Electrochemical and infrared study of electro-oxidation of dimethyl ether (DME) on platinum polycrystalline electrode in acid solutions

Available online at www.sciencedirect.com

Electrochimica Acta 53 (2008) 6093–6103

Electrochemical and infrared study of electro-oxidation of dimethyl ether (DME) on platinum polycrystalline electrode in acid solutions Yi Zhang a , Leilei Lu a,b , Yujin Tong a , Masatoshi Osawa a , Shen Ye a,c,∗ a

b

Catalysis Research Center, Hokkaido University, Sapporo, Japan Department of Chemistry, Faculty of Science, Harbin Institute of Technology, Harbin, China c PRESTO, Japan Science and Technology Agency (JST), Japan

Received 14 October 2007; received in revised form 11 January 2008; accepted 31 January 2008 Available online 21 February 2008

Abstract Electro-oxidation of dimethyl ether (CH3 OCH3 , denoted as DME below) on Pt polycrystalline electrode has been investigated by electrochemical and in situ infrared (IR) measurements in acid solutions. A reaction intermediate species, (CH3 OCH2 –)ad , has been observed in the low potential region as an initial product for dehydrogenation process of DME on Pt electrode surface. This species is subsequently decomposed to adsorbed carbon monoxide (CO) and finally oxidized to carbon dioxide (CO2 ) in higher potential region. The time-resolved IR measurement is employed to follow the transient process of the formation and decomposition of the intermediate on Pt electrode surface. Based on these electrochemical and IR spectroscopic results, a reaction scheme for DME electro-oxidation process is proposed. © 2008 Elsevier Ltd. All rights reserved. Keywords: Electro-oxidation; Dimethyl ether (DME); Fuel cell; Platinum electrode; In situ IR spectroscopy; Flow-cell

1. Introduction Fuel cells are promising alternative power generation devices that generate electricity with higher efficiency and lower pollution than traditional fuel combustion based power systems. As a direct-fed liquid fuel with high energy density, methanol has been paid much attention in the past decades. However, the application of methanol in proton exchange membrane fuel cell (PEMFC) system still suffers from some barriers, such as degradation of electrode catalysts, fuel loss due to the methanol crossover to cathode as well as safety problem by high toxicity of methanol. Dimethyl ether (DME, CH3 –O–CH3 ) presents some interesting options for electrochemical power generation acting as fuel in PEMFC [1–10]. As the simplest ether, DME contains two C O bonds but no C C bond and is expected to show high electro-oxidation activity. DME can be stored in liquid state under a modest pressure (ca. 5 atm). This combines the advantages of high energy density and easy introduction into fuel

∗

Corresponding author at: Catalysis Research Center, Hokkaido University, Sapporo, Japan. E-mail address: [email protected] (S. Ye). 0013-4686/$ – see front matter © 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.electacta.2008.01.109

cell in gas phase without pumping system. Furthermore, DME is safe in handling due to its low toxicity in comparison with methanol. The performance of the direct DME fuel cell has been evaluated in the past few years. M¨uller et al. reported that high Faradic efficiency for the fuel cell was obtained at relatively high temperature (130 ◦ C) [1]. Although part of DME molecules crossed over to the cathode side, no oxidation occurred there. They found that DME was mainly oxidized to CO2 with trace quantities of methanol as by-product. They proposed that the electro-oxidation of DME is initiated from the hydrolysis of one methyl group followed by cleaving the C O C bond. Tsutsumi et al. reported that formic acid was the main by-product in the DME electro-oxidation [10]. They suggested that DME initially decomposes through the two methyl groups, which are hydrolyzed simultaneously on the Pt surface. Further oxidation involves the cleavage of the C O C bond. Recently, Mizutani et al. observed a small amount of methyl formate and methanol in the anode exhaust of the direct DME fuel cell [6]. Most of these works showed that main product in the DME electro-oxidation process is CO2 while reaction intermediates are quite different. In situ IR vibrational spectroscopy combined with the electrochemical characterization is a powerful probe to clarify the reaction scheme in electrocatalytic reaction, including molecu-

6094

Y. Zhang et al. / Electrochimica Acta 53 (2008) 6093–6103

lar information about the reaction intermediate. By using in situ IR reflectance absorption spectroscopy (IRRAS) in a thin-layer cell, Kerangueven et al. reported that the intermediates of DME electro-oxidation on a Pt electrode in H2 SO4 solutions are linearly bonded carbon monoxide (COL ), bridge bonded CO (COB ), and adsorbed COOH species. They proposed that DME was hydrolyzed and partially oxidized to CH3 OH, HCOOH and CO2 [11]. Shao et al. investigated the electro-oxidation process of DME on a Pt thin film electrode in 0.1 M HClO4 solutions by ATR-IR measurement [12]. They reported a number of reaction intermediates for DME decomposition near the hydrogen adsorption region and found that reaction intermediates and products are dependent on the DME concentration. Recently, Liu et al. investigated electrochemical behavior of DME on a sputtered Pt electrode by ATR-IR measurement [13] and reported a number of reaction intermediates such as adsorbed CO, –COOH and –CHO. Although these measurements were carried out in acid solutions under similar conditions, most of the in situ IR characterization results are quite different. Apparently, this is not due to the different kinds of electrolyte used (HClO4 or H2 SO4 ) in these works. One of possible reasons is considered to be the diffusion problem for these measurements, which may change the IR spectral features. It is well known that the concentration of reactant and products as well as soluble intermediates can be easily changed during electrochemical reaction in an environment where diffusion process is restricted. As discussed previously [14], this is considered as main reason that formate, which is observed as a reaction intermediate for many C1 molecules, such as methanol [15] and formic acid [16], could not be observed by in situ IR measurement using the traditional thin-layer geometry. Normally, in situ IR measurement using Kretschmann ATR configuration can greatly improve the diffusion and mass transport processes but may also suffer from it if the cell volume is too small for long-time electrolysis in a high current density. Recently, Chen et al. reported a kinetic study for electro-oxidation of formic acid by in situ ATR-IR spectroscopy using a flow-cell [17]. In the present study, we aim to clarify the reaction scheme of DME electro-oxidation on Pt electrode in acid solutions, by using electrochemical and in situ IR methods under more reliable and reproducible conditions. The potential dependence of DME decomposition has been electrochemically evaluated. In order to avoid the influence from diffusion and accumulation problems from soluble species, a spectroscopic flow-cell with Kretschmann ATR configuration is newly designed and employed in the in situ IR measurement in comparison with conventional one used in our group. Supply of reactant with constant concentration and quick removal of soluble products in solution side will enable us to exactly investigate the species on the electrode surface. Through this in situ IR measurement, the transient processes of DME decomposition process at different potentials are clearly observed. Based on these experimental results, the reaction mechanism of DME on Pt electrode is discussed. We expect that the present research can provide a clue for designing new electro-catalysts for direct DME fuel cells.

