LIC-CGAN: fast lithography latent images calculation method for large-area masks using deep learning.

Area of Science:

• Computational lithography
• Deep learning applications
• Semiconductor manufacturing

Background:

• Latent image calculation for large-area masks is critical but time-consuming in lithography simulation.
• Existing methods lack efficiency for complex, large-scale mask patterns.

Purpose of the Study:

• To develop a fast and accurate method for 3D latent image calculation of large-area masks.
• To accelerate lithography simulations using deep learning.

Main Methods:

• A deep learning approach, LIC-CGAN (Latent Image Calculation using Conditional Generative Adversarial Networks), is proposed.
• Large mask layouts are divided into clips; latent images are retrieved from a library or generated by CGANs.
• Local latent images are synthesized for the complete simulation.

Main Results:

• The LIC-CGAN method achieves high accuracy in latent image calculation.
• A speed-up factor of 2.5 to 4.7 times is observed compared to rigorous simulation methods.
• The approach is successfully applied to lithography simulations for display panels.

Conclusions:

• LIC-CGAN offers a significantly faster alternative for 3D latent image calculation of large-area masks.
• Deep learning effectively addresses the computational bottleneck in lithography simulations.
• This method enhances efficiency in display panel manufacturing processes.