Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.

E-Field inside a metal slab with a spherical void

Please login with a confirmed email address before reporting spam

So, I have two electrodes in my structure. The structure is given in the attached file. Basically, it shows a copper anode (blue) on top and a copper cathode (blue) at the bottom -- at a distance of 200 Angstroms. The bottom cathode has a spherical cavity close to the surface. Cathode is grounded and anode voltage is 10 V, yielding an "average" field of 5e8 V/m. Since the cavity is close to the cathode surface, there is a very thin "covering" of copper.

The sphere inside the copper cathode is of radius 10 angstroms and centered at (0,0,39.22) angstroms. The bottom surface of the cathode is at z = 0 angstroms and the top surface is at 54.22 angstroms. So, the center of sphere (hole) is at a depth of 39.22 A, and has a radius of 10 A ==> the top of the hole is at 49.22A. This leaves 54.22-49.22 = 5 A thin slab above the hole roughly. Process used to create spherical void/hole: I created the copper slab first, then the sphere and used the difference boolean operator to remove the sphere from the copper slab/block. (This is the correct process to create the void I hope?) In case of the space between the cathode and anode, I created another block between these and assigned its material to air.

I find no field in the cathode. I was thinking that since the "cover" above the sphere is so small (Angstrom scale), some E-field penetration could be expected even if the cathode is a metal (Copper).

I was wondering if my use of the built-in model or boundary conditions lead me to E=0 in the metal? Or is it real and the field should indeed be zero?

Also, attaching my model herewith.



3 Replies Last Post 18 sept. 2020, 19:32 UTC−4
Robert Koslover Certified Consultant

Please login with a confirmed email address before reporting spam

Posted: 4 years ago 18 sept. 2020, 16:33 UTC−4
  1. You are using the simple electrostatics model. Copper may not be a perfect conductor, but it does conduct. And thus, those moveable charges can move, and thus you should indeed find E=0 inside the volume of your metal.
  2. If you want to get a different result, use the electric currents (ec) module. This will support a non-zero current in your metal (J = sigma x E). and thus a non-zero E field, if sigma is finite. :-)
-------------------
Scientific Applications & Research Associates (SARA) Inc.
www.comsol.com/partners-consultants/certified-consultants/sara
1. You are using the simple *electrostatics* model. Copper may not be a perfect conductor, but it does conduct. And thus, those moveable charges can move, and thus you should indeed find E=0 inside the volume of your metal. 2. If you want to get a different result, use the electric currents (ec) module. This will support a *non-zero* current in your metal (J = sigma x E). and thus a non-zero E field, if sigma is finite. :-)

Please login with a confirmed email address before reporting spam

Posted: 4 years ago 18 sept. 2020, 17:59 UTC−4

Hello Dr. Koslover!

Thank you very much for the kind response. Your first point does align with what my general idea dictates, i.e. electric field lines won't penetrate a conductive metal surface. The only reason we were expecting a minor penetration was because of the fact that the copper surface right above the void/hole is very thin. (~angstroms)

  • One query about the model that I have created is, the space between the anode and cathode is supposed to be air in our case. In order to do so, I created a block in the space between top copper slab (anode) and bottom copper slab (cathode) and formed a union of it all. Then only for the block that is between anode and cathode (touching them both) I assigned the material as air. Is this a reasonable way of creating a sandwich of anode, air and cathode and then studying the E-Field profiles?

  • Another query is about your second point of using Electric Current (ec) Module. I tried using it as per your suggestion but I have been getting an error "Feature: Stationary Solver 1 (sol1/s1) **Failed to find a solution. Singular matrix. For mesh case 0 there are 87722 void equations (empty rows in matrix) for the variable comp1.V2."

and similarly for the degrees of freedom (empty columns in matrix). Returned solution is not converged. Not all parameter steps returned.

I did manage to solve it by assigning a non-zero value of 1e-100 for sigma of air (was 0 by default). But now in order to compute the study of E-Field profile in this module, could you give me a guideline please?

