Wafer Map Defect Detection Using Unsupervised Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In semiconductor manufacturing, manual inspection of wafer maps is impractical due to the vast number of wafers scanned daily, and traditional rule-based systems struggle to detect all types of defects, especially since normal wafers outnumber defective ones, requiring significant manpower and resources for identification.
Innovation Solution
A wafer map recognition method utilizing unsupervised algorithms to generate feature datasets from wafer maps, followed by a clustering algorithm to identify potential defects without human intervention, thereby automating the detection process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual inspection of wafer maps is performed, then defect detection accuracy is improved, but productivity deteriorates due to the vast number of wafers scanned daily
Solution Approach 1:
The patent replaces manual mechanical inspection with an automated image processing system that uses computer algorithms to analyze wafer maps. The system automatically detects defects by comparing wafer maps against reference patterns and identifying anomalies, eliminating the need for human inspectors while maintaining detection accuracy and significantly increasing throughput capability.
2Ease of manufacture
If rule based systems are used to identify defects, then common defect types can be detected, but the system cannot detect all types of abnormalities due to limited rule coverage
Solution Approach 1:
The patent transforms the defect detection approach from fixed rule-based parameters to adaptive image processing parameters. By using image processing techniques that analyze visual patterns, colors, and spatial relationships in wafer maps, the system can automatically adapt to various defect types without requiring pre-programmed rules for each specific abnormality, thereby significantly expanding defect type coverage.
3Reliability
If traditional methods are used to find defective wafers, then all wafers can be inspected, but significant manpower and resources are required due to the large number of normal wafers
Solution Approach 1:
The patent extracts and identifies only the defective wafers from the large population of normal wafers using automated image processing. The system efficiently scans through all wafers, applies defect detection algorithms, and extracts only those wafer maps that contain abnormalities, thereby maintaining inspection completeness while dramatically reducing the resources needed compared to manual inspection of every single wafer.
Data Source
AI summary
A wafer map recognition method using artificial intelligence includes obtaining wafer maps of a plurality of wafers; performing an unsupervised algorithm on the wafer map of each wafer in the plurality of wafers to generate a feature data set for the corresponding wafer map; and performing a clustering algorithm according to a plurality of feature data sets for the plurality of wafer maps to find a wafer map with a potential defect.