Discussion Forum

See the files I included in my May 31, 2016 6:10pm EDT entry in the discussion at: www.comsol.com/community/forums/general/thread/25662 . You can/should replace the time-domain function (which I named "mypulse") that is used to launch the wave with one that corresponds to the actual wave you wish to launch. And of course, you'll need to make the physical problem and mesh much smaller, and also the time steps much smaller, since you are operating at THz frequencies. And of course, you'll also need to add your target object into the problem... Hope that helps.

Thanks Robert for your reply. Honestly, I have seen your post on that. But I got COMSOL 5.1, thus the file was not opening. Do you have any idea how I can open that file if I don't have the same version here. I don't have experience with Java so changing Java of that file won't help me out as I checked another post concerned with this issue.

If you could kindly let me know what I have to do in the time -domain physics and how to set the boundary condition only to launch that kind of wave that would have been convenient for me. Thanks again Robert.

-RA

OK, I'll try to cover enough of the key items for you to more or less recreate my example file.
1. Model wizard --> 3D --> Radio Frequency --> Electromagnetic Waves, Transient (temw)
2. For the Study, choose Time Dependent, then Done.
3. Right-click Global Definitions, choose Functions --> Analytic.
4. Function name: mypulse
Under Definition --> Expression: exp(-2*t/1.0e-9)*sin(2.0*pi*1e9*t)
and Arguments: t
5. Right-click Geometry under Component 1 and choose "Block"
6. Enter block dimensions. Width: 1, Depth: .25 Height: .25
7. Click Build selected.
8. Right-click materials. Choose Air from Built-in. Right-click on it and Add to Component 1.
9. Right-click on Electromagnetic Waves, Transient (temw) and choose Group by Space Dimension.
9a. I also recommend changing to linear, in the Discretization setting within the Settings tab of Electromagnetic Waves, Transient.
10. Right-click on Boundaries, and add a Perfect Magnetic Conductor to the list.
11. Right-click on Boundaries, and add a Scattering Boundary Condition to the list.
12. Repeat step 11 (to create a second Scattering Boundary Condition in the list)
13. Now assign the surfaces as follows:
The two surfaces parallel to the XY plane should be PECs.
The two surfaces parallel to the XZ plane should be PMCs.
The surface parallel to the YZ plane and located at z=1 should be set to a Scattering BC.
The surface parallel to the YZ plane and located at z = 0 will be used to launch the wave. First, assign it to the other (second) Scattering BC.
14. Now, let's set up that (second) scattering BC further, for launching the wave. Under Incident Field, choose "Wave given by E field." For the incident field, enter mypulse(t) for the z component of E0. Leave the other components zero.
15. Mesh the geometry with a mesh that is sufficiently fine. (Free tetrahedral with size max element size = .03 will work ok.)
16. Go to the Study stage. Under Step 1: Time Dependent, go to Study Settings. Enter for Times: range(0,1.0e-10,4.5e-9) .
17. Under Time-Dependent Solver 1, Time Stepping, Steps Taken by Solver, I recommend you choose "strict" instead of "free."
18. Compute. Plot your results at various times. Plotting temw.Ez at various times in a multi-slice plot should be instructive.
19. Modify as you see fit for your application.

Hope that helps. Good luck.

Thank you very much Robert ! I appreciate what you did here. I did this thing in my own way previously but somehow that was taking more than 5 hours to compute but then I have done in the way you suggested, took at max 3 mins to compute. I really appreciate it !

However, if I want to put an antenna in the box then the plane wave doesn't go through the antenna. I couldn't even get the electric field along the direction that was estimated previously. I have attached the Ex Vs t graph. I don't know for which reason the plane wave is not propagating through the whole way to the end of the Air box. Is it because the plane wave doesn't get enough time to come to steady state condition before the antenna interrupted the propagation?

I was wondering what is the problem and how it can be solved ? Please let me know. Thanks in advance !

