Reliability prediction of semiconductor devices using modified physics of failure approach

Traditional approaches like MIL-HDBK, Telcordia, and PRISM etc. have limitation in accurately predicting the reliability due to advancement in technology, process, materials etc. As predicting the reliability is the major concern in the field of electronics, physics of failure approach gained considerable importance as it involves investigating the root-cause which further helps in reliability growth by redesigning the structure, changing the parameters at manufacturer level and modifying the items at circuit level. On the other hand, probability and statistics methods provide quantitative data with reliability indices from testing by experimentation and by simulations. In this paper, qualitative data from PoF approach and quantitative data from the statistical analysis is combined to form a modified physics of failure approach. This methodology overcomes some of the challenges faced by PoF approach as it involves detailed analysis of stress factors, data modeling and prediction. A decision support system is added to this approach to choose the best option from different failure data models, failure mechanisms, failure criteria and other factors.