US20250390025
2025-12-25
Physics
G03F7/70325
The patent application discusses a method for generating optimal near-field patterns and a corresponding mask manufacturing process. This approach involves creating a mutual interference complex diffraction pattern through the interaction of spherical waves with edge segments of a design layout. By applying a Kirchhoff boundary condition, a complex near-field is derived, which is then optimized to closely match a rigorous near-field of the design layout. The aim is to enhance photolithography processes in semiconductor manufacturing by improving pattern transfer onto substrates.
In semiconductor manufacturing, photolithography is crucial for transferring circuit patterns onto wafers. Masks, which carry these patterns, are pivotal in this process. The patent outlines a method to refine the near-field generation, which is essential for accurate pattern transfer. The process begins with designing a circuit layout and applying optical proximity correction (OPC) to prepare the mask data. The improved method aims to mitigate overfitting and expedite the generation of near-field patterns, thereby enhancing the efficiency of mask manufacturing.
The method involves obtaining a mutual interference complex diffraction pattern from a design layout's edge segments. This pattern is formed by the interaction of spherical waves with the edge segments when a plane wave is incident. A complex near-field is created by considering a mask's three-dimensional effects, which vary with the direction of wave incidence. The optimization process involves reducing discrepancies between the complex and rigorous near-fields using an artificial neural network, ensuring a precise match with the design layout.
Another aspect of the method includes correcting the complex near-field using a Volterra series to address errors between the complex and rigorous near-fields. This correction further refines the near-field generation, ensuring accuracy in the pattern transfer process. The method also encompasses generating an OPC model based on the optimized near-field, which is crucial for simulating and preparing the mask tape-out design data. This data is then used in the mask exposure process.
The implementation process starts with converting the design layout into a binary format, followed by extracting and differentiating the edge segments. A complex diffraction pattern is formed by scattering spherical waves across these segments. The method reflects the mask's 3D effects and corrects any errors, ultimately optimizing the near-field generation. The final steps involve creating an OPC-ed design layout, preparing mask data, and using this data to expose a substrate, thus completing the mask manufacturing process.