gix_d
Newbie
Hi,
I am trying to simulate the radiation absorption in a thin-film patch on my wafer. I modeled it as a rectangle finite conductivity (boundary, solved both sides, DC thickness given) patch and calculated the integrated loss. I am most interested in 100s GHz to a few THz. The result I got matches very well to the theoretical calculation (modified Hagen-Rubens equation) up to the frequency whose wavelength is equal to the size of the rectangle, and then the absorption/loss spectrum becomes very noisy and nonsensical. I can adjust the size of the patch to prove the correspondence of the starting frequency of the artifact and the size of the patch.
I later found in HFSS documentation that, to avoid unphysical results, one side of the finite conductivity boundary needs to be smaller than the wavelength of interest. I don't quite understand this limitation. What's the point if a finite element simulation can't mesh the model and propagate the solution to a larger-than-wavelenght model? Does anyone know the reason of the artifact and any solution to get around it? I am playing with the meshing options now. While it does significantly change the artifact, I haven't found a good configuration to remove it.
Thanks in advance
- Yen-Yung
I am trying to simulate the radiation absorption in a thin-film patch on my wafer. I modeled it as a rectangle finite conductivity (boundary, solved both sides, DC thickness given) patch and calculated the integrated loss. I am most interested in 100s GHz to a few THz. The result I got matches very well to the theoretical calculation (modified Hagen-Rubens equation) up to the frequency whose wavelength is equal to the size of the rectangle, and then the absorption/loss spectrum becomes very noisy and nonsensical. I can adjust the size of the patch to prove the correspondence of the starting frequency of the artifact and the size of the patch.
I later found in HFSS documentation that, to avoid unphysical results, one side of the finite conductivity boundary needs to be smaller than the wavelength of interest. I don't quite understand this limitation. What's the point if a finite element simulation can't mesh the model and propagate the solution to a larger-than-wavelenght model? Does anyone know the reason of the artifact and any solution to get around it? I am playing with the meshing options now. While it does significantly change the artifact, I haven't found a good configuration to remove it.
Thanks in advance
- Yen-Yung