GAN-Based Super-Resolution in Linear R-SAM Imaging for Enhanced Non-Destructive Semiconductor Measurement

This paper proposes LinearTGAN, a GAN-based super-resolution model specifically designed to improve the resolution of linear acoustic wafer images acquired by the rotary scanning acoustic microscopy (R-SAM) system for non-destructive semiconductor measurement.

Paper

Abstract

The precise identification and non-destructive measurement of structural features and defects in semiconductor wafers are essential for process integrity and high yield in advanced manufacturing. Scanning acoustic microscopy (SAM) provides detailed visualizations of both surface and internal wafer structures, but scanning time and image quality significantly impact its performance. This study proposes LinearTGAN, an improved GAN-based model designed to improve the resolution of linear acoustic wafer images from the breakthrough rotary scanning acoustic microscopy (R-SAM) system. Empirical evaluations demonstrate that the proposed model significantly outperforms conventional GAN-based approaches in Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index Measure (SSIM) metrics.

Key Contributions

• LinearTGAN: improved GAN-based SR model for linear R-SAM acoustic images
• Addresses production bottlenecks caused by prolonged scanning durations
• Enables high-quality semiconductor defect detection with reduced scanning time
• Validated on real-world industrial SAM data from the R-SAM system

Recommended citation: T. T. H. Vu, T. H. Vo, T. N. Nguyen, J. Choi, L. H. Tran, V. H. M. Doan, V. B. Nguyen, W. Lee, S. Mondal and J. Oh, "GAN-Based Super-Resolution in Linear R-SAM Imaging for Enhanced Non-Destructive Semiconductor Measurement," *Applied Sciences*, vol. 15, no. 12, p. 6780, June 2025, doi: 10.3390/app15126780.