Hf0.5Zr0.5O2-based ferroelectric bionic electronic synapse device with highly symmetrical and linearity weight modification
Bionic electronic synapses with weight-modulation properties are promising alternative candidates when compared with traditional CMOS based neuromorphic hardware when building at-scale neural networks for efficient neuromorphic computing. In this Letter, due to the partial polarisation switching of...
Saved in:
Published in | Electronics letters Vol. 56; no. 16; pp. 840 - 843 |
---|---|
Main Authors | , , , , , , , , |
Format | Journal Article |
Language | English |
Published |
The Institution of Engineering and Technology
06.08.2020
|
Subjects | |
Online Access | Get full text |
ISSN | 0013-5194 1350-911X 1350-911X |
DOI | 10.1049/el.2020.0423 |
Cover
Loading…
Summary: | Bionic electronic synapses with weight-modulation properties are promising alternative candidates when compared with traditional CMOS based neuromorphic hardware when building at-scale neural networks for efficient neuromorphic computing. In this Letter, due to the partial polarisation switching of multi-domain properties in polycrystalline Zr-doped HfO2 (HZO) ferroelectric materials, a capacitor-based bionic synapse was proposed, which showed multi-level capacitance modulation characteristics. By switching the voltage-controlled polarisation, the capacitance was gradually modulated via the variation of net polarisation vectors in the HZO layer. The potentiation and depression properties can be achieved with better linearity, symmetry and multiple states. This Letter laid the groundwork for realising the high-density integration neuromorphic computing system by using ferroelectric capacitors as the bionic synapse. |
---|---|
Bibliography: | Guo‐Liang Tian, Jin‐Shun Bi, Gao‐Bo Xu, Yan‐Nan Xu, Xue‐Qin Yang, Hua‐Xiang Yin and Yong‐Liang Li: Also with The University of Chinese Academy of Sciences, Beijing 100049, People's Republic of China |
ISSN: | 0013-5194 1350-911X 1350-911X |
DOI: | 10.1049/el.2020.0423 |