15074
J. Phys. Chem. C 2007, 111, 15074-15083
Electrocatalytic Oxidation of Formaldehyde on Platinum under Galvanostatic and Potential Sweep Conditions Studied by Time-Resolved Surface-Enhanced Infrared Spectroscopy Gabor Samjeske´ , Atsushi Miki, and Masatoshi Osawa* Catalysis Research Center, Hokkaido UniVersity, Sapporo 001-0021, Japan ReceiVed: June 4, 2007; In Final Form: July 21, 2007
When formaldehyde is oxidized on a Pt electrode under galvanostatic (constant current) conditions, the electrode potential oscillates spontaneously. The oscillatory behavior has been examined by time-resolved surfaceenhanced infrared absorption spectroscopy (SEIRAS) for understanding the mechanism and kinetics of the reaction at the molecular scale. SEIRAS reveals that CO and formate are adsorbed on the electrode surface and that their band intensities (coverages) oscillate synchronously with the oscillation of potential. SEIRAS coupled to cyclic voltammetry or linear sweep voltammetry suggests that formaldehyde is oxidized to CO2 via two parallel processes: the direct path via adsorbed formate and the indirect path via adsorbed CO. The two processes are kinetically coupled and autocatalytically activate and deactivate formaldehyde oxidation to yield the potential oscillations. The oxidation of the electrode surface also contributes to the oscillations for large applied currents.
COads + OHads f CO2 + H+ + e-
1. Introduction Electrochemical oxidation of small organic molecules, including methanol, formaldehyde, and formic acid, on Pt and Pt alloys has received great interest due to their possibilities as fuels for low-temperature fuel cells.1 Although formaldehyde is toxic and not very suitable for fuel cells, the study of its oxidation is important for the full understanding of methanol oxidation because formaldehyde is produced by partial oxidation of methanol.2 In an acidic electrolyte, most of the formaldehyde is hydrated to methylene glycol (Keq ) 2280).3
H2CO + H2O f CH2(OH)2
(1)
The overall electrochemical oxidation of methylene glycol is represented by the following chemical equation:
CH2(OH)2 f CO2 + 4H+ + 4e-
(2)
The oxidation process is believed to consist of at least two paths: a direct path and an indirect path.1 In the direct path, methylene glycol is directly oxidized to CO2 via short-lived intermediate(s) adsorbed on the electrode surface. In the indirect path, CO is formed first
CH2(OH)2 f COads + H2O + 2H+ + 2e-
(3)
and then oxidized to CO2 at high potentials by a reaction with adsorbed water
COads + H2Oads f CO2 + 2H+ + 2e-
(4)
or with adsorbed OH
H2O f OHads + H+ + e* Corresponding author. E-mail:
[email protected].
(5a)
(5b)
The adsorption of CO on the electrode, eq 3, has been wellestablished by infrared reflection absorption spectroscopy (IRAS),4-8 while the reactive intermediate in the direct path has not been identified definitely yet. Several reactive intermediates, including HC(OH)2,9 COOH,9-11 formic acid (HCOOH),9,12 formate (HCOO),13 and methylene glycolate (H2COO),14 have been proposed (or speculated) so far, where the underlined portions show atoms involved in bonding with the electrode surface. Among the possible reactive intermediates, formate and methylene glycolate were proposed from in situ infrared studies. By using surface-enhanced infrared absorption spectroscopy in the attenuated total reflection mode (ATR-SEIRAS),15 Miki et al.13 found for the first time that formate is adsorbed on a polycrystalline Pt surface via two oxygen atoms in a bridging conformation and that the oxidation current increases as the band intensity of formate increases. From these results, formate was proposed to be the reaction intermediate in the direct path of formaldehyde oxidation. Recently, Batista and Iwasita14 examined this reaction on Pt(111) by IRAS. In contrast to the SIRAS study, they did not observe the adsorption of formate and instead observed a very weak band around 1470 cm-1. The authors assigned this band to the symmetric CH2 bending mode of methylene glycolate adsorbed on the electrode surface and concluded that methylene glycolate is the main reaction intermediate. The adsorption of formate was not reported also in the preceding IRAS studies on polycrystalline Pt electrodes.4-8 Therefore, the discrepancy between SEIRAS and IRAS studies is not likely to arise from the difference in the crystallographic orientations of the electrode surfaces. Regarding to this issue, we have suggested in a previous publication13 that IRAS is not suitable for detecting short-lived intermediates due to limited mass transport arising from the use of a thin solution layer confined between the electrode surface and a cell window. Since mass transport between the thin layer and the reservoir is severely hindered, the concentration of the reactant in the thin layer is largely reduced, and the products are accumulated during
10.1021/jp0743020 CCC: $37.00 © 2007 American Chemical Society Published on Web 09/21/2007
Electrocatalytic Oxidation of Formaldehyde
J. Phys. Chem. C, Vol. 111, No. 41, 2007 15075
