ru(bpy)3 2+ and (a) 30 mM AA or (b) THF. The data were obtained in the second cycle.

due to the absence of the requisite surface sites needed for the adsorption of reaction intermediates on diamond. [1]. For the direct AA oxidation to form the reducing radical, it might be necessary higher overpotential at BDD (∼1.7 V) than that for the catalytic oxidation by Ru(bpy)3 3+ (1.25 V). At BDD, the separation of the catalytic route and the direct oxidation route has been observed for TPrA in Ru(bpy)3 2+ /co-reactant system of ECL [7]. Ascorbic acid oxidation at 2.30 V might be different from the irreversible oxidation reaction obtained at 0.85 V. Clarification of the detailed oxidation scheme at 2.30 V is now in progress. In the CV of Fig. 1, the anode peak for AA oxidation could not be observed at approximately 2.5 V. This might be because the oxidation potential of AA is close to that for oxygen evolution and the decrease of the light intensity beyond the second ECL peak might include the quenching of Ru(bpy)3 2+* by electrogenerated oxygen [8]. Therefore, the potential value for AA oxidation (approximately 1.7 V) estimated from the second ECL peak might include some uncertainty. However, the observation of the oxidation potential for the co-reactant in the region for oxygen evolution could be achieved using ECL reaction at the BDD electrode. 3.4. ECL route in the potential region over 2.6 V at the BDD electrode In the highest potential region (>2.6 V versus Ag/AgCl), the ECL could be observed only at the BDD electrode. Finally, we will discuss the reaction process of this ECL. 3.4.1. Comparison of the light intensity for the first ECL with that for the third ECL Fig. 4 shows (A) voltammetric curves and (B) voltage–ECL curves at the BDD electrode in 0.1 M PBS

594

K. Honda et al. / Electrochimica Acta 51 (2005) 588–597

containing 0.1 mM Ru(bpy)3 2+ and (a) 30 mM AA and (b) 30 mM 2-propanol. In the absence of the Ru(bpy)3 2+ , the third ECL was not obtained. Therefore, the third ECL was supposed to be the emission from the exited state of Ru(bpy)3 2+* . In addition, in the absence of the co-reactants (AA and 2propanol), the third ECL peak was also not observed (not shown), so the co-reactant molecule is necessary for the reaction process of the third ECL. In the ECL process of the Ru(bpy)3 2+ /co-reactant system, the reducing radicals to generate the excited state of Ru(bpy)3 2+* are necessary. In this potential region (>2.6 V), as the co-reactant oxidation also might occur (at the first and second ECL reaction), there is the possibility for the contribution of the oxidized forms of the co-reactants to form the reducing radicals in addition to the co-reactant molecules. If the reducing radical is generated through the direct oxidation route, the oxidation current for the co-reactant should be added to the current for O2 evolution in the potential region for the third ECL. However, there is no difference between the current density for the third ECL and that for oxygen evolution in PBS. To clarify the contribution of the oxidation potential of the co-reactant to the third ECL behavior, an analysis using a different co-reactant molecule (2-propanol) was carried out. In the highest potential region over 2.6 V, in CV for AA and 2-propanol (Fig. 4A), the anodic currents associated with the oxygen evolution reaction were only observed. In contrast, in the voltage–ECL curves (Fig. 4B) the ECL peaks (the third ECL peak) for AA and 2-propanol were obtained at 4.4 and 3.9 V, respectively. The onset potential for the third ECL (approximately 2.6 V) and peak potential for 2-propanol are found to be similar to those for AA, respectively. However, the light intensity at the third ECL peak for AA is approximately 0.3 of that for 2-propanol, suggesting that the rate of the third ECL reaction for 2-propanol is approximately 3 times higher than that for AA. In contrast, the light intensities at the first (1.25 V) and second (2.3 V) ECL peaks for AA are one to four times higher than those for 2-propanol. In the first and second ECL reaction, the co-reactants are oxidized to the radical cations through the catalytic and direct oxidation routes, respectively. The light intensities at the first and second peaks have been reported to be affected by the degree of the difficulty of the oxidation of co-reactions [7]. In fact, the light intensities at the first ECL peaks for amines (co-reactants) increase with decreasing in the first ionization potentials [7]. Consequently, the peak intensities of the third ECL for AA and 2-propanol no longer follow the same order as that for the first and second ECL. Then, the radical formation process for the third ECL reaction might not include the co-reactant oxidation through the homogeneous and heterogeneous electron-transfer reaction. 3.4.2. Concentration dependence of the third ECL Fig. 4 clearly shows that the peak potential for the third ECL varied with the type of the co-reactant species. The dependence of the peak potential of the third ECL on the co-reactant concentration was examined. The voltage–ECL

