Van der Waals Antiferroelectric CuCrP <sub>2</sub> S <sub>6</sub> ‐Based Artificial Synapse for High‐Precision Neuromorphic Computation
Zhipeng Yu, Qinan Wang, Tianle Zeng, Kun Ye, Houjian Zhou, Zishuo Han, Yuxuan Zeng, Bin Fang, Weiming Lv, Lin Geng, Chun Zhao, Zhongyuan Liu, Zhongming Zeng
Abstract
Abstract 2D van der Waals heterostructure‐based artificial synapses have emerged as a compelling platform for next‐generation neuromorphic systems, owing to their tunable electrical conductivity and layer‐engineered functionality through controlled stacking of 2D materials. In this work, an engineered SnS₂/h‐BN/CuCrP₂S₆ van der Waals antiferroelectric field‐effect transistor (AFe‐FET) is presented that implements synaptic weight modulation through the synergistic interplay of charge trapping dynamics and electric‐field‐controlled ferroelectric polarization switching. The AFe‐FET architecture successfully emulates essential neuroplasticity features, including paired‐pulse facilitation, short‐term plasticity, and long‐term plasticity. The device exhibits exceptional long‐term potentiation (LTP) and long‐term depression (LTD), with an ultralow nonlinearity coefficient of 1.1 for both LTP and LTD operations, high symmetricity (30), and broad dynamic range (G max /G min = 10). The AFe‐FET‐based neuromorphic system demonstrates an outstanding computational efficacy, i.e. a classification accuracy of 97.7% on the MNIST benchmark. Furthermore, implementing reservoir computing architectures enables cognitive process emulation, attaining 94.7% task recognition accuracy in brain‐inspired decision‐making simulations. This investigation establishes new design paradigms for high‐fidelity synaptic devices, providing a strategy for energy‐efficient neuromorphic computing systems with biological plausibility.