Date of Graduation
Fall 2025
Degree
Master of Science in Materials Science
Department
Physics, Astronomy, & Materials Science
Committee Chair
Sakidja, Ridwan Physics, Astronomy & Material Science CNAS
Abstract
This thesis presents a multiscale computational investigation into the atomic-scale mechanisms governing defect energetics and resistive switching in intrinsic and MgO-doped Ga₂O₃ systems, with a particular focus on the role of interposed MgO layers in memristive device architectures. By combining Density Functional Theory, Moment Tensor Potentials, and Molecular Dynamics simulations, the study captures both quantum-level accuracy and large-scale dynamical behavior. DFT calculations were used to compute total energies, atomic forces which served as training data for a Moment Tensor Potential capable of reproducing DFT-level precision while enabling simulations over nanosecond timescales and in supercells containing thousands of atoms. The trained MTP revealed that MgO doping enhances vacancy mobility by introducing local strain relief and energetically favorable diffusion pathways. Further ML-MTP simulations demonstrated that MgO atomic layers in interposed MgO/Ga₂O₃ structures promote vacancy diffusion by enabling bond angle reorganization and providing more compliant local environments under applied electric fields. These effects facilitate faster and more uniform conductive filament formation, contributing to improved switching speed, device yield, and endurance. The findings establish MTP as a powerful tool for modeling defect-driven phenomena in oxide electronics and highlight the critical role of atomic layer design, particularly MgO insertion in optimizing the structural and transport properties of sub-2 nm memristors.
Keywords
memristors, resistive switching behavior, density functional theory, moment tensor potential, conductive filament, oxygen vacancy
Subject Categories
Semiconductor and Optical Materials
Copyright
© Anika Tabassum
Recommended Citation
Tabassum, Anika, "AI-Driven Interatomic Potentials for Modeling Defective Gallium Oxide" (2025). Graduate Theses/Dissertations. 4130.
https://bearworks.missouristate.edu/theses/4130
