I did some tests, and found that it can reach convergence by modifying the following:
1: use narrowband instead of broadband (the refractive index does not change much)
2:increase xy simulation span
3: use 36 or 64 PML layers
4: delete the current varFDTD object, add a new one and modify the settings
5: increase simulation time.
Mostly the divergence is caused by PML, as you can extend the Profile monitor to cover PDF and record the data inside the PML. There is strong field intensity/power density in front of PML.
You are navigating away from the AIS Discovery experience