Module and system for, and method of detecting defects in ultra-thin glass

A method of automatically inspecting ultra-fine glass for Pt-derived ultra-fine defects, comprising:

• inspecting a predetermined surface area of the ultra-thin glass for scratches, foreign matter, dents, and stains when the cassette is loaded into the automatic inspection system for poor appearance of the ultra-thin glass in units of cells;
• performing a Pt-derived ultra-fine defect inspection on the ultra-thin glass using a holographic restoration device and a scanning device that moves the holographic restoration device along a predetermined scan path;
• performing edge defect detection for detecting edge defects on the ultra-thin glass on a cell-by-cell basis or inspecting surface stains or depressions of the ultra-thin glass through phase change measurement;
• and
• inspecting the ultra-thin glass by bending or folding the ultra-thin glass in units of cells.

A method of inspecting ultra-thin glass for Pt-derived ultra-fine defects, comprising:

• bringing the ultra-thin glass into the inspection module;
• scanning a first line with respect to a dummy formed at one end of the ultra-thin glass with a holographic restoration device;
• determining a unit area (UA) that can be photographed at one time by the holographic restoration device and resolution, depth, speed, and tact time of the holographic restoration device based on a measurement value obtained through a first line scan of the dummy;
• determining and controlling a predetermined scan path of the scanning device, an objective lens magnification and a tact time of the holographic restoration device;
• determining the number of effective areas for the entire area of the ultra-thin glass according to a predetermined scanning path of the scanning device and scanning the second to Nth lines;
• restoring the shape of the entire ultra-thin glass by storing partial images of the ultra-thin glass through scanning of the second to Nth lines in the image sensor;
• determining a 2D Pt-derived defect-specific moiré pattern image on the restored 3D glass shape using a deep learning algorithm;
• and
• recognizing the nucleus of the Pt foreign material by magnifying the unique pattern of the Pt-derived defect.