Fully functional perpendicular STT-MRAM macro embedded in 40 nm logic for energy-efficient IOT applications

Yu Lu,Tom Zhong, W. Hsu, S. Kim, X. Lu,J. J. Kan,C. Park,W. C. Chen,X. Li,X. Zhu,P. Wang,M. Gottwald, J. Fatehi, L. Seward,J. P. Kim, N. Yu,G. Jan, J. Haq, S. Le,Y. J. Wang, L. Thomas,J. Zhu,H. Liu,Y. J. Lee,R. Y. Tong,K. Pi, D. Shen,R. He,Z. Teng, V. Lam, R. Annapragada, T. Torng,Po-Kang Wang,S. H. Kang

2015 IEEE INTERNATIONAL ELECTRON DEVICES MEETING (IEDM)(2015)

引用 67|浏览11
暂无评分
摘要
We present for the first time a fully functional 40 nm perpendicular STT-MRAM macro (1 Mb, ×32/×64 IO) embedded into a foundry standard CMOS logic platform. We achieved target design specifications of 20 ns read access time and 20–100 ns write cycle time without redundancy repair at standard core and IO voltages. The full 1 Mb macro can be switched reliably with write pulse as short as 6 ns, which results in full-chip write power of ∼ 3.2 μW/Mbps at ×64. This is the lowest eNVM write power reported at a full-chip level and about three orders of magnitude smaller than that of eFLASH. The 0.5 Mbit high-density bitcell array also demonstrates good Rp distribution and 100 % STT switching. Our results demonstrate superior power-area-feature attributes of perpendicular STT-MRAM as a best-in-class unified eNVM solution for Internet-of-Things (IOT) applications at 40 nm as well as the scalability of these advantages to 28 nm and beyond.
更多
查看译文
关键词
fully functional perpendicular STT-MRAM macro,energy-efficient Internet-of-Things applications,foundry standard CMOS logic platform,read access time,write cycle time,full-chip level,high-density bitcell array,size 40 nm
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要