Studies On Rotationally Inelastic Scattering Dynamics Of NO From Ag(111)

A rich variety of chemical reaction types and energy transfer processes occur on gas-solid surfaces,and further research and exploration of these interaction reactions can help us to gain a deeper understanding and appreciation of industrial production regulation.The scattering process between gas molecules and metal surfaces is a common reaction type in gas-solid reactions,and energy transfer usually occurs during scattering,where the energy of gas molecules is transferred to metal surface phonons,causing phonon excitation upon collision with metal surfaces,or to other degrees of freedom within the molecule,causing rotational or vibrational excitation.While first-principles ab initio molecular dynamics can directly simulate this process,it is expensive for rare event scattering processes that require a large number of trajectories to compute and statistics.With the development of computational power and machine learning methods,quasi-classical trajectory calculations can be performed using a neural network-based approach to fit high-precision potential energy surfaces without loss of accuracy for large-scale scattering calculations at lower cost.In this thesis,we use this approach to study the effect of rotational excitation of NO molecules during inelastic scattering on the Ag(111)surface.State-to-state molecular scattering from surfaces represents a sensitive probe of the molecule-surface interaction potential.Due to the abundant experimental data on the scattering of NO molecules on the Ag(111)surface,this system serves as a benchmark for our study of gas-surface energy transfer processes.Early theoretical simulations were carried out by constructing empirical and semi-empirical potential energy functions,and most of them did not consider the degrees of freedom of surface atoms,which undoubtedly led to significant errors.Although they could qualitatively reproduce the experimental trends in the distribution of final rotational states,they still lacked effective first-principles theoretical simulations.Here,we use neural network method to construct two high-dimensional first-principles potential energy surfaces(PESs)based on density functional theory(DFT)single points using PBE and revPBE functionals respectively,for describing the rotationally inelastic scattering of NO from Ag(111).Quasi-classical trajectory calculations based on the two PESs have been compared with a diversity of experimental data and the influence of scattering angle,initial molecular orientation,incidence energy on the rotational state distribution is discussed.It is found in general that revPBE-based results agree better than PBE-based ones with the measured scattering angle and final rotational state distributions,capturing reasonably the rotational rainbow feature at a small incidence energy and the steric effect.But neither PES reproduces all experimental data quantitatively.The different performance of the two PESs can be attributed to their difference in the attractiveness,anisotropy,and corrugation in the entrance channel.Our results suggest the need of further improvement of the accuracy of DFT with commonly-used density functionals towards a chemically accurate description of molecule-surface interactions.