Scholarlink

<p>Developing versatile and reliable memristive devices is crucial for advancing future memory and computing architectures. The years of intensive research have still not reached and demonstrated their full horizon of capabilities, and new concepts are essential for successfully using the complete spectra of memristive functionalities for industrial applications. Here, we introduce two-terminal ohmic memristor, characterized by a different type of switching defined as filament conductivity change mechanism (FCM). The operation is based entirely on localized electrochemical redox reactions, resulting in essential advantages such as ultra-stable binary and analog switching, broad voltage stability window, high temperature stability, high switching ratio and good endurance. The multifunctional properties enabled by the FCM can be effectively used to overcome the catastrophic forgetting problem in conventional deep neural networks. Our findings represent an important milestone in resistive switching fundamentals and provide an effective approach for designing memristive system, expanding the horizon of functionalities and neuroscience applications.</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

Electrochemical ohmic memristors for continual learning

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年3月8日

查看全文
  • Shaochuan Chen,
  • Zhen Yang,
  • Heinrich Hartmann,
  • Astrid Besmehn,
  • Yuchao Yang,
  • Ilia Valov

摘要

Developing versatile and reliable memristive devices is crucial for advancing future memory and computing architectures. The years of intensive research have still not reached and demonstrated their full horizon of capabilities, and new concepts are essential for successfully using the complete spectra of memristive functionalities for industrial applications. Here, we introduce two-terminal ohmic memristor, characterized by a different type of switching defined as filament conductivity change mechanism (FCM). The operation is based entirely on localized electrochemical redox reactions, resulting in essential advantages such as ultra-stable binary and analog switching, broad voltage stability window, high temperature stability, high switching ratio and good endurance. The multifunctional properties enabled by the FCM can be effectively used to overcome the catastrophic forgetting problem in conventional deep neural networks. Our findings represent an important milestone in resistive switching fundamentals and provide an effective approach for designing memristive system, expanding the horizon of functionalities and neuroscience applications.