Second-principles methods for large-scale simulations of realistic functional oxides

The application of Condensed Matter theory via simulation has been over the last decades a solid approach to research in Materials Science. In particular for the case of Perovskite materials the research has been extensive, and customarily (but not only) performed using Density Functional Theory. The collective effort to develop lighter simulation techniques and exploring different theoretical approaches to computationally study materials has provided the scientific community with the possibility to strengthen the interaction between experimental and theoretical research. However, the access to large-scale simulations is still nowadays limited due to the high computational cost of such simulations. In 2013 J. C. Wojdel et al. presented a theory of modelization of crystals known as second-principles models, and which are the central point of the development of my work.
In this Thesis I develop in depth a novel methodology to produce second-principles models efficiently and in a quasi-automatic way from Density Functional Theory data. The scheme presented here identifies, given a set of reliable data to be fit, the most relevant atomic couplings of a system. The fitting process that I present is also analytical, which translates into a fast and accurate model production. I also explore the modelization of chemically inhomogeneous or nanostructured systems using second-principles models. Moreover, I present a heuristic procedure to produce models of the inhomogeneous material which is efficient and sound. Finally, I also show examples of complex problems that can be tackled thanks to the second-principles models, such as the character of 180º anti-phase domain walls in SrTiO3, thermodynamical studies of heat transport across 180º domain walls in PbTiO3 and the reproduction of experimentally-observed polarization vortices in (PbTiO3)n/(SrTiO3)n superlattices.