Ring Resonator Model and desing


#1

Hi , my name is James i’ve been trying to Model a Ring resonator, with 2.5D, 3D and also with the SiEPIC Library Model to compare with my experimental results. and i have some questions.–the complete report and questions is in this link:

Report -Ring Resonator Analysis and Questions.

Brief:
• Why it is the SiEPIC library model more accurate than the 3D model for the Q value? Is it because of mesh accuracy or a script is apply to obtain a more realistic behaviour?
• Considering what the Lumerical’s 2.5D FDTD Propagation Method [1] says about considering negligible the slabsmodestocalculatetheeffectiveindex.Itspossibleto obtain a more accurate results in 2.5D simulations if the geometry used would allow a more pure TE-like mode, for example 440 nm of width and 180 of thickness?
• In order to measure the thermo-optic response, how can a temperature variation can be simulated in FDTD solutions?


#2

Hi James,

It looks like the SiEPIC library you are using contains components characterized using 3D FDTD simulations so you should be able to get the same results for the Q factor result using the same method for both cases (by using Q = f_R/delta_f, where f_R is the resonant frequency and delta_f is the FWHM from the transmission spectrum of the device). It could be possible that the results are different if the mesh used in the simulation is too coarse since the coupling efficiency is sensitive to the meshing around the gap. It is also possible that there are some other errors, eg. if the simulation time is too short.

It could be true that using a device which supports modes with higher TE polarization fraction could give more accurate results, however I’m not sure. In my opinion I think it would be best to compare the results with 3D FDTD to confirm the results.

To simulate the thermo-optic response, it could be possible to use the index perturbation material like described here:
https://kb.lumerical.com/en/index.html?ref_sim_obj_index_perturbation_material.html


#3

Hi , thanks for your response, i have been traing to improve my simulation results,
and I can not even get a Q value close to the experimental results.
i show you my results exporting the S parameters to INTERCONNECT:

2.5D S parameters

Experimental results vs simulated results. overlay

My Q value for experimental resuls are 5690,6100,5320,4733,4050
but the simulation gives me 3676,3146,2467,2437,2107

My simulation file is here:
Ring-r5-Best Q Obtained.lms (1.2 MB)

i have put some new mesh in the coupling region.
if you can tell me how can i improve my simulation i’ll be very thankful


#4

Hi,

I checked the simulation and I found two potential issues. I noticed that the simulation did not reach the auto shutoff threshold so there could be artifacts in the transmission spectrum due to ending the simulation early which could affect the calculated Q factor:
https://kb.lumerical.com/en/index.html?ref_sim_obj_frequency_monitors_simulation_time.html

I also noticed that the mode expansion monitor y position is not centered with respect to the waveguide. This could cause error in the expansion results since the overlap between the field profiles of the monitors for expansion and the calculated modal fields from the expansion monitor are done with respect to the center of each monitor. The mode expansion monitor and monitor for expansion need to be centered at the same position with respect to the waveguide cross section.

With these two changes, there may still be some discrepancy due to approximations of the 2.5D method, so for final results I would recommend 3D FDTD.