the spectral measurements.13 Therefore, the reaction cannot be observed by IRAS. It is known that potential and current periodically oscillate under galvanostatic (constant current) and potentostatic (constant potential) formaldehyde oxidation, respectively.10-12,16-18 The oscillations are described in terms of cycles of autocatalytic adsorption and the consumption of some species (including poisoning species).10-12,16-18 For a full understanding of oscillatory behavior, deep insight into the reaction mechanism and dynamics is required. Although the oscillations have been investigated by using several techniques including electrochemical methods,10-12,17,18 UV-vis reflection spectroscopy,19,20 quartz crystal microbalance (QCM),12 and second harmonic generation (SHG),20,21 the details have not yet been understood well. The aim of the present study is to understand formaldehyde oxidation at the molecular scale through the analysis of potential oscillations by using time-resolved ATR-SEIRAS. As has been demonstrated in our previous studies of potential and current oscillations in formic acid oxidation,22 the high sensitivity, high molecular specificity, and free mass transport advantage of ATRSEIRAS enable us to monitor the reaction with millisecond time resolutions. 2. Experimental Procedures Experimental details of ATR-SEIRAS are described elsewhere.15,23 The electrochemical cell used in most experiments was a glass one with a three-electrode design.23b In some measurements under flow conditions, a cell made of Kel-F with a small volume (1.5 mL) was used. A thin (∼50 nm) Pt film deposited on a hemicylindrical Si prism (1 cm radius and 2.5 cm length, Pier Optics) by an electroless plating process13,24 was used as the working electrode. Pt-gauze served as the counter electrode, and a reversible hydrogen electrode (RHE) in the supporting electrolyte solution (0.5 M H2SO4) was used as the reference electrode. Potential (E) or current (i) was controlled by a potentiostat/galvanostat (EG&G PARC model 263A). Prior to use, the electrode surface was cleaned by repetitive potential sweeps between 0.05 and 1.55 V in the supporting electrolyte. The geometrical area of the surface in contact with the electrolyte solution was 1.77 cm2, while the real surface area was estimated to be ∼10 cm2 from the hydrogen adsorption/desorption peaks in the cyclic voltammogram by assuming 210 µC cm-2 for monolayer adsorption. The large surface roughness factor of ∼6 stems from a surface roughness in the nanometer scale.13 The resistance (Rsol) of the electrochemical system determined by the current interruption method25 was only 5-10 Ω, but it gives large ohmic drop for large current. The ohmic drop was compensated in some voltammetric measurements by using the iR-compensation mode of the potentiostat. SEIRA spectra were recorded on a Bio-Rad FTS-60A/896 Fourier transform infrared spectrometer equipped with a MCT detector and a home-built single-reflection accessory (incident angle of 70°). The spectrometer was operated in the rapid scan mode with a spectral resolution of 4 cm-1. Time resolution used in recording spectra was either 80 ms (single-interferometer scan) or 0.38 s (four-scan averaging). Spectra are shown in the absorbance units defined as A ) -log(I/I0), where I and I0 represent the intensities of the IR signals reflected from the Pt surface in the electrolyte solution with and without formaldehyde, respectively. The reference spectrum was recorded at 0.05 V before adding formaldehyde into the supporting electrolyte solution.
Figure 1. Typical potential oscillations observed in 0.5 M H2SO4 + 0.1 M formaldehyde for applied currents of 2.5 (a), 5 (b), and 10 (c) mA. Bars in the figure represent the time regions where time-resolved SEIRAS measurements were carried out.
The supporting electrolyte solution was prepared from Milli-Q water (>18 MΩ, TOC 1.1 V corresponds to the negative potential side of peak III. Figure 8 shows a series of SEIRA spectra of the electrode surface acquired simultaneously with the cyclic voltammogram (solid trace in Figure 7). Four interferograms were coadded to each spectrum to enhance the signal-to-noise ratio (it required 0.38 s). As in the spectra collected during the oscillations (Figure 3), the bands of adsorbed CO and formate and of interfacial water are identified. The band at 1800-1820 cm-1 may be CO adsorbed at hollow sites (COH). As the potential is scanned positively, the CO bands disappear due to the oxidative removal of CO, and the νs(OCO) mode of formate emerges at 1324 cm-1. The formate band completely disappears around 1.3 V. The spectral changes on the positive-going potential sweep are nearly reversed on the negative-going potential sweep.
Samjeske´ et al.
Figure 8. Series of SEIRA spectra of the Pt electrode acquired simultaneously with the solid cyclic voltammogram in Figure 8. Time resolution used was 0.38 s.
Figure 9. Comparison of cyclic voltammogram (same as the solid trace in Figure 8) and potential dependence of the band intensities of COL (b), COB (c), and formate (d) taken from Figure 8.
