Langmuir–Hinshelwood Mechanism Including Lateral Interactions and

Jan 15, 2014 - Fax: +57 4 425-5254. ... The electrochemical oxidation of CO on metallic surfaces following a lattice-gas model, including effective la...
0 downloads 0 Views 4MB Size
Article pubs.acs.org/JPCC

Langmuir−Hinshelwood Mechanism Including Lateral Interactions and Species Diffusion for CO Electro-Oxidation on Metallic Surfaces Ana M. Gómez-Marín† and Juan P. Hernández-Ortiz*,‡,§ †

Departamento de Química y Petróleos and ‡Departamento de Materiales, Universidad Nacional de Colombia, sede Medellín, Carrera 80 # 65-223, Bloque M3-050, Medellín, Colombia § Biotechnology Center, University of WisconsinMadison, Madison, Wisconsin 53706, United States ABSTRACT: The electrochemical oxidation of CO on metallic surfaces following a lattice-gas model, including ef fective lateral interactions between one of the adsorbates and species diffusion, is studied. The reaction occurs through a Langmuir−Hinshelwood (LH) mechanism in which adsorbed CO reacts with adsorbed hydroxyl species. The mean field approximation and dynamic Monte Carlo simulations have been compared. The effect of the molecular distribution and surface mobility of reacting species on the potential dependence of the CO oxidation rate is analyzed. The inclusion of lateral interactions into the reaction mechanism reconciles different experimental observations, such as island formation and fast CO diffusion. Results highlight the importance of ef fective interactions in the reaction kinetics and suggest that they should be taken into account when interpreting experimental data. Simulations are useful for an improved qualitative understanding of the kinetics of CO oxidation and other electrochemical LH reactions. growth, N&G, of these islands.2,10−15 The second approach is the mean-field approximation (MFA).8,16−19 It assumes a fast COads and OHads surface diffusion and their perfect mixing at the electrode surface. Recently, it was proposed that the oxidation proceeds at specific surface sites (surface defects, kinks, edges, etc.) and both COads and OHads diffuse to the reaction zone created by these sites.18 It should be mentioned that while spectroscopic data support the N&G mechanism,2,3,20−23 accurate fitting of experimental electrochemical current transients validate the MFA.16,17,24−26 There are, however, other intrinsic factors that may be responsible for the observed experimental results. As has been highlighted before,1 physical properties of COads are quite different from those of other adsorbates on the surface, such as H2Oads, OHads, and other anions, such as HSO4,ads, whose hydrogen bonds can be appreciably stronger than CO−CO lateral interactions. Hence, the formation of these bonds may induce COads segregation, knowing that electrochemical oxidation is usually studied at room temperature (RT). In consequence, the influence of all these interactions between adsorbates on the COads oxidation can determine the surface structure, adsorption/desorption kinetics, diffusion, and chemical reaction dynamics. Specifically, the effective interactions inside the adlayer may be attractive or repulsive, depending on the balance between interparticle and intraparticle interactions.27−29 Influence of lateral interactions on the electrochemical oxidation of CO has been suggested from CO

1. INTRODUCTION CO oxidation and adsorption reactions are one of the cornerstones of surface science and one of the most intensively studied processes in electrocatalysis.1 They comprehend complex physical and chemical phenomenology that provides fundamental motivation in research and technology. The exact mechanism of the CO oxidation reaction continues to be a subject of debate. Although in situ electrochemical measurements, such as scanning tunneling microscopy (STM) and Fourier transform infrared (FTIR)2−4 as well as information obtained in ultrahigh vacuum (UHV) environments,5 have yielded a wealth of structural data on CO adsorption on platinum single-crystal electrodes, it is relatively unclear how these microscopic structural features affect the overall macroscopic electrocatalytic activity. The most widely accepted mechanism for electrochemical CO oxidation on metallic surfaces is the Langmuir−Hinshelwood (LH) scheme or reactant pair. In this mechanism, adsorbed CO (COads) reacts with oxygen-containing species located at an adjacent site on the surface (see ref 6). In contrast, some authors have proposed that the stripping of a saturated layer of COads follows an Eley−Rideal (ER) scheme in the early stage of the process. Thus, the reaction proceeds without adsorption of anions, or H2O molecules (H2Oads), and, after some time, it follows a LH mechanism.7 In any case, the real identity of the oxygenated species is not fully established yet.8,9 In the LH scheme, two different models have been invoked to describe the reaction kinetics. The first one assumes that the reaction proceeds at the perimeter of water islands containing adsorbed hydroxyl species, OHads, within the monolayer (ML) of COads. The kinetics is controlled by the nucleation and © 2014 American Chemical Society

Received: September 19, 2013 Revised: December 19, 2013 Published: January 15, 2014 2475

dx.doi.org/10.1021/jp409385k | J. Phys. Chem. C 2014, 118, 2475−2486

The Journal of Physical Chemistry C

Article

oxidation studies on Pt single crystals30 and Nafion coated Pt(111) electrodes.31 In contrast, most of the studies that model the electrochemical CO oxidation on metallic surfaces treat adsorbed reactants as Langmuir particles, that is, no lateral adsorbate− adsorbate interactions between the species were considered.8,25,26,32−34 Trying to further this approach, Zhdanov and Kasemo1 simulated a two particle system, A and B, with B−B attractive interactions using Monte Carlo (MC) simulations, proving that current transients for CO oxidation are fairly sensitive to adsorbate−adsorbate lateral interactions. Specifically, they showed that with realistic values of these interactions (∼4.2 kJ/mol) the reaction kinetics might be accompanied by CO segregation, provided that surface CO diffusion is fast. Recently, an LH model including ef fective attractive and repulsive COads interactions was solved by the MFA.35 However, under this approximation, the effect of repulsive/ attractive short-range interactions may be overestimated or underestimated, depending on the process.36 By construction, the MFA assumes that the surface is homogeneous and so adsorbate−adsorbate correlations cannot be taken into account. Therefore, average lateral interactions and infinite diffusion rates are assumed.36 On the other hand, different works have shown that surface diffusion of COads and/or OHads may play a key role on the whole reaction dynamics.12,13,16−18,24,37−46 The first theoretical efforts to describe the effect of COads diffusion on the reaction kinetic were made by Petukhov25 and Koper et al.8 Both works simulated an LH model by MFA and MC to illustrate the role of COads diffusion on the COads oxidation. Later, the MC technique was also employed by other groups to simulate other aspects of the reaction.26,32−34,47,48 For an extensive review of MC of catalytic reactions, see refs 49−51. The goal of the present study is to analyze the effect of lateral adsorbate−adsorbate interactions and OHads/COads diffusion on the potential dependence of the CO oxidation rate constant. An LH mechanism between adsorbed COads and OHads, resulting from water dissociation, is solved by dynamic MC (DMC). The model fully incorporates the nearest neighbor (nn) lateral interactions and the correct time dependence of rate constants.52 Although this model ignores higher order interactions, such as three-body interactions and long-ranged interactions, within a defined interaction model, the DMC results are accurate. Noting that this model is quite simple, we present only general results and do not try to interpret any experiment in detail. Simulated results are considered useful for an improved qualitative understanding of the kinetics of CO oxidation and in the general theory for other electrochemical LH reactions. The paper is organized as follows: In section 2, the reaction kinetic model is described as well as the methodology of solution based on the DMC method. In section 3, results start with the noninteracting model, continuing to systems with attractive and repulsive interactions. The results are presented in terms of stripping voltammetry and chronoamperometry. At the end of this section, a brief comparison between theoretical simulations and experimental data is made. The paper ends with a summary of the most important observations and conclusions.

but, quantitatively, the difference is related to the number of nn sites: 4 for square lattices and 6 for triangular ones.53 Due to the complexities in the real system, the proposed model is not sufficiently realistic to understand the detailed structural aspects of CO adsorption at the metal/electrolyte interface; however, the model is able to provide interesting insights into the importance of microscopic features such as mixing, lateral diffusion, and island formation. The basic features of our model are explained below; for a more detailed description, see ref 8. Following the Gilman model,6 the reaction is described with an LH mechanism with participation of adsorbed oxygencontaining species. The exact nature of this species is not known.1,18 Traditionally, the two most likely candidates are H2Oads and OHads, but atomic oxygen, Oads, has also been recently suggested.9,54 Considering that electrochemical measurements indicate that both reactions, the formation of the oxygen-containing species and CO oxidation, are potential dependent, OHads is assumed to be the oxidizing species in the present model, resulting from water dissociation. The reaction model is8 k1, k −1

H 2O + * ←⎯⎯→ OHads + H+ + e− k2

COads + OHads → CO2 + H+ + e− + 2*

(1) (2)

where the asterisk (*) denotes a free surface site (or H2Oads). Strictly speaking, complete schemes of CO oxidation including lateral interactions should comprise εCO/OH, εCO/W, εOH/OH, εOH/W, and εW/W parameters (with W for water molecules), where εi/j is the interaction energy between molecule i and molecule j when they occupy neighboring sites on the surface. However, these interaction parameters are unknown or poorly defined, or including all of them is too complicated to be fully described in theoretical models. For example, OHads formation on metal surfaces is complex,55 even on monocrystalline surfaces.56 Density functional theory (DFT) calculations have shown that εOH/OH, εOH/W, and εW/W energies are highly directional; that is, they can be strongly repulsive or attractive depending on the surface angle between the new OHads molecule and H2Oads and OHads molecules already adsorbed.57 Macroscopically, however, lateral interactions between different adsorbates may give rise to adsorbate−adsorbate correlations and even island formation, which in turn increases or decreases the overall oxidation reaction rate. Hence, similar to previous studies,1,35 an effective interaction energy inside the adlayer between surface neighboring CO−CO molecules, ϵCO−CO, will be considered, instead of a full description of each lateral interaction among different adsorbed species, to keep the model as simple as possible, with a measurable physical meaning and close to observable phenomena. Two types of lateral interactions are usually distinguished:1,53,58 Adsorbate−adsorbate interactions, in an adsorbed state, and adsorbate−transition state interactions, in an activated state. An adsorbed state means an adsorbate sitting at an optimized adsorption site, whereas an activated state refers to an intermediate reaction at the transition state. Lateral interactions of the adsorbed state have a twofold influence on the local adsorbate configuration at the active site: (i) they change the coverage of different adsorbates and (ii) they introduce correlation among adsorbates. At the activated state, lateral interactions depend on the local adsorbate configuration at the active site and modify the transition state energy.

