Dynamics of H2O Adsorption on Pt(110)-(1 × 2) Based on a Neural Network Potential Energy Surface

Water adsorption on metal surfaces is of both fundamental and practical relevance to heterogeneous and electrocatalysis. Here, we report a comprehensive theoretical study on the dynamics of H2O adsorption on Pt(110)-(1 × 2) based on a neural network potential energy surface (PES) fit to thousands of density functional theory data points. This PES allows us to investigate the dynamics of molecular and dissociative adsorption of water on Pt(110)-(1 × 2) from first principles with no adjustable parameters. Specifically, we find a molecular adsorption well depth that is close to the experimental estimate with no entry barrier, leading to a monotonic decrease in initial sticking probability (S0) with increasing translational energy. While this translational energy dependence of S0 is consistent with the conventional mechanism for nonactivated adsorption and with earlier theoretical results, it is in qualitative disagreement with experimental data. Using an electronic friction model implemented with the local density friction approximation, we find that the low-energy electron–hole pair excitation mildly increases the trapping probability, whereas it has little influence on the translational energy dependence. Other possible factors, such as the rotational alignment and dynamical steering, which may be relevant to the discrepancy between theory and experiment, have also been discussed. On the other hand, the dissociative sticking probability is predicted as complementary information awaiting experimental confirmation. We hope that these results could motivate future experimental and theoretical studies to explore this puzzling gas–surface system.