EUVL Flare Modeling Using Multi-Sampling Kernels
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional modeling techniques are inaccurate and computationally inefficient when used to model extreme ultraviolet lithography (EUVL) flare, due to its high frequency components and long-range effects, which require high sampling rates and large convolutions, leading to significant performance issues.
Innovation Solution
The EUVL flare is modeled using kernels discretized at different sampling rates and ambits, with short-range, medium-range, and long-range components, where high sampling rates are used for high frequency components over small areas and low sampling rates for low frequency components over large areas, reducing computational resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a high sampling rate is used to discretize the point spread function to capture high frequency components, then model accuracy is improved, but computational complexity and processing time increase significantly
Solution Approach 1:
The patent segments the point spread function into multiple kernels, each discretized at a different sampling rate according to its spatial frequency characteristics. High-frequency components are captured with high sampling rates while low-frequency components use lower sampling rates, thereby maintaining model accuracy while reducing overall computational complexity.
Solution Approach 2:
Different regions of the point spread function are discretized with different sampling rates based on local frequency content. Areas with high-frequency variations use higher sampling rates, while areas with smooth variations use lower sampling rates, optimizing the balance between accuracy and computational efficiency.
2Measurement precision
If a large convolution area is used to model long-range EUVL flare effects, then model accuracy is improved, but computational resources and processing time increase
Solution Approach 1:
The large convolution area is segmented into multiple smaller kernels with different ambits (ranges). Each kernel handles a specific spatial range, allowing the model to capture long-range effects without requiring a single large convolution operation, thus reducing computational resources while maintaining accuracy.
Solution Approach 2:
The patent transforms the problem from a single large 2D convolution into a combination of multiple smaller convolutions with different spatial extents. This dimensional decomposition allows efficient processing by leveraging the fact that different spatial frequencies decay at different rates.
3Ease of manufacture
If conventional modeling techniques are used for EUVL flare, then implementation simplicity is maintained, but model accuracy and computational efficiency deteriorate
Solution Approach 1:
The conventional single-kernel approach is replaced with a segmented multi-kernel representation. Each kernel is independently discretized and convolved, making the implementation straightforward while dramatically improving model accuracy and computational efficiency through the segmentation strategy.
Data Source
AI summary
One embodiment of the present invention provides techniques and systems for modeling long-range extreme ultraviolet lithography (EUVL) flare. During operation, the system may receive an evaluation point in a layout. Next, the system may receive an EUVL model which includes kernels that are discretized at different sampling rates, and which have different sized ambits. Specifically, a kernel that is discretized using a low sampling rate may have a longer range than a kernel that is discretized using a high sampling rate. The system may then convolve the kernels with the layout at the evaluation point over their respective ambits. Next, the system may use the convolution results to determine an indicator value. The indicator value can be used for a number of applications, e.g., to predict pattern shapes that are expected to print on a wafer, to perform optical proximity correction, or to identify manufacturing problem areas in the layout.