Physics based Modeling of Emerging Ferroelectric Devices and Performance Benchmarking of Memory Circuits
Author(s)
Adnaan, Mohammad
Advisor(s)
Editor(s)
Collections
Supplementary to:
Permanent Link
Abstract
A comprehensive framework for the performance analysis of ferroelectric-based memory systems, encompassing from device modeling to system-level analysis is proposed. Initially, a computationally efficient phase-field physics-based compact model for ferroelectric capacitors is developed. The model self-consistently solves the time-dependent Landau-Ginzburg and Poisson's equations to capture polarization dynamics. Analytical equations for the time-dependent kinetic coefficient and voltage-dependent gradient energy coefficient are derived, which are crucial for accurately modeling the transient characteristics of ferroelectric capacitors. This framework is then extended for ferroelectric, antiferroelectric, and dielectric mixed phase capacitors based on Kittel's two sublattice theory. It allows the model to capture endurance effects due to phase evolution during cycling and the effect of depolarization electric field due to the presence of dielectric phases. The developed models are calibrated with experimental results of low switching voltage ferroelectric materials for circuit level analysis. A detailed analysis is conducted on ferroelectric random access memory (FERAM) circuit arrays, examining the impact of various design parameters. The performance of these memory arrays is compared to other competing memory technologies, particularly magnetic memories, in terms of read/write latency and energy consumption. Finally, the dissertation discusses a framework for system level analysis under real workloads, exploring the potential of using FERAM as main memory.
Sponsor
Date
2024-12-02
Extent
Resource Type
Text
Resource Subtype
Dissertation