Fellowship Student Honeywell/University of Missouri Cleveland, MO, United States
This paper uses a survey of literature method to lay the foundation for recommendations that minimize the probability of unintended supply disruptions, shortages, or sabotage of microelectronic assemblies. This paper describes manufacturing (microchip manufacturing, electronic design, wafer preparation, photolithography, material addition/removal/modification, chiplet assembly, packaging and testing) of a microelectronic assembly, including the semiconductor chip. This paper reviews the micro assembly industry standards (via Appendix) according to general test method categories (functional, parametric, reliability, structural, burn-in and packaging) and by major standards organization (JEDEC, MIL-STD, IEC, IEEE, AEC and IPC) as well as specialty standards organizations involved in specialized methods and applications such as chiplets (UCIe) and defense procurement (DFARS). With the above background, the author introduces technical and non-technical market trends threatening assurance of microelectronics capacity and supply such as geopolitics, international and domestic tariff and subsidy policies, supplier concentration, dis-intermediation of integrated fabs, electronic design automation startups and advanced node (photolithography-limit-driven) technologies like 3D, FinFET and factory optimization tools. With a basis of understanding of the manufacturing processes, quality standards and market forces, the thesis introduces artificial intelligence, and machine learning (via Appendix) in the context of human-included, or human-in-the-loop decision processes focused on microelectronic assembly capacity, supply and assurance. The author then demonstrates the benefits of commercial large language models for the testing of queries on such critical items as: current standard coverage and gaps in technical standards for custom integrated circuits and ‘comparison of performance parameters for advanced testing equipment. The thesis then considers mixed human-in-the-loop and model-based exercises that have value for future group supply chain problem solving and enhanced human-included artificial intelligence.