Discussion Forum

Can someone please help? Or maybe this is impossible? In COMSOl , in the quantities to plot, Current and Charge, despite its name, I can see only current. Even when I use operator divergence on current j, I still an not plot the charge density, or even the evolution of charge density with respect to time.

Hi
To have better chances to get a reply, try to be more explicit:
i.e.
- which physics are you using ?
- and which solver: in stationary, time dependent or frequency domain or ...

Without this info its really difficult for someone out here to understand what you are doing and how to best reply.

--
Good luck
Ivar

Thank for your reply.

I am using the RF module with Frequency Domain study. My model include a thin layer of metal, about 100 nm thick with the shape of split ring resonator, similar to COMSOl example of Frequency Selected Surface (but the U shape, not circle) and the period and dimension of the split ring now is 50 to 100 um)), on a glass substrate. Electromagnetic wave shines normally incidence on it.

Actually I found the resonance of the structure, and plot the E electric vector inside and out side the metal are ok. Even when I tried to plot the surface charge density using the formula
epsilon0_const*(nx*emw.Ex+ny*emw.Ey+nz*emw.Ez)*n_glass^2
I could draw it.

But when I tried to calculate and plot the charge density inside the metal, I did not see thing which is clear and make physical sense. As I works in Frequency domain, I think charge current density can be calculate via the contiunity equation d(rho)/dt=divj. So this is what I used to calculate it:
rho=1/i/emw.omega *(d(Jix,x)+d(Jiy,y)+d(Jiy,z)).

Best regards

Hi

OK that makes it easier to follow, but still you have a complex model. I would need to do quite some homework to check several issues such as skin depth versus sizes, and true continuity of the equations you are using. But you might have done these too.

so my first two concerns, that must be checked, are:
1) skin depth versus conducting material layer thickness, & related mesh and resolution thereof.
2) true continuity of the FEM derived variables you are deriving further.

Do not forget that FEM in COMSOL uses 1st or 2nd order discretization (depending on your physics and settings), which allows to drive at most twice the dependent variable, and anyhow the second derivative is going to be step wise smooth. To get a better resolution on the derived variables and their higher you must increase the discretization order my one or 2 levels. This will also increase the amount of RAM and time to solve significantly, a price to pay for increased precision.
One of the most common errors when looking for details is to ask for a too high derivative of the dependent variable, that is in fact not resolved by the FEM setting used by default. To see the advanced features of COMSOL, be sure you have checked them in the eye/view/preference lists.

Now this might not be enough to solve your issue, but could be a way on the path

--
Good luck
Ivar

Dear Ivar,

I don:t get this 'To see the advanced features of COMSOL, be sure you have checked them in the eye/view/preference lists'. Can you give some detailed instruction about this.

Also please check the model, I think maybe the reason is the mesh in z direction is only one layer of element, so the div operator give abrupt result? is it the same as you say '1) skin depth versus conducting material layer thickness, & related mesh and resolution thereof'

.