Learning Forum

[alessandro.brugnoni01]

Hello,
I'm Alessandro and I'm currently working on thin-film lithium niobate electro-optic phase modulator.
Considering the example:
https://optics.ansys.com/hc/en-us/articles/19435937674387-Thin-Film-Lithium-Niobate-Electro-Optic-Phase-Modulator

, I set the lithium niobate telecom permittivities of the CHARGE script to:
eps_o = 2.2111^2;
eps_e = 2.1376^2;
and I retrieved an optical loss of almost 10dB (slightly varying according to the applied voltage) for the TE00 mode.

Then, I tried to unlock the mesh of FEEM and CHARGE and the loss was quite different (1.5 dB). Why did this value change so much?

I read from another post (https://innovationspace.ansys.com/forum/forums/topic/various-charge-monitor-results-for-the-same-problem/) that the mesh varies if it is not locked. However, this is quite a big change.

Thank you in advance.

[kghaffari]

Hi Alessandro,

As you correctly mentioned, the mesh will be different if it not locked. I have the following recommendations to verify the effect that the mesh is having on your results:

A) Once you obtain the 10 loss with your permittivities, please keep the mesh locked but change the permittivity to obtain the loss again. The reason I’m suggesting this step is to make sure that your permittivity values are actually taking effect, while the mesh is locked. If the results don’t change it could mean that the original permittivity was not taking effect while the mesh was locked. Otherwise, please proceed to the next step.

B) If while the mesh is locked changing the permittivity demonstrates change in loss, then please repeat the simulation of unlocked mesh with a smaller mesh size. I expect the loss difference to be smaller for a more refined mesh (compared to the original 10db vs 1.5 db). In this step, it’s important to visualize the mesh to make sure a smaller mesh is actually generated. If you see no difference in the new mesh (while mesh is unlocked), try increasting the “max refine steps” under the mesh tab (in addition to reducing min and max edge length).

Hope this helps. Best regards,

Khash

[alessandro.brugnoni01]

 

Hi,
I went deeper into simulations during the last days and I verified your concerns. Specifically:
a) the change in the permittivity demonstrates change in loss. Therefore, changes are taking effect.
b) to investigate the convergence of the optical loss, I took a step back. I considered the Ansys example with the previous parameters:

eps_o = 2.21^2;
eps_e = 2.14^2;

I added a mesh to the FEEM solver (referred to the lithium niobate waveguide) and I considered a max edge length of 0.005 micrometer. 
The CHARGE parameters were left unchanged (max edge length = 1um, min edge length = 0.05um, mesh max edge length = 0.01um).
Additionally, the CHARGE and the FEEM solver are left unlocked.

Acting on the FEEM settings (polynomial order and edges per wavelength and considering a “refined mesh based on the material properties”), the following results are retrieved for the optical loss [dB/cm]:

Therefore, the optical loss approaches to almost 2 dB/cm when the FEEM parameters are correctly set. 

In conclusion, I must assume that there was some inaccuracy in the computation of the loss. Do you agree on this?

I also have two additional questions:

1) I tried to replicate the Ansys example by using my own script (same parameters are set). However, I’m able to retrieve an effective refractive index at 0V that is 2.0895 (different from the Ansys example where the effective refractive index is 2.02046).
An effective refractive index equal to 2.0895 is also obtained when:
               i) the nk material is removed from the Ansys example
               ii) the FDE solver is used
I suppose there should be some error/change given by the insertion of the nk material in the Ansys example. Do you agree?

2) A x/y cut configuration is considered for the lithium niobate waveguide and the TE00 mode is computed out of it. Anyway, the ordinary refractive index (2.21) is selected for the lithium niobate configuration. Shouldn’t be used the extraordinary index (2.14) since the r33 coefficient is excited?

Thank you in advance.