This video is taken from the EME Learning Track on Ansys Innovation Courses.
Transcript
In FDE, metal boundary conditions are used when finding modes of a waveguide or fiber
where there is no radiative loss.
If there is enough distance between the boundaries and the structures, the supported modes that
the Eigensolver finds is not affected by the choice of the boundary conditions - metal
or PML.
Since metal boundaries allow for relatively faster mode calculations compared to PML,
metal boundaries are generally the best choice for devices with well-confined modes.
Like when using the FDE solver, metal boundaries can be used in EME when the propagating field
is well-confined in the transverse directions.
For waveguides with radiation loss, the scattered field can couple into the artificial modes
formed by the metal boundaries.
Although some of the scattered light can couple back into the guided modes of the physical
structure of interest, the amount of coupling is often negligible for a weakly scattering
waveguide, so the user s-matrix result will not be compromised.
If there is a lot of scattering in the device, the coupling becomes more evident.
If you look at the field profile of the propagating field, you may see the reflection from the
metal boundaries.
To avoid this, you can move the metal boundaries further away from the waveguide to minimize
the coupling back into the structure.
But a more general solution would be to use PML boundaries instead of metal boundaries.
In some cases, you might want to characterize the radiative loss from your device, such
as for this edge coupler example where light leaks into the silicon substrate over the
length of the taper.
In this case, you need to use PML boundaries to absorb the radiating light and account
for the loss due to radiation.
Usually, we want to optimize devices with low loss (low scattering), so it is a good
idea to use metal boundaries and avoid PML boundaries whenever possible, since PML requires
more memory and time to simulate and it can introduce some artificial gain or loss.