Embedded cluster model for chemisorption using density functional calculations: Oxygen adsorption on the Al(100) surface

An embedded cluster model to study chemisorption on metal surfaces is presented. The approach is based on a method proposed by Whitten and Pakkanen [Phys. Rev. B 21, 4357 (1980)], which constructs an embedding potential using a set of molecular orbitals (MO) obtained for a large cluster considered appropriate to describe the adsorption site correctly. The embedded cluster approach combined with the Kohn–Sham density functional theory (KS-DFT) formalism has special features that are discussed in this work. There are several possible approximations that can be explored in practical implementations: the localization procedure used to localize the KS MOs on the surface, the choice of the cluster, the way the embedding potential is constructed and the possibility to use fractional MO occupancies and even fractional charges in order to set up the embedded cluster, are some of the points that are discussed. We have implemented this approach in a modified version of deMon-KS. Results are presented for oxygen adsorption on the Al(100) surface. The embedding potential is constructed from the MOs obtained for an Al70 cluster. The atop, bridge and four-fold sites are described by the Al10, Al20 and Al21 clusters, respectively. The bare cluster results are also presented and compared to those for the embedded cluster. The embedded cluster calculations are in better agreement with the available experimental results. The four-fold site is the most favored. The oxygen atom is found to be in the surface plane, while for the bare cluster model the oxygen lies 0.5 Å above the surface. The local density of states of the oxygen at the four-fold site of Al70, and the bare and embedded clusters have been evaluated and compared with the available UPS data.