Machine Learning Substrate Defect Pattern Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional systems rely on human expertise for classifying substrate defects, which is time-consuming, costly, and prone to inconsistencies, leading to incorrect diagnoses and inefficient corrective actions, resulting in increased production costs and equipment wear.
Innovation Solution
A method utilizing a trained machine learning model to classify substrate defect patterns automatically, processing data from inspection tools to generate images and provide standardized inputs, enabling consistent and efficient identification of defect patterns and recommended corrective actions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If human expertise is used for classifying substrate defects, then classification can be performed, but it is time-consuming, costly, and prone to inconsistencies
Solution Approach 1:
The patent replaces the mechanical system of human visual inspection and manual classification with an automated machine learning-based classification system. The system uses inspection tool data to generate images and feeds them to a trained machine learning model that automatically classifies defect patterns, eliminating human involvement in the classification process and thereby reducing time consumption while improving consistency and reliability.
Solution Approach 2:
The machine learning model enables the system to perform self-classification of defect patterns without requiring human expertise. The model has been trained on defect data and can autonomously identify and classify different defect types, making the system self-sufficient in the classification task and eliminating the need for continuous human intervention.
2Reliability
If human expertise is used for classifying substrate defects, then classification can be performed, but it is costly and leads to incorrect diagnoses
Solution Approach 1:
The patent replaces human-based classification with an automated machine learning system that processes inspection data and generates classifications. This substitution eliminates human errors and inconsistencies in diagnosis while maintaining high productivity through automated processing speeds.
Solution Approach 2:
The system incorporates feedback mechanisms where the machine learning model is trained on defect classifications and can learn from patterns in the data. The feedback loop ensures that the model continuously improves its diagnostic accuracy by analyzing defect patterns and adjusting its classification criteria, thereby reducing incorrect diagnoses.
3Manufacturing precision
If conventional inspection methods are used, then defect locations can be identified, but defect pattern classification is inconsistent and leads to inefficient corrective actions
Solution Approach 1:
The patent replaces conventional manual inspection methods with an automated machine learning-based system that not only identifies defect locations with high precision but also automatically classifies defect patterns. This substitution streamlines the process by eliminating manual classification steps, thereby improving the efficiency of corrective actions while maintaining or enhancing defect identification precision.
Data Source
AI summary
A method includes obtaining, by a processing device, data indicative of locations of defects of a substrate. The method further includes generating an image indicating the locations of the defects. The method further includes providing the image indicating the locations of the defects to a trained machine learning model. The method further includes obtaining, as output from the trained machine learning model, a classification of the locations of the defects. The method further includes performing a corrective action in view of the output from the trained machine learning model.