Electrochemical Reduction of Oxygen on Gold Surfaces: A Density

Paramaconi Rodriguez , Marc T. M. Koper. Physical Chemistry ... Simantini Nayak , P. Ulrich Biedermann , Martin Stratmann , Andreas Erbe. Electrochimi...
0 downloads 0 Views 285KB Size
J. Phys. Chem. C 2007, 111, 2607-2613

2607

Electrochemical Reduction of Oxygen on Gold Surfaces: A Density Functional Theory Study of Intermediates and Reaction Paths Peter Vassilev Schuit Institute of Catalysis, Laboratory of Inorganic Chemistry and Catalysis, EindhoVen UniVersity of Technology, P.O. Box 513, 5600 MB EindhoVen, The Netherlands

Marc T. M. Koper* Leiden Institute of Chemistry, Leiden UniVersity, P.O. Box 9502, 2300 RA Leiden, The Netherlands ReceiVed: July 17, 2006; In Final Form: October 25, 2006

Electron density functional theory simulations are used to model the electrochemical reduction of oxygen on gold surfaces. Adsorption energies and geometry specifications of various intermediates are reported for gas phase adsorption. The elementary reactions of the electrochemical reduction of oxygen are discussed, including the effects of the electrode potential. A general reaction pathway diagram is presented. It is found that the relative stability of the intermediates, although able to provide insight into the elementary steps, cannot fully describe the observed experimental data of structural selectivity of Au(100) and Au(111) electrodes in alkaline solution. To comprehend the complexity of the system, it is suggested that one has to consider in detail the activation energies of the elementary steps including the role of water. By ab initio molecular dynamics simulations including water, it is also shown that water plays an important role in the process of reduction of oxygen via proton exchange, resulting in an intermediate hydroperoxyl OOH and hydroxyl OH on the surface, making OOH one of the most important intermediates in the oxygen reduction reaction.

Introduction The electrochemical reduction of oxygen is one of the most important reactions in fuel cell processes and electrochemistry in general and has received considerable attention in the last several years.1-13 It has been established that depending on the cathode material and the solution composition the reaction can occur as two- or four-electron processes with hydrogen peroxide or water, respectively, as final products:

O2 + 2 H+ (aq) + 2 e- f H2O2

(1)

O2 + 4 H+ (aq) + 4 e- f 2 H2O

(2)

In the case of a four-electron reduction (2), the reaction can occur via the so-called “direct” pathway without detection of the intermediate hydrogen peroxide, or via the “series” pathway which occurs via (1) and followed by (3) -

H2O2 + 2 H+ (aq) + 2 e f 2 H2O

(3)

or combinations of the two pathways called “parallel” or “interactive” pathways.4 Although ample experimental data has been collected, the elementary steps of the electrochemical oxygen reduction, which would ultimately determine the intermediates and the final products (H2O2 and/or H2O), are still under debate. It is believed, however, that the stability of the oxygen molecule O2 adsorbed on the surface and especially the dissociation barrier for the O-O bond are the most important factors determining the outcome of the reduction process.1-3,5,8 * To whom correspondence should be addressed.E-mail: m.koper@ lic.leidenuniv.nl.

One of the most active catalysts for the electrochemical reduction of oxygen is Au(100) in alkaline solution, which exhibits an even higher activity than polycrystalline Pt. It is remarkable however that gold single-crystal electrodes show exceptional structural selectivity toward water or hydrogen peroxide. On Au(111) electrodes, the reduction of oxygen proceeds as a two-electron process to hydrogen peroxide, whereas on Au(100), the product will depend on the pH of the solution. In acidic media, the reduction proceeds as a twoelectron process, whereas in alkaline solution, it is a fourelectron process. Speculations have been raised that the determining factors for this surface and pH-dependent selectivity may be the orientation of the adsorbed O2 molecule, the relative stability of the intermediates, including OH, or an even more complicated process of further reduction of H2O2 catalyzed by coadsorbed OH. The nature of this process however is still not fully understood. Au(111) and Au(100) surfaces would present an excellent choice for computational modeling due to their structural simplicity. In this work, we study the electrochemical reduction of oxygen on gold by means of electron density functional theory (DFT) simulations, aiming at understanding the elementary steps in this process. We start by using simplified models for the interface in which the solvent (water) is not present; the possible effects of the hydration and the role of water in the reactions are further investigated through molecular dynamics simulations of a solvated oxygen molecule adsorbed on the surface. Considering the electrochemical steps, we use the approach introduced by Nørskov et al.,14 which relies on the assertion that for a normal hydrogen electrode (NHE) the chemical potential of (H+ + e-), with H+ being in solution and the electron being on the electrode, is equal to the chemical

