Date of Award

1-3-2003

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

First Advisor

Mark R Hoffman

Abstract

While electronic structure methods that are able to describe accurately excited states exist (e.g., MRCI), they are often prohibitively expensive for all but the smallest systems (i.e., up to three atom systems with a moderate-sized basis set). This is because the inclusion of dynamic electron correlation, which is necessary for obtaining accurate energies and often required for obtaining reliable geometries, is computationally demanding. Since reactions typically involve more than three atoms, techniques able to describe excited states and include dynamic correlation in a much more computationally reasonable fashion than MRCI are highly desirable. Multi-reference perturbation theories (MRPT) are able to account for dynamic correlation at a fraction of the cost of MRCI. However, only a limited number of studies have utilized MRPT. This dissertation focuses on applications involving one variant of MRPT, namely second-order generalized Van Vleck perturbation theory (GW PT2). The current implementation of G W PT 2 utilizes a number of theoretical and computational techniques that make it superior to other versions of MRPT. G W PT 2 calculations were performed for two important atmospheric systems in which excited state surfaces had been studied experimentally. One system, CIO, has been studied thoroughly by both experimental and theoretical techniques. G W P T 2 calculations were performed to determine accuracy and applicability of the method. Results suggest that G W PT 2. sufficiently describes dynamic correlation and could be used to study larger systems where MRCI is prohibited by computational expense. The other system, S2O, has received much less theoretical analysis. Thus, G W PT 2 calculations were performed to determine the nature of the excited state surfaces of S2O. Our results offer the first qualitatively complete description of the excited state potential energy surfaces near the ground state’s geometrical structure. They also describe semi-quantitatively the ground and excited state equilibrium structures. Given the usefulness of the GVVPT2 method in describing excited states, a formulation for analytic energy gradients was proposed and implemented. The formalism is based on a similar approach to calculating MRCI energy gradients. Thus, a number of tools that will prove to be useful for studying large regions of electronically excited potential energy surfaces has been developed and tested.

Share

COinS