Directional coupler with a splitting ratio of 50/50

[felme1]

Hey,

I am not sure if this is the right place to ask, but I wouldn't know where else to go.
I participated in the edX course at the end of 2024 and I would like to use the upcoming production run to manufacture a specific circuit that involves a directional coupler with a splitting ratio of 50/50.
Since I don't have the lumerical license anymore I tried to simulate the default Ebeam directional coupler (10um coupling length, 0.2um gap).
However the result looked like this:

I was always, perhaps wrongly so, under the impression, that the default dc would yield a coupling ratio of 50/50.
But as long as my simulation, which I was able to crudely confirm using tidy3d, is correct this is not the case.
Therefore I have two questions:

1. Am I right in thinking that the default Ebeam coupler does not have a splitting ratio of 50/50? Or am I just not taking into account, that the component will be a little bit larger and therefore the gap will be smaller on the final manufactured component due to the lithography process?

2. Have you been able to test a directional coupler with specific dimensions for which I can be reasonably certain that it will have a coupling ratio of 50/50 in my final design?

Thank you in advance for the answers!

[lukasc-ubc]

Hi @felme1

The device a parameterized cell for which measurement results are here: https://github.com/SiEPIC/SiEPIC_EBeam_PDK/blob/master/docs/components/ebeam/directional_coupler/directional_coupler.md

The default is not 50/50. For the default parameters, the compact model we have shows about -0.8 dB / -8.6 dB splitting. You can find the raw data for the s-parameters in this folder: https://github.com/SiEPIC/SiEPIC_EBeam_PDK/tree/master/klayout/EBeam/CML/EBeam/source_data/ebeam_dc_te1550
It looks like about L = 22 µm gives closer to 50/50.

It would be great if you can simulate the design using tidy3d, even for a range of length values.

I suggest including a process bias of -15 nm (fabrication “shrink”) for ANT fabrication. For example, you can design for 500 nm but expect a fabricated 485 nm width.

Note that the simulations in the folder probably did not take process bias into account, so an updated / validated model would be appreciated.

If you have more data and info to add to the documentation file, that would be great!

thank you

[felme1]

Hi @lukasc-ubc ,
sorry for the late reply, I was occupied with finalizing my design for last weekends design tapeout. I used Meep to calculate the s parameters of a small range of coupling lengths, taking into account the fabrication shrink that you mentioned. I am happy to add all of my data to the documentation file.
As I ran the simulation with meep the output looks different from the one you would get with lumerical. However I will try to match this format.