Fabrication and Characterization of Spintronic Devices for Energy Efficient Computing

Date

2022

Authors

Cherian, Hebin Roy

Journal Title

Journal ISSN

Volume Title

Publisher

Abstract

Abundant data applications such as image recognition, voice-activated assistance, self-driving cars, genomic sequencing, etc., are rapidly rising due to the innovations and advances in deep learning. However, running the deep learning algorithms on the conventional von Neumann computing architecture may offset the advantages in low-power consumption. Modern electronic device technologies aim to keep power density constant while increasing speed, but this is becoming increasingly difficult as we approach the fundamental scaling limits. In this dissertation, the author explores novel spintronic devices for improving the energy-efficiency of the future computing system. The first part focuses on characterization of spin-transfer torque magnetic random-access memory (STT-MRAM) for deep learning (brain-inspired computing). A novel observation of non-uniform gray-scale resistance levels was made on a single STT-MRAM cell, along with comprehensive characterization studies on cell size dependence and reliability. For neuromorphic computing purposes, it was found that the STT-MRAM cell with a steeper slope in the AP-state (anti-parallel) and a higher AP-state resistance is preferred as a synaptic device. This multi-level STT-MRAM cell was further utilized as a synaptic element in a crossbar structure to implement a deep learning accelerator with much enhanced energy efficiency compared with the state-of-the-art. In the second part of the dissertation, the author focuses on the design and fabrication of a spintronic tunnel device that can be possibly explored as a testbed for spin injection experiments. In particular, the author successfully develops a process flow for fabricating a graphene inserted tunneling device (GiTD) in the stack of ferromagnet/tunnel barrier (including graphene)/superconductor. The developed process flow is used to create and propose multiple crossbar device structures for studying spin polarization. A few challenges in fabricating such a delicate device structure and how they could be overcome are discussed in detail. This device will enable subsequent STS (superconducting tunneling spectroscopy) measurements to study how graphene and interfaces introduced by graphene (both FM/graphene and graphene/tunnel barrier interfaces) impact tunneling spin polarization and how one can use graphene to enhance the spin injection efficiency in spin devices.

Description

This item is available only to currently enrolled UTSA students, faculty or staff. To download, navigate to Log In in the top right-hand corner of this screen, then select Log in with my UTSA ID.

Keywords

Fabrication, Graphene inserted tunneling device, Neuromorphic computing, Spintronics, STT-MRAM

Citation

Department

Electrical and Computer Engineering