Open cgubbin opened 3 years ago
cgubbin
commented
3 years ago Update: I partially solved this issue by changing the eigenfrequency solution "Scaling of Output Vectors" parameter to maximum. I still find a lot of spurious solutions but now also find the physical modes.
Tong-Wu1121
commented
3 years ago Hello cgubbin, It's normal to find the spurious modes. As mentioned by you, this is the problem of numerical stability. From many numerical tests, we find this problem may be mitigated by using a swept mesh across a thin background layer attached to the resonator, as that has been done in the spherical resonator modal. Also, be sure to enter a suitable guessed frequency in the Settings of Eigenfrequency. The latest version of the program will be updated on https://www.lp2n.institutoptique.fr/en/equipes-de-recherche-du-lp2n/light-complex-nanostructures. Tong WU
First thank you for developing these excellent tools!
I am trying to use your models to study the QNMs of polar dielectric structures but I am having some issues getting useful results.
To begin I utilised the spherical resonator model you provided, and updated it with my material dielectric functions. This worked fine and the results were in agreement with Mie theory, as shown here for a sphere above a dielectric substrate:
The problem I am having is that when I want to look at non-spherical resonators I find that I do not get any useful solutions in spectral regions where the resonator has a negative dielectric function. Instead I get a large amount of spectrally close eigenfrequencies which have strong field hotspots localised in a single mesh element. I have tried filleting the edges of my resonator to remove geometrical singularities and improving the mesh quality. I don't think the PML is an issue because calculations proceed successfully for spheres in the environment. An example is given below:
I guess this is a problem with numerical stability in the eigenfrequency solver rather than with your models because the only thing that has changed in the geometry. Did you encounter these kinds of issues when developing the models and if so do you have any tips or tricks to over come them?
I am also not sure why these issues occur for polar resonators. I do not see the same effects for metallic ones, and none of your example models exhibit these effects. Maybe it is because of the stronger material dispersion compared to metals?