Plasma Etching III

The continued advancement of plasma-assisted etching technologies requires a fundamental understanding of plasma-surface interactions (PSIs). Due to the difficulty of experimental studies of PSIs, as well as continually shrinking device critical dimensions, molecular dynamics (MD) simulations can be a powerful tool to understand the properties of such systems by resolving the positions and velocities of a collection of atoms. Empirical potentials for plasma processes can be difficult to generalize to complex combinations of multiple elements. However, recent advances in machine learning (ML) methods have enabled the development of ab initio-based models which could greatly extend the range of chemical systems that can be modeled. In this work, we have developed a ML model, trained on data from quantum density functional theory calculations, for the etching of Si by neutral and ion species. Results from MD simulations using ML models are compared to simulation data with empirical potentials, as well as to experimental measurements. Etch yields as a function of flux ratio and ion energy for simultaneous Cl and Ar⁺ impacts are in good agreement with previous simulation results and experiment. Further, we use ML potentials to simulate Si etching by Cl⁺ ions, as well as simultaneous Cl and Cl⁺ impacts. The etch yield as a function of Cl⁺ ion energy shows good agreement with experiment and previous results using empirical force fields. Finally, we analyze the product distribution during the etch process.

Funding acknowledgement

This work was supported by the US Department of Energy OFES (contract # DE-AC02-09CH11466), Samsung Electronics (Project Code # IO 210224-08446-01), and the Onassis Foundation - Scholarship ID: F ZS 029-1/2022-2023.