Add to BibSonomy
Abstract
In recent years, neuromorphic computing is recognized as a promising path to further improve the efficiency of integrated computing system in the post-Moore era, relying on its high parallelism. As a key fundamental element in hardware-implementing neuromorphic system, the synaptic device has made substantial research progress. Among these, SiO2 trapping-based memristive devices generally have systematically integrated merits, such as ease of fabrication and high CMOS process compatibility, but electrochemical activity to oxygen makes them unreliable for operating in air. Here, by using ultrathin Si3N4 as a physical isolation layer, we have obtained a robust memristive device based on SiO2 trapping although operating in air. Further study of Si3N4 thickness dependence has demonstrated that 7 nm is suggested as the most favorable thickness for reliable and flexible programming, and that an inherent isolating mechanism is ‘switching-on’ for an electron but ‘switching-off’ for large-sized oxygen molecules. Based on a device with 7 nm Si3N4, we have mimicked various modes of synaptic plasticities. These results could thus not only increase the prospects of using SiO2 trapping in memristive applications but also provide an effective path to improve the robustness of these SiO2-based applications against ambient air.
© 2024 Optica Publishing Group
Full Article | PDF ArticleMore Like This
Longxing Su, Sudong Wu, Xiaofei Wang, Ke Sun, Teng Yun, Yongping Du, and Jianguo Lu
Opt. Lett. 49(3) 474-477 (2024)
Yilin Sun, Yingtao Ding, Dan Xie, Jianlong Xu, Mengxing Sun, Pengfei Yang, and Yanfeng Zhang
Opt. Lett. 46(7) 1748-1751 (2021)
Chengdong Yang, Linlin Su, Kaipeng Xia, Xinwei Li, Yilong Liu, and Hui Li
Opt. Express 31(19) 31061-31071 (2023)