An Integrated Physics-Based and Data-Driven Framework for Defect Prediction in Advanced Nanoimprint Lithography...

Area of Science:

• Materials Science
• Nanotechnology
• Semiconductor Manufacturing

Background:

• Advanced nanoimprint lithography (NIL) offers high-resolution semiconductor patterning but faces yield loss due to accumulated distortions.
• Defects in NIL arise from complex interactions during coating, exposure, etching, and imprinting processes.

Purpose of the Study:

• To develop an integrated framework for pre-manufacturing defect-risk prediction in NIL.
• To enable early identification and mitigation of yield risks in advanced semiconductor fabrication.

Main Methods:

• An integrated framework combining a physics-based process twin and a data-driven risk prediction model (physics-augmented CNN with conformal uncertainty calibration).
• Utilizing an NDA-safe layout database and simulating sequential process steps: spin coating, electron-beam lithography (EBL), reactive ion etching (RIE), and imprinting.
• Generating physically consistent parameter maps from design layouts for downstream machine learning.

Main Results:

• Demonstrated an end-to-end virtual inspection flow converting layouts into spatially resolved risk maps prior to fabrication.
• Revealed that local pattern density significantly influences risk distribution, beyond nominal geometry.
• Highlighted the importance of the local pattern environment in governing manufacturability.

Conclusions:

• The proposed framework provides a physically interpretable method for early yield-risk screening in advanced NIL.
• Pre-manufacturing virtual inspection is crucial for optimizing semiconductor fabrication processes.
• Understanding pattern environment interactions is key to improving NIL yield and reliability.