Nano-crystalline ZnO memristor for neuromorphic computing: Resistive switching and conductance modulation

Abstract

In this work, a nano-crystalline (NC) ZnO-based memristor was fabricated to investigate the short-term memory characteristics for reservoir computing systems. The crystalline structure of the ZnO film was confirmed through transmission electron microscopy (TEM) and X-ray diffraction pattern (XRD), while X-ray photoelectron spectroscopy (XPS) confirmed the chemical and bonding states of each element. The NC-ZnO-based memristor exhibited remarkable endurance, enduring more than 200 DC cycles, and had a high to low resistance (RH/RL) ratio of 102. Furthermore, it displayed long data retention of 104 s and consistent resistive switching (RS) with restricted variation in the set and reset voltage, showing its excellent per-formance characteristics. By controlling the pulse amplitude and the time interval between pulses, it was possible to effectively replicate the key features of short-term synaptic plasticity, including potentiation, depression, and paired-pulse depression, through conductance modulation. An artificial neural network (ANN) simulation achieved a pattern recognition accuracy of approximately 90.1% for a 28 x 28-pixel image after 100 training epochs. Based on this extensive study, NC-ZnO-based memristor exhibits immense po-tential as a crucial element in constructing high-performance neuromorphic computing systems.& COPY; 2023 Elsevier B.V. All rights reserved.