To facilitate the comparison of the spectral and voltammetric data recorded simultaneously, the integrated band intensities of COL, COB (including COH), and formate are plotted in Figure 9 as a function of the applied potential (panels b-d, respectively) and are compared to the cyclic voltammogram (panel a). In the beginning of the measurement (at 0.05 V), the intensity of the COL band is close to that observed for a saturated CO monolayer, and thus, the electrode surface is almost fully covered by CO. The band intensities of COL and COB decrease gradually for the increase in potential up to 0.7 V and then rapidly decrease at more positive potentials. They completely disappear around 0.9 V, which is about 0.1 V more positive than the potential for the complete oxidation of a saturated CO monolayer in the pure electrolyte measured at the same sweep rate,13 suggesting that CO oxidation is suppressed by the adsorption of formate. Associated with the rapid oxidation of CO at E greater than ∼0.7 V, the formate band appears and
Electrocatalytic Oxidation of Formaldehyde
J. Phys. Chem. C, Vol. 111, No. 41, 2007 15079
grows. It is worth noting that the oxidation current begins to flow concomitantly with the appearance of the formate band. The band intensity of formate reaches a maximum around 0.9 V (close to peak II) and completely disappears at 1.3 V. This band reappears at ∼0.85 V on the negative potential sweep due to the reduction of the electrode surface (vide infra) and reaches another maximum at 0.6 V (close to peak IV). The CO bands reappear around 0.6 V and grow at less positive potentials. The complex behavior of COB may be due to the conversion to COL.31 When the potential sweep rate was decreased, the intensity profiles of the CO bands on the positive- and negativegoing sweeps shifted to negative and positive directions, respectively, and the hysteresis became smaller as in the case of formic acid oxidation26 (data not shown). The results indicate three important aspects: (1) adsorbed CO prevents the adsorption of formate and suppresses formaldehyde oxidation; (2) the formation and removal of CO are equilibrated around 0.6 V (the center potential of the hysteresis of the COL band intensity plot) under stationary conditions, and the CO formation is faster than oxidation at lower potentials; and (3) the oxidation current is well-correlated with the band intensity of formate, although they are not in a linear relation. Since the adsorption of formate
CH2(OH)2 f HCOOads + 3H+ + 3e-
(6)
carries three electrons, the intimate relation between oxidation current and formate band intensity is a reasonable consequence. For the flow of the observed large current, however, adsorbed formate must be oxidized (or decomposed) to CO2
HCOOads f CO2 + H+ + e-
(7)
or desorbed from the surface to yield formic acid in the solution.
HCOOads + H+ + e- f HCOOH
(8)
In the SEIRAS studies of methanol32 and formic acid22,24,26 oxidation, we showed that formate is adsorbed on Pt and that the oxidation current increases as the formate band intensity increases, as in formaldehyde oxidation. By using isotope labeled methanol (12CH3OH and 13CH3OH) and formic acid (H12COOH and H13COOH), we demonstrated that the adsorbed formate is readily replaced by formate newly supplied from methanol and formic acid in solution.22b,26,32 The result clearly shows that adsorbed formate is oxidized to CO2 or desorbed to yield formic acid. Although the same experiment could not be performed in the present study because 13C labeled formaldehyde was not variable, the situation will be the same in formaldehyde oxidation. Although adsorbed formate is very likely to be the main reaction intermediate in the direct path as discussed previously, there still remain two questions. First, Batista and Iwasita14 clearly showed by IRAS that formic acid is produced in the solution by the partial oxidation of formaldehyde (the bands characteristic to formic acid are not seen in Figure 8 because ATR-SEIRAS selectively probes the interface15). Two possibilities are considered for the adsorption of formate: (1) formate is directly formed from methylene glycol on the surface and then desorbed to yield formic acid in the solution and (2) formic acid is formed first and then adsorbed on the surface as formate. To make clear the route for formate adsorption, we carried out the same experiments as in Figures 7 and 8 by flowing a fresh formaldehyde solution at different flow rates aiming to wash away formic acid derived from formaldehyde from the interface.
Figure 10. Proposed reaction scheme for formaldehyde oxidation on Pt in acid. The direct path via adsorbed formate and indirect path via adsorbed CO are kinetically coupled (for details, see text).
