As-grown-Generation Model for Positive Bias Temperature Instability

Abstract

Positive bias temperature instability (PBTI) is poised to cause significant degradation to nFETs with deep scaling into nanometers. It is commonly modeled by a power law fitted with measured threshold voltage shift. For the first time, this paper shows that such models do not warrant PBTI prediction outside the stress conditions used for the fitting. The underlying cause for this failure is the errors in the extracted power exponent. Based on the understanding of different types of defects, we develop a robust as-grown-generation model and demonstrate its capability for accurate prediction of PBTI under both dc and ac conditions. The generation-induced degradation is found to play a key role. Analysis reveals that although PBTI is usually smaller than negative BTI (NBTI) within the typical test time window, it can exceed NBTI by the end of device lifetime.