神经形态工程学
材料科学
记忆电阻器
灵活性(工程)
纳米技术
实现(概率)
计算机科学
基质(水族馆)
接口(物质)
人工神经网络
人工智能
突触
电子工程
复合材料
工程类
神经科学
生物
统计
海洋学
数学
毛细管数
毛细管作用
地质学
作者
Qifeng Lu,Yizhang Xia,Fuqin Sun,Yixiang Shi,Yinchao Zhao,Shuqi Wang,Ting Zhang
标识
DOI:10.1002/admt.202300059
摘要
Abstract The realization of a neuromorphic system with high power efficiency and flexibility is of great significance in constructing bionic interaction systems. Although great improvement has been achieved, the systems based on hardware level remain with great challenges in device performance and stability under deformation due to the inferior interface quality. As a result, the neuromorphic systems are incompatible and incomparable with the biological ones. Therefore, an in situ growth method to synthesize the dielectric materials on flexible substrates, which can be used for the fabrication of flexible memristor‐type artificial synapses, is proposed to overcome the issues. A superiority in interface quality between the synthesized Cu 3 (BTC) 2 Metal‐organic framework (MOF) material and the Cu foil substrate is observed benefiting from the in situ growth method, which will contribute to the improvement in device stability under deformation. In addition, the device shows continuous conductance states due to the highly accessible sites of the MOF. As a proof of concept, an artificial neural network based on the synaptic characteristics of the artificial synapse is designed, and high recognition accuracy of >80% is achieved, even with 30% allows conductance states. These results indicate that the proposed artificial synapse has great potential in the application of neuromorphic systems.
科研通智能强力驱动
Strongly Powered by AbleSci AI