Computational Metrology for EUV Mask Defect Printability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods struggle to accurately predict defect printability on extreme ultraviolet (EUV) scanners due to differences in optics and illumination between EUV mask inspection and EUV scanners, leading to discrepancies in defect detection and wafer analysis, which is resource-intensive and expensive.
Innovation Solution
Utilizing neural networks trained with simulated mask inspection and scanner aerial images to predict scanner aerial images, leveraging models of mask inspection and scanner optics, and employing techniques like convolutional neural networks and generative adversarial networks to improve accuracy and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If conventional optical lithography methods are used to transfer patterns, then manufacturing process is simple, but manufacturing precision deteriorates when critical dimensions approach resolution limits
Solution Approach 1:
The patent introduces an intermediary computational metrology system that includes a scanner model, illumination model, and aerial image calculation engine. This intermediary system acts as a bridge between the physical mask inspection process and the actual scanner printing process, enabling accurate prediction of defect printability without requiring physical test printing. The computational model simulates the optical physics of the scanner to predict how defects will appear on wafers, thereby improving pattern transfer accuracy while avoiding the complexity and cost of extensive test printing.
2Measurement precision
If physical test printing is performed to verify defect printability, then measurement accuracy is high, but time consumption and cost increase
Solution Approach 1:
The patent creates a virtual copy of the scanner's optical system through computational modeling. Instead of performing physical test printing to verify defect printability, the system uses a scanner model that replicates the optical physics, illumination conditions, and aerial image formation processes. This virtual copy enables accurate prediction of defect printability on wafers without the time and cost of actual test printing, while maintaining measurement precision through physics-based simulations.
Solution Approach 2:
The patent performs preliminary computational analysis of defect printability before physical test printing or wafer fabrication. The system pre-calculates aerial images and defect printability metrics using the scanner model and illumination model, allowing engineers to identify and prioritize critical defects beforehand. This preliminary action eliminates the need for time-consuming trial-and-error test printing and accelerates the overall defect analysis process.
3Measurement precision
If mask inspection is performed with high precision, then defect detection accuracy improves, but discrepancy with actual scanner printing increases due to optical differences
Solution Approach 1:
The patent implements a feedback mechanism that connects mask inspection results with scanner printing predictions through computational modeling. The system takes mask inspection data as input, processes it through the scanner model and illumination model to predict actual printing outcomes, and compares these predictions with reference data or actual wafer results. This feedback loop continuously refines the computational models to better match the specific scanner's optical characteristics, thereby improving prediction reliability while maintaining high defect detection accuracy.
Solution Approach 2:
The patent adjusts key parameters in the computational model to match the specific scanner's optical characteristics. This includes modifying the scanner model parameters (numerical aperture, wavelength, focus conditions) and illumination model parameters (source shape, sigma values) to reflect the actual scanner being used. By dynamically adjusting these parameters based on the specific scanner configuration, the system bridges the gap between generic mask inspection and scanner-specific printing behavior, improving prediction reliability.
Data Source
AI summary
Systems for determining a scanner aerial image from a mask inspection image include a computer processor configured to receive the mask inspection image, wherein the mask inspection image has been generated by a mask inspection machine; and a computer processor configured to generate the scanner aerial image from the mask inspection image using a neural network. Systems include a computer processor configured to train a neural network with a set of images, such as with a simulated scanner aerial image and another image selected from a simulated mask inspection image, a simulated Critical Dimension Scanning Electron Microscope (CD-SEM) image, a simulated scanner emulator image and a simulated actinic mask inspection image.