Two-dimensional (2D) material MoS2 holds great promise for neuromorphic applications due to its outstanding electrically tunable resistive switching properties. However, challenges remain in achieving reliable, durable, and high-retention performance in MoS2-based memristors. In this work, we present a heterostructure memristor, Au/Ag/MoS2/HfO2/Pt, which synergistically combines the advantages of 2D MoS2 with the high stability of HfO2. The device demonstrates reconfigurable performance, including multi-level resistance states enabled by current limitation, a high on/off switching ratio of 10⁸, and an ultrafast response time of 19 ns. Furthermore, the memristor successfully emulates both non-volatile synaptic plasticity—such as long-term potentiation (LTP), long-term depression (LTD), paired-pulse facilitation (PPF), and spike-timing-dependent plasticity (STDP)—as well as volatile neuronal functions like leaky integrate-and-fire (LIF) behavior. This study provides a novel pathway for reconfigurable and multifunctional MoS2-based memristor applications in brain-inspired computing systems and demonstrates the immense potential of neuromorphic computing.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Neuromorphic Computing Based on MoS2/HfO2 Memristors

  • Yi Sun,
  • Jing You,
  • Wenxuan Xu,
  • Ziyue Li,
  • Jinxia Xu,
  • Xingjuan Song

摘要

Two-dimensional (2D) material MoS2 holds great promise for neuromorphic applications due to its outstanding electrically tunable resistive switching properties. However, challenges remain in achieving reliable, durable, and high-retention performance in MoS2-based memristors. In this work, we present a heterostructure memristor, Au/Ag/MoS2/HfO2/Pt, which synergistically combines the advantages of 2D MoS2 with the high stability of HfO2. The device demonstrates reconfigurable performance, including multi-level resistance states enabled by current limitation, a high on/off switching ratio of 10⁸, and an ultrafast response time of 19 ns. Furthermore, the memristor successfully emulates both non-volatile synaptic plasticity—such as long-term potentiation (LTP), long-term depression (LTD), paired-pulse facilitation (PPF), and spike-timing-dependent plasticity (STDP)—as well as volatile neuronal functions like leaky integrate-and-fire (LIF) behavior. This study provides a novel pathway for reconfigurable and multifunctional MoS2-based memristor applications in brain-inspired computing systems and demonstrates the immense potential of neuromorphic computing.