2. Experimental The details of the in situ IR measurements with the Kretschmann ATR configuration have been described elsewhere [14–16]. A Pt thin film was chemically deposited on a hemicylindrical silicon (Si) prism surface as a working electrode. The fabrication procedure is briefly described as follows. Si surface was firstly activated by contacting with 0.5% HF containing 1 mM PdCl2 , which can improve the adhesion of the deposited metal film to the Si substrate. Then, 50-nm thick Pt thin film was deposited with a commercially available plating solution (LECTROLESS PT100, Electroplating Engineering of Japan) at 60 ◦ C. The surface morphology and thickness of the Pt thin film are critical to get a reproducible high-quality IR spectrum. For example, Shao et al. investigated DME electro-oxidation on the Pt film electrode by ATR-IR spectroscopy, however, their in situ IR spectra for adsorbed CO show a unusual bipolar band shape making further analysis difficult [12]. Potentials were controlled by a potentiostat (EG&G PARC model 263A) and presented with respect to the reversible hydrogen electrode (RHE). The in situ IR measurements were carried out using a BioRad FTIR-575C spectrometer equipped with a MCT detector. The IR spectra were recorded in the dynamitic mode with a spectral resolution of 4 cm−1 . Five interferograms were co-added to each spectrum. It takes 0.98 s to get one spectrum in present work. All IR spectra shown were normalized to the spectrum taken at 0.05 V in electrolyte solution without DME. A spectroscopic flow-cell for in situ IR measurements was newly made to remove the possible influence from accumulation of reaction intermediates and products dissolved in the electrolyte. Fig. 1 shows its schematic structure for the flowcell with three-electrode design. The flow of solution in the cell is driven by pressure of Ar or DME gas through a threeway stopcock. The cell volume is ca. 4 ml and solution can be exchanged as fast as 20 ml min−1 , i.e., the solution inside could be exchanged by more than five times per minute. A conventional static spectroscopic cell [15] with a cell volume of ca. 30 ml is also employed for a comparison in the experiment. The supporting electrolyte solutions were prepared from Milli-Q water (>18 M) and analytical grade HClO4 or H2 SO4 . Electrolyte solutions were deaerated with Ar for 30 min prior to use. After electrochemically cleaning the electrode surface by successive potential cycles between 0.05 and 1.5 V in the blank electrolyte, an IR reference spectrum was recorded at 0.05 V. Then, the solution was saturated with DME (ca. 1.65 M)

Fig. 1. A simplified scheme for the flow IR cell used in the work.

Y. Zhang et al. / Electrochimica Acta 53 (2008) 6093–6103

[1] by bubbling DME (99.99%, Sumitomo Chemicals Corp.) through the solution. All experiments were carried out at room temperature (22 ◦ C). Recently, Watanabe and co-workers reported that dimethoxymethane, CH2 (OCH3 )2 , which has a similar ether-like structure to DME, can be largely hydrolyzed to CH3 OH and HCHO in acidic solution within 10 days [18]. We found that the hydrolysis of DME is negligible under the present experimental conditions. 3. Results and Discussion 3.1. Electrochemical results of DME electro-oxidation Fig. 2 shows the cyclic voltammograms (CVs) of Pt thin film electrode between 0.05 and 1 V in 0.1 M HClO4 solution before (dashed) and after (solid/dash-dotted) DME saturation in (a) a spectroscopic flow-cell (Fig. 1) and (b) a conventional static spectroscopic cell. In the blank electrolyte solution, the Pt electrodes in both cells give typical CVs behaviors for hydrogen adsorption/desorption (0.05–0.4 V) as well as the oxide formation and reduction (0.6–1 V). The real surface area of the Pt electrode was estimated to be ca. 9.05 cm2 , assuming that an average coverage of hydrogen adsorption on Pt surface is 210 ␮C cm−2 . The roughness factor for the Pt electrode was estimated to be ca. 6 from its apparent area (ca. 1.50 cm2 ). In the first positive-going sweep for DME-saturated HClO4 solution from 0.05 V in the flow-cell (Fig. 2a, solid), the Pt electrode gives a broad anodic current with two peaks around 0.10 and 0.35 V superimposed on the hydrogen desorption wave. The excess charge over the blank CV (Fig. 2a, dashed) in the hydrogen desorption region is approximately 250 ␮C cm−2 . Moreover, an oxidation peak at 0.75 V with a shoulder at 0.68 V is observed in the positive-going sweep while two oxidation peaks are found around 0.70 and 0.55 V in the reversed potential sweep from 1.0 V. In the second cycle (Fig. 2a, dash-dotted), the anodic current observed between 0.05 and 0.40 V is largely decreased while other voltammetric features observed in high potential region are similar to those in the first sweep. It should be mentioned here that the second and successive potential cycles

