Non-Destructive Imaging and Structural Techniques for Reliability Assessment of Power Electronic Modules: A Literature Review of Conventional, Automated and Machine Learning-Based Approaches

The rapid expansion of electrified transport, renewable energy systems and high-efficiency industrial power converters relies on the robustness of power electronic modules and multi-chip packaging architectures. As module power densities rise and operating environments become more extreme, reliability becomes a key determinant of system lifetime, safety and cost. Thermo-mechanical degradation mechanisms such as die-attach delamination, bond-wire lift-off often initiate within material layers, making non-destructive testing (NDT) essential for quality and health monitoring. This literature review assesses state-of-the-art non-destructive imaging modalities for power electronic modules, with primary focus on IGBT packaging architectures. It covers X-ray imaging, scanning acoustic microscopy (SAM), optical metrology and infrared thermography. Rather than treating these modalities in isolation, it evaluates their compatibility, limits and fault-oriented application ranges. Emphasis is placed on the industry transition from manual evaluation to automated, machine learning-driven workflows for defect segmentation, void quantification and multimodal data fusion. It concludes by outlining industrial best practices, software and hardware bottlenecks and directions for embedding machine learning into autonomous packaging inspection. No single NDT technique provides complete assessment of module health, necessitating the complementary use of multiple modalities. Automation and machine-learning approaches improve defect detection but reamin dependant on physics-based measurements, high-quality data and application-specific workflows.