反冲
记忆电阻器
神经形态工程学
弹道传导
物理
蛋白质丝
量子
量子点接触
非平衡态热力学
电导
电子
载流子
凝聚态物理
材料科学
纳米技术
原子物理学
量子力学
计算机科学
量子阱
机器学习
人工神经网络
复合材料
激光器
出处
期刊:Jetp Letters
[Pleiades Publishing]
日期:2023-03-01
卷期号:117 (5): 384-391
被引量:1
标识
DOI:10.1134/s0021364022603323
摘要
Memristive devices, whose resistance can be controlled by applying a voltage and further retained, are attractive as possible circuit elements for neuromorphic computing. This new type of devices poses a number of both technological and theoretical challenges. Even the physics of the key process of resistive switching, usually associated with formation or breakage of conductive filaments in the memristor, is not completely understood yet. This work proposes a new resistive switching mechanism, which should be important in the thin-filament regime and take place due to the back reaction, or recoil, of quantum charge carriers, independent of the conventional electrostatically-driven ion migration. Since thinnest conductive filaments are in question, which are only several atoms thick and allow for a quasi-ballistic, quantized conductance, we use a mean-field theory and the framework of nonequilibrium Green’s functions to discuss the electron recoil effect for a quantum current through a nanofilament on its geometry and compare it with the transmission probability of charge carriers. Namely, we first study an analytically tractable toy model of a 1D atomic chain, to qualitatively demonstrate the importance of the charge-carrier recoil, and further proceed with a realistic molecular-dynamics simulation of the recoil-driven ion migration along a copper filament and the resulting resistive switching. The results obtained are expected to add to the understanding of resistive switching mechanisms at the nanoscale and to help downscale high-retention memristive devices.
科研通智能强力驱动
Strongly Powered by AbleSci AI