Impact of RDF and RTS on the performance of SRAM cells

Abstract

This paper investigates electrical effects due to reliability phenomena associated with the downscaling of transistors in advanced technologies, particularly Random Telegraph Signal (RTS) and Random Dopant Fluctuations (RDF). RDF and RTS are becoming important issues in sub-60 nm technologies, mainly in memories, where transistor sizes are smaller. This paper presents a comprehensive evaluation of the impact of the joint effect of RTS and RDF on 6T-SRAM cells. Based on 3-D atomistic simulation of the device structure, doping profile, and trap location, the random threshold voltage variation caused by the joint effect of RDF and RTS is evaluated. The derived threshold voltage variation is then used to statistically evaluate the performance variation at the circuit level. Finally, the impact of assuming a normal distribution for the threshold voltage variation caused by RDF and RTS is studied. The results obtained running Monte Carlo simulations raffling the value of the variation of the threshold voltage from a database obtained by 3-D atomistic simulation are compared to the results obtained running Monte Carlo simulations considering the variation of the threshold voltage as a Normal distribution. The results show that for a 6T-SRAM designed with 45 nm technology the read noise margin distributions are very similar between the two cases, making the Normal distribution a good approximation for the V th variations. Nevertheless, the performance variation induced at the circuit level does not necessarily follow a normal distribution.