Electrocatalysis of methanol oxidation

Transcription

1 Electrochimica Acta 47 (2002) 3663/ Electrocatalysis of methanol oxidation T. Iwasita Instituto de Química de São Carlos, Universidade de São Paulo, Cx.P. 780, São Carlos, SP, Brazil Received 10 January 2002; received in revised form 10 February 2002 Abstract Most relevant aspects of methanol electrooxidation on Pt electrodes, including voltametric as well as spectroscopic data on the system are presented and discussed. Parallel reaction pathways have been demonstrated, producing CO 2 and HCOOH or HCHO as soluble products, to an extent that depends on several parameters of the systems like methanol concentration, electrode roughness, and time of electrolysis. Several issues like the intermediates of the reaction and the yields of oxidation products remain yet unclear. A status of-the-art on these and other aspects of methanol oxidation is presented, which may be useful for future investigations. This is the aim of the present contribution and referenced papers were accordingly selected. # 2002 Published by Elsevier Science Ltd. Keywords: Methanol; Fuel cell; Electrocatalysis; Electrooxidation; FTIR 1. The methanol system From the different small organic molecules (HCOOH, H 2 CO, CH 3 OH), methanol is the one being most intensively investigated at present. Methanol oxidation reaction has been the subject of a large number of studies in the past [1,2]. Early works [3 /6] already showed up a complex reaction mechanism, the electrocatalysis of methanol oxidation being the most difficult task in the realization of a direct acid methanol fuel cell (DMFC). Since methanol oxidation reaction has been reviewed several times, e.g. [7 /10], in the present paper we intend just to present the most relevant aspects of the electroacatalysis of methanol oxidation, thus giving a status of-the-art on the system. The thermodynamic potential for methanol oxidation to CO 2, lies very close to the equilibrium potential of hydrogen: CH 3 OHH 2 O 0 CO 2 6 H 6e E 0:02 V However, compared with hydrogen oxidation, this reaction is by several orders of magnitude slower. As early suggested by Breiter [5], the total oxidation process consists of a pattern of parallel reactions which can, in principle, be formulated as follows: Both of these pathways require a catalyst, which should be able to (a) dissociate the C/H bond and (b) facilitate the reaction of the resulting residue with some O-containing species to form CO 2 (or HCOOH). On a pure Pt electrode, which is known to be the best catalyst for breaking the C /H bond, complete oxidation takes place via two processes occurring in separate potential regions: / The first process, involving adsorption of methanol molecules, requires several neighboring places at the surface. Since methanol is not able to displace adsorbed H atoms, adsorption can only begin at potentials where enough Pt sites become free from H, i.e. near 0.2 V versus RHE for a polycrystalline Pt electrode. / The second process requires dissociation of water, which is the oxygen donor of the reaction. On pure Pt electrode, a strong interaction of water with the catalyst surface is only possible at potentials above 0.4/0.45 V versus RHE. Thus, on a pure Pt catalyst methanol oxidation to CO 2 cannot begin below, say 0.45 V. However, the adsorbate layer does not exhibit a good reactivity below /02/$ - see front matter # 2002 Published by Elsevier Science Ltd. PII: S ( 0 2 )

2 3664 T. Iwasita / Electrochimica Acta 47 (2002) 3663/3674 approximately 0.7 V, i.e. a high rate of oxidation at pure Pt occurs at potentials without technological interest. 2. Methanol adsorption It was suggested that methanol adsorption takes place in several steps, forming different species due to dissociation of the molecule: CH 3 OH 0 C H 2 OHHe 0 C HOHHe x xx 0 C xxx OHHe 0 C x OHe where x stands for a Pt site [6]. It was suggested that formaldehyde and formic acid could be formed from the intermediates CH 2 OH and CHOH, respectively [6]. If a cyclic voltammogram is started after contacting a polycrystalline Pt electrode with a methanol containing solution at a potential of 0.05 V or less, methanol adsorption can be observed as soon as hydrogen coverage decreases to a certain extent. The dissociation process gives rise to a current peak in the H-region (Fig. 1), which can be observed only during the first potential scan, i.e. when the surface is free from organic residues. The experiment in this figure was performed using the DEMS technique [11]. Briefly, in this technique the electrode is a porous Pt layer on a PTFE membrane, lying on a porous plate at the entrance of a mass spectrometer. This setup allows the entrance of any volatile product in the MS in fractions of a second after being produced. During the experiment in Fig. 1, the recording of mass signals did not show any volatile product from methanol oxidation. As an example, the Fig. 1. First potential scan for a porous polycrystalline Pt electrode in 0.1 M CH 3 OH/0.05 M H 2 SO 4 solution (upper part) and simultaneouly recording of mass intensity for CO 2 production (lower part); 10 mv s 1. Dashed lines: current and MS signals in supporting electrolyte. recording of mass (m/e/44) corresponding to CO 2 is given. Thus the current peak can only be due to Faradaic processes occurring during methanol adsorption. 