Robust and efficient inverse mask synthesis with basis function representation.

Area of Science:

• Optical Lithography
• Semiconductor Manufacturing
• Computational Imaging

Background:

• Mask optimization is crucial for resolution scaling in optical lithography.
• It effectively mitigates optical proximity effects but faces computational challenges.
• Existing methods often involve complex nonlinear optimization with numerous variables.

Purpose of the Study:

• To develop a more efficient mask optimization technique.
• To reduce the computational burden associated with mask pattern design.
• To improve the manufacturability of optimized lithographic masks.

Main Methods:

• Representing mask patterns using a set of basis functions.
• Incorporating this representation into optimization under nominal and defocus conditions.
• Utilizing an adaptive, two-step gradient-based optimization with varying kernel numbers.

Main Results:

• The proposed method significantly improves optimization efficiency, achieving several-fold speedups.
• Optimized patterns exhibit enhanced manufacturability compared to traditional pixel-based methods.
• The gradient calculation is simplified by using coefficients of basis functions.

Conclusions:

• Basis function representation offers an efficient alternative for mask optimization in optical lithography.
• The adaptive two-step optimization strategy enhances computational performance.
• This approach leads to improved pattern manufacturability, crucial for semiconductor fabrication.