Data-Driven Prediction of the Remaining useful Life of QFN Components Mounted on Printed Circuit Boards

Abstract

Prognostics and Health Management (PHM) introduces in-situ monitoring of health parameters to the reliability of electronics. In this paper we adopt a data-driven PHM approach to predict delamination in QFN components. The signal of on-chip stress sensors reacts to thermal and mechanical loads and alters under degradation processes. We track the sensor signal in an accelerated life test, which combines thermal cycling and four-point bending. The obtained run-to-failure data-sets reveal correlation to delamination and furthermore solder joint fatigue.