Plasma Modeling Framework on Dielectric Surfaces in Hybrid Bonding Technology

Atomistic modelling is used to study a variety of material properties, many of which are important for hybrid bonding technology. We created a computational platform, referred to as ‘Plasma Dynamics Simulations’, to study dielectric surface activation and improve the covalent bonding environment. In this approach, the force-field based classical molecular dynamics simulation technique is chosen to establish the interaction between active plasma species and target dielectric material. Parametrization of the force field is one of the critical tasks to ensure the model mimics all reaction conditions during plasma activation procedures. To address this problem, a machine learning force field algorithm is developed to generate an accurate potential, referred to as the moment tensor potential. Moment tensor potentials are highly transferable for large scale plasma simulations. They enable dynamical simulations with ab-initio accuracy and low computational cost. In this study, we chose SiO2 based dielectric materials and studied the effect of a nitrogen plasma to create reactive dielectric surfaces.