I have made a hexagonal solid-core (Silica surrounded by air-holes) photonic crystal fiber. The parameters are:
Pitch: 2 um
diameter of air holes: 0.5 um
number of rings: 5
mesh size: 20 um x 20 um
number of cell: x=180 , y=180
and mesh refinement: Conformal variant 1.
but the dispersion results in the range of (1.2~1.8 um) aren't those which expected and also noticeably change with any fine changing in meshing parameters. what should I do to extract reliable curves? Is there any definite way to get the meshing exact parameters?
If you use a triangular meshing, a good hand-on rule is to use a maximum grid dimension of lambda / (mesh_factor*n_material) , with mesh_factor=10. However, since this dimension imposes a cut-off, your results would probably improve alot sweeping over this grid parameter and extrapolating for the "mesh factor" going to "infinity".
Furthermore, with an adaptive meshing refinement your mesh may change at different lambdas. If you are interested into the dispersion, it would be much better to have a fixed mesh grid. (small variations in the mesh produce small variations into the n_eff, which are nevertheless aplified by the derivative...). I'd suggest to make a single mesh, considering the above criterion at the shorter lambda you are considering, eventually adding an adaptive refining once, and then using this mesh for all your lambda sweep.
From my experience lumerical can give a first view for a structure. But their mesh-dependent numerical technique is pretty memory dependent and the accuracy is questionable. I would recommend RSoft.
- Badges
- Science topic
- Similar topics
- Chemistry
- Nanotechnology
- Nano