Development of atom-centered potentials for efficient and accurate electronic structure modeling of large molecular systems

Accurate quantum mechanical (QM) modeling of large molecular systems is computationally challenging due to the dramatic increase in the demand for computational resources with increasing system size. To tackle this problem, atom-centered potentials (ACPs) were developed to mitigate the errors of Hartree–Fock (HF) and density-functional theory (DFT) methods, particularly when used with small basis sets. The objective behind developing ACPs for such methods was to improve their accuracy in predicting various molecular properties without impacting their low computational cost. ACPs are optimized one-electron Gaussian-type functions that share the same mathematical form as generally used effective-core potentials, except they do not replace any electrons, making them easily usable with many quantum chemistry software packages. Besides, ACPs allow for a convenient means to simultaneously correct the absence of correlation (or deficiencies in exchange-correlation density functionals), basis set incompleteness, and other shortcomings in HF or DFT methods with small basis sets. The overall research conducted for this dissertation demonstrates the gradual transition from the development of proof-of-concept ACPs to final ACPs with more general applicability. In particular, the final ACPs are presented for ten elements in the first and second rows (H, B, C, N, O, F, Si, P, S, Cl), extending the applicability to various organic and biochemical molecules. The ACP-corrected methods have been shown to predict the target molecular properties with slightly less accuracy than very computationally expensive QM methods but at a much lower computational cost. It is anticipated that the methods presented in this dissertation will assist in applications such as supramolecular host-guest complexation, enzymatic catalysis, drug-target binding, protein folding, and others. This dissertation also contributes towards filling the gap in the literature regarding benchmark data sets by presenting new diverse data sets of molecular properties such as polypeptide conformational energies (PEPCONF), bond separation energies (BSE49), barrier height energies (BH9), and reaction energies (BH9-RE). These data sets have been generated using a significant amount of manual and computational effort to address the need for reference data in the ACP development process and other applications.Science, Irving K. Barber Faculty of (Okanagan)Chemistry, Department of (Okanagan)Graduat