Minimum simulation region size

fdtd

#1

Hi,
I am studying the reflectance of a periodic structure (silver cylinders raised above a gold substrate). The structure is very closed pack (the distance between cylinders is 5 nm) and my simulation wavelength range is (300 – 900 nm). To simulate such a structure, the length and width of FDTD simulation region

Ag-disk-on-Au.fsp (280.2 KB)
should be 65 nm (a 60 nm diameter cylinder with 5 nm distance with the adjacent cylinder) and I am using periodic boundary conditions.

• My main concern is the minimum FDTD simulation box size that gives reliable results knowing that my simulation range is (300 – 900 nm).
• Based on mesh refinement options I am using “Conformal variant 1” to account for the metal-metal interface. Is that the right choice?
• Could you kindly check the simulation setup specifically source distance from the structure and the monitor distance from the source (which is used for measuring reflectance)
Please find the simulation file and the structure setup attached.


#2

Hi, @soroosh!

Why don’t you try using the TFSF source instead ?
(https://kb.lumerical.com/en/ref_sim_obj_sources_tfsf.html
https://kb.lumerical.com/en/ref_sim_obj_tfsf_source_in_periodic_boundaries.html)
It should handle the scatterring problem all right.


#4

Now I understand why you use the plain wave source. You need the reflectivity of the structure as a whole, but not the effect of nanostructuring. Here you are right.


#5

Hello,
I checked your simulation setup.
Some problem found in it.
I changed materials.
I modified Ag-disk-on-Au-mod.fsp (296.1 KB)


#6

As you mentioned I want to measure the reflectance. I have extracted the scattering and extinction cross sections in a different setup. Since along the x- and y-direction the cylinder is too close to the simulation boundary, I have concern regarding the reliability of my measurement.


#7

Many thanks for the modification.


#8

I suggest simulating a small array of cells, say 2-by-1, 2-by-2, etc, with the same mesh size and compare the results.
Your simulation takes fairly short time to try this (at least on my laptop though).


#9

Thank you for the suggestion.