3. The nature of the adsorbed methanol species Establishing the nature of the adsorbed species formed during the adsorption of small organic molecules is a difficult task. The issue was approached in a number of studies over many years [10]. Except from a study using gas chromatography [5] earlier papers were mainly based in data of charge measurements during adsorption of methanol and oxidation of the adsorbed residue [2,4,12]. The use of analytical methods for the in situ, ex situ and on line analysis of the electrode surface began in the decade of 1980 [7]. Different adsorbed species were suggested on the basis of data from infrared spectroscopy [13,14], thermal-desorption MS [15] and DEMS [7]. Infrared spectra obtained during methanol adsorption at 0.35 V on polycrystalline Pt show well characterized bands for linearly adsorbed CO at approximately 2040 cm 1 together with other bands in the 1200/1300 cm 1 region, which have been interpreted in terms of the C/ OH stretching of an hydrogenated species (as, e.g. COH [14] or HCOH). These bands were also observed in spectra obtained on single crystal Pt(100) and Pt(111) (see below). Another approach to establish the nature of the adsorbate was made by thermal desorption mass spectrometry, performed on electrodes transferred into UHV [15]. These results also confirmed the presence of hydrogenated species after adsorption of methanol. Moreover, these data have shown that the ratio between the amount of CO and (hydrogenated) species depends on methanol concentration [15]. It should be noticed that large discrepancies of results among different groups may lie, at least in part, in the experimental approaches. The necessity of thoroughly eliminating traces of oxygen from the solution and from the gas atmosphere above the solution must be emphasized, since adsorbed oxygen can interact with organic residues. Recall that O 2 reduction on a Pt surface produces adsorbed peroxide species, which can act as an oxidizing agent for species present at the surface. Other sources of discrepancies may originate in the fact that the reaction is surface sensitive and, consequently, results for polycrystalline Pt may depend on the method of surface pretreatment, electrode roughness and the time scale of the experiments, since adsorbed species can undergo slow transformations.

3 T. Iwasita / Electrochimica Acta 47 (2002) 3663/ Methanol oxidation products The oxidation products of CH 3 OH are well known since early works of Pavela [16] and Schlatter [17]. These authors used long-term electrolysis at potentials between 0.5 and 0.6 V versus RHE and found CO 2, HCHO, HCOOH and HCOOCH 3. The latter product, methyl formate, originates in a reaction: HCOOH CH 3 OHHCOOCH 3 H 2 O The yields of oxidation products depend on methanol concentration, temperature, electrode roughness and time of electrolysis [18,19]. The study of the products of methanol oxidation during a potential scan was the first goal of on-line mass spectrometry, DEMS [20]. In Fig. 2, the potentiodynamic formation of CO 2 and methyl formate on a Pt electrode was followed during the potential scan by recording the corresponding ion currents in the MS: (m/ e/44) and (m/e/60), respectively. No mass signals for HCHO were observed, but a weak ion current for methylal (CH 2 (OCH 3 ) 2 ), indicated its formation via reaction of HCHO with CH 3 OH [7]. However, there must be some problem with the volatility of formaldehyde or its hydration product in aqueous solution (gem diol, CH 2 (OH) 2 ), which makes somehow difficult its direct detection using the DEMS technique [21]. These difficulties extend to other modern analytical methods like in situ FTIR as pointed out by Korzeniewski and Childers [22]. This could be the reason why formaldehyde remained almost disregarded in the methanol fuel cell literature. Korzeniewski and Childers determined formaldehyde yields fluorometrically after applying different constant potentials on a smooth polycrystalline Pt electrode, during 5 min in a micro cell. They report for formaldehyde a yield of 38% under following conditions: 0.25 V versus Ag/AgCl (ca V vs. RHE), 15 mm CH 3 OH/0.1 M HClO 4. The yield decays at higher potentials [22]. On porous Pt electrodes, Wang et al. found at 0.65 V versus RHE 50% of HCHO, 34% of HCOOH and only 16% of CO 2. It is worth noting that Ota et al. also found relatively high yields of HCHO on platinized Pt electrodes at 0.6 V versus RHE [19]. 5. Structural dependence of methanol oxidation 5.1. Results of cyclic voltammetry As shown by the cyclic voltammograms in Fig. 3 for the three low index surfaces of platinum, adsorption and oxidation of methanol present a strong sensitivity to the surface structure [23]. The Pt surfaces in these experiments were contacted with methanol at 0.05 V (a potential where methanol adsorption is negligible) and then the first potential sweep was recorded. From the three surfaces, Pt(100) is the only one presenting a well defined current peak at 0.35 V for methanol dissociative adsorption. This process is superimposed on the current for H-desorption (compare with Fig. 1). Almost no activity towards methanol oxidation is observed until the potential reaches approximately 0.72 V. Contrasting with this result, no indication of dissociative adsorption of methanol parallel to hydrogen desorption is observed at Pt(111). On the other hand, at Pt(110) the lowering of current in the H region may indicate that the methanol adsorption begins already at the initial potential of the voltammogram (0.05 V). This behavior is expected from the low potential for H desorption in this surface. For a comparison, the first part of the potential scan is magnified in Fig. 4. Here, one sees that a CV for Pt(111) shows up the highest oxidation current in the potential range between 0.45 and 0.65 V, followed by Pt(110) and Pt(100). However, before establishing a ranking for the activity of the three surfaces, infrared results should be taken into consideration, as we shall do in the coming section Results of infrared spectroscopy Fig. 2. DEMS experiment: current and mass signals (ion current) for volatile products during methanol oxidation at porous polycrystalline Pt, surface roughness: ca. 50; 0.1 M CH 3 OH/1 M HClO 4 ;20mVs 1. [7]. Infrared spectra during methanol adsorption and oxidation are shown in Fig. 5, for Pt(111), Pt(100) and Pt(110) [23]. Bands at approximately 2060 and 1850 cm 1 are assigned to linearly (CO L ) and bridge (CO B )

