Machine Learning-Assisted FDTD Design of Broadband Terahertz On-Chip Antennas Utilizing Surface Plasmon Polaritons
Plasmonics · 13.10.2026 · 21 (12) · 4287-4299
DOI: 10.1007/s11468-026-03492-z ↗Kurz gefasst
Fachartikel in der Zeitschrift Plasmonics, online erschienen am 13.10.2026. Verfasst von Anis Omidniaee und 2 weiteren Autor:innen. Die Arbeit nennt 63 Literaturangaben.
Abstract (Originalsprache)
Terahertz (THz) on-chip plasmonics antennas are promising candidates for next-generation integrated communication and sensing systems; however, their practical implementation remains challenging due to substrate losses, impedance mismatch, and limited radiation performance at sub-millimeter wavelengths. In this work, a physics-informed machine learning approach is proposed for the optimization of broadband THz on-chip antennas based on a GaAs–TiN heterostructure platform. The proposed framework combines full-wave electromagnetic simulations with machine learning-assisted surrogate modeling to efficiently explore the complex relationship between antenna geometry, material characteristics, and radiation performance. The integration of high-permittivity GaAs with conductive TiN provides an effective platform for controlling electromagnetic confinement and improving energy coupling at THz frequencies. By employing the developed optimization strategy, the antenna parameters are optimized to achieve enhanced impedance matching, broadband operation, and improved directional radiation within a compact footprint. Numerical results demonstrate improved impedance matching and radiation performance, with a minimum reflection coefficient of − 56.53 dB at 193.44 THz, a − 10 dB bandwidth of 8.01 THz corresponding to a fractional bandwidth of 4.14%, a peak directivity of 17.28 dBi, a peak gain of 15.99 dBi, and an estimated radiation efficiency of 76.09%. The presented approach provides an efficient pathway for accelerating the design of advanced THz integrated devices by combining artificial intelligence techniques with engineered heterostructure materials. The proposed GaAs–TiN antenna architecture can offer potential applications in future high-capacity wireless communication, chip-scale sensing, and sub-terahertz photonic-electronic systems.
Zitieren
APA
Omidniaee, A., Barati, K., & Farmani, A. (2026). Machine Learning-Assisted FDTD Design of Broadband Terahertz On-Chip Antennas Utilizing Surface Plasmon Polaritons. Plasmonics, 21(12), 4287-4299. https://doi.org/10.1007/s11468-026-03492-z
Vancouver
Omidniaee A, Barati K, Farmani A. Machine Learning-Assisted FDTD Design of Broadband Terahertz On-Chip Antennas Utilizing Surface Plasmon Polaritons. Plasmonics. 2026;21(12):4287-4299. doi:10.1007/s11468-026-03492-z
BibTeX
@article{Omidniaee2026,
author = {Omidniaee, Anis and Barati, Keyvan and Farmani, Ali},
title = {Machine Learning-Assisted FDTD Design of Broadband Terahertz On-Chip Antennas Utilizing Surface Plasmon Polaritons},
journal = {Plasmonics},
year = {2026},
volume = {21},
number = {12},
pages = {4287--4299},
doi = {10.1007/s11468-026-03492-z}
}
EUVIMED