Fig. 5. Voltage–ECL curves at the BDD electrode in 0.1 M PBS containing 0.1 mM Ru(bpy)3 2+ and 10, 20 and 40 mM ascorbic acid. The data were obtained in the second cycle.

curves obtained in the Ru(bpy)3 2+ + AA solution (Fig. 5) show the range in behavior as AA was varied in concentration from 10 to 40 mM. In Fig. 5, the onset potentials for the third ECL are approximately 2.6 V, independent of the types of the co-reactant (AA or 2-propanol) and the concentrations (Figs. 4 and 5), suggesting that the species that started to be electrogenerated at 2.6 V might contribute to the formation of the co-reactant radicals. Light intensities at the third ECL peak increased with increasing co-reactant concentration (Fig. 5). Surprisingly, the peak potentials were also positively shifted with increasing concentration. Fig. 6 shows the dependence of the light intensities and the potentials at the third ECL peaks on the AA concentrations. It was confirmed that both values show the linear relation with the AA concentration. To form the exited state of Ru(bpy)3 2+* , the oxidation of the co-reactant or oxidized form of the co-reactant is necessary, as previously mentioned. If the oxidation reaction occurs on the electrode surface, the shift of the peak potential of the third ECL might not be observed. The peak potential in the CV for the reversible charge-transfer reactions (e.g. the oxidation reaction for the inorganic redox analytes like Fe(CN)6 −3/−4 ) is known to show the constant value, inde-

Fig. 6. The dependence of the light intensities () and the peak potentials ( ) on ascorbic acid concentration in voltage–ECL curves of Fig. 5. The values of the slop and the correlation coefficient (r2 ) for the linear regression are (1.67 ␮A cm−2 mM−1 , 0.993) and (0.0440 V mM−1 , 0.997), respectively.

K. Honda et al. / Electrochimica Acta 51 (2005) 588–597

pendent of the bulk concentration of the reactant (CR∗ ) and the sweep rate ν [16]. For the totally irreversible processes, the peak potential in the CV also does not depend on CR∗ at constant sweep rate ν [16]. Those results also support that the third ECL process does not include the direct oxidation reaction of the co-reactant. The CV curves (not shown) show the currents associated with oxygen evolution reaction and are independent of the AA concentration. The results of the constant onset potential for third ECL (approximately 2.6 V) for any co-reactant and the linear increase of the peak potential with AA concentration, suggest that at the third ECL, the co-reactant radicals were formed by the reaction between the products or intermediates of oxygen evolution and the co-reactant. The rates of these intermediates production are thought to increase with the potential. The peak for the third ECL might be obtained when the rate of the intermediates production reached the same values as the mass transport limit of the co-reactants. Therefore, the potentials of the third ECL peak might be shifted to a positive direction with increasing co-reactant concentration. Moreover, as the third ECL reaction was observed only at the BDD electrode, the products or intermediates necessary for the formation of the co-reactant radical are confined to that generated only at BDD. The hydroxyl radical is well known to be easily generated during oxygen evolution reaction at the BDD surface and the generation of the OH radical in the high-potential region (>2.6 V) has been confirmed by ESR measurement with the spin-trap method [10]. The initial step of water discharge is the formation of adsorbed hydroxyl radicals on the electrode surface (M (HO• )): M + H2 O → M(OH• ) + H+ + e−

(14)

