Measuring scattering and absorption with TF/SF sources

fdtd
scattering
tfsf

#1

Hello,

I am trying to characterize a single antenna which is excited by a TF/SF source at normal incidence. My objective is to explicitly get the transmission by using the trans_box analysis group that I have modified myself. I am aware of the fact that percentage of the optical power transmitted are not reliable as with other sources as indicated on Scattering simulation using TFSF source shows more power

However, I am concerned with the comparison of transmission on different sides(x normal, y normal, z normal and in different directions) of the transmission cube and I think there will not be any problems without proper normalization. This antenna is a forward scatterer and I want to maximize the transmission at 1550 nm of incident free space wavelength by varying the base width of the triangular antenna.

I proceeded as follows:

  1. I put a modified transmission box with x-normal and y-normal faces residing inside the TF region of the TF/SF source.
  2. I put a modified transmission box with x-normal and y-normal faces residing outside the TF region of the TF/SF source, in other words this time they record the scattered fields.
  3. The z-monitors are in the same place in both cases: the reflection monitor is inside the SF region just before the source injection plane, but I placed the transmission monitor inside the TF region after the beam passes through the substrate and the gold film perforated with the antenna.

I am not sure if my approach is correct and I am pretty confused. In order to clarify matters I am attaching my fsp file to this post. I appreciate your assistance and feedback.

DeltaAntennaSweep.fsp (299.0 KB)


multiple TFSF sources,with different illumination angle
#2

Hi @oarisev14

In the scattering field calculation where we use TFSF, transmission varies with the size of injected source (TFSF source here). Thus we rather use scattering cross section.

In your example since gold is lossy, you need to decrease the size of TFST and trans-box so that their planes/faces are as close as possible to the scatterer. This box will calculate all the scattered light.

If you want to optimzie the geometry for maximum scattered light in forward propagation, you can place a monitor on top and outside the TFSF region (similar to as Top monitor in your simulation file) and make sure that it is large enough to capture all the scattered including the light that is coming at an steep angle.

Please see modified simulation file below and let me know if you have further questions.

DeltaAntennaSweep_BK.fsp (309.1 KB)

Thanks


#3

Thanks for the response, if I understood correctly it is best to use the smallest TFSF source possible since the material is lossy. Are my current simulation results unreliable since they had a large TFSF source and not normalized(I think this norrmalization is not a big issue as it does not fluctuate much with wavelength)?

The analysis box will capture the total scattered light, but in order to quantify the light scattered in +y direction I need to place a transmission monitor with y-normal outside the source region which has a bigger area than the transmission box face so light at grazing angles are also captured, am I correct? I was a little confused that is why I ask this.

Do you think the simulation domain is too big, is it possible to shrink it further without causing complications?

Thanks


#4

Hi @oarisev14

Yes, you are right.In your case, since you have a stand alone scatterer (your system is not periodic), and you want to study the scattered light from a single scatterer, you need to choose a smaller region for the source.

Let me clarify why normalization is a bit challenging here and why we use source intensity rather than source power. The size of the source effects the total power injected into your simulation region. When you increase your source region, the total scattered light remains the same, however your injected power increases. Thus, we need to use a source region around the scatterer.

Yes, you are right.

The best practice is to leave a minimum half the wavelength between the object and PML layer. You can also do some convergence testing: start by a small simulation region and then make it bigger. If by changing the simulation region results remain the same it means that the size of your simulation region is ideal.


#5

Actually I paid attention to all these point, but in my opinion I am still getting unreliable results in these simulations. I have a triangular scatter and I sweep its height along a large range that is from 300 nm to 2000 nm, say in 50 nm increments. This means 18 different files for my sweep since I run them on a cluster without a GUI; I know that the scatterer has to be fully contained in the TF region of the source and it is for all heights. I have two transmission boxes, one inside the TF region and another inside the SF region.

My aim is to maximize the forward scattering of the antenna which corresponds to radiated fields in the +y direction on my simulation geometry. I tried with resizing the TF/SF source along with the inner and outer transmission boxes to achieve more reliable results, but the results were not promising. In the publications it is hinted that these antennas should have a scattering peak around 1550 nm of incident wavelength when the base is 955 nm and the height is 1360 nm.

I am uploading the design file together with the sweep file for your information and comments. Note that I have extended the simulation region a bit to accommodate the large height of the antenna. The last picture which has the largest height range corresponds to the case where the monitors, TF/SF source were displaced along with rescaling of the height. The other two plots are from simulation data of fixed position monitors. The value plotted is Q factor which is scattering cross section over geometric cross section and hence unitless.

Delta900baseDriver.lsf (1.2 KB)

Delta900base.fsp (300.2 KB)


#6

Dear @oarisev14

I couldn’t go over the details of this paper, but I had two questions for you:

  1. In paper, they are studying the results for a periodic array of antennas. Please note that our results might not match with the results of the paper. In your first post you mentioned that you want to run simulations for a single scatterer. Can you please confirm that a single scatterer is what you want to simulate?

  2. Why you want to optimize scattered light in the y-direction? Originally I thought that you are interested in scattered light above the structure (+z direction).

Your simulation file looks good. The only concern is the Surface monitor that need to be located outside the TFSF region if you want to study the scattered field in the z-direction. This holds for scattering along the y-direction (If you want to optimize scattering in +y direction, your monitor needs to be outside the TFSF region).

Thanks


#7

I confirm that before moving on the multiple antennas I want to simulate a single antenna.

I am interested in the SPP modes generated by the antenna on the metal-dielectric boundary that and these triangular antennas are unidirectional(supposed to radiate the most in the +y direction) that is why I optimize according to how much light is scattered in the + y-direction.

I did not understand the placement of the surface monitor: can you say an appropriate position in simulation geometry. Actually I use the transmission boxes for optimization purposes, and the surface monitor is there to just observe the fields on the surface.

Thanks


#8

Dear @oarisev14

Thank you for clarification.

Transmission boxes are constructed out of 6 frequency-domain field and power monitors. If you are interested in scattering only in the +y direction, you can simply use a single monitor outside the TFSFSource region. Please see my simulation file:

Delta900base_BK.fsp (301.8 KB)

Here is the scattered field in the +y direction for device geometry of base 955 nm and height of 1360 nm:

Unfortunately I am not seeing an increase in transmission around 1550, but rather a smooth behavior. I am not seeing any mystery in simulations but I am not convinced yet that these devices will have a maximum scattering at 1550nm in the +y direction.

Please let me know if you have any evidences to prove your claim.

Thanks


#9

Okay thanks for the clarification, I add the other monitors just to compare +y direction scattering with scattering in other directions; hence I will retain them… What do you think about changing monitor positions to accommodate for the change in the height of the scatterer? Will I get more accurate results?


#10

Dear @oarisev14

Yes, for more precise results you can place your monitor as close as you can to the TFSF region. I placed 3 monitors on the +y direction to study your idea. They were located outside the TFSF region and were positioned 50nm, 150nm, and 300nm away from it. The results vary by 5-10% based on the location of your monitors. Here is the plot for your review: