A DFT and MD study of aqueous-phase dehydrogenation of glycerol on Pt(1 1 1): comparing chemical accuracy versus computational expense in different methods for calculating aqueous-phase system energies

Abstract

Glycerol, which is one of the most abundant by-products in biodiesel production, can be converted into H2 through aqueous-phase reforming (APR). Dehydrogenation is one of the main processes in glycerol APR. In this work, we use computational methods to study Pt(1 1 1)-catalysed glycerol dehydrogenation under aqueous conditions. There are 84 intermediates and 250 possible reactions in the dehydrogenation network. Inclusion of the liquid environment adds computational expense, especially if we are to study all the reaction intermediates and reactions under explicit water solvation using quantum methods. In this work, we present a method that can be used to efficiently estimate reaction energies under explicit solvation with reasonable accuracy and computational expense. The method couples a linear scaling relationship for obtaining adsorbate binding energies with Lennard-Jones + Coulomb potentials for obtaining water–adsorbate interaction energies. Comparing reaction energies calculated with this approach to reaction energies obtained from a more extensive approach that attains quantum-level accuracy (published previously by our group), we find good correlation (R2 = 0.84) and reasonable accuracy (the mean absolute error, MAE = 0.28 eV).