| PhD Seminar

Name of the Speaker: Mr. Salil Chourasia (EE21D403)
Guide: Dr. Bhaswar Chakrabarti
Online meeting link: https://meet.google.com/aoc-wzoc-gup
Date/Time: 21st May 2025 (Wednesday), 4 PM
Title: Investigation of resistive switching in HfOx Resistive Random Access Memory devices.

Abstract :

With the ever-increasing data and data-centric applications, the demand for energy consumption is also increasing at a significant rate. Artificial neural networks (Deep Learning) have demonstrated exceptional accuracy in these applications. CMOS-based architectures have been widely employed in neural networks. However, there is a delay and energy consumption since the processing unit needs data from the off-chip memory. The non-volatile memory allows for local storing of network parameters and one-step vector-matrix multiplication, in-memory computing can greatly reduce latency and energy consumption. A promising option for new non-volatile memory technologies is Resistive Random Access Memory (ReRAM), which offers low write/read latency, multi-bit storage, and high device density.

In this presentation, we will discuss the impact of data-centric applications. Next, we will discuss how in-memory computing is a better way to minimise the energy demand of these data-centric applications and how it can be achieved. Next, we go over the many new non-volatile memory technologies that we can use for in-memory computing. Next, we go over the HfOx ReRAM technology and how it can be improved. Next, we will discuss a non-linear capacitance behaviour observed in HfOx ReRAM that is not explored by many researchers. Detailed investigations have been carried out to improve the performance of HfOx ReRAM and to understand the observed non-linear capacitance behaviour.