4 3666 T. Iwasita / Electrochimica Acta 47 (2002) 3663/3674 Fig. 4. Comparison of current for methanol oxidation on Pt(hkl). Conditions as in Fig. 3. Fig. 3. First potential scan for methanol oxidation on Pt(hkl); 1.0 M CH 3 OH in 1.0 M HClO 4 ; sweep rate: 50 mv s 1 [21]. bonded CO, respectively. The band at 1260 cm 1 was assigned to some H containing intermediate, possibly COH [14]. Other H-containing adsorbate may be responsible for the feature at 2950 cm 1 in the spectra for Pt(100). This band, which can be assigned to the C/ H stretching of a CH 2 group appears already at 0.2 V and becomes better defined at more positive potentials. All spectra exhibit a band at 2341 cm 1, due to the oxidation product CO 2. Although not shown here, a weak band at 1230 cm 1 has been observed on Pt(111) at potentials above 0.45 V, which was assigned to the C /O /C stretching of methyl formate [23]. Surface sensitive effects become more evident when analyzing the integrated band intensities for CO and CO 2 for all three surfaces shown in Fig. 6a and b. In comparing these results with those of the cyclic voltammograms, one has to be aware that the time scale for the spectra is much larger than for the CV. Thus, the voltammetric curves in Fig. 4 were taken at 50 mv s 1, i.e s mv 1 ). On the other hand, spectra collection requires about 70 s at each potential and considerable adsorption can take place during this time. The behavior of Pt(111) and Pt(110) highlights some interesting features of the mechanism of methanol oxidation. We let at first Pt(100) out of consideration because of the complication of presenting two forms of adsorbed CO (on top and bridge), thus making it difficult to estimate relativecoverages on the basis of band intensities [24].In the case of Pt(111) the intensity of the band for CO B is so small that we shall simply neglect it in the following discussion. Measurable amounts of CO are observed at Pt(110) already at 0.1 V. The CO signal rapidly grows with potential and reaches a maximum at around 0.3 V. At Pt(111), the initial adsorption can be extrapolated to around 0.2 V, judging from the band for linear bonded CO. This feature grows somewhat slower than that for Pt(110). Also, a weak band for bridge bonded carbon monoxide is apparent at potentials above 0.35 V. But the most important observation here, is that the dissociative adsorption of methanol at Pt(111) takes place at potentials within the H region. Therefore we can conclude that the lack of methanol adsorption in the H-region during the potential scan at 50 mv s 1 (Fig. 3) is due to a kinetic limitation of the adsorption process at Pt(111) which is not observed on the other two surfaces. In Fig. 6b is presented the integrated band intensity for the CO 2 observed in the potential region below 0.8 V. We can state that in this region CO 2 production proceeds with almost the same intensity at both Pt(111) and Pt(110) and is much lower at Pt(100). Both former surfaces have almost the same catalytic activity for CO 2 production, at least at potentials below approximately 0.7 /0.75 V. However, the relative oxidation currents

5 T. Iwasita / Electrochimica Acta 47 (2002) 3663/ Fig. 5. Infrared spectra in 1.0 M CH 3 OH in 0.1 M HClO 4 at Pt(100), Pt(111) and Pt(110). Reference spectra collected at 50 mv; sample spectra collected at the indicated potentials [21]. observed in the voltammograms of Fig. 4 for Pt(110) and Pt(111) indicate a higher activity of the latter towards methanol oxidation. We can thus conclude, that in the interval of potentials between, say, 0.4 and 0.6 V, the potenciodynamic current at Pt(111), originates to a large extent from the dissociative adsorption of methanol and/or from the parallel pathway mentioned in the Introduction, ending in other products than CO 2. Our present knowledge of the yields of HCOOH and/or HCHO at single-crystal electrodes is insufficient for establishing the extent to which these parallel pathways contribute to the current. Also Pt(100) presents evidences for high yields of other products via a parallel reaction: the pronounced increase of current at 0.7 V (Fig. 4), contrasts with the moderate variation of the CO 2 band intensity in Fig. 6b. 6. On the mechanism of methanol oxidation The mechanism of methanol oxidation is an issue which can be considered to be at its very beginning. The slow progress along many years can be easily understood in the light of the difficulties created by the existence of parallel reaction paths with yields depending on potential, time, surface structure, etc. At present, only two global processes are distinguishable as already said in the Introduction namely, adsorption of methanol molecules and oxidation of adsorbed residues. Analizing the IR results presented above it is possible to follow the pathway leading to CO 2 via formation of adsorbed CO and to identify for this pathway, whether adsorption of methanol or removal of CO is the rate determining process. Due to the fact that the interface components of the respective Pt surfaces depend on potential and admitting the necessity of oxygen donor species at the surface, for oxidation of the adsorbed residues, discussing the total oxidation is only meaningful at potentials above 0.4 V or higher. However, what the adsorption process concerns, we should start the discussion by considering the methanol adsorption at potentials below 0.4 V. It is well known that methanol cannot displace adsorbed hydrogen from the Pt surface. This is a well known phenom-

