Investigation of the NbOx phase change self-selective memristor-based photoelectric synapse features

Abstract

A nonlinear resistor with memory function, the memristor, is regarded as an ideal tool for simulating neuronal synapses. Due to the crosstalk problem caused by the bypass current, the scale of the memristor array is limited. A self-selective memristor is regarded as one of the solutions to the latent current problem. This study describes a self-selective NbO_x memristor that has improved sneak current suppression. Both light and electrical stimulation were used to replicate the biological synapse, and both had a 4-bit computing accuracy under such conditions. Using fitting calculations, a phase change-based resistance change mechanism model was built. The array's ability to write and erase data while simultaneously regulating photoelectricity demonstrated the benefits and possibilities of this tool for neuromorphic computing.