ARTICLE pubs.acs.org/JPCC
Oxygen-Assisted Water Dissociation on Metal Surfaces: Kinetics and Quantum Effects Ernst D. German* Department of Chemical Engineering, Technion - Israel Institute of Technology, Haifa 32000, Israel
Moshe Sheintuch Department of Chemical Engineering, Technion - Israel Institute of Technology, Haifa 32000, Israel ABSTRACT: A model of oxygen-assisted water dissociation reaction (OWD, H2O þ O f 2OH), based on a tunnel mechanism of H transfer, is presented and analyzed to show that the corresponding activation energies are much lower than those of water dissociation on a clean surface. The Morse and 00 potentials are used to describe the initial and the PoschlTeller final energy wells. The reaction probability is analyzed in the framework of the nonadiabatic theory that also allows considering the hydrogen transfer in the case of a strong electron coupling. It is shown that the main contribution to the rate constant is due to quantum transition between the ground vibrational levels of the H atom in the initial and final potential wells. Numerical analysis of the rate constants and kinetic isotope effects is performed for the OWD proceeding on platinum, copper, nickel, and rhodium metal surfaces.
1. INTRODUCTION The catalytic water dissociation reaction (OWD) is an elementary step of several commercially important processes like the watergas shift or steam reforming reactions, as well as in the emerging technology of photocatalytic water splitting. It was the subject of many experimental and theoretical works (refs 1 and 2 and references therein), suggesting that it may proceed on a clean metal surface H2 O f H þ OH
ð1Þ
as well as on an oxygen preadsorbed metal surface H2 O þ O f OH þ OH
ð2Þ
The latter reaction is expected to proceed with lower activation energy, and moreover, since OH can split further to O þ H, it is an autocatalytic process. Most theoretical studies of reactions 1 and 2 calculate the minimum energy paths using DFT methods by considering both the gradual extension of the HO bond of a water molecule that has to be broken and the changing water molecule orientation to reach the transition state, while the preexponential factor of the rate constant is assumed to be equal to its classical value kBT/h. In our previous work2 we have calculated the rate constants of reaction 1 proceeding on several clean metal surfaces using an approach based on the approximate construction of an adiabatic potential energy surfaces (APES) which, in turn, depends on several reaction coordinates (including the H atom coordinate) r 2011 American Chemical Society
from which the saddle point coordinates can be determined analytically. In this paper, the kinetics of reaction 2 is considered. We apply the nonadiabatic theory3,4 developed by Dogonadze, Kuznetsov, and coauthors for describing chemical reactions in condensed media, particularly for proton transfer (PT) reactions. A similar approach was also used later by Cukier who considered PT, proton-coupled electron transfer (PCET), and hydrogen atom transfer (HAT) reactions.5 In the PCET and HAT approaches for the reaction AH þ B f A þ HB, one electron of the AH bond, which has to be broken, is transferred “together” with the proton to the neighboring acceptor B atom to form the final BH bond. Further extension of the PCET and HAT theory was performed by Soudakov with coauthors (refs 68 and references therein) and Georgievskii and Stuchebrukhov;9 the latter authors obtained an accurate expression for a tunneling matrix element which is correct for small and large values of the electronic coupling. According to refs 3 and 4, the proton coordinate is in fact identified with the reaction coordinate, while the reactants and media molecules are fixed in their position. The motion along this coordinate has a quantum character. The rate constant is calculated by considering quantum transitions between different vibration states of the proton in donor and acceptor molecules with subsequent averaging over the initial state distribution and summation over the final vibrational states. We apply these Received: January 16, 2011 Revised: April 10, 2011 Published: April 29, 2011 10063
dx.doi.org/10.1021/jp200457h | J. Phys. Chem. C 2011, 115, 10063–10072
The Journal of Physical Chemistry C
ARTICLE
Table 1. Geometric Characteristics of Adsorbed Reactants distancesa
Pt
Cu
Rh
Ni
h(H2O)
2.37
2.34
2.31
2.15
h(O)
1.28
1.28
1.25
1.16
r(OH)
0.98
0.98
0.98
0.98
L
1.56
1.47
1.51
1.44
t
0.58
0.49
0.53
0.46
a
In Å. See Figure 1 for definitions of the distances; h(H2O), h(O), and r(OH) values were adopted from references cited in 1 and 2, and the value of the geometric (maximal) tunneling √ distance t was estimated using the hollow radius L = d(M M)/ 3 on the (111) metal lattice (where d(MM) distances were taken from ref 12) and the OH bond length r: t = L r(OH).
