掃描式電子顯微鏡（SEM）由於其高解析度，被廣泛用於觀察微觀結構。為了分析 SEM 影像中的細節結構，我們可以使用深度學習神經網絡進行精確的估測。要訓練這些神經網絡，我們需要大量的訓練資料集，包括 SEM 影像以及其對應的微觀結構。然而，SEM 成像成本高、程序複雜，且受樣本條件限制，這些因素阻礙了神經網絡在大規模缺陷檢測上的應用。為了克服這些挑戰，我們使用 Nebula 模擬 SEM 影像，以生成合成的 SEM 圖像及其對應的深度圖，從而構建一個用於缺陷檢測的訓練資料集。在我們的方法中，Blender 被用來建立 3D 模型並生成三角網格，這些網格以 *.tri 檔的形式表示，包含了模型的幾何資訊與深度資訊。
這些網格接著被輸入至 Nebula 蒙地卡羅模擬器，以產生 SEM 影像，從 Blender 中提取的深度圖則作為監督式學習的真實標籤（ground truth）。我們在這些合成影像與深度圖上訓練一個以 ResNet 為骨幹的 CNN 模型，以擷取深度變化的特徵並自動識別潛在的缺陷區域。由於表面缺陷通常呈現明顯的深度異常，我們的方法提升了缺陷檢測的準確率與效率。透過將 SEM 模擬與深度學習相結合，我們的方法減少對真實 SEM 影像的依賴，並為晶圓檢測及其他高解析度缺陷分析應用於半導體製造領域提供了一個具成本效益且高效率的解決方案。

This study proposes a grey-level algorithm-assisted SEM framework for reconstructing buried structures in advanced semiconductor devices. By combining Monte Carlo simulation, BSE signal analysis, and deep learning with ResNet-UNet architecture, the system predicts 3D morphology from single SEM images. The model incorporates atomic number-weighted grey-level mapping and multi-angle image analysis to enhance depth estimation accuracy, particularly for high-aspect-ratio features and subsurface defects.
The results demonstrate strong agreement between simulation and experimental data, validating the model’s accuracy in interface detection and defect profiling. This method offers a non-destructive, cost-efficient solution for nanoscale inspection, suitable for integration into industrial metrology workflows.

[1] Q. Zhang, Y. Zhang, Y. Luo, and H. Yin, “New structure transistors for advanced technology node CMOS ICs,” National Science Review, vol. 11, no. 3, 2024.
[2] J. Teshima and J. J. Clarke, “From transistors to bumps: Preparing SEM cross-sections by combining site-specific cleaving and broad ion beam milling,” Hitachi High Technologies, Clarksburg, MD, Tech. Paper
[3] Y. Y. Illarionov et al., “Process implications on the stability and reliability of 300 mm FAB MoS₂ field-effect transistors,” npj 2D Materials and Applications, vol. 8, no. 1, p. 8, 2024.
[4] A. Jalilvand et al., “Gate-oxide-short defect analysis and fault modeling in FinFETs”, IEEE TCAD, 2020
[5] K. Cheng et al., “3D CMOS-memristor hybrid circuits: devices, integration, architecture, and applications”, 2012.
[6] H. Radamson et al., “State of the Art and Future Perspectives in Advanced CMOS Technology”, Nanomaterials, 2020.
[7] Q. Zhang et al., “Optimization of structure and electrical characteristics for four-layer vertically-stacked horizontal gate-all-around Si nanosheets devices,” unpublished, 2024.
[8] J. Hu, T. Liu, P. Ji, G. Zhang, Y. Huang, C. F. Cheung, and S. Yang, “High‑sensitivity, high‑throughput inspection of nanoscale defects using a laser confocal positioning‑assisted darkfield imaging system,” unpublished.
[9] Which Electron Detector is Right for Your Application?, SEM Tech Blog, Scanning Electron Microscopy, Thermo Fisher Scientific. [Online]. Available: https://www.thermofisher.com/blog/microscopy/which-electron-detector-is-right-for-your-application/
[10] D. Samak, A. Fischer, and D. Rittel, “3D reconstruction and visualization of microstructure surfaces from 2D images,” Laboratory for CAD & Lifecycle Engineering, Technion–Israel Institute of Technology, Haifa, Israel, Tech. Rep., 2007.
