Enhancing Semiconductor Manufacturing: Automated Optical Systems for Wafer Defect Detection

Abstract

Purpose: The purpose of this study is to design and evaluate a low-cost, automated optical inspection (AOI) system for detecting defects in silicon semiconductor wafers using a repurposed Ender-3 3D printer and microscope camera.

Hypothesis: I hypothesized that a consumer-grade, programmable platform can be modified to produce consistent, high-resolution wafer images that support effective manual defect inspection, offering a cost-effective alternative to traditional AOI systems.

Method: The system was constructed by replacing the Ender-3 printer’s print head with a microscope camera mounted on a custom CAD-designed bracket. G-code programming enabled automated movement, while Python libraries including ToupCam and OpenCV facilitated image capture, stitching, and autofocus. High-resolution images were captured at defined positions across the wafer and stitched into mosaics for visual analysis. Test wafers containing surface-level defects such as contamination were used to validate system performance.

Results: The system consistently captured sharp, high-resolution images of silicon wafers and successfully stitched them into coherent mosaics. Dynamic autofocus allowed for image clarity across variable surface heights, and G-code automation ensured repeatable imaging. While subsurface and low-contrast defects remained difficult to detect, the system effectively identified visible irregularities like surface contamination.

Conclusion: This study demonstrates that a low-cost AOI system can fulfill essential imaging tasks for wafer inspection. Though not yet suitable for industrial use, the setup lowers the barrier to defect detection in research and educational settings. Future iterations may integrate machine learning to enhance functionality, supporting broader adoption in resource-limited environments.