Triangular nanoprism plasmonic solar cell (UPDATED)

fdtd
plasmonics

#1

Dear all ,
I am simulating a plasmonic solar cell using gold triangular nanoprisms, the triangle is equilateral with sides= 9nm and prism’s height is 20nm.

Here is my strcture:

I want to calculate Scattering Cross Section, I used the cross section analysis group. + TFSF source on the whole array, I used linear wavelength spacing in the frequency monitor with 600 points (from 400nm to 1000nm wavelength)

From the literature, Only one peak should appear( since the prism’s dimensions are smaller than 100nm) . (Link of paper attached in the end).

In my simulation I have one peak and small peaks at the right tail :

I think this is the problem, the staircase causes many peaks in the simulation, but I can’t get rid of them! (it will increase my simulation time tremendously )

If anyone has simulated triangles before, do they have meshing problem because of staircasing?

Paper’s link:
https://www.osapublishing.org/oe/abstract.cfm?uri=oe-18-3-3035

" For nanoparticles with sizes down to the optical wavelength, the surface plasmon resonances
are clearly dipolar and can be satisfactorily described by the lowest order of the series of spherical
harmonics of Mie theory. For larger particles, typically in the 100 nm range, retardation
effects inside the plasmonic particles lead to the appearance of higher order multipolar plasmon
resonances that show off as new features in the optical spectra "

Thank you :slight_smile:


#2

I was able to use the following file to reproduce the results from figure 1 of the paper:
mie_triangular_prism.fsp (276.5 KB)

The mesh step size I used was 2.5 nm, which is relatively small but still requires a reasonably short simulation time and memory. The mesh step size relative to the total width of the triangle is similar to what it looks like you have based on your index monitor preview plot.

I don’t it should be necessary to go to a much finer mesh, so a few reasons I could think of for why you might be seeing some extra unexpected ripples could be:

  1. Since you are simulating an array of particles instead of a single standalone particle, there may be some extra modes due to coupling between neighbouring periods.
  2. The material fit of the gold may not be good.
  3. There may be some reflection from the PML at longer wavelengths, so you might try increasing the distance between the particles and the PML boundary or increasing the number of PML layers.
  4. The simulation time may be too short for the field to fully decay. In this case you can try increasing the simulation time setting in the FDTD simulation region object.

Let me know if one of these suggestions helps.


#3

Dear nlui,

Thank you for your help

Here is my fit, is it good enough?

I used PML width to be 2000 while the largest wavelength is 800.

The simulation terminates using the auto-shut off , so I think this is not a problem.

Thank you again :slight_smile:


#4

The material fit appears to be fine. Sometimes the results can be very sensitive to the fit so if you want to confirm that the problem is not related to the material fitting, one thing you could do is to run the broadband simulation that uses the material fit, and also run several single frequency simulations over the broadband range. You can then compare the results from the single frequency simulations to the results at those frequencies from the broadband simulation to make sure that they match.

The PML distance should also not be a factor, since it sounds like the distance is larger than the wavelength.

You may want to try simulating the single particle instead of the array to see if there are still unexpected ripples in that case. If you are still seeing any unexpected results, please let me know.