pubs.acs.org/Langmuir © 2010 American Chemical Society
CO Electroxidation on Gold in Alkaline Media: A Combined Electrochemical, Spectroscopic, and DFT Study Paramaconi Rodriguez,*,† Nuria Garcia-Araez,†,‡ Andrey Koverga,† Stefan Frank,† and Marc T. M. Koper† †
Leiden Institute of Chemistry, Leiden University, P.O. Box 9502, 2300 RA Leiden, The Netherlands, and ‡FOM Institute for Atomic and Molecular Physics (AMOLF), P.O. Box 41883, 1009 DB Amsterdam, The Netherlands Received April 8, 2010. Revised Manuscript Received June 4, 2010
The aim of the present work is to provide a deeper understanding of gold catalysis for CO electrooxidation in alkaline media, through a combined electrochemical, spectroscopic, and DFT study. Voltammetric and spectroscopic measurements evidence that the amount of CO irreversibly adsorbed on gold increases as the adsorption potential becomes more negative (vs SHE). This explains why higher CO coverages can be achieved in more alkaline solutions, since the value of adsorption potential vs RHE becomes more negative vs SHE with increasing pH. On the other hand, the combination of FTIRRAS experiments and DFT calculations shows that the adsorption site of irreversibly adsorbed CO on Au(111) depends on the value of the adsorption potential. It is concluded that CO adsorption on top sites takes place at all studied potentials, and hollow and bridge sites also become occupied for adsorption potentials lower and higher than 0 V vs RHE, respectively. However, it should be noted that our DFT calculations give values of the CO binding energies that are not strong enough to explain CO irreversible adsorption. This may be partly attributed to the fact that OH coadsorption is not included in the calculations. Indeed, this work presents two experimental facts that suggest that CO adsorption on gold promotes the coadsorption of OH species: (i) CO irreversibly adsorbed on Au(111) and Au(100) leads to an unusual voltammetric feature, whose charge indicates the stabilization of one OH species per adsorbed CO species; (ii) the apparent transfer coefficient of this unusual state is close to unity, suggesting that it is due to a presumed structural transformation coupled to OH adsorption. Finally, the effect of the adsorption potential on the bulk CO electrooxidation is also studied. It is found that, on Au(111), an increased occupation of CO on multifold (hollow) sites seems to result in a less efficient catalysis. However, on Au(110), an increased coverage of CO on top sites does not produce any significant change in catalysis.
1. Introduction The catalytic oxidation of carbon monoxide (CO) is one of the most intensively studied catalytic reactions in both heterogeneous and homogeneous media. Interestingly, gold nanoparticles exhibit exceptional catalytic activity toward gas-phase CO oxidation and other reactions, as shown in 1989 by Haruta’s work.1 Ever since then, gold catalysis has attracted increasing interest, due to its potential applications in important technological processes, such as the water-gas shift reaction2 and the oxidation of hydrocarbons.3 Noteworthy, as opposed to the situation in the gas phase, the electrocatalytic activity of gold is also present for the bulk material and not just for the supported nanoparticles. Indeed, CO electrooxidation on gold polycrystalline electrodes in alkaline media takes place at ca. 0.5 V lower overpotentials than in the case of platinum,4 the latter being the most common catalyst in fuel cells.5 Another striking aspect of gold electrochemistry, which is not applicable to gas-phase reactions, is the very important effect of the pH. Alkaline media appear to be optimum for CO electrooxidation *Corresponding author. FAX þ31-71-5274451, E-mail: rodriguezperezpb@ chem.leidenuniv.nl. (1) Haruta, M.; Yamada, N.; Kobayashi, T.; Ijima, S. J. Catal. 1989, 115, 301. (2) Fu, Q.; Saltsburg, H.; Flytzani-Stephanopoulos, M. Science 2003, 301, 935– 938. (3) Hughes, M. D.; Xu, Y.-J.; Jenkins, P.; McMorn, P.; Landon, P.; Enache, D. I.; Carley, A. F.; Attard, G. A.; Hutchings, G. J.; King, F.; Stitt, E. H.; Johnston, P.; Griffin, K.; Kiely, C. J. Nature 2005, 437, 1132–1135. (4) Kita, H.; Nakajima, H.; Kimitaka, H. J. Electroanal. Chem. 1985, 190, 141– 156. (5) Fuel Cell Catalysis: A Surface Science Approach; Koper, M. T. M., Ed.; John Wiley and Sons, Inc: Hoboken, NJ, 2009.
