High-Fidelity Synthetic Transmission Electron Microscopy Image Generation Using Diffusion Probabilistic Models for Data-Limited Semiconductor Metrology
2026-06-23 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors address challenges in making detailed microscope images of advanced semiconductors, which are hard to get because the process is slow, costly, and damages the samples. They developed a new AI method that can create realistic synthetic microscope images using very few real examples—just 15. Their approach carefully builds images piece by piece, keeping important structural details needed for accurate analysis. Additionally, they use features from their AI model to help segment these images, which aids in identifying different regions. Their synthetic images are shown to be highly similar to real ones, supporting further use in defect detection and measurement tasks.
Transmission Electron Microscopy (TEM)Denoising Diffusion Probabilistic Model (DDPM)Synthetic data generationPatch-based trainingTrivialAugmentClassifier guidanceImage segmentationMS-SSIMMetrologyData scarcity
Authors
Johannes Boehm, Bappaditya Dey
Abstract
Advanced semiconductor nodes drastically increased demand for Transmission Electron Microscopy (TEM), yet destructive sample preparation, slow imaging and high costs severely limit the availability of diverse datasets needed for downstream machine learning (ML). Synthetic data generation is becoming essential, but current generative models often miss TEM-specific noise, structural detail, and stochastic variability crucial for evaluation. We present a Denoising Diffusion Probabilistic Model (DDPM) framework for synthetic TEM image generation under extreme data scarcity. A progressive patch-based training strategy scales from low-resolution patches to full images, enabling from-scratch training with only 15 samples. We integrate a custom TrivialAugment adaptation, cross-process domain transfer, classifier guidance, and RePaint-style inpainting, culminating in full-image generation that preserves global structural and spatial relationships in compliance with FAB metrology requirements. Beyond synthesis, we repurpose DDPM feature representations for segmentation, partitioning encoder feature maps to obtain coherent region masks. Our synthetic images achieve up to MS-SSIM > 0.98 and qualitative expert assessment consistent with structural similarity results, facilitating downstream ML training for defect detection, segmentation, and metrology while preserving statistical and physical realism.