Failure Rate Abnormality Detection for Low-Defect Semiconductor Processes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting abnormalities in semiconductor manufacturing struggle to accurately manage low failure rates due to underestimation issues, as they do not account for variability in quality information updates, sample size, and changing trends, leading to unnecessary failure rate control and reduced reliability.
Innovation Solution
A method that calculates a detection failure rate using an exact test method and weighting based on differences between failure data and target failure data, allowing for preemptive management of failure rates by defining state areas and adjusting manufacturing process parameters to reduce defective products.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional failure rate management methods are used, then the system can operate with simple monitoring, but the failure rate detection accuracy deteriorates due to underestimation in low failure rate fields
Solution Approach 1:
The patent changes the detection parameters by introducing detection failure rate (dfar) and abnormality index (AI) calculations that account for variability in quality information updates, sample sizes, and changing trends. This transforms the traditional static failure rate monitoring into a dynamic parameter-based detection system that accurately identifies abnormalities even in low failure rate fields.
Solution Approach 2:
The patent performs preliminary calculations of reference failure rate, detection failure rate, and abnormality index before making abnormality determinations. By pre-calculating control limits and weighting factors based on historical data, the system prepares detection thresholds in advance, enabling accurate real-time abnormality detection without reactive adjustments.
2Reliability
If traditional intuitive abnormality detection is used, then the system is easy to operate, but the response timing deteriorates due to inability to detect abnormalities in advance
Solution Approach 1:
The patent implements a feedback mechanism where the abnormality index calculation incorporates historical failure data, reference failure rates, and control limits. The system continuously compares current failure data against these feedback-derived thresholds, enabling early abnormality detection and timely responses before failures exceed target levels.
Solution Approach 2:
The system performs preliminary establishment of control limits and reference failure rates based on historical data before actual abnormality detection occurs. This pre-preparation of detection criteria enables the system to immediately identify abnormalities when they occur, eliminating response delays associated with establishing thresholds reactively.
3Measurement precision
If detection sensitivity is increased to detect low failure rates, then the abnormality detection capability improves, but false positive rate increases leading to unnecessary control actions
Solution Approach 1:
The patent introduces multiple layered parameters including detection failure rate (dfar), abnormality index (AI), and control limits to transform simple binary detection into a multi-stage evaluation process. This parameter transformation enables the system to distinguish between normal variability and true abnormalities, reducing false positives while maintaining high detection sensitivity for actual failures.
Solution Approach 2:
The abnormality index serves as an intermediary metric between raw failure data and abnormality determination. By introducing this intermediate calculation that incorporates weighting factors and control limits, the system filters out noise and false signals while preserving true abnormality indicators, thereby reducing false positive rates.
Data Source
AI summary
A method of detecting abnormalities includes: calculating a reference failure rate using failure data at a plurality of points in time included in a particular period; calculating a detection failure rate and weighting, corresponding to failure data at a detection time point after the particular period, using the reference failure rate; calculating an abnormality index based on multiplying the detection failure rate by the weighting; comparing the abnormality index with an index corresponding to a control limit for stably controlling a failure rate; and detecting whether the failure data at the detection time point is abnormal, based on a result of the comparison of the abnormality index with the index corresponding to the control limit.