AI image enhancement for failure analysis in 3D quantum information technology

Area of Science:

• Advanced semiconductor manufacturing and failure analysis.
• Quantum computing hardware development and defect characterization.

Background:

• Increasing use of 3D integration and miniaturization in integrated circuits and quantum computing.
• Growing need for efficient detection of micro-scale defects on wafers.

Purpose of the Study:

• To develop a time-efficient and accurate method for measuring, localizing, and classifying micro-scale defects.
• To enhance image quality using machine learning super-resolution for self-supervised data enhancement.

Main Methods:

• Integration of scanning acoustic microscopy (SAM) with You Only Look Once (YOLO) object detection and semantic segmentation.
• Application of machine learning super-resolution (ML-SR) for image quality enhancement.
• Testing various ML-SR approaches for self-supervised quality improvement.

Main Results:

• Achieved time-efficiency enhancement of approximately 4x for through-silicon via (TSV) analysis and 6x for delamination analysis.
• Demonstrated accurate defect detection, localization, and statistical classification down to the micrometer regime.
• Validated the effectiveness of ML-SR in improving the quality of SAM image data.

Conclusions:

• The developed AI-powered workflow significantly accelerates defect analysis in semiconductor manufacturing.
• The approach is versatile and applicable to various fields requiring rapid failure analysis beyond SAM data.
• This method supports the advancement of miniaturized electronics and quantum computing technologies through improved defect detection.