神经形态工程学
材料科学
光电子学
异质结
晶体管
紫外线
计算机科学
电压
物理
人工神经网络
量子力学
机器学习
作者
Yao Ni,Shuo Zhang,Lin Sun,Lu Liu,Huanhuan Wei,Zhipeng Xu,Wenlong Xu,Wentao Xu
标识
DOI:10.1016/j.apmt.2021.101223
摘要
Abstract We demonstrate a photoelectric neuromorphic transistor (PENT), consisting of a hybrid p-i-n heterojunction channel of vertically phase-separated 2,7-dioctyl[1]benzothieno[3,2-b][1]benzothiophene (C8-BTBT) /poly(methyl methacrylate) (PMMA) as a two-dimensional (2D) sheath and highly-aligned array of long continuous zinc oxide (ZnO) nanowires (NWs) as a quasi-one dimensional (Q1D) core. This is the first report of a hybrid p-i-n heterojunction being developed on a NW-based PENT. The device integrates both optical sensing and electrical processing functionalities. In optical sensing, the device could respond to ultraviolet (UV) light with low intensity down to microwatts per square centimeter, representing so far the most UV-sensitive nonvolatile optoelectronic neural device. In electrical-processing, the PENT demonstrates a mechanically immediate switchable plasticity between short-term and long-term by altering charge carriers in the conductive channel; this emulates the selective release of different neurotransmitters, i.e. dopamine and noradrenaline, from a same axon, for pain-avoiding and pleasure-inducing patterns of Skinner Box for the first time. Furthermore, a new strategy for gesture recognition was developed. This approach can be applied to control and design multipurpose neuromorphic systems by combining brain-like processing and artificial sensory nerves.
科研通智能强力驱动
Strongly Powered by AbleSci AI