Modeling, Characterization and Simulation of Memristors-Based Devices for Sensing and Computing Applications

Abstract

The Memristor – memory resistor (MR)- is a non-linear passive two-terminal element that can be programmed to a certain resistance between two boundary levels: the Low Resistance State (LRS) and High Resistance State (HRS). The MR is a promising candidate to play a great role in the field of electronic technology owing to its high integration density, low power consumption, and relatively fast read/write times. Moreover, the ability of non-volatile MRs to retain their resistance state makes them suitable in Resistive Random-Access Memory, programmable logic and In-Memory Computing (IMC) applications. Additionally, MR devices response to environment can be utilized for sensing application. Despite the great advantages of MRs devices and their future potential there are still some challenges that stand between their design, fabrication, reliably, and their practical implementation. MRs devices have different physical structures and fabricated using different techniques and methods resulting in different characteristics and performance that are difficult to compare. Additionally, developing a general model that reliably describes the I-V behavior of various MR devices is one of the most challenging problems in MR technology. Moreover, one of the most challenging problems in crossbar resistive RAM (ReRAM) memories that limit their scalability and efficiency is the sneak path current. Finally, the exploration of the memristor element in applications beyond computing and in-memory like sensing and security has become an attractive changeling topic among researchers. This thesis encompasses five major parts containing original contributions. Part I Electrical Characterization, Part II Modeling and Simulation, Part III High Density ReRAM Crossbar with Selector Device for Sneak Path Reduction, Part IV Hybrid Memristor-CMOS Based Flash Analog to Digital Converter and Part V Memristor In Radiation Sensing. Part I contains Electrical Characterization and parameter extraction for various Memristor-based devices using cutting edge testing techniques and procedures. Part II describes various mathematical models in the literature that are used to model the Memristor behavior. Additionally, extracted parameter from fabricated Memristor has been used to tune and fit the mathematical and spice model for our group MR devices. Part III presents a novel fabricated memristor devices to synthesize a one selector one memristor (1S1R) configuration used in ReRAM crossbar. The proposed design aims to reduce the sneak path current in crossbar architecture which limits the array size, power, and noise margin in Resistive memories. The proposed configuration increases the crossbar read margin and device selectivity, as well as it enables high memory array density compared to the state-of-the art solutions. Part IV presents a novel hybrid Memristor-CMOS based flash Analog to Digital Converter (ADC). The proposed design solves the issue of resistor mismatching and its effect in the flash ADC encoding error using post-fabrication tuning property of MR technology. Part V report on novel approaches to utilize the electrical characteristics of the fabricated memristive devices for radiation sensing. This part reports the impacts of radiation on the electrical characteristics of fabricated memristors.

Date of Award	May 2019
Original language	American English