10.1021/jp064515+ CCC: $37.00 © 2007 American Chemical Society Published on Web 01/23/2007

2608 J. Phys. Chem. C, Vol. 111, No. 6, 2007

Vassilev and Koper

potential of 0.5 H2 at 1 bar and 298 K. Essentially this method allows for an approximative evaluation of the reaction energies of electrochemical reactions using computed data obtained for gas phase adsorption. At the same time, it also introduces electrode potential dependence which substantiates in a shift of the energy levels depending on whether production or consumption of electrons takes place at the cathode during the course of the reaction.14 Computational Details The simulations were performed using the vasp15,16 electron density functional theory simulation package. The Kohn-Sham electron wave functions are expanded in a plane wave basis set. The ions are described using the PAW approach.17,18 We used the Perdew-Wang 91 (PW91) density functional.19 The computational models consisted of 7 metal layers Au(100) and Au(111) slabs constructed using 4.17 Å fcc lattice parameters as computed with vasp for bulk Au. The size of the unit cell in the direction perpendicular to the surface was 25 Å. Adsorbed species at 0.25 coverage were considered on both sides of the slab in a centrosymmetric configuration with the center of symmetry in the middle of the metal layers. The cutoff for the plane wave basis set was set to 325 eV. For the integration of the band energies, a 5 × 5 × 1 Monkhorst-Packtype mesh20 was used. The gas-phase O2 molecule and O2 adsorbed on the surface were simulated spin- unrestricted in order to account for the unpaired electrons on the molecule. All other systems were computed spin-restricted as essentially they are closed-shell systems due to the formation of bonds between the intermediates and the metal surface, leading to partial electron charge transfer to the adsorbates. Various adsorption sites were considered including top, bridge, and hollow sites with different orientation of the adsorbates (flat, end-on, and tilted). For each of the adsorption sites, only the results for the most stable orientation are presented. Harmonic frequencies were evaluated excluding coupling with the metal layer. Supporting molecular dynamics (MD) simulations of O2 coadsorbed with water were performed using a 3-layer Au(100) slab and a 4 × 4 unit cell at 300 K with a time step of 0.5 fs. The energies presented in the current paper correspond in particular to an electrode potential of U ) 0 V vs NHE. The potential dependence however can easily be evaluated from the numerical data as discussed in ref 14. All reported data for the hollow sites refer to the fcc-type hollow site on the Au(111) surface, which is the more stable one. In eqs 4-14, ∆H is computed using the most stable adsorption sites of both reactants and products. The enthalpies of formation of the intermediates given in Figure 3-5 are according to the formation reactions given in the text. Results and Discussions In the following sections, we first discuss the individual intermediates on the Au(100) and Au(111) surfaces as well as the various reaction steps which could lead to their formation. We present the computed reaction enthalpies ∆H and the enthalpies of formation (from O2 and H2) used to generate the data for ∆H. In eqs 4-14, Au denotes an empty adsorption site on the surface. In the last section, we summarize the data in a unified reaction scheme for both the two-electron and the four-electron electrochemical reduction of oxygen. Adsorption of Reaction Intermediates. Molecular Oxygen. Molecular oxygen on the (100) or (111) gold surfaces O2/Au

