Ab initio molecular dynamics with enhanced sampling for surface reaction kinetics at finite temperatures: CH2⇌ CH + H...

A comprehensive understanding of surface thermodynamics and kinetics based on first-principles approaches is crucial for rational design of novel heterogeneous catalysts, and requires combining accurate electronic structure theory and statistical mechanics modeling. In this work, ab initio molecular dynamics (AIMD) combined with the integrated tempering sampling (ITS) method has been explored to study thermodynamic and kinetic properties of elementary processes on surfaces, using a simple reaction CH2⇌CH+H on the Ni(111) surface as an example. By a careful comparison between the results from ITS-AIMD simulation and those evaluated in terms of the harmonic oscillator (HO) approximation, it is found that the reaction free energy and entropy from the HO approximation are qualitatively consistent with the results from ITS-AIMD simulation, but there are also quantitatively significant discrepancies. In particular, the HO model misses the entropy effects related to the existence of multiple adsorption configurations arising from the frustrated translation and rotation motion of adsorbed species, which are different in the reactant and product states. The rate constants are evaluated from two ITS-enhanced approaches, one using the transition state theory (TST) formulated in terms of the potential of mean force (PMF) and the other one combining ITS with the transition path sampling (TPS) technique, and are further compared to those based on harmonic TST. It is found that the rate constants from the PMF-based TST are significantly smaller than those from the harmonic TST, and that the results from PMF-TST and ITS-TPS are in a surprisingly good agreement. These findings indicate that the basic assumptions of transition state theory are valid in such elementary surface reactions, but the consideration of statistical averaging of all important adsorption configurations and reaction pathways, which are missing in the harmonic TST, are critical for accurate description of thermodynamic and kinetic properties of surface processes. This work clearly demonstrates the importance of considering temperature effects beyond the HO model, for which the AIMD simulation in combination with enhanced sampling techniques like ITS provides a feasible and general approach.