Menu

Discussion Forum

New Discussion
Filter

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 The information provided may be out of date.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.

membrane simulation problem

Posted Feb 27, 2012, 4:11 a.m. EST Studies & Solvers Version 3.5a 12 Replies

Jianan Wang
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Greetings. I'm a technical research currently using the COMSOL software, and I'm confronting a very confusing point for this software simulation ability.
My question is: It is said that if the different layer of structure has a thickness contrast bigger than 1000 times, the results would be very bad and couldn't be considered. I just wonder whether this idea is right or not. Since my research field mainly lies in the brand-new nano-scale membrane, I just wonder whether this kind of simulation would be effective?
12 Replies Last Post Feb 28, 2012, 3:41 a.m. EST
Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Feb 27, 2012, 4:37 a.m. EST
Hi

high aspect ratio are delicate to simulate as bulk's as you need many elements across the thickness. but on the other hand, such "membranes" can also be considered as shells, with that thickness ratio, as what is happening in the thikness, compared to the surface is of different scales to. But the approach to choose depends strongly on what you are after: shapes of deflection or detailed stress in the membrane thickness

--
Good luck
Ivar
Hi high aspect ratio are delicate to simulate as bulk's as you need many elements across the thickness. but on the other hand, such "membranes" can also be considered as shells, with that thickness ratio, as what is happening in the thikness, compared to the surface is of different scales to. But the approach to choose depends strongly on what you are after: shapes of deflection or detailed stress in the membrane thickness -- Good luck Ivar
Jianan Wang
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Feb 27, 2012, 5:18 a.m. EST

Hi

high aspect ratio are delicate to simulate as bulk's as you need many elements across the thickness. but on the other hand, such "membranes" can also be considered as shells, with that thickness ratio, as what is happening in the thikness, compared to the surface is of different scales to. But the approach to choose depends strongly on what you are after: shapes of deflection or detailed stress in the membrane thickness

--
Good luck
Ivar


Just as you have said, the simulation is quite undesirable, because of the error I have confronted during the simulation as follows:
Error: 4100
Failed to generate mesh for subdomain.
- Subdomain: 1
Internal error in advancing front mesher.
Thanks for your response, and I just wonder if there is a method to simulate that kind of material? or some other way I could carry on my meshing simulation, while on the same time maintain the material uniqueness, which means I don't want to treat it as a shell, without thickness.
[QUOTE] Hi high aspect ratio are delicate to simulate as bulk's as you need many elements across the thickness. but on the other hand, such "membranes" can also be considered as shells, with that thickness ratio, as what is happening in the thikness, compared to the surface is of different scales to. But the approach to choose depends strongly on what you are after: shapes of deflection or detailed stress in the membrane thickness -- Good luck Ivar [/QUOTE] Just as you have said, the simulation is quite undesirable, because of the error I have confronted during the simulation as follows: Error: 4100 Failed to generate mesh for subdomain. - Subdomain: 1 Internal error in advancing front mesher. Thanks for your response, and I just wonder if there is a method to simulate that kind of material? or some other way I could carry on my meshing simulation, while on the same time maintain the material uniqueness, which means I don't want to treat it as a shell, without thickness.
Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Feb 27, 2012, 5:59 a.m. EST
Hi

yes I do regularly up to 1:10'000 in aspect ratios, but you need >10GB ram and a very fine mesh, I mostly use mapped mesh And I work often in 2D or 2D axi as far as possible since this allows me to reduce the model size and get replies in a reasonable time ;)

but I'm in 4.2a there meshing is easier, also as meshing strategies are saved

--
Good luck
Ivar
Hi yes I do regularly up to 1:10'000 in aspect ratios, but you need >10GB ram and a very fine mesh, I mostly use mapped mesh And I work often in 2D or 2D axi as far as possible since this allows me to reduce the model size and get replies in a reasonable time ;) but I'm in 4.2a there meshing is easier, also as meshing strategies are saved -- Good luck Ivar
Jianan Wang
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Feb 27, 2012, 6:29 a.m. EST
Thank you for your response, Ivar.

What i just simulate now is the SAW device simulation with a ultra-thin electrode, nearly 1nm, comparing to the 130um substrate, like Lithium Niobate, and I wonder this kind of simulation is accurate, even if I successfully manage the simulation process. Therefore, this is not the matter of whether I could do it or not, it is a matter of whether I need to do that.

