Developments of models to simulate Atomic Layer Deposition of Gallium Oxide

In this study, we developed interatomic potential models to model surface reactions between the precursors of Atomic Layer Deposition (ALD) by using Artificial Intelligence (AI). The objective of the deposition is to generate a thin layer of highly conforming Gallium Oxide. The ALD employs two half-cycle processes: 1) the deposition of water that delivers Oxygen and 2) the deposition of [Ga(NMe₂)₃]₂ dimer that delivers Gallium are alternated during each a half cycle during the deposition. We constructed the Machine Learning Interatomic Potential (MLIP) by employing the ab-initio molecular dynamics (AIMD) data. We then subsequently evaluated the accuracy of the MLIP by assessing the surface reactions and dynamics during each half-cycle of ALD process. As a part of a broader study, we also employed a quantum computing simulator, Qiskit-Aer, to simulate the deposition process. The computing work has been performed using NERSC supercomputing facility through QIS@Perlmutter program and Missouri State University’s AI workstation.