Photonics

Photonics

Discrepancy between RCWA and FDTD

    • Fan Yang
      Subscriber

      I'm using RCWA to simulate diffraction effiicency of different orders with a grating structure, but the result is inconsistent with the FDTD simulation. My structure is attached here. Thanks a lot if anyone can help me find out what's the problem!

    • Guilin Sun
      Ansys Employee

      Please elaborare more regarding to the discrepancy: what the mesh accuracy and PML thickness you used? does the simulation run long enough and reach the autoshutoff min? how many k-vectors you used in RCWA?

      They are two different solvers with different source of errors. Some discrepancy is common, if both of them are not approaching theorerical results, eg, due to using finite sampling.

      In addition,what result is different? the diffraction angles, the diffraction orders, or the diffraction efficiency or anything else?

    • Fan Yang
      Subscriber

      Thank you for your response! Here are my settings for the mesh accuracy and PML thickness in FDTD. I ran it long enough to reach the autoshutoff min. For RCWA, I tried different number of k-vectors ranging from 5 to 300, but I found every time it gives me different result.

      As for the difference, it's about the diffraction efficiency. They can have 30% difference on the first order. I also found that if I simulate 2 periods of the grating, FDTD gives me almost the same value as 1 period, but RCWA gives me different values, and it also varies depending on the number of k-vectors. So I think the problem is most probably coming from RCWA. I also attached a table summarizing all the results.

      Thank you!

    • Guilin Sun
      Ansys Employee

      When every time it gives different result, it means the result has not converged.

      As you said, FDTD is pretty consistent. you may need to spend more time on RCWA convergenging tests. I would not suggest you to use two priods, as the result should not be better than one period if all the parameters are properly set.

      Your structure can be considered as 4 different gratings working together, so it might be somewhat challenge. Please do more tests. If you have premium support, please contact us through email. thank you!

Viewing 3 reply threads
  • You must be logged in to reply to this topic.