神经形态工程学
记忆电阻器
材料科学
保形涂层
钙钛矿(结构)
电阻随机存取存储器
纳米尺度
光电子学
纳米技术
电阻式触摸屏
帕利烯
非易失性存储器
薄膜
电介质
气凝胶
电压
卤化物
化学气相沉积
涂层
柔性电子器件
人工神经网络
电子工程
共形映射
沉积(地质)
纳米孔
MNIST数据库
作者
Qianwen Zhang,Fangchao Li,Ruijie Bu,Jianlong Shao,Tikai Zhang,Lijuan Ge,Bing Wang,Guodong Zhang,Yi Zhang,Shijie Chen,Shiqing Zhao,Chunlan Ma,Gaoyuan Chen,Cheng Zhang,Yang Li
摘要
ABSTRACT Halide perovskites are promising memristive materials for neuromorphic computing, yet their practical deployment in hardware is severely hindered by instability against moisture, oxygen, and reactive metallic ions. Here, we develop an ultra‐robust perovskite memristor with all‐weather adaptability protected by a ∼100 nm conformal polymeric nanoscale coating. Deposited by low‐temperature, solvent‐free chemical vapor deposition (CVD) technique, the polymeric Parylene serves as an encapsulation, interfacial passivation, and homogeneous dielectric layer. The memristor exhibits excellent bipolar resistive switching under ambient conditions with high uniformity, an endurance of 1.5 × 10 4 cycles (over 12.5‐fold improvement), and a low set voltage of ∼0.27 V. It maintains stable operation under 100% relative humidity, thermal treatment at 348 K, 6‐month ambient storage, and 300 bending cycles at 90°, and even passes the stringent IPX7 waterproof standard. A hardware‐mapped convolutional neural network leveraging the synaptic properties of these memristors achieves a highly reliable MNIST recognition accuracy of 90%±0.4% across 10 independent test cycles. This work demonstrates that conformal Parylene encapsulation offers a comprehensive solution to the perovskite instability, paving the way for flexible and low‐power neuromorphic systems.
科研通智能强力驱动
Strongly Powered by AbleSci AI