Adsorbate Free Energies from DFT-Derived Translational Energy Landscapes

Adsorption free energies are fundamental to surface chemistry and catalysis. Standard models combine some assumed analytical form of the translational potential energy surface, often parametrized against density functional theory (DFT) calculations, with an analytical expression for the resultant translational densities of states (DOS), free energy, and entropy. Here we compare the performance of such models against numerical evaluations of the DOS and thermodynamic functions derived from solutions to the translational Schrödinger equation. We compare results for a translational potential energy surface (PES) derived from nudged eleastic band calculations with those obtained from adsorbate rastering across a series of monatomic (O, S, C, N, and H) and polyatomic (NH_x) adsorbates on (100) Pt and Au facets. We find that analytical models as commonly parametrized have mixed performance for describing the translational PES and that the consequences for computed free energies are modest but potentially significant in microkinetic models. Numerical solutions are possible for modest to no additional computational cost over analytical models and thus should be considered when reliable free energy estimates are needed or translational potential energy surfaces are available.