Binocular Wafer Pre-Alignment for Precise Edge Notch Positioning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing wafer pre-alignment systems, both mechanical and optical, face challenges in achieving high precision and efficiency due to complex structures and limited positioning capabilities.
Innovation Solution
A vision-based wafer pre-alignment platform utilizing binocular camera groups with adaptive multiscale morphological filtering and deep convolution feature networks for precise edge feature point extraction and alignment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If a mechanical pre-alignment system is used, then the structure is stable, but the positioning precision is limited to micron level due to mechanical design constraints
Solution Approach 1:
The patent replaces the traditional mechanical pre-alignment system with a vision-based system using binocular cameras and image processing algorithms. The mechanical contact-based positioning is substituted by optical detection and computational alignment, eliminating mechanical precision limitations and achieving sub-pixel positioning accuracy through software-based edge detection and feature point matching.
Solution Approach 2:
The patent creates a virtual copy of the wafer structure by detecting and reconstructing the notched rod pattern in the image domain. By copying the geometric features (notch position, rod endpoints) from the physical wafer to a digital model, the system performs alignment calculations in the virtual space, achieving high precision without mechanical contact.
2Productivity
If traditional optical detection is used, then the detection capability is sufficient, but the alignment efficiency is low due to sequential processing
Solution Approach 1:
The patent performs preliminary detection of all critical features (notch position, rod endpoints, wafer edge) simultaneously in a single imaging operation. By capturing all necessary alignment information in one shot rather than sequential measurements, the system eliminates multiple adjustment steps and achieves rapid one-time alignment, significantly improving throughput.
Solution Approach 2:
The patent merges multiple detection functions into a unified vision system that simultaneously captures and processes notched rod features, wafer edge contours, and alignment mark positions. By combining these separate detection tasks into a single integrated image processing workflow, the system reduces total alignment time while maintaining comprehensive positioning accuracy.
3Measurement precision
If simple edge detection algorithms are used, then the system is easy to implement, but the positioning precision and edge feature extraction accuracy are insufficient
Solution Approach 1:
The patent employs dynamic, adaptive image processing that adjusts detection parameters based on image quality and feature characteristics. The system uses iterative refinement algorithms that adaptively adjust edge detection thresholds and feature matching criteria, allowing high precision extraction of notched rod endpoints and notch center positions while maintaining computational efficiency through conditional processing branches.
Solution Approach 2:
The patent introduces intermediate processing steps including adaptive thresholding, morphological operations, and feature point validation as mediators between raw image data and final alignment coordinates. These intermediary algorithms refine edge detection accuracy by filtering noise, enhancing contrast, and validating detected features against geometric constraints, achieving sub-pixel precision without requiring complex hardware modifications.
Data Source
AI summary
A vision-based wafer pre-alignment platform and alignment method perform image acquisition and processing based on two groups of binocular cameras. To maintain an outer edge contour of a wafer, an adaptive image denoising method of a wafer image is proposed pertinently. Further, the present disclosure defines edge feature points of the wafer according to shape features of the wafer to form feature points constituting a triangle, facilitating extracting the feature points according to an AAM model and matching the feature points according to an isosceles triangle rule. Finally, a position deviation of the wafer is calculated according to spatial coordinates of the feature points and positions of the feature points, an alignment deviation e is calculated according to the obtained spatial position coordinates of the feature points, and the wafer is pre-aligned based on the deviation value e.