Dr. Ye Cao, Assistant Professor from Department of Materials Science and Engineering, University of Texas at Arlington will give a seminar titled "Phase-field Simulation and Machine Learning Study of Conductive Filament Dynamics in Metal Oxide-based Resistive Random-Access Memory" to the interested faculty and students at Discovery Park.

Abstract

The emergence of big-data and data centric applications, such as machine learning and neuromorphic computing, require new paradigms to process information. The present von Neumann paradigm in which data is cycled between separate units for memory and processing is inefficient and energy costly. In this regard, the phenomenon of resistive switching enables the creation of resistive random-access memory (RRAMs), which are devices that can maintain (memorize) and change their electrical resistance based on the history of an applied voltage, providing a platform for so-called processing-in-memory that greatly enhances the data processing efficiency.

Oxides are particularly promising candidates for nonvolatile RRAMs, due to its lower power consumption, fast switching speed, simple structure, and compatibility to the complementary metal oxide semiconductor (CMOS) technologies. The functionality of an oxide-based RRAM stems from the voltage-driven partial rupture and reform of oxygen vacancy (V_O^(∙∙))−rich filaments acting as conductive pathways for electrons. One of the critical challenges that limit the broad applications of metal oxides-based RRAM is the stochastic nature in the formation and spatial distributions of the conductive filaments (CFs), which causes the behavior of memristors to vary not just from device to device, but within a single device over switching cycles. This is due to the extremely complicated multi-physics processes that accompany the CF evolutions, including V_O^(∙∙) formation/evolution, heat transfer, electrical conduction, and local strain effect. These processes are highly coupled, and their complicated interactions are not fully understood.

In this presentation, I will demonstrate how a phase-field model is developed and utilized to fundamentally understand the multi-physics behaviors in resistive switching, and to deterministically control the CF formation and evolution. We use HfO2-x as a prototypical model system. I will first discuss the effect of electrode properties on the oxygen vacancy segregation in bare HfO2 during the electroforming process. The role of the electrical bias, heat transport, strain, and the intrinsic electrical and thermal properties of the switching layers will be systematically explored. Next, high-throughput phase-field calculations and machine learning method are performed to derive interpretable analytical models for device performance metrics in terms of key material parameters. Finally, I will demonstrate how phase-field simulation can design the embedded oxide heterostructure to regulate the CF formation and reduce the stochasticity in terms of switching speed, operational voltages, and current on/off ratios etc. during resistive switching. This work provides a fundamental understanding and control of the CF behavior in metal oxide-based RRAMs, and demonstrates a computational data-driven methodology of materials selection for improved RRAM performance.

Biography

Dr. Ye Cao is an Assistant Professor in the Departments of Materials Science and Engineering at University of Texas at Arlington (UTA). Before joining UTA, he was a Postdoc Research Associate in the Center of Nanophase Materials Sciences at Oak Ridge National Laboratory. He obtained his Ph.D. degree in Materials Science and Engineering from the Pennsylvania State University. His research focuses on the mesoscale phase-field simulations. His current topics include charge transport in oxide-based resistive random-access memories, interfacial stability in Li-ion batteries, and ferroelectric domain structure and switching in multi-functional ferroelectric oxides. He has authored/co-authored more than 50 journal publications. His research projects have been sponsored by National Science Foundation and American Chemical Society.