Three-dimensional X-ray imaging and reliability prediction of flip-chip interconnects in infrared focal plane arrays

Advanced focal plane arrays (FPAs) feature diffraction-limited pixel sizes and high-density interconnect arrays, which require high-integrity solder joints and flip-chip bonding with sub-micrometer alignment precision. Therefore, comprehensive defect inspection and accelerated reliability assessment of the flip-chip solder joints are essential for ensuring robust FPA performance. Here we demonstrate that X-ray computed laminography (CL) enables non-destructive three-dimensional (3D) imaging of planar large-format and mesa-structured dual-color infrared FPAs, simultaneously capturing device-scale topography and micron defect features in indium bump interconnects. By combining machine learning based on multidimensional statistical analysis and K-means clustering, we develop a reliability assessment framework for indium bump interconnects in infrared FPAs. Our approach leverages a reliable X-ray 3D reconstruction database to evaluate thousands of indium bump interconnects and simultaneously predict invalid pixels across entire detector arrays. It achieves high prediction accuracy, showing strong agreement with real defect maps in both large-format planar and dual-color mesa detectors, with processing times of less than 10 minutes per array. Through combined 3D imaging and reliability prediction, this work provides fundamental insights for advancing the development of next-generation FPAs featuring larger array formats, reduced pixel pitch, and multi-band detection capabilities.