6 3668 T. Iwasita / Electrochimica Acta 47 (2002) 3663/3674 Fig. 6. Integrated band intensities from IR spectra at Pt(111) and Pt(100) and Pt(110) in 1.0 M CH 3 OH in 0.1 M HClO 4. (a) Linear (CO L ) and bridge (CO B ) bonded carbon monoxide; (b) carbon dioxide. enon and is now documented by the data in Fig. 6a, where the structure and potential dependence of methanol adsorption can be analyzed from the band intensity for adsorbed CO. The above data were obtained by applying potential steps of approximately 70 s of duration, i.e. they are neither stationary nor dynamic, but since the procedure for collecting spectra was the same for all surfaces, it should be valid to compare with the band intensities in order to have an idea of the rate of methanol adsorption in the low potential region. Thus, it can be stated that adsorption of methanol occurs at a higher rate at Pt(110) than at Pt(111) (Fig. 6a). While CO L coverage at Pt(110) rapidly increases reaching a saturation value between 0.3 and 0.4 V, the band for CO L at Pt(111) grows more slowly and does not present a real maximum but a sudden falling of intensity at a potential of 0.6 V. Additionally, in the whole range of potentials the CO band intensity is substantially higher on Pt(110) than on Pt(111). More about the methanol adsorption process can be learnt by analyzing in parallel the band intensity for CO in Fig. 6a and the hydrogen coverage according to the voltammograms of these surfaces measured in HClO 4 (Fig. 7). Thus, e.g. for Pt(110) at 0.2 V, u H /0.2 and the band intensity for CO is large (about 80% of its maximum value). On the other hand, for the same u H at Pt(111), (E/0.29 V), the band intensity is much lower. In fact, it grows slowly even beyond 0.4 V, where the surface is totally free from adsorbed hydrogen. It can thus be stated that the differences in methanol adsorption rate for these two surfaces originates in some property inherent to the surface itself and not in the availability of H-free sites: adsorption of methanol is intrinsically a slow process at Pt(111). We proceed now to discuss the potential region above 0.4 V, where CO 2 is produced. Considering that independent of the respectivecocoverage, the rate of CO 2 formation is the same for both surfaces, we can state that the reaction rate for CO 2 formation is the same for both surfaces Pt(111) and Pt(110). The most simple interpretation of this result is that the oxidation of the adsorbed residue is the rate determining process. Or, with other words, at all potentials above 0.4 V, there is enough adsorbed methanol on both surfaces as to reach the maximum rate of oxidation at the respective potential. Also Gasteiger et al. [25] suggested that oxidative removal of CO is the rds in this potential region. Following a somewhat different approach, Christensen et al. [26] arrived to the same conclusion for polycrystalline Pt. Summarizing, oxidation of adsorbed CO and not adsorption of methanol is likely to be the rate determining process in the potential region between, say 0.5 and 0.70 V. It has been shown that interaction of water with the Pt surface increases as the potential is made more positive [27] and competition of methanol with water for adsorption sites becomes important. Therefore, at high potentials (above, say, 0.7 V) methanol adsorption becomes again rds [28] and for this reason the reaction rate passes through a maximum and then decays. For the pathway analyzed here, leading to CO 2 formation via adsorbed CO, we can state that CO is Fig. 7. Potentiodynamic current/potential response for between 0.05 and 0.5 V vs. RHE for Pt(111) and Pt(100) in 0.1 M HClO 4 ; sweep rate 50 mv s 1.

7 T. Iwasita / Electrochimica Acta 47 (2002) 3663/ an intermediate of the reaction and the role of surface poison usually ascribed to CO ad should be revised. Inspection of Fig. 6 indicates that CO indeed accumulates on the surface at low potentials (see the result for Pt(110)), however, the reason for the lack of CO 2 formation lies in the inability of Pt to dissociate water and not in the degree of surface blocking by CO. Otherwise it would be difficult to understand that two surfaces covered with CO to different extent produce CO 2 at identical rates. Concerning the nature of the oxygen donor, it was originally suggested by Gilman that it is an adsorbed OH species coming from water dissociation [29]. But, according to Wieckowski et al. [30] the oxygen donor is simply some activated water molecule adsorbed on the Pt surface. For oxidative stripping of CO adlayers on platinum, Koper et al. [31] have shown that the dissociation of water is a necessary step in order to harmonize experimental data with results of Monte Carlo simulations. Although the real nature of the oxygen donor for the oxidation of CO ad was not demonstrated via spectroscopic methods, we are inclined to accept the more or less general consensus that it is OH ad formed through the dissociation of water Supporting electrolyte effects on the rate of methanol oxidation Perchloric and sulfuric acids are the commonly used supporting electrolytes for studies of methanol electrooxidation. It is, however noteworthy, that methanol oxidation can be affected by the anion of the supporting electrolyte used. This is clearly the case for Pt(111) and Pt(100) electrodes, where cyclic voltammograms of methanol exhibits much larger currents in HClO 4 than in H 2 SO 4 [32,33]. The difference was justified by the specific adsorption of sulfate species at Pt electrodes, which partially hinder methanol oxidation. Kita et al. [32] observed for Pt(111) a factor of ten higher current in HClO 4 than in H 2 SO 4, while for Pt(100) the enhancement factor was about two. The different behavior can be rationalized in terms of a stronger specific anion adsorption at Pt(111) than at Pt(100) [34]. No difference between HClO 4 and H 2 SO 4, was observed at Pt(110) [32]. 