Figure 1. (a) Geometrical conceptual arrangement of reactants on a surface. (b) Potential wells for the H atom in the initial and final states. (c) Schematic representation of the top view of the adsorbed water molecule and the neighboring metal atoms of the (111) lattice. t is the distance between the initial and final positions of the H atom.
concepts to reaction 2 which in principle is not different from the usual PT or PCET reactions; distinctions are rather of quantitative or of terminological character. The set of catalyst metal atom oscillators plays the role of a classical subsystem in reaction 2. This approach explains several features of the system, particularly why reaction 2 is characterized by the lower activation energy when compared with reaction 1. In Section 2 we consider our model of reaction 2. A description of the H transfer kinetics in terms of this model is presented in Section 3. Results of the numerical calculations of the activation energies and the tunneling effect for the reaction proceeding on various metals are outlined and discussed in Section 4.
2. REACTION MODEL AND PHYSICAL MECHANISM OF H ATOM TRANSFER We start with the geometrical picture of the adsorbed reactants to explain our model and to use this information to approximately draw the potential field from which the transition state and the corresponding energetic barrier are calculated. It is assumed that a H2O molecule behaves like a quasi-diatomic molecule AH, with the center of its mass lying at a distance h from a surface. This assumption involves neglecting the internal structure of the OH fragment and replacing it with a hypothetical atom A of equivalent mass. Then, due to the discrete character of adsorption on a catalyst surface with a regular structure, we consider that the adsorbed reactants (H2O and O) are located at neighboring sites with a minimum energy, and the equilibrium distance L between the reactants in this precursor complex is determined by the specific lattice structure (Figure 1).
Because of the favorable geometry of the coadsorbed reactants, reaction 2 is considered to proceed by direct H tunneling from the H2O to the neighboring oxygen along a line coinciding with the initial OH bond parallel to a surface. This model neglects motion of the reactants in a perpendicular direction. We justify this linear model as follows. According to published DFT calculations (see references in paper 2), a H2O molecule prefers to be adsorbed at the on-top site of the (111) surfaces of platinum, copper, nickel, and rhodium, with the O atom lying over the metal atom, while the H2O plane is located almost parallel to the metal surface. The oxygen atom prefers to be adsorbed at the hollow position on the surface. A typical geometric arrangement of reactants in a precursor complex on a (111) surface based on DFT calculations for four metals (Pt, Ni, Rh, and Cu) is presented in Figure 1a. The quantum chemical computations mentioned above show that the distance of the oxygen atom of an adsorbed molecule from the metal surface, h(H2O), is approximately equal to the sum of the adsorbed oxygen atom distance from the surface, h(O), and of the OH bond length, r(OH), to be formed (see Table 1), i.e. h(H2O) ≈ h(O) þ r(OH). A possible mismatch between the h(H2O) and h(O) þ r(OH) distances is only of about an order of magnitude of the zero vibration amplitudes of the H2O and O in the perpendicular direction, and it is within the accuracy of experimental methods. This fact allows approximately considering the initial and final position of the transferred H atom to be located along a line parallel to a surface. The simplification is not a restriction of our approach since the model may be extended by introducing an additional coordinate perpendicular to a surface to describe matching the distances h(H2O) and h(O) þ r(OH) leading to the arrangement shown in Figure 1a. Because the perpendicular vibrational frequencies are approximately the same for both reactants (∼