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.
Port Boundary Conditions (Fabry-Perot Resonance between them?)
Posted Mar 29, 2012, 10:36 a.m. EDT Mesh 5 Replies
Please login with a confirmed email address before reporting spam
Good morning Community,
I just posted another query on periodic boundary conditions and after working with some other colleagues we determined we have more problems.
I am doing a simple grating structure in the RF module with air on top and silicon on bottom. The response should only be dependent on the grating area and should not change with thicker air or substrate layers. However, when I change the thickness of the air and the substrate my response changes dramatically. I am looking at the reflection coefficient abs(emw.S11)^2. If the layers are very thin then I have only a couple oscillations, if I increase the distance then I get more oscillations/resonances. I thought it may be because my mesh is insufficient but I have gone to an extremely fine mesh and am still seeing the same response.
Does anybody have any idea why this might be?
Thanks for your time.
I just posted another query on periodic boundary conditions and after working with some other colleagues we determined we have more problems.
I am doing a simple grating structure in the RF module with air on top and silicon on bottom. The response should only be dependent on the grating area and should not change with thicker air or substrate layers. However, when I change the thickness of the air and the substrate my response changes dramatically. I am looking at the reflection coefficient abs(emw.S11)^2. If the layers are very thin then I have only a couple oscillations, if I increase the distance then I get more oscillations/resonances. I thought it may be because my mesh is insufficient but I have gone to an extremely fine mesh and am still seeing the same response.
Does anybody have any idea why this might be?
Thanks for your time.
5 Replies Last Post Mar 11, 2013, 4:11 p.m. EDT
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
First, you need to make sure you define separate BC for each of the existing diffraction orders.
Second, evanescent waves are generally reflected by port BC. For gratings, certain evanescent waves can propagate much farther then the wavelength. When reflected of port BC, such waves cause interference patterns similar to what you described. For me, there is no acceptable solution to this problem. It's hard to predict when this waves appear and increasing size of air to something very large is computationally expensive.
Finally, port BC give completely wrong results in the vicinity of wood's anomalies. Luckily enough, wavelength range where anomalies exist is quite small and you can always exclude this interval. Plasmonic Wire Grating model has this type of anomaly at lower wavelength and reflectivity results are basically not correct in that region.
And one suggestion. In you are not interested in transmission, consider adding PML + scattering BC on the bottom. It eliminates half of the problems :)
Second, evanescent waves are generally reflected by port BC. For gratings, certain evanescent waves can propagate much farther then the wavelength. When reflected of port BC, such waves cause interference patterns similar to what you described. For me, there is no acceptable solution to this problem. It's hard to predict when this waves appear and increasing size of air to something very large is computationally expensive.
Finally, port BC give completely wrong results in the vicinity of wood's anomalies. Luckily enough, wavelength range where anomalies exist is quite small and you can always exclude this interval. Plasmonic Wire Grating model has this type of anomaly at lower wavelength and reflectivity results are basically not correct in that region.
And one suggestion. In you are not interested in transmission, consider adding PML + scattering BC on the bottom. It eliminates half of the problems :)
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Alexander,
Thank you for the thorough response. These are all things I have not considered. I just started working on plasmonic gratings and had no idea the port boundary conditions would be insufficient. For each diffraction order I need a port at the top and the bottom? I will have to look at my diffraction theory to see what I expect here then.
How do you circumvent the interference then? Your post makes it sound like there is no answer to it. Do you use a different software or simulation method?
I will have to explore the PML+scattering boundary condition. I need the transmittance and reflectance but just having a confident answer for the reflectance would help me out.
Thanks again.
Thank you for the thorough response. These are all things I have not considered. I just started working on plasmonic gratings and had no idea the port boundary conditions would be insufficient. For each diffraction order I need a port at the top and the bottom? I will have to look at my diffraction theory to see what I expect here then.
How do you circumvent the interference then? Your post makes it sound like there is no answer to it. Do you use a different software or simulation method?
I will have to explore the PML+scattering boundary condition. I need the transmittance and reflectance but just having a confident answer for the reflectance would help me out.
Thanks again.
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Plasmonic Wire example is a good start. It explains how to set up higher diffraction orders. I'm sure most of your problems will disappear after you take them into account.
If you encounter number 2,3 I described earlier, the best solution I found is to exclude those wavelengths from simulations. The problem with making transparent boundary for evanescent waves is not specific to COMSOL. In general for FEM it's hard to come up with such a thing.
There is a workaround in COMSOL that uses background+scattered field calculations. There is no problem with evanescent waves there. But it has its own much bigger disadvantages, i.e. relatively low accuracy.
If you encounter number 2,3 I described earlier, the best solution I found is to exclude those wavelengths from simulations. The problem with making transparent boundary for evanescent waves is not specific to COMSOL. In general for FEM it's hard to come up with such a thing.
There is a workaround in COMSOL that uses background+scattered field calculations. There is no problem with evanescent waves there. But it has its own much bigger disadvantages, i.e. relatively low accuracy.
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi,
I am trying to model a coupler with grating ,but I have some questions:
1- To get reflection & transmission spectrum I use "dBS21" & " dBS31" ( I have one port as source and two other ports ) .Is it correct or I should use abs(S)^2 ?
2-I don't get good result , should I use PML? If so, how should I determine the width of gratings?
3- I have to use different coupling coefficients (C.Cs) ,but I don't know how I can introduce different C.Cs in comsol?
Thanks,
Sara
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi,
I am trying to model a coupler with grating ,but I have some questions:
1- To get reflection & transmission spectrum I use "dBS21" & " dBS31" ( I have one port as source and two other ports ) .Is it correct or I should use abs(S)^2 ?
2-I don't get good result , should I use PML? If so, how should I determine the width of gratings?
3- I have to use different coupling coefficients (C.Cs) ,but I don't know how I can introduce different C.Cs in comsol?
Thanks,
Sara