Scholarlink

<p>Neuromorphic computing, drawing inspiration from the human brain’s architecture and functionality, offers a transformative approach to surmount the limitations of conventional von Neumann systems through in-memory processing and minimal energy dissipation. Herein, we fabricated a BTT-H/CsPbBr<sub>3</sub> heterojunction-based device exhibiting exceptional operational characteristics. These include a high ON/OFF current ratio (∼10<sup>3.8</sup>), a low set voltage (+0.60 V), robust cycling endurance (448 cycles), and superior retention capability (1 × 10<sup>4</sup> s). These attributes stem from the synergistic interplay between the built-in electric field and interfacial interactions, which optimize carrier separation and transport, facilitate silver ion migration, and promote the formation of stable and reversible conduction pathways. Moreover, the device faithfully emulates essential synaptic behaviors, including excitatory postsynaptic current, spike-voltage-dependent plasticity, paired-pulse facilitation, long-term potentiation, and long-term depression. Notably, the incorporation of CsPbBr<sub>3</sub> markedly improves the intrinsic conductivity of the covalent organic framework (COF) layer, reducing the nonlinearity factor of the conductance modulation from 0.77 to 0.27. When integrated into a Hopfield neural network, the device achieved an image recognition accuracy of 98.4%, establishing a new benchmark for COF-based synaptic devices. This study pioneers a novel paradigm for COF utilization in neuromorphic hardware, heralding advancements in efficient, intelligent computing systems.</p>

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

High linearity COF/CsPbBr3 heterojunction neuromorphic memristors for ultrahigh precision image recognition

  • 期刊论文

  • 发表日期：

    2026年4月3日

查看全文
  • Ziyue Yu,
  • Yang Li,
  • Pan-Ke Zhou,
  • Zhipeng Luo,
  • Yifan Lin,
  • Jiawen Qiu,
  • Fengtao Zhang,
  • Chao Lin,
  • Qianqian Tan,
  • Tao Zeng,
  • Chaoxing Wu,
  • Xiong Chen

摘要

Neuromorphic computing, drawing inspiration from the human brain’s architecture and functionality, offers a transformative approach to surmount the limitations of conventional von Neumann systems through in-memory processing and minimal energy dissipation. Herein, we fabricated a BTT-H/CsPbBr3 heterojunction-based device exhibiting exceptional operational characteristics. These include a high ON/OFF current ratio (∼103.8), a low set voltage (+0.60 V), robust cycling endurance (448 cycles), and superior retention capability (1 × 104 s). These attributes stem from the synergistic interplay between the built-in electric field and interfacial interactions, which optimize carrier separation and transport, facilitate silver ion migration, and promote the formation of stable and reversible conduction pathways. Moreover, the device faithfully emulates essential synaptic behaviors, including excitatory postsynaptic current, spike-voltage-dependent plasticity, paired-pulse facilitation, long-term potentiation, and long-term depression. Notably, the incorporation of CsPbBr3 markedly improves the intrinsic conductivity of the covalent organic framework (COF) layer, reducing the nonlinearity factor of the conductance modulation from 0.77 to 0.27. When integrated into a Hopfield neural network, the device achieved an image recognition accuracy of 98.4%, establishing a new benchmark for COF-based synaptic devices. This study pioneers a novel paradigm for COF utilization in neuromorphic hardware, heralding advancements in efficient, intelligent computing systems.