Looking forward to hearing from you!
Thank you for your response, Ivar. What i just simulate now is the SAW device simulation with a ultra-thin electrode, nearly 1nm, comparing to the 130um substrate, like Lithium Niobate, and I wonder this kind of simulation is accurate, even if I successfully manage the simulation process. Therefore, this is not the matter of whether I could do it or not, it is a matter of whether I need to do that. Looking forward to hearing from you!
Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Feb 27, 2012, 8:12 a.m. EST
Hi

normally one uses thin film layers for that, entering the phyiscs of the thickness into the boundary. COMSOl has several options for that, check the doc and the list of BC's

--
Good luck
Ivar
Hi normally one uses thin film layers for that, entering the phyiscs of the thickness into the boundary. COMSOl has several options for that, check the doc and the list of BC's -- Good luck Ivar
Jianan Wang
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Feb 27, 2012, 11:08 a.m. EST

Hi

normally one uses thin film layers for that, entering the phyiscs of the thickness into the boundary. COMSOl has several options for that, check the doc and the list of BC's

--
Good luck
Ivar


Thanks for your sincere and considerate response!

By the way, another question I want to raise is that why most of the relative permittivity of metal is 1, when actually most of the material has various relative permittivity? I just get really confused about that, and couldn't do further simulation with this doubt....
[QUOTE] Hi normally one uses thin film layers for that, entering the phyiscs of the thickness into the boundary. COMSOl has several options for that, check the doc and the list of BC's -- Good luck Ivar [/QUOTE] Thanks for your sincere and considerate response! By the way, another question I want to raise is that why most of the relative permittivity of metal is 1, when actually most of the material has various relative permittivity? I just get really confused about that, and couldn't do further simulation with this doubt....
Jianan Wang
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Feb 27, 2012, 11:15 a.m. EST

Hi

normally one uses thin film layers for that, entering the phyiscs of the thickness into the boundary. COMSOl has several options for that, check the doc and the list of BC's

--
Good luck
Ivar


By the way, for the "thin film layers", is it applicable to the 2D schemes? or only designed for the 3D simulation?
[QUOTE] Hi normally one uses thin film layers for that, entering the phyiscs of the thickness into the boundary. COMSOl has several options for that, check the doc and the list of BC's -- Good luck Ivar [/QUOTE] By the way, for the "thin film layers", is it applicable to the 2D schemes? or only designed for the 3D simulation?
Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Feb 28, 2012, 1:05 a.m. EST
Hi

my word, always check the material properties, these are often defined for a given domain (frequency domain for the permittivity), but if you are in the RF or optical domain then its no the same as in the DC or low frequency ACDC domain, but that your material does not know.

Normally the thin films are there for 3D and 2D, but only some where in 3.5 and I do not believe all were defined in 4.0 they came later and are still being updated the way I understand it

--
Good luck
Ivar
Hi my word, always check the material properties, these are often defined for a given domain (frequency domain for the permittivity), but if you are in the RF or optical domain then its no the same as in the DC or low frequency ACDC domain, but that your material does not know. Normally the thin films are there for 3D and 2D, but only some where in 3.5 and I do not believe all were defined in 4.0 they came later and are still being updated the way I understand it -- Good luck Ivar
Jianan Wang
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Feb 28, 2012, 1:44 a.m. EST

Hi

my word, always check the material properties, these are often defined for a given domain (frequency domain for the permittivity), but if you are in the RF or optical domain then its no the same as in the DC or low frequency ACDC domain, but that your material does not know.

Normally the thin films are there for 3D and 2D, but only some where in 3.5 and I do not believe all were defined in 4.0 they came later and are still being updated the way I understand it

--
Good luck
Ivar


Thanks for your sincere response for my inquiry.

I'm currently engaged in the different electrode material simulation of SAW sensor device. However, as I have said in the previous question, the permittivity of gold, Aluminum and copper are all defaulted as 1, whereas from the data of several paper, the value of relative permittivity of those metals in the normal condition, like tempreture, are varied from 1-10. I just wonder why they are all defaulted as 1 for different material. Does it not really matter the final simulation?

As for the thin film simulation, I'm now exchange the model of electrode from a rectangle area of in the 2D scheme into a line, and set the boundary conditions. In this means, the mesh problem is solved. However, I couldn't apply the special material condition into the simulation. I just wonder how to apply the material information in this scheme.

Thanks again for helping me so much!

