Bias Temperature Instability of MOSFETs: Physical Processes, Models, and Prediction

Abstract

CMOS technology dominates the semiconductor industry, and the reliability of MOSFETs is a key issue. To optimize chip design, trade-offs between reliability, speed, power consumption, and cost must be carried out. This requires modeling and prediction of device instability, and a major source of instability is device aging, where defects gradually build up and eventually cause malfunction of circuits. This paper first gives an overview of the major aging processes and discusses their relative importance as CMOS technology developed. Attentions are then focused on the negative and positive bias temperature instabilities (NBTI and PBTI), mainly based on the early works of the authors. The aim is to present the As-grown-Generation (AG) model, which can be used not only to fit the test data but also to predict long-term BTI at low biases. The model is based on an in-depth understanding of the different types of defects and the experimental separation of their contributions to BTI. The new measurement techniques developed to enable this separation are reviewed. The physical processes responsible for BTI are explored, and the reasons for the failure of the early models in predicting BTI are discussed.