We have developed a novel, rapid, robust and non-destructive experimental technique for in-situ monitoring of delamination of interfaces for electronic packages. The method is based on a simple thermal transducer matrix of so-called THIXELS (thermal pixels) which allows a spatially resolved real-time image of the current status of delamination. The transducers are small metal wire meanders which are driven and electrically read out using the well-known 3-omega method. This method has special advantages over other thermal contrast methods with respect to robustness, sensitivity and signal-to-noise ratio. Notable is the absence of cross-effects. The proof of concept has been furnished on an industry-grade flip-chip package with underfill on an organic substrate. The technique is especially powerful for buried interfaces, where time-honoured methods like scanning acoustic microscopy (SAM) cannot be applied. As the technique effectively performs a thermal diffusivity sensitive scan, it may not only be useful for stress testing during package qualification, but sensor applications on other fields of health monitoring seem also possible

Today, there is no non-destructive, simple, inexpensive and yet accurate in-situ monitoring technique for cracks and delamination available for routine use in electronic package testing. However, such a method is highly desirable, as delamination testing is part of every qualification programme in industry. Rapid failure analytical techniques which allow introspect and easy-to-interpret information on adhesion loss during stress testing, analogous to e.g. resistance testing for solder joint reliability, would enable considerable speed up of the development process of advanced packaging technologies, especially also for situations where not even classical ex-situ methods like scanning acoustic microscopy or pulse phase thermography are applicable. So this paper proposes a radically new approach using local thermal impedance variations caused by cracks to generate electrically detectable signals by the 3ω-Method, designed as highly local thermal sensors array in analogy to a pixel matrix for readout as image from a delamination camera. We show the proof of concept of the method on an industry-grade flip-chip package, its robustness with respect to electrical readout and the very good correlation between experiment and simulation, enabling unequivocal detection of thermal impedance changes and its insensitivity to cross effects, e.g. moisture ingress. Guidelines and estimations for future applications are given

No abstract

No abstract

No abstract