Kruskal-Wallis Test for Defect Source Identification in IC Manufacturing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current defect detection techniques in integrated circuit manufacturing are inadequate for identifying sources of defects, particularly due to assumptions of normal distribution and the inability to detect latent defects, leading to inefficiencies and increased costs in semiconductor fabrication.
Innovation Solution
The use of non-parametric statistical tools, specifically the Kruskal-Wallis test, to analyze defect characteristics and identify likely sources by processing steps and tools, enabling the detection of abnormal distributions and latent defects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If non-parametric statistical tools (Kruskal-Wallis test) are used to analyze defect characteristics, then defect detection accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent transforms the defect analysis approach by changing the statistical parameters from parametric assumptions (normal distribution) to non-parametric methods (Kruskal-Wallis test). This parameter change enables accurate detection of latent defects and abnormal distributions without requiring defects to follow normal distribution patterns, thereby improving defect detection accuracy while managing computational complexity through established statistical algorithms.
2Productivity
If automated defect detection system is implemented, then productivity is improved, but system complexity increases
Solution Approach 1:
The patent implements an automated defect detection system that performs self-service by automatically collecting defect data from multiple sources, applying the Kruskal-Wallis test algorithm, and generating analysis results without requiring manual intervention. The system autonomously identifies latent defects and abnormal distributions, improving productivity while containing system complexity through modular architecture that integrates data collection, statistical analysis, and result generation in an automated workflow.
3Measurement precision
If multiple processing steps and tools are monitored, then measurement precision is improved, but data processing complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the complex data processing task into distinct modules: defect data collection from multiple processing steps and tools, Kruskal-Wallis test application for statistical analysis, and result interpretation for identifying latent defects and abnormal distributions. This segmentation allows precise monitoring of multiple processing steps and tools while managing data processing complexity through structured, modular analysis that processes data from each source systematically.
Data Source
AI summary
Method and system for defect detection in manufacturing integrated circuits. In an embodiment, the invention provides a method for identifying one or more sources for possible causing manufacturing detects in integrated circuits. The method includes a step for providing a plurality of semiconductor substrates. The method includes a step for processing the plurality of semiconductor substrates in a plurality of processing steps using a plurality of processing tools. The method additionally includes a step for providing a database, which includes data associated with the processing of the plurality of semiconductor substrates. The method further includes a step for testing the plurality of semiconductor wafers after the processing of the plurality of semiconductor substrates. Additionally, the method includes a step for detecting at least one defect characteristic associated with the plurality of the semiconductor substrates that have been processed. Moreover, the method includes a step for identifying a set of processing steps. For example, the set of processing step are possibly associated with the defect characteristic.