Langmuir 2010, 26(14), 12425–12432
on gold.4 Nevertheless, despite the extensive research in this field, the origin of gold catalysis is not yet fully understood. Recently, our group reported experimental evidence on the unexpected irreversible adsorption of CO on gold single-crystal electrodes in alkaline media.6 In the absence of CO in solution, CO remains chemisorbed on the gold surface, leading to the appearance of an unusual adsorption state on Au(111) and hexagonally reconstructed Au(100) electrodes. This adsorption state is associated with the appearance of near-reversible voltammetric peaks at potentials close to the onset of the oxidation of CO in solution. We tentatively proposed that this voltammetric feature is associated with the coadsorption of OH- from the alkaline solution at low potentials, in view of its marked pH dependence. This picture is supported by recent DFT calculations that show that CO promotes the adsorption of its own oxidant (most probably, OH species).7 Furthermore, voltammetric experiments with hanging meniscus bead-type single-crystal electrodes show that the reaction order in CO concentration achieves values higher than unity,7,8 in agreement with the self-promotion mechanism for CO electrooxidation suggested from the DFT calculations. In acidic media, CO oxidation on gold is shifted by ca. 0.5 V toward higher potentials.4 Besides, CO does not remain chemisorbed on the gold surface in the absence of CO in solution.9-11 Consequently, the higher catalytic activity in alkaline media seems (6) Rodriguez, P.; Feliu, J. M.; Koper, M. T. M. Electrochem. Commun. 2009, 11, 1105–1108. (7) Rodriguez, P.; Koverga, A. A.; Koper, M. T. M. Angew. Chem., Int. Ed. 2010, 122, 1263–1265. (8) Rodriguez, P.; Garcia-Araez, N.; Koper, M. T. M. Phys. Chem. Chem. Phys. 2010, DOI: 10.1039/b926365a.
Published on Web 06/18/2010
DOI: 10.1021/la1014048
12425
Article
Rodriguez et al.
to be related to the apparently stronger CO chemisorption. In the present work, this hypothesis will be tested as follows: first, the properties of the irreversibly adsorbed CO on gold electrodes will be characterized in detail, by the combination of voltammetric and spectroscopic measurements and DFT calculations. It will be shown that the extent of CO irreversible adsorption critically depends on the value of the CO adsorption potential. In addition, the effect of this variable on the kinetics of solution CO electrooxidation will be evaluated. In this way, it will be shown that the relationship between CO chemisorption and CO catalysis is more complex than previously considered.
2. Experimental and Computational Details The experiments were performed with bead-type gold singlecrystal electrodes purchased from icryst. Prior to each experiment, the electrodes were annealed in a propane-air flame to red heat and quenched with ultrapure water. A two-compartment electrochemical cell was employed, with a gold wire as the counter electrode and a reversible hydrogen electrode (RHE) as the reference electrode. All potentials in this work are referred to the RHE scale, except when otherwise stated. Electrochemical measurements were performed with an Autolab PGSTAT12. Solutions were prepared from NaOH (99.998%, Sigma-Aldrich) and ultrapure water (Milli-Q gradient A10 system, 18.2 MΩ cm). Ar (N66) was used to deoxygenate all solutions and CO (N47, stored in an aluminum cylinder and connected through aluminum valves, in order to prevent iron carbonyl contamination) was used to dose CO. The normalization of currents to current densities has been performed by using the geometric area of the electrodes. The FTIRRAS experiments were carried out with a Bruker Vertex80 V IR spectrophotometer. A CaF2 prism bevelled at 60° was used. Thirty interferograms were averaged for each spectrum, with a resolution of 6 cm-1, resulting in an acquisition time of 10 s per spectrum. All the spectra were recorded using p-polarized light and with an angle of incidence of 60°. First, a reference spectrum was acquired with the CO-modified gold electrode at the potential of CO adsorption. Then, several sample spectra at different potentials were collected, and the difference with respect to the reference spectrum was evaluated as ΔR/R. In these difference spectra, negative bands (pointing down) correspond to the generation of species at the gold-electrolyte interface and positive bands (pointing up) to the consumption of species. It should be mentioned that the voltammetric experiments on bulk CO electrooxidation on gold electrodes with the hangingmeniscus configuration were performed with particular caution. The presence of CO in the atmosphere of the electrochemical cell may produce the appearance of an additional oxidation current, whose origin is likely related to the fast diffusion of gas-phase CO to the meniscus solution layer. In order to avoid these complications, previous works embedded the gold electrodes in Teflon.12,13 Unfortunately, this procedure may damage the surface structure of the single-crystal electrodes. For this reason, in the present work, the effect of gas-phase CO has been eliminated by performing the experiments with a blanket of argon. In addition, the flow and duration of the argon flow within the cell atmosphere were minimized, in order to avoid the decrease of the dissolved CO concentration in solution by argon purging of the cell atmosphere. (9) Shue, C.-H.; Yang, L.-Y. O.; Yau, S.-L.; Itaya, K. Langmuir 2005, 21, 1942– 1948. (10) Sun, S.-G.; Cai, W.-B.; Wan, L.-J.; Osawa, M. J. Phys. Chem. B 1999, 103, 2460–2466. (11) Pronkin, S.; Hara, M.; Wandlowski, T. Russ. J. Electrochem. 2006, 42, 1177–1192. (12) Edens, G. J.; Hamelin, A.; Weaver, M. J. J. Phys. Chem. 1996, 100, 2322– 2329. (13) Blizanac, B. B.; Arenz, M.; Ross, P. N.; Markovic, N. M. J. Am. Chem. Soc. 2004, 126, 10130–10141.
