(62k) Pynta: An Automated Quantum Chemistry Workflow for Surface and Gas-Surface Kinetics Including Coverage Dependence

Heterogeneous catalysis is vital to many industrially important chemical processes. Running experiments on these systems at many different conditions can be very expensive and time consuming so it is usually most practical to instead construct microkinetic models that are able to describe the chemistry at an elementary level. However, microkinetic models require accurate estimates of many kinetic and thermodynamic parameters. Using quantum chemistry methods to calculate these parameters manually can be incredibly time consuming and error prone. We present our software Pynta (https://github.com/zadorlab/pynta) a workflow code for calculating thermochemical and kinetic parameters for surface and gas-surface reactions including coverage dependence. Pynta can be applied to any surface with known sites. Pynta generates saddle point guesses using its novel Harmonically Forced Saddle Point Search (HFSP) algorithm that we have augmented training subgraph isomorphic decision trees (SIDTs) to improve saddle point guess accuracy. Coverage dependent thermochemistry and rate coefficients can be calculated by leveraging pair-wise filtering and active learning using SIDTs to predict coverage dependent effects calculating a small fraction of possible surface configurations. We demonstrate Pynta’s ability to predict the coverage dependence of rate coefficients on a small set of reactions.