Semiconductor Process Abnormality Detection With Sensor Factor Search
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current abnormality detection methods in semiconductor manufacturing struggle to promptly identify the cause of abnormalities in the process, as they rely on multivariate analysis which calculates an abnormality degree by adding changes in sensor values, making it difficult to analyze the specific abnormality factors.
Innovation Solution
An information processing apparatus that combines multivariate abnormality detection with abnormality factor search methods, using a learned abnormality detection model to infer abnormality degrees and search for univariate and correlation abnormality factors, providing comprehensive monitoring and visualization of sensor data to quickly identify abnormality causes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multivariate analysis is used to calculate abnormality degree by adding changes in sensor values, then comprehensive monitoring of the process is achieved, but it becomes difficult to analyze specific abnormality factors
Solution Approach 1:
The patent segments the abnormality analysis into two distinct parts: (1) multivariate analysis for comprehensive abnormality degree calculation, and (2) univariate analysis for specific factor identification. This segmentation allows each method to serve its optimal purpose without interference, resolving the contradiction between comprehensive monitoring and factor identifiability.
Solution Approach 2:
The patent introduces an intermediary mechanism that connects multivariate and univariate analyses. The system uses the multivariate abnormality degree as a trigger to activate univariate analysis, which then identifies specific factors. This intermediary approach allows the system to maintain both comprehensive monitoring capability and specific factor identification.
2Loss of information
If only univariate analysis of single sensor changes is performed, then specific abnormality factors can be identified, but comprehensive monitoring and early detection capability is reduced
Solution Approach 1:
The patent merges univariate analysis and multivariate analysis into a unified abnormality detection system. The univariate analysis component identifies specific factors from individual sensor changes, while the multivariate analysis component provides comprehensive monitoring by considering correlations between multiple sensors. The merging of these approaches resolves the contradiction between factor identifiability and detection accuracy.
Solution Approach 2:
The patent adds another dimension to the analysis by introducing correlation-based multivariate analysis alongside traditional univariate analysis. This dimensional expansion allows the system to simultaneously capture both individual sensor abnormalities and their interrelationships, improving early detection capability while maintaining factor identifiability through the univariate component.
3Loss of information
If multiple abnormality factor search methods are used, then comprehensive factor identification is achieved, but system complexity increases
Solution Approach 1:
The patent implements a dynamic abnormality factor search system that adapts the analysis depth and methods based on the detected abnormality degree. When the multivariate abnormality degree exceeds a threshold, the system dynamically activates univariate analysis and correlation analysis. This dynamic approach achieves comprehensive factor identification while controlling system complexity by only activating additional analysis methods when necessary.
Solution Approach 2:
The patent performs preliminary multivariate analysis to calculate the abnormality degree before activating more complex univariate and correlation analyses. This preliminary action allows the system to determine whether further detailed analysis is needed, achieving comprehensive factor identification only when abnormalities are detected, thus controlling overall system complexity.
Data Source
AI summary
An information processing apparatus includes an acquisition unit that acquires a plurality of sensor values output from a plurality of sensors installed in a semiconductor manufacturing apparatus while a process is running; an inference unit that infers an abnormality degree of the process from the acquired sensor values, using an abnormality detection model that has learned a correspondence relationship between the sensor values and the abnormality degree of the process using learning data; an abnormality detection unit that detects an abnormality occurring in the process based on the inferred abnormality degree of the process; and an abnormality factor search unit that searches for a univariate abnormality and a correlation abnormality that are candidates of abnormality factors occurring in the process, using abnormality factor search methods; and an abnormality determination result output unit that outputs the detected abnormality and the searched univariate abnormality and correlation abnormality, as an abnormality determination result.