Yiqin Gao

Computer simulations of chemical reactions in solution

We developed a sampling strategy for rare events to generate chemical reaction trajectories using which a statistical investigation can be performed on the thermodynamics, kinetics, dynamics and mechanisms of the chemical reactions in solution. Free of predefined CVs or RCs, the post-analysis of reaction mechanisms can be performed. As an example, the reaction coordinate(s) of a (retro-)Claisen rearrangement in bulk water was variationally optimized based on a Bayesian learning algorithm as well as a machine learning method. It was found that the hydrogen-bonding of water molecule to the charge-enriched site of the reactant state changes the dynamics of reaction. A characteristic shrinkage of the solvent shell during the chemical transition was identified, suggesting a necessary energy transferring process during the thermal activated chemical reaction, reminiscent of the “cage-effect”. The transition path time of the reaction and the rate constants for the forward and backward reaction were calculated independent of RCs, and we were able to self-consistently connect the kinetics to the thermodynamics. The diffusion coefficient over the energy barrier was also determined according to Kramers’ theory, showing the limitation of canonical transition state theory in dealing with the condensed phase reactions. Therefore, this method allows us to investigate the solvent participation and energy transfer processes in solution chemical reactions.