Plasmonic lithography optical proximity correction method based on a decomposed machine-learning fast imaging model
Optics Express · 16.11.2026 · 34 (23) · 43674
DOI: 10.1364/oe.611390 ↗Kurz gefasst
Fachartikel in der Zeitschrift Optics Express, online erschienen am 04.11.2026. Verfasst von Zhenyu Xing und 3 weiteren Autor:innen. Die Arbeit nennt 21 Literaturangaben. Gefördert von: Research on Resolution Enhancement Technology for Deep Ultraviolet Band Plasmonic Imaging Lithography.
Abstract (Originalsprache)
Strong near-field coupling in plasmonic lithography readily induces optical proximity effects such as line-end shortening, corner rounding, and local linewidth variation. In contrast, conventional model-driven optical proximity correction (OPC) requires repeated rigorous electromagnetic field calculations in each iteration, which severely limits computational efficiency. To address this issue, this paper proposes a plasmonic lithography OPC method based on decomposition-based machine-learning fast-imaging model. The method uses a pre-trained imaging transfer matrix to rapidly predict the photoresist image (PRI) of mask patterns. It embeds this prediction into a closed-loop optimization process driven by edge placement error (EPE) feedback. The signed EPE determines the movement direction of each edgex`, and an adaptive correction coefficient that varies with the overall EPE controls the movement distance, thereby enabling an iterative correction flow of fast imaging, threshold exposure, EPE calculation, edge movement, and mask update. Simulation results on seven complex two-dimensional patterns show that the proposed method converges stably under different illumination modes, pattern sizes, and polarization states, with the main errors decreasing rapidly within the first six iterations. After ten correction iterations, the average reductions in the sum of absolute EPE and pattern error are 75.03% and 63.54%, respectively, under TE polarization, and 73.77% and 63.02%, respectively, under TM polarization. Compared with OPC, which is based on a rigorous electromagnetic field imaging model on the same hardware platform, the overall computational efficiency, including the one-time training cost, is approximately 4 times higher. These results indicate that the proposed method can significantly reduce computational cost while maintaining correction accuracy and convergence stability, providing an effective route for fast plasmonic lithography OPC of complex two-dimensional patterns.
Förderung
- Research on Resolution Enhancement Technology for Deep Ultraviolet Band Plasmonic Imaging Lithography
Zitieren
APA
Xing, Z., Ding, H., Ren, P., & Wei, Y. (2026). Plasmonic lithography optical proximity correction method based on a decomposed machine-learning fast imaging model. Optics Express, 34(23), 43674. https://doi.org/10.1364/oe.611390
Vancouver
Xing Z, Ding H, Ren P, Wei Y. Plasmonic lithography optical proximity correction method based on a decomposed machine-learning fast imaging model. Optics Express. 2026;34(23):43674. doi:10.1364/oe.611390
BibTeX
@article{Xing2026,
author = {Xing, Zhenyu and Ding, Huwen and Ren, Pengyu and Wei, Yayi},
title = {Plasmonic lithography optical proximity correction method based on a decomposed machine-learning fast imaging model},
journal = {Optics Express},
year = {2026},
volume = {34},
number = {23},
pages = {43674},
doi = {10.1364/oe.611390}
}
EUVIMED