DFT+U-Inspired Functional for Improved Modeling of Molecules and Solids

Accurate description of chemical bond energetics is a pivotal part of computational chemistry and catalysis. It ultimately determines how successful quantum mechanical models are and how reliable their predictions. Density functional theory (DFT) together with the generalized gradient approximation (GGA) is inarguably the method of choice for modeling surface phenomena in heterogeneous catalysis due to its optimal accuracy-to-cost ratio. Despite the success of GGA functionals (PBE, RPBE, PW91) in describing chemistry on metal surfaces, the GGA-inherent self-interaction error poses challenges for accurate treatment of processes on semiconductor surfaces, such as oxides.

Herein, we reveal the inadequacy of the PBE functional for describing experimental trends on metal oxide surfaces (e.g., RuO₂, TiO₂) for the vacancy-catalyzed¹ furfuryl alcohol hydrodeoxygenation, using first principles microkinetic modeling. We attribute the error to nonsystematic deviations of metal-oxygen bond energetics from experimental values. By utilizing adiabatic connection and the nth order perturbation theory, we arrive at an orbital-occupancy-dependent correction to the PBE functional that holds promise in mitigating the DFT accuracy problem. Corrections are formulated in the spirit of the DFT+U method. The new PBE-based parameter-free functional demonstrates clear improvement over pure PBE in describing dissociation energetics of molecules, NiO local magnetic moment and density of states, and cohesive energies of metal oxides, at minimal extra computational cost.

References:

1. Mironenko, A.V., Vlachos, D.G. "Conjugation-Driven “Reverse Mars–van Krevelen”-Type Radical Mechanism for Low-Temperature C–O Bond Activation." Journal of the American Chemical Society 138.26 (2016): 8104-8113.