Influence of Al2O3 layer on InGaZnO memristor crossbar array for neuromorphic applications. (March 2022)

• Record Type: Journal Article
• Authors: Choi, Woo Sik; Jang, Jun Tae; Kim, Donguk; Yang, Tae Jun; Kim, Changwook; Kim, Hyungjin; Kim, Dae Hwan
• Is Part Of: Chaos, solitons and fractals. Volume 156(2022)
• Publication Date: 2022-03
• Subjects: Memristor -- Memristor crossbar array -- Switching Mechanism -- Schottky barrier modulation -- Pattern classification -- Winner-takes-all; Chaotic behavior in systems -- Periodicals; Solitons -- Periodicals; Fractals -- Periodicals; Chaotic behavior in systems; Fractals; Solitons; Periodicals; 003.7
• DOI: 10.1016/j.chaos.2022.111813
• Languages: English
• ISSNs: 0960-0779

Abstract

Highlights: IGZO memristor devices for neuromorphic system. Insertion of Al2 O3 layer for symmetric switching characteristics and low-current operation. Analyzing conduction mechanisms of IGZO memristor with and without Al2 O3 layer. Programming the 10 × 10 IGZO memristor crossbar array with V/2 and V/3 programming schemes. Evaluating the classification performance of the IGZO memristor crossbar array with binarized weight values. Abstract: In this study, we studied the effect of Al2 O3 layer insertion on InGaZnO (IGZO) memristors by fabricating two kinds of crossbar arrays according to the presence or absence of the Al2 O3 layer. It was confirmed that the asymmetric current-voltage ( I - V ) characteristics come from the bias-polarity-dependent Schottky barrier and improving the symmetry of the I - V properties plays an important role in programming the array. The on/off ratio and endurance characteristics of the device are improved by the inserted Al2 O3 layer due to the voltage divide by high resistance of the layer. In addition, a pattern classification is experimentally verified in a 10 × 10 fabricated memristor crossbar array and the improvement of classification rate and power consumption by the insertion of the Al2 O3 layer is discussed with binarized weight values.