Key Role of Density Functional Approximation in Predicting M–N–C Catalyst Activities for Oxygen Reduction

Metal–nitrogen–carbon (M–N–C) motifs present intriguing structural and electronic properties for a number of applications, including as oxygen reduction catalysts. However, computational investigations of M–N–C-catalyzed reactions must grapple with their complex electronic structures. In the present study, we evaluate the impact of the density functional approximation on calculated M–N–C catalyst activities for oxygen reduction. Using metalloporphyrins as model catalysts, we find a significant split between pure (GGA) and hybrid functionals, with hybrid functionals, in particular B3LYP, showing greater agreement with DLPNO-CCSD(T) reaction energies. Notably, double-hybrids offered no noticeable improvement over the much more computationally efficient B3LYP and PBE0. Other discrepancies between functionals, as well as an in-depth analysis of ground state spin and geometry, are also considered in this work. Finally, both hybrid and double-hybrid functionals greatly reduced the gas phase errors associated with the main group molecules in the oxygen reduction reaction relative to GGA calculations, leading us to question the application of widely used empirical corrections to O2.