Interface Structure Prediction Via CALYPSO Method

The atomistic structures of solid-solid interfaces are of fundamental interests for understanding physical properties of interfacial materials. However, determination of interface structures faces a substantial challenge, both experimentally and theoretically. Here, we propose an efficient method for predicting interface structures via the generalization of our in-house developed CALYPSO method for structure prediction. We devised a lattice match toolkit that allows us to automatically search for the optimal lattice-matched superlattice for construction of the interface structures. In addition, bonding constraints (e.g., constraints on interatomic distances and coordination numbers of atoms) are imposed to generate better starting interface structures by taking advantages of the known bonding environment derived from the stable bulk phases. The interface structures evolve by following interfacially confined swarm intelligence algorithm, which is known to be efficient for exploration of potential energy surface. The method was validated by correctly predicting a number of known interface structures with only given information of two parent solids. The application of the developed method leads to prediction of two unknown grain boundary (GB) structures (r-GB and p-GB) of rutile TiO Σ5(2 1 0) under an O reducing atmosphere that contained Ti as the result of O defects. Further calculations revealed that the intrinsic band gap of p-GB is reduced to 0.7 eV owing to substantial broadening of the Ti-3d interfacial levels from Ti centers. Our results demonstrated that introduction of grain boundaries is an effective strategy to engineer the electronic properties and thus enhance the visible-light photoactivity of TiO.