AI-Assisted Composite Etch Model for MPT

Area of Science:

• Semiconductor Manufacturing
• Computational Modeling
• Artificial Intelligence

Background:

• Advanced semiconductor nodes require high-precision patterning for complex foundry circuits.
• Lithography-Etch-Lithography-Etch (LELE) is a key Multiple Patterning Technology (MPT) in Deep Ultraviolet (DUV) processes.
• Modeling etch contour simulation and hotspot detection in LELE is challenging due to inter-LE cycle interactions.

Purpose of the Study:

• To develop an Artificial Intelligence (AI)-assisted composite etch model for LELE processes.
• To accurately capture inter-LE cycle interactions for improved simulation.
• To enable Etch Rule Check (ERC)-based detection of After Etch Inspection (AEI) hotspots and predict etch bias for After Develop Inspection (ADI) target generation.

Main Methods:

• An AI-assisted composite etch model was developed to directly output the final post-LELE etch contour.
• The model captures inter-LE cycle interactions inherent in LELE processes.
• The model predicts etch bias for various pattern types, including complex 2D patterns.

Main Results:

• The AI-assisted model successfully captures inter-LE interactions, enabling direct output of the final post-LELE etch contour.
• Etch Rule Check (ERC) based simulation detection of After Etch Inspection (AEI) hotspots is enabled.
• Accurate prediction of etch bias for complex 2D patterns facilitates auto retargeting for After Develop Inspection (ADI) target generation.

Conclusions:

• The proposed AI-assisted composite etch model effectively addresses the challenges of LELE process simulation and hotspot detection.
• The model enhances precision in patterning complex foundry circuits for advanced semiconductor nodes.
• The framework shows potential for adaptation to other complex MPT challenges like Self-Aligned Reverse Patterning (SARP) + Cut.