Wafer Defect Prediction via Multi-Source Data Correlation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current silicon wafer manufacturing processes face inefficiencies due to defects such as misalignment, which require costly corrective actions like stripping and re-coating, and lack a systematic approach to predict and address defects effectively.
Innovation Solution
A method that aggregates and analyzes data from manufacturing tools to predict defect occurrences and likelihoods in critical areas, providing predictive information to the manufacturing system for proactive corrective actions, using stochastic-based methods and correlations between defects and tool parameters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional manufacturing processes are used without predictive analysis, then the manufacturing process is simple to operate, but defect detection and corrective action are delayed, leading to lower yield and higher costs
Solution Approach 1:
The system performs preliminary actions by collecting and analyzing manufacturing data before defects actually occur. Statistical models and machine learning algorithms predict potential defects by examining trends in alignment data, defect data, and process parameters, enabling preventive corrective actions to be taken before yield is impacted.
Solution Approach 2:
The system implements continuous feedback loops where manufacturing data from multiple sources (alignment tools, defect inspection equipment, process control systems) is constantly collected, analyzed, and fed back to adjust process parameters. This closed-loop feedback mechanism enables real-time optimization of manufacturing parameters to prevent defects and improve yield.
2Measurement precision
If comprehensive data collection and analysis systems are implemented, then defect prediction accuracy improves, but the complexity of the manufacturing system increases
Solution Approach 1:
The data management system is designed as a universal multi-functional platform that handles multiple types of manufacturing data (alignment, defect, process parameters, tool performance) through a single integrated architecture. The system performs multiple functions including data collection, storage, analysis, prediction, and corrective action recommendation, eliminating the need for separate specialized systems for each function.
Solution Approach 2:
The system introduces intermediary components including data normalization layers, standardized communication protocols, and intermediate storage structures that bridge different data sources and analysis algorithms. These intermediaries simplify the integration of diverse manufacturing data sources and enable seamless data flow between collection, analysis, and action stages without requiring complex point-to-point connections.
3Reliability
If predictive analysis is performed on all manufacturing data, then defect detection capability improves, but processing time and computational resources increase
Solution Approach 1:
The system applies local quality by focusing computational analysis resources on critical areas and parameters rather than uniformly analyzing all data. Statistical models identify and prioritize analysis of high-risk process parameters, critical manufacturing steps, and areas with historical defect patterns, allocating more computational power to these localized critical points while reducing analysis intensity in low-risk areas.
Solution Approach 2:
The system implements partial action by performing comprehensive predictive analysis only when triggered by specific conditions such as detected anomalies, historical defect patterns, or critical process deviations. For routine manufacturing conditions, the system uses streamlined analysis protocols that process only essential parameters, reducing computational overhead while maintaining adequate defect detection capability for normal operating conditions.
Data Source
AI summary
Implementations described herein generally relate to improving silicon wafer manufacturing. In one implementation, a method includes receiving information describing a defect. The method further includes identifying a critical area of a silicon wafer and determining the probability of the defect occurring in the critical area. The method further includes determining, based on the probability, the likelihood of an open or a short occurring as a result of the defect occurring in the critical area. The method further includes providing, based on the likelihood, predictive information to a manufacturing system. In some embodiments, corrective action may be taken based on the predictive information in order to improve silicon wafer manufacturing.