2. REACTION MODEL AND COMPUTATIONAL METHOD The model follows an ‘‘A + B’’ reaction scheme on a square lattice. For triangular lattices, results are qualitatively the same, 2476

dx.doi.org/10.1021/jp409385k | J. Phys. Chem. C 2014, 118, 2475−2486

The Journal of Physical Chemistry C

Article

According to DFT calculations,59 the reaction may actually occur via two steps: first COads and OHads react to form COOHads, and then the latter species reacts to form CO2. Hence, the structure of the corresponding transition state is close to that of the initial state and the difference of lateral interactions in the transition and initial states is expected to be relatively small. To investigate the effect of lateral interactions, the activation energy, Ea, is required to depend on the local adsorbate configuration i at the active site (only lateral interactions between nn sites are considered). This is included by assuming that every occupied nn site modifies Ea by the same amount. In this context, for the CO oxidation reaction, Ea is decreased or increased by an amount nCOγ2ϵCO−CO, with nCO being the number of COads occupying nn sites of the reacting COads and γ2 is a Brønsted factor specifying to which extent the transition state for adsorption/desorption is influenced by the interactions (in this work, γ2 = 0.5). The reaction rate constants obey the Butler−Volmer law for electrochemical reactions,60 and they are given by ⎛ β e0E ⎞ k OH,ads = k1 = k10 exp⎜ 1 ⎟ ⎝ kBT ⎠ k OH,des

⎛ (1 − β )e0E ⎞ 1 = k −1 = k −01 exp⎜ − ⎟ kBT ⎝ ⎠

The surface diffusion in DMC is included, assuming that COads and/or OHads surface diffusion occurs via the following mechanism: COads + * → * + COads

OHads + * → * + OHads

(8) −1 8

This “reaction” rate is defined by D (s ), corresponding to a surface diffusion coefficient of around 10−15D cm2 s−1. In this work, D is varied to assess the influence of COads and OHads surface diffusion. This parameter implicitly incorporates the role of the interfacial solvation, and detailed solvent structure at the Pt/solution interface is not considered. This rate may depend on the electrode potential, but this effect is expected to be weak compared to that of the reaction steps because COads or OHads diffusion does not involve charge transfer. Therefore, the jump rate constants are considered to be independent of the electrode potential. When lateral interactions are taken into account, considerations explained in previous paragraphs also apply to diffusion. The rate constants were chosen such that detailed balance was satisfied, that is, ⎛ ζdiff ΔNεij ⎞ kdiff = D exp⎜ − ⎟ kBT ⎠ ⎝

(3)

(9)

where ζdiff = 0.5 is the Brønsted−Polanyi coefficient for diffusion and ΔN is the difference in the number of occupied nn neighbor shell sites before and after the hop (i.e., nCO,after − nCO,before). Tables 1 and 2 resume the main model equations and parameters employed in this study.

(4)

⎛ ⎞ ⎛ β FE ⎞ γ e0E k CO2,des = k 2 = k 20 exp⎜ 2 ⎟ exp⎜nCO 2 ϵCO − CO⎟ ⎝ RT ⎠ kBT ⎝ ⎠ (5)

Table 1. Model Parameters for the Electrooxidation of Saturated CO Adlayers on a Squared Lattice

where β’s are transfer coefficients (taken to be 0.5), E is the electrode potential, e0 is the electron charge, kB is the Boltzmann constant, and T is the absolute temperature. Subscripts “1″, “−1”, and “2” denote OH adsorption, OH desorption, and CO oxidation reactions, respectively. Reaction rate constants k01, k0−1, and k02 are independent of the lateral interactions and have the dimension of reciprocal seconds. The rate constants k01 and k0−1 include surface concentrations of water and proton, respectively, while k02 is the rate constant at the low-coverage limit. Faraday’s law gives the total current density from these reactions, that is, j = e0(v1 − v−1 + v2)

(7)

process

equation

OH adsorption

k OH,ads

(6)

where vj denotes the rate of the jth reaction (cm−2 s−1). The oxidation of a certain amount of preadsorbed CO, with no CO present in the solution, is carried out by a voltammetric potential scan or a potential step. On Pt(111) and Pt(100), the absolute saturation coverage of CO is between 0.6 and 0.7 per Pt site.61,62 Instead, here and below, coverages are defined as the ratio of the actual coverage to the absolute saturation coverage. Thus, the relative coverage used in calculations may be close to unity. In this work, a relative coverage of 0.99 is considered as the saturation coverage. In addition, following previous studies,8,33,35,47 rate constants for OH adsorption and desorption were chosen such that for 20 mV s−1, OH adsorption appears quite reversible, and are equal to 0.02 and 10 000 s−1, respectively. The CO oxidation reaction constant can adopt different values, as indicated. Hence, the potential axis used is, to some extent, arbitrary. Values of lateral interactions evaluated are in the range from +0.1 (repulsion) to −0.1 eV/atom (attraction), equivalent to ca. ± 9.6 kJ/mol.

⎛ β e0E ⎞ = k1 = k10 exp⎜ 1 ⎟ ⎝ kBT ⎠

OH desorption

⎛ (1 − β )e0E ⎞ 1 k OH,des = k −1 = k −01 exp⎜− ⎟ kBT ⎝ ⎠

CO oxidation

⎛ ⎞ ⎛ β FE ⎞ γ e0E k CO2,des = k 2 = k 20 exp⎜ 2 ⎟ exp⎜nCO 2 ∈CO − CO ⎟ kBT ⎝ RT ⎠ ⎝ ⎠

Total current density CO/OH diffusion

j = e0(v1 − v−1 + v2) ⎛ ζdiff ΔNεij ⎞ kdiff = D exp⎜− ⎟ kBT ⎠ ⎝

Experimentally, COads diffusion on Pt/gas phase interface is known to be very fast (the activation barrier is about 0.3 eV63), lying in the range of 10−9−10−11 cm2 s−1 at RT.64 Under electrochemical conditions, COads diffusion is expected to be slower due to CO interactions with the solution. Accurate data for this regime are still missing. Estimated values for the COads diffusion coefficient from potentiostatic transients were above 1 × 10−11 cm2 s−1.12,13,16,17,19 Instead, electrochemical nuclear magnetic resonance (EC-NMR) spectroscopy studies found lower values, ∼3.6 × 10−13 cm2 s−1.44 Similarly, the OHads diffusion is supposed very fast. However, recent works have suggested the formation of a hydrogen bonded water/OHads network as the rate-determining step (RDS) in the CO oxidation.31,40 In the absence of strongly adsorbing anions on 2477

dx.doi.org/10.1021/jp409385k | J. Phys. Chem. C 2014, 118, 2475−2486

The Journal of Physical Chemistry C

Article

Table 2. Model Parameters for the Electrooxidation of Saturated CO Adlayers on a Squared Lattice symbol

description

values

nn ϵCO−CO nCO γ2 β k01 k0−1 k02 D ζdiff T

nearest neighbor molecules ef fective interaction energy between surface neighboring CO−CO molecules. no. of COads occupying nn sites of the reacting COads Brønsted factor specifying to which extent the transition state is influenced by the interactions transfer coefficients OH adsorption reaction rate constant OH desorption reaction rate constant CO oxidation reaction rate constant diffusion rate constant Brønsted−Polanyi coefficient for diffusion Absolute temperature

4 +0.1 to −0.1 eV/atom 0−3 0.5 0.5 0.02 10 000 8.234 × 10−5 to 82.34 s−1 0−10 000 0.5 300 K

