Care Areas for SEM Defect Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current defect detection methods in semiconductor manufacturing, particularly using scanning electron microscopes (SEM), are inefficient due to lack of region-specific selectivity, high throughput costs, misclassification issues, and the need for processing entire field of view images, which leads to increased computation and nuisance detection problems.
Innovation Solution
Implementing a care area defined by an inspector tool, such as a broad band plasma tool, to focus defect detection within specific areas of interest on the SEM, optimizing the detection process by using design-referenced defect location accuracy and bounding boxes to reduce unnecessary image processing and improve defect classification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the entire field of view is processed for defect detection using SEM, then comprehensive defect coverage is achieved, but computation time and processing cost increase significantly
Solution Approach 1:
The field of view is segmented into multiple regions of interest (ROIs) based on defect probability maps. Instead of processing the entire FOV uniformly, the system divides it into high-probability ROIs that require detailed inspection and low-probability regions that can be processed more quickly or skipped, thereby reducing overall computation time while maintaining detection reliability in critical areas.
Solution Approach 2:
Different processing qualities are applied to different regions of the field of view. High-probability regions receive full-resolution, detailed defect detection processing, while low-probability regions receive reduced processing. This local differentiation optimizes the balance between detection reliability and computation time by concentrating resources where they are most needed.
2Reliability
If automatic defect detection processes complete SEM images within FOV, then all potential defects are captured, but nuisance defects and false positives increase
Solution Approach 1:
A defect probability map is generated before performing detailed defect detection. This preliminary step identifies regions with high probability of containing actual defects versus regions likely to contain nuisance or false positive defects. The detection process then focuses on high-probability regions, preventing nuisance defects in low-probability regions from being falsely identified, thereby reducing false positives while maintaining detection completeness.
Solution Approach 2:
The defect probability map serves as an intermediary between the raw SEM image and the defect detection algorithm. This intermediate representation guides the detection process by highlighting where to look for actual defects and where nuisance defects are likely to occur, allowing the system to distinguish between significant defects and false positives more effectively.
3Measurement precision
If manual defect classification is performed on all detected defects, then classification accuracy is maintained, but throughput and efficiency decrease
Solution Approach 1:
The defect review process is segmented into automated classification for low-probability regions and manual classification for high-probability regions. The system automatically classifies defects in regions where the probability map indicates low likelihood of actual defects, maintaining throughput while accepting lower accuracy for these less critical cases. Manual classification is reserved for high-probability regions where accuracy is most important, thus optimizing the balance between productivity and measurement precision.
Solution Approach 2:
Different classification qualities are applied locally based on defect probability. High-probability defects receive full manual classification attention for maximum accuracy, while low-probability defects receive automated classification to maintain throughput. This local differentiation allows the system to maintain high overall accuracy for critical defects while preserving productivity through automated processing of less critical cases.
4Measurement precision
If optical inspection tools use recipes optimized for defect of interest, then DOI detection sensitivity is improved, but nuisance defects may be misclassified
Solution Approach 1:
The defect probability map acts as an intermediary that reconciles the conflict between DOI-optimized inspection and overall classification accuracy. By using the probability map to guide which regions to inspect with DOI-optimized recipes versus which regions to inspect with more general recipes, the system maintains high DOI detection sensitivity in critical areas while reducing misclassification of nuisance defects in other areas through selective application of inspection strategies.
Data Source
AI summary
Use of care areas in scanning electron microscopes or other review tools can provide improved sensitivity and throughput. A care area is received at a controller of a scanning electron microscope from, for example, an inspector tool. The inspector tool may be a broad band plasma tool. The care area is applied to a field of view of a scanning electron microscope image to identify at least one area of interest. Defects are detected only within the area of interest using the scanning electron microscope. The care areas can be design-based or some other type of care area. Use of care areas in SEM tools can provide improved sensitivity and throughput.