Best regards.
[QUOTE] Hi my word, always check the material properties, these are often defined for a given domain (frequency domain for the permittivity), but if you are in the RF or optical domain then its no the same as in the DC or low frequency ACDC domain, but that your material does not know. Normally the thin films are there for 3D and 2D, but only some where in 3.5 and I do not believe all were defined in 4.0 they came later and are still being updated the way I understand it -- Good luck Ivar [/QUOTE] Thanks for your sincere response for my inquiry. I'm currently engaged in the different electrode material simulation of SAW sensor device. However, as I have said in the previous question, the permittivity of gold, Aluminum and copper are all defaulted as 1, whereas from the data of several paper, the value of relative permittivity of those metals in the normal condition, like tempreture, are varied from 1-10. I just wonder why they are all defaulted as 1 for different material. Does it not really matter the final simulation? As for the thin film simulation, I'm now exchange the model of electrode from a rectangle area of in the 2D scheme into a line, and set the boundary conditions. In this means, the mesh problem is solved. However, I couldn't apply the special material condition into the simulation. I just wonder how to apply the material information in this scheme. Thanks again for helping me so much! Best regards.
Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Feb 28, 2012, 3:26 a.m. EST
Hi

the epsilonr issue is that you are in the RF and optical domain, I understand that there epsilonr is complex and quite different from "1", and is dependent on the frequency. But most materials have been defined for steady state fields ACDC not RF and there using epsilonr = 1 is often a good approximation

Also be sure you use the same sign convention if you operate inreal/complex er or real complex "n" index, check the doc carefully

This is my explanation for the difference,
hence always enter or at least check all material data each time;)

--
Good luck
Ivar
Hi the epsilonr issue is that you are in the RF and optical domain, I understand that there epsilonr is complex and quite different from "1", and is dependent on the frequency. But most materials have been defined for steady state fields ACDC not RF and there using epsilonr = 1 is often a good approximation Also be sure you use the same sign convention if you operate inreal/complex er or real complex "n" index, check the doc carefully This is my explanation for the difference, hence always enter or at least check all material data each time;) -- Good luck Ivar
Jianan Wang
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Feb 28, 2012, 3:40 a.m. EST

Hi

the epsilonr issue is that you are in the RF and optical domain, I understand that there epsilonr is complex and quite different from "1", and is dependent on the frequency. But most materials have been defined for steady state fields ACDC not RF and there using epsilonr = 1 is often a good approximation

Also be sure you use the same sign convention if you operate inreal/complex er or real complex "n" index, check the doc carefully

This is my explanation for the difference,
hence always enter or at least check all material data each time;)

--
Good luck
Ivar


Thanks for your insight about this problem, and for model simplicity I would just use the default value, and if I have time, I would try the complex ones~

Besides, I'm just very curious about the modeling of thin film, like the width of merely several nano-meters. I really want to know exactly if there are some techniques for 2D scheme simulation of SAW sensor model(COMSOL provided) to achieve the thickness like that. If there is no other way but increase the ram, then I could focus more on other issues that I'm engaged in.

The simple model is attached in the question, and look forward to your response~!

Best regards.
[QUOTE] Hi the epsilonr issue is that you are in the RF and optical domain, I understand that there epsilonr is complex and quite different from "1", and is dependent on the frequency. But most materials have been defined for steady state fields ACDC not RF and there using epsilonr = 1 is often a good approximation Also be sure you use the same sign convention if you operate inreal/complex er or real complex "n" index, check the doc carefully This is my explanation for the difference, hence always enter or at least check all material data each time;) -- Good luck Ivar [/QUOTE] Thanks for your insight about this problem, and for model simplicity I would just use the default value, and if I have time, I would try the complex ones~ Besides, I'm just very curious about the modeling of thin film, like the width of merely several nano-meters. I really want to know exactly if there are some techniques for 2D scheme simulation of SAW sensor model(COMSOL provided) to achieve the thickness like that. If there is no other way but increase the ram, then I could focus more on other issues that I'm engaged in. The simple model is attached in the question, and look forward to your response~! Best regards.
Jianan Wang
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Feb 28, 2012, 3:41 a.m. EST

Hi

the epsilonr issue is that you are in the RF and optical domain, I understand that there epsilonr is complex and quite different from "1", and is dependent on the frequency. But most materials have been defined for steady state fields ACDC not RF and there using epsilonr = 1 is often a good approximation

Also be sure you use the same sign convention if you operate inreal/complex er or real complex "n" index, check the doc carefully

This is my explanation for the difference,
hence always enter or at least check all material data each time;)

--
Good luck
Ivar


There is something wrong with uploading, sorry....
[QUOTE] Hi the epsilonr issue is that you are in the RF and optical domain, I understand that there epsilonr is complex and quite different from "1", and is dependent on the frequency. But most materials have been defined for steady state fields ACDC not RF and there using epsilonr = 1 is often a good approximation Also be sure you use the same sign convention if you operate inreal/complex er or real complex "n" index, check the doc carefully This is my explanation for the difference, hence always enter or at least check all material data each time;) -- Good luck Ivar [/QUOTE] There is something wrong with uploading, sorry....

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.

Suggested Content