“ideal” platinum surfaces, this process is fast and reversible, but in their presence OHads adlayer formation may be slow. Fast OHads and COads diffusion was suggested when comparing CO stripping voltammograms (SV) on bare and Nafion covered Pt(111) electrodes with theoretical curves, calculated by a MFA model including attractive CO−CO interactions.35 A discussion about the role of OHads diffusion in the current response during CO oxidation by either stripping voltammetric or potential-step chronoamperometric experiments, under an LH scheme has been done in a previous work.8 In the next sections, this discussion will be extended to the case when CO−CO lateral interactions between the species are considered. 2.1. Methodology of Solution. Dynamic Monte Carlo simulations were done using the software CARLOS.65 This is a general-purpose program for DMC simulations of surface reactions. CARLOS’s possibilities and features are described in detail in ref 65 and on CARLOS’s Web site.66 The program accounts for time dependence of reaction rate constants in either cyclic voltammetry or potential step chronoamperometry.8 Simulations were performed on 128 × 128 and 256 × 256 square lattices with periodic boundary conditions. Apart from a lower noise level on the larger lattice, there is not any observable difference between both lattices. Simulations started with an initial random configuration generated by the CARLOS program for each system. CO particles are adsorbed on the lattice up to a desirable coverage (usually 0.99 ML, unless stated otherwise), and then DMC steps of diffusion are performed to equilibrate the adsorbed overlayer. All voltammograms, potential step transients, and snapshots to be presented below were carried out on the smaller lattice size. The temperature of the system was always fixed at 300 K.

current maximum is reached. The descending part of the peak is described by the OHads formation, eq 2. These asymmetric transients cannot be explained by the theory, derived by Bewick, Fleischmann, and Thirsk (BFT),67,68 for nucleation and growth (N&G) processes. This is because the reaction occurs on already existing “islands”,8 which implies a breakdown of the Avrami theorem.69−71 Instead, if the COads oxidation reaction is the slowest process, current transients can be properly described by either the BFT theory for a progressive nucleation,67,68 or a MFA simplified model, derived from eqs 1 and 2, and assuming a reversible, fast reaction for the oxygen-containing species, so that its coverage, θOH, is always proportional to (1-θCO).1,16,25 Role of COads and OHads Surface Mobility. Figure 1 shows the current maximum time, tmax, from the current evolution, j−t

3. RESULTS AND DISCUSSION For the CO oxidation on metallic surfaces following an LH mechanism features like peak shape and Tafel curves, in cyclic voltammetry and potential-step chronoamperometry, have already been fully described.8,35 Nevertheless, a short discussion about the influence of the lateral diffusion and island formation in absence of adsorbed molecules’ interactions is given, as a starting point to understand the effect of adsorbate−adsorbate correlations. 3.1. Noninteracting System. The shape of potential step current transients depends on the controlling process, eq 1 or 2.8,35 Without diffusional effects, if the COads oxidation reaction is the fastest process, current transients are asymmetric, with a stepped current rise at the beginning and a fast decay after the

Figure 1. Logarithm of the current maximum time, tmax, as a function of the final potential, Ef. Lower curve:, k2 = 8.234 × 10−1 s−1 (olive left triangle, purple right triangle,); upper curve, k2 = 8.234 × 10−5 s−1 (black square, red circle, blue up triangle, magenta down triangle). MFA curves are also included (empty symbols).

profile, as function of the final potential, Ef, and at different DCO and DOH values. Two different sets of rate constants are used: A fast set, where the CO oxidation is faster than OH adsorption, k2 = 8.234 × 10−1 s−1, and a slow set, where OH adsorption is faster than the oxidation, k2 = 8.234 × 10−5 s−1. Figure 2 illustrates normalized current transients, j/jmax, after stepping from a potential at which a 0.99 CO ML is stable to Ef, for the slow set of rate constant. Normalizations are made for both 2478

dx.doi.org/10.1021/jp409385k | J. Phys. Chem. C 2014, 118, 2475−2486

The Journal of Physical Chemistry C

Article

change occurs when the diffusion length of the adsorbates is about the average cluster distance.72,73 In the case in Figure 2C and D, the crossover is understandable because the rate of the RDS, that is, CO oxidation, changes with the potential while DCO does not. Correspondingly, OH diffusion may also be the driving force of the crossover if OHads adsorption is the RDS in the mechanism (see below). Tafel slopes of 60 to 80 mV have been found in potentiostatic studies of CO oxidation.11,16,17,74,75 It has been suggested that such Tafel slopes, close to 60 mV, might indicate the presence of a slow chemical step in the reaction scheme as RDS.74−76 Here, Tafel slopes lower than 119 mV have been calculated in certain Ef ranges, without considering a chemical step in the mechanism. Besides, Tafel slopes of 240 mV,11,24 with a change from progressive to instantaneous nucleation at high potentials11 have been also reported for the CO oxidation on Pt surfaces. Although such a high Tafel is not reproduced in simulations, slopes higher than 119 mV can be obtained in narrow potential ranges at high Ef, by including the role of adsorbate diffusion. However, in this work, the reactive OH can absorb in all the empty sites once oxidation begins. It is expected that with a fixed number of active sites, COads and OHads diffusion will play a more important role. 3.2. System Including Lateral Interactions. Attractive Interactions. Figure 3 shows typical stripping voltammograms

Figure 2. Normalized DMC potential step current transients for k2 = 8.234 × 10−5 s−1, from a 0.99 CO ML at different final potentials: (blue up triangle) 0.65 V, (red circle) 0.50 V, and (black square) 0.35 V. MFA curves are also included (solid lines).

scales using jmax and tmax. Simulated results by the MFA are also included. Suppressing adsorbate diffusion, DCO = DOH = 0, enables surface island formation and COads molecules can be only oxidized at the perimeter of existing holes.8 Both sets of rate constants, fast and slow, produce current transients different to those calculated by the MFA. However, they show similar log tmax vs Ef curves, but shifted upward, with Tafel slopes at low Ef of 40 mV and at high Ef of 119 mV.8,35 The magnitude of the shift is larger for slower k2 (Figure 1). The shape of normalized current transients follows for the fast set, a progressive nucleation mechanism, with an initial quadratic rise in current,8 while for the slow set, a instantaneous nucleation scheme,67,68 with an initial linear rise in current, independent of Ef (Figure 2A). OHads diffusion does not change the system response (Figures 1 and 2B). COads diffusion, DCO = 100, enhances mixing and suppresses the formation of large holes, which also effectively diffuse. Thus, OHads adsorption can take place at many different places, making more COads available to react. In this case, log tmax vs Ef curves have an intermediate response (Figure 1): at Ef > 0.6 V, j−t profiles are similar to DCO = 0 curves,8 with Tafel slopes of ∼119 mV. For the slow set, transients are described by an instantaneous nucleation scheme (Figure 2C and D). At Ef < 0.4 V, curves follow the MFA dynamics, with Tafel slopes of 80−105 mV and current transients explained by a progressive nucleation scheme,8 Figures 2C and D. Between 0.4 < Ef < 0.6 V, Tafel slopes are ca. 145−155 mV, and transients have an intermediate shape between both cases (Figure 2C and D at 0.50 V). The slope in this region depends on the DCO/k2 ratio. The change in the transient shape with Ef is an effect of the relation between the ef fective molecular adsorbate mobility and the ef fective constant rate of the RDS (kdiffusion/kRDS) on the LH scheme. Current transients, for a reaction following an LH scheme, are asymmetric at low adsorbate diffusion (instantaneous nucleation) and become more symmetric when the hopping rate (diffusion) is increased (progressive nucleation), compared to the constant rate of the RDS.25 This crossover from a continuous to an instantaneous nucleation has been reported before,72 even in dynamics with local barriers.73 This

Figure 3. Stripping voltammograms at 50 mV s−1 and attractive lateral interactions ϵCO−CO. Initial CO coverage = 0.99 ML. k2 = 8.234 × 10−2 s−1, (purple square) MFA; DCO = 0−DOH = 0 (black up triangle); DCO = 100−DOH = 0 (red down triangle). k2 = 82.34 s−1, (olive circle) MFA; DCO = 0−DOH = 0 (maroon left triangle); DCO = 100 and DOH = 0 (green right triangle); (blue star) DCO = 100 and DOH = 10 000. DMC curves without interactions are also given (empty symbols).

for CO oxidation on a squared lattice, calculated taking into account attractive lateral nn CO−CO interactions (negative ϵCO−CO values) and neglecting shifts of equilibrium positions of adsorbed particles, for different DCO and DOH values. In Figure 4, SVs at different scan rates, v, for k2 = 82.34 s−1 and ϵCO−CO = −0.1 eV are given. Attractive interactions shift the CO oxidation peak, Ep,CO/OH, to higher potentials,35 proportional to their magnitude; that is, greater |ϵCO−CO| values produce larger shifts. MFA predicts faster reaction rates than DMC, although the effect of attractive interactions is overestimated in this approach. This highlights the important role of surface 2479

dx.doi.org/10.1021/jp409385k | J. Phys. Chem. C 2014, 118, 2475−2486

The Journal of Physical Chemistry C

Article

Figure 5. Snapshots during the stripping voltammetry in Figures 3 and 7. Green, OH; red, CO; blue, empty site (or H2Oads). (A, B) ϵCO−CO = −0.1 eV, k2 = 8.234 × 10−2 s−1, DCO = 100−DOH = 0 (A); k2 = 82.34 s−1, DCO = 100−DOH = 10 000 (B). (C−F) ϵCO−CO = 0.1 eV, k2 = 8.234 × 10−4 s−1, DCO = 100−DOH = 0 (C); k2 = 8.234 × 10−2 s−1, DCO = 0−DOH = 0 (D); DCO = 100−DOH = 0 (E); DCO = 0−DOH = 100 (F).