6095

give exactly same CVs, indicating that the second CV can be regarded as a steady state during potential sweep. These anodic currents clearly indicate that DME molecules are electrochemically oxidized on Pt electrode surface in a wide potential region. Recently, we also investigated the electrochemical behaviors of DME on platinum single crystal electrodes and found that the large anodic currents in the high potential region in both positiveand negative-going sweeps, were only observed on Pt(1 0 0) surface or Pt(h k l) high index surfaces with (1 0 0) terrace[19], implying that (1 0 0) facets on the polycrystalline Pt thin film electrode may give major contribution to the electro-oxidation activity in the high potential region. The voltammetric behaviors in flow-cell (Fig. 2a) with those observed in convention static-cell (Fig. 2b) are generally similar to each other. In fact, it is found that the differences in the electrochemical behaviors between the two cells are small unless for case of a long electrolysis at a high potential, where more soluble intermediates and products from the electro-oxidation of the DME molecules can be accumulated in the convention static-cell. The voltammetric behaviors observed in Fig. 2 are obviously different from those recently reported by Shao et al. [12] in a DME-saturated HClO4 solution where much higher oxidation currents in contrast to anodic current in hydrogen adsorption in blank solution were obtained. The volume of IR spectroscopic cell used by Shao et al. is ca. 5 ml [12], smaller than typical one used in our group (ca. 30 ml), therefore, may receive more influence from soluble intermediates and products even the surface area of the Pt electrode under the Kretschmann ATR configuration is comparable with the present experiment. The hydrogen desorption/adsorption wave on Pt electrode is partially suppressed during potential sweep between 0.05 and 1.0 V in DME-saturated solution (Fig. 2). In order to clarify this issue in details, electrochemical behaviors of DME on Pt electrode surface are evaluated by potential cycle only in the hydrogen region between 0.05 and 0.4 V (Fig. 3, solid lines). The sequence number of the potential cycle is also shown in Fig. 3. In the first positive-going sweep, an anodic peak is observed around 0.2 V on the hydrogen desorption wave. The currents

Fig. 2. The first two consecutive cyclic voltammograms of the Pt electrode in 0.1 M HClO4 in (a) a flow-cell and (b) a static-cell; scan rate is 25 mV/s. Without DME (dashed); first cycle (solid) and second cycle (dash-dotted) in saturated DME solution.

6096

Y. Zhang et al. / Electrochimica Acta 53 (2008) 6093–6103

Fig. 3. The consecutive cyclic voltammograms of the Pt electrode in 0.1 M HClO4 saturated by DME; scan rate is 25 mV/s; blank (dashed).

in the hydrogen region observed in the second and following successive CVs become fairly small than that in blank solution (Fig. 3, dashed line) and decrease slightly for repeated potential cycles. Assuming the anodic currents in the second and successive CV here are mainly attributed to the hydrogen desorption process, hydrogen coverage is estimated as ca. 0.2 after 10 potential cycles between 0.05 and 0.40 V. This suggests that certain amounts of intermediates are formed and adsorbed on the Pt electrode surface in the dissociation process of DME. As will be shown later in our in situ IR measurement, (CH3 OCH2 –)ad and COL are formed as reaction intermediates through stepwise decomposition reaction in the potential region. These species partially block the hydrogen adsorption/desorption. It is interesting to note that the decrease of the hydrogen coverage becomes very slow after the first potential cycle. The identical CV behaviors in the first positive-going sweep can be observed repeatedly by holding potential at 1.5 V for a short time, for example, 2 s, which can remove all the surface species from the Pt surface by oxidation reaction. Fig. 4 shows the dependence on the scan rate for the first CV (10–500 mV/s) from 0.05 V on the Pt electrode in a

Fig. 4. The first potential sweep of the Pt electrode in 0.1 M HClO4 saturated by DME; scan rate is 10, 25, 50, 100, 200, 500 mV/s, respectively; blank (500 mV/s, dashed). The current density is normalized by scan rate.

DME-saturated HClO4 solution between 0.05 and 1.0 V. The current densities are normalized by dividing the scan rate for comparison. Generally, the normalized currents should be identical if the corresponding reactions are completely reversible processes of surface absorbed species, such as hydrogen adsorption/desorption and oxide formation/reduction on Pt electrode surface. As shown in Fig. 4, when the scan rate is lower than 50 mV/s, an additional anodic peak can be clearly observed on the hydrogen desorption wave. With decrease of the scan rate, the anodic peak height increases and the peak position also shifts to negative direction. On the contrary, as the scan rate is higher than 100 mV/s, the additional anodic current almost disappears and current profiles in the hydrogen adsorption region are similar to that observed in blank solution, indicating that the DME decomposition reaction on the Pt electrode surface proceeds to a very low extent under this condition. Potential sweep with a lower scan rate, which will polarize for longer time in the potential region passed, thus enables more DME molecules to be decomposed if active potential for DME reaction exist in the potential region. Moreover, the oxidation peaks in high potential region also present similar dependence of scan rate. The lower the scan rate, the higher the peak height and the more negative the peak position for both positive- and negative-going sweeps. As the scan rate is decreased to 10 mV/s, large oxidation peaks in the high potential region can be observed in either positive- and negative-going sweeps, indicating that the direct oxidation process takes place at these regions in addition to the oxidation of the adsorbed species formed on the low potential region. These results clearly demonstrate that the decomposition and oxidation process of DME on Pt surface is a kinetically controlled process. To quantitatively understand the effect of absorption potential on the DME decomposition process, a potential program (Fig. 5a, inset) is applied on Pt electrode to evaluate the amounts of surface adsorbed species at various potentials in 0.1 M HClO4 solution saturated by DME. After the sequential polarization at 1.5 V (2 s) and 0.05 V (2 s), the Pt electrode is kept at a certain potential for 300 s, and then the CV is recorded with a scan rate of 500 mV/s between 0.05 and 1.0 V. The potential is firstly swept to negative direction to characterize the hydrogen adsorption state. Fig. 5 shows first (solid lines) and steady CVs (dash-dotted lines) after Pt electrode is kept at (a) 0.05 V, (b) 0.2 V, (c) 0.4 V and (d) 0.6 V. A CV observed in blank solution (Fig. 5, dashed line) is also shown for comparison. The relatively high scan rate (500 mV/s) is employed here to reduce the contribution from the direct oxidation of solution species and/or the adsorbed species generated during the potential sweep. It is found that all CVs show almost identical steady profile after the first potential cycle regardless of adsorption potential, where the suppression of the hydrogen wave and increase of an oxidation current still exist in comparison with blank CV. This is understood since surface adsorbed species are repeatedly produced in the continuous potential cycles even at such high scan rate (Figs. 4 and 5). After the potential is held at 0.05 V for 300 s, the subsequent potential sweep shows that the hydrogen adsorption is partially suppressed and a broad oxidation peak appears around 0.8 V

