Multifunctional ZnO-based optical memristor for synapse-neuron integration and neuromorphic vision system

Abstract

This work presents a comprehensive analysis of ZnO-based memristive devices integrated with ITO electrodes, focusing on optical and neuromorphic properties. The precise layered structure and chemical composition of the ITO/ZnO/ITO stack were identified by cross-sectional Transmission Electron Microscope (TEM) and energy-dispersive X-ray (EDX) analysis. Synaptic properties were investigated using light stimulation at 405 nm. The device exhibits distinct functionalities depending on the light intensity. At high light intensity, the device exhibited short-term memory (STM) behavior and pair-pulse facilitation (PPF), which are critical for mimicking biological synaptic plasticity. This study also reproduced key features such as threshold, no adaptation, relaxation, and nociceptive sensitization with respect to nociceptor. These results demonstrate the potential of ZnO-based devices as visual pain detectors for advanced neuromorphic systems. Additionally, using reservoir computing (RC) for MNIST image classification reached an accuracy of 97.35%, showing that the device is capable of handling real-time tasks like image recognition and processing. At low light intensity, the device exhibited computational capability at the neuron level through the implementation of the Restricted Boltzmann Machine (RBM) model. By integrating neuromorphic, nociceptive and computational features, this work paves the way for the development of biological system-inspired multifunctional devices.