Figure 4. Scan rate study for k2 = 82.34 s−1 and ϵCO−CO = −0.1 eV. Initial CO coverage = 0.99 ML. MFA (■); DCO = 100−DOH = 0 (▲); DCO = 100−DOH = 10 000 (▼); DCO = 0−DOH = 100 (●). For 50 mV s−1, view Figure 3D.

diffusion on the oxidation dynamics, which decreases the CO reaction rate. This is not taken into account into the MFA (empty symbols, Figure 3). In the MFA, the intensity and sharpness of Ep,CO/OH increase at stronger attractions. Instead, in DMC, they depend on k2 and DCO, or DOH, values. At k2 = 8.234 × 10−2 s−1 and DCO = 0, Ep,CO/OH is broad, regardless of ϵCO−CO or DOH. Contrarily, for DCO = 100, Ep,CO/OH broadens and decreases in intensity when decreasing ϵCO−CO (Figure 3). For strong attractions (−0.1 eV), Ep,CO/OH for DCO = 100 coincides with DCO = 0 (Figure 3D), even at k2 = 8.234 × 10−1 s−1. Instead, for strong attractions and fast CO constant rates, k2 = 82.34 s−1, the DOH may play a non-negligible role (Figure 3D). At slow v, Ep,CO/OH occurs in a potential region where OH desorption is the fastest reaction and so the effective OHads diffusion is high, even for DOH = 0. Thus, Ep,CO/OH for DCO = 100−DOH = 0 and DCO = 100−DOH = 10 000 are similar (Figure 4A−C). At faster v, Ep,CO/OH becomes broader and shifts to higher potentials if DOH = 0 at the OHads adsorption region (Figure 4D−F). So, the effective adsorption/desorption OH reaction is not so fast to fill all the “holes” in the adlayer and DOH is fast enough to increase OHads concentration at the perimeter of the reaction front. In any case, the full width at half-maximum (fwhm) of the peak slightly increases at faster v. Therefore, OHads diffusion has influence on the mechanism if it is faster than the ef fective OH adsorption/desorption rate and this latter reaction controls the entire oxidation rate. This is clearly seen in the equilibrium distributions of the species over the surface (snapshots) given in Figure 5. If the CO oxidation controls the mechanism, the OHads adsorption is in quasiequilibrium, and large regions of COads can only be oxidized at the perimeter of the existing holes. These holes are fully occupied by OHads, impeding the diffusion (Figure 5A). Instead, if OHads adsorption controls mechanism, the OHads can diffuse through the empty regions around COads islands and increase the oxidation rate (Figure 5B). With attractive interactions, fast COads diffusion may facilitate island formation, and the average island size becomes greater as ϵCO−CO decreases.1 If the attraction is weak, the adlayer equilibration does not always keep up with the oxidation and diffusion does not compete with the reaction for adsorbed

COads. The holes inside the adlayer effectively diffuse along island domain boundaries and reduce the formation of large holes, increasing the oxidation rate. If the attraction is strong, COads diffuses instead to react and islands are almost as large as when DCO = 0. In other words, attractive interactions suppress “hole” diffusion, which is important to make more COads available to react with OHads.8 A plot of Ep,CO/OH vs log v is also nonlinear, but slopes can be higher than those in noninteracting systems.8,35 At v < 3 mV s−1, both simulations, MFA and DMC, predict slopes of 50 and 70 mV, while for 3 < v < 100 mV s−1 slopes are between 60 and 85 mV. At faster v, MFA slopes are around 107 and 111 mV, while DMC slopes are higher than 130 mV. Figure 6 shows normalized potential step current transients to different Ef, for two k2 and negative ϵCO−CO values. Transients from MFA are quite different to those from DMC, but similar to those without lateral interactions and fast rate constants.8 Analogous to Figure 2, some current transients in Figure 6 apparently show a change, from an instantaneous to a progressive nucleation at decreasing Ef, and become asymmetric at Ef = 0.35 V. At slow k2 and moderate attraction, DCO is at the origin of the crossover (Figure 6A−C). Contrarily, at fast k2 and strong attractive interactions (Figure 6D−F), DOH is the driving force for it. A simple explanation follows the same arguments discussed above. In addition, transients with an initial quadratic current rise show an extended plateau. This is because at the beginning of the oxidation, OHads replace COads species (no empty sites) and the reaction rate is almost constant.35 Instead, transients with an initial linear current rise have a small tailing in the descending part, especially at slow k2 and moderate attraction. When log tmax vs Ef is plotted, again nonlinear Tafel curves are obtained. At slow k2 and moderate ϵCO−CO, curves are similar to Figure 1. However, for fast k2 and strong attractions, at Ef < 0.25 V and 0.25 < Ef < 0.45 V, slopes between 45−50 and 84−125 mV are calculated, respectively. At Ef > 0.45 V, slopes of 119 mV are observed for DOH = 0, and for DOH = 100 they are higher. Differences between MFA and DMC are consequence of the inability of MFA to describe surface 2480

dx.doi.org/10.1021/jp409385k | J. Phys. Chem. C 2014, 118, 2475−2486

The Journal of Physical Chemistry C

Article

F. In this case, Ep,CO/OH broadens and shifts to lower potentials (Figure 7). The MFA again overestimates the oxidation rate and the shift in Ep,CO/OH is slightly higher than that in DMC. Stronger repulsions do not significantly change the SVs (Figure 7A and B). For instance, doubling ϵCO−CO produces a small shift in height and position of Ep,CO/OH. Additionally, if Ep,CO/OH occurs in a potential region where OHads desorption is important, DOH has little influence on the SV (Figure 8), because the effective DOH is high. For ϵCO−CO = 0.1 eV and k2 = 8.234 × 10−4 s−1, there is a shoulder in the positive side of the main peak for both MFA and DMC with DCO = 0.35 For DCO = 100 there are two distinct peaks in the SV (Figure 7C). This splitting of Ep,CO/OH is also seen at slow v and faster k2 (Figure 8). Thus, the fwhm of

Figure 6. Normalized DMC potential step current transients from a 0.99 CO ML. DCO = 0-DOH = 0 (red square); DCO = 0−DOH = 100 (magenta circle); DCO = 100−DOH = 0 (blue up triangle); DCO = 100− DOH = 10 000 (olive down triangle). (A−C) k2/ϵCO−CO = 8.234 × 10−4 s−1/−0.04 eV. (D−F) 82.34 s−1/−0.1 eV. MFA transients are also given (solid line).

heterogeneities, like island formation, and its assumption of infinite adsorbate diffusion. Repulsive Interactions. Figure 7 shows several stripping voltammograms once repulsive interactions are considered, and Figure 8 presents SVs at different scan rates, 0.3 < v < 100 mV s−1, for k2 = 8.234 × 10−2 s−1 and ϵCO−CO = 0.1 eV (for 50 mV s−1, see Figure 7B). Typical snapshots are given in Figure 5C−

Figure 8. Scan rate study for k2 = 8.234 × 10−2 s−1 and ϵCO−CO = 0.1 eV. Initial CO coverage = 0.99 ML. MFA (blue square); DCO = 0− DOH = 0 (black up triangle); DCO = 100−DOH = 0 (red down triangle); DCO = 0−DOH = 100 (pink circle). For 50 mV s−1, view Figure 7B.

Ep,CO/OH always increases at slower v. Faster v only produces single and tailed peaks. If DCO = 100, the transformation occurs at faster v than for DCO = 0. Increasing v decreases the charge under the first peak and increases it under the second one. These two peaks occur because of two different domains inside the adlayer (Figure 5C−F). The first peak represents domains with high packing density and faster oxidation rate due to repulsive interactions. The second peak corresponds to domains with low packing density, where the reaction occurs without the effect of interactions, so its potential dependence is independent of DCO. The MFA does not predict these two peaks because it cannot describe the nonrandom molecular distribution of the species on the surface. Both approximations MFA and DMC with DCO ≠ 0, predict similar Ep,CO/OH vs log v plots at slow v. However, the dynamics is slightly different at faster v. For DCO = 0, the curve shifts upward, because of the poor mixing inside the adlayer. Slopes around 50 and 70 mV are obtained for v < 30 mVs−1; but between 3 < v < 100 mVs−1 slopes are higher, ∼90−110 mV. At v > 100 mV s−1 and k2 slower than OHads adsorption, the COads diffusion is hindered and slopes are higher than 145 mV. Interestingly, for v > 30 mV s−1, the maximum current in SVs for DCO = 100 is higher than for DCO = 0 (Figure 8A−D). This is because the ef fective oxidation rate also depends on the OHads concentration, and if the oxidation is very fast, the reaction

