神经形态工程学
石墨烯
材料科学
电阻随机存取存储器
可靠性(半导体)
光电子学
纳米技术
计算机科学
电气工程
电压
功率(物理)
工程类
人工神经网络
物理
量子力学
机器学习
作者
Seonjeong Lee,Sookyeong Kim,Boram Kim,Rayoung Park,Dong‐Wook Park,Yoon Kim
标识
DOI:10.1002/sstr.202500056
摘要
Flexible parylene‐C (PPXC)‐based resistive random access memory (RRAM) has garnered attention as a synaptic device suitable for flexible neuromorphic systems due to its low‐power consumption and high‐speed switching characteristics. However, Negative‐SET has been a key factor contributing to reliability degradation by inducing the overgrowth of conductive filaments (CFs). To address this issue, this study fabricates a PPXC‐based RRAM crossbar array with a graphene barrier inserted between the inert electrode and the resistive switching layer. The incorporation of a graphene barrier layer effectively mitigates the excessive diffusion of metal ions, thereby significantly improving the stability of CF. Furthermore, the graphene layer plays a critical role in regulating the RESET process, ensuring enhanced device reliability. The fabricated devices, featuring a Pt/Graphene/PPXC/Cu structure, demonstrate superior electrical and mechanical performance, including a low operating voltage <2 V, endurance cycles >104, retention time >104 s, conductance ON/OFF ratio >102, and mechanical bending durability exceeding 500 cycles at a bending radius of 3 mm. Furthermore, artificial neural network simulations using the Modified National Institute of Standards and Technology database confirm its applicability as a neuromorphic system. This study establishes a technological foundation for the development of next‐generation flexible neuromorphic systems.
科研通智能强力驱动
Strongly Powered by AbleSci AI