Figure 1. Hollow, bridge, and top configurations, enthalpies of adsorption, geometry specifications, and electron state of an oxygen molecule on a Au(100) surface.

could be formed upon adsorption of O2 molecules from the solution or from the gas phase in eq 4

{

O2 +Au f O2/Au, ∆H )

Au(100) : -7 kJ‚mol-1 Au(111) : +2 kJ‚mol-1

}

(4)

Although not made explicit in eq 4, upon adsorption there might be a (partial) electron charge transfer from the surface toward the molecule effectively resulting in the charging of the adsorbed O2, as will be discussed below. In the case of Au(100), we were able to find local minima for all of the three adsorption sites: hollow, bridge, and top (Figure 1). The most stable appears to be the hollow site followed by the bridge site, but the differences and the absolute values of the enthalpies are very small. In either of these adsorption configurations, the molecules lie flat on the surface. We were not able to find stable adsorption states, i.e., local minima on the potential energy surface, for configurations in the hollow or at the bridge sites with the molecule perpendicular or tilted with respect to the surface. On the top adsorption site on the (100) surface, the molecule is in the tilted configuration. This state is characterized by a longer Au-O distance from the surface of 2.59 Å, suggesting a rather weak chemical interaction compared to the hollow or bridge sites (Figure 1). Frequency analysis of the optimized structure of O2 in the hollow site of Au(100) confirmed that the configuration is a relatively well-defined minimum on the potential energy surface (PES) with the frustrated-translation vibrational modes in the plane of the surface on the order of 179 and 195 cm-1. The frustrated translation in the direction perpendicular to the surface, i.e., the direction of desorption of the molecule, is 233 cm-1. Additional calculations of configurations with the O2 at various distances while keeping the molecule parallel to the surface lead to an estimated desorption activation energy on the order of 40 kJ‚mol-1. One has to point out, however, that this may not be the most probable adsorption-desorption path for the oxygen molecule, as also will be discussed further below. Considering the bridge adsorption site, the frustrated translations along the surface have much lower frequencies on the order of 76 cm-1 along the bridge and an essentially flat potential energy surface in the direction perpendicular to the bridge. The frustrated translation perpendicular to the surface itself is also with a lower but still moderate frequency of 137 cm-1, and we note that it is still a well-defined local PES minimum. The

Electrochemical Reduction of Oxygen

J. Phys. Chem. C, Vol. 111, No. 6, 2007 2609

Figure 2. Bridge configuration, enthalpy of adsorption, geometry specification, and electron state of an oxygen molecule adsorbed on a Au(111) surface.

frequencies of the top site were not explicitly considered in this work, but one would expect that the molecule is in a very shallow PES minimum, as indicated by the elongated O-Au distance [compare also the case of O2 on Au(111) discussed below]. In the case of O2 on Au(111), we were able to find only one adsorption site, which is the bridge site (Figure 2). This adsorption configuration has a positive enthalpy of adsorption (i.e., energetically unstable). Frequency analysis of this state revealed that it is a very shallow minimum on the PES with frustrated-translation modes on the order of 43 and 41 cm-1 (in the plane of the metal surface) and 72 cm-1 (perpendicular to the surface). Simulations of top and hollow adsorption sites resulted in the desorption of the molecule (i.e., translation toward the vacuum region during the geometry optimization), indicating positive enthalpies of adsorption and no desorption barrier for these sites. The strength and the intramolecular stretch frequency of the O-O bond are also dependent on the surface and the adsorption site. On the bridge sites, the O-O stretch modes are 1153 and 1252 cm-1 respectively on Au(100) and Au(111) surfaces, to be compared to 1505 cm-1 computed for the isolated molecule using an equivalent computational setup. In the hollow site on Au(100), the stretch mode is with an even lower frequency of 825 cm-1, indicating a significantly weakened O-O bond. It is remarkable that in all of the three casesstop, bridge, and hollows the oxygen molecule is charged differently due to electron charge transfer from the metal surface (Figure 1). As estimated from the projected density of states and comparing the bond lengths and frequencies, on the top site, the oxygen molecule retains its original charge state, whereas on the bridge, it is in a super-oxo state (O2-). In the hollow site O2 is in a peroxo-like state (O22-). The electron charge transfer occurs from the conduction band of the metal to the 2π* molecular orbitals of O2, as also discussed for example by Panchenko et al.9 The change of the electronic state is also clearly reflected in the O-O bond distance, which increases in the same order of states from 1.26 Å, which is comparable to the gas phase isolated O2 molecule, to 1.43 Å (Figure 1). The electron charge transfer from the surface toward the molecule should not be associated with an electrochemical process but is rather a result of a formation of a chemical bond between the oxygen and the gold surface atoms. Different “charged” states of the molecule are observed as a direct result of the very close proximity (on eigenvalue energy scale) of the 2π* molecular orbitals (MOs) of O2 to the Fermi level (i.e., the highest occupied electron band of the metal). Depending on the adsorption site, and thus the local chemical and electrostatic environment, the 2π* MOs are above or (partially) below the Fermi level, resulting in a (partial) electron transfer. It is to be noted that the charged state on different adsorption

