Limits of BOSS DFT: O2 + Al(111) Dynamics on a Screened Hybrid Van der Waals DFT Potential Energy Surface

Abstract

The activated dissociative chemisorption (DC) of O₂ on Al(111) is a thoroughly studied benchmark system for oxygen–metal interactions. However, research based on density functional theory (DFT) has not yet been able to accurately determine the electronic structure, and theory as a whole has so far been unable to reproduce measured sticking probabilities with chemical accuracy. Previous work has argued that this is likely due to the inability of DFT at the generalized gradient approximation (GGA) level to describe the barriers to DC of O₂ on Al(111) correctly. The argument is that the most commonly applied electronic structure approach in surface science, which involves the use of GGA-DFT, yields too low reaction barriers for the DC of O₂ on Al(111). Moreover, it seems that GGAs will generally fail to accurately predict barriers for systems with low charge transfer energy, i.e., systems for which charge transfer from metal to molecule at the transition state is likely. Subsequent work on both O₂ + Al(111) and O₂ + Cu(111) has suggested that screened hybrid density functionals (DF) yield more accurate barrier heights for DC on metal surfaces. However, so far the use of only a screened hybrid DF was not enough to ensure a highly accurate description for O₂ + Al(111). Even though the onset of the sticking probability (S₀) curve was correctly described, the slope, or width, of the curve was not. The use of a nonlocal correlation DF combined with an increased fraction of exact exchange in the screened hybrid exchange DF was believed to further improve the description of the electronic structure by increasing the energetic corrugation of the barrier. This approach was assumed to increase the width of the sticking curve without lowering the incidence energy for the reaction onset, thus reducing the slope of the sticking curve. To test this, we present quasi-classical trajectory (QCT) calculations on the O₂ + Al(111) system based on a potential energy surface (PES) computed with the HSE06-1/2x-VdWDF2 screened hybrid van der Waals DF, using the Born–Oppenheimer static surface (BOSS) model. The resulting PES shows the presence of shallow van der Waals wells in the entrance channel. Furthermore, the barriers to DC show a slightly higher energetic corrugation than the previously used HSE03-1/3x screened hybrid DF, although most differences are smaller than 1 kcal/mol. These minor alterations in the PES with respect to previous work mean that the S₀ computed for O₂ + Al(111) using the HSE06-1/2x-VdWDF2 DF are somewhat improved over the previous results. Specifically, the onset of the S₀ curve is now somewhat better described and the curve is broadened a little compared to the HSE03-1/3x description. These results, in combination with previous studies, imply that future electronic structure methods would need to provide larger changes in the PES, or a different dynamical model would need to be used to bring theory in better agreement with the experiment. Moreover, future higher-level theory also needs to address the currently very demanding computational costs of screened hybrid plane-wave-DFT for molecule-metal interactions.