However, no noticeable changes were observed in the spectral features and band intensities. The result strongly suggests that formate is formed directly from methlyene glycol on the surface and that formic acid is produced by formate desorption. Otherwise, the band intensity should become smaller as the flow rate increases. The second question is about the main route after formate production. Batista and Iwasita14 reported that formic acid is produced more efficiently than CO2 (63% formic acid and 37% CO2 after 2 min polarization at the constant potential of 0.6 V). The result suggests that formate is mostly desorbed from the surface before being oxidized to CO2 under the given conditions. However, formic acid should be oxidized to CO2 at this potential,22,26 and thus, the high formic acid yield indicates that the electrode surface was highly covered by CO so that the oxidation of formic acid was significantly suppressed. We suggested in previous studies of formic acid oxidation22,26 that the rate of formate oxidation is expressed by the function of coverages of formate and vacant sites (θf and θv, respectively). If we assume that the oxidation of formate occurs randomly, the rate is given by
dθf ) koxθfθv dt
(9)
where kox is the rate constant. The coverage of vacant sites, θv ) 1 - θf - θCO, was introduced to represent the necessity of an adjacent vacant site for the C-H scissoring of formate by an analogy of formate and acetate decomposition on metal surfaces in UHV.33 This rate equation suggests that the oxidation of formate to CO2 becomes faster as the CO coverage decreases and that formate suppresses its oxidation at high formate coverage exceeding (1 - θCO)/2 (self-poisoning). That is, formate is a reaction intermediate in the direct path and also can be poison depending on the conditions. The kinetics analysis of the isotopic substitution of adsorbed formate during formic acid oxidation at 0.6 V suggests that adsorbed formate is mostly oxidized to CO2 when the coverage of CO is small.22b,26 Although the contributions of the two competitive processes to the total current for formaldehyde oxidation cannot be evaluated from the present SEIRAS experiment, it is stressed again that formic acid is finally oxidized to CO2 via adsorbed formate.22b,26 On the basis of the aforementioned experimental results and discussion, we propose a reaction scheme of formaldehyde oxidation in Figure 10. Methylene glycol, the hydrated form of formaldehyde, gives adsorbed CO and formate. Adsorbed CO is oxidized to CO2 by adsorbed water or OH (eqs 4 and 5). Formate is oxidized to CO2 (eq 7) or desorbed to yield formic acid (eq 8). Formic acid is finally oxidized to CO2 via adsorbed formate or adsorbed CO produced by dehydration.22,26 It is not clear at present whether CO and formate are produced from
15080 J. Phys. Chem. C, Vol. 111, No. 41, 2007 the same precursor. Olivi et al.7a observed an absorption band at 1710 cm-1 by IRAS and assigned this band to the CdO stretching mode of non-hydrated formaldehyde, from which they concluded that the non-hydrated formaldehyde is a precursor for the formation of CO. However, the corresponding band is totally missing in the SEIRA spectra shown in Figures 3 and 8 despite the very high sensitivity of SEIRAS. Taking into account that SEIRAS selectively observes the interface,15 this band should arise from a species in the bulk solution and is assigned more reasonably to the CdO stretching mode of formic acid accumulated in the solution as Batista and Iwasita showed.14 The reaction scheme is totally different from that proposed by Batista and Iwasita.14 They did not observe the νsym(O-CO) mode of adsorbed formate and instead observed two bands at ∼1470 and ∼1230 cm-1 on a Pt(111) electrode. They assigned these bands to the symmetric CH2 deformation of methyl glycolate (H2COO) and COOH adsorbed on the electrode surface, respectively, and concluded that adsorbed methylene glycolate is the main reactive intermediate in formaldehyde oxidation. In marked contrast, no corresponding bands were observed in our SEIRA measurements. A possible reason for this discrepancy has been described in the Introduction. Even if the assignment of the 1470 cm-1 band to the symmetric CH2 deformation were reasonable, this band cannot be ascribed definitely to methyl glycolate because its characteristic C-O stretching mode was not observed in the expected spectral range of 1000-1200 cm-1. Other species such as hydrocarbon contaminants also can be candidates. Nevertheless, methyl glycolate is very likely to be a short-lived precursor for formate: adsorbed formate can be formed from methyl glycolate by breaking a C-H bond. Finally, it should be noted that the direct and indirect paths are kinetically coupled. We have already mentioned that the direct path via formate is affected by adsorbed CO (eq 9) and that the oxidation of CO is suppressed by coadsorption of formate. Since CO oxidation requires an oxygen source (H2O or OH) and its rate is proportional to θCOθv in the LangmuirHinshelwood mechanism, adsorbed formate suppresses CO oxidation by reducing θv. Adsorbed formate also affects the formation of CO because its rate should be proportional to θvn (n is the number of adjacent sites required for the dehydration of methylene glycol yielding CO). We will show later that the kinetic coupling of the direct and indirect paths plays an important role in the oscillations. 3.4. Negative Differential Resistance (NDR). In the sawlike oscillations (Figure 4), the potential steeply rises from ∼0.6 V to the maximum (0.75-0.85 V). The result suggests that formaldehyde oxidation is suppressed more strongly at higher potentials (i.e., NDR12,16,18). This process is autocatalytic, and the increase further rises the potential. Two models have been proposed: surface oxidation12,19,20 and adsorption of water on the surface.18a The surface oxidation model has been proposed from the periodic changes in mass (measured by QCM)12 and reflectivity (in the UV-vis region19,20) of the electrode observed during oscillations. The periodic reflectivity change is observed also in our SEIRA spectra shown in Figure 3 (the periodic shift of the baseline). In fact, the Pt surface is oxidized at E >0.8 V in the pure electrolyte. Nevertheless, the change in mass and reflectivity of the electrode can be explained also by the adsorption of formate. We show next that the NDR caused by surface oxidation contributes to the oscillations only for a large applied current.
