仿真
材料科学
记忆电阻器
长时程增强
分子动力学
生物系统
动力学(音乐)
纳米技术
电子工程
声学
物理
生物
工程类
经济
受体
量子力学
生物化学
经济增长
作者
Phu‐Quan Pham,T. Ngoc Tran,Tai Anh,Thanh Danh Nguyen,Juergen Brügger,Yoshiyuki Kawazoe,Thắng Bách Phan,Thuat Nguyen-Tran,Nam Hoang Vu,Ngoc Kim Pham
标识
DOI:10.1002/adfm.202516695
摘要
Abstract This study presents a combined experimental–theoretical approach to understanding and optimizing CrO x ‐based self‐rectifying memristor (SRM) for neuromorphic computing and provides a promising platform for selector‐free, low‐power crossbar arrays. By using a stencil‐assisted DC sputtering technique, a 16 × 16 crossbar array is fabricated and exhibits reliable analog resistive switching, minimal sneak current (≈150 nA), and an excellent rectification ratio up to 2.56 × 10 3 , demonstrating excellent scalability and array‐level uniformity. Charge–voltage measurements reveal capacitor‐coupled behavior, including nonlinear, time‐dependent charge accumulation. The memristor supports key synaptic functions such as long‐term potentiation/ depression (LTP/LTD) and excitatory postsynaptic current modulation, which is tunable by pulse amplitude and width. First‐principles calculations show that the Ti/CrO x /TiO 2 /Cr structure exhibits strong asymmetry in both band alignment and interfacial barrier heights, which is plausible for self‐rectifying behavior. Furthermore, the band offset at the CrO x /TiO 2 interface can be modulated by the presence and migration of oxygen vacancies. This enables dynamic modulation of shallow and deep trap states across the oxide layers. This mechanism facilitates analog and multilevel switching, making the proposed structure highly suitable for future in‐memory computing architectures.
科研通智能强力驱动
Strongly Powered by AbleSci AI