At the active electrode like Pt, as there is strong interaction of the hydroxyl radicals with the electrode surface, the absorbed hydroxyl radical forms the higher oxide (M–O) and oxygen evolution occurs through this higher oxide. M(OH• ) → MO + H+ + e−

(15)

MO → M + 21 O2

(16)

where M is an active site at the electrode surface. In nonactive electrodes like BDD, there is a weak interaction between the hydroxyl radicals and the electrode surface, resulting in oxidation of organic molecules mediated by HO• . Hydroxyl radicals are discharged to O2 . M(OH• ) → M + 21 O2 + H+ + e−

595

Fig. 7. Voltage–ECL curves in 0.1 M PBS containing 0.1 mM Ru(bpy)3 2+ and 30 mM co-reactant. The co-reactants are THF, AA, 2-propanol and 1propanol.

to that for oxygen evolution. However, the light emission started at the slightly higher potential than oxygen evolution (Fig. 4). This is because the rate for the OH radical formation at the potential under 2.6 V was thought to be lower than that for the reducing radical to emit the observable ECL intensity. 3.4.3. Third ECL behavior for various co-reactant molecules By assuming that the hydroxyl radical plays an important role in the third ECL reaction, an analysis of the third ECL intensity using the co-reactant molecules that show various reactivity for the hydrogen abstraction reaction by HO• was carried out. Fig. 7 shows the voltage–ECL curves for tetrahydrofurane, AA, 1-propanol and 2-propanol at the BDD electrode. The oxidation potentials reported in literature for 1-propanol, 2-propanol and THF are 2.96, 2.90 and 2.20 V versus SCE, respectively [25,26]. The light intensities and potentials at the third ECL peak varied with the co-reactant species. Among the co-reactants used in this study, the light intensity for THF exhibits the highest value and the value is a factor of 4 and 2 higher than those for AA and 1-propanol, respectively (Fig. 8). The light intensity for 2-propanol was slightly lower than that for THF, but roughly in the same range. It has been reported that hydrogen is easily abstracted from THF and 2-propanol by the halogen radical [27] and forms the

(17)

In the absence of the co-reactants, the third ECL was not observed, so the co-reactant molecule or the oxidized forms of the co-reactants are necessary for the reaction process of the third ECL. Therefore, there is the possibility for the formation of the reducing radical (Co–R• ) thorough the reaction between OH• and the co-reactants or oxidized forms of the co-reactants. If the third ECL includes the formation of the coreactant radicals by the reaction between the co-reactant and HO• , the onset potential for the third ECL should correspond

Fig. 8. Comparison of the peak intensities of the third ECL for co-reactants examined in Fig. 7.

596

K. Honda et al. / Electrochimica Acta 51 (2005) 588–597

Fig. 9. Relationship of the light intensities at the third ECL peak () and the rate constants of hydrogen abstraction reaction ( ) for alcohols (coreactants).

corresponding radicals. The OH radical is well known as a radical initiator and abstracts hydrogen from the molecules like THF. By using alcohols as a co-reactant, the behaviors of the third ECL intensities were examined. The kinetics of the hydrogen abstraction reaction for alcohols have been well studied [28]. In Fig. 9, the intensities of the third ECL peak for 2-propanol, 1-propanol, ethanol and methanol are compared with the rate constants of the hydrogen abstraction reaction, reported in previous literature [28]. The reported values of the oxidation potential for methanol and ethanol are 3.13 and 3.01 V versus SCE, respectively [25]. In Fig. 9, it was confirmed that the order of the peak intensity for the third ECL is consistent with that for the rate constants of the hydrogen abstraction reaction. Therefore, we were able to conclude that the reaction process of the third ECL includes the radical formation through the hydrogen abstraction reaction by HO• . As shown in Fig. 8, the third ECL intensity for 2-propanol is approximately 1.5 times higher than that for 1-propanol. It is known that the stability of the 2-propanol radical is higher than that for 1-propanol due to the lowering of the C–H bond dissociation energies for a carbon bound to an alcohol oxygen [27,28]. Hence, the rate for the radical formation reaction for 2-propanol was supposed to be faster than that for the 1-propanol radical due to the lower dissociation energy of the C–H bond. Moreover, the rate for the generation of Ru(bpy)3 2+* might be also higher due to the high stability of the 2-propanol radical, resulting in the higher peak intensity for the third ECL. Therefore, the process of ECL reaction at the third ECL peak can be described by following equation: M(OH• ) + Co–R–H → M + Co–R• + H2 O

