Effective multi-objective inverse lithography technology at full-field and full-chip levels with a hybrid dynamic...

Area of Science:

• Semiconductor manufacturing
• Optical lithography
• Computational imaging

Background:

• Current inverse lithography technology (ILT) typically uses single-objective optimization, leading to suboptimal performance across different field points due to varying system aberrations.
• Existing multi-objective optimization algorithms (MOAs) struggle with incomplete target prioritization, causing over- or under-optimization.
• High-performance imaging with both fidelity and uniformity is critical for advanced lithography, particularly extreme ultraviolet lithography (EUVL).

Purpose of the Study:

• To develop a novel multi-objective ILT approach capable of achieving high-performance imaging across the entire die.
• To address the limitations of current MOAs in handling complex prioritization for multiple imaging targets.
• To introduce and validate a hybrid dynamic priority (HDP) algorithm for advanced ILT applications.

Main Methods:

• Investigated and developed a multi-objective ILT framework incorporating a hybrid dynamic priority (HDP) algorithm.
• Designed a hybrid criterion for balanced target completion and prioritization.
• Applied the HDP algorithm to multi-field wavefront error-aware source mask optimization (SMO) and multi-clip source optimization (SO) problems.

Main Results:

• Achieved high-fidelity and high-uniformity images across multi-field and multi-clip areas.
• The HDP algorithm improved image uniformity at full-field points by up to 31.1% compared to current MOAs in wavefront error-aware SMO.
• Demonstrated the HDP algorithm's universality and superior performance in multi-clip source optimization.

Conclusions:

• The developed HDP algorithm effectively addresses the challenges of multi-objective ILT by enabling balanced prioritization of imaging targets.
• HDP significantly enhances imaging uniformity and fidelity across diverse lithography conditions, outperforming existing MOAs.
• The HDP algorithm is a more qualified approach for multi-objective ILT optimization, paving the way for improved semiconductor manufacturing processes.