FEM Segmentation Rules
I am trying to simulate scattering from a metallic material(i.e Pt) in the Thz range with the FEM. I have uploaded platinum's relative permitivity and loss tangent. The Mesh required for this material is much greater then that of a dielectric material like Silicon/SiO2. I was not able to find any material specific segmentation rules only frequency dependent rules. Are their material specific segmentation rules based on the relative permitivity of a material for the FEM?
Answers
-
Hello pmcardle,
Are you modelling the platinum as a sheet or a volume? If you can model it as a sheet, the meshing requirements for the sheet would be similar to those for PEC. If you model it as a volume (dielectric with equivalent material properties), the meshing requirements with auto meshing could be quite high (the mesh size is determined by the wavelength in the medium for dielectrics, but the equivalent wavelength in the medium will be very small (infinite for PEC). The better way to model this would be to model the bounding region with sheet metal faces.
0 -
I am trying to model the platinum as a thin layer around another dielectric(silicon). I have done this already with the SEP but wanted to get the results with another solution method, to make sure it's right. Any suggestions on how I could accomplish this ?
0 -
Hi
Please share your model or an equivalent model, then we can make suggestions.
0 -
Here is a model that is very similar to what I have. The real issue is the mixing of scales. The length of the antenna is ~15um whereas the area of interest is ~60nm. Near the apex it is necessary to have a pt-refinement at a mesh size much smaller than the layer thickness. In the model attached is how I originally simulated the platinum layer, by modeling the layering as an identical model embedded inside. The inner geometry is silicon and the outer is the ~50nm layer of platinum, I have another model that has a layer of ~15nm layer thickness which I have not included. I used the SEP for the platinum and silicon regions and obtained nearly exactly the same results to that of an PEC. Obviously simulating a PEC is much more ideal but I need to be sure that there are no errors with what I have done, which is why I wanted to confirm with the FEM. Thanks for the assistance.
Cheers,
Patrick
0 -
Hi pmcardle
Model.cfx that you shared above has the Platinum dielectric defined as an active material. An active material has negative losses and therefore “generates” power. This is not supported in FEKO and the simulation terminates with:
ERROR 49031: Active materials are currently not supported
I noticed that you entered your loss tangent values as negative values. Please note that FEKO uses the positive time convention with the time dependency given by e+jωt (and not e-jωt). As such, a lossy dielectric should be defined with positive loss tangent values.
In general, metals in FEKO are defined as a Metallic medium and applied to a (2D) surface rather than a Dielectric medium applied to a finite thickness layer (or Region). When modelling a metallic medium applied to a surface FEKO allows you to enter a thickness value which would then be applied using the skin effect approximation to account for the thickness analytically. (This then does not require meshing very finely “inside” the metal as only a surface is represented.)
You mention an antenna in your description above, but I cannot see an antenna in your model. Could you please clarify and provide some more information on your application? We might then be able to provide more detailed recommendations.
0 -
My apologies the relative permitivity values were supposed to have negative signs. As for my application, I am simulating the scattering from an AFM tip in the THZ frequency range. The cone is a rough approximation of an AFM tip without the cantilever. Since the height of the AFM is on the order of the incident wavelengths, antenna resonances can be seen in the scattered far field. What I am really interested in though is placing a dielectric slab beneath the the apex of the cone (in the updated file). As the cone is brought closer to the dielectric, through a near field mediation, the scattered far field changes thus encoding the local properties from the dielectric. What I would like to prove is that instead of simulating a platinum coated silicon cone, I can simply replace with a PEC. I was attempting to do this without a sample, which is why in the original file there was no dielectric slab. I have done this with the SEP, but I am not sure this is sufficient. I am simulating metals as dielectrics because my meshing requirements near the apex of the cone are such that the skin effect approximation does not work since the mesh is much smaller than the thickness of the coating.
Cheers,
Patrick
0
Altair Employee