(18)

Ru(bpy)3 3+ + Co–R• → Ru(bpy)3 2+∗ + product

(2)

Ru(bpy)3 2+∗ → Ru(bpy)3 2+ + hν

(3)

Co–R–H is the co-reactant, and M is an active site at the BDD electrode. In the case of water discharge at the BDD electrode, absorbed HO• might abstract a hydrogen atom from the coreactant, resulting in the reducing radical Co–R• to form the excited state of Ru(bpy)3 2+* .

The oxidation of organic compounds can be mediated by HO• and the organics could be fully oxidized to CO2 . By use of this unique characteristic at BDD, numerous studies related to the electrochemical treatment of toxic substances (e.g. phenol and 4-chlorophenol) have been performed [9,11]. However, at the BDD electrode, the rate for the formation of HO• has not been reported. The rate for the HO• generation is thought to increase with increasing potential. In this study, the increase of the rate for the formation of OH radical with potential was confirmed by the dependence of the peak potential of the third ECL on the co-reactant concentration (Figs. 5 and 6). The rate of the HO• formation can be roughly estimated from the light intensity at the third ECL peak. By use of the number of photons corresponding to the light intensity at the third ECL peak and the reaction schemes of Eqs. (2) and (18), the ratio of the current density for the HO• formation was calculated to be approximately 1.1 × 10−6 %. The more accurate values of the rate constant for the HO• generation at BDD will be discussed using the results for the analysis of the potential dependence of HO• generation and the reaction efficiency for ECL at the BDD electrode. Meanwhile, in this study, the behavior of the potential dependence of the HO• generation that starts at approximately 2.6 V could be clearly visualized using the ECL system with Ru(bpy)3 2+ and AA.

4. Conclusion An ECL reaction of the Ru(bpy)3 2+ /co-reactant system in the extremely high-potential region (>2.6 V versus Ag/AgCl) was proved using the BDD electrode. At the BDD electrode, three ECL waves (1.25, 2.3 and 3.72 V) were observed in cyclic voltammograms for 20 mM ascorbic acid. At each peak potential, the process of the formation of the reducing radical that is necessary to generate the excited state of Ru(bpy)3 2+* is different. At the ECL peak observed at 1.25 and 2.3 V, the reducing radicals were found to be generated through homogeneous electron-transfer between Ru(bpy)3 3+ and AA and the direct oxidation of AA at the electrode surface, respectively. For the ECL reaction at 3.72 V that was observed only at the BDD electrode, it was clarified that the co-reactant radical was formed through the hydrogen abstraction reaction with the hydroxyl radical (HO• ) generated during the oxygen evolution reaction by analysis using various co-reactants that show different reactivity for the hydrogen abstraction reaction by OH radical. In consequence, we could first visualize the behavior of the potential dependence of the OH• generation by use of the ECL system. Moreover, it was confirmed that the HO• plays an important role for the electrochemical treatment of the organics (i.e. aromatic compounds) at the high-potential region (>3.0 V). By use of the HO• generation at BDD, it is possible to oxidize various organics and form the corresponding radicals with high efficiency. By combining these radicals with the ECL system using Ru(bpy)3 2+ , it can be possible to detect the different types of co-reactant molecules that can not be observed in the typical ECL at

K. Honda et al. / Electrochimica Acta 51 (2005) 588–597

other electrodes (e.g. polycrystalline Pt). Propanol that gives higher intensity for the third ECL is a promising candidate for the high-sensitivity detection. Moreover, by introducing the ECL detection system in HPLC, there is a possibility to improve the performance of the selective detection by controlling the electrode potential in the higher potential region over 2.6 V.

