神经形态工程学
材料科学
纳米技术
不对称
金属
光电子学
计算机科学
物理
人工神经网络
人工智能
冶金
量子力学
作者
Tian Tan,Maheswari Sivan,Kai Zhou,Haoyue Guo,Yiting Wu,Linfeng Sun,Yida Li,Xuewei Feng
出处
期刊:Small
[Wiley]
日期:2025-06-29
标识
DOI:10.1002/smll.202503716
摘要
Abstract Memtransistors, integrating the resistive switching behavior of memristors with the gate tunability of transistors, offer significant promise for neuromorphic computing and in‐memory processing. However, their scalability in crossbar arrays is limited by sneak leakage currents. In this study, it is reported that a self‐rectifying Molybenum Disulfide (MoS 2 ) memtransistor is enabled by asymmetric metal contacts, where a Schottky Platinum (Pt) contact and a quasi‐ohmic Bismuth (Bi) contact are employed. The asymmetric Schottky barrier, coupled with drain voltage‐induced barrier narrowing, induces highly asymmetric current characteristics. The device exhibits exceptional performance metrics, including a high rectification ratio of 10 4 , a switching ratio of 10 4 , and a retention time exceeding 10 5 seconds. The dynamic modulation of the Schottky barrier height is validated through temperature‐dependent studies, energy band analysis, and technology computer‐aided design (TCAD) simulations. Compared to non‐rectifying configurations, the asymmetric memtransistors reduce crossbar array leakage currents by five orders of magnitude while enhancing power efficiency by 61 times. Simulations using an Echo State Network (ESN) highlight the memtransistor's robustness under low‐precision and noisy conditions. Overall, the approach presents a scalable, energy‐efficient approach for memtransistor‐based in‐memory computing and neuromorphic architectures.
科研通智能强力驱动
Strongly Powered by AbleSci AI