7. Catalyst promoters for methanol oxidation Several binary and ternary catalysts were proposed for methanol oxidation, most of them based in modifications of Pt with some other metal. This metal must fulfill the requirement of forming O-containing surface species at low potentials. Among others, Sn, Re, Ru, Ge and Mo, were suggested [7,25,35 /54]. There are, of course, several practical factors limiting the choice of the metal. Many O-adsorbing metals can produce negative effects, e.g. inhibit methanol adsorption or may be not sufficiently stable for long-term use, as required for a fuel cell. At present, there is a general consensus that PtRu offers the most promising results. Methanol oxidation on PtRu binary catalysts has been the matter of a number of papers [25,35/54]. The catalytic effect has been observed using different kinds of PtRu materials, such as PtRu alloys, PtRu electrodeposits, Ru evaporated on Pt, Ru adsorbed on single-crystal Pt(hkl), etc. and on technical (carbon supported electrodes) as well [54,55]. When discussing the reason for the catalytic effect of PtRu, the bi-functional mechanism is often invoked [41]. The term was suggested to give emphasis on the joint activities of both metals, Pt being the one adsorbing and dissociating methanol and Ru, the one oxidizing the adsorbed residues at low potentials. This description of the mechanism is based in the observation that at potentials below 0.4 V, Pt is a good catalyst for methanol adsorption, but not for water dissociation while Ru is able to dissociate water but it cannot adsorb methanol. However, establishing a role for each metal as in the bi-functional mechanism is of limited use, since it is well known that at high potentials Pt dissociates water and, as shown in refs. [43,53] at high temperatures (60, 80 8C) Ru adsorbs methanol. Moreover, even for conditions where methanol adsorption occurs only on Pt, CO can move on the surface and occupy sites on Ru atoms. Altogether, several adsorbed species could be involved in the oxidation process at the PtRu catalyst, namely, Pt(CO) ad, Ru(CO) ad, Ru(OH) ad and Pt(OH) ad. In a simplified manner, we shall describe the bifunctional mechanism as follows [25]. The first step of the reaction is adsorption of methanol: CH 3 OH (sol) 0 (CO) ad 4 H 4 e (1) (CO) ad represents an adsorbed CO species either on Pt or on Ru. Then, both Pt and Ru dissociate water to form adsorbed OH: RuH 2 O 0 (OH) ad H e (2) PtH 2 O 0 (OH) ad H e (3) Finally, following a Langmuir /Hinschelwood mechanism adsorbed CO reacts with adsorbed OH to give CO 2 : (CO) ad (OH) ad 0 CO 2 H e (4) For CO adlayers obtained via adsorption of dissolved CO on PtRu, Koper et al. [56] analyzed reaction (4) for all possible species mentioned above and found that an enhanced effect is only possible if the final oxidation step occurs between CO adsorbed at Pt and OH adsorbed at Ru. Therefore, reaction (4) can be specifically written as

8 3670 T. Iwasita / Electrochimica Acta 47 (2002) 3663/3674 Pt(CO) ad Ru(OH) ad 0 CO 2 H e (5) Moreover, as a necessary condition for the Langmuir /Hinschelwood mechanism, CO must diffuse on the surface to the places where the OH ad partner is formed. Koper et al. [56] suggested that the relative rate of CO diffusion on the surface must be high. Besides the promoter effect of Ru through reaction (5), there are experimental evidences of additional effects of Ru on the reaction. Thus, the potentiodynamic curves (first sweep) of Fig. 8, obtained at PtRu electrodeposits in the course of DEMS experiments, show a negative shift of the dissociative adsorption of methanol depending on the Ru content of the electrodes. Further evidence of the earlier adsorption was obtained via infrared spectroscopy as shown in Section 5.1. The catalytic effect of PtRu materials is usually presented in the form of current /time plots at constant potential, as observed in Fig. 9, [46,50,51]. Similar responses have been reported by other authors under comparable conditions [43,47]. In order to measure the data represented in this plot Pt(111)/Ru electrodes were prepared by adsorption of Ru onto Pt(111) [46] and alloys were cleaned by sputtering and heating in UHV following the pre-treatment technique suggested by Gasteiger et al. [56]. This is a reliable method for producing smooth surfaces of identical composition as the bulk. The quality of having a low roughness factor can be easily checked via comparison of the current in voltammograms in base electrolyte measured with the alloy with those obtained using Pt(111)/Ru electrodes [46]. As observed in Fig. 9, when smooth PtRu materials are polarized at constant potential in methanol solutions, the current decays continuously indicating a pronounced loss in activity. The current does not reach a stationary state even after several hours. Two different origins of this effect causing electrode deactivation have been identified. One of these is reversible and seems to Fig. 9. Current/time curves for comparing the catalytic activity of Pt(111) and PtRu alloys, towards methanol oxidation in 0.5 M CH 3 OH/0.1 M HClO 4. Potential 0.5 V vs. RHE. Room temperature, electrodes surfaces cleaned in UHV. Fig. 8. First potential scan in 1.0 M CH 3 OH/0.5 M H 2 SO 4, starting at 50 mv (RHE); scan rate 1 mv s 1. Porous electrodes prepared by depositing Pt and Ru at 0.2 V vs. RHE up to a total charge of 1200 mc, from solutions containing x mm RuCl 3 and y mm H 2 Cl 6 Pt in 0.5 M H 2 SO 4. The x/y ratios are indicated in each curve. Fig. 10. Plot of the current density for methanol oxidation as function of Ru coverage, taken from current/time curves at 0.5 V as in Fig. 9. Data for UHV prepared PtRu alloys, measured after 20 min; data for Pt(111)/Ru formed by spontaneous adsorption, measured after 5 min. Solutions: 0.5 M H 2 SO 4 /0.5 M CH 3 OH (data form ref. [43]), 0.5 M CH 3 OH/0.1 M HClO 4 [46].

