(443g) Data-Driven Strategy for Ab-Initio Microkinetic Modeling of C1-Reaction on Copper (111) from Disparate Experimental Datasets

Microkinetic models based on energies derived from density functional theory (DFT) (with or without re-parameterization with kinetic data) are often used to elucidate mechanisms and quantify reaction fluxes of catalytic systems. However, such analyses are often carried out for a specific system of interest and independent of related chemistries.

Here, we present a statistical framework that enables a systematic and active improvement of a microkinetic model by making use of a variety of thermokinetic data (e.g. adsorption calorimetry, DFT energies, steady-state kinetics, and temperature-programmed studies). To achieve this, we build a statistical framework that systematically calibrate the intrinsic error of DFT and quantify the uncertainty. Then, Bayesian experimental design is used to improve the prediction of kinetic model by sequentially select the most “informative” kinetic experiments of different type and from different resources.

We demonstrate how this approach allows us to build a unified model describing the reactions of C1 oxygenates on Cu(111). This work demonstrates the feasibility (and caveats) of kinetic modeling of large reaction network using disparate datasets.