Strategies for parameter extraction of the time constant distribution of time-dependent variability models for nanometer-scale devices

Abstract

Time-dependent variability phenomena are stochastic and discrete for nanometer-scale technologies, and, hence, must be statistically characterized. These phenomena are attributed to the emission and capture of charges in device defects. This paper explores two different strategies to extract, from experimental data, the distribution parameters of the time constants of the defects. It delves into the accuracy of each strategy, showing how the extraction strategy can have a huge impact on the accuracy and the amount of characterization data required, and, therefore, on the amount of (expensive) characterization time in the lab.