材料科学
神经形态工程学
记忆电阻器
光电子学
双层
电阻随机存取存储器
薄膜
制作
突触重量
氧化物
纳米技术
兴奋剂
降级(电信)
GSM演进的增强数据速率
卷积神经网络
电压
再现性
非易失性存储器
纳米点
高效能源利用
电压降低
逻辑门
作者
Subarna Pramanik,Rajarshi Chakraborty,Pushpendra Prakash Maurya,Swati Suman,Parasuraman Swaminathan,Bhola Nath Pal
标识
DOI:10.1002/adfm.202514732
摘要
Abstract A solution‐processed, high‐performance bilayer‐structured memristor is developed by linking metallic and oxygen‐deficient channels. The device comprised a thin film of Ag(Ag⁺)‐V 2 O 5 , synthesized through Ag⁺ ion‐exchange and subsequent reduction of LiV 3 O 8 , sandwiched with a SnO 2 film followed by a 10 nm gold electrode, which also functions as light‐transmitting window. The memristor exhibits robust bipolar switching, sustaining over 3000 repeatable cycles with minimal degradation and pulse endurance approaching 10⁶ cycles. It achieves low‐to‐high resistance state ratio near 10⁵, and data retention exceeding 10⁵ s, maintaining an on/off ratio of ≈10⁴. Device‐to‐device uniformity is confirmed by intra‐batch variation within ±2% and inter‐batch variation across 50 fabrication runs within ±7%, highlighting excellent process reproducibility. Beyond standard ReRAM functions, the device supports logic operations and 4‐bit edge computing, validated across 20 intra and inter‐batch devices for operational reliability. Furthermore, it emulates long‐term synaptic plasticity under optical excitation, maintaining stable performance under ageing and repeated light on/off cycling, confirming strong environmental stability and reproducibility of the photo response behavior. Cyclic I–V sweeps induce potentiation and depression behavior, driving convolutional neural network‐based image recognition with ≈98% training accuracy. These results collectively position the bilayer architecture as a promising platform for multifunctional neuromorphic, logic, and optoelectronic applications.
科研通智能强力驱动
Strongly Powered by AbleSci AI