Dear all,
I recently meet a problem when I use RCWA codes.
In the same structure, it tooks fewer time when using the FDTD solution.
I need set a lot of orders to calcuate the structure which can reach the similar result.
So I have a question that how can I judge the accuracy of the simulation when I use RCWA codes? and How to judge the orders I need?
Thanks
Dear Sai,
this depends mainly on the material properties of your structure and whether you can calculate the Fourier Transform of the structured layer analytically or have to use a numeric algorithm.
In your numerics, you can try to track how much time the different parts need.
A good article discussing some aspects of convergence and the formulation of the necessary Fourier Transform is Liu & Fan (2012)Article S4 : A free electromagnetic solver for layered periodic structures
In my case, when I use a dielectric structure, see e.g.Article Multi-type particle layer improved light trapping for photov...
orArticle TiO2 Self-Assembled, Thin-Walled Nanotube Arrays for Photoni...
I typically achieve convergence for N=(2nG+1)^2 plane waves, where nG=6 and reciprocal vectors -G(nG)...G(0)...+G(nG) lead to H(magnetic field)-matrices of the size 2Nx2N.
However, if the structure is metallic, due to the large contrast in the dielectric permittivity at metal-dielectric interfaces, nG becomes much larger and, hence, the matrices in the program become much larger quickly. Here, I typically use nG=11..15, but check convergence for tests cases up to nG=23.
This was done, e.g. for holey metal films:
Article Spatial dispersion in two-dimensional plasmonic crystals: La...
I hope this helps!
Best regards,
Christin
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Optics