Adiabatic Vertical Taper Coupler EME Simulation Issue

[dbose]

1. I am trying to simulate an adiabatic vertical taper coupler in EME, that couples from a 80 nm thick Silicon nitride core waveguide to a 800 nm thick silicon nitride core one. There is SiO2 cladding all around; specifically about 320 nm of it between the two silicon nitride cores. The width of the 80 nm core at the start of its taper is 6 um, while the width of the 800 nm core at the start of its taper from its end is 2 um. 
2. This is paper is a good example of the structure to what I'm trying to simulate - https://opg.optica.org/abstract.cfm?uri=iprsn-2013-IT2A.4, but obviously the parameters in my simulation are changed.
3. The EME simulation is ran initially with a 100 um taper length; with extrapolation to longer lengths after simulation is complete. And the CVCS cell in the same region is divided into 200 subcells. The mesh step in vertical direction in taper region is 3 nm.
4. The problem I'm seeing is then that the simulation results now(image below) do not seem to agree with results of a similar simulation I had done about a year ago wherein I got more almost 100% coupling within several mm of taper coupler length. Additionally, the coupling % seems to reduce with increased length now which does not seem to be correct physics. I have though seen the results get more accurate (increase in coupling %) for longer lengths as I increase subcell count of the CVCS cell; but the RAM usage is already almost a 100 GB. Previously I did not even have to use more than 20 subcells for the results to be accurate.

Q. I was wondering then what to change in the simulation now to get accurate results and find the taper length at which the coupling would go to like 99%.?

[Dimitris Polyzos]

Hello 

Thank you for contacting us. I wil lbe happy to help you with your queries.

Unfortunately I cannot open the reference link you shared. Could you please share the title of the publication and I will try to find it through google.

Something thta might affect your resutls is the version of Lumerical you use. Could you please let me know which version do you use? Did you create your model in an older version and now you open it with a newer one?

You could have a look at thic link to check the changes between Lumerical versions: https://support.ansys.com/Components/WhatsNew

Let me know if you have access to this link. 

Something more is that increasing the number of cells you increase the accuracy of your model in propagation direction. I reckon that longer lengths should have less coupling according to theory due to losses. For a length of mm range you should increase accordingly the cells to test the accuracy as well. I would be interested to have a look at your model in order to have a better understanding of your simulation model. A convergence test should be conducted as well to know that your model is correclty build and the results you get are valid with a correct mesh selection.

PLease have a look at these articles as well:

MODE-EigenMode-Expansion-EME-solver-introduction

EME-Convergence-Testing-An-Intuitive-Approach

Convergence-testing-process-for-EME-simulations

I hope I helped.

Please let me know if you have any queries. I will be happy to help.

Kind regards

Dimitris

 

[dbose]

Hi Dimitris,

This is the title of the paper in the link - "Optical Interconnect for 3D Integration of Ultra-Low Loss Planar Lightwave Circuits".

We are using Lumerical 2023 R2.0, version: 7.22.3457. 

The issue described initially is something I experienced only after an update earlier last year. But I see that there is a new version now and will report to you our results.

It is asking me to register for customer support space again at the version changes link -  https://support.ansys.com/Components/WhatsNew. Not sure if that should be the case.

Thank you again for your help. I will report back to you on this thread if results improve.

Best,
Debapam.

[Dimitris Polyzos]

Hello Debapa, 

Thanks for sharing the publication reference. I will try to find it online and have a look at it.

Regarding the version of Lumerical, I assume that this might be the reason of this issue. You could upgrade the Lumerical version to 25R1.1 and check if the issue persists.

You could register for customer support. In that way you could share your model and me or one of my collegaues will be able to check the results of your model for various versions. 

Please let me know if you have any findings with your issue. 

Regards

Dimitris