Design of experiments and integer linear programming-assisted conjugate-gradient optimisation of high-κ/metal-gate nano-complementary metal-oxide semiconductor static random access memory

Abstract

Low-power consumption and stability in static random access memories (SRAMs) is essential for embedded applications. This study presents a novel design flow for power minimisation of nano-complementary metal-oxide semiconductor SRAMs, while maintaining stability. A 32 nm high-κ/metal-gate SRAM has been used as an example circuit. The baseline circuit is subjected to power minimisation using a dual-threshold voltage assignment based on novel combined design of experiments and integer linear programming (DOE-ILP) approach. However, this leads to a 15% reduction in the static noise margin (SNM) of the cell. The conjugate gradient optimisation overcomes this SNM degradation, while reducing the power consumption. The final SRAM design shows 86% reduction in power consumption (including leakage) and 8% increase in the SNM compared with the baseline design. The variability analysis of the optimised cell is performed by considering the effect of 12 parameters. SRAM arrays of different sizes are constructed to demonstrate the feasibility of the proposed SRAM cell. To the best of the authors’ knowledge, this is the first study which makes use of DOE-ILP and conjugate gradient method for simultaneous stability and power optimisation in high-κ/metal-gate SRAM circuits.