---

A11: Predicting the atomic structure of oxide interfaces

Our main objective in this project is to develop a formalism that allows, in an completely unbiased way, for the prediction of ground-state geometries of interfaces and of its low-energy defects, and to use this formalism to predict the structure of certain oxide interfaces. The key ingredients of this formalism are state-of-the-art global structural prediction methods. These methods were developed to find the lowest energy crystal structure given a certain chemical composition of the system. We will be using the minima hopping method which was designed to explore the low-enthalpy phases of materials. Within this method, the system is moved from one configuration to the next by performing consecutive molecular dynamics escape steps and geometry relaxations. In order to extend this method to interfaces, we will use supercell geometries where the relative positions of the atoms of the “bulk” materials are frozen, and where the atoms close to the interface are allowed to reconstruct in order to minimize their energy. We note that the output of the minima-hopping method is not only the ground-state geometry, but also several of the low-lying configurations. In this context, these low-lying states correspond to defects at the interface.

With this new method, we will start by studying perovskite grain boundaries and oxide-oxide heterostructures such as LaAlO₃/SrTiO₃. In this particular system, various experiments show that atomic reconstructions, such as cation intermixtures and oxygen vacancies, do exist at the interfaces. With our simulations we will be able to give a comprehensive survey of all possible low-energy geometries of the interface and of their influence on the electronic properties of the system. We will then turn to structure of oxides deposited on metals. In particular, we plan to investigate the intriguing quasi-crystal phase recently found in BaTiO₃ deposited on Pt(111). It is true that a quasi-crystal can not be represented exactly in a computer, but large supercells are enough to obtain the so-called approximant structures. Moreover, we will also be interested in performing structural prediction runs for smaller supercells, as we expect the appearance of competing (crystalline) phases with 3- and 4-fold symmetry with similar energy. We hope that these studies will allow us to extract simple rules that enable us to predict the emergence of 2D quasicrystals in other oxide layers.

Principal Investigators

Up