Mesh Refinement


Hello Lumerical Community,

I had a question about how best to refine the mesh region within the software with respect to simulating a resonance cavity between four metal (silver) plates. A rough overview of the orientation of the plates is shown below where each vertical dash mark represents a single silver plate:

                                                                                                                               |  |
                                                                                                                               |  |

I am wondering why, when I adjust the number of mesh cells, the resonant frequency of the cavity shifts. I have the mesh refinement option set to ‘conformal variant 1’ since my simulation involves metals and have the boundary conditions set to ‘PML’, but I don’t understand why a change in number of mesh cells would change the resonance of the cavity. I am unsure if the accuracy of my results is dependent on a specific number of mesh cells, if I have chosen the wrong mesh refinement option, or if I even need to utilize mesh refinement.

I have been using an electric dipole source in the TE polarization mode for my simulations. My goal has been to range the emission wavelength of the dipole and observe the wavelength corresponding to the greatest magnitude of electric field and power within the cavity in order to determine the resonant frequency of said cavity. Once I have determined the cavity’s resonance, I will then move the dipole outside of the cavity at various distances and monitor the power transmitted through one opening of the cavity out through the other. I am simulating with a light source at a wavelength of approximately 31 mm.

Any help would be greatly appreciated!

Mesh Refinement Continued..

A couple of things I could think of are if the gap distances between the plates aren’t being resolved accurately, or if the resonant fields in the cavity aren’t decaying by the end of the simulation time.

For the first point, ideally the mesh lines will fall at the edges of the structures and there should be at least 2 mesh cells over a gap to accurately resolve the thickness of a gap.

For the second point, if the fields don’t fully decay by the end of the simulation time, the Fourier transform of the time domain fields could have artifacts like the ones discussed here:

If you’re still seeing unexpected results you could also upload a copy of your simulation file.


Attached is my simulation file that I have been working on. Thank you, Nancy, for your feedback on the matter.

Fig_2a_Ag_Plates_Dipole_1mm_Outside_No_Apod_Al.fsp (269.6 KB)



I checked the simulation file and I think one main issue is that the Al material that you were using does not have any data in the wavelength range that you are simulating. Since in the GHz frequency range, metals can be approximated at perfect electrical conductors, I would recommend changing the material to “PEC (Perfect Electrical Conductor)” as discussed here:

As well, inside the cavity, there will be light that bounces back and forth at angles very close to the x-direction, so when these fields reach the PML at the y max and y min boundaries, since PML can have degraded performance for absorbing light at steep angles, it can cause artificial reflections. I would recommend using the “steep angle” PML profile with a large number of PML layers to avoid this. The different PML settings are discussed on this page:

Finally, you may want to try out different apodization settings for your monitor since in order to get the fields of the resonant mode, the filter window should be at a late enough time to fully exclude the initial source pulse and transient fields.

If you are still finding that the results aren’t as expected please let me know.

Another further suggestion is to consider using the high Q analysis group discussed on this page in order to extract the resonant frequency of the cavity:

This analysis group uses time monitors placed at different locations in the cavity and performs some post processing to calculate both the resonant frequencies and the quality factors of the resonances of a cavity.