(659f) Theoretical Investigation of the Pt Catalyzed Hydrodeoxygenation of Succinic Acid to 1,4-Butanediol

The surface reaction network of the hydrodeoxygenation (HDO) of succinic acid (SUCC) to 1,4-butanediol (BDO) exhibts enormous complexity with hundreds of intermediates and thousands of possible elementary reaction steps. To investigate all theoretically possible elementary reactions at DFT/BEEF-vdW level of theory is not only cubersome, but likely also a waste of computational resources, considering that most likely only a small part of the total reaction network is of kinetic relevance. We present here a framework for optimizing the reaction network using models that are trained on the fly.

Our model framework consists of a database of reaction intermediates and transition states from both this and our previous studies on the HDO of propanoic acid, stacked Gaussian processes (GP) that predict both the unmeasured adsorption energies and transition state energies, and a mean-field microkinetic model which can identify key reaction intermediates, rate controlling steps and dominant catalytic cycles. The model is iteratively used until all relevant reaction intermediates and transition states are explicitly calculated at DFT/BEEF-vdW level of theory.

We applied this modeling framework to the HDO of SUCC to BDO on Pt (111) and Pt (100) and identified the dominant reaction pathway. Next, we predicted the uncertainties of quantities of interest such as the TOF to identify the active site. Finally, a comparison between our predicted values with experimental measurements is made to prove the applicability of our model.