Photo‐Induced Valence Changed Memristor Based on WO 3 and Polyvinyl Alcohol Nanocomposites for In‐Sensor Reservoir Computing

ABSTRACT In‐sensor reservoir computing based on optoelectronic synapses is a promising candidate to develop high‐efficiency neuromorphic vision, owing to their ability to directly sensing and processing optical signals. Here, a photo‐induced valence change memristor based on WO 3 nanoparticles and polyvinyl alcohol (WO 3 @PVA) is developed, in which PVA functions as a proton donor, effectively enhancing the photocurrent response and photocurrent relaxation time of the device. Several fundamental synaptic functions have been emulated by utilizing ultraviolet light stimulation, including excitatory postsynaptic currents (EPSCs), short‐term/long‐term synaptic plasticity (STP/LTP) and learning‐experience behaviors. In situ x‐ray photoelectron spectroscopy analysis confirms that the photo‐induced valence changed in tungsten oxide is the intrinsic mechanism underlying memristive switching. Furthermore, an in‐sensor reservoir computing system was constructed by leveraging the photo‐induced nonlinear dynamics and short‐term memory characteristics of device, enabling static image classification and motion recognition with high precision. Our work provides a feasible pathway toward the development of optoelectronic synapse devices for application in high‐efficiency neuromorphic visual systems.