A physics-based simulation framework for ion transport and surface interaction analysis in high-aspect ratio etching

As the growth of the semiconductor industry accelerates, a precise understanding of ion-surface interactions within high aspect ratio (HAR) structures is essential. In particular, in the HAR etching process, complex physical behaviors such as ion scattering, energy transfer, and trajectory develop nonlinearly depending on the patterning characteristics within the structure, which is a key factor that makes accurate prediction and control of the etching process difficult.

In this study, molecular dynamics (MD) simulations were performed to theoretically elucidate the physical mechanisms of ion collision behavior in this process. The scattering characteristics and energy-angle distributions (EADs) for single bombardments were analyzed. Based on this data, a collision-based probability model was trained in a deep neural network (DNN) to build an AI-based encoding structure that can predict the physics-based probability distribution of ion-surface interactions.

The probability distribution function predicted by the DNN is then applied to a kinetic monte carlo (KMC) simulator, which is configured to dynamically track the ion trajectory, energy loss, sidewall collision frequency, and energy transfer as a function of position inside the trench. We propose a physics-based simulation framework to quantitatively analyze how ion transport pathways and energy transport properties change with structure under different HAR conditions.