Process Trace Window Generation for Semiconductor Fault Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The detection of equipment faults in semiconductor manufacturing through process trace analysis is hindered by the manual and costly process of defining windows for statistical analysis, which requires extensive manual intervention and does not effectively capture rapid changes in sensor data.
Innovation Solution
An automated method for defining windows using stability and summary statistics, where trace data is aligned from both the start and end of each process step, and statistics such as median, standard deviation, and rate of change are calculated to identify optimal windows, allowing for the separation of stable and transition periods.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual window definition is used, then window quality can be controlled based on engineering knowledge, but the process requires extensive manual intervention and is costly
Solution Approach 1:
The system performs self-service by automatically defining windows using statistical analysis of the trace data itself. The algorithm identifies stable periods and transition periods autonomously without requiring manual intervention, while still producing high-quality windows through objective statistical criteria such as standard deviation and rate of change thresholds.
2Measurement precision
If manual window definition is used, then window quality can be controlled, but the process is time-consuming and reduces productivity
Solution Approach 1:
The patent replaces the manual mechanical process of window definition with an automated computational system. Statistical algorithms calculate window boundaries based on objective criteria (standard deviation, rate of change), eliminating the need for manual analysis while maintaining or improving window quality and significantly increasing processing speed.
3Extent of automation
If conventional automatic windowing algorithms are used, then automation is achieved, but extensive manual intervention is still required to produce high quality windows
Solution Approach 1:
The patent changes the parameters used for window definition from conventional approaches to statistical parameters including standard deviation, rate of change, and stability metrics. This allows fully automated window definition that produces high-quality windows by objectively identifying stable periods where parameters meet predefined thresholds, eliminating the need for manual intervention while maintaining quality.
4Reliability
If traditional FDC methods are used, then fault detection can be performed, but the quality of indicators is limited by manual window definition
Solution Approach 1:
The patent applies preliminary action by automatically defining high-quality windows before statistical indicators are calculated. This ensures that the foundation for indicator computation is optimized, leading to higher quality indicators and improved fault detection reliability. The window definition step is performed first using statistical criteria, then indicators are computed on these pre-optimized windows.
Data Source
AI summary
Automatic definition of windows for trace analysis. For each process step, the trace data are aligned to both the start of the process step and the end of the process step, and statistics including rate of change are calculated from both the start of the process step and the end of the process step. Windows are generated based on analysis of the calculated statistics.