Modeling Analysis of BTI-Driven Degradation of a Ring Oscillator Designed in a 28-nm CMOS Technology

With tightening reliability margins, product-level aging analysis is gradually gaining impetus and is set to become an integral part of the modern design flow. Increased emphasis is placed on the development of physics-based compact models for the phenomena responsible for transistor degradation and their integration into EDA environments. Commercial Process Design Kits (PDKs) of advanced technologies have also started to include compact models for transistor degradation along with a dedicated reliability simulation framework. In this work, we present a comprehensive study of the compact aging models in one such commercial PDK, with the help of extensive measurement data from individual devices as well as Ring Oscillators (RO). Then, we perform our own extraction of model parameter shifts from full $I_{d}$ - $V_{gs}$ fitting of the measured BTI-stressed devices to gain insight into the modeling procedure adopted by the foundry. We finally use the parameters extracted thusly to investigate the impact of the time-0 and the time-dependent device-to-device variability on RO degradation.

Automated summary

With the increasing importance of reliability margins, product-level aging analysis has become an integral part of modern design flow. The focus has shifted towards developing physics-based compact models for transistor degradation and integrating them into EDA environments. The inclusion of compact models for transistor degradation in commercial Process Design Kits (PDKs) has also gained traction. This study presents a comprehensive analysis of compact aging models in a commercial PDK, utilizing extensive measurement data from individual devices and Ring Oscillators (RO). Model parameter shifts are extracted through full $I_{d}$ - $V_{gs}$ fitting of BTI-stressed devices, providing insight into the modeling procedure adopted by the foundry. The extracted parameters are then used to investigate the impact of time-0 and time-dependent device-to-device variability on RO degradation.