simulating graphene with PEC contacts


i want to simulate a graphene device using lumerical to see the optical response of the different contact shapes. when i try to make the PEC contacts to touch the graphene layer (which is a 2d rectangle) lumerical gives an error indicating that PEC cannot be in the same cell with a material modeled by surface conductivity model. if i use the finite thickness graphene model i think the simulation will need 15 days to complete with 2nm graphene thickness (contact thickness is about 60 nm and the other metal features have 2um thickness, frequency window of the simulation is 0.2THz-1THz)

one more question, are the material fits for gold reliable in the freq. region that i mentioned since there are no experimental data? i attached the picture below. using gold prevents the error occurring but i want to know if there are any possibility of using PEC as contact material.


It is likely that the error message indicating that “PEC cannot be in the same cell with a material modeled by surface conductivity model” is related to a known issue in an older versions of FDTD. I would recommend downloading the latest version of FDTD from Lumerical website and running the simulation again. If you get the message even with the latest version, share your simulation file with us and we can try to reproduce the error and check the detailed settings of your simulation from more information.

Also, if you want, lets us know what you are trying to simulate and what is your simulation setup. We could try to provide additional simulation tips. For example, it might not be necessary to include the contacts in the optical part of your simulation. For example, you can include a link to a reference paper if you based your simulation on a published research.

As for the material data, the general response would be no, the material fit is not reliable outside of the range of available material data. You should try to get coefficients for some material model or sampled data that are valid for your frequency to make sure that the results are reliable. On the other hand, the values in this frequency range are likely so high that it will not have major impact on the results. Then again, you should be able to avoid the error and use PEC.


Dear mbenes, I added the simulation file. My aim is to simulate the magnetic field on the graphene device near the resonator. Since there will be contacts on the fabricated device I want to see the effects of the contact distribution (contacts are for electrical measurements). For my project I’ll study the transport properties of graphene under the time and position dependent magnetic field. We will send terahertz beam into the resonator and there will be a time varying enhanced magnetic field in the vicinity of the resonator. Therefore I need to find the magnetic field distribution on the graphene to investigate the physical properties. Is what I am doing correct? Can I get the correct magnetic field distribution on the graphene with this simulation file? I would be very very glad if you could help me with this. Thanks in advance

My structure is deposited on SiO2/Si substrate. I added a picture with exaggerated details for you to see the structure.

The following picture is for some parameters in the simulation file

Finally the simulation file:

simulation.fsp (283.3 KB)

Hi @kkucuk,

Before discussing the simulation file, one important question is if you need to take into account magnetic properties of graphene. In Lumerical’s optical solvers we model graphene using the surface conductivity from the Kubo formula. Do you need a model that also includes magnetic properties for your application?


Thank you for your reply. I think ignoring magnetic properties in the conductivity model would not be problem. But it would be great for me to learn how to put magnetic properties also into the model.

As I indicated in my previous message I want to obtain the magnetic field distribution on the graphene flakes between the contacts. Can I get reliable results with my simulation file? Your help will be very appreciated.

Hi @kkucuk,

You can use the TFSF source if you want to avoid the edge effects of a finite size plane wave. However, you need to be aware that the usual normalization using the source power will depend on the size of the source as explained here. An alternative would be to use a source that is intrinsically finite in size, such as a Gaussian beam.

I have some additional comments and suggestions:

  • It is probably a good idea to increase the size of the simulation region so that the distance between the structure and the PML is at least half the maximum wavelength. The purpose is to avoid evanescent fields at the PML.
  • Inside the TFSF source it is better to have an uniform mesh in the directions perpendicular to the propagation direction. Since you want a fine mesh in the region where the structure is present, you could reduce the size of the TFSF source.
  • Is the structure symmetric with respect to the x=0 plane? In other words, do you have contacts on both sides?
  • I would try a simulation with a coarser mesh first. At the moment your simulation requires a lot of memory and will probably take a long time to run. It is always good to start with a coarse mesh for checking purposes and then refine it as necessary.

Hope this helps!