EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Neural Adjoint Inverse Design of Three‐Dimensional Nanophotonic Devices

Chanik Kang iD, Hyewon Suk iD, Ikbeom Jang iD, Haejun Chung iD

Nanophotonics · 27.10.2026 · 15 (20)

DOI: 10.1002/nap2.70334 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Nanophotonics, online erschienen am 16.10.2026. Verfasst von Chanik Kang und 3 weiteren Autor:innen. Beteiligte Einrichtung: Department of Artificial Intelligence Hanyang University Seoul Korea. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 69 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Ministry of Science and ICT, South Korea, National Research Foundation of Korea, Ministry of Culture, Sports and Tourism.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Volumetric meta‐optics have emerged as a compact high‐performance platform for advanced imaging, color routing in image sensors, and on‐chip wavefront control. Realizing such devices depends on solving a high‐dimensional nonconvex inverse‐design problem governed by Maxwell's equations. Adjoint optimization is the standard route to gradients, yet broadband three‐dimensional (3D) design typically requires thousands of full‐wave forward/adjoint simulations, making industrial‐scale optimization prohibitively slow. To address this bottleneck, a neural adjoint method is introduced, in which a solver‐supervised Fourier neural operator (FNO) directly predicts dense, per‐voxel 3D adjoint gradient fields from a voxelized permittivity volume. To preserve the sharp, resonance‐driven sensitivity peaks that single‐pass operator networks tend to oversmooth, a stage‐wise FNO is proposed, in which successive stages refine residual errors with progressively higher‐frequency emphasis. A domain‐specific dataset is curated from paired forward/adjoint finite‐difference time‐domain (FDTD) simulations spanning three volumetric tasks: spectral sorting (color routers), achromatic focusing (metalenses), and waveguide mode conversion. Across all benchmarks, the surrogate‐driven optimization attains figures of merit within 86%–99% of rigorous FDTD‐based adjoint optimization whilst reducing end‐to‐end design time from hours to seconds, with speedups exceeding . These results establish a pathway toward fast large‐scale volumetric meta‐optical design enabled by neural‐operator surrogates.

Schlagwörter

Stichwörter der Autor:innen

adjoint methodfourier neural operatorinverse designmeta‐opticsscientific machine learningsurrogate modeling

Autor:innen und Institutionen

  • Department of Artificial Intelligence Hanyang University Seoul Korea.
  • Department of Artificial Intelligence Semiconductor Engineering Hanyang University Seoul Korea.
  • Department of Computer Engineering Hankuk University of Foreign Studies Yongin Korea.
  • Department of Electronic Engineering Hanyang University Seoul Korea.

ORCID-Profile: Chanik Kang, Hyewon Suk, Ikbeom Jang, Haejun Chung

Förderung

  • Ministry of Science and ICT, South Korea
  • National Research Foundation of Korea
  • Ministry of Culture, Sports and Tourism

Zitieren

APA

Kang, C., Suk, H., Jang, I., & Chung, H. (2026). Neural Adjoint Inverse Design of Three‐Dimensional Nanophotonic Devices. Nanophotonics, 15(20). https://doi.org/10.1002/nap2.70334

Vancouver

Kang C, Suk H, Jang I, Chung H. Neural Adjoint Inverse Design of Three‐Dimensional Nanophotonic Devices. Nanophotonics. 2026;15(20). doi:10.1002/nap2.70334

BibTeX
@article{Kang2026,
  author = {Kang, Chanik and Suk, Hyewon and Jang, Ikbeom and Chung, Haejun},
  title = {Neural Adjoint Inverse Design of Three‐Dimensional Nanophotonic Devices},
  journal = {Nanophotonics},
  year = {2026},
  volume = {15},
  number = {20},
  doi = {10.1002/nap2.70334}
}
Kostenlos starten →