Efficient Hihg-Sigma Yield Analysis with Yield Optimization Method for Near-Threshold SRAM Design

Abstract

The enhancement of process, the feature size of device becomes smaller, and the deductions of power supply require the design of SRAM should be consummated. For the design of 6-T SRAM, although we have many methods to optimize it in order to overcome lots of challenges. But the smaller feature size and lower power supply will produce a decline in the performance of Static Noise Margin (SNM) and increase of leakage current. They also produce much more uncertain factors than before. To make matters worse, 6-T SRAM may not work properly. So a new structure with good design-for-yield (DFY) to work at lower supply voltage becomes more important. We have found a alternative structure, which we call it 8-T SRAM. And we can prove its performance is more suitable for the demand in the future SRAM design. But there is still no comprehensive method to optimize 8-T SRAM design, and we don't know whether its yield is acceptable or not. In this paper, we propose a comprehensive method for statistical 8-T SRAM design relying on an efficient combination of yield analysis and performance optimization. To the best of our knowledge, this is the first time that such a comprehensive method is applied to analyzing and optimizing the 8-T SRAM design with lower power supply.