Date of Award

Summer 8-2014

Embargo Period

1-13-2015

Degree Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Electrical and Computer Engineering

Advisor(s)

Larry Pileggi

Second Advisor

Jimmy Zhu

Abstract

Magnetic logic has entered the spotlight as an intriguing candidate for future electronic systems. Recently we proposed a magnetic logic technology (“mLogic”) based on a current driven four terminal device (“mCell”) with isolated read- and write- paths. The first step with this nascent technology is to understand the device limitations and performance in response to input stimuli and material properties. In this thesis we explore the design, micromagnetic modeling, and experimental verification of mCell devices. The concept of an mCell is best described as a “black box” device with four terminals. Two terminals constitute a write-path, wherein the direction of input current flows to program the digital state of the device. The other two terminals constitute a read-path that is electrically isolated from the write-path. The state of the device is read out as a high or low resistance through the read-path terminals. Because multiple nanomagnetic phenomena (including spin transfer torque and the spin Hall effect) can be used to program a magnetization (logic) state based on a current direction, we introduce several mCell designs that all satisfy the conceptual 4- terminal mCell model. For each design we describe the operating principles and key features, followed by a presentation of modeling results that indicate performance trends. Of particular focus is the influence of material properties and device geometry on the current density required to instigate state switching. It is found that with appropriate design choices in scaled devices, sub-10 μA switching currents are achievable. Furthermore, we explore other mCell designs that can accommodate switching times below 1 ns. As part of this device exploration work, we experimentally demonstrate successful domain wall motion, tunnel magnetoresistance, and coupling through a magnetic oxide to validate writepath, read-path, and interlayer components. These components are then integrated into a prototype device. We show this prototype can be reliably switched into one of two (binary) states by pulsing current through the write-path, thereby demonstrating the fundamental mCell concept. We conclude this thesis by proposing future research directions in device design and fabrication to improve this device to enable logic circuits and all-magnetic MRAM bitcells.

Share

COinS