Electrostatics simulation of Periodic structure in Device



Hi all,

I’m new to Device, and I’m currently trying to use Device to simulate the electric field distribution in a periodic structure. I’ve done a test run, but the result doesn’t seem to be right (see the attached file).

The structure is simple, it’s just periodic electrodes on top of an oxide layer. The electrodes are periodic in y direction, and the period is 5 um. I applied +400 V to the electrode on the left, and the electrode on the right is ground (0 V). I want to know the electric filed distribution in both xz plane (side view) and xy plane (top view). I wonder how can I correctly setup the boundary conditions and mesh for this periodic structure?

Thanks a lot for the help!


test_efieldsimulation.ldev (5.8 MB)


Dear @Joyzzz,

Can you elaborate on what seems to be wrong with the field results? What is exactly the result that you’d expect?

Regarding the periodic structure, DEVICE does not have any boundary conditions that can account for the periodic nature of the structure. First you need to consider the expected electrostatic behavior of the device you are trying to simulate. If you expect the field distribution to vary in a quasi 2D manner (not being affected by the adjacent unit cells) then you can only simulate one unit cell in the 2D simulation similar to the setup in your file. However, if you expect otherwise, then you’d need to perform a 3D simulation including a few unit cells but most likely only consider the results for unit cells closer to the center since they are more representative of the unit cells in a periodic structure.
Hope this helps!


Hi @mmahpeykar,

Thanks for the reply!

Can you elaborate on what seems to be wrong with the field results? What is exactly the result that you’d expect?

  1. As you can see in the previous attached picture that visualizes the simulated E-field, there’s a weird E-field enhancement near one side of the Si waveguide. I’m not sure whether it is real, or there’s something wrong in my setting (e.g. boundary conditions or mesh);

  2. The contours of the E-field also don’t seem to be as smooth as I expected, especially in the region between the left electrode and the Si waveguide;

  3. The E-field distribution that I expect should be something like what is shown in the picture attached below, maybe with a little bit distortion near the Si waveguide.expected

I also have some questions about the simulation setting:

  1. Boundary conditions in the simulation region
    I don’t quite understand the differences between the “open” and “closed” boundary, and which one should I choose in this simulation.
    It’s also not quite clear to me whether the “norm length” value will influence my simulation time and results.

  2. Mesh setting
    Is there any way to set up the mesh so that it has different “min edge length” in x and z directions?


Based on the plot provided, you only need to do an electrostatic simulation not a self-consistent charge and potential simulation since your system does not have any charge transport. This can be done from the advanced tab of the CHARGE solver as shown below and also makes your simulation much easier to converge even at high voltages:

However, the plot you provided seems to be more of a vector type plot which specifies the path of the fields. This can’t be provided by the CHARGE solver since it does not report field as a vector quantity and it only reports its magnitude.

Regarding the boundary conditions, see the link below (the last video):
and this page:

Norm length won’t affect the simulation time but may affect the results which depend on the 3rd dimension of the simulation such as the contact currents (which is not applicable to your simulation)

No, this is because the mesh is in the triangular shape (not rectilinear like FDTD) and edge length is forced by other factors such as triangle quality etc.


Hi @mmahpeykar,

Thanks a lot for the help! I was able to get reasonable results by setting the “dc update mode” to “electrostatic”. But in my real device, I have four electrodes, with the two outer electrodes connected to the power supply, while leaving the two middle electrodes floating. As is shown in the following picture:
When I tried to run the simulation, one error showed up, which was: “The program terminated due to an error: A conductor was found with no specified contact properties”. I guess I have to add boundary conditions to the middle two electrodes.
I wonder what boundary conditions should I apply to these electrodes?
Thanks again for the help!


As discussed in the following topic, floating contacts are not allowed in DEVICE. As for your question, it all depends what the purpose of those electrodes is in your device. If they are not participating in charge transport, you might be able to model them using a highly doped semiconductor (which would be equivalent to a conductive material) and then leave them float.


Hi @mmahpeykar,
In my case, the two electrodes in the middle are just two pieces of thin conductors. I want to see how the electric field looks like with the two floating electrodes in the middle. I did what you suggested, set the two electrodes in the middle as highly doped semiconductors, but the result shows that there’s no e-field in between the two floating electrodes, which doesn’t seem right to me.

Here is my simulation file:
test.ldev (6.1 MB)
Do you know what went wrong?


Not sure why you expect to see any field between two floating electrodes if they don’t have any potential applied to them. I think the only way you might get field in between them is through the outer electrodes if you apply a voltage high enough so the field resulting from outer electrodes would be expanded to that area.


Hi @ mmahpeykar,

I think as long as there’s insulating material in the middle, there should be electric field in the middle. It’s like a capacitor. I expect the electric field distribution to be like the picture shown below:
BTW, the voltage difference between the outer two electrodes is 400 V.


To me, this diagram implies that the middle electrodes are biased too not floating. Have you tried biasing the middle electrodes to see what you get?


Hi @mmahpeykar,

If you check the test.ldev file that I uploaded, you’ll know that I didn’t bias the middle two electrodes.
I’ve tried to bias the two middle electrodes (by applying arbitrary electrical potentials to them ), and the E-field looks like this:

The problem is in the real case, I do not know the potential in the middle two electrodes.


This looks closer to what you expect for field distribution but CHARGE solver is not able to simulate the potential induced onto a conductor due to existing electric field if that’s what you are trying to simulate. So you would have to provide the potential for the electrodes yourself.


Hi @mmahpeykar,

I see. Unfortunately, this is what I’m trying to simulate.
Thanks a lot for your help though!

closed #14

This topic was automatically closed 3 days after the last reply. New replies are no longer allowed.