Gaussian source having rectangular cross section


I’m trying to do a simulation which needs illuminating only a part of my structure which has a rectangular shape, meaning if I use the Gaussian source for my simulation, I need the source to have an oval cross section( not circular). Is there any way I can have such thing? Or any other suggestion like using a simple plane wave, but blocking the part of the structure that I don’t want to illuminate, with a strong absorbing material( I don’t know if we have such thing in fdtd) ?

Thank you ,

Hi @nzaraee

Can you please explain what is your experimental setup? Are you blocking a Gaussian beam profile with some other objects?

Yes, there is one example in KB which basically uses a mask material made from metal (Chromium). Please note that diffraction occurs in the boundaries of the etched cross shape.

Another example would be to use PEC material as in the double slit experiment. PEC is an artificial material and provides a 100% reflection at its boundaries. This means that if you have a light that is reflected from your geometry and is incident on PEC, will be reflected back into your simulations.

Gaussian beam can be injected only in the circular shape from built-in Gaussian source. While there are techniques to inject arbitrary shapes of light, it requires some expertise and are generally challenging.




Thank you very much for your response. Unfortunately, I cannot use the
reflecting mask, since my structure is metal and I’m looking at the
farfield scattering of the structure, I can’t have any other reflecting
component in the simulation because it will affect my response.
I’m basically trying to simulate a grating made of gold( limited number of
periods, PML boundary) on a silicon substrate. But this grating is placed
at the top section of a big chip, that’s why I only want to illuminate the
top section of the chip that only includes the grating( I’ve attached the
FDTD file )
I had two questions in this simulation:
1- my setup needs illuminating from two angles simultaneously, meaning I
need two Gaussian source at exactly the same place, but different
illumination angles. Is there any specific setting that I should use?
2 - Is there any way that I can have the oval cross section of the Gaussian
source, so that it only illuminates the grating alone?( from your email, I
understand that it is a complicated setting, But is there a strong
absorbing material( not reflecting) that I can use to cover the rest of my
chip except the grating?)
3- since this is my first time using gaussian source, would you please help
me figure out the proper setting for my setup( scalar or thin lens, waist
radius and distance )

Thank you very much for your help,

NAND2-OR2_Y.fsp (314 KB)

Hi Negin,

Thank you for clarification and sending your simulation file, I have a better understanding of what you are trying to simulate.
As you said, if you are planning to look at farfield profile, adding any other material to your simulation will affect your results. Here is my response for your questions:

  1. No, these two sources will behave independently. You can take a look at this KB example to learn more.

  2. In Lumerical we can export a custom beam profile as explained here. However, the approach for creating a Gaussian source with oval shape will be challenging as you need to set it properly so that it is a focused beam. If this is what you are planning to do, we can work on it together.

I don’t think we can add a material that completely blocks the field and gives no absorption. Every material will give a reflection (unless it is dark matter!), and unfortunately we don’t have artificial material to use in our simulations to do this task.

These are different ways of setting your Gaussian source. For example, if you know the beam waist and offeset you can use scalar approximation, and if you know your lens settings in an experiment, you can use thin lens which you will need your lens NA and distance from focus. Please refer to this page for more information.

If you are only interested in reflection from your gold geometry and a Gaussian distribution of field on gold is not important, I will recommend using TFSF source. We can set it in a way that it covers only the metal section and capture only the scattering light from metal section.

Please let me know what you think and I am glad to discuss it to come up with the best possible solution for your problem.



Thank you for the suggestions. I’ve used the TFSF source when I was
simulating the grating alone. But now that I put the grating in the actual
bigger chip, I wanted the simulations to be more realistic and closer to
the experiment. If I don’t get my desired response using the Gaussian
source( circular cross section), I’ll switch to the TFSF source.
just a quick question in setting up the Gaussian source. if I use the
scalar approximation and want my metal structure to be at the waist of the
beam, is the source settings in the file attached correct?
I basically put the offset equal to the z0 location of the source, and the
beam waist equal to my metal grating whole width.
Thank you for your help,

NAND2-OR2_y_send.fsp (316 KB)

Hi @nzaraee

Regarding setting your Gaussian beam source:

  1. I recommend you to use thin lens rather than scalar approximation. Scalar approximation is more suitable for fast simulation, and for cases where the wavelength of light (630nm here) is comparable to beam waist (800nm) (highly confined field) we recommend to use thin lens. Add the lens NA in dedicated box If you know the lens NA (that you are using in experimental setup) or you can choose an NA of 0.5 to start with.

Advanced note: You can check the fill lens box for initial simulations but bear in mind that this might create ripples and some artificial effect in some cases. Generally in most cases these ripples are not very important but if you wanted to remove them completely, we will need to study how to set lens diameter and beam diameter properly in your case by unchecking the fill lens box.

  1. If you want your beam to be focused on the device, you will need to set distance from focus to a negative value (- 2000nm). Negative value means that your beam is still converging:

Please let me know if you have any other questions and I am glad to be of help.



Thank you for your response. I’ll use the thin lens settings. but, how does
the NA relate to the waist radius of the beam at focus? I mean with this
setting how do I know the area that the source is illuminating?

Also, in case I want to use the TFSF source instead to illuminate just the
top part of my chip, does it matter if there are reflecting material( gold)
passing through the boundary of the source as long as the reference line(
top right corner) edge is only passing through substrate? ( because some of
the bars in the gating are connected to longer metallic structures that are
not part of the grating and I only want to illuminate the grating, so
unavoidably those connecting metals will pass through some boundaries of
the source).

thanks again

Dear @nzaraee

I expect that they will follow the theoretical value as the Wikipedia link (that I provided earlier in my comments) (waist radius=lambda/(NA*pi)). However, if you want to study it, you can prepare a test simulations with a Gaussian beam and monitors (such as Frequency domain field and profile and movie monitors).

TFST source uses two simulations as is explained in this page. You need to make sure that TFSF planes never cut through your scattering object and all sides of the TFSF source must ‘see’ the same refractive index profile along the direction of propagation. For more information regarding how to set it up correctly please refer to this page. Also bare in mind that you need to take extra steps as your light is injected in an angle (Non-normal angles of incidence).

If you expect bars to cause scattering, I think you need to cut them just before the edge of the side-planes of TFSF source. Otherwise if you want to completely ignore their effect, you can remove them from your simulations. I believe cutting them will give more precise results of what you actually expect in your simulations.