Learning Forum

[Will Eshbaugh]

Hi all,

I am running some "reflectivity" simulations for an inverse-designed photonic cavity and encountered a normalization issue with the transmission through a field/power monitor. I am plotting spectra for the light coupled out of the grating coupler and comparing that to the reference spectra when there is no grating coupler present (ideally to see if the dip at the cavity can still be detected amidst the fabry-perot modes that appear with the grating couplers).

Here is the resulting transmission "T" through the field monitors (monitor is placed halfway between structure surface and FDTD boundary):

• the reference spectra (Blue) looks like what I expect (dip near cavity resonance and roughly >90% power is reflected back into the waveguide)
• the grating coupler simulations have two issues:
  1. Transmission peaks >1
  2. Expected transmission should remain < 0.5 for a fully-etched, suspended grating coupler

 

Simulation screenshot(s) for visualization. (Left - reference sim without grating, Right - example with grating)

Some other details:

• Lumerical 2023R2.2
• FDTD Mesh: accuracy 3, conformal variant 0
• Auto Shutoff: 1e-7 (all 3 simulation logs confirm this was satisfied)
• Boundaries: All PML except zmin=symmetric
• Material: 150nm thick suspended AlGaAs (40% Al), imported sampled 3D n,k data

 

Does anyone know what might be causing this? Let me know if I can provide additional info for clarification. Thank you.

[Guilin Sun]

When transmission/reflection is larger than 1, most likely it is due to lack of simulation time to let the signal completely decay. You can significantly increase the simulation time, and let autoshutoff min to terminate the simulation. When resonance is strong, you may also need to reduce autoshutoff min.

In addition, make sure the waveguide extendes to outside of PML. Make sure PML is thickness enough, say half wavelength.

 

[Will Eshbaugh]

Thanks Guilin, that makes sense. The grating couplers do create some fairly long-lived resonances between the grating and cavity. I will try extending the runtime and decreasing the autoshutoff min.