SRAF Seed Map Generation for Lithography Mask Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing SRAF placement techniques in semiconductor manufacturing, such as rule-based and inference mapping lithography, suffer from unsatisfactory accuracy due to inadequate consideration of exposure conditions and mask 3D effects, leading to inefficiencies in lithography image quality.
Innovation Solution
The method involves generating a sub-resolution assist feature (SRAF) seed map that considers exposure conditions like illumination intensity, numerical aperture, depth of focus, and aberrations, and addresses polarization due to mask 3D effects, using a comprehensive approach that includes all eigenvalues and eigenfunctions of the transmission cross coefficient to improve SRAF placement accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If rule-based SRAF placement techniques are used, then turn-around time is short, but accuracy is far-from-ideal
Solution Approach 1:
The patent pre-calculates and stores the transmission cross coefficient (TCC) kernel for the specific exposure tool before SRAF placement. This preliminary action allows the SRAF placement algorithm to directly use the pre-computed optical characteristics without performing complex calculations during the placement process, thereby maintaining short turn-around time while achieving high accuracy through comprehensive optical modeling.
Solution Approach 2:
The patent introduces the TCC kernel as an intermediary that bridges the exposure tool characteristics and the SRAF placement decision-making process. The TCC kernel serves as a pre-computed intermediary that encapsulates the complex optical interactions, allowing the placement algorithm to make accurate decisions without directly simulating the full optical physics during placement, thus balancing speed and accuracy.
2Device complexity
If conventional SRAF placement techniques are used, then process is simpler, but accuracy is unsatisfactory due to inadequate consideration of exposure conditions and mask 3D effects
Solution Approach 1:
The patent incorporates multiple exposure tool parameters (illumination intensity, numerical aperture, depth of focus, aberrations) and mask 3D effect parameters into the TCC kernel calculation. By changing and including these additional parameters in the optical model, the patent achieves comprehensive accuracy without significantly increasing process complexity, as all parameters are integrated into a unified TCC-based framework.
3Manufacturing precision
If comprehensive optical modeling including all eigenvalues and eigenfunctions of TCC is used, then SRAF placement accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent performs the computationally intensive decomposition of the TCC into eigenvalues and eigenfunctions in advance, storing the results in a TCC kernel. This preliminary computation eliminates the need to perform these calculations during SRAF placement, maintaining high accuracy through comprehensive optical modeling while avoiding the computational burden during the actual placement process.
Solution Approach 2:
The patent creates a simplified copy of the complex optical system in the form of the TCC kernel, which contains pre-computed eigenvalues and eigenfunctions. This kernel copy allows the placement algorithm to work with simplified data structures while still capturing the essential optical physics, thereby reducing computational complexity during placement while maintaining accuracy.
Data Source
AI summary
In a pattern formation method for a semiconductor device fabrication, an original pattern for manufacturing a photomask is acquired, a modified original pattern is obtained by performing an optical proximity correction on the original pattern, a sub-resolution assist feature (SRAF) seed map with respect to the modified original pattern indicating locations where an image quality is improved by an SRAF pattern is obtained, SRAF patterns are placed around the original pattern, the SRAF patterns and the modified original pattern are output as mask data, and the photo mask is manufactured using the mask data.