Semiconductor Image Registration Using Distance Maps for Noise Robustness
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current examination methods for semiconductor wafers struggle with accurately registering and comparing design images with inspection images, particularly in the presence of noise and variations, which hinders precise monitoring of fabrication processes and detection of defects.
Innovation Solution
A system and method utilizing distance transform operations on design and inspection images to generate transform images, followed by matching operations to identify corresponding areas, enabling accurate registration and detection of dimensional changes in structural elements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional image registration methods are used to compare design images with inspection images, then the process is simpler, but accuracy deteriorates due to noise and SEM signal intensity variations
Solution Approach 1:
The patent segments the image registration process into distinct stages: applying distance transform to convert images into distance maps, performing matching operations on transformed images, and using matching data to identify corresponding areas. This segmentation allows each stage to optimize for its specific function, improving overall registration accuracy while managing complexity through modular processing
Solution Approach 2:
The patent introduces transform images as an intermediary representation between the original design and inspection images. By converting both images into distance transform space, the system creates a common intermediate format that is more robust to noise and intensity variations, enabling more accurate matching operations while maintaining manageable processing complexity
2Measurement precision
If distance transform operations are applied to generate transform images and perform matching operations, then registration accuracy improves, but processing time increases
Solution Approach 1:
The patent applies distance transform operations as a preliminary step before matching operations. By pre-processing both design and inspection images into distance maps, the system prepares the data in advance to facilitate more efficient and accurate matching, reducing the computational burden during the actual comparison phase
Solution Approach 2:
The patent extracts key structural information from the original images by applying distance transform, which converts images into distance maps that emphasize geometric features while suppressing noise and intensity variations. This extraction of essential structural data enables accurate registration without processing the full complexity of the original images
Data Source
AI summary
There are provided systems and methods comprising obtaining an inspection image of a semiconductor specimen and a second image, applying a distance transform operation to at least part of the second image, or to at least part of an image derived from the second image, thereby obtaining at least one transform image, applying a distance transform operation to at least part of the inspection image, or to at least part of an image derived from the inspection image, thereby obtaining a transformed inspection image, performing a matching operation using data informative of the at least one transform image and data informative of the transformed inspection image, thereby obtaining matching data, and using the matching data to identify an area of the inspection image which matches the second image according to a matching criterion.