No abstract

Density-functional tight-binding methods stand out as a very good compromise between accuracy and computational efficiency. These methods rely on parameter sets that have to be determined and tabulated for every pair of chemical elements. We describe an efficient, and to a large extent automatic, procedure to build such parameter sets. This procedure includes the generation of unbiased training sets and subsequent optimization of the parameters using a pattern search method. As target for the optimization we ask that the formation energy and the forces on the atoms calculated within tight-binding reproduce the ones obtained using density-functional theory. We then use this approach to calculate parameter sets for group IV elements and their binaries. These turn out to yield substantially better results than previously available parameters, especially in what concerns energies and forces.

We present a practical procedure to obtain reliable and unbiased neural network based force fields for solids.

We develop a procedure for the investigation of the phase diagram of materials under strain. This is based on a global structural prediction method where the volume is constrained to predefined values. Our method is more general than other available techniques, and it avoids at the same time numerical instabilities. As a first example, we investigate the phase diagram of two-dimensional carbon as a function of the area per atom. As expected, we find that graphene is stable for a large range of biaxial strains. However, at large areas there appear novel carbon allotropes containing decagons and higher order polygons. These phases are thermodynamically stable for strains below the breaking point of graphene, indicating that they could be accessible experimentally.