2025 AIChE Annual Meeting

(707c) Correcting Surface Reaction Rate Constants for Anharmonicity with Adtherm: Dissociation of H*CO on Co(0001)

Authors

David Bross, Argonne National Laboratory
Khachik Sargsyan, Sandia National Laboratories
Branko Ruscic, Argonne National Laboratory
C Franklin Goldsmith, Brown University
Partition functions of adsorbed intermediates and first-order saddle points (FOSPs) are evaluated to assess thermochemical and kinetic parameters. It is commonplace to assume that the potential energy surfaces near these stationary points are harmonic. However, work by Blondal et al. shows that this assumption can cause adsorbate entropy, and in turn adsorption equilibrium constants to be severely underestimated [1]. Accounting for anharmonicity causes a fixed equilibrium constant temperature shift of 130 K for methanol on Cu(111). This work was done using the open-source software: AdTherm. This tool generates and calculates energies for displacements corresponding to rigid body rotation and translation of adsorbates and uses the energies to train a minima-preserving neural network (MPNN). This results in a surrogate energy surface with enforced accuracy near the minima. Monte Carlo integration is performed over the phase space to get the anharmonic partition function corresponding to external motion. This tool was developed initially for reactive intermediates, but here it is extended to FOSPs. While the anharmonic modes of reactive intermediates correspond to rotation and translation, that may not be true for FOSPs. In this update, AdTherm will assess hessian eigenvalues at the FOSP and determine which modes are anharmonic. Samples will be generated along these modes, and energies will be calculated. These energies will be used to train MPNN and perform phase space integration. As a case study for this update, the anharmonic rate constant for the dissociation of formyl on Co(0001) will be calculated and compared to the harmonic result. It is anticipated that both formyl and the corresponding FOSP will behave anharmonically, but it is the goal of this work to see which is more anharmonic, and whether accounting for anharmonicity will increase or decrease the rate constant.

[1] Blondal et al. ACS Catal. 2023, 13, 19-32