References [1] J. Xu, M.C. Granger, Q. Chen, J.W. Strojek, T.E. Lister, G.M. Swain, Anal. Chem. News Features 1 (1997) 591. [2] W. Siangproh, P. Ngamukot, O. Chailapakul, Sens. Actuators B 91 (2003) 60. [3] T.N. Rao, A. Fujishima, Diamond Relat. Mater. 9 (2000) 3. [4] M. Komatsu, A. Fujishima, Bull. Chem. Soc. Jpn. 76 (2003) 927. [5] K. Yoo, B. Miller, R. Kalish, X. Shi, Electrochem. Solid-State Lett. 2 (1999) 233. [6] R. Uchikado, T.N. Rao, D.A. Tryk, A. Fujishima, Chem. Lett. 2 (2001) 144. [7] K. Honda, M. Yoshimura, T.N. Rao, A. Fujishima, J. Phys. Chem. B 107 (2003) 1653. [8] Y. Zu, A.J. Bard, Anal. Chem. 72 (2000) 3223. [9] M.A. Rodrigo, P.A. Michaud, I. Duo, M. Panizza, G. Cerisola, C. Comninellis, J. Electrochem. Soc. 148 (2001) 60. [10] B. Marselli, J. Garcia-Gomez, P.-A. Michaud, M.A. Rodrigo, C. Comninellis, J. Electrochem. Soc. 150 (2003) 79.

597

[11] J. Iniesta, P.A. Michaud, M. Panizza, G. Gerisola, A. Aldaz, C. Comninellis, Electrochim. Acta 46 (2001) 3573. [12] A. Kraft, M. Stadelmann, M. Blaschke, J. Hazard. Mater. B 103 (2003) 247. [13] T. Yano, D.A. Tryk, K. Hashimoto, A. Fujishima, J. Electrochem. Soc. 145 (1998) 1870. [14] H. Notsu, I. Yagi, T. Tatsuma, D.A. Tryk, A. Fujishima, J. Electroanal. Chem. 429 (2000) 31. [15] M.C. Granger, M. Witek, J. Xu, J. Wang, M. Hupert, A. Hanks, M.D. Koppang, J.E. Butler, G. Lucazeau, M. Mermoux, J.W. Strojek, G.M. Swain, Anal. Chem. 72 (2000) 3793. [16] D.T. Sawer, A. Sobkowiak, J.L. Robert, Electrochamistry for Chemist, John Wiley & Sons Inc., NY, 1995 (Chapters 3 and 12). [17] R.P. Akkermans, M. Wu, C.D. Bain, M. Fidel-Su´arez, R.G. Compton, Electroanalysis 10 (1998) 613. [18] P. Karabinas, D. Jannakoudakis, J. Electroanal. Chem. 160 (1984) 159. [19] L.C. Hian, K.J. Grehan, R.G. Comptomn, J.S. Foord, F. Marken, Diamond Relat. Mater. 12 (2003) 590. [20] J. Maruyama, I. Abe, Electrochim. Acta 46 (2001) 3381. [21] G.M. Swain, J. Electrochem. Soc. 141 (1994) 3382. [22] F. Kanoufi, Y. Zu, A.J. Bard, J. Phys. Chem. B 105 (2001) 210. [23] R.S. Nicholson, I. Shain, Anal. Chem. 36 (1964) 706. [24] M. Yoshimura, K. Honda, T. Kondo, T.N. Rao, D.A. Tryk, A. Fujishima, Electrochim. Acta 47 (2002) 4387. [25] A. Fujishima (Ed.), A Hand Book of Electrochemistry, Maruzen Co. Ltd., Tokyo, 2000 (Chapter 7). [26] M. Ue, K. Ida, S. Mori, J. Electrochem. Soc. 141 (1994) 2989. [27] L. Nelson, O. Rattigan, R. Neavyn, H. Sidebottom, J. Treacy, O.J. Nielsen, Int. J. Chem. Kinet. 22 (1990) 1111. [28] A. Hapipo˘glu, Z. Cinar, J. Mol. Struct. (Theochem.) 631 (2003) 189.