MATRIX: Microelectronics Assurance through X-Ray Imaging of Advanced Packaging

Advanced packaging techniques for microelectronics such as heterogenous 2.5D/3D integration enable numerous advantages for improved semiconductor performance and low form factor, but their complex structures introduce new challenges when it comes to inspecting unvalidated components and verifying assurance. As distributed and horizontal supply chains continue to supply U.S. consumers and entities with vital components and devices, the safety and security of digitized systems becomes increasingly threatened and microelectronic assurance becomes even more crucial. Current physical inspection techniques for microelectronic analysis, such as optical and X-ray, require manual assessment by a subject matter expert once the imaging data is collected, which is time consuming and prone to human error. With the condensed nature of advanced packaging structures, the features become even smaller and more difficult to assess. To this end, the proposed work MATRIX: Microelectronics Assurance through X-Ray Imaging of Advanced Packaging by Caspia Technologies will provide artificial intelligence (AI) algorithms that embed deep domain knowledge to automatically assess advanced packaging features for trust and assurance. This innovative approach deploys AI and deep learning to assess 3D X-ray computed tomography (CT) scans, which can be generated from standard X-ray CT imaging equipment. MATRIX requires no specialized image acquisition. Embedded within the algorithms are numerous processes for denoising images, reducing the impact of various artifacts such as beam hardening, and can automatically extract the individual interposer layers. In Phase I, Caspia successfully demonstrated the capabilities of AI algorithms in optical image-based inspection. This included the development of a real-time system architecture that is quick and easy-to-use for operators with no imaging nor testing expertise. It also validated the commercial need for reduced human-in-the-loop inspection. In Phase II, Caspia will pivot this technology and expand it towards internal validation with X-ray imaging. In the Base Phase, MATRIX will provide segmentations for substrate interconnects, including traces and vias, and in Option I functionality will be integrated to support automated anomaly detection and metrology validation. The development and refinement of MATRIX aims to ensure the trustworthiness and reliability of complex components, which are vital in maintaining the readiness and functionality of essential military and civilian technologies.