Discussion Forum

Hello Comsolers,
I am trying to model focusing Gaussian wave in optics module with substrate and air geometry to study the optical characteristics of the scatterer on substrate. Gaussian wave focus at the interface of air and Si. I am solving with 2 step procedure in 2D similar to the one in modal library except it is in 2D and with Gaussian wave. To verify my model, I calculated relEz without scatterer ( as explained in srdjancomsol.weebly.com) and it is 0.03 V/m where background Ez is about 1.2 V/m. In the absence of scatterer, relEz (=Ez-Ebz) is supposed to zero (ideally) but it should be as minimum as possible (actually), say in my case 0.00005 would be acceptable (as per srdjan). He explained that improving the mesh quality reduces this a lot.
I tried several ways to reduce the relE to minimum possible: disabling the scatterer for uniform mesh, reducing the mesh element size by 3 times (maximum mesh element is lambda_freespace/15), changing to cyllindrical wave in scattering BC assuming focussing Gaussain is closer to cylindrical wavefronts in 2D, reducing the substrate refractive index from 3.4 to 1.6 guessing any TIR at focussing happens, increasing w0 ( beam waist) size to reduce the effect of TIR (if presents). These have not given any much improvements. So, I changed model materials all to air (no substrate in this case) and tried even with plane wave, just I am curious if it works at east for plane wave in free space in 2-step solvers; still I see same amplitude without scatterer. If Ebz is 1.47 then Ez is 1.50 approx in second solver, which means that field is amplifying a bit in scattering formalism (second solver). relE is about 3% in my case.
If anybody has such experience or any ideas on how to model gaussian wave in substrate case and verifying model would be appreciated.

Thanks
Mahendar