Unbiased Wafer Defect Sampling via Attribute-Based Binning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for generating defect samples on wafers are inadequate in achieving maximum defect type diversity, particularly in noisy inspections, and lack flexibility and adaptability, often missing defect types and requiring prior knowledge and manual tuning.
Innovation Solution
A computer-implemented method that identifies common defects across multiple scans, selects sampling parameters independently for each scan, and creates a diverse defect sample by choosing defects with the most diversity in user-defined attributes, allowing for unbiased or biased sampling without relying on pre-constructed classifiers or binning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If random sampling is used to select defects, then the sampling process is simple and fast, but defect type diversity is insufficient and many defect types are missed
Solution Approach 1:
The patent segments the defect population into multiple bins based on defect attributes (such as size, shape, location) before sampling. This segmentation allows the sampling algorithm to select defects from different bins, ensuring diverse defect type representation while maintaining efficient sampling speed through automated bin-based classification.
Solution Approach 2:
The patent implements dynamic sampling where the selection criteria can adapt based on user-defined attributes and weighting factors. The sampling algorithm dynamically adjusts which defects are selected based on the specified attributes, allowing flexibility in achieving diverse defect type representation without manual intervention.
2Adaptability or versatility
If iDO binning is used for diversification, then defect classification is achieved, but prior knowledge and manual tuning are required which increases complexity
Solution Approach 1:
The patent enables the sampling system to automatically classify defects into bins based on user-defined attributes without requiring pre-constructed classifiers or manual tuning. The system self-adapts to the defect data by using the specified attributes directly for binning, eliminating the need for complex classifier training and tuning processes.
Solution Approach 2:
The patent allows users to define binning parameters and weighting factors dynamically based on their specific needs. Instead of requiring fixed, pre-tuned classifiers, the system accepts user-defined parameter settings that can be adjusted according to the particular inspection requirements, simplifying the overall system configuration.
3Measurement precision
If manual tuning of sampling parameters is required, then sampling accuracy can be optimized, but time to result increases and productivity decreases
Solution Approach 1:
The patent performs preliminary binning of defects into categories based on user-defined attributes before the sampling process begins. This preliminary classification is done automatically without requiring iterative manual tuning, allowing the sampling algorithm to immediately generate accurate results based on pre-computed bins, thus reducing time to result while maintaining sampling accuracy.
Solution Approach 2:
The patent incorporates feedback mechanisms where the sampling algorithm uses the specified attributes and weighting factors to automatically adjust the sampling process. The system feedback-loop allows it to optimize sampling accuracy by adjusting selection criteria based on the input parameters without requiring manual intervention, thereby reducing time to result while maintaining precision.
4Adaptability or versatility
If feature vector space diversification is used, then some defect diversity is achieved, but the large dimensional space with correlated features makes diversification inefficient
Solution Approach 1:
The patent extracts and focuses on specific user-defined attributes that are most relevant for defect diversification, rather than using the entire high-dimensional feature vector space. By selecting and working with only the necessary attributes (such as size, shape, location), the system achieves efficient diversification without being overwhelmed by the complexity of correlated features in the full feature space.
Solution Approach 2:
The patent transforms the diversification problem from the original high-dimensional feature vector space into a lower-dimensional attribute-based binning structure. By reorganizing defects into bins based on user-defined attributes, the system reduces the effective dimensionality while maintaining diversification capability, making the process more efficient and manageable.
Data Source
AI summary
Methods and systems for generating unbiased wafer defect samples are provided. One method includes selecting the defects detected by each of multiple scans performed on a wafer that have the most diversity in one or more defect attributes such that a diverse set of defects are selected across each scan. In addition, the method may include selecting the defects such that any defect that is selected and is common to two or more of the scans is not selected twice and any defects that are selected are diverse with respect to the common, selected defect. Furthermore, no sampling, binning, or classifying of the defects may be performed prior to selection of the defects such that the sampled defects are unbiased by any sampling, binning, or classifying method.