Note: This discussion is about an older version of the COMSOL Multiphysics® software. The information provided may be out of date.

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.

PML implementation for Drude type of material

Please login with a confirmed email address before reporting spam

Hi,

I recently encountered a problem where I need to terminate a Drude type material (negative epsilon) with a PML region. I tried the built-in PML, but it doesn't seem to work very well and I get reflections at the interface. I am just wondering if there is any special techniques that I need to use when terminating a media with negative epsilon. thanks.

Yan W.

16 Replies Last Post Mar 30, 2011, 8:17 a.m. EDT

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Jan 22, 2011, 1:15 a.m. EST
Try setting width of your PML to lambdaS_xyz where xyz is the suffix of the application mode you are using.
Try setting width of your PML to lambdaS_xyz where xyz is the suffix of the application mode you are using.

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Jan 22, 2011, 1:18 a.m. EST
Just to clarify, I don't mean the actual geometrical width of the object, but in the subdomain settings where you enable PML, there is an option of width. Remove the original thing with the one I mentioned above.
Just to clarify, I don't mean the actual geometrical width of the object, but in the subdomain settings where you enable PML, there is an option of width. Remove the original thing with the one I mentioned above.

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Jan 24, 2011, 10:39 a.m. EST
Hi,

Thanks for replying. I am guessing that'll stretch the coordinate of the PML region and make it one lambda thick, right? That doesn't seem to be working, unfortunately. it did not fix my problem and I am still getting reflections from the PML.

I think the reflection would have something to do with trying to terminate a negative epsilon medium. But I am not sure what I can change to fix that.

Yan
Hi, Thanks for replying. I am guessing that'll stretch the coordinate of the PML region and make it one lambda thick, right? That doesn't seem to be working, unfortunately. it did not fix my problem and I am still getting reflections from the PML. I think the reflection would have something to do with trying to terminate a negative epsilon medium. But I am not sure what I can change to fix that. Yan

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Jan 24, 2011, 10:45 a.m. EST
Hi

For your PML "material properties" are you using "air" or the same as the surrounding material?

You might need to split your PML and adapt to the non PML boundary material to avoid reflexions, if I remember right

--
Good luck
Ivar
Hi For your PML "material properties" are you using "air" or the same as the surrounding material? You might need to split your PML and adapt to the non PML boundary material to avoid reflexions, if I remember right -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Jan 24, 2011, 12:17 p.m. EST
Upload your model. May be we can discuss after having a look at it. I have in the past dealt with the issue of metallic waveguides with negative permittivity extending into PMLs. And reflections did disappear, at least to the point of becoming invisible in the absolute field plot. And that is always as far as I have gone. Numerical effects always make sure that something comes back, however strong or weak.
Upload your model. May be we can discuss after having a look at it. I have in the past dealt with the issue of metallic waveguides with negative permittivity extending into PMLs. And reflections did disappear, at least to the point of becoming invisible in the absolute field plot. And that is always as far as I have gone. Numerical effects always make sure that something comes back, however strong or weak.

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Jan 24, 2011, 1:10 p.m. EST
Hi, thanks a lot for taking a look. I have to reset the model because it was too big. You will have to remesh it before running. I used 'mapped mesh' for domain 1,2,3&9. The rest is just 'free mesh'. To summarize the model, I am looking for the eigenmode of the plasmonic waveguide (it's a leakymode), which is a metallic slot sandwiched in between air and glass. thanks.
Hi, thanks a lot for taking a look. I have to reset the model because it was too big. You will have to remesh it before running. I used 'mapped mesh' for domain 1,2,3&9. The rest is just 'free mesh'. To summarize the model, I am looking for the eigenmode of the plasmonic waveguide (it's a leakymode), which is a metallic slot sandwiched in between air and glass. thanks.


Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Jan 25, 2011, 11:41 a.m. EST
Sorry Yan, I didn't realize you were talking with respect to mode analysis problem. I have dealt with harmonic propagation problems in which metals conveniently protrude into PMLs without trouble. But this case does seem strange and I could not find any meaningful solution despite tweaking with various parameters.
Sorry Yan, I didn't realize you were talking with respect to mode analysis problem. I have dealt with harmonic propagation problems in which metals conveniently protrude into PMLs without trouble. But this case does seem strange and I could not find any meaningful solution despite tweaking with various parameters.

