Skip to Main Content
In the paper under discussion, the Van der Pol article referenced by this author and also by the commentator is a study of a large number of devices used in commercial products where both the long term reliability in system application (field reliability) and long term life test were compared to sample burn-in testing. The authors of this reference found that life test data and system reliability were both well fitted to Weibull distribution. The formulae in the IEEE forum article are based on an empirical fit to this very large number of experimental data points. These are not assumptions but simply an empirical model well grounded in an extensive experimental data set. This author does not assert that a single failure mechanism is appropriate to a complete explanation of infant mortality in integrated circuits. I concur that many failure mechanisms actually apply with different failure characteristics (best detected by failure analysis of infant mortals) and different acceleration.