Method for positioning defects of epitaxial wafer based on machine vision

By collecting control parameters during the epitaxial wafer growth process, a defect size prediction network is constructed. Machine vision is used to identify dislocations and cracks, predict the propagation direction, and process the influence coefficient. This solves the problem of insufficient efficiency and accuracy in epitaxial wafer defect detection in existing technologies, and achieves efficient and accurate defect localization.

CN120655618BActive Publication Date: 2026-05-01ZHEJIANG LISHUI XIN WAFER SEMICON TECH CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
ZHEJIANG LISHUI XIN WAFER SEMICON TECH CO LTD
Filing Date
2025-06-13
Publication Date
2026-05-01

AI Technical Summary

Technical Problem

Existing machine vision-based epitaxial wafer defect detection methods struggle to comprehensively consider control parameters, defect propagation patterns, and mutual influences during epitaxial wafer growth, resulting in insufficient detection efficiency and accuracy. In particular, when dislocations and cracks coexist, it is difficult to accurately assess the actual scale of cracks affected by dislocations.

Method used

The sequence of control parameters during the growth of epitaxial wafers is collected. A network for predicting the scale of dislocations and cracks is constructed using machine learning technology. The network predicts the scale parameters of dislocations and cracks, configures the microscopic image acquisition area, uses machine vision to identify dislocations and cracks, predicts their propagation direction when a dislocation is identified, processes the crack influence coefficient, and so on until the defect identification and location are completed.

Benefits of technology

It enables precise location of defects in epitaxial wafers, significantly improves detection efficiency and accuracy, reduces interference from human factors, and ensures the scientific nature of quality assessment and the reliability of the production process.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120655618B_ABST
    Figure CN120655618B_ABST
Patent Text Reader

Abstract

The present application relates to a machine vision-based epitaxial wafer defect positioning method, and relates to the field of defect positioning. The method predicts dislocation and crack scale parameters by collecting control parameter sequences in the epitaxial wafer growth process, then configures the number of microscopic image collection areas, frames the recognition areas on the epitaxial wafer and collects microscopic images, uses machine vision to identify dislocations and cracks, predicts the propagation direction of the dislocations when the dislocations are identified, simultaneously processes the crack influence coefficient when the dislocations and cracks coexist to correct the number of crack areas, and finally continues to identify and correct according to the dislocation propagation direction until the preset number of dislocation and crack areas is reached or the identification area is traversed, thereby solving the technical problem that it is difficult to comprehensively consider the control parameters in the epitaxial wafer growth process, the defect propagation law and the mutual influence, resulting in insufficient defect detection efficiency and accuracy, realizing accurate positioning of epitaxial wafer defects, and significantly improving the efficiency and accuracy of epitaxial wafer defect detection.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Metal material surface microcrack positioning method

    CN119413899A

  • System and method of analyzing a crystal defect

    US20200020506A1