Photosensitive resistive switching in parylene–PbTe nanocomposite memristors for neuromorphic computing

Abstract

Resistive switching (RS) memory devices with incorporated capabilities of in situ data sensing, storing and processing are promising for artificial intelligence applications. In this respect, controlling resistance not only by electrical but also optical stimulations provides attractive opportunities for the development of novel neuromorphic sensing and computing systems. Here, we demonstrate the RS of Cu/parylene–PbTe/ITO memristive devices and the dependence of RS on optical excitation for efficient neuromorphic computing with high classification accuracy. The main memristive characteristics (multilevel resistive states, RS voltages, endurance, retention, RS time, RS energy, etc.) are evaluated with account of temporal and spatial variations. Additionally, the devices demonstrate a range of synaptic plasticity behaviors, such as spike-timing (amplitude, width)-dependent plasticity, long-term potentiation and depression. A qualitative model that describes photosensitive RS and takes into account the influence of photogenerated charge carriers on conductive filament growth is proposed based on the experimental results. This work presents an appealing approach towards the development of photosensitive memristive devices for upcoming neuromorphic sensing and computing systems.