Discussion Forum

Sounds like you are generally on the right track, but details matter. Are you using ports to drive your dipoles? Some types of ports support setting the phase explicitly (e.g., you could set one port to 0 deg and another to 90 deg). I encourage you to post your .mph file to the forum so that others can examine it and offer more specific and helpful suggestions.

Thanks for the quick response.

I want to be able to move the dipole around a more complicated structure, so I thought that the point dipole current source would be the most convenient way to set things up. Here's a 2D version of what I'm trying to do for a simple planar interface. I don't see a way to set up the phase of the dipole current source.

It is possible to set up a point dipole source at an arbitrary location using ports?

Also note I'm plotting sqrt(real(Ex)^2 + real(Ey)^2 +real(Ez)^2) instead of the standard normE in the plot I generate.

Noah, FYI, I have the RF module but not the wave optics module, so I can't directly execute your model. Maybe another reader here can help you better? Meanwhile, if you haven't already found it, I noticed that there is a Comsol article about doing this kind of model at: https://www.comsol.com/blogs/modeling-surface-plasmon-polaritons-in-comsol/ . Good luck. Added later: As noted in the article I just cited, a surface plasmon is basically only a TM wave. This makes sense to me from my classical EM perspective, since silver is such a good conductor, so it would greatly suppress a tangential electric field component, even at optical frequencies -- after all, it is a great optical mirror. Anyway, in your model, the x component of E at the surface is the tangential component of E, and will (thus) be nearly zero in amplitude. So, you likely can't in practice generate a CP dipole on the surface with ~equal amplitudes of E in x and y. ) Rather, if I am understanding the physics correctly, the plasmon is basically a surface wave involving the electric field component that is normal to the conducting surface and does not significantly include tangential components.

I've seen that blog before, but the linked mph files don't seem to include the models that excite the plasmons with the dipole. I'll check again and leave a comment if need be.

What you said is a correct description of the plasmon field itself. However, what I want is to create a CP dipole source, which then couples into the plasmon. You're right in that it would be hard to make a CP plasmon field.

If I simulate a simple vertically polarized dipole current in free space, I get exactly the textbook radiation field I expect. My issue is just in adding a second one 90 degrees out of phase. Is this possible with the frequency domain solver? Should I look into time explicit instead?

Thanks again.

Well, you should certainly be able to setup two dipoles at a reasonable distance from the surface, and then set the phases of the excitations of those dipoles to differ by 90 deg. I have the RF module, but since I don't have the wave optics module, I'm not confident about the details of implementing it in the latter. Perhaps another reader here can offer a more specific suggestion to you.

How would you do it in the RF module? The implemention might be similar. Even with the advanced physics options, I see no way to control the phase of the dipole with the frequency domain physics, and I don't even see any point source option in the time explicit interface. I'm extremely confused.

With the RF module, the "port" boundary condition lets you set the "Mode phase." In 2D, this setting appears for a variety of port sub-types, including: "User defined," "Numeric," "Transverse electromagnetic (TEM)," "Rectangular," and "Periodic." In 3D, it also includes sub-types of "Coaxial," and "Circular." In both 2D and 3D, the "Lumped port" settings (FYI: a lumped port might be a better way to drive a dipole) also offer "Port phase" settings. As I said, I don' t know how this looks in the wave optics module. Even so, and years ago, prior to Comsol specifically implementing "ports" and "lumped ports," it was possible to specify complex quantities as various boundary conditions (bc's) in frequency-domain models. You should be able to specify complex-valued quantities as bc's to achieve your objective. If you can't get sufficient advice on this discussion forum (from someone who has the wave optics module!) then I suggest you contact Comsol Support directly.

I figured out a way to do it using ports. In this 2D case, I set up 2 line segments perpendicular to one another, aligned parallel to the x and y axes. On the x-oriented line segment, I put 2 ports, each with an x-oriented electric field of equal amplitude. The only difference was that on the two ports, the direction of power flow was reversed. This sets up a linear dipole. I repeated with the y-oriented line segment, but changed the orientation of the field to the y direction. On either pair of ports corresponding to a line segement, I changed the phase to +/- pi/2, and got the circular dipole I needed.