Hybrid capture layer synaptic transistors with visible spectrum perception for artificial vision

Abstract

Photosynaptic transistors, which integrate visual perception, processing, and memory functions, have shown great advantages in recognizing and processing visual information and accomplishing complex learning and memory tasks. Developing optoelectronic synapse transistors with superior visible spectrum sensing capabilities can mimic the retina's color information processing characteristics, offering a candidate solution for constructing neuromorphic visual systems. This study designed a floating-gate organic optoelectronic synaptic transistor (FG-OOST), whose capture layer is composed of a mixture of poly(N-vinyl carbazole) (PVK) and poly(9,9-dioctylfluorene-alt-benzothiadiazole) (F8BT). By leveraging the capture and release of photo-generated carriers by the mixed capture layer, along with the efficient carrier separation mechanism generated at the heterojunction interface, the transistor exhibits distinguishable light-dependent response behavior across the visible light spectrum for blue (B), green (G), and red (R) wavelengths, demonstrating exceptional performance in processing visual information in this process. The device successfully simulated photostimulation-induced injury receptors through the application of optoelectronic stimuli, realizing short-term and long-term plasticity, learning, forgetting, and re-learning behaviors, as well as photo-modulated dynamic filtering of basic synaptic plasticity. Complex biological associative learning behaviors and logical functions were simulated using the optoelectronically co-modulated synaptic plasticity. In addition, a 5 × 5 synapse array was constructed to achieve memory storage and image recognition. Neuromorphic computation simulations were performed, achieving a handwritten digit recognition accuracy of 85.5%, which demonstrates its potential for application in the development of next-generation artificial visual systems.