SiC Wafer Defect Characterization Using Photoluminescence AI
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for defect characterization in silicon carbide (SiC) wafers are destructive, expensive, and limited, making it difficult to accurately count dislocations and other defects, which hinders the improvement of crystal growth processes and increases the cost of wafer characterization.
Innovation Solution
The development of nondestructive defect characterization methods using deep convolutional neural networks (DCNNs) that analyze photoluminescence microscopy data to detect and identify defects in SiC wafers, allowing for the reuse of characterized wafers and increased sampling without significant cost increases.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If destructive etching is used to characterize defects in SiC wafers, then defect detection accuracy is improved, but wafer usability deteriorates and characterization cost increases
Solution Approach 1:
The patent replaces the mechanical/chemical etching process with optical imaging techniques. Specifically, it uses optical microscopes equipped with specialized filters and illumination sources to directly image defects on the wafer surface without chemical treatment. This substitution eliminates the need for destructive etching while maintaining defect detection capability, allowing wafers to be reused for device fabrication after characterization.
Solution Approach 2:
The patent creates optical copies (images) of the defect structures on the wafer surface through nondestructive optical imaging. By capturing optical images with specific filters that highlight defect features, the system produces a visual replica of the defect landscape without physically altering the wafer. This allows multiple wafers to be characterized and all characterized wafers to remain usable, effectively copying the information rather than consuming the substrate.
2Measurement precision
If destructive etching is used for defect characterization, then defect identification is improved, but processing complexity and cost increase
Solution Approach 1:
The patent replaces the complex multi-step etching process with a simpler optical imaging system. Instead of requiring chemical preparation, etching baths, safety protocols, and waste disposal procedures, the system uses an optical microscope with filtered illumination and detection. This substitution dramatically reduces process complexity while maintaining or improving defect identification accuracy through automated image analysis.
Solution Approach 2:
The patent enables the wafer to reveal its defect structure directly through its inherent optical properties without requiring external chemical treatment. By using appropriate illumination wavelengths and filter combinations, the defect features naturally manifest in the optical images, eliminating the need for etching chemicals and complex processing equipment. The wafer essentially serves itself by providing defect information through its native optical characteristics.
3Loss of substance
If only a few wafers per crystal are sampled for characterization, then characterization cost is reduced, but information availability for process improvement deteriorates
Solution Approach 1:
The patent creates accurate optical copies of defect patterns across entire wafer surfaces, enabling comprehensive characterization without consuming the wafers. Because the optical imaging process is nondestructive, information can be extracted from every wafer in a crystal batch rather than just a few sampled wafers. This multiplying effect of information collection—characterizing all wafers instead of sampling—provides much richer data for process improvement while the wafers remain valuable for device fabrication.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enables fast, accurate, and cost-effective nondestructive defect characterization, reducing the expense of wafer characterization and enhancing feedback loops between growth process development and production by allowing SiC wafers to be used for device fabrication after characterization.
Implementation Method 1
capturing the nondestructive data comprises photoluminescence microscopy
Data Source
AI summary
Nondestructive characterization of crystalline wafers is provided, including defect detection, identification, and counting. Certain aspects relate to development of nondestructive, high fidelity defect characterization and/or dislocation counting methods based on deep neural networks. Certain aspects relate to nondestructive methods for defect characterization of silicon carbide (SiC) wafers. By subjecting SiC wafers to nondestructive defect characterization, SiC wafers in their final state may be characterized and subsequently used for device fabrication, vastly reducing the expense of the characterization process. Nondestructive defect characterization also allows for increased sampling and improved feedback loops between crystalline growth process development and subsequent device production.