Neural Network Potential Energy Surface for the low temperature Ring Polymer Molecular Dynamics of the H2CO + OH reaction

TL;DR

This paper develops a new neural network-based potential energy surface for low-temperature reaction dynamics of H2CO + OH, addressing spurious interactions and validating with RPMD and trajectory studies that align well with experimental data.

Contribution

It introduces a novel neural network scheme that prevents spurious long-range interactions in PES modeling for low-temperature reactive dynamics.

Findings

The new PES accurately predicts reaction rate constants across temperatures.

RPMD reveals a trapping mechanism linked to ring polymer normal mode excitations.

The neural network approach improves low-temperature dynamical simulations.

Abstract

A new potential energy surface (PES) and dynamical study are presented of the reactive process between H2CO + OH towards the formation of HCO + H2O and HCOOH + H. In this work a source of spurious long range interactions in symmetry adapted neural network (NN) schemes is identified, what prevents their direct application for low temperature dynamical studies. For this reason, a partition of the PES into a diabatic matrix plus a NN many body term has been used fitted with a novel artificial neural networks scheme that prevents spurious asymptotic interactions. Quasi-classical trajectory and ring polymer molecular dynamics (RPMD) studies have been carried on this PES to evaluate the rate constant temperature dependence for the different reactive processes, showing a good agreement with the available experimental data. Of special interest is the analysis of the previously identified trapping mechanism in the RPMD study, which can be attributed to spurious resonances associated to excitations of the normal modes of the ring polymer.