Figure 7. Stripping voltammograms at 50 mV s−1 and repulsive interactions. Initial CO coverage 0.99 ML. (A, B) k2 = 8.234 × 10−2 s−1: MFA (purple square); DCO = 0−DOH = 0 (black up triangle); DCO = 100−DOH = 0 (red down triangle). (C) k2 = 8.234 × 10−4 s−1: MFA (olive circle); DCO = 0−DOH = 0 (pink left triangle); DCO = 100−DOH = 0 (purple right tiangle); DCO = 0−DOH = 100 (red star). DMC curves without interactions are also given (empty symbols). 2481

dx.doi.org/10.1021/jp409385k | J. Phys. Chem. C 2014, 118, 2475−2486

The Journal of Physical Chemistry C

Article

become sharper and, if DCO = 100, they are similar to MFA curves. However, if Ef is lower than Ep,OH, the transient shape is determined by DOH, despite DCO. In this case, both MFA and DMC transients are asymmetric, with a tailing in the descending part and, even when the current tends to zero, there is still a significant amount of COads adsorbed on the surface, especially if DOH ≠ 0 (see insets to Figure 10). MFA and DMC with DCO = 0 predict nearly the same log tmax vs Ef dynamics, with an upward shift for the DMCS, similar to Figure 1. Instead, if DCO = 100, curves show a mixed dynamics. At low Ef, they follow MFA predictions, while at higher Ef, all holes in the adlayer are immediately filled by OHads, leaving no room for COads migration, and transients merge the DCO = 0 curves. In the transition region, Tafel slopes are higher than 119 mV. 3.3. Theoretical and Experimental Comparison. In acidic media, the most appealing features of SVs for CO oxidation on Pt(111) are the following:10,14,77 At low coverages, when a significant fraction of the surface is COads free and available for hydrogen adsorption, Hads, a single oxidation peak is observed. As COads coverage increases, the intensity of this peak decreases and a second oxidation peak appears at higher potentials. Finally, at high coverages, where no sign of Hads is detected, only the latter, sharp and narrow peak remains. In the latter case, similar to Pt(100), the narrower the peak the greater the oxidation overpotential, with increasing fwhm at faster v and Tafel slopes of 60 and 80 mV for Pt(100) and Pt(111), respectively.77 These results were associated with the presence of two types of CO-arrangements,10,14 one easily oxidizable at low coverages, and a second one with compact COads domains, which is more difficult to oxidize, at high coverages. In addition to stripping results, temperature programmed desorption (TPD), 78 low energy electron diffraction (LEED),5,78 in situ IR spectroscopy,3,20−23,61,78 and optical second harmonic generation (SHG)78 studies have related the adsorbate adlayer structure with the possible formation of COads islands at high coverages, and the appearance of two peaks in the SVs with CO adsorbed in different surface sites. In consequence, an N&G kinetics during the oxidation,2,10−15 together with low CO surface mobility, has been assumed. In contrast, symmetric current transients, that become slightly asymmetric at lower Ef, with tailing in the descending part, during potentiostatic studies on Pt(111) and its stepped surfaces (in acidic16,17 and alkaline media41−43,79), have been interpreted as a clear indication of a LH mechanism including fast CO diffusion.8,16−19 Notice, however, that, from simulations, these transients could also be reproduced with low k2 and either moderate or low lateral interactions, regardless of DCO (attractive, Figure 6, and repulsive, Figure 9). In addition, the exponential current contribution before the main peak during potential-step experiments on Pt surfaces, reported by some authors and explained in terms of an Eley−Rideal scheme in the early stage of the process,7 can also be obtained under some conditions through a LH schema including effective lateral interactions; sees Figure 6A−C and 9. In alkaline solutions, SVs between 5 < v < 500 mV s−1 on Pt basal planes43,45,79 show a different dynamics, similar to curves depicted in Figure 8. In this case, results have been explained considering COads diffusional restrictions.37,38,42,43,45 On Pt(110), similar to Figure 8 for DCO = 0, two peaks can be differentiated in the SV at slow v, that transform into one broad peak at faster v, with a slope of 99 mV in the Ep,CO/OH vs log v curve. On Pt(111),45 SVs shown a prewave, attributed to the

inhibits itself: at faster rates, the OHads consumption is so high that its surface concentration is nearly zero. For faster v, Ep,CO/OH shifts to higher potentials where the OHads formation is faster and the oxidation rate increases. Figures 9 and 10 show normalized current transients for two k2 at repulsive interactions. For slow k2 and low ϵCO−CO, the

Figure 9. Normalized DMC current transients for k2 = 8.234 × 10−3 s−1 and ∈CO−CO = 0.02 eV, from a 0.99 CO ML. DCO = 0−DOH = 0 (red square); DCO = 0−DOH = 100 (pink circle); DCO = 100−DOH = 0 (blue up triangle); DCO = 100−DOH = 10 000 (olive down triangle). MFA transients are also given (solid line).

Figure 10. Normalized DMC potential step current transients for k2 = 8.234 × 10−2 s−1 and ϵCO−CO = 0.1 eV, from a 0.99 CO ML. DCO = 0− DOH = 0 (red square); DCO = 0−DOH = 100 (magenta circle); DCO = 100−DOH = 0 (blue up triangle); DCO = 100−DOH = 10 000 (green down triangle). MFA transients are also given (solid line). Insets: CO coverages during step current transients.

change in shape of current transients with the applied potential (Figure 9) and its potential dependence are quite similar to curves in absence of CO−CO interactions, Figures 1 and 2. However, at lower Ef and DCO = 100, there is a noticeable change and transients have an initial quadratic rise in current. In all curves, transients are tailed on their descending part. For ϵCO−CO = 0.1 eV and k2 = 8.234 × 10−2 s−1 (Figure 10), MFA predicts sharper transients than DMC. At high Ef, all transients are symmetric and DMC curves are similar, regardless of DCO or DOH. When decreasing Ef, transients 2482

dx.doi.org/10.1021/jp409385k | J. Phys. Chem. C 2014, 118, 2475−2486

The Journal of Physical Chemistry C

Article

active sites for the reaction still remains, that is, preferential surface sites where OH adsorption begins. It is worth to mention that rate constants for OH adsorption/desorption were chosen in such a way that at 20 mV s−1 OHads adsorption appears quite reversible in this work. But at the same time, they have been kept deliberately low in order to avoid long simulation times. Higher OHads adsorption/ desorption rates, as expected on Pt(111), will lead to sharper Ep,CO/OH at moderate k2. Indeed, even repulsive interactions can lead to a sharp Ep,CO/OH, although its position would be shifted to lower potentials before OHads adsorption occurs. Additionally, as seen from Figures 4 and 8, under some conditions it is possible that Ep,CO/OH appears at potentials as low as 0.2 V before Ep,OH. Therefore, experimental CO oxidation peaks on Pt stepped surfaces between the hydrogen adsorption/ desorption region and the OHads adsorption process on the terrace, do not necessary imply that the characteristic voltammetric peak in the hydrogen adsorption/desorption region of stepped surfaces correspond to OH adsorption on steps, as suggested before.87 Similarly,35 experimental results on Rh(111) and its vicinal Rh[n(111) × (111)] surfaces could be also explained in light of repulsive interactions, especially in absence of CO diffusion. Features, such as tailings, pre- and post-shoulders on voltammetric peaks,46,88 symmetric current transients, that become slightly asymmetric at lower potentials, with a tailing in its descending part, and the high amount of COads remaining after a potential step at low Ef on sulfuric acid,46,89 can be observed in Figures 7−10. Indeed, on Rh(111), ef fective repulsive interactions are expected, taking into account that average attractive CO−H2O interactions have been reported.80 Of course, this is not the entire picture and other important phenomena also play an important role. However, results here, and from previous work,35 highlight the importance of effective interactions in the reaction kinetics and suggest that they should be taken into account when interpreting experimental data. Finally, some words should be said about the role of surface diffusion on the CO electro-oxidation on Rh surfaces. Experimentally, less tailing of the main peak in perchloric than in sulfuric solutions, in chronoamperometric and voltammetric data, has been interpreted in terms of a higher COads mobility in the former solution.46,89 However, DMC transients for k2 = 8.234 × 10−2 s−1 and ϵCO−CO = 0.1 eV show that those results could also be explained considering the role of the OHads diffusion. For potentiostatic experiences, if DCO = 0 and Ef = 0.5 V, transients have more tailing with DOH = 0 than DOH = 100 (Figure 10B), and if DCO = 100 and Ef = 0.15 V, transients have more pronounced tailing with DOH = 0 than DOH = 10000 (Figure 10D). More experimental efforts are necessary to clarify this point.