12426 DOI: 10.1021/la1014048
DFT calculations have been performed using the Vienna ab initio simulation package (VASP).14,15 The Au(111) surface was represented as a six-layer slab in a p(2 2) supercell, with a calculated bulk FCC lattice constant of 4.17 A˚, and with 14.61 A˚ of vacuum. Adsorbates under consideration were placed on one side of the metal slab with each having a coverage θ = 1/4 ML. Geometry optimizations were performed using a cutoff for the plane wave basis set of 400 eV. Brillouin-zone sampling has been performed on 6 6 1 k-mesh using a Methfessel-Paxton integration scheme. The projector-augmented-wave (PAW) core potentials and the Perdew-Burke-Ernzerhof (PBE) full electron frozen core approximation to the exchange-correlation energy was used. This set of parameters assures an energy convergence of 0.001 meV/atom. In the structural optimization of adsorbate molecules, the adsorbate, as well as the two top metal layers, was relaxed simultaneously until forces were converged below 0.001 eV/A˚. In all cases, the vibrational frequencies were calculated after the structure optimization. Vibrational frequencies were obtained by evaluating the Hessian matrix as built in the VASP program.
3. Results and Discussion 3.1. CO Irreversible Adsorption on Gold. CO adsorbs irreversibly on gold electrodes in alkaline solutions.6 Figure 1 shows the voltammograms of Au(111), Au(100), and Au(110) in 0.1 M NaOH, as obtained after dosing CO into the electrochemical cell for 2 min, followed by purging with argon for 2030 min, with the electrode under potential control (Eads = 0.1 V) during the whole CO irreversible adsorption process. The blank voltammograms, as obtained before the CO irreversible adsorption, are included in the figure for the sake of comparison. It is observed that the presence of irreversibly adsorbed CO produces the following effects: (a) CO adsorption results in an increased capacity within the low potential region, since higher current densities are recorded at E < 0.5 V for the three gold basal planes, in comparison to the blank voltammograms, (b) the voltammograms of the CO-modified Au(111) and Au(100) electrodes exhibit an additional near-reversible pair of peaks at around E = 0.4 V, which are absent on the bare surfaces, and (c) an oxidation wave at around 1 V is observed for all the CO-modified surfaces, associated with the oxidative stripping of the irreversibly adsorbed CO. 3.2. Effect of Adsorption Potential on the CO Irreversible Adsorption on Gold. The fact that CO remains irreversibly adsorbed on gold in alkaline media introduces the possibility of characterizing the properties of adsorbed CO without the interference of the response of dissolved CO in solution. Figure 2 shows the effect of the variation of the value of the applied potential during the CO irreversible adsorption on Au(111). It is clear that the application of lower adsorption potentials produces (a) an increase of the current density at E < 0.5 V, (b) an increase of the charge under the peak at 0.4 V, and (c) an increase of the CO stripping charge. This indicates that the amount of irreversibly adsorbed CO increases as the CO adsorption is performed at lower potentials. The effect of adsorbing CO at more negative potentials has been studied quantitatively by the evaluation of the charge density of the process at E < 0.5 V and the CO stripping charge density. These charge density values have been evaluated by integration of the voltammograms at 0 < E < 0.55 V and 0.8 < E < 1.07 V, respectively, after subtraction of the blank voltammogram. The results are shown in Figure 3, plotted as a function of the CO (14) Kresse, G.; Hafner, J. Phys. Rev. B 1993, 47, 558–561. (15) Kresse, G.; Furthm€uller, J. Phys. Rev. B 1996, 54, 11169–11186.
Langmuir 2010, 26(14), 12425–12432
Rodriguez et al.
