Sensor Anomaly Detection via Correlation Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing anomaly detection methods in semiconductor processing are recipe-dependent, costly, and time-consuming, leading to inefficiencies and high false positive rates, especially when scaling to larger datasets.
Innovation Solution
A method involving z-normalization and filtering of sensor data from paired sensors to identify anomalies, using a reference correlation and low-frequency band path filtering to reduce noise and false positives, allowing for efficient adjustment of process parameters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If recipe-dependent anomaly detection methods are used, then anomaly detection can be performed for specific processes, but the complexity increases and scalability decreases when dealing with different tools and processes
Solution Approach 1:
The patent applies universality by developing a generic anomaly detection framework that works across different semiconductor processing tools and recipes. Instead of creating separate detection methods for each process, the system uses a unified approach based on sensor trace analysis that can be applied universally. The framework processes sensor data from multiple chambers and sensors using consistent algorithms, making the system multi-functional and adaptable to various processing scenarios without requiring recipe-specific customization.
2Reliability
If multiple different anomaly detection techniques are developed for different tools and processes, then detection accuracy improves for each specific case, but development cost and time increase
Solution Approach 1:
The patent implements preliminary action by pre-processing sensor trace data to extract meaningful features and patterns before anomaly detection. The system performs preliminary analysis on historical sensor data to establish baseline behaviors and correlations between sensors. This pre-processing step creates a standardized foundation that accelerates subsequent anomaly detection across different processes, eliminating the need to start from scratch for each new tool or recipe.
3Reliability
If comprehensive sensor data analysis is performed to improve anomaly detection, then detection capability improves, but false positive rate increases
Solution Approach 1:
The patent applies feedback by continuously monitoring sensor traces and comparing current readings against historical patterns and correlations. The system uses feedback loops to adjust detection thresholds and parameters based on observed data patterns. When anomalies are detected, the system can validate findings by checking multiple sensor correlations and historical contexts, reducing false positives. The feedback mechanism allows the system to learn from false alarms and improve its discrimination between true anomalies and normal variations.
Data Source
AI summary
A method for detecting an anomaly in sensor data generated in a substrate processing apparatus is disclosed herein. A plurality of data sets is received. A first data set from a first sensor and second data set from a second sensor are selected. The first second sensors are defined as a sensor pair. A reference correlation is generated by selecting a subset of values in each data set for each of the first and second data sets. A difference of remaining data correlation outside the subset of values in each data set to the reference correlation is normalized. The normalized data set is filtered to smooth the normalized difference to avoid isolated outliers with high chance of false positive candidates. One or more anomalies are identified. Process parameters of the substrate processing apparatus are adjusted, based on the one or more identified anomalies from the filtered data set.