A fast method for predicting the formation of crystal interfaces and heterocrystals

Abstract

In order to address an important problem in Computational Materials Design, we have demonstrated the feasibility of an algorithm which rapidly scans through various surface configurations of two single crystals selected from a crystal structures database and identifies those pairs which are likely to form stable heterointerfaces. Any two crystals are cut along different planes and all possible heterocrystal interfaces are generated based on geometric criteria and predicted bond directions of atoms on both constituent surfaces. Each configuration is assigned two scores derived using deviations of interfacial bond lengths from ideal and electronegativity differences between atoms on each side of the interface. We present some results to illustrate our method for PtNi3 on Pt3Co, GaP on Si, and Si on SiO2. This technique can be used as a fast filter for further analysis by more detailed ab initio-based methods, and is meant to address the higher throughput methods needed in the Materials Genome Initiative for complex material structures.