Ion-gating synaptic transistors with long-term synaptic weight modulation

Neuromorphic devices that emulate the human brain are required for efficient computing systems. To develop efficient neuromorphic devices, artificial synapses that are capable of linear and symmetric synaptic weight updates are necessary. Artificial synapses that exploit ion dynamics are suitable fo...

Full description

Saved in:
Bibliographic Details
Published in:Journal of materials chemistry. C, Materials for optical and electronic devices Materials for optical and electronic devices, 2021-04, Vol.9 (16), p.5396-542
Main Authors: Park, Youngjun, Kim, Min-Kyu, Lee, Jang-Sik
Format: Article
Language:English
Subjects:
Artificial neural networks
Electrolytes
Handwriting recognition
High resistance
Ion dynamics
Ion injection
Low resistance
Modulation
Semiconductor devices
Synapses
Transistors
Weight
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:Neuromorphic devices that emulate the human brain are required for efficient computing systems. To develop efficient neuromorphic devices, artificial synapses that are capable of linear and symmetric synaptic weight updates are necessary. Artificial synapses that exploit ion dynamics are suitable for achieving these properties, but stable and long-term synaptic weight modulation is difficult to be achieved because ions can be easily dissipated at the interfaces or in electrolytes. To prevent spontaneous ion dissipation, we design synaptic transistors that operate by ion injection into the channel layer; this process allows long-term synaptic weight updates. We also use a threshold switch as an access device for synaptic transistors. The threshold switch shows low resistance during weight updates of a synapse, and high resistance otherwise to prevent self-discharge of injected ions into the channel layer, which can improve the data retention of synaptic transistors. Linear and symmetric synaptic weight updates are achieved with a large dynamic range (>20), which enables high recognition accuracy (91.4%) of handwritten digits by artificial neural networks. These results provide insights into applications of synaptic transistors for future neuromorphic systems. This paper presents synaptic transistors that show long-term synaptic weight modulation via injection of ions. Linear and symmetric weight update is achieved, which enables high recognition accuracy in artificial neural networks.
ISSN:2050-7526
2050-7534
DOI:10.1039/d1tc00048a