Wafer Defect Detection Using Standard Reference Image
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing semiconductor wafer inspection methods, such as multi-die automatic thresholding (MDAT), face instability due to color and process variations, leading to inconsistent defect detection and classification results across different dies and wafers, as they rely on reference images that can be affected by gray level distribution differences.
Innovation Solution
A method utilizing a standard reference image to stabilize defect detection and classification, where a median or average image is created from multiple dies to account for color variations, and a 2D scatter plot is generated to identify defects, allowing for consistent segmentation and attribute calculation across wafers and dies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multi-die automatic thresholding (MDAT) is used for defect detection, then defect detection capability is provided, but inspection stability and consistency deteriorate due to color and process variations across different dies and wafers
Solution Approach 1:
The patent introduces a standard reference image as an intermediary between the inspection system and the wafers being inspected. This reference image, created from a clean wafer or by removing die-repeater defects, serves as a stable mediator that eliminates the impact of color and process variations. By comparing all test images against this fixed reference rather than varying reference images, the system achieves consistent and stable defect detection across different dies and wafers.
Solution Approach 2:
The patent changes the reference parameter from dynamic, die-specific reference images to a static, standardized reference image. This parameter change stabilizes the comparison baseline, ensuring that segmentation breaks and thresholds remain consistent across all inspections. The standard reference image fixes the gray level distribution, preventing the shifting of 2D histograms that occurs with traditional methods.
2Measurement precision
If standard reference images are used to detect die-repeater defects, then sensitivity to single-die reticle defects is improved, but the complexity of creating and maintaining standard images increases
Solution Approach 1:
The patent performs preliminary action by creating the standard reference image in advance, during recipe setup or from a known clean wafer. This pre-established reference eliminates the need for complex real-time reference management during actual inspection. The standard image is prepared beforehand with all die-repeater defects removed, so that during production inspection, only simple comparison operations are needed, reducing operational complexity.
Data Source
AI summary
Systems and methods for classifying defects detected on a wafer are provided. One method includes detecting defects on a wafer based on output generated for the wafer by an inspection system. The method also includes determining one or more attributes for at least one of the defects based on portions of a standard reference image corresponding to the at least one of the defects. The method further includes classifying the at least one of the defects based at least in part on the one or more determined attributes.