Low-Power Optoelectronic Synaptic Transistors with Multimodal Neuromorphic Computation and Retinal-Inspired Multiband Optical Binary Communication.

Abstract

Biomimetic neuromorphic optoelectronics exude tempting attraction in multimodal interaction and visual applications because of their capability of integrating sensing, memorizing, and processing in a single device. Herein, a natural dextran film that is intrinsically green and transparent is employed as the dielectric of the optoelectronic synaptic transistors (OSTs). The resulting dextran-OSTs that operate at an ultralow energy consumption (15.89 aJ) exhibit multimodal neuromorphic computation ability with excellent synaptic plasticity, including pair-pulse facilitation (PPF, as high as 494%), spike voltage/frequency/duration/number-dependent plasticity, and a high recognition accuracy of 89.95% by handwritten digital datasets. Furthermore, the device exhibits visual self-adaptation ability and audiovisual fusion effect, showcasing the immense potential in self-adaptation and synergy sensing. More importantly, the dextran-OSTs can significantly advance the capabilities of binary optical information processing and memorizing. This demonstrates the great advantages of dextran-OSTs in multimodal neuromorphic computation, visual self-adaptation, synergy sensing, and multiband optical communication.