Digital Matched Filter for Semiconductor Defect Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current digital filter techniques for substrate inspection are limited by a predefined number of templates, which insufficiently increases defect detection sensitivity for challenging inspection settings, particularly in semiconductor wafer inspection.
Innovation Solution
A system and method that acquires multiple inspection images from different locations, generates an aggregated defect profile, calculates noise correlation characteristics, and creates a matched filter based on these profiles and characteristics to enhance defect detection sensitivity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If predefined digital filter templates are used, then the inspection system operates with simple and fast processing, but defect detection sensitivity is insufficient for challenging inspection settings
Solution Approach 1:
The system performs preliminary actions by acquiring multiple inspection images from different locations and generating an aggregated defect profile before actual defect detection. This pre-processing step creates a customized filter template that captures the specific characteristics of defects in the inspected sample, thereby improving detection sensitivity without requiring complex real-time processing during inspection
Solution Approach 2:
The system changes the parameters of the filter by generating a customized matched filter based on aggregated defect profiles and noise correlation characteristics calculated from actual inspection images. This dynamic parameter adjustment allows the filter to adapt to specific defect types and noise conditions, significantly improving defect detection sensitivity compared to static predefined templates
2Measurement precision
If multiple inspection images are acquired and processed to generate customized matched filters, then defect detection sensitivity improves, but processing time and computational load increase
Solution Approach 1:
The system performs filter generation as a preliminary action using a representative set of inspection images acquired from different locations. By completing the computationally intensive filter generation process before actual defect detection, the system enables fast real-time inspection while maintaining high sensitivity. The matched filter can be generated once and reused for multiple inspections under similar conditions
Solution Approach 2:
The system acquires multiple inspection images from different locations to generate the aggregated defect profile, which provides a more comprehensive representation of defect characteristics. This excessive sampling approach ensures the filter captures all relevant defect patterns, improving sensitivity while the computational load is distributed across multiple images rather than requiring excessive processing of a single image
3Measurement precision
If a matched filter is generated based on aggregated defect profiles and noise correlation characteristics, then targeted defects are detected more effectively, but nuisance events may increase without proper filtering
Solution Approach 1:
The system changes the filter parameters by incorporating noise correlation characteristics calculated from the inspection images into the matched filter generation process. This allows the filter to distinguish between actual defects and noise patterns by adapting to the specific noise characteristics of the inspection system and sample, thereby reducing nuisance events while maintaining high defect detection accuracy
Solution Approach 2:
The system uses feedback from the calculated noise correlation characteristics to refine the matched filter parameters. By analyzing the noise patterns in the inspection images and incorporating this information into the filter design, the system creates a more selective filter that responds to actual defects while suppressing noise-induced false positives, thus reducing nuisance events
Data Source
AI summary
Enhanced defect detection of a sample includes acquiring two or more inspection images from a sample from two or more locations of the sample for a first optical mode. The defect detection also generates an aggregated defect profile based on the two or more inspection images from the two or more locations for the first optical mode for a selected defect type and calculating one or more noise correlation characteristics of the two or more inspection images acquired from the two or more locations for the first optical mode. Defect detection further includes the generation of a matched filter for the first optical mode based on the generated aggregated defect profile and the calculated one or more noise correlation characteristics.