Comparative analysis of RD and Atomistic trap-based BTI models on SRAM Sense Amplifier

Abstract

Bias Temperature Instability (BTI) in transistors has become a key reliability bottleneck with sub-45nm CMOS technologies. The most common models to characterize BTI are the Reaction-Diffusion (RD) and Atomistic trap-based models. This paper presents comparative impact analysis of RD and Atomistic trap-based BTI models for the SRAM Sense Amplifier. The evaluation metric, the sensing delay is analyzed for both models for the different workloads and supply voltages for 45nm technology node. The results show that the sensing delay degradation is slightly higher in RD model than Atomistic trap-based model for different workloads. Nevertheless, we observe a similar trend for both models. For example the BTI impact degradation is 6:69% for RD model and 6:57% for Atomistic trap-based model when worst case workload is applied for a 108s life time.