Figure 3. Configurations, enthalpies of formation, and geometry specifications of OOH species on Au(100).

Figure 4. Configurations, enthalpies of formation, and geometry specifications of OOH species on Au(111).

sites must only be considered as tentative and can be influenced by the presence of the polar solvating water molecules. Hydroperoxyl OOH Species. In the course of oxygen reduction, hydroperoxyl OOH intermediates on the metal surface could be formed in a couple of ways

O2/Au + H2O +Au f HOO/Au + HO/Au, ∆H )

{

Au(100) : -10 kJ‚mol-1 Au(111) : +54 kJ‚mol-1

O2/Au + H+ (aq) + e- f HOO/Au, ∆H )

{

Au(100) : -88 kJ‚mol-1 (at E ) 0 V vs NHE) Au(111) : -73 kJ‚mol-1 (at E ) 0 V vs NHE)

}

(5)

}

(6)

Reaction 5 is a purely chemical reaction (i.e., no charge transfer through the interface), whereas reaction 6 is an electrochemical step involving a hydronium ion H+(aq) from the solution and an electron e- from the cathode. The overall reaction of formation of HOO/Au from O2 and H2 is

O2 + 0.5H2 + Au f HOO/Au and the enthalpies for the different gold surfaces and adsorption sites are given in Figures 3 and 4. All three adsorption sitess top, bridge, and hollowson either of the surfaces are local minima. In all cases, as it can be expected, the bonding is realized primarily through the non-H-bonded oxygen atom of the OOH. For the bridge and hollow configurations, one could speculate that an additional bond is being formed through the

2610 J. Phys. Chem. C, Vol. 111, No. 6, 2007

Vassilev and Koper

second oxygen atom. In the case of Au(100), there is a sufficient difference in the adsorption energies between the adsorption sites, whereas in the case of Au(111), the three sites are practically indistinguishable. In all cases, the OOH species are charged (OOH-) as a result of a electron charge transfer from the metal surface, as estimated from projected electron density of states (but see also the discussion on charge transfer in the previous section). Atomic Oxygen. Atomic oxygen could be formed either after the dissociation of adsorbed oxygen molecules or after decomposition of hydroperoxyl HOO/Au

O2/Au + Au f 2 O/Au, ∆H )

{

}

Au(100) : +14 kJ‚mol-1 Au(111) : -40 kJ‚mol-1 (7)

HOO/Au + Au f HO/Au + O/Au, ∆H ) Au(100) : -67 kJ‚mol-1 Au(111) : -75 kJ‚mol-1

{

}

(8)

