Thin Metal
I have been trying to simulate the scattering of an antenna above a thin layer of gold ~100 nm. I am a plane wave source, the plane wave is focused using a mirror. After the mirror, the focused plane wave is used to illuminate the antenna and layer of gold. The simulation is using frequencies in the ~ THz, where the skin depth of Au is comparable to the thickness of the layer. I originally tried the multilayered Green's function and received this warning.
WARNING 3489: Too many integration intervals are required for the potentials
See also message in the output file
The results were not consistent with experimental results. I than tried to simplify the problem by using a half plane PEC, and received results that were also inconsisent . Is there a better way to proceed? Should I increase the number of layers in the multilayered Green's function to reduce the number of integration intervals? Is the half plane PEC always reliable? Is there a way to check its reliability for a particular simulation?
Answers
-
Hi Pmcardle
It is unfortunately too difficult to make even general comments without some kind of image or model file.0 -
I have attached an image of the simulation. A plane wave is incident and focused by the mirror onto a conical conducting antenna, which is above a metallic layer. It is relevant to have a light source which converges to a focus, which necessitates the mirror. The antenna is above a thin layer of gold ~100 nm. I am simulating the gold by using the multilayered Green's function. I am calculating the scattered electric field from the antenna, in the far field. I have also performed the same simulation with a half planar PEC. For both I am getting inconsistent results for both the half planar PEC and multilayered green's function. Any suggestions on how to proceed?
<?xml version="1.0" encoding="UTF-8"?>
0 -
The warning about too many integration intervals indicates that the calculations behind the Multilayer Green's Functions are not converging. The combination of material, thickness and frequency must be taking this capability outside its validity range.
If the thin gold layer is reflecting, then with the PEC approximation you should see results that are not very different from the expected results. If the simulation results with PEC are totally different, then something else isn't right. There could be something with the mesh (approximation of curved surfaces?), the frequency (note the frequency you enter is in Hertz, not MHz or GHz), connections between objects (is the connection between the conical object and the 'ground' similar to what you have in reality? One point touching may not be enough). To investigate, plot currents on the objects and fields between the objects. you might notice where the results start to deviate from expectations.0 -
Other thoughts:
1. If the circular object represents a parabolic reflector that you're analyzing with curvilinear elements and RL-GO, you might also want to try, for comparison, flat mesh elements. Those would then need to be a lot smaller, and you'd have to choose the best surface approximation under advanced mesh settings.2. You might also want to try, for comparison, a finite metal 'ground' object instead of the infinite structure.
3. The concept of scattered field can sometimes trip up a user. In the laboratory, you always measure total field (and the extent of the incident field is finite). In a simulation, you have to be careful whether you're displaying total field or scattered field. The scattered field is a mathematical concept: the difference between incident and total field. In a shadow region, to produce small total field, a simulation tool will show a large scattered field (counter-intuitive), out of phase with the incident field (which is everywhere at all times).
0 -
Thank you for the replies
0
