Scholarlink

<p>As an emerging computing paradigm, neuromorphic computing seeks to achieve highly efficient and low-power information processing by emulating the intricate structure and sophisticated functions of biological nervous systems. In this study, an Au/HfO₂/Bi<sub>2</sub>FeCrO<sub>6</sub> (BFCO)/HfO₂/FTO artificial photomemristor was fabricated using the sol–gel method. This device exhibits excellent memristive characteristics at room temperature, with a stable switching ratio of at least 10<sup>2</sup>. Moreover, the device demonstrates exceptional optoelectronic synaptic properties effectively mimics the essential synaptic behaviors observed in biological systems under electrical stimulation. These behaviors include STP (short-term plasticity), PPD (paired-pulse depression), LTP (long-term plasticity), and STDP (spike-timing-dependent plasticity). Through precise control of the power, duration, and number of light pulses during optical stimulation, the device can seamlessly transition between short-term and long-term plasticity states. In the ultimate phase of the study, an advanced stochastic adaptive method was employed to train an integrated CNN (Convolutional Neural Network), thereby assessing the device's neuromorphic computing capabilities. The results are highly encouraging: the device achieved a recognition accuracy of 97.6% on the MNIST dataset. This remarkable performance underscores the device's significant potential for practical applications in the burgeoning domain of brain-like computing.</p> Graphical abstract <p></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

The sandwich structure with BFCO in HfO2 shows photoelectric synaptic properties for neuromorphic computing

  • 期刊论文

  • 发表日期：

    2025年7月14日

查看全文
  • Weiyu Li,
  • Yanping Jiang,
  • Jianhao Feng,
  • Xingui Tang,
  • Zhenhua Tang,
  • Xiaobin Guo,
  • Wenhua Li,
  • Yichun Zhou

摘要

As an emerging computing paradigm, neuromorphic computing seeks to achieve highly efficient and low-power information processing by emulating the intricate structure and sophisticated functions of biological nervous systems. In this study, an Au/HfO₂/Bi2FeCrO6 (BFCO)/HfO₂/FTO artificial photomemristor was fabricated using the sol–gel method. This device exhibits excellent memristive characteristics at room temperature, with a stable switching ratio of at least 102. Moreover, the device demonstrates exceptional optoelectronic synaptic properties effectively mimics the essential synaptic behaviors observed in biological systems under electrical stimulation. These behaviors include STP (short-term plasticity), PPD (paired-pulse depression), LTP (long-term plasticity), and STDP (spike-timing-dependent plasticity). Through precise control of the power, duration, and number of light pulses during optical stimulation, the device can seamlessly transition between short-term and long-term plasticity states. In the ultimate phase of the study, an advanced stochastic adaptive method was employed to train an integrated CNN (Convolutional Neural Network), thereby assessing the device's neuromorphic computing capabilities. The results are highly encouraging: the device achieved a recognition accuracy of 97.6% on the MNIST dataset. This remarkable performance underscores the device's significant potential for practical applications in the burgeoning domain of brain-like computing.

Graphical abstract