Scattering simulation using TFSF source shows more power

I am using the TFSF source to simulate scattering from a random distribution of spheres (r=100nm) at a central wavelength of 600nm. I noticed some anomalies with a power transmission box placed outside the TFSF source (it gave me values>1, which should not be the case) and decided to add power monitors to study the power distribution at various points in the simulations.

In particular, one power monitor was placed just inside the TFSF source, and the other was placed, just outside it. The power measured by the power monitor outside the TFSF source is larger than the power monitor placed inside it. Since the field outside the TFSF source consists of only the scattered field, I expected that it would have less power than within the TFSF source. Could I please find out why the results are as such?

Simulation layout:


Plot of the power inside and outside of the source:

Simulation file: RandomMediaTest_v3.fsp (314.9 KB)


It appears to me that the problem you described may be related to power normalization of TFSF discussed here. This page has some good discussion about why and what suggestions Lumerical can offer. I will suggest you to take a look at it and see if it makes sense to you.

1 Like


Thank you for the reply. I have read the page, and I understand that normalising to source power may be tricky. However, in this case, I am comparing the power from the power monitor placed at the far end of the TFSF source (with respect to the injection plane) versus the power from the power monitor located next to it but just outside the TFSF source. Perhaps I have misunderstood something… isn’t the incident field subtracted from the field outside of the TFSF source? If so, shouldn’t the power measured outside the TFSF source be lower?

I have to admit that it seems logical to think that way. However, with the complication introduced by the TFSF source, the default transmission function becomes less intuitive to interpret. Therefore, we usually do not recommend using the default transmission function in these cases. Indeed, when you think about the experiment, the transmission is also not so intuitive. The fact that the structure is standalone, the scattered power must be almost negligible compared to the input power.

Instead, in a standalone scattering situation, we usually recommend re-normalizing the transmission function with the sourceintensity command. That way, the results returned will be a cross-section which is usually a more meaningful physical quantity.

If you can tell us more what experimental results you are hoping to compare with, I can try to give comment specific to it.

1 Like

I’m trying to calculate the mean free path of the scattering medium. To do this, I require the unscattered component of the field, and I was trying to calculate it by considering how much light has been scattered. Is there a better way to do this?

In addition, could I please understand, in the scenario I have described earlier, where has the normalisation been conducted? I thought that looking at the power (from DFTMonitor), I was comparing actual power, rather than normalised power.

Thank you for your explanation. I did a quick Wikipedia search it looks like you may be able to calculate the mean free path from the scattering cross section. We can output the cross section using a cross section analysis group. This example is a good place to learn how to measure the scatt cross section for a standalone structure. Make sure you have the analysis group outside the TFSF region in order to properly capture scattered light.

As far as the transmission for TFSF source, this is not a recommended approach like I said. My understanding is that Transmission is a concept for periodic arrays. So frankly I did not fully understand its physical or simulation meaning for a standalone structure like yours.