Discussion Forum

For infinite boundary, you can use PML.
Can you upload your mph file?

Hi, Zhe Wang,

The attached is my mph file. I put PML around the simulation domain.

As shown in the simulated results, the simulated TM plane wave is not perfect. The plane wave on the top and bottom boundary is not so good.

Do you have any suggestion?

Many thanks,

Can not open your attached mph file.
The the boundaries you set are not symmetry, so the result should as it is.

Hi, Zhe Wang,

Thanks for your kind reply. The mph file is the version 4.0. If you work with Comsol 3.5a, you can download a package from Comsol webpage which can be used to open version 4.0.

I double check my model. I just put PML around the simulated area. And on the inner edge of the left PML boundary, I put the magnetic field condition(Hz=1). It is not symmetric?

Can you explain how to set the boundary condition as you said?

Many thanks!

I haven't taken look at your model as I don't have 4.0 but there is one word of caution: Do NOT specify your input field by specifying magnetic field in the interior boundary. The magnetic field specification creates a so-called 'hard boundary' whose magnetic field value can NOT change as a result of any superposition of back-scattered waves. What happens instead is that this 'hard-bound' boundary of yours acts as a mirror for any back-reflected/scattered light.

The proper way of introducing incoming wave is to use scattering boundary condition and there specify your magnetic/electric field. In this case, it will be a 'soft-source' meaning it won't automatically repel your backscattered light. However, in Comsol, scattering boundaries can be placed only at the exterior of geometry limiting their applicability for excitation of a soft source. If possible, I would suggest you to use the application mode dedicated to scattering problems, if you are not using already, and use the field of incident wave to define excitation source. This makes it possible to surround your entire geometry with PML and excite a 'soft-source' as well. In 3.5a, we find it in physical properties menu.

All the best!

Hi, Shakeeb,

Thanks very much for your precise explaination. I learned a lot of physics from your post. In my under standing, you can use the field of incident wave to define excitation source which is surrounded by PML, in 3.5a, and do not need to use scattering boundary or field.

Can you upload a simple mph file to explain this? My version is 4.0, which can not see the same memu as we did in 3.5a.
And this new version can read the 3.5a mph files.

Many thanks, Hope can hear from you soon.

First, my apologies for a slight misinformation. In 3.5a, we define incident field in "Physics->Scalar variables" and not "Physics->Properties" as I earlier suggested. I haven't used for any application of my own scattered harmonic mode and that is why I made that slip.

Rest, you got me quite right actually. Just to add to what I said before, Scattering boundary is perfect for absorbing only reflected plane waves so in some cases it might not be the best choice. And besides scattering boundary, you could also used matched boundary and port boundary conditions (again at exterior of geometry) but both of them are suitable to launch eigen modes in waveguides.

Anyhow, I am attaching a tutorial model which computes scattered field from a dielectric object. I don't where in 4.0 the distinction between simple harmonic and scattered harmonic analysis sets in, but in 3.5a it is defined in "Physics->Properties->Field type". At the moment it is set to scattered TE which is why we have the luxury of having incident field whilst the whole geometry is enveloped by PMLs. Otherwise, and this is to the best of my knowledge, if the option of TE/TM/Hybrid waves is selected then we will have to specify one of the exterior boundaries for incoming wave leaving that facet out of the cover of PMLs.

Please feel free to write again if the need be.

Forgot to attach the model. Here it is now