0.5 O2 + Au f O/Au and enthalpies are +4 and +5 kJ‚mol-1 for the hollow and bridge sites on the Au(100) surface, respectively, and -19 kJ‚mol-1 for the hollow site on the Au(111) surface. The bridge site on Au(111) is not a stable minimum on the potential energy surface and relaxes to the hollow site. The top sites are also not stable adsorption sites on neither of the surfaces. Hydroxyl OH Species. Hydroxyl OH on the surface could be formed via

{

Au(100) : -84 kJ‚mol-1 Au(111) : +16 kJ‚mol-1

HOO/Au + H2O f H2O2 + HO/Au, ∆H )

{

Au(100) : +16 kJ‚mol-1 Au(111) : +31 kJ‚mol-1

H2O2 + 2 Au f 2 HO/Au, ∆H )

{

Au(100) : -168 kJ‚mol-1 Au(111) : -89 kJ‚mol-1

O/Au + H+(aq) + e- f HO/Au, ∆H )

{

Au(100) : -170 kJ‚mol-1 (at E ) 0 V vs NHE) Au(111) : -107 kJ‚mol-1 (at E ) 0 V vs NHE)

}

(9)

}

(10)

}

(11)

}

can be considered as negatively charged (OH-) due to electron charge transfer from the metal surface. Hydrogen Peroxide. Formation of H2O2 could occur as a result of a (chemical) reaction between OOH and water (10) or after electrochemical reduction of OOH:

HOO/Au + H+(aq) + e- f H2O2 + Au, ∆H )

Additionally, atomic oxygen could be formed by the reverse of reaction 9. The overall reaction of formation of atomic oxygen on the surface from O2 is

O/Au + H2O + Au f 2 HO/Au, ∆H )

Figure 5. Configurations, enthalpies of formation, and geometry specifications of OH species adsorbed on a Au(100) surface.

{

Au(100) : -69 kJ‚mol-1 (at E ) 0 V vs NHE) Au(100) : -93 kJ‚mol-1 (at E ) 0 V vs NHE)

}

(13)

We were not able to find stable adsorption sites of H2O2 on neither Au(100) nor Au(111) surfaces. During the geometry optimization of the initial structures, the molecules either desorbed or dissociated on the surface forming two OH. It is interesting to note that dissociation during the relaxation calculations occurred only on some of the adsorption sites on Au(111). However, this result might be partially influenced by the specific geometry characteristics of the initial guess configurations and may not originate from differences in the properties of the two surfaces. In addition, we must point out that previous density functional theory calculations of H2O2 adsorbed on Au(100) and Au(111) have suggested adsorption energies on the top sites on the order of -17 and -22 kJ‚mol-1,7 which if incorporated in the present reaction schemes will stabilize further the products of reactions 10 and 13, but naturally will also add an additional step of desorption of hydrogen peroxide from the surface related with exactly the same energy loss. The overall reaction of formation of hydrogen peroxide is

O2 + H2 f H2O2 with a computed enthalpy of formation of -164 kJ‚mol-1, in reasonable agreement with the experimental value of -188 kJ‚mol-1.21 Water. Water is the final product of the four-electron reduction of oxygen and can be formed from hydroxyl adsorbed on the surface

HO/Au + H+ (aq) + e- f H2O + Au, ∆H ) (12)

as well as via reactions 5 and 8 and the reverse of reaction 14 (i.e., oxidation of water). The overall reaction of formation of HO/Au is

0.5 O2 + 0.5 H2 + Au f HO/Au and the enthalpies for the bridge and top sites on the (100) gold surface are given in Figure 5. The enthalpies of formation for the bridge and top sites on Au(111), which also exhibit tilted configurations of OH on the surface (not shown), are -126 and -111 kJ‚mol-1, respectively. The hollow adsorption sites on either of the two surfaces are unstable and converge toward the bridge sites during geometry optimization. The hydroxyl species

{

Au(100) : -85 kJ‚mol-1 (at E ) 0 V vs NHE) Au(111) : -124 kJ‚mol-1 (at E ) 0 V vs NHE)

}

(14)

The overall reaction of formation of water is

0.5 O2 + H2 f H2O with computed enthalpy of formation of -251 kJ‚mol-1 to be compared with the experimental value of -286 kJ‚mol-1.21 Reaction Paths for Oxygen Reduction. On the basis of the above data, one can construct a reaction path scheme for the (electrochemical) reduction of oxygen on Au(100) and Au(111) surfaces (Figure 6). The graph incorporates all of the above reaction steps and includes also the various possible adsorption sites for the intermediates.

