Oscillations in reflected power at butt-joint interface

[valeria.rustichelli]

Hello,

Now I am facing a new problem.

I add 90 nm of MQW in the buried waveguide and when I vary the angle of small values I obtain the graph shown in this figure.

I also added a monitor to check the first order mode and it is even worst.

Is this normal or I am missing something?

Below the simulation without the wells

Thanks again

Valeria

[valeria.rustichelli]

Reflections at butt-joint interface

It is usually better to start with metal boundary conditions everywhere except for x min and x max. The reason is that the PML can introduce additional effects. If the fields decrease to very small values before they reach the boundary it should be safe to use metal boundaries. However, in some cases (for example for taking into account radiative losses) it is necessary to use PML.

I have just noticed that the simulation has PML in the y min and y max.

If I change them to metal, the simulation changes and I have more reflections at a larger angle…I am a bit confused about this!

Thanks a lot

Valeria

[valeria.rustichelli]

In case you need I upload also the script

Valeriareflections_modified_MQW.lsf (7.4 KB)

[fgomez]

Hi Valeria,

Thanks for waiting. I have a couple of comments:

New simulation with the added MQW layer:

This is a situation where varFDTD would not be the right tool to use. The reason is that the vertical profile of the waveguide changes drastically since the MQW layer is only present on one side of the structure. This can lead to vertical mode coupling as explained here:

varFDTD vs EME

I am trying to decide which solver on MODE solution can be suitable for my 3D structure since simulations using 3D FDTD aren’t time efficient. I just can’t understand this sentence from the knowledge base “For planar geometries where there is negligible coupling between vertical modes, varFDTD can achieve accuracy comparable to 3D FDTD with computation times comparable to 2D FDTD. For 3D geometries where there is coupling between vertical modes, EME is a frequency domain method that offers a g…

In this case EME would be the right tool to use. Without too much extra effort you could also try the previous structure without MQW with EME.

Ultimately, you can try a full 3D-FDTD simulation to check your results.

Oscillations in the reflection:

I am not sure if the oscillations you see are an artifact of the simulation. They are more pronounced for the net reflection in the case with MQW, where varFDTD is definitely not appropriate. My suggestion would be to check if you see the same behavior using the EME solver. There are some additional things to check:

• Distance between the waveguides and the simulation boundaries: As you increase this distance you might find that the simulation results change. It is a good idea to do a convergence testing for one angle to find the distance such that when you increase the distance further the results don’t change significantly. Once you find this, you can use this setting for all the angles in the sweep. Remember to make sure the structure is large enough so that the boundaries are inside the structure.
• Metal or PML boundaries: The plots for total and TE reflected power show that some power is not reflected back in the fundamental TE mode. Probably, there are radiative losses at the interface between the waveguides. If that is the case, you should use PML boundaries everywhere.

[valeria.rustichelli]

Hi thanks a lot for your feedback.

I was wondering if with an EME simulation I can really see the influence of the tilted interface.

Thanks a lot

Valeria

[fgomez]

Hi Valeria,

Yes, you can simulate the effect of the tilted interface. Since the structure profile perpendicular to the propagation direction changes with position at the tilted interface you need to use several cells in that region. I suggest taking a look at this getting started example of a spot size converter. As shown there, you can approximate the continuously varying structure profile with a number of cells; the staircasing effect can be minimized using the CVCS method.