神经形态工程学
材料科学
兴奋剂
多重性(数学)
光电子学
接触电阻
图层(电子)
化学物理
纳米技术
化学
计算机科学
人工神经网络
数学分析
数学
机器学习
作者
Ai Kassai,Tsuyoshi Hasegawa
标识
DOI:10.35848/1347-4065/ab7f59
摘要
The development of hardware-based neuromorphic systems requires devices that change synaptic weight in an analog manner. Molecular-gap atomic switches exhibit higher on/off ratios compared to other, Redox-based, non-volatile memory devices, and also show analog change in resistance. However, several issues remain with these devices, such as those related to the stability and multiplicity of analog resistance. In this study, we employed Cu-doped Ta2O5 as an ionic transfer layer, since it was thought that the higher concentration of Cu in the Ta2O5 layer, from pre-doping, would stabilize the Cu atoms precipitated on the Ta2O5 layer even after removal of the bias application. The fabricated Pt/PTCDA/Cu-doped Ta2O5/Cu molecular-gap atomic switches showed analog change in resistance over a range of six (6) orders of magnitude, with high multiplicity. The stability of the analog resistance was also improved.
科研通智能强力驱动
Strongly Powered by AbleSci AI