High-fidelity source mask optimization for suppressing line-end shortening.

Area of Science:

• Computational lithography
• Semiconductor manufacturing
• Optical physics

Background:

• Source mask optimization (SMO) is crucial for correcting lithographic distortion.
• Line-end shortening remains a significant challenge, impacting image fidelity in advanced semiconductor nodes.
• Existing SMO methods struggle to effectively address line-end shortening.

Purpose of the Study:

• To propose a novel source mask optimization method for suppressing line-end shortening.
• To enhance lithographic fidelity, particularly in critical line-end regions.
• To improve the accuracy of pattern transfer in semiconductor manufacturing.

Main Methods:

• An adaptive hybrid weight method was developed to prioritize line-end regions during optimization.
• Weights were dynamically updated based on edge placement error (EPE) in each iteration.
• A cost function incorporating a normalized image log slope (NILS) penalty term was designed.
• The penalty term's scope was controlled by widening and extending split contours to mitigate line-end shortening.

Main Results:

• The proposed SMO method effectively suppressed line-end shortening.
• Lithographic fidelity was significantly improved compared to traditional SMO techniques.
• Adaptive weighting and NILS-based penalties demonstrated superior performance in critical areas.
• Simulation results validated the method's efficacy in advanced node lithography.

Conclusions:

• The developed adaptive SMO approach offers a robust solution for line-end shortening.
• This method enhances overall lithographic fidelity and pattern accuracy.
• It represents a significant advancement for semiconductor manufacturing at advanced nodes.