Electrochemical Reduction of Oxygen

J. Phys. Chem. C, Vol. 111, No. 6, 2007 2611

Figure 6. Schematic representation of the reaction intermediates formed in the course of electro-chemical oxygen reduction on the Au(100) (upper part) and Au(111) (lower part) surfaces and their relative enthalpies at zero potential (vs NHE, normal hydrogen electrode). No activation energies/ barriers are indicated. “X/Au” denote species X adsorbed on the corresponding surface. The thick solid horizontal levels depict the various adsorption sites. Shaded rectangles depict those intermediates which can still lead to hydrogen peroxide. The thin connecting lines represent the reaction steps. “+W” indicates that a water molecule is involved in the reaction as a reactant. “-W” means that water is produced as a result of the reaction. The vertical doted lines indicate that the reaction steps crossing the line are electrochemical reactions (i.e., involving charge transfer through the interface). The potential dependence can in principle be included in the graph and it effectively results in a shift of the relative positions of the energy levels in the regions I, II, III, and IV, in a similar way as illustrated for example in ref 14. The electrochemical reactions will reverse when the shift (upward) due to the electrical potential results in the energy levels of the products being above those of the reactants.

As shown in Figure 6, on both Au(100) or Au(111) for each of the intermediates O2 and OOH, which could potentially lead to hydrogen peroxide, one can find a purely chemical exothermic step breaking the O-O bond and leading to water as a final product. One can therefore conclude that the relative stability of the intermediates is not the determining factor for the selectivity of Au(100) and Au(111) surfaces toward four-electron or two-electron reduction of oxygen. It is also apparent from the relatively broad distribution of the adsorption energies of the individual intermediates on (100), that Au(100) exhibits greater differences between the adsorption energies of the different adsorption sites compared to Au(111). Although this site-selectivity cannot explain directly the differences in the catalytic behavior of the two surfaces, it may play a role in the liquid phase, where the presence of water will affect the mobility and/or stability of the intermediates on the surface. The relatively rigid structure of interfacial water, dictated by the extensive hydrogen-bond network, could for example result in only the top site being accessible for adsorption [after exchange with an interfacial water molecule, see for example ref 22 for the case of solvated OH on Rh(111)]. In essence this means that the hollow or the bridge sites may not always be available, and therefore, the reaction enthalpies will vary. Water will also have an effect on the stability of the intermediates through solvation effects, but one could also speculate that the solvation effects of equally charged species will be comparable. Role of Water in the Electrochemical Oxygen Reduction. An important conclusion from the results presented above is

that water molecules, apart from contributing with solvation effects, could also play an “active role” in the process of oxygen reduction and engage in chemical reactions like 5, 9, and 10. These reaction involve breaking of the strong O-H bond in the water molecule, which is an endothermic process, but this energy loss can be counter-balanced by the exothermic adsorption energies of the newly formed intermediates [see eqs 5, 9, and 10]. To verify this possibility, particularly with respect to the activation barriers, we performed a series of molecular dynamics simulations at 300 K of O2 molecules coadsorbed with a monolayer of water on the Au(100) surface. The length of the individual simulations were on the order of 0.5-0.8 ps. [Note: The molecular dynamics simulations were performed for relatively short simulation times as due to the very weak interaction between the water molecules and the surface at longer simulation times the H2O molecules will essentially desorb from the surface.] Two possibilities for the absorption configuration of O2 were considered: (i) with the molecule flat on the surface and (ii) with the molecule perpendicular to the surface. The molecules approximately retained their initial orientation during the simulations due to the presence of solvating water molecules, and both configurations are comparable in terms of total energies. As derived from the electron density of states, in either case the oxygen molecule is effectively charged as a peroxolike species O22- and exhibit a preference toward adsorption on a bridge site. The simulations revealed that a reaction of type (5) as well as its reverse, both of which involve H-O bond breaking (Figure

