Explorations in solvaton models of solvation within a quantum chemical framework

Abstract

The purpose of this thesis is to investigate the potential of developing a tractable model for solvation of large systems within a quantum chemical framework. Candidate computational methods are reviewed ranging from ab initio quantum mechanics to the more straightforward force field methods. Existing methods for modelling solvation are also reviewed. Based on the dual criteria of reasonably capturing the central physical features of solvation and retaining a computationally tractable method, the Virtual Charge Model (VCM), a continuum model, is selected for development within the CNDO framework. The strengths and weaknesses of the VCM are addressed and its shortcomings are found to be due to the use of solvatons (virtual charges) to simultaneously effect both polarisation and dielectric attenuation. The apparent success of the VCM in predicting relative protonation energies of solvated amines is also found to be fortuitous, and due to the cancellation of large errors. Removing this physically unreasonable coupling, while retaining physically meaningful results, proves to be a formidable challenge. On the basis of the analysis of the VCM behaviour, a new solvation model, the Virtual Charge Model with Explicit Attenuation (VCMX) is proposed. The complex thermodynamic behaviour of substituents on amine basicities and alcohol acidities are used as strict benchmark calculations. Preliminary studies on some other systems of biological interest, glycine, the glycyl dipeptide and isoGuanine, are presented. The method requires refinement, but the inherent viability of the VCMX is demonstrated.