Y. Zhang et al. / Electrochimica Acta 53 (2008) 6093–6103

Fig. 5. The Pt electrode is held at the absorption potential of (a) 0.05 V, (b) 0.2 V, (c) 0.4 V and (d) 0.6 V, respectively, and followed by a fast CV (500 mV/s) adsorption time: 300 s. Blank (dashed); CV in steady state (dash-dotted). The inset shows the potential program. Electrolyte: 0.1 M HClO4 saturated by DME.

(Fig. 5a). The suppression of hydrogen wave after adsorption at 0.05 V is close to that of the steady CV. In addition, the oxidation peak around 0.8 V is slightly higher than that of the steady CV. These results indicate that a very small amount of DME molecules are decomposed at 0.05 V. As the adsorption potential is higher than 0.1 V, changes observed in the hydrogen adsorption and the oxidation peak around 0.8 V become more obvious than that of 0.05 V. CVs after adsorption at 0.2 and 0.4 V are quite similar (Fig. 5b and c). Large changes are observed at these potentials in comparison with those of blank and steady CVs. However, as the adsorption potential is kept at 0.6 V for 300 s (Fig. 5d), the oxidation peak is largely decreased and the CV obtained is almost same as that of steady CV, indicating that the coverage of surface species formed at this potential is low and comparable to that generated in the continuous potential cycles.

6097

Fig. 6a summarizes the charge quantities for suppressed hydrogen adsorption (QSH , circles) and oxidation peak (Qoxi , squares) at various adsorption potentials. The ratio (RQ ) of Qoxi to-QSH is calculated at each potential (Fig. 6b) to estimate the number of electrons exchanged per Pt site during the oxidation reaction. As shown in Fig. 6a, Qoxi and QSH are almost constant and found to be approximately 280 and 180 ␮C cm−2 , respectively, in the potential region between 0.1 and 0.5 V. Consequently, a nearly constant value of RQ (ca. 1.5) is obtained between 0.1 and 0.5 V. As will be shown later in our in situ IR results, the adsorbed CO (COL and COB ) on Pt electrode surface are mainly observed after a polarization for 300 s in the hydrogen adsorption region (for example, see Figs. 12 and 13). Thus, this value of 1.5 is reasonable if the CO adlayer is a 2:1 mixed structure of COL and COB species on Pt electrode surface if one assume that COL and COB can give the ratio of 2 and 1, respectively, for their oxidation to CO2 . At the potential of 0.55 and 0.6 V, Qoxi decreased to 225 and 130 ␮C cm−2 companying with the decrease of QSH to 125 and 50 ␮C cm−2 , respectively. In fact, a steady anodic current is already observed around 0.6 V while such steady anodic current is almost negligible below 0.5 V (results are not shown here). As also shown from in situ IR results, the IR band intensity of CO on Pt electrode surface is largely decreased if the potential is kept on the 0.6 V for 300 s. The oxidation of intermediates from decomposition process occurs simultaneously at the potential higher than 0.5 V, giving lower QSH and Qoxi . Deviation of RQ from 1.5 at 0.6 V (Fig. 6b) can be considered as a possible contribution from direct oxidation current of DME at high potentials. Furthermore, relatively large value of RQ at 0.05 V (ca. 3.4) can be ascribed to non-CO intermediates generated during DME decomposition at very low potential region, which is partially verified from the in situ IR results (see below, Figs. 12 and 13). It is well known that the anion adsorption can greatly affect the electrochemical activities of Pt electrode toward small organic molecules oxidation (ca. CH3 OH [20], HCOOH [21]). The electro-oxidation of DME is also evaluated in sulfuric acid solution. Fig. 7 shows first (solid line) and second CVs (dashdotted line) of Pt electrode in 0.5 M H2 SO4 solution saturated by DME and blank solution (dashed line) in the flow-cell. In comparison with the CV observed in 0.1 M HClO4 (Fig. 2a), a similar anodic current peak at near 0.4 V could also be observed in the first potential sweep and disappear in the second potential cycle, which is consistent with the results of M¨uller et al. [1] Additionally, the existence of the anodic currents initiated from 0.05 V presents in a wide potential region. However, the anodic current intensity in the high potential region (ca. 0.6–0.8 V) is much smaller. Similar profiles of voltammograms were also reported previously by other groups in H2 SO4 solution [1,5,11]. It is known that bisulfate anion adsorbs more strongly on the Pt electrode surface than perchlorate anion. This suggests that adsorption of bisulfate can also significantly affect the oxidation rate of DME on Pt electrode. In the present paper, the studies mainly focus on the behavior of DME electro-oxidation in perchloric acid solution, in which relative higher activity is revealed.

6098

Y. Zhang et al. / Electrochimica Acta 53 (2008) 6093–6103

Fig. 6. (a) The charge calculated from suppressed hydrogen (QSH ) and oxidation current (Qoxi ) in the CVs observed in Fig. 5. (b) Ratio of Qoxi and QSH at different adsorption potentials. Electrolyte: 0.1 M HClO4 saturated by DME.

