Metz, Michael P.2022-10-102022-10-102020https://udspace.udel.edu/handle/19716/31456Modern electronic structure methods are able to describe molecular systems accurately enough to predict most observed phenomena. Unfortunately, the compu- tational cost of such methods becomes prohibitively high as the size of the system grows. This problem can be alleviated by the use of relatively simple models which are fitted to data from high-level calculations. By maintaining the predictive accuracy of methods based on first principles at a computational cost which is many orders of magnitude smaller, these intermolecular potential energy surfaces (PESs) provide an important bridge to many applications in chemistry and materials science. Despite their importance and widespread use, these PESs are typically constructed in an ad hoc manner. The focus of this work is to generalize and improve upon existing method- ology for the generation of intermolecular PESs. These advancements are consolidated into a software package called autoPES, which is designed to be sufficiently automated that it is accessible to researchers with only a basic understanding of the underlying theory. The validity of this approach is demonstrated by applying it to predict the crystal structures of various organic molecules, and to compute accurate vibrational energies of the water, methane, and water-methane dimers. Finally, we extend the PES generation methodology to handle the difficult case where the monomers are allowed to deform substantially from their gas phase geometries.CrystalIntermolecularPotentialThesis1347115595https://doi.org/10.58088/hs7q-z3132022-08-11en