Applications of science-driven AI in the semiconductor manufacturing

Non-destructive metrology plays a critical role in the semiconductor manufacturing process, enabling precise monitoring and control of complex device architectures without compromising the integrity of wafers. As device dimensions continue to shrink and 3D architectures become increasingly complex, traditional metrology methods face significant challenges in terms of resolution, accuracy, and throughput. This presentation provides a comprehensive review of recent advancements in non-destructive metrology, with a particular focus on the integration of simulation and artificial intelligence (AI). By leveraging physics-based simulations to model complex processes and utilizing AI to analyze vast datasets and predict critical parameters, innovative methodologies have emerged that enhance measurement accuracy and process efficiency. Case studies highlighting the application of simulation-driven AI in areas such as SEM3D will be discussed, showcasing how these approaches address key challenges in advanced semiconductor manufacturing. This review aims to provide insights into the current state of the field, identify remaining limitations, and propose directions for future research and development