Quantum Monte Carlo simulation of carbon monoxide reactivity when adsorbed at metal and oxide catalyst surfaces: Trial wave‐functions with exponential type basis and quasi‐exact three‐body correlation

Abstract

AbstractReview publications show the advantages of using quantum Monte Carlo (QMC) methods when bonding interactions are modified and consequently, where the electron correlation energy varies and needs to be evaluated accurately. This article considers a model efficient metal catalyst for reactions of carbon monoxide. The ultimate aim is to help design industrially useful catalysts, since the reaction with water produces hydrogen gas selectively, that is, a clean fuel and a sustainable energy source. To narrow the gap between the industrial process and calculation, a model of platinum/oxide is studied. It is mimicked here by copper with and without adsorbed oxygen. Periodicity as well as a validated pseudo‐potential are used to limit the number of active electrons considered. A suitable code to carry out this task is CASINO. Furthermore, this code scales linearly to 100,000 processors and possesses a shared memory facility so that it is well‐suited to runs on the Blugene/P. It has been in production on such computers since early in 2011. A mixed plane‐wave/exponential type orbitals (ETO) basis is used for the systems studied. Plane waves are well‐suited to periodic solid substrates and a linear combination of ETOs for molecules. The atomic orbitals have direct physical interpretation, i.e., Coulomb Sturmians and hydrogen‐like orbitals. Their radial nodes are shown to be essential in obtaining the vitally important accurate QMC trial wave‐functions. Until 2008, ETO products on different atoms were difficult to manipulate for the evaluation of two‐electron integrals. Coulomb resolutions provide an excellent approximation that reduces these integrals to a sum of one‐electron overlap‐like integral products that each involve orbitals on at most two centers. They are thus readily evaluated. Only these integrals need to be re‐evaluated to change basis functions. In this article, QMC‐VMC variational optimisation is used with a quasi analytic two‐electron and nucleus correlation factor. The QMC diffusion Monte Carlo methodology is applied to adsorbed carbon monoxide and some of its reactions. © 2012 Wiley Periodicals, Inc.