Scholarlink

<p>The growing gap between the rapidly increasing demand for computing power and the slowing improvements in computing speed is becoming more noticeable within the von Neumann architecture. Ferroelectric materials, such as hafnium-based ferroelectrics and two-dimensional (2D) van der Waals ferroelectrics, are promising for neuromorphic computing because the partial ferroelectric domain switching behaviour can emulate the temporal dynamics of biological neurons and synapses. Because ferroelectric devices are driven by electric fields, their writing energy is much lower than that of other efficient materials used for memory, such as phase change memory and resistive random-access memory. In this Review, we discuss the advances in ferroelectric neuromorphic devices and arrays, and their in-sensor applications. We summarize the device structure and principles of ferroelectric synaptic devices and neuronal circuits. Furthermore, we emphasize the key role of ferroelectric devices in building efficient and scalable synapse and neuron arrays, including topologies of various structures, the potential for more physical domain computing and high-density 3D integration. Finally, we discuss ferroelectric materials as a key component in supporting workloads that are unattainable using complementary metal-oxide semiconductor (CMOS)-based memory technology.</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

Ferroelectric-based neuromorphic memory devices for bio-inspired computing

  • 期刊论文

  • 发表日期：

    2025年10月24日

查看全文
  • Yihan Liu,
  • Weiyi Tang,
  • Jinhua Zeng,
  • Chongyang Bai,
  • Keji Zhou,
  • Xumeng Zhang,
  • Qi Liu,
  • Zhangcheng Huang,
  • Guangjian Wu,
  • Jianlu Wang

摘要

The growing gap between the rapidly increasing demand for computing power and the slowing improvements in computing speed is becoming more noticeable within the von Neumann architecture. Ferroelectric materials, such as hafnium-based ferroelectrics and two-dimensional (2D) van der Waals ferroelectrics, are promising for neuromorphic computing because the partial ferroelectric domain switching behaviour can emulate the temporal dynamics of biological neurons and synapses. Because ferroelectric devices are driven by electric fields, their writing energy is much lower than that of other efficient materials used for memory, such as phase change memory and resistive random-access memory. In this Review, we discuss the advances in ferroelectric neuromorphic devices and arrays, and their in-sensor applications. We summarize the device structure and principles of ferroelectric synaptic devices and neuronal circuits. Furthermore, we emphasize the key role of ferroelectric devices in building efficient and scalable synapse and neuron arrays, including topologies of various structures, the potential for more physical domain computing and high-density 3D integration. Finally, we discuss ferroelectric materials as a key component in supporting workloads that are unattainable using complementary metal-oxide semiconductor (CMOS)-based memory technology.