EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Data-driven insight into propagation and bend losses in silicon nitride waveguides using classification and regression trees and design of experiments

Jakob Wilhelm Hinum-Wagner, Peter Scheibelhofer, Samuel Marko Hörmann, Christoph Schmidt, Gandolf Feigl, Jochen Kraft, Alexander Bergmann

Optical Materials Express · 01.06.2025 · 15 (6) · 1278

DOI: 10.1364/ome.564676 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Optical Materials Express, online erschienen am 28.05.2025. Verfasst von Jakob Wilhelm Hinum-Wagner und 6 weiteren Autor:innen. Die Arbeit wurde bisher 4-mal zitiert (Crossref) und nennt 36 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Photonic integrated circuits play a pivotal role in high-speed telecommunications and sensing, yet minimizing optical losses in scalable fabrication remains a persistent challenge. This study focuses on silicon nitride waveguides and applies a full-factorial design of experiments combined with regression trees to identify the layer-specific origins of propagation and bend excess losses. Regression tree modeling revealed that waveguide width is the dominant predictor of propagation loss, followed by material composition—particularly the upper cladding at lower widths and the waveguide core material at higher widths—highlighting the confinement-dependent interplay of scattering and absorption. Bend excess loss analysis indicated a shift in dominant predictors from bend radius and width to waveguide material at smaller radii. Although these dependencies have been previously established through physical modeling and experimental studies, most prior work covers smaller datasets or relies on idealized fabrication assumptions. The results of this work establish a statistically grounded and interpretable framework for waveguide loss analysis and underscore classification and regression trees’ utility in photonic process optimization. Our approach complements existing physical frameworks by providing a statistically grounded, data-driven method to optimize SiN photonic platforms, offering predictive insight into complex layer—geometry interactions and fabrication-induced variations that are difficult to model physically. By enabling predictive insight into layer—geometry interactions, this approach paves the way for more robust photonic integrated-circuit design in sensing and next-generation optical systems.

Zitieren

APA

Hinum-Wagner, J. W., Scheibelhofer, P., Hörmann, S. M., Schmidt, C., Feigl, G., Kraft, J., & Bergmann, A. (2025). Data-driven insight into propagation and bend losses in silicon nitride waveguides using classification and regression trees and design of experiments. Optical Materials Express, 15(6), 1278. https://doi.org/10.1364/ome.564676

Vancouver

Hinum-Wagner JW, Scheibelhofer P, Hörmann SM, Schmidt C, Feigl G, Kraft J, et al. Data-driven insight into propagation and bend losses in silicon nitride waveguides using classification and regression trees and design of experiments. Optical Materials Express. 2025;15(6):1278. doi:10.1364/ome.564676

BibTeX
@article{HinumWagner2025,
  author = {Hinum-Wagner, Jakob Wilhelm and Scheibelhofer, Peter and Hörmann, Samuel Marko and Schmidt, Christoph and Feigl, Gandolf and Kraft, Jochen and Bergmann, Alexander},
  title = {Data-driven insight into propagation and bend losses in silicon nitride waveguides using classification and regression trees and design of experiments},
  journal = {Optical Materials Express},
  year = {2025},
  volume = {15},
  number = {6},
  pages = {1278},
  doi = {10.1364/ome.564676}
}
Kostenlos starten →