Singularity problem in periodic plasmonic structure

Hi, I was trying to simulate a periodic plasmonic structure and l used 1 nm mesh for most of my simulation region.

  1. when I looked at the power transmission of my bottom monitor, at certain wavelength the power flow is positive(meaning power flowing from outside into the fdtd region?). What caused this problem?
  2. I was using (1-reflection-transmission) to calculate the power absorbed in the structure but sometimes this gives Absorption greater than 1. I think this might come from problem1 but I was also wondering if this is a reasonable method to find the resonance for plasmonic structures.

Thank you,
sweep_4.fsp (991.8 KB)

When there are grating orders travelling at close to 90 degrees there can be evanescent fields of the structure which can extend out to the PML boundary and this can degrade the performance of the PML. In these cases, it can help to extend the distance between the structure and the PML so that the evanescent fields decay before reaching the PML. Typically about half a wavelength’s distance is required.

I tested your file using an increased z span of the FDTD simulation region of 1500 nm, and the result was improved (none of the transmission values exceeded 1). I think you can try even further increasing the z span and number of PML layers until the results converge. Please try it out and let me know how it goes!

Thank you very much @nlui. I was able to get transmission all below 1. But when I look at the R or T spectrum, I can still see a really sharp peak at 729nm. How can I tell if this is a real peak resulting from the plasmonic structure or simply from some non-converging problem?

-sweep_4.fsp (993.1 KB)

I checked the resulting spectrum and it looks like the feature could be a real physical effect due to the Wood’s anomaly which occurs at frequencies where the number of grating orders supported by the structure changes so the new grating order has light travelling basically at 90 degrees along the surface of the structure.

If you perform convergence testing of the PML and the feature doesn’t change, then it indicates that it is physical.

Hopefully this helps!