• Photonics Research
  • Vol. 10, Issue 4, A57 (2022)
Carlos Pérez-Armenta1、*, Alejandro Ortega-Moñux1, José Manuel Luque-González1, Robert Halir1、2, Pedro J. Reyes-Iglesias1, Jens Schmid3, Pavel Cheben3, Íñigo Molina-Fernández1、2, and J. Gonzalo Wangüemert-Pérez1、2
Author Affiliations
  • 1Telecommunication Research Institute (TELMA), Universidad de Málaga, CEI Andalucía TECH, E.T.S.I. Telecomunicación, 29010 Málaga, Spain
  • 2Bionand Center for Nanomedicine and Biotechnology, Parque Tecnológico de Andalucía, Málaga 29590, Spain
  • 3National Research Council Canada, Ottawa, Ontario K1A 0R6, Canada
  • show less
    DOI: 10.1364/PRJ.446932 Cite this Article Set citation alerts
    Carlos Pérez-Armenta, Alejandro Ortega-Moñux, José Manuel Luque-González, Robert Halir, Pedro J. Reyes-Iglesias, Jens Schmid, Pavel Cheben, Íñigo Molina-Fernández, J. Gonzalo Wangüemert-Pérez. Polarization-independent multimode interference coupler with anisotropy-engineered bricked metamaterial[J]. Photonics Research, 2022, 10(4): A57 Copy Citation Text show less
    (a) Schematic of the proposed polarization-insensitive 2×2 MMI (SiO2 cladding is not shown for clarity). (b) Top view of the taper from input silicon wire to bricked SWG waveguide. (c) Top view of the bricked SWG multimode waveguide.
    Fig. 1. (a) Schematic of the proposed polarization-insensitive 2×2 MMI (SiO2 cladding is not shown for clarity). (b) Top view of the taper from input silicon wire to bricked SWG waveguide. (c) Top view of the bricked SWG multimode waveguide.
    (a) Procedure employed for modeling the bricked SWG. (b) Calculated refractive index components for a bricked SWG waveguide with h=220 nm, Λx=200 nm, DCx=50%, Λz=240 nm, and DCz=50% at the wavelength λ=1.31 μm. (c) Beat length for wMMI=3 μm using the 2D anisotropic model (solid curves) and full 3D simulations (dotted curves).
    Fig. 2. (a) Procedure employed for modeling the bricked SWG. (b) Calculated refractive index components for a bricked SWG waveguide with h=220  nm, Λx=200  nm, DCx=50%, Λz=240  nm, and DCz=50% at the wavelength λ=1.31  μm. (c) Beat length for wMMI=3  μm using the 2D anisotropic model (solid curves) and full 3D simulations (dotted curves).
    Design methodology used in this work, comprising two main stages: (a) design of the bricked SWG multimode waveguide and (b) optimization of the complete device using a 3D-FDTD simulator.
    Fig. 3. Design methodology used in this work, comprising two main stages: (a) design of the bricked SWG multimode waveguide and (b) optimization of the complete device using a 3D-FDTD simulator.
    Absolute value of the relative difference between TE and TM beat lengths |ΔLπ¯|=|ΔLπ/Lπ¯|, where ΔLπ=Lπ(TE)−Lπ(TM) and Lπ¯=(Lπ(TE)+Lπ(TM))/2, calculated using the anisotropic 2D model. Simulation parameters of the unit cell are h=220 nm, DCz=DCx=50%, Λx=200 nm, and λ=1.31 μm. The dashed line represents the points (dz,Λz) for which ΔLπ¯=0. The optimal parameters Λz=240 nm and dz=Λz/2 are marked with a green dot.
    Fig. 4. Absolute value of the relative difference between TE and TM beat lengths |ΔLπ¯|=|ΔLπ/Lπ¯|, where ΔLπ=Lπ(TE)Lπ(TM) and Lπ¯=(Lπ(TE)+Lπ(TM))/2, calculated using the anisotropic 2D model. Simulation parameters of the unit cell are h=220  nm, DCz=DCx=50%, Λx=200  nm, and λ=1.31  μm. The dashed line represents the points (dz,Λz) for which ΔLπ¯=0. The optimal parameters Λz=240  nm and dz=Λz/2 are marked with a green dot.
