Index k saddles and dividing surfaces in phase space, with applications to isomerization dynamics

TL;DR

This paper develops a phase space geometric framework using normal form theory to analyze index k saddles, enabling the sampling of dividing surfaces that distinguish different isomerization pathways in complex molecular systems.

Contribution

It provides explicit formulas for dividing surfaces near index k saddles and introduces a numerical sampling method, advancing the understanding of multi-saddle dynamics in chemical reactions.

Findings

Successfully sampled dividing surfaces in a 4-minima potential energy surface.

Distinguished concerted crossing from sequential isomerization trajectories.

Extended the method to systems with bath modes, maintaining effectiveness.

Abstract

In this paper we continue our studies of the phase space geometry and dynamics associated with index k saddles (k > 1) of the potential energy surface. Using normal form theory, we give an explicit formula for a "dividing surface" in phase space, i.e. a co-dimension one surface (within the energy shell) through which all trajectories that "cross" the region of the index k saddle must pass. With a generic non-resonance assumption, the normal form provides k (approximate) integrals that describe the saddle dynamics in a neighborhood of the index k saddle. These integrals provide a symbolic description of all trajectories that pass through a neighborhood of the saddle. We give a parametrization of the dividing surface which is used as the basis for a numerical method to sample the dividing surface. Our techniques are applied to isomerization dynamics on a potential energy surface having 4 minima; two symmetry related pairs of minima are connected by low energy index one saddles, with the pairs themselves connected via higher energy index one saddles and an index two saddle at the origin. We compute and sample the dividing surface and show that our approach enables us to distinguish between concerted crossing ("hilltop crossing") isomerizing trajectories and those trajectories that are not concerted crossing (potentially sequentially isomerizing trajectories). We then consider the effect of additional "bath modes" on the dynamics, which is a four degree-of-freedom system. For this system we show that the normal form and dividing surface can be realized and sampled and that, using the approximate integrals of motion and our symbolic description of trajectories, we are able to choose initial conditions corresponding to concerted crossing isomerizing trajectories and (potentially) sequentially isomerizing trajectories.