Article
Figure 3. Comparison of the CO stripping charge with the charge of the process at 0 < E < 0.55 V on CO-modified Au(111), plotted as a function of the CO adsorption potential, Eads. Error bars indicate the dispersion between several measurements.
Figure 2. Cyclic voltammograms of Au(111) modified by CO irreversible adsorption at different adsorption potentials, Eads, as indicated, in 0.1 M NaOH. The blank voltammogram is also included for comparison purposes. Scan rate: 50 mV/s.
This demonstrates that the extent of CO irreversible adsorption increases with lower CO adsorption potentials. Besides, it is also observed that both charge densities attain the same values, within the rather high experimental dispersion of these experiments. In view of Figure 3, it is tempting to assign the process at E < 0.5 V to the early adsorption of OH on the CO-modified gold electrode, as we have proposed in previous works.6-8 In this regard, DFT calculations suggest that one adsorbed CO molecule may stabilize the coadsorption of one OH species.7 Therefore, if the process at E < 0.5 V is associated with the early adsorption of OH species, promoted by irreversibly adsorbed CO, then this process will involve the exchange of one electron per CO adsorbed species. On the other hand, the oxidation of CO to CO2 would involve the exchange of a second electron. In conclusion, this explanation is in agreement with the fact that the charge of the process at E < 0.5 V is the same as the CO stripping charge, within the experimental error. 3.3. Effect of pH on the CO Irreversible Adsorption on Gold. CO irreversible adsorption is markedly sensitive to the pH of the solution.6 The reversible pair of peaks at around 0.4 V for CO-modified Au(100) and Au(111) electrodes shift toward lower potential values as the NaOH concentration decreases (see Supporting Information, Figure SI1). In addition, the charge under these peaks increases with the NaOH concentration, evidencing an increase on the amount of irreversibly adsorbed CO, in such a way that the CO irreversible adsorption is negligible in 0.02 M NaOH solutions. The effect of pH on the extent of CO irreversible adsorption can be understood with the help of the results in section 3.2 on the effect of the adsorption potential. This is related to the fact that, as the NaOH concentration increases, the value of the applied potential during CO adsorption (Eads = 0.1 V vs RHE) becomes more negative when referred to the SHE scale. That is, the adsorption potential Eads = 0.1 V vs RHE corresponds to Eads = -0.70, -0.66, and -0.62 V vs SHE for NaOH solutions with concentration 0.5, 0.1, and 0.02 M, respectively.16 Consequently, it seems more reasonable to compare the cyclic voltammograms of gold surfaces modified by CO adsorption at the same adsorption potential on the SHE scale. This comparison is shown in Figure 4, for two different values of the adsorption potential on the SHE scale: Eads =-0.70 V (a,b) and Eads = -0.62 V (c,d), for
adsorption potential. Error bars indicate the dispersion between different measurements. Noteworthy, it is observed that both charge densities clearly increase as the applied potential decreases.
(16) The conversion of potential values to the SHE scale has been performed by using values of the activity coefficients calculated with the program Aq-solutions, supported by the Analytical Division of IUPAC, and developed by I. V. Sukhno, V. Y. Buzko, and L. D. Pettit.
Figure 1. Cyclic voltammograms of Au(111), Au(100), and Au(110) in 0.1 M NaOH, before (a) and after (b,c) CO irreversible adsorption at Eads = 0.1 V. Scan rate: 50 mV/s.
Langmuir 2010, 26(14), 12425–12432
DOI: 10.1021/la1014048
12427
Article
Figure 4. Comparison of the cyclic voltammograms of Au(111) modified by CO irreversible adsorption at Eads = -0.70 V vs SHE (a,b) and Eads = -0.62 V vs SHE (c,d), in 0.5 M NaOH (a), 0.1 M NaOH (b,c) and 0.02 M NaOH (d). The corresponding CO adsorption potentials on the RHE scale are 0.1 V (a,d), 0.06 V (b), and 0.14 V (c). Scan rate: 50 mV/s.
