Journal of Semiconductors, Volume. 46, Issue 6, 061301(2025)

Revolutionizing neuromorphic computing with memristor-based artificial neurons

Yanning Chen1, Guobin Zhang2,3,4, Fang Liu1, Bo Wu1, Yongfeng Deng1, Dawei Gao2,3,4、*, and Yishu Zhang2,3,4、**
Author Affiliations
  • 1Beiiing Smartchip Microelectronics Technology Co., Ltd, Beijing 102200, China
  • 2College of Integrated Circuits, Zhejiang University, Hangzhou 311200, China
  • 3Zhejiang ICsprout Semiconductor Co., Ltd, Hangzhou 310027, China
  • 4ZJU-Hangzhou Global Scientific and Technological Innovation Center, Hangzhou 310027, China
  • show less
    Figures & Tables(5)
    (Color online) (a) 1T1R structure based on volatile diffusive memristors. The transistor is placed in series to obtain better control of the maximum flowing current. (b) Schematic diagram of the arrangement of the silver atoms inside hafnium oxide with switching between LRS and HRS. (c) Schematic diagram of the arrangement of parallel cells based on the 1T1R cell. (d) Examples of filtered temporal current traces collected in the experiment with different frequency. (e) Histograms showing the normalization of ON device counts across various frequencies reveal a shift in the peak towards the right with frequency increase, indicating a rise in the quantity of active devices. (f) Linear mapping of log-spaced frequencies in our systems and in the cochlea. Reprinted from Ref. [26], with permission of Nature.
    (Color online) (a) Multiple types of spiking behavior such as sustained spiking, bursting, phasic spiking, and mixed-mode spiking realized by VO2-based Mott memristors. (b) VO2-based Mott memristors for artificial neurons resembling the structure of the cerebral cortex. Reprinted from Ref. [29], with permission of Nature.
    (Color online) (a) Schematic of the work-cognition tasks realized by reservoir computing systems based on optoelectronic memristive neurons. (b) Schematic of an optoelectronic memristive neuron stimulated by hybrid electrical and optical inputs. Reprinted from Ref. [9], with permission of Science.
    (Color online) (a) Diagram of the structure of the 1TFT1R volatile memristor serves as a time-surface neuron with temporal kernel function. (b) Sample image of the packaged 32 × 36 1T1R memristor array (1 kb) with the TiN/TaOx/HfO2/TiN structure wire boned in a chip holder. (c) PCB subsystem hardware for the audio recognition consists of the 1TFT volatile memristors, 1T1R memristors array, analog circuits, and communication interface. Reprinted from Ref. [69], with permission of Science.
    (Color online) Future directions of memristor-based artificial neurons for neuromorphic applications and strategies to address challenges.
    Tools

    Get Citation

    Copy Citation Text

    Yanning Chen, Guobin Zhang, Fang Liu, Bo Wu, Yongfeng Deng, Dawei Gao, Yishu Zhang. Revolutionizing neuromorphic computing with memristor-based artificial neurons[J]. Journal of Semiconductors, 2025, 46(6): 061301

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category: Research Articles

    Received: Nov. 25, 2024

    Accepted: --

    Published Online: Jun. 30, 2025

    The Author Email: Dawei Gao (DWGao), Yishu Zhang (YSZhang)

    DOI:10.1088/1674-4926/24110006

    Topics