Fig. 7. The first two consecutive cyclic voltammograms for a Pt thin film electrode in 0.5 M H2 SO4 in a flow-cell; without DME (dashed); first cycle (solid) and second cycle (dash-dotted) in saturated DME solution. Scan rate is 25 mV/s.

3.2. In situ IR studies for DME electro-oxidation Fig. 8 shows in situ IR spectra (4000–1300 cm−1 ) on the Pt electrode recorded simultaneously with CV measurement (25 mV/s) as shown in Fig. 2a in the flow-cell, for the (a) first positive- and (b) subsequent negative-going sweeps, respectively. Several IR bands are observed in a wide frequency region

for DME electro-oxidation process. These in situ IR spectra are recorded every 0.025 V and only part of them are shown in Fig. 8. Since the present in situ IR measurement using Kretschmann ATR configuration on Pt thin film is impossible to detect CO2 near the electrode surface [15,16], spectral features between 2500 and 2200 cm−1 , where C O stretching mode of CO2 locates, are cut in the present paper. The IR background spectrum is recorded in a DME-free HClO4 solution at 0.05 V. The upward and downward bands in the in situ IR spectra correspond to the vibrational modes of the surface species generated and lost at the each potential, respectively, in comparison with the blank solution. In the beginning of the first positive-going sweep (Fig. 8a), no IR band can be observed in the potential region below 0.1 V. When the potential becomes more positive than 0.2 V, a number of upward peaks appear at 1322, 1445 and 1480 cm−1 , and grow quickly with potential to a maximum around 0.3 V and then decrease to zero around 0.6 V. As will be described later, the peak position for the band at near 1322 cm−1 changes with coverage and potential (Fig. 13b), indicating that this vibrational mode of the species has strong interaction with electrode surface. For convenience, we assume the peak at 1322 cm−1 locates at same position in the following text. On the other hand, two upward IR bands appear around 2000 cm−1 at 0.3 V and 1810 cm−1 at 0.5 V, respectively. With increase of potential,

Fig. 8. The in situ IR spectra recorded during oxidation of DME on a Pt thin film electrode in 0.1 M HClO4 for the first potential cycle in Fig. 2a; (a) positive-going sweep and (b) negative-going sweep.

Y. Zhang et al. / Electrochimica Acta 53 (2008) 6093–6103

the intensities of these two peaks increase and peak positions shift to higher frequency. These two peaks have been previously observed for electro-oxidation of many C1 molecules, such as CH3 OH [15] and formic acid [16], on Pt electrode surface and can be undoubtedly assigned to the linearly adsorbed CO (COL ) and bridge adsorbed CO (COB ), respectively, on the electrode surface. These peaks have been also reported in previous in situ IR studies of DME electro-oxidation reaction [11,12]. Thus, CO can be regarded as one of reaction intermediates from the decomposition and electro-oxidation processes of DME. CO bands become weaker when potential is higher than 0.6 V and disappear around 0.7 V. Furthermore, two downward bands, a broad one around 3500 cm−1 and a small one at 1625 cm−1 , are observed in pair when potential is higher than 0.3 V, which can be assigned to OH stretching, ␯ (OH), and HOH bending, ␦ (HOH) modes, respectively, of interfacial water on Pt electrode surface [16]. The downward shape indicates that water molecules are repelled from the Pt interface referred to blank solution, where DME decomposition reaction occurs and leads to the adsorption of other surface species. On the other hand, an upward band at 3655 cm−1 can be observed, for example, at 0.6 V, where surface CO coverage is higher (Fig. 8a, also see Fig. 12). This band has been attributed to OH stretching of water molecules embedded in the CO adlayer on Pt electrode surface [16]. No band is observed at 1.0 V except the two downward ones at 3500 and 1625 cm−1 (Fig. 8a). In the subsequent negativegoing sweep from 1.0 V (Fig. 8b), two upward bands at 2024 and 1830 cm−1 , corresponding to COL and COB , respectively, appear from 0.6 V. The intensities of these two peaks increase with deceasing potential and become nearly constant when potential is more negative than 0.2 V. The upward bands around 1480, 1445 and 1322 cm−1 appear from the similar potential and become approximately invariable in the potential region below 0.5 V. The hysteresis for the IR intensity is clearly observed in the first potential sweep (Fig. 8a and b). Fig. 9 summarizes the integrated band intensities for the COL and other two bands at 1480 and 1322 cm−1 as a function of electrode potential during the first potential sweep. The integrated intensities for the later two bands are multiplied by 5 times for easier comparison. COL appears around 0.2 V and reaches its

6099

Fig. 9. The integrated band intensities of main absorbed species taken from a series of in situ IR spectra acquired simultaneously with the first potential cycle. The intensity of bands at 1322 and 1480 cm−1 are multiplied by 5 times.

maximum intensity around 0.6 V and decreases to zero around 0.725 V. When the potential is swept toward negative direction, COL appears around 0.6 V and reaches to maximum and keeps constant from ca. 0.2 V. After one potential cycle in the DME-saturated solution, the Pt surface is highly covered by CO adlayer. It should be mentioned that our results are quite different from those reported by Shao et al. where a very strong CO band was observed even at very low potential (−0.2 V vs. Ag/AgCl) in the first positive-going sweep [12]. On the other hand, it is obvious from Fig. 9 that the intensities of two bands at 1480 and 1322 cm−1 show almost identical potential dependence in either positive- or negative-going potential sweep. Maximum intensity for two bands is observed at same potential (ca. 0.32 V) in the positive-going sweep and becomes weaker but almost constant in the negative-going sweep when potential is less positive than 0.5 V. The band at 1445 cm−1 also demonstrates the similar potential dependence. The synchronous changes with electrode potential for the three bands suggest that these bands should be related to the same non-CO surface intermediate during DME decomposition process. It is interesting to note that, in the first positive-going sweep, the potential for the maximum intensity of the bands at 1480 and 1322 cm−1 locates

Fig. 10. The in situ IR spectra recorded during oxidation of DME on a Pt thin film electrode in 0.1 M HClO4 for the second potential cycle in Fig. 2a; (a) positive-going sweep and (b) negative-going sweep.

