Wafer Defect Heat Mapping from Photoluminescence Signals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing wafer defect detection methods using photoluminescence and electroluminescence face issues such as low accuracy, potential damage to wafers, and slow detection speeds, with limited correlation between the two techniques.
Innovation Solution
A defect detection method that utilizes a photoluminescence detection result to generate a defect heat map using a preset heat map model constructed from electroluminescence-marked photoluminescence samples, allowing for accurate defect detection without electroluminescence scanning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If photoluminescence detection is used alone, then detection speed is improved, but detection accuracy deteriorates
Solution Approach 1:
The patent uses a heat map model trained on electroluminescence data to generate a virtual electroluminescence detection result based on photoluminescence input. This copying approach allows the system to leverage the high accuracy of electroluminescence detection without performing actual electroluminescence scanning, thus maintaining detection speed while improving accuracy
Solution Approach 2:
The heat map model serves as an intermediary that translates photoluminescence detection results into predicted electroluminescence defect locations. This intermediary mechanism bridges the gap between the two detection methods, allowing the system to use the advantages of both methods without combining their physical processes
2Measurement precision
If electroluminescence detection is used, then detection accuracy is improved, but wafer damage increases
Solution Approach 1:
Instead of performing actual electroluminescence detection that may damage the wafer, the system copies the essential information by using a heat map model trained on electroluminescence data. The model generates predicted defect locations based on photoluminescence input, providing accurate defect detection without the harmful physical process
Solution Approach 2:
The patent extracts the critical defect detection capability from electroluminescence detection by training a heat map model on electroluminescence data. Once trained, the model can predict defects using only photoluminescence input, separating the learning phase (where electroluminescence data is used) from the detection phase (where only photoluminescence is needed, avoiding wafer damage)
3Productivity
If traditional photoluminescence analysis is used, then detection speed is maintained, but detection accuracy deteriorates due to ignoring image connections
Solution Approach 1:
The patent merges the fast processing capability of photoluminescence detection with the high accuracy of electroluminescence detection by combining them through a heat map model. The model integrates information from photoluminescence images while incorporating the defect patterns learned from electroluminescence data, achieving both speed and accuracy
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
Improves detection efficiency and accuracy by generating defect heat maps based on photoluminescence results, eliminating the need for electroluminescence scanning and reducing potential wafer damage.
Implementation Method 1
Wafer defect detection is usually performed using a variety of different technical means, such as photoluminescence detection and electroluminescence detection
Implementation Method 2
generating, based on the photoluminescence detection result and a preset heat map model, a defect heat map corresponding to the wafer to be detected, the preset heat map model being constructed based on a photoluminescence detection result sample marked by an electroluminescence defect
Data Source
AI summary
A defect detection method includes: acquiring a photoluminescence detection result of a wafer to be detected; generating a defect heat map corresponding to said wafer according to the photoluminescence detection result and a preset heat map model, the preset heat map model being constructed on the basis of a photoluminescence detection result sample after electroluminescent defect marking; and determining a defect detection result of said wafer according to the defect heat map.


