EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

CTransE: a CNN-transformer network for long-sequence prediction of plasma etching profiles

Zerui Jin, Tingyang Qin, Mengjiao Lu, Quan Liu, Tiantian Shi, Wenbin Zhou, Xiangze Liu, Yiming Liao, Feng Yan, Kai Wang, Xiaoli Ji

Journal of Physics D: Applied Physics · 17.07.2026 · 59 (28) · 285203

DOI: 10.1088/1361-6463/ae830a ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Physics D: Applied Physics, online erschienen am 15.07.2026. Verfasst von Zerui Jin und 10 weiteren Autor:innen. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 42 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Fundamental and Interdisciplinary Disciplines Breakthrough Plan of the Ministry of Education of China, National Natural Science Foundation of China.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Data-driven modeling of etching processes allows for the accurate prediction of etch profiles significantly enhancing the efficiency of integrated circuit (IC) research. In this study, a convolutional Transformer based etching network (CTransE) is developed to extract multi-scale features from etching profiles, enabling accurate modeling and prediction under variable process conditions. Recognizing that etching profile evolution is a physically constrained, long-sequence task, the model exploits the standard Transformer architecture to effectively capture cross-timestep dependencies in physical image sequences through self-attention mechanism. We specifically employed the CTransE model for the etching of SiO 2 using Ar/CF 4 gas mixtures using a simulated dataset of 18 360 2D profiles and a few-shot experimental dataset. Compared to conventional recurrent neural network or long short-term memory, CTransE model reduces the mean squared error by 93.15% and improves the structural similarity index measure by 29.02%, enabling accurate predictions of real-world experimental profiles. The inference time for a single profile is 0.13 s, significantly speedup the prediction efficiency of temporal etching processes. The proposed method delivers a fast, robust, and low-cost strategy for monitoring plasma etching in small-node IC fabrication.

Förderung

  • Fundamental and Interdisciplinary Disciplines Breakthrough Plan of the Ministry of Education of China
  • National Natural Science Foundation of China

Zitieren

APA

Jin, Z., Qin, T., Lu, M., Liu, Q., Shi, T., Zhou, W., Liu, X., Liao, Y., Yan, F., Wang, K., & Ji, X. (2026). CTransE: a CNN-transformer network for long-sequence prediction of plasma etching profiles. Journal of Physics D: Applied Physics, 59(28), 285203. https://doi.org/10.1088/1361-6463/ae830a

Vancouver

Jin Z, Qin T, Lu M, Liu Q, Shi T, Zhou W, et al. CTransE: a CNN-transformer network for long-sequence prediction of plasma etching profiles. Journal of Physics D: Applied Physics. 2026;59(28):285203. doi:10.1088/1361-6463/ae830a

BibTeX
@article{Jin2026,
  author = {Jin, Zerui and Qin, Tingyang and Lu, Mengjiao and Liu, Quan and Shi, Tiantian and Zhou, Wenbin and Liu, Xiangze and Liao, Yiming and Yan, Feng and Wang, Kai and Ji, Xiaoli},
  title = {CTransE: a CNN-transformer network for long-sequence prediction of plasma etching profiles},
  journal = {Journal of Physics D: Applied Physics},
  year = {2026},
  volume = {59},
  number = {28},
  pages = {285203},
  doi = {10.1088/1361-6463/ae830a}
}
Kostenlos starten →