All-Optical Controlled Bidirectional Synaptic Transistors for Motion Perception

Abstract

Bidirectional photoresponsive synaptic devices offer distinct advantages and significant potential for applications in artificial visual systems. However, most existing artificial optical synapses require both optical stimulation and separate electrical control to achieve bidirectional synaptic conductance modulation, thereby increasing processing time and system complexity. In this study, we developed a bidirectional photoresponsive synaptic transistor based on a heterostructure of indium oxide (In2O3) and organic semiconductor, which exhibits positive response to ultraviolet (UV) light and negative response to red-light illumination. Leveraging their reversible conductance tunability, we simulated artificial neural networks for handwritten digit recognition, achieving an accuracy of 94.7%. Furthermore, the synaptic transistors can be utilized for precise motion perception, achieving 100% prediction accuracy.