Volume 21, Issue 7 2409510
Research Article

Contact-Engineered Oxide Memtransistors for Homeostasis-Based High-Linearity and Precision Neuromorphic Computing

San Nam

San Nam

School of Advanced Materials Science and Engineering, Sungkyunkwan University, Suwon, 16419 Republic of Korea

Search for more papers by this author
Donghyun Kang

Donghyun Kang

School of Advanced Materials Science and Engineering, Sungkyunkwan University, Suwon, 16419 Republic of Korea

Search for more papers by this author
Seong-Pil Jeon

Seong-Pil Jeon

School of Electrical and Electronics Engineering, Chung-Ang University, Seoul, 06974 Republic of Korea

Search for more papers by this author
Dayul Nam

Dayul Nam

School of Electrical and Electronics Engineering, Chung-Ang University, Seoul, 06974 Republic of Korea

Search for more papers by this author
Jeong-Wan Jo

Jeong-Wan Jo

Electrical Engineering Division, Department of Engineering, University of Cambridge, Cambridge, CB2 1PZ United Kingdom

Search for more papers by this author
Sang-Joon Park

Sang-Joon Park

Department of Electronic Materials Engineering, Kwangwoon University, Seoul, 01897 Republic of Korea

Search for more papers by this author
Jiyong Lee

Jiyong Lee

School of Advanced Materials Science and Engineering, Sungkyunkwan University, Suwon, 16419 Republic of Korea

Search for more papers by this author
Myung-Gil Kim

Myung-Gil Kim

School of Advanced Materials Science and Engineering, Sungkyunkwan University, Suwon, 16419 Republic of Korea

Search for more papers by this author
Tae-Jun Ha

Corresponding Author

Tae-Jun Ha

Department of Electronic Materials Engineering, Kwangwoon University, Seoul, 01897 Republic of Korea

E-mail: [email protected][email protected]; [email protected]

Search for more papers by this author
Sung Kyu Park

Corresponding Author

Sung Kyu Park

School of Electrical and Electronics Engineering, Chung-Ang University, Seoul, 06974 Republic of Korea

E-mail: [email protected][email protected]; [email protected]

Search for more papers by this author
Yong-Hoon Kim

Corresponding Author

Yong-Hoon Kim

School of Advanced Materials Science and Engineering, Sungkyunkwan University, Suwon, 16419 Republic of Korea

E-mail: [email protected][email protected]; [email protected]

Search for more papers by this author
First published: 05 January 2025
Citations: 2

Abstract

Homeostasis is essential in biological neural networks, optimizing information processing and experience-dependent learning by maintaining the balance of neuronal activity. However, conventional two-terminal memristors have limitations in implementing homeostatic functions due to the absence of global regulation ability. Here, three-terminal oxide memtransistor-based homeostatic synapses are demonstrated to perform highly linear synaptic weight update and enhanced accuracy in neuromorphic computing. Particularly, by leveraging the gate control of contact-engineered indium-gallium-zinc-oxide (IGZO) memtransistor, synaptic weight scaling is enabled for high-linearity and precision neuromorphic computing. Moreover, sinusoidal control of gate voltage is demonstrated, possibly enabling the emulation of higher-order synaptic functions. The device structure of IGZO memtransistor is optimized regarding the source/drain electrode materials and an interfacial layer inserted between the IGZO channel and source electrode. As a result, memtransistors exhibiting high current switching ratio of >104 and reliable endurance characteristics are obtained. Furthermore, through the adaptation of synaptic scaling, emulating the homeostasis, non-linearity values of 0.01 and −0.01 are achieved for potentiation and depression, respectively, exhibiting a recognition accuracy of 91.77% for digit images. It is envisioned that the contact-engineered IGZO memtransistors hold significant promise for implementing the homeostasis in neuromorphic computing for high linearity and high efficiency.

Conflict of Interest

The authors declare no conflict of interest.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.

The full text of this article hosted at iucr.org is unavailable due to technical difficulties.