Contour Based Defect Detection Using Deep Learning Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for detecting defects in semiconductor patterns are slow, require careful setup, and struggle with complex patterns and pattern intersections, lacking flexibility and robustness, especially in imaging conditions, and fail to provide quantitative pattern characterization and hot spot detection.
Innovation Solution
A system utilizing deep learning-based models to generate simulated and actual contours of semiconductor patterns, allowing for robust and generalized pattern fidelity measurements, including the use of first and second learning-based models to compare expected and actual contours for defect detection, which can adapt to different pattern types and image modalities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional defect detection methods are used, then defects can be detected on wafers, but the inspection process is slow and requires careful setup for each site
Solution Approach 1:
The patent transforms the inspection approach by changing from traditional optical imaging parameters to electron beam imaging parameters, enabling faster inspection speeds while reducing setup complexity through automated site generation from design data
Solution Approach 2:
The system performs preliminary actions by automatically generating inspection sites from design data before actual inspection begins, eliminating the need for manual setup and accelerating the inspection process
2Reliability
If traditional contour extraction methods are used, then pattern contours can be identified, but the methods fail for complex patterns and pattern intersections
Solution Approach 1:
The patent replaces traditional mechanical/optical contour extraction methods with electron beam-based imaging and advanced image processing algorithms, achieving reliable contour extraction for complex patterns and intersections that were previously impossible
3Measurement precision
If high magnification optical systems or SEM are used for defect review, then higher resolution defect information can be obtained, but the process is slow and requires discrete location scanning
Solution Approach 1:
The system implements periodic action by using electron beam scanning at optimized scan rates that balance resolution and throughput, enabling fast defect review without sacrificing measurement precision through intelligent sampling strategies
4Ease of operation
If metrology processes are performed independently of inspection results, then process monitoring can be done at predetermined locations, but the locations cannot be optimized based on actual defect patterns
Solution Approach 1:
The system implements feedback by using inspection results to dynamically identify and prioritize hot spot locations for metrology processes, optimizing measurement precision while maintaining operational flexibility through adaptive site selection
Data Source
AI summary
Methods and systems for detecting defects in patterns formed on a specimen are provided. One system includes one or more components executed by one or more computer subsystems, and the component(s) include first and second learning based models. The first learning based model generates simulated contours for the patterns based on a design for the specimen, and the simulated contours are expected contours of a defect free version of the patterns in images of the specimen generated by an imaging subsystem. The second learning based model is configured for generating actual contours for the patterns in at least one acquired image of the patterns formed on the specimen. The computer subsystem(s) are configured for comparing the actual contours to the simulated contours and detecting defects in the patterns formed on the specimen based on results of the comparing.