Samjeske´ et al.
Figure 11. (a) Cyclic voltammogram for a Pt electrode in 0.5 M H2SO4 at 50 mV s-1 (dashed trace) and the plot of the baseline level of the SEIRA spectra at 2500 cm-1 (closed circles) taken from the set of spectra acquired simultaneously with the voltammogram. (b) Corresponding baseline level data in 0.5 M H2SO4 + 0.1 M formaldehyde taken from Figure 8. The thin solid curve is the intensity plot for the formate band (Figure 9d) multiplied by 0.01. Open circles are the difference between the observed baseline shift and the scaled formate band intensity, representing oxidation and reduction of the Pt surface.
In Figure 11a, the baseline level of the SEIRA spectra of a Pt electrode (at 2500 cm-1) recorded during a potential cycling at 50 mV s-1 in 0.5 M H2SO4 is plotted as a function of potential. As compared to the cyclic voltammogram recorded simultaneously (Figure 11, dashed trace), it is clear that the upshift of the baseline level (that is, the decrease in the reflectivity of the surface) at E >0.8 V arises from the oxidation of the surface. On the other hand, the down-shift of the baseline at E 0.7 V on the positive-going sweep and 0.3 < E < 0.85 V on the negative-going sweep (Figure 9), the data show that the adsorption of formate also can change the baseline profile. The contributions of formate adsorption and surface oxidation to the baseline shift were separated by assuming that the baseline shift caused by adsorbed formate is proportional to formate coverage (band intensity). More precisely, the formate band intensity plot in Figure 9d was subtracted from the observed baseline plot by scaling the former so that the broad tail at 0.40.7 V on the negative-going sweep disappears (scaling factor of 0.01, thin curve). The result, the baseline shift caused by surface oxidation, is shown by open circles in Figure 11. The plot shows that the onset potential of surface oxidation is positively shifted from ∼0.8 V in the pure supporting electrolyte to ∼0.9 V in the formaldehyde solution, while the reduction of surface oxide on the negative-going scan is scarcely affected by the addition of formaldehyde. Adsorbed formate is believed to suppress the surface oxidation by blocking the adsorption of OH, the precursor for surface oxidation.34 It is worth noting that the surface is not oxidized in the potential range between
Electrocatalytic Oxidation of Formaldehyde
Figure 12. (a) Linear sweep voltammogram for a Pt electrode in 0.5 M H2SO4 + 0.1 M formaldehyde recorded at 50 mV s-1 after stepping the potential from 1.3 to 0.4 V (upper panel). (b) Potential dependence of integrated band intensities of COL, COB, and formate taken from the set SEIRA spectra simultaneously acquired with the voltammogram. The ohmic drop (Rs ) 6 Ω) was not compensated for so that peak II observed at 0.85 V in Figure 9 was shifted to 0.93 V in this measurement.
0.35 and 0.85 V where the quasi-sinusoidal and saw-like potential oscillations occur for small and medium applied currents. However, surface oxidation can contribute to the reverse saw-like oscillations observed for larger currents in which the maximum potential (φ) reaches 0.9 V (Figure 2). The second model for the NDR was proposed by Okamoto and Tanaka.18a They clearly showed the NDR by linear sweep voltammetry and proposed that the adsorption of water suppresses the direct path (assuming that COOH is the reactive intermediate) at 0.6-0.7 V to yield the NDR. In fact, the NDR can be observed in the voltammogram recorded soon after stepping the potential from 1.3 to 0.4 V as shown in Figure 12 (the positive side of a weak peak centered at ∼0.6 V, peak I). The polarization at 1.3 V was aimed to remove adsorbed CO by oxidation. However, we do not think that water plays the decisive role because adsorbed formate is the reactive intermediate in the direct path and removes water from the surface (Figure 8). Instead, we suggest next that the kinetic coupling of the direct and indirect paths gives the NDR. The spectral data taken from a set of time-resolved spectra recorded simultaneously with the voltammogram in Figure 12 are shown in the same figure (lower panel). Formate and COB are quickly adsorbed on the surface, and their band intensities are nearly constant in the peak I region, while the band intensity of COL increases as the potential is increased up to 0.73 V and then decreases to zero at ∼0.9 V. The decay of current at 0.4 V is apparently not due to double-layer charging