Gravity Point Image Matching for SEM Pattern Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image processing systems for scanning electron microscopes face challenges in achieving accurate pattern matching between deformed semiconductor wafer patterns and CAD data, leading to insufficient positional accuracy and increased manual effort in re-registering patterns, which decreases inspection efficiency.
Innovation Solution
An image processing system that generates a gravity point distribution image from one image and a group of edge line segments from another, calculating a matching score based on their alignment to determine the optimal positional match, thereby avoiding edge matching and reducing deformation-related inaccuracies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If edge-based pattern matching methods (normalized correlation method or generalized Hough conversion method) are used to match CAD data with observation images, then the inspection process can be automated, but the positional accuracy becomes insufficient when patterns are deformed due to manufacturing variations
Solution Approach 1:
The patent changes the matching parameter from edge-based to gravity point-based. Instead of comparing edge positions and shapes between CAD data and observation images, the system calculates gravity points (center of mass) for both the registered pattern and detected patterns, then matches these gravity points. This parameter transformation makes the matching process robust to manufacturing deformations while maintaining automation.
2Measurement precision
If manual re-registration of patterns is performed to improve matching accuracy, then the positional accuracy can be enhanced, but the inspection efficiency decreases
Solution Approach 1:
The system performs self-alignment by automatically calculating and matching gravity points without requiring manual intervention for re-registration. The gravity point matching algorithm inherently compensates for deformations, enabling the system to self-correct positioning errors while maintaining high inspection efficiency through full automation.
3Adaptability or versatility
If edge matching is performed between deformed observation patterns and CAD data, then the pattern recognition can be achieved, but the matching fails or produces inaccurate results due to deformation
Solution Approach 1:
The patent extracts the gravity point information from both the registered pattern and the observed pattern, separating this critical positioning information from the deformed edge details. By extracting only the gravity point coordinates and ignoring the deformed edge shapes, the system achieves reliable matching even when patterns are significantly deformed by manufacturing variations.
Data Source
AI summary
A scanning electron microscope comprises an image processing system for carrying out a pattern matching between a first image and a second image. The image processing system comprises: a paint-divided image generator for generating a paint divided image based on the first image; a gravity point distribution image generator for carrying out a smoothing process of the paint divided image and generating a gravity point distribution image; an edge line segment group generation unit for generating a group of edge line segments based on the second image; a matching score calculation unit for calculating a matching score based on the gravity point distribution image and the group of edge line segments; and a maximum score position detection unit for detecting a position where the matching score becomes the maximum.