University of Illinois Urbana-Champaign

Understanding order dynamics in magnetic and ferroelectric materials and devices for next generation computing

Shukla, Ankit

Content Files
SHUKLA-DISSERTATION-2024.pdf

Permalink

https://hdl.handle.net/2142/124472

Description

Title
Understanding order dynamics in magnetic and ferroelectric materials and devices for next generation computing
Author(s)
Shukla, Ankit
Issue Date
2024-04-10
Doctoral Committee Chair(s)
Rakheja, Shaloo
Committee Member(s)
  • Zhu, Wenjuan
  • Shanbhag, Naresh
  • Ravaioli, Umberto
Department of Study
Electrical & Computer Eng
Discipline
Electrical & Computer Engr
Degree Granting Institution
University of Illinois at Urbana-Champaign
Degree Name
Ph.D.
Degree Level
Dissertation
Keyword(s)
  • Spintronics
  • Magnetism
  • Ferroelectricity
  • Ferroelectric Materials
  • Ferromagnets
  • Antiferromagnets
  • Modeling Dynamics
Abstract
The von Neumann computing architecture comprises three main components: memory, logic, and interconnect. Traditionally, improvements in logic and processor performance, namely increased speed, reduced power consumption, footprint and cost, accomplished by Moore's law---the continuous down scaling of physical dimensions of complementary metal oxide semiconductor (CMOS) transistors---have driven computing performance. On the other hand, recent advancements in silicon (Si)-based specialized hardware accelerators, such as graphics processor units (GPUs) and tensor processing units (TPUs), that can process information at a much faster rate than central processing units (CPUs), have helped revolutionize modern-day data-intensive applications like machine learning (ML), artificial intelligence (AI), big data, and the internet of things (IoTs). The next generation of computing, however, faces two main challenges. First, Moore's law is anticipated to slow down significantly by the end of this decade as Si reaches its fundamental limits, thereby limiting processor performance, including that of GPUs. Second, the `Memory wall', characterized by excessive time and power consumption in the transfer of large sets of data between memory and logic units, due to a performance gap between the processor and the main memory, poses a restriction on the overall performance of the system. Sustainable computing for the future requires investigating the physics of CMOS-compatible materials, followed by building novel devices, and architectures that co-locate memory and logic. These devices and architectures should be compact, fast, energy-efficient, and scalable with problem size. This work focuses on investigating the dynamics of emerging materials, including ferromagnets (FMs), antiferromagnets (AFMs), and doped hafnia-based ferroelectrics (FEs). Ferromagnetic materials are non-volatile, scalable down to nanometer (nm) size, and capable of exhibiting various dynamics in the megahertz (MHz) to gigahertz (GHz) range when driven by electric current. Nanoscale-sized magnetic tunnel junction (MTJ) devices, comprising two ferromagnetic layers sandwiching an insulator layer, are CMOS-compatible and can operate as a single-bit memory, source of random numbers, or signal generators. AFMs constitute another class of magnetically ordered materials with negligible net magnetization. When integrated into a tunnel junction device, they could potentially offer electric current-driven switching and oscillation dynamics in the terahertz (THz) regime. Hafnia-based FE materials form yet another category of CMOS-compatible materials that could exhibit switching dynamics in the hundreds of MHz frequency regime when driven by electric voltage. These materials can be utilized in ferroelectric tunnel junctions (FTJs) or in the gate stack of ferroelectric field-effect transistors (FEFETs), enabling the operation of multi-state tunable memory or sources of random numbers. The research presented encompasses three main themes: numerical modeling frameworks to investigate material dynamics, exploration of various dynamics and their dependence on material parameters and external stimuli, and leveraging these dynamics for developing CMOS-compatible and energy-efficient devices and circuits. In this context, FM and AFM dynamics are modeled using the Landau-Lifshitz Gilbert (LLG) equation to study current-driven switching and oscillation dynamics. Analytic models, in agreement with numerical results, are developed as a function of material parameters and external stimuli. On the device front, applications such as true random number generators and spiking neuron emulators are explored for FMs and AFMs, respectively. Additionally, the nucleation-limited switching (NLS) model is employed to investigate field-driven dynamics in FE materials and devices. Finally, a FE oscillator-based room temperature Ising machine circuit for solving combinatorial optimization problems is proposed.
Graduation Semester
2024-05
Type of Resource
Thesis
Handle URL
https://hdl.handle.net/2142/124472
Copyright and License Information
Copyright 2024 Ankit Shukla

Owning Collections

Graduate Dissertations and Theses at Illinois PRIMARY
Graduate Theses and Dissertations at Illinois