but to the formation of adsorbed COB and formate (eqs 3 and 6). The band intensity of formate is almost constant at 0.4-0.8 V, and thus, the background current of 4-5 mA in this range is mostly ascribed to the direct path via formate. On the other hand, the profile of the COL band intensity suggests that the accumulation and oxidation of COL (i.e., the indirect path) constitute peak I. The accumulation of COL is slowed down with potential due to its potential dependence, the decrease in free sites, and also the oxidation. We have suggested that formation and oxidation of COL are equilibrated around 0.6 V under stationary conditions. Therefore, the band intensity of COL is expected to be reduced at the positive side of peak I due to the faster oxidation than supply. However, the band intensity of COL is constant in
J. Phys. Chem. C, Vol. 111, No. 41, 2007 15081 this potential range under the potential sweep conditions apparently due to the slow kinetics of CO oxidation. Coadsorbed formate also can slow down CO oxidation by reducing θv and reduces the oxidation current. In addition, adsorbed CO suppresses the oxidation of adsorbed formate by blocking adjacent vacant sites (eq 9) and reduces the current by the direct path. Although the suppression of formate oxidation leads to the desorption of formate to formic acid, this sub-path carries less electrons (2e-). As a result of the kinetic coupling of the direct and indirect paths, the NDR appears. The oxidation of CO at higher potentials facilitates the adsorption and oxidation of formate and increases the oxidation current to give peak II. The second NDR of peak II is ascribed to the oxidation of the electrode surface (Figure 11). The coadsorption of oxygen is known to stabilize adsorbed formate on Pt in UHV.35 The stabilization of formate on the oxidized Pt electrode surface also has been demonstrated by using isotopic labeled formic acids.26 3.5. Mechanism of Potential Oscillations. On the basis of the dual path mechanism and kinetic coupling of the direct and indirect paths, we qualitatively discuss next the mechanism of the oscillations shown in Figures 1 and 2. First of all, it should be noted that formic acid produced by the desorption of formate is not essential in the oscillations because oscillations are observed with rotating electrodes as well.10-12,17 The electrode surface is almost fully covered by CO, and formaldehyde oxidation is totally suppressed in the beginning. Therefore, the application of current raises the potential and oxidizes adsorbed CO. For small applied currents, CO is oxidized only slightly, and thus, the contribution of the direct path via formate is small. The formation of CO consumes 2eand lowers the potential, but its rate becomes smaller as the CO coverage increases due to the decrease of vacant sites. Accordingly, the potential rises. The rise in potential oxidizes CO and activates the direct path. However, since the coverage of formate is small and the oxidation of formate is highly suppressed by CO, the potential rises up to a value at which the direct path becomes fast enough to consume the applied current. The activation of the direct path decreases the potential again, and this cycle repeats itself to exhibit the quasi-sinusoidal oscillations. Since CO formation is favored at E less than ∼0.6 V and CO oxidation is favored at E greater than ∼0.6, the potential oscillations occur around 0.6 V. The quasi-sinusoidal oscillations for iappl ) 1 mA (Figures 2a and 6a) can be explained in this way. The oxidation of CO at high potentials is faster than its formation at low potentials so that CO coverage gradually decreases with time (Figure 6a). The decrease in CO coverage increases the contribution of the direct path and changes the quasi-sinusoidal oscillation waveform to the saw-like one. For larger applied currents, a large amount of CO is oxidized, and thus, the saw-like oscillations appear immediately (Figures 1 and 4). The initial slow potential rise in the saw-like oscillations is ascribed to the suppression of the direct path by CO accumulation. Once the potential is raised up to ∼0.6 V, the potential is autocatalytically raised by the NDR shown in Figure 12. Although the rise in potential to E greater than ∼0.7 V oxidizes CO quickly and facilities the adsorption of formate, the potential keeps rising until the CO coverage goes down to some level at which formate oxidation is accelerated. Once the CO coverage is reduced to that level, the direct path is explosively accelerated, and the potential suddenly drops. If the potential is raised slightly above the onset potential of surface oxidation (∼0.9 V) by larger applied current as shown