NaOH solutions with concentrations 0.5 M (a), 0.1 M (b,c), and 0.02 M (d). Comparison of curves a,b with curves c,d in Figure 4 shows that the value of the adsorption potential on the SHE scale has a major influence on the extent of CO irreversible adsorption. The CO stripping charge and the charge at E < 0.5 V are clearly higher on curves a,b than in curves c,d. This behavior is the same as that observed by varying the value of the adsorption potential without changing the NaOH concentration in Figure 2. Most importantly, it should be noticed from Figure 4 that, although curves b,c were both obtained in a 0.1 M NaOH solution, they resemble curves a and d, respectively, obtained with different NaOH composition but at the same adsorption potential on the SHE scale. It should also be noted that certain differences are still observed between curves a and b, and c and d. Specifically, higher amounts of irreversible adsorbed CO are attained in more concentrated NaOH solutions for the same values of the adsorption potential on the SHE scale. This indicates that the OH- anions exhibit specific interactions with the CO-modified gold electrodes, in such a way that the adsorption energy of CO species appears to be strengthened. 3.4. FTIRRAS Measurements. The FTIRRAS (Fourier transform infrared reflection absorption spectroscopy) measurements of CO irreversibly adsorbed on gold were performed as follows. After adsorbing CO at a given potential, a reference spectrum was taken at the potential of CO adsorption. Then, several sample spectra were obtained as the potential was gradually increased up to 1.4 V. Potential difference spectra were calculated as ΔR/R. These potential difference spectra are useful to characterize the species formed during the CO stripping. On the other hand, the consumption of irreversibly adsorbed CO species is more clearly characterized by difference spectra referred to the 12428 DOI: 10.1021/la1014048
Rodriguez et al.
Figure 5. Potential difference FTIRRAS spectra of Au(111) modified by CO adsorption at Eads = -0.1 V in 0.1 M NaOH. The spectra on the left are referred to the reference spectrum acquired at Eref = -0.1 V prior to the potential scan up to 1.4 V. The spectra on the right are referred to the spectra acquired at Eref = 1.4 V, at the end of the potential scan.
spectrum acquired at the end of the potential scan, i.e., when all CO has been stripped off the surface. Figure 5 compares both types of difference spectra for a Au(111) electrode modified by CO irreversible adsorption at Eads = -0.1 V. The spectra on the left in Figure 5 (referred to Eref = -0.1 V) show the appearance of a negative band at 1400 cm-1 as the potential is increased above 0.7 V. This band can be ascribed to the asymmetric stretching of carbonate, which is formed by the oxidative desorption of adsorbed CO in alkaline media. Clearly, the amount of carbonate formation increases with potential, up to 1.4 V, where it reaches saturation, meaning that all CO has been stripped off the surface at this potential. The spectra on the left in Figure 5 also show two bipolar bands at around 2000 and 1820 cm-1. These bands become positive at 1.4 V, meaning that they correspond to species that are present at the reference potential (-0.1 V) and that have disappeared at the sample potential (1.4 V). The interpretation of these bands becomes clear when the spectra are referenced to a CO-free surface (spectra on the right in Figure 5). It is seen that CO irreversibly adsorbed on Au(111) leads to two negative adsorption bands, whose frequencies shift with potential. The shift with potential produces the bipolarity of the bands when the reference spectrum is at -0.1 V. The extent of CO adsorption is obviously correlated with the intensity of the positive bands of the spectrum at 1.4 V referenced to -0.1 V. Additional FTIRRAS measurements were performed with Au(111) electrodes modified by CO adsorption at different potentials. Figures 6 and 7 summarize the main results, with additional data to be found in the Supporting Information. Figures 6 and 7 show the spectra acquired at 1.4 and 0.4 V, respectively, for Au(111) modified by CO adsorption at different potentials, between 0.2 and -0.1 V. In both cases, the reference spectra were taken at the potential of CO adsorption. In view of Figure 6, it is clear that the intensity of the carbonate band increases as the potential of CO adsorption is made more negative, evidencing Langmuir 2010, 26(14), 12425–12432
Rodriguez et al.
Article
Figure 8. Experimental potential-dependent CO stretching frequencies for CO adsorbed at different potentials, Eads, as indicated, in 0.1 M NaOH.
Figure 6. Potential difference FTIRRAS spectra of Au(111) modified by CO adsorption at different potentials, Eads, as indicated, in 0.1 M NaOH. The reference spectra were taken at the potential of CO adsorption, and then the sample spectra were measured at 1.4 V.
Figure 7. As in Figure 6, but the sample spectra were measured at 0.4 V.
that the coverage of CO irreversibly adsorbed increases correspondingly. This is in agreement with the voltammetric results in section 3.2. In addition, it is seen that the intensity of the bands associated with the consumption of adsorbed CO, between 2000 and 1800 cm-1, also becomes more intense as the potential for CO adsorption is decreased, corroborating that the extent of irreversibly adsorbed CO increases in this way. The spectra in Figure 7 show the effect of the variation of the potential between 0.4 V and the potential of CO adsorption. These spectra allow one to distinguish small CO bands, that are not observable in Figure 6 because water bands in alkaline solutions result in a highly nonflat background spectrum.17 Three (17) Chen, A. C.; Lipkowski, J. J. Phys. Chem. B 1999, 103, 682–691.
