Applications of Artificial Intelligence in Semiconductor Defect Detection and Process Control: Methods, Progress, and Prospects

Artificial intelligence (AI) is becoming an essential enabler for semiconductor manufacturing, where stringent requirements on accuracy, reliability, and efficiency pose significant challenges at advanced technology nodes. This review focuses on the application of AI to defect detection and process control, two critical domains for yield enhancement and reliability assurance. The scope covers device-level inspections, process-level monitoring, and equipment-level maintenance, as well as multivariate modeling, virtual metrology, and adaptive optimization in dynamic manufacturing environments. The reviewed studies employ deep learning, image-based modeling, physics-informed simulations, and data-driven frameworks to improve precision, accelerate failure analysis, and support real-time decision making. In addition, emerging platforms such as federated learning, digital twins, and neuro-symbolic AI are discussed as pathways to industrial-scale deployment, addressing challenges of data privacy, interpretability, and cross-enterprise collaboration. Overall, this work highlights the potential of AI to bridge laboratory research and large-scale production, providing a foundation for future semiconductor manufacturing that is more accurate, automated, and resilient, while identifying open challenges in scalability, generalization, and robust validation.