Discussion Forum

As a last resort, how about rotating your scatterer?

Of course, you can try to understand how the Gaussian beam implemeted in COMSOL works and find out the expressions of the electric fields. Then you can input the analytical expressions as a user defined background field.

As a last resort, how about rotating your scatterer?

Of course, you can try to understand how the Gaussian beam implemeted in COMSOL works and find out the expressions of the electric fields. Then you can input the analytical expressions as a user defined background field.

I input the formula for Gaussian beam using new coordinates in a rotated frame and it works. Now a different problem arises: I want the beam to hit a metal grating at an angle, and I expect part of it to be reflected/diffracted, part of it to couple to SPP, part of it be scattered somewhere. But it seems that in the scattered field formulation in wave optics, the solution it calculates is still 'similar' to the inout background field. As the background field is just a tilted gaussian beam without hitting anything (i.e. no reflection/refraction/scattering, etc, just a beam in free space), the calculated total field is clearly wrong. The beam just passes into the material with some wierd pattern. Is this just a limitation of the method? Or are there some ways to model this more accurately? I tried using a plane wave with periodic boundary condition as the background field as that seemed to work, but I would like to use an actual beam so I can see what happens in the region close to but not directly within the beam.

As a last resort, how about rotating your scatterer?

Of course, you can try to understand how the Gaussian beam implemeted in COMSOL works and find out the expressions of the electric fields. Then you can input the analytical expressions as a user defined background field.

I input the formula for Gaussian beam using new coordinates in a rotated frame and it works. Now a different problem arises: I want the beam to hit a metal grating at an angle, and I expect part of it to be reflected/diffracted, part of it to couple to SPP, part of it be scattered somewhere. But it seems that in the scattered field formulation in wave optics, the solution it calculates is still 'similar' to the inout background field. As the background field is just a tilted gaussian beam without hitting anything (i.e. no reflection/refraction/scattering, etc, just a beam in free space), the calculated total field is clearly wrong. The beam just passes into the material with some wierd pattern. Is this just a limitation of the method? Or are there some ways to model this more accurately? I tried using a plane wave with periodic boundary condition as the background field as that seemed to work, but I would like to use an actual beam so I can see what happens in the region close to but not directly within the beam.

It's great you made the bolique Gaussian beam work! As for the follow-up question, I believe it is hard for anyone to figure out what is wrong without actually seeing your model. It would help if you don't mind posting a model file of your project.

>

It's great you made the bolique Gaussian beam work! As for the follow-up question, I believe it is hard for anyone to figure out what is wrong without actually seeing your model. It would help if you don't mind posting a model file of your project.

Here's the mph file. I didn't include my own Gaussian beam, but the difference between mine and the default one (for vertical or horizontal incidence) is of the order of 1e-10.

When you set n2=1 (no material present), you can see a nice Gaussian beam. When you set n2 to something else (a tilted material is present), then you expect some reflected and refracted wave, but the solution I see is clearly not physical (one refracted beam in roughly the right direction, one refracted beam as if n2 is negative, and no reflected beam).

one refracted beam in roughly

Is the attached figure what you expect?

one refracted beam in roughly

Is the attached figure what you expect?

correct. At least qualitatively it makes more sense to me. When I run the above simulation I posted, I get the 'wrong' E-field in the attached fugure.

one refracted beam in roughly

Is the attached figure what you expect?

correct. At least qualitatively it makes more sense to me. When I run the above simulation I posted, I get the 'wrong' E-field in the attached fugure.

The default direction of incidence is positive y, so the Gaussian beam illuminates from the bottom to the top. That's what you got. The easy fix is just the change of sign for the wavevector.