Systems and methods for air pocket defect detection

The computer system analyzes images of silicon ingots using multiple models to detect air pockets, addressing inefficiencies in current methods and enhancing the detection of air pockets, which improves the structural integrity and reliability of silicon wafers.

WO2025006413A1PCT designated stage expired Publication Date: 2025-01-02GLOBALWAFERS CO LTD
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
PCT/US2024/035326
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2023-06-26
Filing Date
2024-06-25
Publication Date
2025-01-02

AI Technical Summary

Technical Problem

Current methods for detecting air pockets in silicon ingots during the Czochralski process are inefficient, leading to potential structural integrity issues and device failures, as they fail to accurately identify air pockets early enough in the manufacturing process.

Method used

A computer system utilizing a processor and memory device executes multiple pre-trained models to analyze images of the ingot, combining predictions to determine the presence of air pockets, thereby improving detection accuracy and preventing further processing of defective materials.

Benefits of technology

This approach enhances the detection of air pockets, reducing material loss and improving wafer quality by accurately identifying defects before they cause failures, thus ensuring higher product reliability.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure US2024035326_02012025_PF_FP_ABST
    Figure US2024035326_02012025_PF_FP_ABST
Patent Text Reader

Abstract

A computer device (108; 802) includes at least one processor (805) in communication with at least one memory device (810). The at least one processor (805) is programmed to: receive (505) at least one image (118, 128) of a material to be analysed (104); execute (510) a plurality of models (410, 420, 430) trained to classify the at least one image (118, 128) to detect a first defect type; receive (515) from each of the plurality of models (410, 420, 430) a prediction (415, 425, 435) that the at least one image (118, 128) includes the first defect type; combine (520) the plurality of predictions (415, 425, 435) to calculate a final prediction (445) of whether or not the at least one image (118, 128) includes the first defect type; and reject or approve (525) the material to be analysed (104) based upon the final prediction (445).
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Wafer defect detection method based on WaferDet network

    CN111696077A

  • Method and System for Iterative Defect Classification

    US20160358041A1

  • Symmetry based air pocket detection methods and systems

    US9317912B2