Analysis of parameter-degradation data using life-data analysis programs

Abstract

A new application of existing reliability software packages is proposed to analyze parameter-degradation data. This new implementation of existing tools is applicable to a large class of degradation phenomena described by the power law or the stretched exponential law. When these laws are used to model parameter-degradation paths, the parametric drift can be treated as an explanatory variable, on equal footing with conventional explanatory variables such as temperature, electrical stress, or humidity. Thin film integrated circuit resistor degradation data and the STAR (Statistical Analysis of Reliability) software package are used to illustrate the authors' approach. Other degradation phenomena that can be studied using the proposed method include hot carrier degradation in metal-oxide-silicon (MOS) transistors, laser degradation, and degradation of light emitting diodes. >