HFSS getting rid of mesh inside conductor ("perfect conductor")

[spark449062]

Hi

I am trying to simulate a parallel plate capacitor with a very thin (~9nm) dielectric.

The shape of the parallel plate capacitor is circular.

The conductors ("perfect conductor" material is assigned to them and "PerfE" boundary conditions are assigned to them) forming the capacitor both have ~1.4um in radius and are 100nm thin.

The dielectric in between has the same radius 1.4um and 9nm thin.

(1) as mesh density is increased by assigning mesh operation on the dielectric, the number of tetrahedra grows very big (several million). Furthermore, the mesh density is pretty high in the conductors forming the capacitor as well. I want to reallocate the computational resource that are being used to form heavy mesh in the conductors to capture behaviors in some other important regions such as corners of the conductor where electric field tends to be intense.

Is there any way that I let HFSS to not mesh inside the conductor to save some resource? They are assigned "perfect conductor" so field inside them are zero anyway.

(2) what would be a mesh size for dielectric to obtain reliable result?

 

[BigBoss]

You can use 2.5D simulator such as Momentum. (or Sonnet)
HFSS is not very appropriate EM simulator for simple planar devcies due to requiring huge computing resources.
Moment's Method is very efficient EM simulator technique for planar type structures.

 

[PlanarMetamaterials]

Is there any way that I let HFSS to not mesh inside the conductor to save some resource? They are assigned "perfect conductor" so field inside them are zero anyway.

Yes, it won't by default. Ensure the "mesh inside" box is unchecked when you select the object.

(2) what would be a mesh size for dielectric to obtain reliable result?

Probably less that 1M, I would guess

Good luck!

 

[spark449062]

Thank you for your responses!

PlanarMetamaterials, can you guide me where I can find the "mesh inside" checkbox option? I have only found the "solve inside" checkbox from the properties window, which was unchecked by default.

 

[volker@muehlhaus]

I have only found the "solve inside" checkbox from the properties window, which was unchecked by default.

That is the option he meant.

From many years support in EM solvers, I agree to BigBoss: use Sonnet for this, it is specialized in such planar structures and much better (more accurate) for this task than HFSS. Ultra thin planar dielectric is nothing you want to solve using FEM.

 

[PlanarMetamaterials]

PlanarMetamaterials, can you guide me where I can find the "mesh inside" checkbox option? I have only found the "solve inside" checkbox from the properties window, which was unchecked by default.

Yes, my apologies, I did mean to say "solve inside". I would also follow Volker's and BigBoss' advice and try using other software, if possible.

 

[spark449062]

Thank you all for the suggestions. It's good to know the options, but it seems I have to stick with HFSS for the time being.

Even with the "solve inside" box unchecked, the conductor inside is still meshed (checked by a mesh plot on non-model plane cutting through the cross section). Am I doing something wrong?

 

[PlanarMetamaterials]

Is the mesh just joining the two two sides of the conductor, or are there actual nodes inside the volume of the conductor? You can check to see if it was actually solved inside by plotting the fields as well.

 

[volker@muehlhaus]

but it seems I have to stick with HFSS for the time being.

Then in any case, double check the accuracy of your modelling approach with a known testcase (square plate capacitor with correction for fringing fields).