Investigating chemical reactions using computer simulations

Although proton-transfer reactions are ubiquitous in chemistry and biology, their theoretical study is limited by number of difficulties. Some of the practical challenges involving the study of proton-transfer reactions include the fact that accurate descriptions of hydrogen-bonded systems require time-consuming ab initio electronic structure methods. Computer time limitations become particularly relevant when investigating “rare events” such as chemical reactions, especially when the reactions are accompanied by substantial differences in the structure of the solvent. In addition, the proton is the lightest nucleus and nuclear quantum effects must be described adequately in proton-transfer reactions.; The problems mentioned above are addressed by proposing methodological improvements to calculate reaction rates for proton-transfer reactions. A new ab initio Monte Carlo algorithm is proposed in which a much faster molecular mechanics potential “guides” the slow ab initio electronic structure simulation resulting in significantly more efficient sampling schemes compared to the more familiar ab initio molecular dynamics methods. The molecular mechanics guided sampling method is generalized to incorporate nuclear quantum effects via centroid transition state theory and solvent effects via hybrid quantum mechanics-molecular mechanics methods.; The use of the molecular mechanics based importance function method is exemplified in a study of model intra-molecular proton-transfer reactions which investigates how molecular mechanics potentials should be created in order to predict kinetic isotope effects and reaction mechanisms correctly, and to understand the conditions in which atoms other than the hydrogen atom should be treated quantum-mechanically.