Defect Classification via Electrical Design Intent Annotation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current wafer inspection methods lack the ability to accurately classify defects based on electrical design intent, leading to misclassification and increased manufacturing costs due to the inability to determine the criticality of defects, especially in cases where electrical structures are missing or incomplete.
Innovation Solution
A system and method that annotates design data with electrical design properties, allowing for the classification of defects by comparing inspection images to annotated design data, which includes patterns of interest with electrical properties, thereby determining the criticality of defects and improving defect classification accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional wafer inspection methods are used to detect defects, then defects can be identified, but the defects cannot be accurately classified based on electrical design intent, leading to misclassification
Solution Approach 1:
The system performs preliminary annotation of design data with electrical design properties before defect classification. Electrical design properties such as voltage, current, resistance, and circuit functionality are pre-assigned to different regions and structures in the design data, enabling subsequent accurate defect classification without losing electrical intent information
Solution Approach 2:
The system introduces an intermediary layer of electrical design property annotations that bridge the gap between physical defect detection and electrical functionality assessment. These annotations act as a mediator that connects inspection images to design intent, enabling accurate classification of defects based on their electrical impact
2Productivity
If defect inspection is performed without electrical design context, then inspection speed is maintained, but manufacturing costs increase due to misclassification and inability to determine defect criticality
Solution Approach 1:
The system applies electrical design properties locally to specific regions, structures, and patterns in the design data. Different areas are annotated with relevant electrical characteristics (e.g., high-voltage regions, sensitive signal lines, redundant structures), enabling rapid assessment of defect criticality without进行全面 analysis, thus maintaining inspection throughput while reducing misclassification costs
3Adaptability or versatility
If electrical structures are missing or incomplete in the design data, then inspection can proceed, but defect criticality cannot be determined, leading to increased misclassification
Solution Approach 1:
The system prepares design data in advance by annotating electrical design properties and establishing default classification rules for various structure types. This cushioning preparation ensures that even when electrical structures are missing or incomplete during inspection, the system can still classify defects accurately using the pre-established annotations and rules
Solution Approach 2:
The system incorporates feedback mechanisms where classification results and defect patterns are continuously analyzed to refine electrical design property annotations. This feedback loop improves the system's ability to handle incomplete design data by learning from actual defect patterns and adjusting classification criteria accordingly
Data Source
AI summary
A method for automatically classifying one or more defects based on electrical design properties includes receiving one or more images of a selected region of a sample, receiving one or more sets of design data associated with the selected region of the sample, locating one or more defects in the one or more images of the selected region of the sample by comparing the one or more images of the selected region of the sample to the one or more sets of design data, retrieving one or more patterns of interest from the one or more sets of design data corresponding to the one or more defects, and classifying the one or more defects in the one or more images of the selected region of the sample based on one or more annotated electrical design properties included in the one or more patterns of interest.