Thermally accelerated aging of semiconductor components

Abstract

The rising interest in the use of elevated temperatures to accelerate the life of semiconductor components warrants a review of the mathematical basis of the technique. Starting from the four reliability probability functions, the lognormal density function is featured showing how the temperature dependence of the associated cumulative failure function leads to a prediction of the component median life under service conditions through the construction of Arrhenius lines. Both steady and incrementally raised temperatures can be employed to gather test data, the latter-the "step-stress" method-involving a correction procedure. The analysis is illustrated by its application to test results from bipolar and MOS transistors.