Simulating the Effects of the Native Oxide Layer During Kinetic Spraying of Tantalum Particles Using a Deep Learning Interatomic Potential

The role of the native oxide layer present on metallic powders on the deformation and film formation mechanisms is not fully understood for powders deposited via kinetic spray processes such as cold spray or micro-cold spray (also referred to as the aerosol deposition method or vacuum kinetic spray). This study compares the deformation behavior of tantalum particles with and without an oxide layer using single-particle impact simulations performed via molecular dynamics (MD) simulations performed with a deep learning interatomic potential representation of the Ta/Ta₂O₅ system. The ability of the deep learning-based potential to accurately reproduce the impact behavior for this complex material system is verified and the computational costs are shown to be significantly lower than for ab initio molecular dynamics and complex reactive potentials previously used to simulate metal–oxide interfaces. The lower computational costs when using the deep learning approach allow for partially oxidized particles that are large enough to be comparable to what can be deposited experimentally (50 nm) to be studied using MD simulations across a wide range of impact velocities (250-750 m/s). These simulations reveal that the presence of the 3-nm-thick oxide layer reduces overall deformation by > 40% across all particle sizes and impact velocities. For 40 nm and 50 nm particles impacted at high velocity, however, the rupture of the native oxide layer allows comparable kinetic energy dissipation compared to particles without an oxide layer of equal diameter despite much more limited overall deformation.