AI-Powered Process Optimization for EUV MOR: Equipment Trace Data Feature Extraction and Machine Learning is Essential for CD Control

Abstract

We explore the application of Artificial Intelligence (AI) and Machine Learning (ML) to enhance lithography processes involving metal-oxide resists (MOR) for Extreme Ultraviolet (EUV) lithography. Through a systematic experimental approach, we investigate the impact of lithography processes and 120 equipment parameters on critical metrology features such as Critical Dimension (CD) Mean and CD Standard Deviation (Std) to accelerate the development of this technology into the industry. Our findings reveal clear insights between experimental settings and metrology outcomes, highlighting the significance of lithography parameter variability and control. Additionally, we uncovered valuable insights into the correlation between sensor-trace features and experimental perturbations, paving the way for enhanced process control and optimization. This collaborative effort underscores the importance of leveraging purpose-built AI platforms and Data Science to drive further continuous improvements in high volume semiconductor manufacturing and research and development pilot lines.