Reply To: Why is the FDTD solusion of the microdisk resonator not converged?

[Guilin Sun]

This could be due to several different causes. One is the device performance its self. For example, the device has high Q resonance, which will need much longer simulation time. You can use a point time monitor and then check its spectrum to know this. If long simulation time does not resolve this, you may also reduce the autoshutoff min. In general, you may need to increase the simulation time, and reduce the autshotoff min (you can check the log file and the transmission curve (if it is smooth) to determine if it is necessary).
The second cause might be due to the PML settings. If you kept the original override mesh with 10nm mesh size, it can be too fine for longer wavelength. I would suggest to disable the override mesh first, and see if the mesh accuracy 2 can resolve the disk well, eg, more than 10 meshes on each axis. You may need to change mesh size. When operating at longer wavelength and finer mesh size, the PML performance may be degraded as it depends on the thickness for the given parameters to absorb light striking on it. So you may 1) increase the number of layers; 2) increase the FDTD spans to make the PML away from the disk, such as half wavelength.
You may also need to change the PML type from the current "uniaxial anisotropic" to "stretched coordinate" PML, then choose "Standard", or "Steep Angle" depending on the disk interaction with light. If light radiates from the disk at near glazing angle, the "Steep Angle" PML may be used.

Please try those methods and see if it works for you.