材料科学
光电子学
神经形态工程学
人工神经网络
计算机科学
人工智能
作者
Longlong Jiang,Lu Yang,Yiming Yuan,Qingshan Zhu,Wenjie Wu,Xiaohong Wang,Wentao Xu,Longzhen Qiu
标识
DOI:10.1021/acsmaterialslett.4c00448
摘要
We demonstrate a flexible optoelectronic synaptic device that uses a donor–acceptor conjugated polymer with excellent electrical properties and a highly optically active block copolymer to form a heterojunction as the active layer, enabling ultrabroad-spectrum perception from deep ultraviolet (DUV), to visible (vis), and to near-infrared (NIR) for the first time. Essential synaptic behaviors have been successfully simulated, such as learning experience behavior simulation, international Morse code communication, and high-pass filtering. The synaptic device shows stable synaptic properties when bent at different radii of curvature and at different numbers of bending cycles. When the device is extremely deformed under a radius of curvature of 4 mm, the postsynaptic current is still maintained above 84%. Moreover, an artificial reflex arc was constructed using the flexible optoelectronic synapse as a key information processing unit. Recognition of digits was achieved by constructing an artificial neural network under DUV light stimulation, achieving the highest recognition rate so far, up to 94%. This work demonstrates a methodology to prepare flexible synaptic devices with tunable synaptic plasticity and broad-spectrum perception that is potentially applicable to building flexible neuromorphic electronics.
科研通智能强力驱动
Strongly Powered by AbleSci AI