A Genetic Algorithm for Structure Prediction of Magnetic Materials

When considering global optimisation of magnetic crystal structures, it is important to consider both the atomic and spin degrees of freedom. This thesis presents a novel genetic algorithm for simultaneously optimising the magnetic and crystal structures of materials. The algorithm was first tested on a new magnetic interatomic potential presented in the thesis, and was shown to be capable of finding the correct atomic and magnetic structure. The algorithm was then used to study mixing the NiO(111)/MgO(111) interface, where the process behind the mixing was unknown. Results from this study suggest that mixing is driven by energetics of the system, rather than kinetic processes. Finally, the interface between the Heusler alloy CFAS and n-doped Ge, where experimental observations suggested an unknown interface phase, was studied. This work proposed the half Heusler structure for this phase, and predicted this to have unfavourable electronic properties.