基于二维有机–无机异质结突触晶体管的研究

Research on Synaptic Transistors Based on 2D Organic-Inorganic Heterojunctions

摘要: 随着传统冯·诺依曼架构因“存–算分离”导致的能效瓶颈问题日益突出，神经形态计算通过模拟生物突触可塑性机制，为突破现有计算范式提供了新思路。本文提出一种基于二硫化钼(MoS₂)-并五苯有机–无机异质结的光电突触晶体管，旨在实现光–电协同调控的仿生突触功能。实验结果表明，该晶体管可模拟多种生物突触行为，包括成对脉冲促进(PPF)及短期记忆(STM)向长期记忆(LTM)的过渡。在460 nm光刺激下，器件表现出高响应度(光强1.464 mW/cm²时PPF指数达158.2%)和非易失性记忆特性，且通过光脉冲持续时间和次数可调控记忆保持时间，模拟类脑学习模式。此外，结合电脉冲“擦除”功能，器件成功实现光写入与电擦除的协同操作。本研究为开发柔性、低功耗神经形态光电子器件提供了新策略，在仿生视觉系统与智能感知领域具有重要应用潜力。

Abstract: With the increasing prominence of energy efficiency bottlenecks caused by the “memory-process separation” in traditional von Neumann architectures, neuromorphic computing has provided new insights to transcend current computational paradigms by emulating biological synaptic plasticity mechanisms. This study proposes an optoelectronic synaptic transistor based on a molybdenum disulfide (MoS₂)-pentacene organic-inorganic heterojunction, aiming to achieve bio-inspired synaptic functions through optoelectronic co-regulation. Experimental results demonstrate that the transistor can mimic diverse biological synaptic behaviors, including paired-pulse facilitation (PPF) and the transition from short-term memory (STM) to long-term memory (LTM). Under 460 nm optical stimulation, the device exhibits high responsivity (PPF index of 158.2% at 1.464 mW/cm²) and non-volatile memory characteristics. The memory retention time can be modulated by adjusting optical pulse duration and repetition frequency, simulating brain-like learning patterns. Furthermore, by integrating an electrical “erase” function, the device successfully realizes synergistic operations of optical writing and electrical erasing. This work provides a novel strategy for developing flexible, low-power neuromorphic optoelectronic devices, with significant application potential in bio-inspired vision systems and intelligent sensing.