Doctoral Dissertations
Date of Award
5-2025
Degree Type
Dissertation
Degree Name
Doctor of Philosophy
Major
Data Science and Engineering
Major Professor
Ganesh Panchapakesan
Committee Members
Bobby Sumpter, Adri van Duin, David Mandrus, Jan-Michael Carrillo, Rama Vasudevan, Russell Zareztki
Abstract
The discovery of multifunctional properties such as ferroelectricity in hexagonal nitride and oxide materials has sparked renewed interest in wurtzite materials, given their compatibility with CMOS integration and tunable optical and piezoelectric properties. However, the origin of ferroelectricity and its switching mechanism under electric fields remain poorly understood, posing challenges for designing and improving low-power microelectronic devices based on these materials. Similarly, in ferroelectrics like BaTiO$_3$, polar domain walls at the nanoscale show promise for memory devices, but their structure and dynamics, particularly in defective materials, are not well characterized. In this work, we address these gaps by performing largescale simulations to investigate both the field-induced switching in wurtzite-ZnO and the nanoscale domain-wall dynamics in pristine and defective BaTiO$_3$. Using reactive force fields of BaTiO$_3$ and a graph dynamical neural network approach, we capture atomistic switching mechanisms, extract energy barriers, and reveal the role of defects such as oxygen vacancies in slowing dipole relaxation and influencing domain-wall dynamics on the hysteretic behavior \cite{dhakane2023graph}. In addition, we analyzed ferroelectric switching trajectories in ZnO with/without Mg-dopants and identified the key switching mechanism in good agreement with recent experimental results \cite{calderon2023atomic, dhakane2025understanding}. Furthermore, large-scale simulations revealed a new design principle on how to lower the coercive fields in wurtzite ferroelectrics, that will revolutionize their adoption in making low-power microelectronic devices. More generally, we show the power of combining large-scale simulations and data analytics, including deep learning techniques to provide a new paradigm of {\em ‘dynamics-by-design’} for discovering novel functional materials. In addition to field-induced switching, advantage of this paradigm in controlling out-of-equilibrium processes such as material synthesis will also be discussed in the outlook section of this thesis.
Recommended Citation
Dhakane, Abhijeet S., "Understanding the Switching Dynamics in Memristor Materials using Large-scale Simulations and Deep Learning Methods. " PhD diss., University of Tennessee, 2025.
https://trace.tennessee.edu/utk_graddiss/12349
Included in
Atomic, Molecular and Optical Physics Commons, Condensed Matter Physics Commons, Data Science Commons, Semiconductor and Optical Materials Commons