Discussion Forum

Hi all,

I am trying to study the evanescent field around a nanophotonic cavity. My structure is very similar to the one from Painter group for the ones who know his work (see painterbis.png). I reproduce his structure as an example here for my question : see 01.png. The structure is in Silicon with a box of air all around (1 micrometer) and scattering boundary conditions all around.

I did 2 steps : - find the resonance mode : with an eigenfrequency study, it works perfectly : see 03.png - With this frequency as an input, I did a frequency domain study (see 10.png : I added a port (one ON sendind a TE10 mode and one OFF) on each side of the cavity, hence I added 2 steps for the study : boundary mode step). For this one frequency, I get 05.png so the right result (I am plotting ewfd.Ey each time) - you can notice that with the frequency domain modal study I get the same : see 11.png

Then, if I look at another quantity, that is to say the phase or here the quantity |Ey|^2/( |Ey|^2+ |Ex|^2), which is just a projection of the field on the vector (0 1).T : if this quantity is equal to 1, the field is linearly polarized, if it not equal to 1 the polarization is elliptic. Here is my problem :

With the result of the eingenfrequency study, when I plot this quantity for the mode found just before, I get 04.png. You can see that there is a lot of red and it seems kinda smooth and nice. With the result of the frequency domain study I get 09.png. My question is : why is it so different ?

I know that for the eigenfrequency study, there is no source, so Comsol just resolves the helmholtz equation on the mesh to find the modes allowed by the geometry. Whereas in the frequency domain study case, there is a source term added. In that case, I am not sure about what comsol resolves in the structure, I haven't find out much informations about that.

My understanding is what follows : to be closer to the experiment : we send light through the source I put, so I should first do an eigenfrequency study to find where is the mode, and then do a frequency study in order to study the field around. Hence, using the eigenfrequency study to study the field itself would not have a real sense, this is just to find the modes allowed by the geometry.

Or am I wrong ? Is there something I am missing ? I know I didn't put PML for this case, but the field is almost dead so far from the structure so the scattering boundary conditions are enough.

Thank you very much for your time,

François