presence of defects in the electrode, and a main peak with a small shoulder at low v. A slope of 67 mV has been reported. Finally, on Pt(100),43 Ep,CO/OH does not exhibit shoulder or tail; however, in this case, a detailed scan rate study is missing. Experimental SVs and current transients in acidic medium at high coverages can be described considering ef fective attractive CO−CO interactions. Sharp and narrow peaks in SVs, with an increasing fwhm at faster v, and symmetric transients are predicted from simulations, together with island formation. The difference in physical properties of adsorbed molecules may give rise to these effective interactions; that is, while H2O/OH species can form a stable network, owing to attractive interactions, CO−CO and CO−H2O interactions are repulsive.10,62,80 At low CO coverages, this balance cannot explain experimental results. It is proposed that COads and water molecules form a mixed phase in which COads and water occupy adjacent sites. As COads coverage increases, coadsorbates segregate into incompressible islands containing only water and compressible, internally repulsive, patches with COads.80 A small percentage of the surface would be covered by a mixed phase, giving rise to two different SV peaks. At high CO coverages, water and COads islands would compose the entire surface. Calculations based on DFT help to shed light on microscopic structures and processes. They indicate that chemical bonding in specific adsorption is hardly influenced by water, because of weak water-metal interactions. However, this is not necessarily true for reaction barriers in electrocatalytic reactions. Molecular simulations of water-metal interfaces at RT indicate that water molecules of the first layer at the interface remain rather localized, forming structures close to an ice-like layer.81 Instead, for CO adsorption on Pt, UHV studies reported a (√3 × √3)R30° superstructure on a triangular lattice at θ < 0.33. This structure, in which nn sites are unoccupied, indicates repulsive nn interactions and, probably, attractive next-nearest-neighbor interactions.82 It is suggested that at low coverages (below 1/ 3), CO adsorbs preferably on the “holes” of the water ice-like structure, forming a mixed CO/water phase. Higher CO coverages disrupt water network and adsorbates segregate in two phases (island formation). Different CO ads /H 2 O ads structures have been reported depending on the coverage and the potential at which CO is adsorbed.83 A full detailed simulation including all the interactions in the adlayer could clarify this point. Interactions between coadsorbates driving the reaction dynamics can explain the strong dependence on the CO adsorption potential of Ea for the CO stripping reaction.84 It could also support, at least qualitatively, higher oxidation rates on stepped surfaces than basal planes and also in alkaline than acidic solutions. For the OH−H2O interaction on Pt(111), two different OHads coverage regimes have been suggested: up to 1/ 3 ML and above 1/3 ML.57 Their origin is twofold: OH−H2O interaction is stronger than both OH−OH and H2O−H2O interactions, and hydrogen scrambling in these overlayers is facile, that is, the relaxation time for finding the most stable overlayer is fast. In alkaline solutions, weakening of the water/ OH network, because of higher OHads coverages, may change effective interactions inside the adlayer from attractive to slightly repulsive. For stepped surfaces, it is well known that defects and steps disturb the OH−H2O network.9,54 Hence, it could make effective interactions less attractive, especially for the more stepped surfaces, which affect the COads packing density, as suggested from experiments.17,39,46,85,86 In any case, the role of

4. CONCLUSIONS A lattice-gas modeling of the electrochemical CO oxidation through a LH scheme including ef fective lateral interactions between one of the adsorbates and species diffusion has been solved. The main aim was to underscore the importance of the molecular distribution and surface mobility of reacting species. Two methods, MFA and DMC have been compared. In the MFA approach, average lateral interactions, which depend on the adsorbate coverages, are used. Contrary, DMC simulations treat the distribution of the adsorbed species in an essentially exact way, provided a sufficiently large lattice. 2483

dx.doi.org/10.1021/jp409385k | J. Phys. Chem. C 2014, 118, 2475−2486

The Journal of Physical Chemistry C

Article

(3) Chang, S.-C.; Weaver, M. Coverage-Dependent Dipole Coupling for Carbon Monoxide Adsorbed at Ordered Platinum (111)-Aqueous Interfaces: Structural and Electrochemical Implications. J. Chem. Phys. 1990, 92, 4582−4594. (4) Villegas, I.; Weaver, M. J. Carbon Monoxide Adlayer Structures on Platinum (111) Electrodes: A Synergy between In Situ Scanning Tunneling Microscopy and Infrared Spectroscopy. J. Chem. Phys. 1994, 101, 1648−1660. (5) Zurawski, D.; Wasberg, M.; Wieckowksi, A. LEED and Voltammetry of Carbon Monoxide Electrosorbed on Platinum (111). J. Phys. Chem. 1990, 94, 2076−2082. (6) Gilman, S. The Mechanism of Electrochemical Oxidation of Carbon Monoxide and Methanol on Platinum. II. The “Reactant-Pair” Mechanism for Electrochemical Oxidation of Carbon Monoxide and Methanol. J. Phys. Chem. 1964, 68, 70−80. (7) Urchaga, P.; Baranton, S.; Coutanceau, C.; Jerkiewicz, G. Evidence of an Eley−Rideal Mechanism in the Stripping of a Saturation Layer of Chemisorbed CO on Platinum Nanoparticles. Langmuir 2012, 28, 13094−13104. (8) Koper, M. T. M.; Jansen, A. P. J.; Van Santen, R. A.; Lukkien, J. J.; Hilbers, P. A. J. Monte Carlo Simulations of a Simple Model for the Electrocatalytic CO Oxidation on Platinum. J. Chem. Phys. 1998, 109, 6051−6062. (9) Van der Niet, M. J. T. C.; Den Dunnen, A.; Juurlink, L. B. F.; Koper, M. T. M. CO−Adsorption of O and H2O on Nanostructured Platinum Surfaces: Does OH Form at Steps? Angew. Chem., Int. Ed. 2010, 49, 6572−6575. (10) Orts, J. M.; Louis, E.; Sander, L. M.; Feliu, J. M.; Aldaz, A.; Clavilier, J. Monte Carlo Simulation of CO Adlayers Electrooxidation on Pt(111). Surf. Sci. 1998, 416, 371−383. (11) Love, B.; Lipkowski, J. Effect of Surface Crystallography on Electrocatalytic Oxidation of Carbon Monoxide on Pt Electrodes. In Molecular Phenomena at Electrode Surfaces; Soriaga, M., Ed.; ACS Symposium Series No 378; American Chemical Society: Washington, DC, 1988; Chapter 33, pp 484−496. (12) Maillard, F.; Eikerling, M.; Cherstiouk, O. V.; Schreier, S.; Savinova, E.; Stimming, U. Size Effects on Reactivity of Pt NanoParticles in CO Monolayer Oxidation: The Role of Surface Mobility. Faraday Discuss 2004, 125, 357−377. (13) Cherstiouk, O. V.; Simonov, P. A.; Zaikovskii, V. I.; Savinova, E. R. CO Monolayer Oxidation at Pt Nanoparticles Supported on Glassy Carbon Electrodes. J. Electroanal. Chem. 2003, 554−555, 241−251. (14) Feliu, J. M.; Orts, J. M.; Femandez−Vega, A.; Aldaz, A.; Clavilier, J. Electrochemical Studies in Sulphuric Acid Solutions of Adsorbed CO on Pt (111) Electrodes. J. Electroanal. Chem. Interfacial Electrochem. 1990, 296, 191−201. (15) Orts, J. M.; Louis, E.; Sander, L. M.; Clavilier, J. Numerical Simulation of the Voltammetric Electrooxidation of CO Adsorbed on Pt(111). Electrochim. Acta 1998, 44, 1221−1227. (16) Lebedeva, N. P.; Koper, M. T. M.; Feliu, J. M.; Van Santen, R. A. Mechanism and Kinetics of the Electrochemical CO Adlayer Oxidation on Pt(111). J. Electroanal. Chem. 2002, 524−525, 242−251. (17) Lebedeva, N. P.; Koper, M. T. M.; Feliu, J. M.; Van Santen, R. A. Role of Crystalline Defects in Electrocatalysis: Mechanism and Kinetics of CO Adlayer Oxidation on stepped Platinum Electrodes. J. Phys. Chem. B 2002, 106, 12938−12947. (18) Koper, M. T. M.; Lebedeva, N. P.; Hermse, C. G. M. Dynamics of CO at the Solid/liquid Interface Studied by Modeling and Simulation of CO Electrooxidation on Pt and Pt/Ru Electrodes. Faraday Discuss. 2002, 121, 301−311. (19) Petukhov, A. V.; Akemann, W.; Friedrich, K. A.; Stimming, U. Kinetics of Electrooxidation of a CO Monolayer at the Platinum/ Electrolyte Interface. Surf. Sci. 1998, 402−404, 182−186. (20) Korzeniewski, C.; Severson, M. W. Applications of Infrared Spectroscopy in the Study of Catalytic Reactions and Related Adsorption Phenomena on Single Crystal Electrodes: Connections between Electrochemical and Ultra High Vacuum Surface Science. Spectrochim. Acta 1995, 51A, 499−518.

For low CO oxidation rates, k2, repulsive interactions between COads species may lead to multiple peaks in the voltammetry, especially if DCO is high. These peaks are a consequence of high and low packing COads density domains inside the adlayer. Each domain has a different ef fective oxidation rate. The MFA only predicts one main oxidation peak either with a shoulder or an asymmetrical tail at its positive potential side. Simulations suggest that some experimental results of the electrochemical COads oxidation, by stripping voltammetry and/or potential step chronoamperometry, on different surface planes of Pt or Rh on acidic medium may be connected with ef fective adsorbate−adsorbate interactions on the adlayer, accompanied, or not, by island formation and rapid adsorbate diffusion, rather to a low COads surface mobility, or a change in the reaction mechanism. Similarly, on alkaline medium COads oxidation on Pt(111) and its vicinal surfaces may be better described considering an increase in the inter/intraparticle balance and low DCO. In particular, values of lateral interactions on different planes may be slightly different and could depend to some extent on the potential. The island size may depend on how the adsorbed overlayer is prepared and on the reaction conditions, which influence the overall reaction kinetics. In summary, using DMC has shown that including lateral interactions into the LH mechanism reconciles different experimental results. Thus, they should be taken into account when interpreting experiments. In addition, results extend the conceptual framework for interpreting transient kinetics of COads oxidation on Pt, and in general on metallic surfaces. In spite of the high level of simplicity, the model holds great promise in the understanding of electrocatalytic activity. However, the structural features of the real system are much more complex, and it is necessary to continue refining the model.