6100

Y. Zhang et al. / Electrochimica Acta 53 (2008) 6093–6103

Fig. 11. The integrated band intensities of main absorbed species taken from a series of in situ IR spectra acquired simultaneously with the second potential cycle. The intensities of bands at 1322 and 1480 cm−1 are multiplied by 5 times.

ca. 0.2 V negative than that of COL , implying that the reaction intermediate is formed before that of COL adlayer. Figs. 10 and 11 show the IR spectra and their band intensities recorded simultaneously with the second potential cycle of CVs (Fig. 2a, dash-dotted line), respectively. It is found that second and successive potential cycles (not shown here) give identical CV and IR spectra, and thus can be regarded as a steady state during the potential cycle. This corresponds well to the electrochemical behaviors observed in CV showed in Fig. 2. If one compares the results shown in Figs. 8 and 10, or Figs. 9 and 11, the differences are mainly observed in the positive-going sweep. In the second positive-going sweep, Pt surface is covered by CO from beginning (0.05 V). As the potential increases, the intensity of CO slightly increases and starts to decrease from 0.6 V and disappear around 0.8 V. The intensities for bands at 1480 and 1322 cm−1 in the second positive-going sweep are nearly constant in the potential region less positive than 0.5 V and then quickly decrease to zero. In the subsequent negative-going sweep, the spectral features for both COL and the other surface species are similar to those observed in the first negative-going one (Fig. 10b). The potential dependences observed for the IR bands between 1500 and 1300 cm−1 are quite unique for decomposition and electro-oxidation process of DME on Pt electrode surface. To our knowledge, this kind of behavior has not been observed in electro-oxidation process of other C1 molecules and should be specially related to the DME decomposition process. In order to understand the reaction scheme for DME electro-oxidation, it is important to assign these peculiar IR bands observed. Several studies on the adsorption and oxidation of DME on heterogeneous metal oxide catalyst surfaces in gas phase have been carried out. Yates and co-workers investigated adsorption of DME on A12 O3 surfaces and found the formation of surface methoxy (–OCH3 ) species at 250 K giving bands at 2962 cm−1 (␯as CH3 ), 2848 cm−1 (␯s CH3 ), 1477 cm−1 (␦as CH3 ) and 1454 cm−1 (␦s CH3 ) as well as 1157 cm−1 (␦s CH3 ) [22]. Solymosi et al. investigated interaction of DME with alumina-

supported Pt metals and reported similar peaks related to adsorbed methoxy species and a new band related to adsorbed CO species on the catalyst surfaces [23]. The electrocatalytic behaviors of DME on Pt electrodes have been studied by in situ IR spectroscopy [11–13]. Since the intensities for C H stretching modes in the potential difference IR spectra on the electrode surface, especially for the small organic molecules in electrolyte solution, are usually very week [15,16], we will mainly focus on vibration modes observed in the low frequency region. In addition to adsorbed CO species, a number of different peaks have been reported. Kerangueven et al. found the bands at 1459, 1250 and 1165 cm−1 [11]. Shao et al. observed a pair of band around 1450 and 1285 cm−1 in DMEsaturated solution [12]. Recently, Liu et al. reported the bands at near 1450, 1217 and 1170 cm−1 on a sputtered Pt electrode surface [13]. All groups reported the existence of the band around 1450 cm−1 and assigned to OCH3 . Based on these works carried out in gas phase and electrolyte solution, we attribute the two bands at 1480 and 1445 cm−1 in Figs. 8 and 10 to ␦as (CH3 ) and ␦s (CH3 ) mode in OCH3 group, respectively. Spectral features of other bands, however, are found to be very different in each group. Shao et al. assigned the band at 1285 cm−1 to ␯(CO) of adsorbed formaldehyde as a reaction intermediate during DME electro-oxidation [12]. In fact, such band has not been observed by our previous in situ IR study for formaldehyde electro-oxidation on Pt electrode [14]. Moreover, Shao et al. showed strong (CO)ad band with a bipolar shape in the low potential region [12]. Such high CO coverage at low potential region in DME solution has never been observed in the present work. Some experimental conditions in their experiment may be not well controlled. Kerangueven et al. [11] and Liu et al. [13] argue some different species but are still not reliable enough from the experimental results. The intensity for the band observed at 1322 cm−1 in the present work is not very weak. As clearly shown in Figs. 8–11, the band intensity at 1322 cm−1 changes together with bands at 1450 and 1480 cm−1 , implying that these modes may come from the same adsorbed species. In fact, the pair of peaks at 1450 and 1295 cm−1 in 0.1 M DME by Shao et al. also showed some similar harmonic dependence with potential in the low potential region [12]. Here, we tentatively assign the peak at 1322 cm−1 to the CH2 wagging mode (␥CH2 ) of the OCH2 group if one assumes that C H cleavage takes place on one OCH3 group interacting with Pt electrode surface [24]. It is known that CH2 wagging bands are intensified when binding with electron affinity group [24]. For example, ␥CH2 mode for the ClCH2 OCH3 appears at 1320 cm−1 while ␦as CH3 and ␦s CH3 are observed at 1464 and 1430 cm−1 , respectively [25]. In the lower potential region, DME is expected to adsorb on Pt electrode to form a surface intermediate of (CH3 OCH2 –)ad . Due to electronic interaction between the electrified Pt surface, OCH2 – side in (CH3 OCH2 –)ad is expected to directly adsorb on the Pt surface, giving a relatively intense band at 1322 cm−1 with higher dependence of electrode potential and coverage. Two possible initial steps are considered for electro-oxidation of DME on electrified Pt surface: cleavage of C H and C O bonds [19]. The (CH3 OCH2 –)ad can be regarded as a product for