2612 J. Phys. Chem. C, Vol. 111, No. 6, 2007

Vassilev and Koper

Figure 7. Snapshots from the molecular dynamics simulations depicting the process of proton exchange (left to right) between the oxygen molecule O2 and a water molecule. The proton involved in the exchange is indicated with a black arrow.

7), can easily occur in the course of an MD run without applying any constraints or specialized MD methods for an acceleration of infrequent events. Essentially, this means that the activation energies of these processes are very low and the oxygen species readily exchange a proton with a neighboring water molecule. The proton exchange occurs between the oxygen molecule and a neighboring water molecule in the first layer next to the metal surface. This result can be attributed to the electron charge transfer to the oxygen molecule, which leads to the formation of the relatively strong base O22-. Possible activation of water molecules next to a metal surface, analogously to the observed increased dynamics of proton hopping in the case of hydroxylwater coadsorption on Rh(111),22 could also contribute to this effect. For a long time, it has been speculated that the stability of O2 on the surface, particularly with regard to the dissociation of the O-O bond, is the most important factor determining the outcome (water or hydrogen peroxide) of the oxygen reduction reaction.1-3 The present ab initio molecular dynamics simulations show, however, that in fact upon adsorption the solvated oxygen molecule can easily be converted to OOH, making hydroperoxyl as one of the most important intermediates on the surface, the stability of which would determine the final products of the reduction process. It is interesting to note that, although water molecules appear as reductive agents for O2, O, or OOH, reactions 5, 9, and 10 are not electrochemical reduction processes. In essence, the process of reduction of O2, O, or OOH on the surface has been “shifted” to reduction of OH on the surface, which will consequently occur via reaction 14 [in alkaline solutions this will be desorption of OH-]. Reaction 14 reinstates a water molecule, which completes a (pseudo-)catalytic cycle of electrochemical reduction involving H2O as a cocatalyst. A consequence of reactions 5, 9, and 10 however is that an OH is formed on the surface in a close proximity to O2, OOH, or H2O2, respectively. It has been speculated that OH species on the surface will catalyze the process of dissociation of the O-O bond, which may explain the (difference of) catalytic behavior of Au(100) in acidic and alkaline solutions. This assumption also prompted a recent investigation including DFT simulations of O2 and OH coadsorbed on Au(100) and Au(111) surfaces.5 The results however were inconclusive and could not explain the experimental data. The comparisons made between single oxygen molecules and an oxygen molecule and hydroxyl adsorbed on neighboring sites did not reveal any obvious variation in the properties of the O2 molecule. Our simulations suggest that in fact even in acidic solutions one could observe OOH (or H2O2) adsorbed in close proximity to HO as a result of reactions 5, 9, and 10. Therefore, in order to fully comprehend the catalytic properties of gold surfaces for the reduction of oxygen, one has to consider models featuring high OH coverage, with OOH and solvating water molecules included. The mechanism of the dissociative steps as a result of the complex

