Extending the Range of Nonadiabatic Processes that Can Be Treated with Analytic Representations of Coupled Potential Energy Surfaces

Johns Hopkins University, Baltimore MD

Investigators

Abstract

David R. Yarkony of Johns Hopkins University is supported by an award from the Chemical Theory, Models and Computational Methods program in the Division of Chemistry. The goal of the award is to build computational models of processes that convert light energy into chemical energy, but can not produce the reverse reaction. These are called nonadiabatic processes. Nonadiabatic processes are involved in many important processes including vision, photosynthesis and protecting our DNA against the dangers of the ultraviolet component of sunlight. Professor Yarkony and his research group produce computational models using high speed digital computers. The tools they develop add subtle but important interactions to their existing model that can alter the outcome of a predicted nonadiabatic event. Scientists frequently use light energy in an attempt to manipulate the outcome of a chemical reaction. If the computational model is sufficiently accurate, computer simulations can guide and inform experiments. If the model is inaccurate, incorrect inferences are obtained. Professor Yarkony has identified a class of light induced molecular breakup processes for which a decades-old standard model fails. His group continues to study this failure to establish its prevalence and the consequences of the errors. Professor Yarkony mentors a group of students and postdocs. Together they develop software that can be used by the community to explore the new methods. The Yarkony group is using fit coupled diabatic state electronic Hamiltonians to study nonadiabatic processes. They take advantage of the key strengths of fit representations, which is the use of wave functions that include both static and dynamic correlations. Nuclear quantum effects including the geometric phase (GP) and tunneling are calculated. In order to treat intense laser fields and spin-changing processes, dipole and spin orbit coupling terms must also be represented. The Yarkony group uses a two-step approach in which Gaussian Process Regression is used followed by the application of neural network techniques. The ability to extend the range of problems that can be treated in this way provides an important new tool to treat nonadiabatic processes. The accuracy of the diabatic potential energy matrix is improved, while quantum tunneling can be treated rigorously and the GP is properly included in full and reduced dimensionality representations. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.