Wafer Defect Detection via Sub-Image Segmentation and Variance Thresholding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for inspecting microscopic objects like wafers are inadequate in detecting defects due to limitations in comparing images and identifying anomalies, particularly in handling noise and variations across images.
Innovation Solution
A method that divides images of wafers into sub-images, compares them to defectless references, estimates variance, and performs further analysis on defect candidates using threshold values and difference images to identify and filter out defects, incorporating techniques like peer-group filtering and same-smear re-sampling to minimize noise and misalignment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional image comparison methods are used for defect detection, then the inspection process is simple, but the defect detection accuracy is insufficient due to noise and variations across images
Solution Approach 1:
The patent divides the wafer image into multiple sub-images representing different regions. Each sub-image is independently compared to its corresponding reference sub-image, enabling localized defect detection with higher precision while managing complexity through modular processing
Solution Approach 2:
The patent performs preliminary actions including image alignment, variance estimation, and reference image generation before actual defect comparison. These preparatory steps reduce noise and variations in advance, improving detection accuracy without requiring complex real-time processing
Solution Approach 3:
The patent applies different processing strategies to different image regions based on their characteristics. Bright portions are handled differently from patternless portions, and defect candidates are selected based on local variance thresholds, optimizing detection precision for each region type
2Reliability
If multiple images are compared to improve defect detection, then the detection sensitivity increases, but the computational complexity and processing time increase
Solution Approach 1:
The patent performs full image alignment and variance estimation on selected sub-images rather than the entire wafer image. This partial action approach maintains high detection sensitivity by comparing multiple images where needed while reducing overall processing time by limiting intensive operations to specific regions
Solution Approach 2:
By segmenting the wafer image into sub-images, the patent enables parallel processing of multiple regions. Each sub-image can be processed independently, reducing total processing time while maintaining high sensitivity through multi-image comparison in each region
3Reliability
If threshold values are used to filter defect candidates, then false alarms are reduced, but subtle defects may be missed
Solution Approach 1:
The patent applies local variance thresholds specific to each image location rather than a global threshold. The threshold at each location is determined by the estimated variance at that location, allowing subtle defects to be detected in low-variance regions while filtering false alarms in high-variance regions
Solution Approach 2:
The patent dynamically adjusts the threshold parameter based on local image characteristics. The threshold value at each location is a function of the estimated variance at that location, changing the detection sensitivity adaptively to maintain both false alarm reduction and subtle defect detection
Data Source
AI summary
A method for inspecting a wafer including a multiplicity of dies, the method including dividing an image of at least a portion of the wafer into a plurality of sub-images each representing a sub-portion of the wafer and selecting at least one defect candidate within each sub-image by comparing each sub-image to a corresponding sub-image of a reference including a representation, which is assumed to be faultless, of the portion of the wafer.