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Jan 25, 2011, 12:20 p.m. EST
Hey, thanks a lot for trying to fix my model. I think despite using eigenmode simulation vs. harmonic propagation, the way that the PML is set up should be the same. It's weird that Comsol help is not responding lately. I think it used to give me a case number before anyone takes a look, but I haven't been getting that automatic reply at all. Does this happen to you lately? I am just curious...
Hey, thanks a lot for trying to fix my model. I think despite using eigenmode simulation vs. harmonic propagation, the way that the PML is set up should be the same. It's weird that Comsol help is not responding lately. I think it used to give me a case number before anyone takes a look, but I haven't been getting that automatic reply at all. Does this happen to you lately? I am just curious...

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Jan 25, 2011, 3:10 p.m. EST
Hi Ivar,

I did not use air as the material property. It's the same as it's non-PML neighbor.

Yan
Hi Ivar, I did not use air as the material property. It's the same as it's non-PML neighbor. Yan

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Jan 26, 2011, 5:25 a.m. EST
Hello,
no changes were made in the way notification messages from COMSOL Support are sent out. Here is how it works:

* If you have no open case tickets, a mail message to support@comsol.com will generate an automatic notification to you with case number included.

* If you already have an open case ticket, a new email to support@comsol.com will be silently appended to that open case. The support engineer who is assigned to the support case, as well as your COMSOL sales rep will get a notification and a copy of your mail.

*If you have more than one open case tickets, you can control which case your email will end up in by using the syntax [Case: NNNN] included in the subject header.

I hope this answers your questions.

Niklas Rom, COMSOL


Hey, thanks a lot for trying to fix my model. I think despite using eigenmode simulation vs. harmonic propagation, the way that the PML is set up should be the same. It's weird that Comsol help is not responding lately. I think it used to give me a case number before anyone takes a look, but I haven't been getting that automatic reply at all. Does this happen to you lately? I am just curious...

Hello, no changes were made in the way notification messages from COMSOL Support are sent out. Here is how it works: * If you have no open case tickets, a mail message to support@comsol.com will generate an automatic notification to you with case number included. * If you already have an open case ticket, a new email to support@comsol.com will be silently appended to that open case. The support engineer who is assigned to the support case, as well as your COMSOL sales rep will get a notification and a copy of your mail. *If you have more than one open case tickets, you can control which case your email will end up in by using the syntax [Case: NNNN] included in the subject header. I hope this answers your questions. Niklas Rom, COMSOL [QUOTE] Hey, thanks a lot for trying to fix my model. I think despite using eigenmode simulation vs. harmonic propagation, the way that the PML is set up should be the same. It's weird that Comsol help is not responding lately. I think it used to give me a case number before anyone takes a look, but I haven't been getting that automatic reply at all. Does this happen to you lately? I am just curious... [/QUOTE]

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Feb 9, 2011, 1:20 a.m. EST
Hi,
I m making a model of metamaterials in COMSOL. In my model PML is used around a glass slab.
Can anyone suggest about the subdomain settings for PML having rank tensor properties??????

Or normal subdomain settings for PML so that glass slab shows reflectivity??

Please help me...
Its urgent for me..
Hi, I m making a model of metamaterials in COMSOL. In my model PML is used around a glass slab. Can anyone suggest about the subdomain settings for PML having rank tensor properties?????? Or normal subdomain settings for PML so that glass slab shows reflectivity?? Please help me... Its urgent for me..

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Feb 9, 2011, 2:09 a.m. EST
Hi

are you saying that you want the PML to be an infinite "metamaterial" ?

I normally always have "air/vacuum" around my models so the PML is air, I have though used one "glass" and then cut up my PML radially (cylindrical symmetry) such to have part of it as an infinite glass slab and the rest air, to minimise any reflection from the transition material to PML. As by beams were roughly radial at PML level the edge and material change in the PML region did not distub me.

