Memristor crossbar array for binarized neural networks

Yong Kim; Won Hee Jeong; Son Bao Tran; Hyo Cheon Woo; Jihun Kim; Cheol Seong Hwang; Kyeong-Sik Min; Byung Joon Choi

doi:10.1063/1.5092177

AIP Advances (Apr 2019)

Memristor crossbar array for binarized neural networks

Yong Kim,
Won Hee Jeong,
Son Bao Tran,
Hyo Cheon Woo,
Jihun Kim,
Cheol Seong Hwang,
Kyeong-Sik Min,
Byung Joon Choi

Affiliations

Yong Kim: Department of Materials Science and Engineering, Seoul National University of Science and Technology, Seoul 01811, Korea
Won Hee Jeong: Department of Materials Science and Engineering, Seoul National University of Science and Technology, Seoul 01811, Korea
Son Bao Tran: School of Electrical Engineering, Kookmin University, Seoul 02707, Korea
Hyo Cheon Woo: Department of Materials Science and Engineering and Inter-University Semiconductor Research Center, College of Engineering, Seoul National University, Seoul 08826, Korea
Jihun Kim: Department of Materials Science and Engineering and Inter-University Semiconductor Research Center, College of Engineering, Seoul National University, Seoul 08826, Korea
Cheol Seong Hwang: Department of Materials Science and Engineering and Inter-University Semiconductor Research Center, College of Engineering, Seoul National University, Seoul 08826, Korea
Kyeong-Sik Min: School of Electrical Engineering, Kookmin University, Seoul 02707, Korea
Byung Joon Choi: Department of Materials Science and Engineering, Seoul National University of Science and Technology, Seoul 01811, Korea

DOI: https://doi.org/10.1063/1.5092177
Journal volume & issue: Vol. 9, no. 4
pp. 045131 – 045131-5

Abstract

Read online

Memristor crossbar arrays were fabricated based on a Ti/HfO2/Ti stack that exhibited electroforming-free behavior and low device variability in a 10 x 10 array size. The binary states of high-resistance-state and low-resistance-state in the bipolar memristor device were used for the synaptic weight representation of a binarized neural network. The electroforming-free memristor was confirmed as being suitable as a binary synaptic device because of its higher device yield, lower variability, and less severe malfunction (for example, hard break-down) than the electroformed memristors based on a Ti/HfO2/Pt structure. The feasibly working binarized neural network adopting the electroforming-free binary memristors was demonstrated through simulation.

Published in AIP Advances

ISSN: 2158-3226 (Online)
Publisher: AIP Publishing LLC
Country of publisher: United States
LCC subjects: Science: Physics
Website: http://aipadvances.aip.org/

About the journal