Modeling Non-Manhattan ILT Masks with Rotated 2D Kernels
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional lithography mask modeling techniques are inadequate for inverse lithography technology (ILT) masks with non-straight or curvilinear edges, as they are optimized for Manhattan patterns, leading to challenges in accurately modeling and fabricating ILT masks with arbitrary angles.
Innovation Solution
The use of two-dimensional kernels that can be quickly rotated to accurately model ILT lithography masks with freeform or arbitrary mask patterns, enabling the generation of a correction field that improves the accuracy of near field simulation and fabrication.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional mask modeling techniques are used for ILT masks, then the modeling process is simple and fast, but the modeling accuracy is insufficient for non-Manhattan patterns with arbitrary angles
Solution Approach 1:
The patent segments the continuous mask pattern into discrete super-pixel units and groups them into super-patterns. This segmentation allows the complex arbitrary-angle patterns to be broken down into manageable components that can be processed using modified conventional techniques, thereby improving modeling accuracy without overwhelming computational complexity.
Solution Approach 2:
The patent introduces a super-pixel dimension that aggregates multiple conventional pixels into larger units. This dimensional transformation allows the model to capture arbitrary-angle patterns more effectively by operating at a coarser resolution level, improving accuracy while reducing the computational burden of processing every individual pixel.
2Manufacturing precision
If conventional mask modeling techniques are used, then the process is easier to implement, but the depth-of-focus and pattern fidelity are insufficient
Solution Approach 1:
The patent performs preliminary aggregation of pixels into super-pixels and super-patterns before the actual mask modeling process. This preliminary action simplifies the subsequent modeling steps by reducing the data volume and highlighting the essential pattern features, thereby improving pattern fidelity while maintaining implementation ease through a structured two-stage process.
Solution Approach 2:
The super-pixel and super-pattern structures serve as intermediary representations between the raw mask design and the final modeled pattern. These intermediaries capture the essential geometric features of arbitrary-angle patterns while filtering out unnecessary fine-grained details, thus improving pattern fidelity without significantly complicating the manufacturing process.
3Measurement precision
If high-accuracy modeling of arbitrary-angle patterns is achieved, then pattern fidelity improves, but computational resources and processing time increase
Solution Approach 1:
The patent applies partial action by focusing computational resources on the essential super-pattern features rather than processing every pixel detail. This selective approach captures the critical arbitrary-angle pattern characteristics with sufficient accuracy while avoiding the excessive computational burden of full-resolution processing, thereby reducing processing time while maintaining pattern accuracy.
Data Source
AI summary
A mask layout containing a non-Manhattan pattern is received. The received mask layout is processed. An edge of the non-Manhattan pattern is identified. A plurality of two-dimensional kernels is generated based on processed pre-selected mask layout samples. The two-dimensional kernels each have a respective rotational symmetry. The two-dimensional kernels are applied to the edge of the non-Manhattan pattern to obtain a correction field for the non-Manhattan pattern. A thin mask model is applied to the non-Manhattan pattern. The thin mask model contains a binary modeling of the non-Manhattan pattern. A near field of the non-Manhattan pattern is determined by applying the correction field to the non-Manhattan pattern having the thin mask model applied thereon. An optical model is applied to the near field to obtain an aerial image on a wafer. A resist model is applied to the aerial image to obtain a final resist image on the wafer.