Thank you again!

Hello Dr. Koslover! Thank you very much for the kind response. Your first point does align with what my general idea dictates, i.e. electric field lines won't penetrate a conductive metal surface. The only reason we were expecting a minor penetration was because of the fact that the copper surface right above the void/hole is very thin. (~angstroms) * One query about the model that I have created is, the space between the anode and cathode is supposed to be air in our case. In order to do so, I created a block in the space between top copper slab (anode) and bottom copper slab (cathode) and formed a union of it all. Then only for the block that is between anode and cathode (touching them both) I assigned the material as air. Is this a reasonable way of creating a sandwich of anode, air and cathode and then studying the E-Field profiles? * Another query is about your second point of using Electric Current (ec) Module. I tried using it as per your suggestion but I have been getting an error **"Feature: Stationary Solver 1 (sol1/s1) **Failed to find a solution. Singular matrix. For mesh case 0 there are 87722 void equations (empty rows in matrix) for the variable comp1.V2."** **and similarly for the degrees of freedom (empty columns in matrix). Returned solution is not converged. Not all parameter steps returned.** I did manage to solve it by assigning a non-zero value of 1e-100 for sigma of air (was 0 by default). But now in order to compute the study of E-Field profile in this module, could you give me a guideline please? Thank you again!

Robert Koslover Certified Consultant

Please login with a confirmed email address before reporting spam

Posted: 4 years ago 18 sept. 2020, 19:32 UTC−4
Updated: 4 years ago 18 sept. 2020, 19:45 UTC−4
  1. Any manner/order in which you can create the geometry that successfully yields the domains, surfaces, edges, points, etc., that you need, is a legitimate one, just as long as you can then assign the physics, define the material properties, boundary conditions, get it to mesh, etc. With time, practice, and experimenting, you'll eventually figure out the tricks and easy ways to build geometries (or you could follow some of the library examples that Comsol provides...)
  2. It doesn't matter at all how thin it is to the code, since the code is solving the classical EM equations. It could be less than one atom thick and, so long as the computer didn't choke on the small numbers, it wouldn't make any difference at all. The code is just doing math and crunching numbers. It has no AI capabilities and it is not a physicist -- all that is your job, i.e., to actually know/understand if the sub-category, or approximation, of physics that you've chosen actually applies to your real-world problem. (If/when the day comes that the code can handle all of that too, then physicists like me will be unemployable, and will have to stand by the road holding signs saying "Will Calculate for Food.") :-).
  3. As you've already noticed, you need to have a finite conductivity in this kind of model. I don't know how super-small a value you can get away with before the software will treat it as equal to zero (which would be bad). I sure wouldn't use a number as small as 1e-100.
-------------------
Scientific Applications & Research Associates (SARA) Inc.
www.comsol.com/partners-consultants/certified-consultants/sara
1. Any manner/order in which you can create the geometry that successfully yields the domains, surfaces, edges, points, etc., that you need, is a legitimate one, just as long as you can then assign the physics, define the material properties, boundary conditions, get it to mesh, etc. With time, practice, and experimenting, you'll eventually figure out the tricks and easy ways to build geometries (or you could follow some of the library examples that Comsol provides...) 2. It *doesn't matter at all* how thin it is *to the code*, since the code is solving the classical EM equations. It could be less than one atom thick and, so long as the computer didn't choke on the small numbers, it wouldn't make any difference at all. *The code is just doing math and crunching numbers.* It has no AI capabilities and it is *not a physicist* -- all that is your job, i.e., to actually know/understand if the sub-category, or approximation, of physics that you've chosen actually applies to your real-world problem. (If/when the day comes that the code can handle all of that too, then physicists like me will be unemployable, and will have to stand by the road holding signs saying "Will Calculate for Food.") :-). 3. As you've already noticed, you need to have a *finite conductivity* in this kind of model. I don't know how super-small a value you can get away with before the software will treat it as *equal* to zero (which would be bad). I sure wouldn't use a number as small as 1e-100.

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.