Cross-Layer Image Alignment Using SVD Noise Offsets
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for aligning images of different layers on a specimen, particularly in semiconductor manufacturing, face challenges due to process variations and low image correlation, leading to poor alignment and subsequent inspection inaccuracies.
Innovation Solution
A system and method utilizing singular value decomposition (SVD) based noise images to compute global offsets for aligning images of different layers, leveraging correlated noise between layers for accurate alignment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional image alignment methods are used for aligning images of different layers, then the alignment process is simple, but the alignment accuracy deteriorates due to low image correlation and process variations
Solution Approach 1:
The patent introduces a reference layer as an intermediary element to facilitate alignment between different layers. The reference layer contains alignment marks that serve as a common reference for multiple layers, enabling accurate alignment even when direct correlation between layers is low. This mediator approach resolves the contradiction by providing a stable reference point that improves alignment accuracy without requiring complex direct layer-to-layer matching algorithms
Solution Approach 2:
The patent transforms the alignment problem from direct image correlation to feature-based correlation by extracting and comparing alignment marks. This parameter transformation changes the basis of alignment from pixel-level image data to extracted feature coordinates, improving robustness against process variations and low image correlation while maintaining computational efficiency
2Reliability
If images are aligned without considering process variations, then the alignment process is fast, but the inspection accuracy deteriorates due to misalignment
Solution Approach 1:
The patent performs alignment mark detection and coordinate extraction in advance during the imaging process, storing these reference coordinates for later alignment operations. This preliminary action separates the time-consuming feature extraction from the actual alignment operation, enabling fast alignment processing while maintaining high accuracy by using pre-computed reference data that accounts for process variations
Solution Approach 2:
The patent replaces complex iterative alignment algorithms with a direct coordinate transformation approach based on detected alignment marks. Instead of using computationally intensive image registration methods, the system uses analytical geometry to compute transformation parameters from mark coordinates, significantly reducing processing time while improving reliability through explicit modeling of process variations
3Measurement precision
If direct image comparison is used without proper alignment, then the process is simple, but defect detection accuracy deteriorates due to misalignment
Solution Approach 1:
The patent introduces aligned coordinate systems as an intermediary framework that maps pixels from different layers to common physical locations. This intermediary coordinate transformation enables accurate defect detection by ensuring that comparisons are made at corresponding positions, resolving the contradiction by adding a systematic alignment layer that improves detection accuracy while maintaining processing clarity through structured coordinate mapping
Data Source
AI summary
Methods and systems for aligning images of different layers on a specimen are provided. The embodiments may be particularly useful for singular value decomposition (SVD) based cross-layer alignment for defect detection. One system includes a computer subsystem configured for generating first and second SVD based noise images for first and second images, respectively, of first and second layers, respectively, on a specimen. The first layer is formed subsequent to formation of the second layer on the specimen. The computer subsystem is also configured for computing one or more global offsets by aligning the first SVD based noise images to the second SVD based noise images, applying the one or more global offsets to the second images to thereby align the second images to the first images, and determining information for the specimen based on the aligned first and second images.