Mask Inspection Apparatus Defect Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional mask inspection methods struggle to detect fine defects while minimizing pseudo defects, as they often require a trade-off between sensitivity and noise reduction, leading to either overlooking real defects or detecting unnecessary shape or line width variations as defects.
Innovation Solution
An inspection method and apparatus that acquires an optical image of a mask pattern, generates a reference image reflecting the dimension distribution of the pattern, and re-compares the optical image with a regenerated reference image using the die-to-database comparison method to differentiate between real and pseudo defects, thereby reducing pseudo defect detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the threshold of defect determination is set to be high, then noise is reduced, but the original target defect of detection is overlooked
Solution Approach 1:
The patent applies preliminary action by performing a first defect detection using design pattern data before the actual inspection. This preliminary detection identifies potential defects and allows the system to adjust the reference image accordingly, enabling more accurate defect determination in the final inspection without requiring an overly sensitive threshold that would capture noise.
Solution Approach 2:
The patent changes the parameter of the reference image by regenerating it based on the dimension distribution of the pattern in the sample. This parameter change allows the system to adapt the reference image to the actual characteristics of the inspected pattern, enabling reliable defect detection without being overly sensitive to normal variations.
2Measurement precision
If the threshold of defect determination is set to be low, then fine defects are detected, but pseudo defects due to noise are detected
Solution Approach 1:
The patent applies local quality by considering the dimension distribution of the pattern in the sample when regenerating the reference image. This allows different regions of the pattern to be evaluated against their local dimensional characteristics, enabling fine defects to be detected while distinguishing them from noise-based pseudo defects that do not conform to the local dimension distribution.
Solution Approach 2:
The patent changes the parameter of the reference image by regenerating it based on the dimension distribution of the pattern in the sample. This parameter change enables the system to detect fine defects while filtering out pseudo defects, as the regenerated reference image reflects the actual dimensional characteristics of the inspected pattern.
3Measurement precision
If conventional mask inspection is performed with high sensitivity, then line width fluctuation is detected, but the number of detected defects becomes large due to shape defects
Solution Approach 1:
The patent applies preliminary action by performing defect detection using design pattern data before the actual inspection. This preliminary step identifies potential issues and allows the system to focus on truly defective regions, reducing the overall number of detected defects while maintaining high sensitivity for line width fluctuations.
Solution Approach 2:
The patent changes the parameter of the reference image by regenerating it based on the dimension distribution of the pattern in the sample. This parameter change enables the system to distinguish between genuine defects and normal variations, reducing the number of false defect detections while maintaining accurate line width measurement capability.
Data Source
AI summary
An inspection method and inspection apparatus comprising, acquiring an optical image of a pattern formed in a sample, generating a reference image corresponding to the optical image, comparing the optical image and the reference image using a die-to-database method to detect a defect in the optical image and storing information on the defect; regenerating a reference image by reflecting a dimension distribution of a pattern in the surface of the sample on the reference image, and re-comparing an optical image in which a defect is detected by the comparison using the die-to-database method and the regenerated reference image which corresponds to the optical image using the die-to-database method to detect the defect in the optical image in which the defect has been detected, storing information on the defect when the defect is redetected, and determining that the optical image has no defect.