• Photonics Research
  • Vol. 4, Issue 2, 0093 (2016)
Yufei Xing1, Domenico Spina2, Ang Li1, Tom Dhaene2, and Wim Bogaerts1、3、*
Author Affiliations
  • 1Photonics Research Group, Department of Information Technology, Center for Nano and Biophotonics, Ghent University imec, Ghent B-9000, Belgium
  • 2Department of Information Technology, Internet Based Communication Networks and Services (IBCN), Ghent University iMinds, Gaston Crommenlaan 8 Bus 201, B-9050 Gent, Belgium
  • 3Luceda Photonics, 9200 Dendermonde, Belgium
  • show less
    DOI: 10.1364/prj.4.000093 Cite this Article Set citation alerts
    Yufei Xing, Domenico Spina, Ang Li, Tom Dhaene, Wim Bogaerts. Stochastic collocation for device-level variability analysis in integrated photonics[J]. Photonics Research, 2016, 4(2): 0093 Copy Citation Text show less
    Upper plot shows the perspective view of a symmetric DC. Red arrows present the flow of light. Part of the light is coupled from bottom waveguide to the above one. Cross section is amplified in the lower plot. The mean width and thickness of the DC are w0 and t0, respectively. The width w and thickness t of the fabricated DC are indicated as dashed boxes. The refractive indexes are nsi=3.44, nSiO2=1.45.
    Fig. 1. Upper plot shows the perspective view of a symmetric DC. Red arrows present the flow of light. Part of the light is coupled from bottom waveguide to the above one. Cross section is amplified in the lower plot. The mean width and thickness of the DC are w0 and t0, respectively. The width w and thickness t of the fabricated DC are indicated as dashed boxes. The refractive indexes are nsi=3.44, nSiO2=1.45.
    2D contour plot of field coupling coefficient versus waveguide width and thickness.
    Fig. 2. 2D contour plot of field coupling coefficient versus waveguide width and thickness.
    Flow chart of the proposed technique.
    Fig. 3. Flow chart of the proposed technique.
    Top: the red exes (×) represent the interpolation nodes for the normalized independent random variables ξ1 and ξ2 used to build the SC model. Bottom: the blue circles (°) are the corresponding values for the correlated random variables w and t used to compute the coupling coefficients in Fimmwave.
    Fig. 4. Top: the red exes (×) represent the interpolation nodes for the normalized independent random variables ξ1 and ξ2 used to build the SC model. Bottom: the blue circles (°) are the corresponding values for the correlated random variables w and t used to compute the coupling coefficients in Fimmwave.
    Sampling points used to perform the MC analysis through direct Fimmwave simulations for the correlated random variables (w,t). The corresponding values for the independent random variables (ξ1,ξ2) are used to evaluate the SC model computed.
    Fig. 5. Sampling points used to perform the MC analysis through direct Fimmwave simulations for the correlated random variables (w,t). The corresponding values for the independent random variables (ξ1,ξ2) are used to evaluate the SC model computed.
    Blue circles (°): coupling coefficient computed via the MC analysis for the 10000 (w,t) samples shown in Fig. 5. Red (×)-markers: corresponding values obtained by evaluating the SC model.
    Fig. 6. Blue circles (°): coupling coefficient computed via the MC analysis for the 10000 (w,t) samples shown in Fig. 5. Red (×)-markers: corresponding values obtained by evaluating the SC model.
    PDF and CDF of the coupling coefficient for λ=1.55 μm. The blue solid and red dashed line are PDF and CDF obtained by means of the SC model, respectively, while the blue circles and red squares represent the same quantities computed by means of the MC analysis.
    Fig. 7. PDF and CDF of the coupling coefficient for λ=1.55  μm. The blue solid and red dashed line are PDF and CDF obtained by means of the SC model, respectively, while the blue circles and red squares represent the same quantities computed by means of the MC analysis.
    PDF and CDF of the 3 dB-coupling length for λ=1.55 μm. The blue solid and red dashed line are PDF and CDF obtained by means of the SC model, respectively, while the blue circles and red squares represent the same quantities computed by means of the MC analysis.
    Fig. 8. PDF and CDF of the 3 dB-coupling length for λ=1.55  μm. The blue solid and red dashed line are PDF and CDF obtained by means of the SC model, respectively, while the blue circles and red squares represent the same quantities computed by means of the MC analysis.
     MCSC
    Mean value65,16065,166
    S.t.d value2616.92631.4
    Table 1. Performance Summary of SC and MC Simulation
    Variability Analysis TechniqueSimulatorNumber of PointsComputation Time
    MCFimmwave FMM Solver1000021 h 53 min 14 s
    SCStochastic modelingFimmwave FMM Solver658 min 32 s
    MC using stochastic modelSC stochastic model1000027 s
    Total time8 min 59 s
    Table 2. Computation Time of SC and MC Simulation
    Yufei Xing, Domenico Spina, Ang Li, Tom Dhaene, Wim Bogaerts. Stochastic collocation for device-level variability analysis in integrated photonics[J]. Photonics Research, 2016, 4(2): 0093
    Download Citation