AUTHOR INFORMATION

Corresponding Author

*Tel.: +57 4 425-5249. Fax: +57 4 425-5254. E-mail: [email protected]; [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This research has been made possible by a fellowship from the Universidad Nacional de Colombia and by a grant from COLCIENCIAS and the Universidad Nacional de Colombia in the framework of the National Program of Formation in Innovation Leaders (Contract No. 472 of 2007). Special thanks go to Professor Marc T. M. Koper for useful and interesting discussions and his comments on the manuscript. J. P. Hernández-Ortiz would like to give thanks for the constant support by NSF through the Nanoscale Science and Engineering Center (UW-NSEC) on Driven Assembly, Grant DMR0425880.



REFERENCES

(1) Zhdanov, V. P.; Kasemo, B. One of the Scenarios of Electrochemical Oxidation of CO on Single-Crystal Pt Surfaces. Surf. Sci. 2003, 545, 109−121. (2) Pozniak, B.; Mo, Y.; Scherson, D. A. The Electrochemical Oxidation of Carbon Monoxide Adsorbed on Pt(111) in Aqueous Electrolytes as Monitored by In Situ Potential Step-Second Harmonic Generation. Faraday Discuss 2002, 121, 313−322. 2484

dx.doi.org/10.1021/jp409385k | J. Phys. Chem. C 2014, 118, 2475−2486

The Journal of Physical Chemistry C

Article

(21) Korzeniewski, C. Infrared Spectroscopy in Electrochemistry: New Methods and Connections to UHV Surface Science. Crit. Rev. Anal. Chem. 1997, 27, 81−102. (22) Korzeniewski, C. Vibrational Coupling Effects on Infrared Spectra of Adsorbates on Electrodes. In Interfacial electrochemistry; Wieckowski, A, Ed.; Dekker: New York, 1999; Chapter 20, pp 345− 352. (23) Iwasita, T.; Nart, F. C. In Situ Infrared Spectroscopy at Electrochemical Interfaces. Prog. Surf. Sci. 1997, 55, 271−340. (24) Bergelin, M.; Herrero, E.; Feliu, J. M.; Wasberg, M. Oxidation of CO Adlayers on Pt(111) at Low Potentials: an Impinging Jet Study in H2SO4 Electrolyte with Mathematical Modeling of the Current Transients. J. Electroanal. Chem. 1999, 467, 74−84. (25) Petukhov, A. V. Effect of Molecular Mobility on Kinetics of an Electrochemical Langmuir−Hinshelwood Reaction. Chem. Phys. Lett. 1997, 277, 539−544. (26) Saravanan, C.; Markovic, N. M.; Head−Gordon, M.; Ross, P. N. Stripping and Bulk CO Electrooxidation at the Pt−Electrode Interface: Dynamic Monte Carlo Simulations. J. Chem. Phys. 2001, 114, 6404− 6412. (27) Armand, D.; Rosinberg, M. L. A Simple Model for the Competitive Adsorption of Anions and Hydrogen on the (100) Orientation of a Platinum Surface in Acid Medium: The 3-State Lattice−Gas Model. J. Electroanal. Chem. 1991, 302, 191−206. (28) Hermse, C. G. M.; Van Bavel, A. P.; Koper, M. T. M.; Lukkien, J. J.; Van Santen, R. A.; Jansen, A. P. J. Modelling the Butterfly: (3 × 7) Ordering on FCC(111) Surfaces. Surf. Sci. 2004, 572, 247−260. (29) Garcia-Araez, N.; Lukkien, J. J.; Koper, M. T. M.; Feliu, J. M. Competitive Adsorption of Hydrogen and Bromide on Pt(100): Mean-Field Approximation vs. Monte Carlo Simulations. J. Electroanal. Chem. 2006, 588, 1−14. (30) Farias, M. J. S.; Tanaka, A. A.; Tremiliosi, G.; Feliu, J. M. On the Apparent Lack of Preferential Site Occupancy and Electrooxidation of CO Adsorbed at Low Coverage onto Stepped Platinum Surfaces. Electrochem. Commun. 2011, 13, 338−341. (31) Gómez-Marín, A. M.; Berná, A.; Feliu, J. M. Spectroelectrochemical Studies of the Pt(111)/Nafion Interface Cast Electrode. J. Phys. Chem. C 2010, 114, 20130−20140. (32) Koper, M. T. M.; Lukkien, J. J.; Jansen, A. P. J.; Van Santen, R. A. Lattice Gas Model for CO Electrooxidation on Pt−Ru Bimetallic Surfaces. J. Phys. Chem. B 1999, 103, 5522−5529. (33) Korzeniewski, C.; Kardash, D. Use of a Dynamic Monte Carlo Simulation in the Study of Nucleation-and-Growth Models for CO Electrochemical Oxidation. J. Phys. Chem. B 2001, 105, 8663−8671. (34) Saravanan, C.; Koper, M. T. M.; Markovic, N. M.; HeadGordon, M.; Ross, P. N. Modeling Base Voltammetry and CO Electrooxidation at the Pt(111)-Electrolyte Interface: Monte Carlo Simulations Including Anion Adsorption. Phys. Chem. Chem. Phys. 2002, 4, 2660−2666. (35) Gómez-Marín, A. M.; Hernández-Ortíz, J. P. Mean Field Approximation of Langmuir−Hinshelwood CO−Surface Reactions Considering Lateral Interactions. J. Phys. Chem. C 2013, 117, 15716− 15727. (36) Hill, T. L. An Introduction to Statistical Thermodynamics; Dover: New York, 1960. (37) Herrero, E.; Chen, Q.-S.; Hernandez, J.; Sun, S.-G.; Feliu, J. M. Effects of the Surface Mobility on the Oxidation of Adsorbed CO on Platinum Electrodes in Alkaline Media. The Role of the Adlayer and Surface Defects. Phys. Chem. Chem. Phys. 2011, 13, 16762−16771. (38) Farias, M. J. S.; Herrero, E.; Feliu, J. M. Site Selectivity for CO Adsorption and Stripping on Stepped and Kinked Platinum Surfaces in Alkaline Medium. J. Phys. Chem. C 2013, 117, 2903−2913. (39) Lebedeva, N. P.; Rodes, A.; Feliu, J. M.; Koper, M. T. M.; Van Santen, R. A. Role of Crystalline Defects in Electrocatalysis: CO Adsorption and Oxidation on Stepped Platinum Electrodes as Studied by In Situ Infrared Spectroscopy. J. Phys. Chem. B 2002, 106, 9863− 9872.

(40) Kucernak, A. R.; Offer, G. J. The Role of Adsorbed Hydroxyl Species in the Electrocatalytic Carbon Monoxide Oxidation Reaction on Platinum. Phys. Chem. Chem. Phys. 2008, 10, 3699−3711. (41) García, G.; Koper, M. T. M. Dual Reactivity of Step-Bound Carbon Monoxide during Oxidation on a Stepped Platinum Electrode in Alkaline Media. J. Am. Chem. Soc. 2009, 131, 5384−5385. (42) García, G.; Koper, M. T. M. Mechanism of Electrooxidation of Carbon Monoxide on Stepped Platinum Electrodes in Alkaline Media: a Chronoamperometric and Kinetic Modelling Study. Phys. Chem. Chem. Phys. 2009, 11, 11437−11446. (43) Spendelow, J. S.; Lu, G. Q.; Kenis, P. J. A.; Wieckowski, A. Electrooxidation of Adsorbed CO on Pt(111) and Pt(111)/Ru in Alkaline Media and Comparison with Results from Acidic Media. J. Electroanal. Chem. 2004, 568, 215−224. (44) Kobayashi, T.; Babu, P. K.; Gancs, L.; Chung, J. H.; Oldfield, E.; Wieckowski, A. An NMR Determination of CO Diffusion on Platinum Electrocatalysts. J. Am. Chem. Soc. 2005, 127, 14164−14165. (45) García, G.; Koper, M. T. M. Stripping Voltammetry of Carbon Monoxide Oxidation on Stepped Platinum Single-Crystal Electrodes in Alkaline Solution. Phys. Chem. Chem. Phys. 2008, 10, 3802−3811. (46) Housmans, T. H. M.; Koper, M. T. M. CO Oxidation on Stepped Rh[n(111) × (111)] Single Crystal Electrodes: Anion Effects on CO Surface Mobility. Electrochem. Commun. 2005, 7, 581−588. (47) Koper, M. T. M.; Jansen, A. P. J.; Lukkien, J. J. Lattice-Gas Modeling of Electrochemical Langmuir−Hinshelwood Surface Reactions. Electrochim. Acta 1999, 45, 645−651. (48) Housmans, T. H. M.; Hermse, C. G. M.; Koper, M. T. M. CO Oxidation on Stepped Single Crystal Electrodes: A Dynamic Monte Carlo Study. J. Electroanal. Chem. 2007, 607, 69−82. (49) Kang, H. C.; Weinberg, W. H. Modeling the Kinetics of Heterogeneous Catalysis. Chem. Rev. 1995, 95, 667−676. (50) Zhdanov, V. P. Monte Carlo Simulations of Oscillations, Chaos and Pattern Formation in Heterogeneous Catalytic Reactions. Surf. Sci. Rep. 2002, 45, 231−326. (51) Liu, D.-J.; Evans, J. W. From Atomic Scale Ordering to Mesoscale Spatial Patterns in Surface Reactions: A Heterogeneous Coupled Lattice-Gas Simulation Approach. Model. Simul. 2005, 4−2, 424−446. (52) Jansen, A. P. J. Monte Carlo Simulations of Chemical Reactions on a Surface with Time-Dependent Reaction-Rate Constants. Comput. Phys. Commun. 1995, 86, 1−12. (53) Zhdanov, V. P.; Kasemo, B. Lattice-Gas Model of NO Decomposition on Transition Metals. J. Chem. Phys. 1996, 104, 2446−2452. (54) Björling, A.; Herrero, E.; Feliu, J. M. Electrochemical Oxidation of Pt(111) Vicinal Surfaces: Effects of Surface Structure and Specific Anion Adsorption. J. Phys. Chem. C 2011, 115, 15509−15515. (55) Conway, B. E. Electrochemical Oxide Film Formation at Noble Metals as Surface-Chemical Process. Prog. Surf. Sci. 1995, 49, 331− 452. (56) Gómez-Marín, A. M.; Clavilier, J.; Feliu, J. M. Sequential Pt(111) Oxide Formation in Perchloric Acid: An Electrochemical Study of Surface Species Inter-conversion. J. Electroanal. Chem. 2013, 688, 360−370. (57) Karlberg, G. S.; Wahnstrom, G. An Interaction Model for OH +H2O-Mixed and Pure H2O Overlayers Adsorbed on Pt(111). J. Chem. Phys. 2005, 122, 194705−1−194705−6. (58) Hellman, A.; Honkala, K. Including Lateral Interactions into Microkinetic Models of Catalytic Reactions. J. Chem. Phys. 2007, 127, 194704−1−194704−6. (59) Anderson, A. B.; Neshev, N. M. Mechanism for the Electrooxidation of Carbon Monoxide on Platinum, Including Electrode Potential Dependence: Theoretical Determination. J. Electrochem. Soc. 2002, 149, E383−E388. (60) Bard, A. J.; Faulkner, L. R. Electrochemical Methods: Fundamentals and Applications, 2nd ed.; John Wiley and Sons: New York, 2001. (61) Weaver, M. J.; Chang, S.-C.; Leung, L.-W.; Jiang, X.; Rubel, M.; Szklarczyk, M.; Zurawski, D.; Wieckowski, A. Evaluation of Absolute 2485

