Inspection Apparatus for Semiconductor Mask Defect Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing inspection apparatuses face challenges in accurately detecting defects in semiconductor masks due to position gaps and spatial distortions, particularly in regions with sparse circuit patterns, leading to erroneous defect detection.
Innovation Solution
The inspection apparatus employs a defect detection device that estimates position differences and reliability values for partial regions, using SSD matching and reliability-weighted local shift amount estimation to correct for global and local shifts, and estimates distortion amounts to suppress false defect detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional defect detection methods are used, then defect detection can be performed, but erroneous defect detection occurs in regions with sparse circuit patterns due to position gaps and spatial distortion
Solution Approach 1:
The image is divided into multiple regions, and for each region, a local shift amount is estimated separately. This segmentation allows the system to handle position gaps and spatial distortions locally, improving measurement precision in sparse circuit pattern regions where global estimation would fail.
Solution Approach 2:
The patent applies different processing strategies to different regions of the image. In regions with sufficient circuit patterns, conventional estimation methods are used, while in sparse regions, the system uses reliability-based filtering and neighboring region information to maintain accurate defect detection.
2Measurement precision
If position gap and spatial distortion estimation is performed in sparse circuit pattern regions, then defect detection can be attempted, but estimation precision deteriorates leading to false defects
Solution Approach 1:
Before performing defect detection, the system pre-estimates shift amounts for all regions and calculates reliability values. This preliminary action allows the system to identify unreliable estimates in sparse regions and correct them using neighboring region information before the actual defect detection process.
Solution Approach 2:
The system uses reliability values as feedback to determine whether to trust an estimated shift amount. When reliability is low in sparse regions, the system feeds back this information and uses alternative methods (neighboring region data, reliability-weighted averaging) to correct the estimation, thereby maintaining measurement precision.
3Reliability
If reliability-weighted local shift amount estimation is applied, then false defect detection is suppressed, but device complexity increases
Solution Approach 1:
The system applies reliability-weighted processing selectively rather than uniformly. In regions with sufficient circuit patterns, conventional methods are used. In sparse regions where false defects occur, the system applies the more complex reliability-weighted estimation using neighboring region information, thus managing device complexity while maintaining reliability.
Data Source
AI summary
An inspection apparatus including an image generation device which generates a second image corresponding to a first image, and a defect detection device which detects a defect in the second image. Each of the first and second image includes partial regions each including pixels. The defect detection device is configured to estimate a first value indicating a position difference between the first and second image for each of the partial regions, based on a luminance difference between the first and second image, estimate a second value indicating a reliability of the first value for each of the partial regions, and estimate a position difference between the first and second image for each of the pixels, based on the first and second value estimated for each of the partial regions.