(415a) The Surprising Accuracy of Dispersion-Corrected Ggas in the Prediction of Dissociation Barriers on Transition Metal Surfaces

The development of density functional theory (DFT) for accurate description of molecule-surface interactions, while essential for heterogeneous catalysis and energy conversion, is severely limited by the absence of well-defined, chemically accurate benchmark databases for reactions on surfaces. Therefore, unlike gas phase barriers for which density functionals have been extensively benchmarked, limited information is available on computational accuracies for surface reactions. We have developed a database of ten experimental apparent barrier heights (SBH10) for dissociation of small molecules such as H₂, N₂, and CH₄ on transition metal surfaces. The barriers are carefully chosen from combined molecular beam experiments and quantum dynamics simulations, laser assisted associative desorption (LAAD) experiments, and thermal rate measurements.

We have also benchmarked the performance of a dispersion-corrected GGA, BEEF-vdW, a metaGGA, MS2, and a screened hybrid functional, HSE06. In the gas phase, GGAs systematically underestimate barriers due to self-interaction errors that lead to incorrect charge separation at the transition state. By including exact exchange, hybrid functionals correct for these errors and lead to better barrier estimates. The trends are reversed for reactions occurring on transition metal surfaces. Our benchmarking study shows that the BEEF-vdW functional (mean error < 0.2 eV) outperforms MS2 and HSE06, which underestimate barriers by 0.3 eV, and 0.5 eV, respectively. This is because transition states for dissociative adsorption closely resemble the final chemisorbed states, and errors in barrier heights mirror errors in chemisorption energies. In other words, dissociation transition states are bound to the surface, while gas phase transition states are isolated species. The key driver for functional accuracy for reactions on surfaces, therefore, is the description of surface-adsorbate interactions, and not charge separation or self-interaction correction. A functional that can accurately predict chemisorption energies, like BEEF-vdW, is also reliable for estimating barrier heights on surfaces.