COMSOL band structure of photonic crystal waveguide

[BrillouinPie]

Setup:
I'm trying to make the band structure for a planar photonic crystal with finite thickness, i.e., a quasi-3D problem.

I only want the x-direction band structure. So, I'm using variable floquet periodic BCs for the x-direction boundaries, and 0 degree floquet periodic BCs for the y-direction. The top and bottom parts of the cell have either PEC or PMC BCs.

I'm trying to adapt the method used here: http://www.comsol.com/showroom/documentation/model/798/

This tutorial shows how to a make band diagram for a purely 2D photonic crystal. The main idea in the tutorial is following a particular band while ramping k by force feeding the previous frequency into the current parametric solver. Their method also uses an integration coupling variable with an ODE on the frequency, by normalizing the z-comp of the electric field (they're using 2D > RF Module > In-Plane Wave > TE waves > eigenfrequency).

Problem:
I try to adapt the problem by normalizing the entire electric field using: Ex*conj(Ex) + Ey*conj(Ey) + Ez*conj(Ez). For subdomain ICs I use Ex(to) = Ex, etc., for all parts of the E-field.

When I try to run the parametric solver and ramp k, the solver immediately halts, saying that it doesn't know what Ex, Ey, and Ez are.

Any ideas?

 

[Gokul43201]

Nothing specific to this problem, but an FYI: COMSOL has started a discussion forum (as of only a week ago) where you might be able to get help from active users.

http://www.comsol.com/community/

You should be able to use the forum if you are a registered user with an account. Be sure to read and follow the Guidelines posted at the bottom of the welcome thread.

 

[BrillouinPie]

Oh great. Thanks!

 

[Davidleee]

Dear Sir,

I am a started using COMSOL. And I would like to do the band structure for a planar photonic crystal as well. Can you be so kind to share the document from COMSOL web you got to me?

Thanks in advance!
David

 

[Sah]

I am also trying to do this. I haven't got the arrors you have but the computation time seems to be never ending, but perhaps that's because the mesh is very fine and it's a 3D problem.

Have you made sure you are using a 3D RF module where the dependent variables are Ex, Ey and Ez? Because in the 2D problems the two mode polarisations decouple and there is only a single variable Ez (or Hz). Such decoupling does not occur in 3D and you need a full vectorial treatment (although from your post it seems as though you already knew that).

I'll keep you posted and if I get the thing working I can send you my scripts if you wish.

 

[lostinsocorro]

Can you please load your code so that I may learn how you were able to accomplish this? Thank you.

 

[Sah]

I didn't actually write any code. I just used the GUI and followed the instructions for the 2D case, but changed the geometry I drew. Unfortunately the file, with the mesh and results is too large to upload here.

The most important thing to realize when doing things in 3D is the the mesh on the periodic boundary condition boundary pairs must be identical. This is not done automatically and you must thus mesh one boundary then manually copy the mesh from it to another when there is a periodic pair. Once you have done all the periodic pairs then mesh what's remaining using "Mesh remaining free". If you do not do this your computation will not converge.