9 T. Iwasita / Electrochimica Acta 47 (2002) 3663/ Fig. 11. Current densities for methanol oxidation at 0.4 V in 0.5 M CH 3 OH/0.5 M H 2 SO 4 vs. Ru surface composition of sputter cleaned Pt/Ru alloys. Electrode immersion at V for 3 min prior to stepping to the indicated potential. Dashed lines are arbitrarily drawn smooth curves to connect the experimental data points. Data from ref. [43], published with permission. be caused by the oxidation of the Ru surface [50,51], forming oxides like RuO 2 and RuO 3, which are not active as oxygen donors for CO oxidation. The electrode activity is thus partially recovered by applying a potential step towards more negative values, where the ruthenium oxide species are reduced. The other factor causing the decay of current is, apparently, a blockage of the surface by some organic residue, which is slowly formed and can only be oxidized at high anodic potentials. No spectroscopic proofs are yet available on the nature of such blocking species. The results of i /t curves obtained at room temperature for the two types of materials (alloys and Pt(111)/ Ru) are collected in Fig. 10, where the current after 20 min of polarization at 0.5 V versus RHE is plotted as a function of the Ru:Pt surface composition. It is noteworthy that in spite of the fact that the data for the alloys in this figure were measured in two different laboratories [43,46], they nicely fit together. Also, methanol oxidation at PtRu alloys seems not to be the same for both supporting electrolytes H 2 SO 4 and HClO 4 Fig. 12. Comparison of in situ FTIR spectra for Pt(111), Pt(111)/Ru 39% and PtRu alloy (85:15) in 0.5 M CH 3 OH/0.1 M HClO 4. Potentials as indicated on each spectrum; reference spectrum taken at 0.05 V (from [46] with added spectra at 0.55 V).

10 3672 T. Iwasita / Electrochimica Acta 47 (2002) 3663/3674 used. Two main points can be extracted from this plot: (i) PtRu alloys are better catalysts than Pt(111)/Ru, and (ii) both materials present a wide maximum (between ca. 10 and 40% Ru for the alloys and ca. 15 and 50% for the Pt(111)/Ru electrodes) at room temperature. In terms of the bi-functional mechanism, these maxima indicate that a Ru percentage of approximately 10/ 45% on the surface is enough to provide an efficient oxidation of adsorbed methanol residues. Within this range one has a possibility of studying the influence on the reaction of other parameters (like e.g. methanol concentration) and possibly, the kinetic of the reaction. The upper limit of Ru coverage, is given by the necessity of having enough Pt sites for adsorbing and dissociating methanol. This limit is somewhat higher for adsorbed Ru than for the alloy (observe the wider maximum). This and the higher currents observed on the alloys may be related to the fact that the latter present a more homogeneous distribution of Ru atoms than the Pt(111)/Ru electrode [57]. STM data of Pt(111)/Ru show that Ru tends to form aggregates (islands) on the surface of Pt [46,58]. Thus, for the same Ru percentage, wider Ru patches can be found on the Pt(111)/Ru surface than on the alloy. It is noteworthy that the maximum of the j /u Ru plot for the alloy is shifted to higher u Ru values at higher temperatures (Fig. 11) [43]. This effect probably originates from the fact that at higher temperatures (e.g. 60 8C), Ru becomes active for adsorbing methanol and oxidizing the residue (see the value of current for pure Ru in Fig. 11). 8. Spectroscopic results of methanol oxidation on PtRu materials Infrared spectra using Pt(111), Pt(111)/Ru and a PtRu alloy [46] are presented in Fig. 12. PtRu alloy electrodes were UHV cleaned, as for the experiments of Fig. 9. The bands observed on both Pt(111)Ru and PtRu (85:15) alloy correspond to CO 2 (2341 cm 1 ) and CO L (2050 cm 1 ). At Pt(111) a band at approximately 1820 cm 1 is due to bridge adsorbed (CO B )(Fig. 13). The PtRu alloy electrode presents the highest production of CO 2 at a given potential (note the differences in scale). It is also noteworthy, that although the CO 2 band markedly grows with potential and also the highest band intensity for linear bonded CO. At PtRu the CO feature remains approximately constant in the range of potentials between 