Validating potential energy surfaces for classical trajectory calculations

Huw O Pritchard

doi:10.1039/c5ra06732d

Abstract

Potential energy distributions for normal and reacting molecules.

Highlights

Since the advent of Eyring–Polanyi transition state theory, it has seemed almost axiomatic that a classical trajectory calculation for a unimolecular reaction on a carefully constructed potential energy (PE) surface could yield an acceptable reaction rate; this assumption rests primarily on the notion that classical and quantum properties converge at high energy, i.e. the correspondence principle will hold for the vibrational motions of a polyatomic molecule, which seems to be a reasonable assumption.[1]
For the foreseeable future, only semi-empirical potential energy surfaces will be available for estimating reaction rate constants of many interesting molecules, this study begins to explore methods for validating such surfaces
Two facets of the problem are addressed, whether energy can ow freely between different parts of the molecule, and the energy pro le of those molecules reacting relative to the whole energy distribution pro le

Summary

Introduction

Since the advent of Eyring–Polanyi transition state theory, it has seemed almost axiomatic that a classical trajectory calculation for a unimolecular reaction on a carefully constructed potential energy (PE) surface could yield an acceptable reaction rate; this assumption rests primarily on the notion that classical and quantum properties converge at high energy, i.e. the correspondence principle will hold for the vibrational motions of a polyatomic molecule, which seems to be a reasonable assumption.[1]. Hundreds of thermal unimolecular reactions have been reported over the past 60 years;[3] among those, three of the most extensively studied, the isomerisations of methyl isocyanide,[4,5] of cyclopropane,[6] and the thermal dissociation of di-tert-butyl peroxide,[7] present clear challenges for computational modelling. Ab initio potential energy surfaces are not available for any of these reactions, but for the simplest case, CH3NC # CH3CN, one can be expected before long. For the other two target reactions, an ab initio potential energy surface for the isomerisation of cyclopropane to propylene may follow later, but probably not for the thermal dissociation of ditert-butyl peroxide, for which one would need to create an empirical surface. As Bowman et al.[8] have said, empirical surfaces are “highly problematic”, and the purpose of the work described below is to probe possible methods for exploring de ciencies

Methods

Discussion

Conclusion