Y. Zhang et al. / Electrochimica Acta 53 (2008) 6093–6103

C H cleavage process. This reaction intermediate of C H cleavage process has been proposed previously for the first oxidation step of DME on electrode surface by several groups, however, no reliable evidence has been given to support this assumption [1,11,12]. In the present experiment, potential dependence of the novel IR bands and anodic current observed in the first positive-going sweep definitely show the correlation between the initial electrochemical reaction and formation of the surface species in the low potential region. Furthermore, comparison of electrochemical behaviors between DME and CH3 OH in the low potential region also supports such assumption. CH3 OH is selected here since it can be regarded as DME where CH3 group is replaced by H. Many studies have been carried out to propose that CH3 OH is first dehydrogenated on the Pt electrode surface through C H cleavage in the low potential region to yield CO [26,27]. However, no spectroscopic evidence has been obtained for this conclusion since –CH2 OH species is hard to be detected by in situ IR measurement. In the present study, after careful comparison of CVs for DME and CH3 OH in the low potential region (0.05–0.5 V) under the same condition, it is found that they are fairly analogous to each other from first potential sweep below 0.4 V [19], implying that the initial decomposition of DME on Pt electrode surface in the low potential region is same with CH3 OH, i.e., dehydrogenation of DME on Pt electrode surface firstly takes place through C H cleavage which can form the first reaction intermediate of (CH3 OCH2 –)ad . Similar behaviors have also been observed on Pt(hkl) surfaces [19]. We believe that this can also be regarded as the first direct evidence for the dehydrogenation process of small organic molecules on Pt electrode surface. In order to quantitatively explore the kinetics behavior of the reaction intermediates during DME decomposition, the timeresolved IR measurement has been carried out to monitor the formation process of (CH3 OCH2 –)ad and CO in the low potential region. The in situ IR spectra are recorded at various target potentials (0.05–0.6 V) as a function of time after a potential sequence of 1.5 V (2 s)/0.05 V (2 s)/target potential. The IR spectra are recorded every one second and part of them is given in the following figures. Fig. 12 shows two typical timeresolved IR spectra at (a) 0.2 V and (b) 0.5 V, respectively, for a period of 300 s. It is interesting to note that peaks related to

6101

(CH3 OCH2 –)ad appear earlier than COL and finally disappear while COL grows monotonously in the beginning and become constant after some time. Fig. 13 summarizes the IR band intensities for the (CH3 OCH2 –)ad (1322 and 1480 cm−1 ) and COL (2050 cm−1 ) at (a) 0.1 V, (b) 0.2 V, (c) 0.4 V and (d) 0.5 V, respectively, as a function of time. At 0.1 V, bands related to (CH3 OCH2 –)ad are still weak and COL grows slowly with time but no saturation coverage is reached. When potential is higher than 0.2 V, formation rate of (CH3 OCH2 –)ad and COL increases quickly. (CH3 OCH2 –)ad reaches maximum intensity soon and then decreases to zero while COL increases monotonously in the beginning and subsequently gets to saturation. For example, as shown Fig. 13c, (CH3 OCH2 –)ad reaches the maximum intensity at 0.4 V for ca. 5 s and disappears after ca. 80 s while COL increases quickly and gets to saturation after 150 s. As mentioned above, formation of COL always has a time delay to that of (CH3 OCH2 –)ad . In fact, from time profile of these IR bands, it is unlikely that (CH3 OCH2 –)ad and COL are formed simultaneously in a random competition process. These two intermediates are expected to be produced in sequence, i.e., (CH3 OCH2 –)ad is firstly formed and then further decomposed to COL . Since stability of COL is much higher than (CH3 OCH2 –)ad , once the surface is covered by COL , further formation of (CH3 OCH2 –)ad on the same site becomes impossible. Although we are not sure whether all of adsorbed CO species are generated from (CH3 OCH2 –)ad path in the first positive-going sweep, only (CH3 OCH2 –)ad has been detected as the non-CO intermediates in the present work. We have also found that the present kinetic profile for DME decomposition process can be modeled mathematically with a consecutive two-step reaction scheme as discussed above [28]. Meanwhile, as described above, it is interesting to demonstrate that peak position for the band at 1322 cm−1 is dependent on both electrode potential and its band intensity, i.e., surface coverage of the intermediates. Here, a typical example for the peak position profile after the potential step to 0.2 V is shown in Fig. 13b. With the increase of band intensity, peak position for the band increases from 1308 to 1320 cm−1 and decreases again with coverage decrease. This result supports that the non-CO reaction intermediate (CH3 OCH2 –)ad strongly interacts with Pt electrode surface as well as intermediates themselves.

Fig. 12. IR spectra recorded during adsorption/oxidation of DME on a Pt thin film electrode combined with the potential step. The employed potential program: (a) 0.2 V; (b) 0.5 V.

6102

Y. Zhang et al. / Electrochimica Acta 53 (2008) 6093–6103

Fig. 13. The integrated band intensities taken from a series of in situ IR spectra simultaneously acquired in the potential step from 0.05 V to (a) 0.1 V, (b) 0.2 V, (c) 0.4 V and (d) 0.5 V; the band center of 1322 cm−1 at 0.2 V (hollow). The intensities of bands at 1322 and 1480 cm−1 are multiplied by 5 times.