dx.doi.org/10.1021/jp409385k | J. Phys. Chem. C 2014, 118, 2475−2486

The Journal of Physical Chemistry C

Article

Saturation Coverages of CO on Low-Index Platinum and Rhodium Electrodes. J. Electroanal. Chem. 1992, 327, 247−260. (62) Marcovic, N. M.; Ross, P. N. Surface Science Studies of Model Fuel Cell Electrocatalysts. Surf. Sci. Rep. 2002, 45, 117−230. (63) Barth, J. V. Transport of Adsorbates at Metal Surfaces: From Thermal Migration to Hot Precursors. Surf. Sci. Rep. 2000, 40, 75− 149. (64) Seebauer, E. G.; Allen, C. E. Estimating Surface Diffusion Coefficients. Prog. Surf. Sci. 1995, 49, 265−230. (65) Lukkien, J. J.; Segers, J. P. L.; Hilbers, P. A. J.; Gelten, R. J.; Jansen, A. P. J. Efficient Monte Carlo Methods for the Simulation of Catalytic Surface Reactions. Phys. Rev. E 1998, 58, 2598−2610. (66) CARLOS website: http://www.win.tue.nl/∼johanl/projects/ Carlos/ and http://www.win.tue.nl/∼johanl/projects/nCarlos/. Last visit: 24 May 2013. (67) Bewick, A.; Fleischmann, M.; Thirsk, H. R. Kinetics of the Electrocrystallization of Thin Films of Calomel. Trans. Faraday Soc. 1962, 58, 2200−2216. (68) Schmickler, W. Interfacial Electrochemistry; Oxford University Press: New York, 1996. (69) Avrami, M. Kinetics of Phase Change. I General Theory. J. Chem. Phys. 1939, 7, 1103−1112. (70) Avrami, M. Kinetics of Phase Change. II Transformation-Time Relations for Random Distribution of Nuclei. J. Chem. Phys. 1940, 8, 212−224. (71) Avrami, M. Granulation, Phase Change, and Microstructure Kinetics of Phase Change. III. J. Chem. Phys. 1941, 9, 177−182. (72) Frank, S.; Roberts, D. E.; Rikvold, P. A. Effects of Lateral Diffusion on Morphology and Dynamics of a Microscopic Lattice-Gas Model of Pulsed Electrodeposition. J. Chem. Phys. 2005, 122, 064705− 1−064705−10. (73) Frank, S.; Rikvold, P. A. Kinetic Monte Carlo Simulations of Electrodeposition: Crossover from Continuous to Instantaneous Homogeneous Nucleation within Avrami’s Law. Surf. Sci. 2006, 600, 2470−2487. (74) Santos, E.; Leiva, E. P. M.; Vielstich, W. CO Adsorbate on Pt(111) Single Crystal Surfaces. Electrochim. Acta 1991, 36, 555−561. (75) Herrero, E.; Feliu, J. M.; Blais, S.; Radovic-Hrapovic, Z.; Jerkiewicz, G. Temperature Dependence of CO Chemisorption and its Oxidative Desorption on the Pt(111) Electrode. Langmuir 2000, 16, 4779−4783. (76) Zhu, Y.; Uchida, H.; Watanabe, M. Oxidation of Carbon Monoxide at a Platinum Film Electrode Studied by Fourier Transform Infrared Spectroscopy with Attenuated Total Reflection Technique. Langmuir 1999, 15, 8757−8764. (77) Palaikis, L.; Zurawski, D.; Hourani, M.; Wieckowski, A. Surface Electrochemistry of Carbon Monoxide Adsorbed from Electrolytic Solutions at Single Crystal Surfaces of Pt(111) and Pt(l00). Surf. Sci. 1988, 199, 183−198. (78) Akemann, W.; Friedrich, K. A.; Stimming, U. Potentialdependence of CO Adlayer Structures on Pt(111) Electrodes in Acid Solution: Evidence for a Site Selective Charge Transfer. J. Chem. Phys. 2000, 113, 6864−6874. (79) Spendelow, J. S.; Goodpaster, J. D.; Kenis, P. J. A.; Wieckowski, A. Mechanism of CO Oxidation on Pt(111) in Alkaline Media. J. Phys. Chem. B 2006, 110, 9545−9555. (80) Wagner, F. T.; Moylan, T. E.; Schmieg, S. J. Hydrophilic versus Hydrophobic CO-Adsorption: Carbon Monoxide and Water on Rh(111) versus Pt(111). Surf. Sci. 1988, 195, 403−428. (81) Gohda, Y.; Schnur, S.; Groß, A. Influence of Water on Elementary Reaction Steps in Electrocatalysis. Faraday Discuss 2008, 140, 233−244. (82) Petrova, N. V.; Yakovkin, I. N. Lateral Interaction and CO Adlayer Structures on the Pt(111) surface. Surf. Sci. 2002, 519, 90− 100. (83) Yan, Y.-G.; Yang, Y.-Y.; Peng, B.; Malkhandi, S.; Bund, A.; Stimming, U.; Cai, W.-B. Study of CO Oxidation on Polycrystalline Pt Electrodes in Acidic Solution by ATR-SEIRAS. J. Phys. Chem. C 2011, 115, 16378−16388.

(84) Rincón, A.; Pérez, M. C.; Cuesta, A.; Gutiérrez, C. Dependence on the CO Admission Potential of the Activation Energy of the Electrooxidation of Adsorbed CO on Pt. Electrochem. Commun. 2005, 7, 1027−1032. (85) Lebedeva, N. P.; Koper, M. T. M.; Herrero, E.; Feliu, J. M.; Van Santen, R. A. Cooxidation on Stepped Pt[n(111) × (111)] Electrodes. J. Electroanal. Chem. 2000, 487, 37−44. (86) Lebedeva, N. P.; Koper, M. T. M.; Feliu, J. M.; Van Santen, R. A. The Effect of the Cooling Atmosphere in the Preparation of FlameAnnealed Pt(111) Electrodes on CO Adlayer Oxidation. Electrochem. Commun. 2000, 2, 487−490. (87) Van der Niet, M. J. T. C.; Garcia-Araez, N.; Hernandez, J.; Feliu, J. M.; Koper, M. T. M. Water Dissociation on Well-Defined Platinum Surfaces: The Electrochemical Perspective. Catal. Today 2013, 202, 105−113. (88) Housmans, T. H. M.; Feliu, J. M.; Koper, M. T. M. CO Oxidation on Stepped Rh[n(111)x(111)] Single Crystal Electrodes: A Voltammetric Study. J. Electroanal. Chem. 2004, 572, 79−91. (89) Housmans, T. H. M.; Koper, M. T. M. CO Oxidation on Stepped Rh[n(111) × (111)] Single Crystal Electrodes: a Chronoamperometric Study. J. Electroanal. Chem. 2005, 575, 39−51.

2486

dx.doi.org/10.1021/jp409385k | J. Phys. Chem. C 2014, 118, 2475−2486