Acta Optica Sinica (Online), Volume. 2, Issue 16, 1602001(2025)

Inverse Design of Ultraviolet Absorber Utilizing Tandem Neural Networks

Xufei Fan, Wenrui Xue*, and Fanyi Meng
Author Affiliations
  • College of Physics and Electronic Engineering, Shanxi University, Taiyuan 030006, Shanxi , China
  • show less
    Figures & Tables(17)
    Schematics of the unit structure of UV broadband absorber. (a) Three-dimensional view; (b) top view
    Dielectric constant of TiN versus wavelength
    Contour plot depicting the structure transmittance as a function of substrate thicknesses t1 and wavelength
    FNN model
    Loss value varies with the number of epochs when training FNN
    Comparison of absorption spectra simulated and predicted by trained FNN
    TNN model
    Loss value varies with the number of epochs when training TNN
    Comparison of absorption spectra simulated and predicted by trained TNN
    Comparison of target absorption spectrum, TNN predicted absorption spectrum, and simulated absorption spectrum at incident angle of 50°
    Contour plots of absorptivity. (a) TM wave incidence; (b) TE wave incidence
    Absorption spectra of TM and TE waves at incident angle of 0°
    Contour plots of absorptivity corresponding to different azimuth angles at incident angle of 0°
    Relative impedance versus wavelength
    Electromagnetic field distributions at 290 nm when TM wave and TE wave are vertically incident. (a)‒(c) Electric field distribution of TM wave; (d)‒(f) magnetic field distribution of TM wave; (g)‒(i) electric field distribution of TE wave; (j)‒(l) magnetic field distribution of TE wave
    • Table 1. Scanning ranges of four structural parameters

      View table
      View in Article

      Table 1. Scanning ranges of four structural parameters

      ParameterRange /nmStep /nmNumber of points
      t280‒11057
      t3140‒17057
      p370‒40057
      d25‒5557
    • Table 2. Comparison of proposed absorber with previously reported adsorbers

      View table
      View in Article

      Table 2. Comparison of proposed absorber with previously reported adsorbers

      Ref.MaterialGeometryAbsorption range /nm

      Average absorptivity /%@α=0°

      (200‒400 nm)

      Minimum absorptivity /%@α=0°

      (200‒400 nm)

      Maximum incident angle /(°) @λ=300 nm(absorptivity>80%)
      TM modeTE mode
      20SiO2-TiN-SiO2-TiNRing-square array200‒120094.8906570
      21TiN-SiO2-TiNCross-shaped array200‒250095.4905050
      22TiN-Si3N4-TiN-Si3N4-Ti-WMulti-layer nanodisk array200‒300096.7936060
      33TiN-SiO2-TiN-SiO2-TiNThree-layer hollow ring array200‒188094.690
      32TiN-TiO2-TiNNanodisk-square ring array250‒3000

      94.3

      (250‒400 nm)

      90

      (250‒400 nm)

      5050
      This workTiN-SiO2-TiNGrid-pattern array200‒40097.494.35353
    Tools

    Get Citation

    Copy Citation Text

    Xufei Fan, Wenrui Xue, Fanyi Meng. Inverse Design of Ultraviolet Absorber Utilizing Tandem Neural Networks[J]. Acta Optica Sinica (Online), 2025, 2(16): 1602001

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category: Photonic and Optoelectronic Devices

    Received: Jul. 7, 2025

    Accepted: Jul. 16, 2025

    Published Online: Aug. 7, 2025

    The Author Email: Wenrui Xue (wrxue@sxu.edu.cn)

    DOI:10.3788/AOSOL250490

    CSTR:32394.14.AOSOL250490

    Topics