I'm not sure how to make a PML fully anisotropic, or if the equations are compatible, one would need to look into the equation view, and make afew tests on simple cases

--
Good luck
Ivar
Hi are you saying that you want the PML to be an infinite "metamaterial" ? I normally always have "air/vacuum" around my models so the PML is air, I have though used one "glass" and then cut up my PML radially (cylindrical symmetry) such to have part of it as an infinite glass slab and the rest air, to minimise any reflection from the transition material to PML. As by beams were roughly radial at PML level the edge and material change in the PML region did not distub me. I'm not sure how to make a PML fully anisotropic, or if the equations are compatible, one would need to look into the equation view, and make afew tests on simple cases -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Feb 9, 2011, 2:26 a.m. EST
Hi,
Thanks for replying.
I have attatched my geomatery file. please saw it.
The entire geometry is made of six parts. R1 is the glass slab, R2-R5 are the PML and R6 is a rectangle whose sub domain properties are those of air.

Now my question is this, what are the subdomain settings for PML, so that it slab shows reflectivity.
Hi, Thanks for replying. I have attatched my geomatery file. please saw it. The entire geometry is made of six parts. R1 is the glass slab, R2-R5 are the PML and R6 is a rectangle whose sub domain properties are those of air. Now my question is this, what are the subdomain settings for PML, so that it slab shows reflectivity.

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Feb 9, 2011, 2:27 a.m. EST
The geomaery file is attached with this.
The geomaery file is attached with this.


Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Feb 9, 2011, 3:38 a.m. EST
Hi

if you have "air" around your PML should also be "air", by the way I mostly use circular PML centered at "0,0" I have noticed these mostly solve better

--
Good luck
Ivar
Hi if you have "air" around your PML should also be "air", by the way I mostly use circular PML centered at "0,0" I have noticed these mostly solve better -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Mar 30, 2011, 8:17 a.m. EDT

are you saying that you want the PML to be an infinite "metamaterial" ?

I normally always have "air/vacuum" around my models so the PML is air, I have though used one "glass" and then cut up my PML radially (cylindrical symmetry) such to have part of it as an infinite glass slab and the rest air, to minimise any reflection from the transition material to PML. As by beams were roughly radial at PML level the edge and material change in the PML region did not distub me.

I'm not sure how to make a PML fully anisotropic, or if the equations are compatible, one would need to look into the equation view, and make afew tests on simple cases

Dear all,

We've also encountered problems with full rank tensor medium PML (in our case, the PML should model a infinite plasma medium), and worked a bit with the support on our case. So, here is what we experienced.

It appears that COMSOL PML works generally fine when dealing with air/vacuum media (as Ivar Kjelberg explained it : metamaterial geometry are often located in the middle of the simulation space, and air-PML are put at the edges).

However, things get more complicated for anisotropic (and/or negative index permittivity) media PML. In some case, it may be interesting to change the sign of the PML equations, for example changing the (1-i) into (1+i). But this may not work in all situations and as far as I know for the moment (v3.5 or v4.1), the COMSOL PML can't deal with all tensorial permittivity (especially when negative index are used).

Best regards
[QUOTE] are you saying that you want the PML to be an infinite "metamaterial" ? I normally always have "air/vacuum" around my models so the PML is air, I have though used one "glass" and then cut up my PML radially (cylindrical symmetry) such to have part of it as an infinite glass slab and the rest air, to minimise any reflection from the transition material to PML. As by beams were roughly radial at PML level the edge and material change in the PML region did not distub me. I'm not sure how to make a PML fully anisotropic, or if the equations are compatible, one would need to look into the equation view, and make afew tests on simple cases [/QUOTE] Dear all, We've also encountered problems with full rank tensor medium PML (in our case, the PML should model a infinite plasma medium), and worked a bit with the support on our case. So, here is what we experienced. It appears that COMSOL PML works generally fine when dealing with air/vacuum media (as Ivar Kjelberg explained it : metamaterial geometry are often located in the middle of the simulation space, and air-PML are put at the edges). However, things get more complicated for anisotropic (and/or negative index permittivity) media PML. In some case, it may be interesting to change the sign of the PML equations, for example changing the (1-i) into (1+i). But this may not work in all situations and as far as I know for the moment (v3.5 or v4.1), the COMSOL PML can't deal with all tensorial permittivity (especially when negative index are used). Best regards

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.