Mesh size for MMI with 0.4 microns access waveguide width

mesh-size

#1

Hello,

I am working on simple MMI. For waveguide width 1.5 microns convergence occurs for the mesh size 0.07 microns and 60 modes (for the range of MMI widths from 3 to 8 microns)
For waveguide width 0.4 microns convergence occurs for mesh size 0.01 at 40 to 50 modes for the range of MMI width from 5 to 8 microns but for low MMI widths (3,4 microns there is no proper convergence was able to figure out till 100 modes) for any mesh size from 0.01 microns to 0.07 microns. Is there any suggestion for this scenario?


#2

Hi @minusunny

Can you please elaborate on your inquiry and share the simulation file with me.

In general mesh size and number of modes are two separate factors. Finer mesh improves the mode spatial distribution and the overlap calculation results, however, you may need to use a large number of modes to make sure that coupling to all the modes are considered.

Thanks


#3

Please see the file attached
https://drive.google.com/open?id=1V5sVLdrwIVNtSRP8LsZ3rDvN1eoh-DiO

Mode%20convergence
This is the mode convergence i obtained for 0.07 micron mesh size. Transmission on 0.13 % @60 MODES

01
This is the mode convergence for 0.02 mesh size. Transmission on 0.3645% @ 60 MODES

Can this be considered as a good mode convergence?


#4

Hi @minusunny

I did some modifications into your simulation file. Most importantly I added mesh override regions, increased the number of modes to 200, and applied anti-symmetric BC along the y-direction. Please see attached:
MMI_BK.lms (356.5 KB)

I performed mode convergence sweep and here is the result that I obtained:

image

This means that you will need to have ~150 modes to make sure that results are reasonable. You can run the simulations with no applied symmetry to compare the results. Below shows the mode profile over the simulation region:

image


#5

Thank you for your response.