Thermally Stable Single-Transistor Neuron for Reliable Neuromorphic System

A thermally stable single-transistor neuron (1T-neuron) that uses a vertical nanowire Si MOSFET is demonstrated. The concept of a single-transistor latch (STL) in a MOSFET, which is used to mimic neural firing in a biological neuron, is sensitive to the temperature ( T ) because it generally occurs...

Full description

Saved in:
Bibliographic Details
Published in:IEEE electron device letters 2022-11, Vol.43 (11), p.1-1
Main Authors: Chung, Young-Woo, Han, Joon-Kyu, Jung, Jin-Woo, Yun, Seong-Yun, Yu, Ji-Man, Lee, Mun-Woo, Choi, Yang-Kyu
Format: Article
Language:English
Subjects:
Handwriting recognition
Logic gates
MOSFET
MOSFETs
Nanowires
Neural networks
Neuromorphic computing system
Neuromorphics
Neurons
Semiconductor devices
Silicon
single-transistor latch (STL)
single-transistor neuron (1T-neuron)
spiking neural network (SNN)
Thermal stability
Transistors
Voltage measurement
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:A thermally stable single-transistor neuron (1T-neuron) that uses a vertical nanowire Si MOSFET is demonstrated. The concept of a single-transistor latch (STL) in a MOSFET, which is used to mimic neural firing in a biological neuron, is sensitive to the temperature ( T ) because it generally occurs due to impact ionization (I.I.). As an alternative to the thermally sensitive I.I., thermally insensitive band-to-band tunneling (BTBT) is used to enable the STL to function. Neuronal operation with thermal stability is thereby achieved. It is verified in spiking neural network (SNN) simulations that the recognition ability of handwritten digits in the MNIST dataset was not sensitive to T .
ISSN:0741-3106
1558-0563
DOI:10.1109/LED.2022.3207187