interplay between intermediates, OH, and activated water molecules however is still to be understood. Summary and Conclusions Density functional theory simulations of the process of electrochemical reduction of oxygen showed that the orientation of the adsorbed on the surface oxygen molecules alone is not the determining factor for the structural and pH selectivity of the Au(100) and Au(111) surfaces. On either of the surfaces, the adsorption energies are very small and the molecule lays predominantly flat on the surface. The relative stability of the other intermediates, while displaying differences in the relative stability, also suggests that on either of the surfaces it is thermodynamically favorable to dissociate the O-O bond of either of O2, OOH, or H2O2. One should therefore search for the mechanism and energy barriers of the dissociation of these intermediates, as influenced by the coadsorbed water molecules in the presence or absence of OH, in order to understand the processes on the surface Apart from solvation, water plays an active role in the electrochemical processes on the surface engaging in proton exchange reactions. This process could occur rather rapidly and results in an apparent cocatalytic role of water for the electrochemical reduction of oxygen. The ab initio molecular dynamics simulations also showed that solvated oxygen molecule adsorbed on the metal surface can easily be converted to hydroperoxyl OOH. In essence, this means that hydroperoxyl OOH, rather than O2, is one of the most important intermediates for this reaction, the stability of which would be the determining factor for the final outcome of the reduction process. Acknowledgment. Financial support from the European Union Under Contract 505906, Project NENA, and from The Netherlands National Computing Facilities foundation (Stichting Nationale Computerfaciliteiten, NCF) is gratefully acknowledged. References and Notes (1) Markovic, N. M.; Ross, P. N., Jr. Surf. Sci. Rep. 2002, 45, 117. (2) Yeager, E. J. Mol. Catal. 1986, 38, 5. (3) Wroblowa, H. S.; Yen-Chi-Pan; Razljmney, G. J. Electroanal. Chem. 1976, 69, 195. (4) Shi, Z.; Zhang, J.; Liu, Z.-S.; Wang, H.; Wilkinson, D. P. Electrochim. Acta 2006, 51, 1905. (5) Kim, J.; Gewirth, A. A. J. Phys. Chem. B 2006, 110, 2565. (6) Li, X.; Gewirth, A. A. J. Am. Chem. Soc. 2005, 127, 5252. (7) Li, X.; Heryadi, D.; Gewirth, A. A. Langmuir 2005, 21, 9251. (8) Kuzume, A.; Herrero, E.; Feliu, J. M.; Ahlberg, E.; Nichols, R. J.; Schiffrin, D. J. Phys. Chem. Chem. Phys. 2005, 7, 1293. (9) Panchenko, A.; Koper, M. T. M.; Shubina, T. E.; Mitchell, S. J.; Roduner, E. J. Electrochem. Soc. 2004, 151, A2016. (10) Sˇ trbac, S.; Adzˇic´, R. R. Electrochim. Acta 1996, 41, 2903. (11) Sˇtrbac, S.; Anastasijevic´, N. A.; Adzˇic´, R. R. J. Electroanal. Chem. 1992, 323, 179. (12) Markovic´, N. M.; Adzˇic´, R. R.; Vesˇovic´, V. B. J. Electroanal. Chem. 1984, 165, 121.

Electrochemical Reduction of Oxygen (13) Xu, Y.; Mavrikakis, M. J. Phys. Chem. B 2003, 107, 9298. (14) Nørskov, J. K.; Rossmeisl, J.; Logadottir, A.; Lindqvist, L.; Kitchin, J. R.; Bligaard, T.; Jo´nsson, H. J. Phys. Chem. B 2004, 108, 17886. (15) Kresse, G.; Furthmu¨ller, J. Comput. Math. Sci. 1996, 6, 15. (16) Kresse, G.; Furthmu¨ller, J. Phys. ReV. B 1996, 54, 11169. (17) Blo¨chl, P. E. Phys. ReV. B 1994, 50, 17953. (18) Kresse, G.; Joubert, J. Phys. ReV. B 1999, 59, 1758.

J. Phys. Chem. C, Vol. 111, No. 6, 2007 2613 (19) Perdew, J. P. Electronic Structure of Solids; Akademie Verlag: Berlin, 1991. (20) Monkhorst, H. J.; Pack, J. D. Phys. ReV. B 1976, 13, 5188. (21) Lide, D. R. ed., CRC Handbook of Chemistry and Physisc, 74th ed.; CRC Press: Boca Raton, FL, 1993. (22) Vassilev, P.; Koper, M. T. M.; van Santen, R. A. Chem. Phys. Lett. 2002, 359, 337.