Analysis of the Evolutionary Process of Spatial Microdamage Propagation in Silicon Nitride Bearings Using Deep Learning-Driven Molecular Dynamics

To reveal the dynamic evolution mechanism of spatial microdamage on the surface of silicon nitride bearings, this study proposes a molecular dynamics modeling method based on 3D point cloud reconstruction and a machine learning potential function. A high-precision simulation framework for the dynamic evolution of silicon nitride microdamage is constructed by fusing three-dimensional morphological reconstruction of spatial microdamage point cloud data with a deep potential atomic interaction model. The simulation results show that during indentation loading, the initial microdamage is in a latent period before the critical depth and then dislocation slip and amorphization synergistically trigger the damage extension. During the unloading phase, the indentation region shows elastic recovery-dominated closure behavior, but the closure of the microdamage region is significantly lower than that of the surrounding matrix. The study reveals for the first time the dynamic coupling mechanism between microdamage extension and elastic-plastic deformation in silicon nitride bearings, elucidates the regional variability of damage closure behavior, and provides atomic-scale theoretical support for the life prediction and damage-resistant design of ceramic bearings.