Machine learning molecular dynamics reveals atomistic mechanisms of ion beam damage in calcium fluoride

Abstract

Calcium fluoride (CaF₂) crystals are regarded as ideal materials for optical components due to their excellent ultraviolet (UV) transmittance and optical anisotropy. High-precision and low-damage CaF₂ crystals can be prepared by ion beam processing, but the atomic-scale surface and interfacial stability under ion beam processing remains poorly understood. In this paper, a machine learning potential (DP) based on the DeePMD method was developed to explore the underlying physical mechanisms of microstructure and property changes in calcium fluoride under ion beam bombardment. The results show that the DP model has higher accuracy and stronger stability in property prediction, and can accurately describe the material properties under various complex experimental conditions. Using the DP potential, we simulated the irradiation process of calcium fluoride by high-energy ion beams to study the irradiation damage, microstructural evolution, and the motion behavior and distribution patterns of particles. The processing mechanism showed significant differences compared with that of low-energy ion beams. This study developed a potential function that can stably operate in high-energy ion beam irradiation systems, clarified the microscopic mechanism of high-energy ion beam processing, and provided theoretical guidance for optimizing ion beam processing techniques. This work has important reference value for the research of ion beam processing and material surface modification.