15082 J. Phys. Chem. C, Vol. 111, No. 41, 2007 in Figure 2b, adsorbed CO is quickly oxidized. However, the surface oxidation suppresses the direct path, and thus, the potential is decreased only slightly. When the potential is decreased to the potential at which surface oxide is reduced (∼0.85 V), the direct path is accelerated, and thus, the potential steeply drops. The accumulation of CO at low potentials raises the potential again, and this cycle repeats itself to give the reverse saw-like oscillation. If the potential is raised well into the second NDR region of peak II where the electrode surface is heavily oxidized, formate is highly stabilized, and thus, the potential is autocatalytically raised further. Accordingly, oscillations do not occur for very large currents. Recently, Karantonis et al.36 proposed a mathematical model in which the adsorption of formate is neglected. The model simulates well the bistability, oscillations, and bifurcations of the electrocatalytic oxidation of formaldehyde. Taking into account that many mathematical models assuming different reaction mechanisms successfully simulate electrochemical oscillations in formic acid oxidation,22b the results do not confirm the validity of the mathematical model. Mokouyama et al.37 incorporated formate adsorption into their mathematical model for the potential oscillations in galvanostatic formic acid oxidation and succeeded in simulating the oscillatory behaviors of both the potential and coverages of adsorbed species. Since the essence of the oscillations (the dual path mechanism and kinetic coupling) is common in formic acid oxidation and formaldehyde oxidation, the oscillations in formaldehyde oxidation will be simulated by extending their model to include the formation of formate from methylene glycol. 4. Conclusion Time-resolved ATR-SEIRAS was used successfully to monitor the dynamics of adsorbates on a Pt electrode surface during potential oscillations under galvanostatic formaldehyde oxidation. SEIRAS revealed that the formaldehyde oxidation occurs via a direct path using formate and an indirect path using CO. Adsorbed formate is oxidized (or decomposed) to CO2 or desorbed to yield formic acid. The oxidation and desorption are competitive, and the former is faster at low CO coverage. Formic acid is finally oxidized to CO2 via formate and CO. The direct and indirect paths are kinetically coupled, and the reaction intermediates suppress their oxidation of each other. The reaction intermediates also suppress their own oxidation at high coverage. Adsorbed formate also suppresses the oxidation of the electrode surface, and surface oxidation suppresses the direct path by stabilizing formate. The observed potential oscillations were qualitatively explained by the dual path mechanism and the interplays among the reaction processes. Finally, oscillations are known to be accompanied by spatiotemporal pattern formation.16 Studies of the pattern formation are required for full understanding of electrocatalytic formaldehyde oxidation. Acknowledgment. This work was supported by the Ministry of Education, Culture, Sports, Science and Technology of Japan (Grant-in-Aid for Basic Research 18350038 and Japan Society for the Promotion of Science (JSPS) Fellows 18‚06343). G.S. acknowledges the JSPS for a Postdoctoral Fellowship for Foreign Researchers. Supporting Information Available: Potential oscillation results. This material is available via the Internet at http:// pubs.acs.org.
Samjeske´ et al. References and Notes (1) (a) Parsons, R.; VanderNoot, T. J. Electroanal. Chem. 1988, 257, 9. (b) Beden, B.; Leger, J. M.; Lamy, C. Electrocatalytic oxidation of oxygenated aliphatic organic compounds at noble metal electrode. In Modern Aspects of Electrochemistry; Bockris, J. O. M., Conway, B., Eds.; Plenum Press: New York, 1992; Vol. 22, p 97. (c) Lipkowski, J.; Ross, P. N., Eds. Electrocatalysis; Wiley-VCH: New York, 1998. (d) Vielstich, W.; Lamm, A.; Gasteiger, H. A., Eds. Handbook of Fuel Cells: Fundamentals Technology and Applications; Wiley: Chichester, U.K., 2003. (2) Korzeniewski, C.; Childers, C. L. J. Phys. Chem. B 1998, 102, 489. (3) Guthrie, J. P. Can. J. Chem. 1975, 53, 898. (4) Avramov-Ivic, M.; Adzic, R. R.; Bewick, A.; Razaq, M. J. Electroanal. Chem. 