NEW IMPLEMENTATION OF THE EXCHANGE-HOLE DIPOLE MOMENT DISPERSION METHOD FOR LARGE-SCALE MATERIALS SIMULATIONS

Abstract

Intermolecular interactions are the forces that exist between molecules and are fundamental to many aspects of chemistry and the natural sciences, including the determination of the phases of matter. A correct theoretical description of intermolecular interactions at the quantum mechanical level is required for quantitative understanding of chemistry. While highly accurate wavefunction-based methods exist, they are impractical or impossible to apply to the large molecular systems relevant to most areas of chemistry, making dispersion-corrected density-functional theory (DFT) the standard class of methods for such applications. This thesis first characterised the shortcomings and laid out the requirements for high accuracy within dispersion-corrected DFT approaches, and highlighted some advantages of the exchange-hole dipole moment (XDM) dispersion method. This thesis then focused on the implementation of XDM within the Fritz Haber Institute ab initio materials simulations package (FHI-aims), which allowed for the first time the routine use of XDM-corrected hybrid functionals for the study of molecular solids. Using a selection of common molecular and solid-state benchmarks as test systems allowed for the validation of the XDM model within FHI-aims. Specifically, XDM-corrected hybrid functionals were shown to yield unprecedented accuracy for predicting energies of forming crystals from their constituent molecules, as well as remarkable predictive capacity for stable crystal packing motifs, or polymorphs, of organic molecules in the solid state. Finally, these methods were then applied to the compounds forming the 7th blind test of first-principles molecular crystal structure prediction. The implementation of XDM within the FHI-aims package presented within this work now allows for the accurate and efficient computational study of a new and interesting range of important chemical compounds and materials.