[11] Logic and Memory Applications in Semiconductor and Electronics, ZEISS Microscopy. [Online]. Available: https://www.zeiss.com/microscopy/us/applications/semiconductors-electronics/logic-memory.html
[12] Liao, S., et al. "First demonstration of monolithic CFET inverter at 48nm gate pitch toward future logic technology scaling." 2024 IEEE International Electron Devices Meeting (IEDM). IEEE, 2024.
[13] HOUBEN, Tim, Huisman, T.: Pisarenco, M.; Van Der Sommen, F,; PHN de With. Depth estimation from a single SEM image using pixel-wise fine-tuning with multimodal data. Machine Vision and Applications, 2022, 33.4: 56.
[14] VAN KESSEL, L. C. P. M.; HAGEN, C. W. Nebula: Monte Carlo simulator of electron–matter interaction. SoftwareX, 2020, 12: 100605.
[15] GODARD, Clément, Aodha O. M.; Firman, M.; Brostow, G. J. Digging into self-supervised monocular depth estimation. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. 2019. p. 3828-3838.
[16] BINNIG, Gerd; QUATE, Calvin F.; GERBER, Ch. Atomic force microscope. Physical Review Letters, 1986, 56.9: 930.
[17] GAUVIN, Raynald. Review of transmission electron microscopy for the characterization of materials. Materials Characterization and Optical Probe Techniques: A Critical Review, 1997, 10291: 196-225.
[18] DEN BOEF, Arie J. Optical metrology of semiconductor wafers in lithography. In: International Conference on Optics in Precision Engineering and Nanotechnology (icOPEN2013). SPIE, 2013. p. 57-65.
[19] GOODFELLOW, Pouget-Abadie, J.; Mirza, M.; Xu, B.; Warde-Farley, D.; Ozair, S.; Courville, A.; Bengio. Y. Generative adversarial nets. Advances in Neural Information Processing Systems, 2014, 27.
[20] KINGMA, Diederik P.; Welling, M. Auto-encoding variational bayes. arXiv preprint arXiv:1312.6114. 2013.
[21] Atomic number and crystallographic contrast images with the SEM: a review of backscattered electron techniques, Cambridge University Press, published 05 July 2018.
[22] M. Čalkovský, E. Müller, and D. Gerthsen, “Quantitative analysis of backscattered-electron contrast in scanning electron microscopy,” Journal of Microscopy, vol. 288, no. 1, pp. 33–44, Jan. 2023, doi: 10.1111/jmi.13148.
[23] VERDUIN, T.; LOKHORST, S. R.; HAGEN, C. W. GPU accelerated Monte-Carlo simulation of SEM images for metrology. In: Metrology, Inspection, and Process Control for Microlithography XXX. SPIE, 2016. p. 122-135.
[24] THIBAULT Lechien, Enrique Dehaerne, Bappaditya Dey, Victor Blanco, Sandip Halder, Stefan De Gendt, Wannes Meert 2023 Automated Semiconductor Defect Inspection in Scanning Electron Microscope Images: a Systematic Review 2308.08376
[25] SAKAKIBARA M, Suzuki M, Tanimoto K, Sohda Y, Bizen D and Nakamae K: Impact of secondary electron emission noise in SEM Microscopy 68 279–88 (2019)
[26] NAKAMAE K, Fujioka H and Ura K Measurements of deep penetration of low-energy electrons into metal-oxide-semiconductor structure J. Appl. Phys. 52 1306–8 (1981)
[27] BUNDAY, B., Solecky, E., Vaid, A., Bello, A.F., Dai, X.: Metrology capabilities and needs for 7 nm and 5 nm logic nodes. In: Metrology, Inspection, and Process Control for Microlithography XXXI, vol. 10145, p. 101450 (2017).
[28] HENAO-LONDOÑO, J.C., Riaño-Rojas, J.C., Gómez-Mendoza, J.B., Restrepo-Parra, E.: 3D stereo reconstruction of SEM images. Modern Appl. Sci. 12(12), 57 (2018).
[29] WANG, T.-C., Liu, M.-Y., Zhu, J.-Y., Tao, A., Kautz, J., Catanzaro, B.: High-resolution image synthesis and semantic manipulation with conditional GANs. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 8798–8807 (2018)
[30] VLADÁR, A.E., Villarrubia, J.S., Chawla, J., Ming, B., Kline, J.R., List, S., Postek, M.T.: 10 nm three-dimensional CD-SEM metrology. In: Metrology, Inspection, and Process Control for Microlithography XXVIII, vol. 9050 (April 2014), p. 90500 (2014).
