Automatic Defect Classifier Readiness Evaluation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current automated defect classification systems in the semiconductor industry face challenges in determining the sufficiency of pre-classified data for training classifiers and maintaining classifier performance over time, leading to inefficiencies in automatic defect classification and potential misclassification of defects.
Innovation Solution
The system evaluates a readiness criterion for defect classes based on benchmark classification results, selects suitable inspection data for training automatic classifiers, and continuously refines the training data by incorporating manual classifications to ensure accurate and consistent automatic defect classification, while alerting operators of potential production issues.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If automated defect classification systems use pre-classified data for training classifiers, then classification speed and productivity improve, but the reliability and accuracy of classification deteriorate due to insufficient or inadequate training data
Solution Approach 1:
The system performs preliminary classification of defects into core classes and subclasses before final classification. This staged approach allows the system to use pre-classified data efficiently while maintaining accuracy by refining classifications in subsequent stages rather than relying on a single automated pass
Solution Approach 2:
Human inspectors serve as intermediaries to review and verify classifications that the automated system cannot confidently determine. This hybrid approach combines the speed of automated classification with the reliability of human judgment for edge cases, resolving the contradiction between productivity and accuracy
2Measurement precision
If the system continuously trains classifiers with more data, then classification accuracy improves, but the complexity of the system increases
Solution Approach 1:
The classification system is segmented into multiple independent classifiers, each specialized for specific defect types or classes. This modular architecture allows the system to achieve high accuracy through specialized models without requiring a single overly complex classifier, managing system complexity through functional decomposition
Solution Approach 2:
The system applies classification only to the extent necessary - using automated classifiers for clear cases and human review only for ambiguous cases. This partial application of complex classification logic reduces overall system complexity while maintaining accuracy where it matters most
Data Source
AI summary
A method for classification includes receiving inspection data associated with a plurality of defects found in one or more samples and receiving one or more benchmark classification comprising a class for each of the plurality of defects. A readiness criterion for one or more of the classes is evaluated based on the one or more benchmark classification results, wherein the readiness criterion comprises for each class, a suitability of the inspection data for training an automatic defect classifier for the class. A portion of the inspection data is selected corresponding to one or more defects associated with one or more classes that satisfy the readiness criterion. One or more automatic classifiers are trained for the one or more classes that satisfy the readiness criterion using the selected portion of the inspection data.