Accessing Numerical Energy Hessians with Graph Neural Network Potentials and Their Application in Heterogeneous Catalysis

TL;DR

This paper demonstrates that pretrained graph neural network potentials can accurately compute energy Hessians for catalytic systems, enabling improved free energy calculations, entropy estimation, and transition state searches in heterogeneous catalysis.

Contribution

It shows that existing machine learning potentials can reliably determine Hessians, facilitating advanced thermodynamic and kinetic studies in catalysis without extensive quantum calculations.

Findings

Achieved 58 cm$^{-1}$ MAE in Hessian prediction for adsorbed intermediates.

Estimated vibrational entropy with 0.042 eV MAE at 300 K using simple offset correction.

Reduced unconverged transition state searches by up to 93% using Hessian-based methods.

Abstract

Access to the potential energy Hessian enables determination of the Gibbs free energy, and certain approaches to transition state search and optimization. Here, we demonstrate that off-the-shelf pretrained Open Catalyst Project (OCP) machine learned potentials (MLPs) determine the Hessian with great success (58 cm$^{-1}$ mean absolute error (MAE)) for intermediates adsorbed to heterogeneous catalyst surfaces. This enables the use of OCP models for the aforementioned applications. The top performing model, with a simple offset correction, gives good estimations of the vibrational entropy contribution to the Gibbs free energy with an MAE of 0.042 eV at 300 K. The ability to leverage models to capture the translational entropy was also explored. It was determined that 94% of randomly sampled systems had a translational entropy greater than 0.1 eV at 300 K. This underscores the need to go beyond the harmonic approximation to consider the entropy introduced by adsorbate translation, which increases with temperature. Lastly, we used MLP determined Hessian information for transition state search and found we were able to reduce the number of unconverged systems by 65% to 93% overall convergence, improving on the baseline established by CatTSunami.