Divalent pseudoatoms for modeling Si(100) surfaces

Abstract

An accurate first-principles treatment of complex systems, such as surfaces, continues to be a major challenge in computational chemistry. A popular approach to treat such systems is the use of cluster models, where a moderately sized model system is constructed by excising a cluster from the extended surface. This requires cutting chemical bonds, creating dangling bonds on the cluster boundary atoms that can introduce unphysical errors. Pseudobond, pseudoatom, and quantum capping potential approaches have been developed to treat such systems using a boundary "design-atom" subject to an appropriately fitted effective potential. However, previous approaches have been developed only for truncation of a single covalent bond. They may not be adequate for many important problems involving surface chemistry or materials chemistry, where multiple covalent bonds are severed between layers. In this paper, we have extended the pseudoatom formulation for divalent silicon, which can be employed to describe accurate Si(100) surface chemistry. The effective core potential parameters of our pseudoatom are obtained by fitting to geometrical parameters and atomic charges of molecules containing Si-Si and Si-O bonds, making our pseudoatom robust for applicability in different bonding environments. We calibrate the performance of our pseudoatom approach in small molecules and surface models, and also discuss its ability to describe heteroatomic bonds using multiple theoretical methods.