Langmuir 2010, 26(14), 12425–12432
different CO bands can be identified in Figure 7, at around 2000, 1900, and 1820 cm-1. The frequencies of these bands are plotted as a function of potential in Figure 8.18 It should be noted that no increase of the intensity of one of these bands, at the expense of a decrease on the intensity of another, is observed as the applied potential is scanned. In the next section, the assignment of these three CO bands will be discussed with the help of DFT calculations incorporating an electric field. 3.5. DFT Calculations. DFT calculations of CO adsorbed on a Au(111) slab were performed in order to assign the CO bands measured by FTIRRAS and to study the influence of an electric field on the adsorption energy and vibration properties of CO adsorbed on different sites. The binding energies and C-O stretching frequencies of CO adsorbed on top, bridge, and hollow sites of a Au(111) slab were calculated by using the Vienna ab initio simulation package (VASP). The effect of the applied potential was evaluated through the introduction of an electric field. Figure 9 shows the calculated binding energies and CO stretching frequencies as a function of the applied electric field. It is clear that the CO stretching frequency decreases in the order top > bridge > hollow and that the CO stretching frequency decreases as the applied electric field becomes more negative, with slopes of 93 ( 7, 98 ( 8, and 66 ( 1 cm-1 (V A˚-1)-1 for CO adsorbed on top, bridge, and hollow sites, respectively. These slopes correspond to a Stark tuning slope of around 32-20 cm-1 V-1, if a double-layer thickness of 3 A˚ is assumed. It is interesting to compare the calculated CO stretching frequencies in Figure 9 with the experimental FTIRRAS results in Figure 8. From this comparison, the CO bands at around 2000, 1900, and 1820 cm-1 can be assigned to CO adsorbed on top, bridge, and hollow sites, respectively. Thus, it is concluded that irreversibly adsorbed CO occupies different sites, depending on the potential of CO adsorption. In our experiments, top sites are always occupied, and bridge and hollow sites become occupied when CO adsorption is performed at Eads g 0 V and Eads e 0 V, respectively. However, the variation of the applied potential during the FTIRRAS measurements does not produce a reorganization of the CO occupation, which might be because these measurements last less than 5 min. On the other hand, the experimental potential-induced shift of the CO stretching frequency (18) When possible, the frequency of the CO bands was obtained from the potential difference spectra referred to the CO-free spectra at 1.4 V. When the intensity of the CO bands was too small, their frequency was estimated from the bipolar bands observed in the potential difference spectra referrred to the spectra at the potential of CO adsorption.
DOI: 10.1021/la1014048
12429
Article
Rodriguez et al.
Figure 10. Bulk CO oxidation on Au(111) in CO-containing 0.1 M NaOH solutions, measured at different values of the immersion potential: -0.10 (a), 0.00 (b), 0.05 (c), 0.10 (d), and 0.18 V (e). Scan rate: 50 mV/s.
Figure 9. Calculated field-dependent binding energies (A) and C-O stretching frequencies (B) for CO adsorbed on top, bridge, and hollow sites of a Au(111) slab.
of CO adsorbed on top, bridge, and hollow sites is around 4020 cm-1 V-1 in all cases, in reasonable agreement with the DFT results.19 It should be noted, though, that our DFT calculations do not provide a consistent explanation of all the experimental data on CO adsorption on gold in alkaline media. In particular, the strongest binding energy we obtain is ca. -0.35 eV, which is not strong enough to lead to irreversibly adsorbed CO at room temperature. Although the DFT calculations indeed predict that CO binding should become stronger with a more negative field/ potential, the calculated binding energies for CO on top, bridge, and hollow sites do not explain the observed preference for the adsorption on top nor the change in preference for bridge and hollow sites as a function of potential. These inconsistencies are most likely related to the fact that CO irreversible adsorption in alkaline media is enhanced by OH coadsorption, as stated above. On the other hand, at negative potentials, one would expect the coadsorption of OH-/OH to be less important. Currently, more accurate DFT calculations incorporating OH coadsorbates are being performed in our laboratory. 3.6. Bulk CO Oxidation. The voltammetric and the spectroscopic results presented in sections 3.2 and 3.4 have demonstrated that the extent of CO irreversible adsorption can be controlled by selecting an appropriate adsorption potential (for a given adsorption time and solution composition). In the following, we will evaluate the effect of this variable on the oxidation of CO in solution. For this purpose, cyclic voltammograms of Au(111) in CO-containing 0.1 M NaOH solutions have been recorded as follows. The initial contact of the electrode with the electrolyte solution was performed at different selected potentials, between 0 and 0.18 V. The CO adsorption time was 20 s in all cases, which is the time required to correctly position the electrode in the meniscus configuration. Figure 10 shows the experimental results. (19) The Stark tuning slopes obtained here for CO on Au(111) are smaller than those reported in ref 20 on polycrystalline gold (64 cm-1 V-1). This is likely due to a surface sensitive effect, since the Stark tuning slopes for CO on Au(110) (around 50 cm-1 V-1, figure SI6) are intermediate between Au(111) and polycrystalline gold.
