Gradient-based inverse lithography for EUV masks via the waveguide method and a physics-informed neural operator

Researchers present a novel gradient-based inverse lithography technology (ILT) for extreme ultraviolet (EUV) masks that uses physics-informed neural operators and automatic differentiation to optimize mask absorber permittivity. The method combines a differentiable waveguide approach with waveguide neural operators (WGNO) to recover mask structures achieving desired field patterns on wafers, demonstrated on realistic 2D and 3D absorbers at 11.2 nm wavelengths.

This research addresses a critical challenge in semiconductor manufacturing: designing EUV masks with optimal absorber properties to achieve precise field patterns on fabricated wafers. The work bridges computational physics and machine learning by treating the complete diffraction model as a differentiable engine, enabling gradient-based optimization of mask structures through automatic differentiation. This approach overcomes traditional inverse design limitations by leveraging neural operators that encode physics constraints directly into the optimization process.

EUV lithography represents the cutting edge of chip fabrication technology, enabling smaller feature sizes essential for advanced semiconductor nodes. The challenge of mask design has historically relied on computationally expensive iterative simulations and expert intuition. By making the entire forward model differentiable and integrating physics-informed neural operators, the researchers enable direct optimization of mask parameters—specifically the permittivity of absorber materials like tantalum boron nitride, lanthanum, and uranium.

The practical implications extend to semiconductor manufacturers and equipment suppliers developing next-generation fabrication tools. More efficient mask design processes reduce time-to-market for new chip technologies and lower development costs. The numerical demonstrations on realistic 3D absorber structures suggest the method translates from theory to practical application, indicating potential adoption in manufacturing workflows.

Future development should focus on validating these designs through actual mask fabrication and testing, scaling the method to full-chip layouts with millions of features, and exploring applicability to next-generation lithography approaches beyond EUV.

• →Gradient-based inverse lithography combines differentiable waveguide methods with physics-informed neural operators for EUV mask optimization
• →The framework successfully recovers optimal absorber permittivity for TaBN, lanthanum, and uranium materials at 11.2 nm wavelengths
• →Automatic differentiation of the full forward diffraction model enables direct optimization without iterative trial-and-error processes
• →Method demonstrated effectiveness on realistic 2D and 3D mask absorber structures relevant to semiconductor manufacturing
• →Technology reduces computational overhead and design iteration cycles for advanced chip fabrication mask development