中国物理B ›› 2018, Vol. 27 ›› Issue (11): 118501-118501.doi: 10.1088/1674-1056/27/11/118501
• INTERDISCIPLINARY PHYSICS AND RELATED AREAS OF SCIENCE AND TECHNOLOGY • 上一篇 下一篇
Jing Liu(刘璟), Xiaoxin Xu(许晓欣), Chuanbing Chen(陈传兵), Tiancheng Gong(龚天成), Zhaoan Yu(余兆安), Qing Luo(罗庆), Peng Yuan(袁鹏), Danian Dong(董大年), Qi Liu(刘琦), Shibing Long(龙世兵), Hangbing Lv(吕杭炳), Ming Liu(刘明)
Jing Liu(刘璟)1,2, Xiaoxin Xu(许晓欣)1,2, Chuanbing Chen(陈传兵)1,2, Tiancheng Gong(龚天成)1,2, Zhaoan Yu(余兆安)2, Qing Luo(罗庆)2, Peng Yuan(袁鹏)1,2, Danian Dong(董大年)2, Qi Liu(刘琦)2, Shibing Long(龙世兵)2, Hangbing Lv(吕杭炳)2, Ming Liu(刘明)2
摘要:
The tail bits of intermediate resistance states (IRSs) achieved in the SET process (IRSS) and the RESET process (IRSR) of conductive-bridge random-access memory were investigated. Two types of tail bits were observed, depending on the filament morphology after the SET/RESET operation. (i) Tail bits resulting from lateral diffusion of Cu ions introduced an abrupt increase of device resistance from IRS to ultrahigh-resistance state, which mainly happened in IRSS. (ii) Tail bits induced by the vertical diffusion of Cu ions showed a gradual shift of resistance toward lower value. Statistical results show that more than 95% of tail bits are generated in IRSS. To achieve a reliable IRS for multilevel cell (MLC) operation, it is desirable to program the IRS in RESET operation. The mechanism of tail bit generation that is disclosed here provides a clear guideline for the data retention optimization of MLC resistive random-access memory cells.
中图分类号: (Semiconductor-device characterization, design, and modeling)