-Rishad

I can't tell from the figure. I suggest that you post your .mph file. (Clear the solution and mesh first, to make the file small enough in size.)

Thanks a lot for your reply. I have attached the mph file here.

- I did the physics and BC in the model.
- the geometry is micron ranges, however the antenna thickness is in nanometer range.
- I have defined the frequency , f=100THz in this case.


I would like to see if a plane wave of 100 THz, in time domain propagates through the antenna in the air box, how the Ex changes with time in the gap of the two antennas. For that I need to do point graph.

But the problem in this model is the plane wave doesn't propagate through the antenna and it gives a plot that I have attached before.

Please let me know. Thanks a lot !

I have attached the simulation gif file for this model when there is NO antenna in the airbox. i would like to get the gif file when there is an antenna in the airbox.

Well, first of all, based on the file you posted, your mesh is much too coarse. Your mesh elements should be much smaller than the wavelength. And your "antenna" would benefit from having a feed (such as a simple lumped element port, which could be based on a line segment connecting your two triangles), even if you only want to consider it as a receiver.

I also notice that your .mph file did not specify any mesh or a solver. Did you intend to use Comsol's defaults? That's seldom the best idea... Anyway, when I suggested earlier to clear the solution and mesh before posting your .mph file, I did not mean for you to remove your specifications of them, but rather for you to choose "Clear All Solutions" and "Clear All Meshes" under the Study and Mesh tabs, respectively. That way, I would still see how you had specified those (that is, assuming that you specified them at all).

Also, I didn't see anywhere under 'Results" where you specified the point/plot at which you wanted to plot Ex vs. time. It's as if you cleared/deleted all your specifications under results too....?

Hello Robert, thanks for your response. I was just waiting for it. I am answering your question one after one so I hope we don't miss points here.

1. Mesh: I know the mesh is much coarse. Technically, it's a default mesh. I am limited to 4GB memory in my working pc so didn'tintend to do finer mesh to early when I still have to find out how to solve the problem. But thanks a lot for your concern about that.

2. Lumped Port: I didn't think of it previously but what kind of feed can I get from this lumped port. Can you suggest how to use it here in this case ?

3. Clear Solution: I misunderstood you in this case. I thought I would rather not specify anything in the mesh and solution so you can find out the problem here.

4. Point Plot: I have attached another mph file which includes mesh (default) and Results sections as well.


I think, I am having the problem because of the material properties in this model. The permittivity of gold here is a complex number and frequency dependent as well. How can I define frequency-dependent material's complex permittivity in time domain simulation? This is suppose to be the main problem here. Any idea from you would be great !

Thanks Robert ! Kindly, let me know.

1. You do need a finer mesh. You have two planes of symmetry. Consider creating a 1/4th of the problem representation, to help you make better use of your very limited memory. (You will have one PEC plane and one PMC plane of symmetry.) You may also need to go to an iterative solver instead, if the problem is still too computationally large for your computer. Consider getting more memory or a newer computer.
2. Consider plotting log10(temw.normE) for your point plot, so you can see the initial tiny values showing up.
3. For a port, consider putting a line segment connecting your two triangles in the work plane, and extrude that along with the triangles. Then in 3D assign the resulting rectangle there as a lumped port, user-defined, and fill in the details. Use the Comsol help system to guide you on what those details mean. That's how I learned it, so you should be able to also.
4. In your material properties for gold, you'll need to specify a non-zero value of the conductivity (what you called sig_g) if you want to represent, in the time-domain, what would otherwise be a complex quantity in the frequency domain. Do not try to assign a complex number to a material property in a time domain problem, since that would be nonsensical. You can find the relationship between complex dielectric constants and a real dielectric constant with a finite conductivity in any good EM textbook. (And looking that up will help you better understand this topic, so you really should do that.) Choose sig_g accordingly, for your frequency of interest, based on literature you have (or can find) about the known properties of gold at the frequency of interest.

Good luck.

Thanks a lot Robert ! It worked apparently. Still need some more works to do on it.

Regards,
Rishad