1988, 240, 161. (5) Kitamura, F.; Takahashi, M.; Ito, M. Chem. Phys. Lett. 1986, 123, 273. (6) Nishimura, K.; Ohnishi, R.; Kunimatsu, K.; Enyo, M. J. Electroanal. Chem. 1989, 258, 219. (7) (a) Olivi, P.; Bulhoes, L. O. S.; Leger, J. M.; Hahn, F.; Beden, B.; Lamy, C. J. Electroanal. Chem. 1994, 370, 241. (b) Olivi, P.; Bulhoes, L. O. S.; Leger, J. M.; Hahn, F.; Beden, B.; Lamy, C. Electrochim. Acta 1996, 41, 927. (8) Sun, S. G.; Lu, G. Q.; Tian, Z. W. J. Electroanal. Chem. 1995, 393, 97. (9) Beltowska-Brzezinska, M.; Heitbaum, J.; Vielstich, W. Electrochim. Acta 1985, 30, 1465. (10) Xu, Y.; Schell, M. J. Phys. Chem. 1990, 94, 7134. (11) (a) Nakabayashi, S.; Kira, A. J. Phys. Chem. 1992, 96, 1021. (b) Nakabayashi, S.; Uosaki, K. Chem. Phys. Lett. 1994, 217, 163. (12) Koper, M. T.; Hachker, M.; Beden, B. J. Chem. Soc., Faraday Trans. 1996, 92, 3975. (13) Miki, A.; Ye, S.; Senzaki, T.; Osawa, M. J. Electroanal. Chem. 2004, 563, 23. (14) Batista, E. A.; Iwasita, T. Langmuir 2006, 22, 7912. (15) (a) Osawa, M. Bull. Chem. Soc. Jpn. 1997, 70, 2861. (b) Osawa, M. Top. Appl. Phys. 2001, 81, 163. (c) Osawa, M. Surface enhanced infrared absorption spectroscopy. In Handbook of Vibrational Spectroscopy; Chalmers, J. M., Griffiths, P. R., Eds.; Wiley: Chichester, U.K., 2002; Vol. 1, p 785. (d) Osawa, M. In-situ Surface-enhanced Infrared Spectroscopy of the Electrode/Solution Interface. In Diffraction and Spectroscopic Methods in Electrochemistry (AdVances in Electrochemical Science and Engineering, Vol. 9); Alkire, R. C., Kolb, D. M., Lipkowski, J., Ross, P. N., Eds.; WileyVCH: New York, 2006; p 269. (16) As reviews: (a) Hudson, J. L.; Tsotsis, T. T. Chem. Eng. Sci. 1994, 49, 1493. (b) Koper, M. T. M. AdV. Chem. Phys. 1996, 92, 161. (c) Krischer, K. In Modern Aspects of Electrochemistry; Conway, B. E., Bockris, J. O. M., White, R. E., Eds.; Plenum: New York, 1999; Vol. 32, p 1. (17) Schell, M.; Albahadily, F. N.; Safar, J.; Xu, Y. J. Phys. Chem. 1989, 93, 4806. (18) (a) Okamoto, H.; Tanaka, N. Electrochim. Acta 1993, 38, 503. (b) Okamoto, H.; Tanaka, N.; Naito, M. J. Phys. Chem. 1998, 102, 7343. (c) Okamoto, H.; Tanaka, N.; Naito, M. J. Electrochem. Soc. 2000, 147, 2629. (19) Hachkar, M.; Demartinez, M. C.; Rakotondrainibe, A.; Beden, B.; Lamy, C. J. Electroanal. Chem. 1991, 302, 173. (20) Nakabayashi, S.; Yagi, I.; Sugiyama, N.; Tamura, K.; Uosaki, K. Surf. Sci. 1997, 386, 82. (21) Mishina, E.; Karantonis, A.; Yu, Q. K.; Nakabayashi, S. J. Phys. Chem. B 2002, 106, 10199. (22) (a) Sameske´, G.; Osawa, M. Angew. Chem., Int. Ed. 2005, 44, 5694. (b) Sameske´, G.; Miki, A.; Ye, S.; Yamakata, A.; Mukouyama, Y.; Okamoto, H.; Osawa, M. J. Phys. Chem. B 2005, 109, 23509. (23) (a) Osawa, M.; Ataka, K.; Yoshii, K.; Yotsuyanagi, T. J. Electron Spectrosc. Relat. Phenom. 1993, 64-65, 371. (b) Ataka, K.; Yotsuyanagi, T.; Osawa, M. J. Phys. Chem. 1996, 100, 10664. (24) Miki, A.; Ye, S.; Osawa, M. Chem. Commun. 2002, 1500. (25) Bard, A. J.; Faulkner, L. R. Electrochemical Methods, 2nd ed.; Wiley: New York, 2001; p 645. (26) Samjeske´, G.; Miki, A.; Ye, S.; Osawa, M. J. Phys. Chem. B 2006, 110, 16559. (27) Nakamoto, K. Infrared and Raman Spectra of Inorganic and Coordination Compounds, 4th ed.; Wiley: New York, 1986; p 232. (28) (a) Columbia, M. R.; Crabtree, A. M.; Thiel, P. A. J. Am. Chem. Soc. 1992, 114, 1231. (b) Ibid, J. Electroanal. Chem. 1993, 345, 93. (29) Osawa, M.; Tushima, M.; Mogami, T.; Sameske´, G.; Yamakata, A., to be published. (30) (a) Nichols, R. J. In Adsorption of Molecules at Metal Electrodes; Lipkowski, J., Ross, P. N., Eds.; Wiley-VCH: New York, 1992; p 347. (b) Iwashita, T.; Nart, F. C. Prog. Surf. Sci. 1997, 55, 271.
Electrocatalytic Oxidation of Formaldehyde (31) Iwasita, T.; Xia, X.; Herrero, E.; Leiss, H.-D. Langmuir 1996, 12, 4260. (32) Chen, Y. X.; Miki, A.; Ye, S.; Sakai, H.; Osawa, M. J. Am. Chem. Soc. 2003, 125, 3680. (33) (a) Falconer, J. L.; Madix, R. J. Surf. Sci. 1974, 46, 473. (b) Li, Y.; Bowker, M. Surf. Sci. 1993, 285, 219. (c) Sharp, R. G.; Bowker, M. J. Phys.: Condens. Matter 1995, 7, 6379.
J. Phys. Chem. C, Vol. 111, No. 41, 2007 15083 (34) Angerstein-Kozlowska, H.; Conway, B. E.; Sharp, W. B. A. J. Electroanal. Chem. 1973, 43, 9. (35) Columbia, M. R.; Thiel, P. A. Chem. Phys. Lett. 1994, 220, 167. (36) Karantonis, A.; Koutsaftis, D.; Kouloumbi, N. Chem. Phys. Lett. 2006, 422, 78. (37) Mukouyama, Y.; Kikuchi, M.; Samjeske´, G.; Osawa, M.; Okamoto, H. J. Phys. Chem. B 2006, 110, 11912.