12430 DOI: 10.1021/la1014048
It is observed that decreasing the value of the immersion potential produces a dramatic shift of the onset of the bulk CO oxidation toward higher potentials. Some variations on the magnitude of the current density at higher potentials (within the diffusion-controlled regime) are also observed, but these changes are related to the different CO concentration in solution. From an experimental point of view, it is difficult to control the CO concentration in solution, since these experiments are performed with a blanket of argon above the solution in order to remove gasphase CO from the cell atmosphere. As a result, the CO concentration in solution will slowly decrease over time during a long experimental session. The effect of CO concentration on the kinetics of bulk CO electrooxidation on gold has been characterized in ref 8 in detail. From these results, it is clear that the variations in CO concentration in Figure 10 cannot explain the marked shift on the onset of CO electrooxidation. Therefore, it is concluded that, when CO adsorption is performed at more negative potentials, the kinetics for CO electrooxidation becomes more inhibited. The adsorption of CO at more negative potentials is accompanied by an increase of the extent of CO irreversible adsorption. In addition, the spectroscopic data in section 3.4 show that the geometry of CO adsorption also changes, and hollow sites become occupied at the expense of bridge sites as the adsorption potential is decreased. Moreover, as shown in section 3.2, the charge under the reversible pair of peaks at around 0.4 V also increases as the CO irreversible adsorption is performed at more negative potentials. With the presently available data, it is difficult to unambiguously identify the origin of the inhibited catalysis of gold surfaces modified by CO adsorption at more negative potentials. However, it is more tempting to relate this to the occupation of hollow sites. Experimental support for this hypothesis is obtained with experiments with the Au(110) surface, to be discussed next. 3.7. CO on Au(110). Figure 11 shows the stripping voltammograms of irreversibly adsorbed CO on Au(110) at different CO adsorption potentials. It is observed that the charge under the CO-stripping wave at 1 V increases as the adsorption potential decreases, evidencing that the extent of CO irreversible adsorption increases, as was previously observed for Au(111). Figure 12 shows the bulk CO oxidation on Au(110) in COcontaining 0.1 M NaOH solutions for different values of the immersion potential. Remarkably, it is observed that the onset potential for bulk CO oxidation is essentially independent of the value of the adsorption potential. Thus, the increase in the CO Langmuir 2010, 26(14), 12425–12432
Rodriguez et al.
Article
Figure 13. Peak potential values of the pair of peaks at around Figure 11. Cyclic voltammograms of Au(110) modified by CO irreversible adsorption at different adsorption potentials, Eads, as indicated, in 0.1 M NaOH. The blank voltammogram is also included for comparison purposes. Scan rate: 50 mV/s.
0.4 V for a CO-modified Au(100) electrode, plotted as a function of the logarithm of the scan rate. Lines are the fits of the experimental data.
against the logarithm of the scan rate. The fits within the high scan rate region give values of the slope of ca. 59 mV/de, corresponding to R ≈ 1.21 This R value corresponds to a mechanism in which the rate-determining reaction is a chemical reaction preceded by a fast electrochemical step. This suggests that the process under study is associated with a relatively slow (structural) process that is in equilibrium with OH- adsorption. We suggest this process to be “structural” on account of its strong dependence on the gold surface structure (the peak is only observed on hexagonal surfaces) and on account of the sharpness of the peak, which is aften associated with structural changes.
Conclusions
Figure 12. Bulk CO oxidation on Au(110) in CO-containing 0.1 M NaOH solutions, measured at different values of the immersion potential: -0.10 (a), 0.00 (b), and 0.10 (c). Scan rate: 50 mV/s.