In addition, we also carried out the in situ IR measurement of the DME oxidation in 0.5 M H2 SO4 solution. The CO bands and the bands at 1480, 1445, 1322 cm−1 as well as their potentialdependant features are similar to that in 0.1 M HClO4 solution, except for lower intensity of these bands. At the same time, adsorption of bisulfate ion on Pt surface can be clearly observed by the in situ IR measurements. The present results also support that the dissociation of DME on Pt electrode in 0.5 M H2 SO4 solution followed similar process as in 0.1 M HClO4 solution, where the (CH3 OCH2 –)ad acts as first dissociation product. In the present work, reliable spectroscopic evidences have been obtained to support the C H cleavage of DME on Pt electrode surface in relative low potential region. Two reaction intermediate species, (CH3 OCH2 –)ad and COad are generated stepwise on Pt electrode surface during DME decomposition and electro-oxidation process. On the contrary, the C O cleavage step is another possible way for direct oxidation of DME on Pt electrode especially at high potentials (ca. 0.6–1.0 V). In fact, the C O cleavage for DME to yield methoxy has been reported on many aluminasupported heterogeneous noble metal catalyst surfaces and the methoxy is further oxidized to CO2 [23,29]. As reported recently by our group for DME electro-oxidation on Pt(h k l) surface[19], this C O cleavage step of DME is anticipated to be facilitated by the existence of the (1 0 0) terrace and, therefore, contributes to the high electro-oxidation activity of DME on (1 0 0) terraces. Detailed measurements are still in progress.

4. Conclusion Electro-oxidation of DME on Pt polycrystalline electrode in acidic solution has been investigated by electrochemical and in situ IR measurement using Kretschmann ATR configuration. The reaction intermediates, (CH3 OCH2 –)ad and CO, have been identified during the decomposition and oxidation process of DME. Time-resolved in situ IR spectroscopy is employed to characterize the prominent transient process from (CH3 OCH2 –)ad to CO at various potentials. Acknowledgments This research is supported by a Grant-in-Aid for Scientific Research (B) 19350099 from MEXT and PRESTO, Japan Science and Technology Agency (JST). LL acknowledges a fellowship from the Chinese government. References [1] J.T. M¨uller, P.M. Urban, W.F. Holderich, K.M. Colbow, J. Zhang, D.P. Wilkinson, J. Electrochem. Soc. 147 (2000) 4058. [2] M.M. Mench, H.M. Chance, C.Y. Wang, J. Electrochem. Soc. 151 (2004) A144. [3] R.H. Yu, H.G. Choi, S.M. Cho, Electrochem. Commun. 7 (2005) 1385. [4] G. Kerangueven, C. Coutanceau, E. Sibert, J.M. Leger, C. Lamy, J. Power Sources 157 (2006) 318. [5] Y. Liu, S. Mitsushima, K. Ota, N. Kamiya, Electrochim. Acta 51 (2006) 6503.

Y. Zhang et al. / Electrochimica Acta 53 (2008) 6093–6103 [6] I. Mizutani, Y. Liu, S. Mitsushima, K.I. Ota, N. Kamiya, J. Power Sources 156 (2006) 183. [7] T.A. Semelsberger, R.L. Borup, H.L. Greene, J. Power Sources 156 (2006) 497. [8] S. Ueda, M. Eguchi, K. Uno, Y. Tsutsumi, N. Ogawa, Solid State Ionics 177 (2006) 2175. [9] J.H. Yoo, H.G. Choi, C.H. Chung, S.M. Cho, J. Power Sources 163 (2006) 103. [10] Y. Tsutsumi, Y. Nakano, S. Kajitani, S. Yamasita, Electrochemistry 70 (2002) 984. [11] G. Kerangueven, C. Coutanceau, E. Sibert, F. Hahn, J.M. Leger, C. Lamy, J. Appl. Electrochem. 36 (2006) 441. [12] M.H. Shao, J. Warren, N.S. Marinkovic, P.W. Faguy, R.R. Adzic, Electrochem. Commun. 7 (2005) 459. [13] Y. Liu, M. Muraoka, S. Mitsushima, K.-I. Ota, N. Kamiya, Electrochim. Acta 52 (2007) 5781. [14] A. Miki, S. Ye, T. Senzaki, M. Osawa, J. Electroanal. Chem. 563 (2004) 23. [15] Y.X. Chen, A. Miki, S. Ye, H. Sakai, M. Osawa, J. Am. Chem. Soc. 125 (2003) 3680. [16] A. Miki, S. Ye, M. Osawa, Chem. Commun. (2002) 1500.

6103

[17] Y.X. Chen, M. Heinen, Z. Jusys, R.J. Behm, Angew. Chem. Int. Ed. 45 (2006) 981. [18] N. Wakabayashi, K. Takeuchi, H. Uchida, M. Watanabe, J. Electrochem. Soc. 125 (2004) A1636. [19] Y. Tong, L. Lu, Y. Zhang, Y. Gao, G. Yin, M. Osawa, S. Ye, J. Phys. Chem. C 111 (2007) 18836. [20] E. Herrero, K. Franaszczuk, A. Wieckowski, J. Phys. Chem. 98 (1994) 5074. [21] S.L. Chen, M. Schell, J. Electroanal. Chem. 504 (2001) 78. [22] J.G. Chen, P. Basu, T.H. Ballinger, J.T. Yates, Langmuir 5 (1989) 352. [23] F. Solymosi, J. Cserenyi, L. Ovari, Catal. Lett. 44 (1997) 89. [24] N.B. Colthup, L.H. Daly, S.E. Wiberley, Introduction to Infrared and Raman Spectroscopy, 3rd ed., Academic Press, Inc., San Diego, CA, 1990. [25] Spectral Database for Organic Compounds SDBS, National Institute of Advanced Industrial Science and Technology (AIST), Japan. [26] T.H.M. Housmans, A.H. Wonders, M.T.M. Koper, J. Phys. Chem. B 110 (2006) 10021. [27] D. Cao, G.Q. Lu, A. Wieckowski, S.A. Wasileski, M. Neurock, J. Phys. Chem. B 109 (2005) 11622. [28] Y. Zhang, Y. Tong, L. Lu, M. Osawa, S. Ye, unpublished. [29] F. Solymosi, J. Cserenyi, L. Ovari, J. Catal. 171 (1997) 476.