A novel physics-based variable NBTI simulation framework from small area devices to 6T-SRAM

Abstract

A novel simulation framework is developed to study NBTI variability in devices and SRAM circuits. Stochastic Reaction Diffusion (RD) model for interface trap generation (ΔN_IT) and stochastic 2 well model for charging of pre-existing bulk traps (ΔN_HT) are interfaced with TCAD for electrostatics and time-zero variability to determine variable NBTI in device level. Distributions of ΔN_IT and ΔN_HT are generated for stress and recovery. ΔV_T and V_T distributions are generated using empirical exponential impact and full TCAD simulations. Simulated ΔV_T distributions match with experiments, analytic models and show correlation between mean and variance. Impact of NBTI variability on read, write and hold failure of 6T-SRAM is simulated by using SPICE. NBTI variability of these SRAM metrics is correlated to device V_T variability.