Waveform Abnormality Detection Using Cumulative Temperature Differences
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing abnormality detection systems in semiconductor wafer processing struggle to accurately distinguish between normally and abnormally processed wafers, leading to false detections and interruptions in manufacturing, as the temperature integrated area method often fails to differentiate between normal and abnormal data when the differences are minimal.
Innovation Solution
An abnormality detection system that includes a waveform acquisition unit, an abnormality degree calculation unit, and an abnormality determination unit, which calculates an abnormality degree based on the cumulative value of absolute differences in data values over time and determines whether the waveform data is normal or abnormal, utilizing the score value obtained from temperature data to differentiate between normal and abnormal processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the temperature integrated area method is used to detect abnormalities, then the detection system can identify processing issues, but it produces false detections when the differences between normal and abnormal data are minimal
Solution Approach 1:
The patent changes the detection parameter from temperature integrated area to cumulative absolute temperature differences. This parameter transformation enhances the system's ability to distinguish between normal and abnormal processes by accumulating deviations over time, making subtle differences more detectable and reducing false positives.
Solution Approach 2:
The patent introduces a time-dimensioned cumulative metric instead of using a single integrated area value. By summing absolute differences across multiple time points, the system adds temporal accumulation as a new dimension, which amplifies subtle abnormalities and improves discrimination precision.
2Measurement precision
If the cumulative value of absolute differences is calculated for each unit time, then the abnormality detection precision is improved, but the computational complexity increases
Solution Approach 1:
The patent divides the temperature data into discrete unit time segments and calculates absolute differences for each segment independently. This segmentation approach simplifies the overall calculation by breaking it down into manageable, repetitive operations that can be processed sequentially, reducing computational complexity while maintaining precision.
Solution Approach 2:
The patent replaces complex pattern recognition or statistical analysis mechanisms with a simpler arithmetic accumulation method. By using basic operations (absolute value calculation and summation) instead of sophisticated algorithms, the system achieves high precision with minimal computational complexity.
3Loss of time
If abnormality detection is performed continuously during processing, then defects can be detected early, but false detections may cause interruptions in manufacturing
Solution Approach 1:
The patent implements continuous feedback through real-time calculation of cumulative absolute differences during wafer processing. This allows early defect detection while maintaining manufacturing continuity, as the system provides ongoing information about process health without requiring interruptions for inspection.
Solution Approach 2:
The patent performs preliminary abnormality detection during the processing itself rather than after completion. By calculating cumulative differences in real-time, the system identifies defects early in the process, allowing for timely intervention without causing unnecessary interruptions, thus preserving productivity.
Data Source
AI summary
An abnormality detection system includes a waveform acquisition unit configured to acquire waveform data that changes over time; an abnormality degree calculation unit, and an abnormality determination unit. The abnormality degree calculation unit is configured to calculate an abnormality degree based on a cumulative value obtained by accumulating absolute values of differences in data values for each unit time from the waveform data acquired by the waveform acquisition unit. The abnormality determination unit is configured to determine whether the waveform data is normal or abnormal based on the abnormality degree calculated by the abnormality degree calculation unit.