Toward Automated Defect Extraction From Bias Temperature Instability Measurements

Abstract

Defects in the gate oxide give rise to bias temperature instability (BTI), which is considered a serious threat to the device reliability of ultrascaled MOSFETs. Extrapolating the device degradation over the operational lifetime, therefore, requires detailed knowledge about the distributions of defects causing BTI. Typically, BTI degradation is modeled by calibrating a predefined defect parameter distribution, such as normally distributed defect bands, by employing measure-stress-measure (MSM) sequences at various temperatures. Here, we present the Effective Single Defect Decomposition (ESiD), a novel method for a semiautomated extraction of defect parameters from MSM experiments which does not require any prior assumptions about their distributions. This technique decomposes the MSM sequences into contributions from dominant effective single-defects and constructs a defect parameter distribution that reproduces the experimental data with high accuracy. We validate this new method by comparing its results to density functional theory (DFT) predictions and single-defect characterizations.