Automated Wafer Inspection for Weak Pattern Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current wafer inspection methods rely on human eyes and are limited by small sample sizes, prone to error, and lack automation, making it difficult to detect and quantify systematic defects effectively, especially with increased data volume and sensitivity.
Innovation Solution
A system with a processor-configured wafer inspection tool that performs pattern grouping, identifies regions of interest, and automatically detects and validates weak patterns by simulating wafer processes and using SEM images, enabling automated hotspot detection and quantification without human intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If human eyes are used to determine the presence of defects, then manual classification can be performed, but the inspection process is limited to small sample sizes and is prone to error
Solution Approach 1:
The patent replaces the mechanical human visual inspection system with an automated computer-based inspection system that uses algorithms to detect and classify defects. This substitution enables processing of large datasets without the limitations of human fatigue and attention span, thereby increasing productivity while maintaining or improving measurement precision through consistent automated classification.
Solution Approach 2:
The inspection system performs self-classification of defects using automated algorithms that analyze defect patterns and characteristics. The system independently identifies weak patterns and classifies them without requiring human intervention, enabling the processing of large sample sizes while maintaining high accuracy through systematic automated decision-making.
2Quantity of substance
If sampling is used to review defects, then manual classification can be performed, but the sample size is typically limited to no more than about 5 thousand defects
Solution Approach 1:
The system performs preliminary automated classification and identification of weak patterns before any potential human review. By pre-processing the entire dataset with automated algorithms to identify suspicious patterns and weak defects, the system enables reliable analysis of large sample sizes without the sampling limitations that constrain manual inspection reliability.
Solution Approach 2:
The patent replaces manual sampling and classification with automated computer-based inspection that can process and reliably analyze tens of thousands of defects. This substitution eliminates the 5,000-defect sample size limitation by using algorithms that maintain reliability across large datasets through consistent application of classification criteria.
3Loss of information
If simulations are used to predict risky sites, then pattern prediction can be performed, but they do not help identify if the sites actually fail post processing
Solution Approach 1:
The inspection system incorporates feedback by comparing simulated predictions with actual post-processing inspection results. The system uses actual defect data from processed wafers to validate and refine predictions, creating a closed-loop system that improves both prediction accuracy and failure detection by continuously learning from actual outcomes rather than relying solely on theoretical simulations.
Data Source
AI summary
Methods and systems for providing weak pattern (or hotspot) detection and quantification are disclosed. A weak pattern detection and quantification system may include a wafer inspection tool configured to inspect a wafer and detect defects present on the wafer. The system may also include at least one processor in communication with the wafer inspection tool. The at least one processor may be configured to: perform pattern grouping on the detected defects based on design of the wafer; identify regions of interest based on the pattern grouping; identify weak patterns contained in the identified regions of interest, the weak patterns being patterns deviating from the design by an amount greater than a threshold; validate the weak patterns identified; and report the validated weak patterns or facilitate revision of the design of the wafer based on the validated weak patterns.