[31] EIGEN, D., Puhrsch, C., and Fergus, R.: Depth map prediction from a single image using a multi-scale deep network. In: Advances in Neural Information Processing Systems, pp. 2366-2374, (2014).
[32] LIU, F., Shen C., and Lin, G.: Deep convolutional neural fields for depth estimation from a single image. In: IEEE Conf. on Computer Vision and Pattern Recognition, pp. 5162-5170, (2015)
[33] Arat, K. T., Bolten, J., Zonnevylle, A. C., Kruit, P., & Hagen, C. W. (2014). Estimating Step Heights from Top-Down SEM Images.
[34] Desai, V., & Reimer, L. (1989). Calculation of the Signal of Backscattered Electrons Using a Diffusion Matrix from Monte Carlo Calculations. Physikalisches Institut, Universitat Miinster, Miinster, Federal Republic of Germany.
[35] Arat, K. T., & Zonnevylle, A. C. (2013). Verification of height and sidewall angle SEM metrology accuracy using Monte Carlo simulation.
[36] Houben, T., Huisman, T., Pisarenco, M., van der Sommen, F., & de With, P. (2022). Training procedure for scanning electron microscope 3D surface reconstruction using unsupervised domain adaptation with simulated data. Journal of Electronic Imaging, 31(4), 043502.
[37] van Kessel, L., Huisman, T., & Hagen, C. W. (2020). Understanding the influence of 3D sidewall roughness on observed line-edge roughness in scanning electron microscopy images. Proceedings of SPIE, 11325, Metrology, Inspection, and Process Control for Microlithography XXXIV, 113250Z. https://doi.org/10.1117/12.2550240
[38] Sweeney, T., Coleman, S., & Kerr, D. (2022). Deep learning for semiconductor defect classification. 2022 IEEE 20th International Conference on Industrial Informatics (INDIN), 1–6. https://doi.org/10.1109/INDIN51773.2022.9976162
[39] B. Chakrabarti et al., “A multiply-add engine with monolithically integrated 3D memristor crossbar/CMOS hybrid circuit”, Scientific Reports, 2017
[40] X. Wu et al., “B-open defect: A Novel Defect Model in FinFET Technology”, IEEE VTS, 2021.
[41] H. H. Radamson et al., “The Challenges of Advanced CMOS Process from 2D to 3D,” Applied Sciences, vol. 7, no. 10, p. 1047, Oct. 2017. [Online]. Available: https://doi.org/10.3390/app7101047
[42] Q. Zhang et al., “Optimization of Structure and Electrical Characteristics for Four-Layer Vertically-Stacked Horizontal Gate-All-Around Si Nanosheets Devices,” ResearchGate, [Online]. Available: https://www.researchgate.net/publication/349832007_Optimization_of_Structure_and_Electrical_Characteristics_for_Four-Layer_Vertically-Stacked_Horizontal_Gate-All-Around_Si_Nanosheets_Devices [Accessed: Jun. 10, 2025].
[43] FinFET to GAA MBCFET: A Review and Insights, IEEE Access, vol. PP, no. 99, pp. 1–1, Jan. 2024, doi: 10.1109/ACCESS.2024.3384428.
[44] J. Lim, J. Lee, and C. Shin, “Probabilistic artificial neural network for line-edge-roughness-induced random variation in FinFET,” IEEE Access, vol. 9, pp. 86786–86794, Jun. 2021, doi: 10.1109/ACCESS.2021.3088461.
[45] G. Espiñeira, D. Nagy, G. Indalecio, A. J. García-Loureiro, K. Kalna, and N. Seoane, “Impact of gate edge roughness variability on FinFET and gate-all-around nanowire FET,” IEEE Electron Device Letters, early access, doi: 10.1109/LED.2019.2900494.
[46] G. E. Lloyd, “Atomic number and crystallographic contrast images with the SEM: a review of backscattered electron techniques,” Microscopy and Microanalysis, vol. 24, no. 5, pp. 1–20, Jul. 2018. doi: 10.1017/S1431927618000434.