Topics related to Lumerical and more

RCWA and Transmission Monitor Convergenec

    • schinta3


      I am trying to simulate the transmission through a photonic crystal slab, however the RCWA solver and standard transmission monitor techniques do not seem to align. Is there something wrong with my approach?

      Additionally, .fsp or .lsf files cannot be attached here. I can send them over if needed.



    • Kyle
      Ansys Employee

      Ansys staff are not permitted to exchange files on the ALF. However, you can include screenshots or sections of script in your post itself.
      In general we would expect that the FDTD and RCWA to be fairly close. They are different methods which make different approximations, however, so some small discrepancies in their results are expected. Some things to double check:
      Make sure the unit cells being simulated are the same. For example, for a unit cell normal to the z direction, the "x min", "x max", "y min", and "y max" properties of the FDTD solver should match the "x_min", "x_max", etc. properties in the "geometry" struct for the RCWA solver. Note that the units of the RCWA geometry struct are meters.
      Check that the layer boundaries are properly defined in the "geometry" struct. For geometries with sloped or graded geometries, you will need multiple layers along the sloped/graded section.
      Check that the polarization of the source matched in RCWA and FDTD. The source in RCWA can be s, p, or unpolarized, while the source in FDTD can be s or p polarized.
      FDTD simulations of periodic structures can be difficult due to diffraction into higher grating orders. You should use a "steep angle" PML profile, and you may need to increase the simulation time and reduce the autoshutoff threshold. Also make sure you are using the right choice between periodic, Bloch, and BFAST boundaries.
      Perform convergence testing for FDTD and RCWA. For FDTD, reduce the mesh, increase the number of PML layers, increase the simulation time and reduce the autoshutoff threshold. For RCWA, increase the number of k-vectors by increasing the "max_N" and increase the number of layers over any graded/sloped geometries.
      When troubleshooting it is helpful to simplify the simulations. For example, you can:
      Change the materials to "" and give them a constant index.
      Use a single frequency instead of a broadband simulation.
      Use a source at normal incidence instead of at an angle.
      Simplify the structure, for example by using uniform layers instead of patterned layers.
      Once the results match for the simpler simulation, you can start making it more complex.
      Viewing 1 reply thread
      • You must be logged in to reply to this topic.