DocumentCode
3174316
Title
Evaluation of static noise margin and performance of 6T FinFET SRAM cells with asymmetric gate to source/drain underlap devices
Author
Hu, Vita Pi-Ho ; Fan, Ming-Long ; Su, Pin ; Chuang, Ching-Te
Author_Institution
Dept. of Electron. Eng., Nat. Chiao Tung Univ., Hsinchu, Taiwan
fYear
2010
fDate
11-14 Oct. 2010
Firstpage
1
Lastpage
2
Abstract
This paper analyzes the stability, performance, and variability of 6T FinFET SRAM cells with asymmetric gate-to-source/drain underlap devices. At Vdd = IV, using asymmetric source-underlap access transistors can improve RSNM while degrading WSNM; using source-underlap pull-up transistors can improve WSNM without sacrificing RSNM. Thus, the conflict between improving RSNM and WSNM in 6T FinFET SRAM cell can be relaxed by using the asymmetric source/drain underlap access and pull-up transistors (PUAX_Asym.). We also show, for the first time, that as Vdd is reduced (e.g. <; 0.6V), the effectiveness of using asymmetric source/drain-underlap access transistors to improve RSNM diminishes due to the worse electrostatic integrity caused by the underlap. At Vdd = IV, the 6T PUAX_Asym. SRAM cell shows 20.5% improvement in RSNM, comparable WSNM, 10% degradation in "cell" Read access time and 36% improvement in Time-to-Write compared with the conventional 6T SRAM cell (Symm.). The PUAX_Asym. SRAM cell also shows adequate μRSNM/σRSNM and μWSNM/σWSNM at Vdd = IV.
Keywords
MOSFET; SRAM chips; circuit noise; FinFET SRAM cell; asymmetric gate; asymmetric source-underlap access transistor; electrostatic integrity; source-drain underlap device; source-underlap pull-up transistor; static noise margin; Electrostatics; FinFETs; Logic gates; Random access memory; Stability analysis; Wireless sensor networks;
fLanguage
English
Publisher
ieee
Conference_Titel
SOI Conference (SOI), 2010 IEEE International
Conference_Location
San Diego, CA
ISSN
1078-621x
Print_ISBN
978-1-4244-9130-8
Electronic_ISBN
1078-621x
Type
conf
DOI
10.1109/SOI.2010.5641392
Filename
5641392
Link To Document