Basic Information

  • Project Title: SOI device optimisation and analysis for operation in Band-to-band tunneling regime
  • Name: Jay Sonawane
  • Project: SRE
  • Semester(s): 7,8 continuing as DDP
  • Guide: Prof. Udayan Ganguly

Abstract

Spiking Neural Networks are 3rd generation neural networks generally used to realize hardware for neuromorphic computing. There are about 10^11 neurons in the human brain. To simulate and realize hardware that mimics such an extensive network of neurons, ultra-low power, and area solutions are necessary. LIF (leaky integrate and fire) Neuron implementation can be based on many device phenomena: impact ionization, single transistor latch phenomenon in SOI, etc. The technologies used to realize these phenomena include SB MOSFET, SOI, 2D materials (nascent), etc. Recently, Band-to-band tunneling (BTBT) based ultra-low power and area realization in the hardware of a neuron akin to biological neurons (spiking freq in kHz) was shown by Prof. Udayan Ganguly and group. The BTBT neuron is promising in terms of power (since MOSFET operates in the OFF state), area (No external capacitor required), and technology (SOI) is easy to integrate into hardware.

My work: TCAD-based modeling to understand underlying physics, design space optimization, and variability analysis. Apart from that, experimental variability and reliability analysis are also a part of this work.

Any courses you completed relevant to the project

EE207, EE620, EE724, EE746, EE672

Describe your experience on the project

Motivation for project topic: I am passionate about Nanoelectronics. My interest is to design and model novel MOSFETS (or devices) in general.

Motivation for guide: I approached Prof. Udayan after doing the EE724 course with him as the instructor. I liked his teaching methodology. Prof. Udayan is an excellent guide, and the group is also the best.

Type of Work: Major: TCAD software-based modeling. A strong grasp of the physics of transistors and literature study of current papers are needed. Minor: Electrical characterization in IITBNF lab for experimental data (Ph.D. student guide Shubham Patil was involved majorly in this).

Workload: Depends on you. Due to the DD course load, I put in the required hours for a general course (~6-10 per week) until December 2022. Jan, Feb 2023 was ~4-5 hours per day on average. In March and April, I was working every waking hour. Post-sem 8 (DDP) requires 8 hours per day. TCAD simulations can be time-consuming. Effective hours provided are definitely lesser.

Publication: Submitted to SISPAD 2023 in April (conference), was rejected. Will be submitting around 2 journal papers in the coming months. Other than that, I received URA01 for my research work.

Extension: The complete project is equivalent to good DDP work: TCAD Calibration followed by: i) Physics analysis ii) design space exploration iii) Variability analysis iv) Reliability analysis. I did complete i, ii, and am about to complete iii. Reliability is done by my Ph.D. student guide Shubham. He trains me while he experiments. Eventually, I would be experimenting independently.

I will be continuing the project. We will be done with the analyses soon. Hence the topic might change.

Describe your experience with the guide

Guide Involvement: 100%. Prof. Udayan reciprocates with your enthusiasm. You work closely with your mentor, who is a group member senior to you (Ph.D. student). My mentor is Shubham Patil (Ph.D. student). He is a fantastic mentor.

Frequency of meetings: Weekly group meetings. Personal meeting with guide whenever needed.

Approachability: Prof. Udayan is very approachable, responsive, and motivating. He is excellent when it comes to providing you direction.

Evaluation: My grades: AA in SRE and RnD. Prof. Udayan has a carefully constructed policy (on his website). For BTP1/SRE/RnD level projects, BB- replication, AB- replication + novelty, AA- a significant, meaningful novelty. This differs for BTP2/ DDP. I didn’t care about the evaluation as much and was involved in research work because I didn’t get involved in research for SRE/RnD and would have done it otherwise, too. SRE/RnD was just a tag.

Any advice for anyone considering a project under the same guide? Any other professors working in similar fields?

Prof. Udayan Ganguly is an outstanding professor to work with. You don’t work with the professor; you work with the group. All the members of the group are very responsive and helpful. The research areas in which the MeLoDe group is involved are:

  1. FeRAM, Flash, SOXI, Variability (my subgroup)
  2. RRAM
  3. Algorithms
  4. Circuits

There are many professors in SSD to work with. Everyone is good and approachable. Check the EE website for the domain in SSD you want to work in.