Frequency domain scanning acoustic microscopy for power electronics: Physics-based feature identification and selectivity

Abstract

In the evaluation of power electronic components non-destructive test methods (NDT) such as scanning acoustic microscopy (SAM) are valuable tools in packaging and failure analysis. As power modules become more compact and power devices thinner, echo overlap and interference can hamper the imaging capability and analysis of the conventional SAM time domain-based imaging techniques. Frequency domain analysis offers increased resolution and contrast but the interpretation and feature identification of waveforms is less obvious and interpretation guidelines have received little attention. In this paper a physics-based analysis of the frequency domain response is presented in a power module case study. The analysis is used to demonstrate physics-based feature selective contrast in such systems and offers guidelines for feature prediction. The approach is verified on full scan datasets from acoustic scans of a hybrid multilayer stack. Distinct features in the echo frequency domain are identified as resonances associated with multiple internal reflections in the layer structure. Where such features may hamper analysis in the conventional approach these can, if properly assigned, be exploited to yield selective and improved contrasts and allow accurate structural and material analysis.