神经形态工程学
材料科学
光电子学
异质结
纳米技术
瓶颈
冯·诺依曼建筑
计算机科学
带宽(计算)
人工神经网络
人工智能
电信
嵌入式系统
操作系统
作者
Yilin Sun,Yingtao Ding,Dan Xie
标识
DOI:10.1002/adfm.202105625
摘要
Abstract Neuromorphic devices provide a hardware platform to implement synaptic functions into artificial electronic devices, which opens a new way to overcome the von Neumann bottleneck from the device level. Optoelectronic synaptic devices are expected to break the limitations of electrically stimulated synapses due to wider bandwidth, higher speed, and lower crosstalk. However, most optoelectronic synaptic devices are enabled by the defect‐dominant photo‐generated carrier trapping/de‐trapping. Therefore, designable device structure and controllable synaptic functions are urgently desirable in optoelectronic synaptic devices. Among various functional materials, low‐dimensional materials exhibit excellent optical and electrical properties and can be easily applied to build van der Waals (vdW) heterostructures with ideal surface characteristics. Herein, the basic morphology and characteristics of low‐dimensional materials have been introduced and the typical constitution of mixed‐dimensional (MD) vdW heterostructures has been reviewed to highlight their unique light‐matter interaction. Then, optoelectronic synaptic devices are classified into three categories by the role of light as input, modulated and output signals based on different photoelectric conversion mechanisms. Furthermore, a bridge between neuromorphic devices and practical applications is established to illustrate their potential in neuromorphic systems. Finally, great challenges and possible study directions are presented to guide the development of MD vdW heterostructures in future neuromorphic systems.
科研通智能强力驱动
Strongly Powered by AbleSci AI