0.3 and 0.55 V, indicating that the intermediate CO reaches a stationary coverage balancing the rates of formation and oxidation. In a recent paper [53], methanol oxidation was studied on two PtRu alloys having 10 and 90% Ru, in the range of temperatures between 25 and 80 8C. For the 90% Ru alloy, this study shows a negligible activity towards methanol oxidation at 25 8C but an appreciable CO 2 Fig. 13. DEMS experiment on PtRu porous electrode in 1.0 M CH 3 OH/0.5 M H 2 SO 4 ; v: 20 mv s 1. Electrode prepared by electrodepositing Pt and Ru during 7 min at 0.05 V, on a gold substrate. Solution composition: 10 mm RuCl 3 /10 mm H 2 Cl 6 Pt in 0.5 M H 2 SO 4 [45] published with permission. production and also a high CO coverage (near saturation) at 80 8C. It is noteworthy that, although no features for other intermediates or soluble products were observed under the conditions for taking the spectra on the PtRu materials shown in Fig. [12], DEMS experiments in 1.0 MCH 3 OH using PtRu electrodes do exhibit a potential dependent mass signal for HCOOCH 3 (Fig. 12) [45]. As pointed out before, the yields of different products on a Pt electrode depend on several experimental parameters such as methanol concentration and electrode roughness [19]. This condition seems to extent to PtRu materials. So far, IR spectra on well prepared PtRu alloys in concentrated methanol solutions (above 0.5 M) have not been published and the question on the parallel paths on PtRu should be let open to further discussion. 9. Concluding remarks In the last years, considerable progress has been made in the knowledge of methanol electrooxidation reaction and on its catalysis. As for other small organic molecules, parallel pathways occur during methanol oxidation. In addition to CO 2 which is the major product, formic acid (methyl

11 T. Iwasita / Electrochimica Acta 47 (2002) 3663/ formate) and formaldehyde are formed, their yields depending on experimental conditions such as surface roughness and methanol concentration and time of electrolysis. In spite of noticeable structural effects of the Pt surface, which are reflected on the amount and rate of surface coverage by CO, comparable quasi-stationary rates of CO 2 production are observed at Pt(111) and Pt(110) in the potential region between approximately 0.4 and 0.75 V. This result indicates that the rate determining process is more likely the oxidation of the adsorbed species and not the adsorption of methanol. At room temperature PtRu alloys having a Ru content between 10 and 40% are, at present, the best catalysts for the reaction. This is probably due to a good distribution of Pt and Ru atoms as compared with other materials where the atoms tend to segregate forming patches of pure Ru or Pt, thus physically separating the bi-functional partners. Experimental data show that in addition to the bi-functional action of both metals, the promoter effect of Ru may also lie in other effects, such as lowering the potential for methanol adsorption. Some deactivation effects are observed during polarization at constant potential, the origin of these being probably, the formation of inactive ruthenium oxides of and blockage of the surface by organic residues of unknown nature. Acknowledgements Financial support from FAPESP and CNPq, Brazil and DFG, Germany, is gratefully acknowledged References [1] W. Vielstich, Fuel Cells, Wiley Interscience, Bristol, [2] M.W. Breiter, Electrochemical Processes in Fuel Cells, Springer- Verlag, Berlin, [3] O.A. Petrii, B.I. Podlovchenko, A.N. Frumkin, H. Lal, J. Electroanal. Chem. 10 (1965) 253. [4] V.S. Bagotzki, Y. Vassileiv, Electrochim. Acta 12 (1967) [5] M. Breiter, Electrochim. Acta 12 (1967) [6] V.S. Bagotzki, Y.B. Vassiliev, O.A. Kazova, J. Electroanal. Chem. 81 (1977) 229. [7] T. Iwasita, in: H. Gerischer, C. Tobias (Eds.), Advances in Electrochemical Science and Engineering, vol. 1, Verlag Chemie, 1990, p [8] R. Parsons, T. Van der Not, Electroanal. Chem. 257 (1988) 1. [9] S. Wasmus, A. Küver, J. Electroanal. Chem. 461 (1999) 14. [10] A. Hamnett, in: A. Wieckowski (Ed.), Interfacial Electrochemistry. Theory, Experimental and Applications, Marcel Dekker, New York, 1999, p [11] B. Bittins-Cattaneo, E. Cataneo, P. Königshoven, W. Vielstich, in: A.J. Bard (Ed.), Electroanalytical Chemistry a Series of Advances, Marcel Dekker, 1991, p [12] T. Biegler, J. Phys. Chem. 72 (1968) [13] B. Beden, C. Lamy, A. Bewick, K. Kunimatsu, J. Electroanal. Chem. 121 (1981) 343. [14] T. Iwasita, F.C. Nart, J. Electroanal. Chem. 317 (1991) 291. [15] S. Wilhelm, T. Iwasita, W. Vielstich, J. Electroanal. Chem. 238 (1987) 383. [16] T.O. Pavela, Ann. Acad. Sci. Fenn. Ser. K A2 (1954) 1. [17] M.J. Schlachter, in: G.Y. Yong (Ed.), Fuel Cells, Reinhold, New York, 1963, p [18] M. Breiter, Disc. Faraday Soc. 45 (1968) 79. [19] Y. Ota, M. Nakagawa, Takahashi, J. Electroanal. Chem. 179 (1984) 179. [20] (a) O. Wolter, C. Giordano, J. Heitbaum, W. Vielstich, Proceedings of Symposium on Electrocatalysis, The Electrochemical Society, Pennington, 1982, p. 235; (b) O. Wolter, J. Heitbaum, J. Electrochem. Soc. 132 (1985) [21] H. Wang, T. Löffler, H. Baltruschat, J. Appl. Electrochem. 31 (2001) 759. [22] C. Korzeniewski, C. Childers, J. Phys. Chem. B 102 (1998) 489. [23] X.H. Xia, T. Iwasita, F. Ge, W. Vielstich, Electrochim. Acta 41 (1996) 711. [24] I. Villegas, M.J. Weaver, J. Chem. Phys. 101 (1994) [25] H. Gasteiger, N. Markovic, P.N. Ross, E.J. Cairns, J. Phys. Chem. 97 (1993) [26] P.A. Christensen, A. Hamnett, G.L. Troughton, J. Eletroanal. Chem. 362 (1993) 207. [27] X.H. Xia, T. Iwasita, J. Electroanal. Chem. 411 (1996) 95. [28] T. Iwasita, X.H. Xia, H.-D. Liess, W. Vielstich, J. Phys. Chem. B 101 (1997) [29] S. Gilman, J. Phys. Chem. 68 (1964) 70. [30] H. Kim, I. Rabelo de Moraes, G. Tremiliosi-Filho, R. Haasch, A. Wieckowski, Surf. Sci. 474 (2001) L203. [31] M.T.M. Koper, A.P.J. Jansen, R.A. van Santen, J.J. Lukkiens, P.A.J. Hibers, J. Chem. Phys. 109 (1998) [32] H. Kita, Y. Gao, H. Hattori, J. Electroanal. Chem. 373 (1994) 177. [33] N. Markovic, P.N. Ross, J. Electroanal. Chem. 330 (1992) 499. [34] (a) F.C. Nart, T. Iwasita, M. Weber, Electrochim. Acta 39 (1994) 961; (b) T. Iwasita, F.C. Nart, M. Weber, Electrochim. Acta 39 (1994) [35] K.J. Cathro, J. Electrochem. Soc. 116 (1969) [36] M.M.P. Jansen, J. Moolhuysen, J. Catal. 46 (1977) 289. [37] O.A. Petrii, B.I. Podlovchenko, A.N. Frumkin, H. Lal, J. Electroanal. Chem. 10 (1965) 253. [38] H. Binder, A. Köhling, G. Sandstede, in: G. Sandstede (Ed.), From Electrocatalysis to Fuel Cells, University of Washington, Seatle, [39] M. Watanabe, S. Motoo, J. Electroanal. Chem. 69 (1976) 429. [40] T. Iwasita, F.C. Nart, W. Vielstich, Ber. Bunsenges Phys. Chem. 94 (1990) [41] M. Watanabe, S. Motoo, J. Electroanal. Chem. 60 (1975) 267. [42] N. Gasteiger, Markovic, P.N. Ross, Electrochim. Acta 41 (1996) [43] H.A. Gasteiger, N. Markovic, P.N. Ross, E.J. Cairns, J. Electrochem. Soc. 141 (1994) [44] N. Markovic, H.A. Gasteiger, P.N. Ross, L. Villegas, M.J. Weaver, Electrochim. Acta 40 (1995) 91. [45] M. Krausa, W. Vielstich, J. Electroanal. Chem. 379 (1994) 307. [46] T. Iwasita, H. Hoster, A. John-Anacker, W.F. Lin, W. Vielstich, Langmuir 16 (2000) 522. [47] W. Chrzanowski, A. Wieckowski, Langmuir 14 (1998) [48] W. Chrzanowski, W.H. Kim, A. Wieckowski, Catal. Lett. 50 (1998) 69. [49] G. Tremiliosi-Filho, H. Kim, W. Chrzanowski, A. Wieckowski, B. Grzybowska, P. Kulesza, J. Electroanal. Chem. 467 (1999) 143. [50] H. Hoster, T. Iwasita, H. Baumgärtner, W. Vielstich, Phys. Chem. Chem. Phys. 3 (2001) 337.

12 3674 T. Iwasita / Electrochimica Acta 47 (2002) 3663/3674 [51] H. Hoster, T. Iwasita, H. Baumgärtner, W. Vielstich, J. Electrochem. Soc. 148 (2001) A496. [52] A. Kabbaby, R. Faure, R. Durand, B. Beden, F. Hahn, J.-M. Leger, C. Lamy, J. Electroanal. Chem. 444 (1998) 41. [53] D. Kardash, C. Korzeniewski, N. Markovic, J. Electroanal. Chem. 500 (2001) 518. [54] W.H. Lizcano-Valbuena, E.C. Bortholin, A. Oliveira Neto, V.A. Paganin, E.R. González, 200th Meeting of the Electrochemical Society, 52nd ISE Meeting Ext. Abstract 330, San Francisco, [55] A.O. Neto, J. Peres, W.T. Naporn, E. Ticcianelli, E. González, J. Braz. Chem. Soc. 11 (2000) 39. [56] M.T.M. Koper, J.J. Lukkien, A.P.J. Jansen, R.A. van Santen, J. Phys. Chem. B 105 (1999) [57] H. Gasteiger, P.N. Ross, E.J. Cairns, Surf. Sci. 293 (1993) 67. [58] E. Herrero, J. Feliu, A. Wieckowski, Langmuir 15 (1999) 493.