Substrate Inspection Filter Image Creation for Pseudo Defect Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current substrate inspection methods for semiconductor manufacturing face challenges in distinguishing real defects from pseudo defects, leading to incorrect detection of defects, especially with varying mottle states on substrates, which complicates the setting of appropriate threshold values for defect detection.
Innovation Solution
The method involves creating a filter image by replacing pixel values on the circumference of a circle in a registered image with the maximum value of selected pixel elements, and using this filter image to subtract from inspection images, allowing for judgment of defects based on reference values without decreasing sensitivity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a relatively large threshold value is set to reduce pseudo defects, then the number of pseudo defects detected in error is reduced, but real defects which are supposed to be detected cannot be detected
Solution Approach 1:
The patent applies local quality by setting different threshold values for different regions of the substrate. Specifically, the inspection area is divided into a first area (center region) and a second area (peripheral region), with different threshold criteria applied to each. This allows the system to maintain high sensitivity in the center area while being more lenient in peripheral areas where mottles are common, thereby resolving the contradiction between detecting all real defects and filtering out pseudo defects.
Solution Approach 2:
The patent segments the substrate inspection area into multiple distinct regions (first area and second area) with different defect detection criteria. By dividing the inspection space and applying region-specific thresholds, the system can simultaneously achieve high sensitivity for real defects in critical areas while reducing false positives from mottles in peripheral areas, thus resolving the technical contradiction.
2Reliability
If a relatively large threshold value is set to reduce pseudo defects, then the number of pseudo defects detected in error is reduced, but the sensitivity of detecting defects decreases
Solution Approach 1:
The patent implements local quality by assigning different threshold values to different spatial regions of the substrate. The first area (center) uses a lower threshold to maintain high detection sensitivity, while the second area (periphery) uses a higher threshold to reduce false positives from mottles. This regional differentiation allows the system to maintain overall high productivity by not sacrificing sensitivity in critical areas while still reducing pseudo defects overall.
3Ease of operation
If the threshold value is set uniformly across the entire substrate surface, then the inspection process is simple, but it is difficult to appropriately set the threshold value taking the variation in the state of the mottle into consideration
Solution Approach 1:
The patent applies local quality by establishing region-specific threshold values that account for the varying mottle characteristics across different substrate areas. The center region (first area) and peripheral region (second area) have different threshold criteria matched to their respective mottle patterns, improving detection accuracy while maintaining operational simplicity through automated regional classification.
Solution Approach 2:
The patent segments the substrate into multiple inspection areas with different threshold settings. This segmentation allows the system to address the variability in mottle states across different regions by applying appropriate thresholds to each area, thereby improving reliability without significantly complicating the inspection process through automated region identification.
Data Source
AI summary
An image creation method of creating a filter image for removing a pseudo defect to inspect presence/absence of a defect on a substrate includes a filter image creation step of creating the filter image by replacing a picture element value of any one of picture elements located on a circumference of a circle about a center position of a registered image with a maximum value of picture element values of a plurality of picture elements selected from among the picture elements located on the circumference.