coverage on Au(110) at lower immersion potentials does not produce any significant effect on the catalysis of bulk CO oxidation. Previous work6 has shown that the irreversible adsorption of CO on Au(110) gives FTIRRAS spectra with a single CO band at ca. 2000 cm-1, ascribed to CO adsorbed on top sites (see also Supporting Information). The adsorption at more negative potentials does not produce the appearance of additional CO bands that would indicate the adsorption of CO on multifold sites. In conclusion, it appears that the absence of inhibition of the kinetics of bulk CO oxidation by CO irreversible adsorption on Au(110) (Figure 12) may be due to the fact that, on this surface, CO only adsorbs on top sites. 3.8. Apparent Transfer Coefficient. The apparent transfer coefficient, R, describes the effect of potential on the kinetics of reactions, and it is widely used to obtain insight into the mechanism of the reaction. Here, we evaluate R for the voltammetric pair of peaks at around 0.4 V associated with CO irreversible adsorption on Au(111) and Au(100). This is done by measuring the peak potential values as a function of the scan rate (with scan rates between 0.05 and 11 V/s). These experiments were performed with Au(100), instead of Au(111), since the former shows a somewhat higher irreversibility. Figure 13 plots the peak potential values (20) Kunimatsu, K.; Aramata, A.; Nakajima, H.; Kita, H. J. Electroanal. Chem. 1986, 207, 293–307.
Langmuir 2010, 26(14), 12425–12432
This paper was focused on providing a more complete understanding of gold catalysis for CO electrooxidation in alkaline media, through a combined electrochemical, spectroscopic, and DFT study. First, CO irreversible adsorption on gold single crystals was characterized in detail by voltammetric measurements. It was concluded that the amount of CO irreversibly adsorbed markedly increases as the value of CO adsorption potential decreases. This explains the fact that higher CO coverages are achieved in CO irreversible adsorption experiments performed in more alkaline solutions, because the value of the CO adsorption potential vs RHE becomes more negative (vs SHE) with increasing pH. However, specific interactions between OH- and the gold surface also appear to contribute to stabilizing CO adsorption. Next, CO irreversible adsorption was studied by FTIRRAS measurements, corroborating that the extent of CO irreversibly adsorbed critically depends on the value of the adsorption potential. Furthermore, comparison with DFT calculations showed that CO adsorption on Au(111) takes place on top and bridge sites for CO adsoption potentials higher than 0 V vs RHE, while top and hollow sites are occupied when the CO irreversible adsorption is made at lower potentials. However, the simple DFT model of CO on Au(111) with an applied electric field provides values of binding energies that are not consistent with the experimental fact that CO irreversibly adsorbs on gold in alkaline media. This is most likely related to the fact that OH coadsorbates are not included in the DFT calculations. The kinetics of bulk CO electrooxidation were studied by voltammetric experiments in CO-containing alkaline solutions. In order to study the effect of the CO adsorption potential, the (21) Laviron, E. J. Electroanal. Chem. 1979, 101, 19–28.
DOI: 10.1021/la1014048
12431
Article
contact of the electrode with the CO-containing electrolyte solution was performed at different potentials. It was found that the kinetics of bulk CO electrooxidation on Au(111) become more inhibited when the immersion potential was made more negative. This behavior is tentatively ascribed to the occupation of hollow sites. Conversely, the variation of the value of the immersion potential does not produce any effect on the kinetics for bulk CO electrooxidation on Au(110). This behavior may be related to the fact CO on Au(110) only occupies top sites, regardless of the adsorption potential. Finally, this work presented two experimental observations that indicate that irreversible adsorption of CO on gold promotes the coadsorption of OH: (i) For Au(111) modified by CO adsorption at different potentials, it was found that the charge associated to the unusual voltammetric feature at E < 0.5 V is the same as the CO stripping charge, within the experimental uncertainty. This suggests that one electron per CO adsorbed species is involved in the process at E < 0.5 V, associated to OH adsorption, while the CO stripping oxidation to CO2 at E ≈ 1 V involves the exchange of a second electron. (ii) From the effect of the scan rate, it was concluded that the pair of peaks at around
12432 DOI: 10.1021/la1014048
Rodriguez et al.
0.4 V have an effective transfer coefficient of R ≈ 1. This result can be explained by considering that the process under study is associated to a relatively slow (structural) process in equilibrium with OH coadsorption. Acknowledgment. The authors acknowledge financial support from The Netherlands Organization for Scientific Research (NWO) and the European Commission (through FP7 Initial Training Network “ELCAT”, Grant Agreement No. 214936-2). NG acknowledges the European Commission (FP7) for the award of a Marie Curie fellowship. The calculations were partly performed using supercomputer facilities provided by the National Computing Facilities (NCF) of The Netherlands Organization for Science Research (NWO). Supporting Information Available: Cyclic voltammograms of Au(111) and Au(100) modified by CO irreversible adsorption at different pH. Potential difference FTIRRAS spectra of Au(111) and Au(110) modified by CO adsorption. This material is available free of charge via the Internet at http:// pubs.acs.org.
Langmuir 2010, 26(14), 12425–12432
