An Adaptive Denoising System for Sub-nm Scale Failure Analysis Based on TEM Image

摘要

TEM (Transmission Electron Microscopy) image-based failure analysis is one of the key Physical Failure Analysis (PFA) methods in DRAM (Dynamic Random Access Memory) manufacturing. Noise caused by surface damage or front material will corrupted the small features of defect which increases the difficulty for analyzing it. The system described in this paper is designed to enhance small features and reduce background noise for TEM images. The system is formed by 2-Dimensional Fourier Transformation (2D FT) with orientation and other features detection, Gabor filter and Wiener filter.