DocumentCode
2351306
Title
Variability in sub-100nm SRAM designs
Author
Heald, Raymond ; Wang, Ping
Author_Institution
Sun Microsystems Inc., Sunnyvale, CA, USA
fYear
2004
fDate
7-11 Nov. 2004
Firstpage
347
Lastpage
352
Abstract
Many components of variability become larger percentage design factors with decreasing feature size. Hence, the small transistors in SRAM cells are particularly sensitive to these variations. The SRAM cell transistors in sub-100-nm designs may contain fewer than 100 channel dopant atoms. To achieve a robust design with such variability, one must enhance the normal static-noise-margin and write-trip-point analysis, often with Monte Carlo simulations using statistical transistor models including the process and mismatch fluctuations. Similar challenges exist for the sense amplifiers normally used with SRAM arrays. Except with very low speed designs, yield to speed can be substantially reduced by variations between nominally matched sense amplifier transistors as well as by the variability resulting in a very worst memory cell low read current. This also increases the hazards of delay timing with dummy paths and dummy cells and increases the need for at-speed testing prior to repair.
Keywords
Monte Carlo methods; SRAM chips; integrated circuit design; integrated circuit noise; nanoelectronics; transistors; Monte Carlo simulation; SRAM cell transistors; SRAM design variability; at-speed testing; channel dopant atoms; memory cell low read current; mismatch fluctuation; nominally matched sense amplifier transistors; normal static-noise-margin; process fluctuation; statistical transistor models; sub-100nm SRAM designs; write-trip-point analysis; Circuits; Fluctuations; Monte Carlo methods; Noise measurement; Random access memory; Robustness; Semiconductor process modeling; Sun; Timing; Voltage;
fLanguage
English
Publisher
ieee
Conference_Titel
Computer Aided Design, 2004. ICCAD-2004. IEEE/ACM International Conference on
ISSN
1092-3152
Print_ISBN
0-7803-8702-3
Type
conf
DOI
10.1109/ICCAD.2004.1382599
Filename
1382599
Link To Document