    Beat length as a function of wavelength obtained from 3D Floquet–Bloch simulations of the (bricked) SWG waveguide. Solid lines correspond to dz=0, while dashed lines correspond to dz=0.48Λz. Other geometrical parameters are: wMMI=3 μm, h=220 nm, Λx=200 nm, DCx=50%, Λz=220 nm, and DCz=50%.
    Fig. 5. Beat length as a function of wavelength obtained from 3D Floquet–Bloch simulations of the (bricked) SWG waveguide. Solid lines correspond to dz=0, while dashed lines correspond to dz=0.48Λz. Other geometrical parameters are: wMMI=3  μm, h=220  nm, Λx=200  nm, DCx=50%, Λz=220  nm, and DCz=50%.
    Final design performance of the optimized polarization-independent 2×2 MMI.
    Fig. 6. Final design performance of the optimized polarization-independent 2×2 MMI.
    Dependence of the bandwidth and center wavelength of the MMI on fabrication error δ. The bandwidth is defined as the wavelength range in which PDL, EL, and |IB| are simultaneously below 1 dB, and |PE| is less than 5°. The inset shows the effect on the in-plane geometry of positive and negative values of δ. The original geometry (top view, i.e., in the chip plane) is represented with a continuous line, while the over/under-etched geometry is represented with a dashed line.
    Fig. 7. Dependence of the bandwidth and center wavelength of the MMI on fabrication error δ. The bandwidth is defined as the wavelength range in which PDL, EL, and |IB| are simultaneously below 1 dB, and |PE| is less than 5°. The inset shows the effect on the in-plane geometry of positive and negative values of δ. The original geometry (top view, i.e., in the chip plane) is represented with a continuous line, while the over/under-etched geometry is represented with a dashed line.
    Calculation of |ΔLπ¯| as performed in Fig. 4 for different transversal periods Λx and duty cycles (DC=DCx=DCz) and the wavelength of λ=1310 nm. The dashed line corresponds to |ΔLπ|=0. The design point is marked with a green dot in (d). For Λx>200 nm or DC≠50%, the design space for |ΔLπ¯| is closer to the Bragg zone [see cases (a), (e), and (g)] or no solution can be found, as shown in (b), (c), (f), (h), and (i).
    Fig. 8. Calculation of |ΔLπ¯| as performed in Fig. 4 for different transversal periods Λx and duty cycles (DC=DCx=DCz) and the wavelength of λ=1310  nm. The dashed line corresponds to |ΔLπ|=0. The design point is marked with a green dot in (d). For Λx>200  nm or DC50%, the design space for |ΔLπ¯| is closer to the Bragg zone [see cases (a), (e), and (g)] or no solution can be found, as shown in (b), (c), (f), (h), and (i).
    ParameterwMMIΛz (DCz)Λx (DCx)dzNMMIwawsNT
    Value2.9 μm220 nm (50%)200 nm (50%)110 nm1211.2 μm0.8 μm21
    Table 1. Optimized Design Parameters of Polarization Insensitive 2×2 MMI Coupler
    ReferencePerformanceλ0BW (Δλ/λ0)TypeFootprintSi Thickness (nm)
    [20]EL<1dB1550 nm80 nm (5%)1×23.9  μm×15.3  μm400
    [36]EL<1dB, |IB|<1dB, PDL<0.1dB1550 nm200 nm (13%)1×22  μm×7.7  μm250
    [21]EL<0.3dB, |IB|<0.7dB1550 nm100 nm (6%)2×22  μm×14.4  μm340
    This workEL<1dB, |IB|<1dB, PDL<1dB, |PE|<5°1310 nm160 nm (12%)2×23  μm×35.9  μm220
    Table 2. Comparison of Our Device with Other State-of-the-Art Polarization-Insensitive SOI-based MMI Couplers
    Carlos Pérez-Armenta, Alejandro Ortega-Moñux, José Manuel Luque-González, Robert Halir, Pedro J. Reyes-Iglesias, Jens Schmid, Pavel Cheben, Íñigo Molina-Fernández, J. Gonzalo Wangüemert-Pérez. Polarization-independent multimode interference coupler with anisotropy-engineered bricked metamaterial[J]. Photonics Research, 2022, 10(4): A57
    Download Citation