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.
Implementing a two-temperature-model (electronic/lattice) for fast laser-excitation
Posted 2014年7月11日 GMT-4 03:56 Modeling Tools & Definitions, Parameters, Variables, & Functions Version 5.1, Version 5.2 16 Replies
Please login with a confirmed email address before reporting spam
I want to model a pulsed laser that hits a solid, excites electrons which "thermalize" then transferring their energy to the lattice (solid, atoms). I think most people do not use COMSOL for this purpose, but I want to, because I started with classical heat conduction and now want to setup my model in greater detail.
The two-temperature model is basically a model where you define two temperatures for electrons and atoms in a material respectively. This is necessary if you take into account that the electrons react much faster on an external excitation (femtosecond laser pulse [1E-15 s]) than the atoms.
The question is: how to realize this in COMSOL . In principal I would expect to be able to add and couple two heat transfer physics with two different dependent variables Te (electronic) and T (atoms) and custom material parameters (thermal conductivity, source terms). But how do I couple these?
Or would it be easier to go directly for custom PDEs and without predefined physics terminals?
Please help me here.
Best regards,
Henning Hollermann
Please login with a confirmed email address before reporting spam
Hello,
I want to model a pulsed laser that hits a solid, excites electrons which "thermalize" then transferring their energy to the lattice (solid, atoms). I think most people do not use COMSOL for this purpose, but I want to, because I started with classical heat conduction and now want to setup my model in greater detail.
The two-temperature model is basically a model where you define two temperatures for electrons and atoms in a material respectively. This is necessary if you take into account that the electrons react much faster on an external excitation (femtosecond laser pulse [1E-15 s]) than the atoms.
The question is: how to realize this in COMSOL . In principal I would expect to be able to add and couple two heat transfer physics with two different dependent variables Te (electronic) and T (atoms) and custom material parameters (thermal conductivity, source terms). But how do I couple these?
Or would it be easier to go directly for custom PDEs and without predefined physics terminals?
Please help me here.
Best regards,
Henning Hollermann
Dear Sir,
I am also trying similar problem. i found one thesis related this work. he used the phonon coupling factor.
please excuse me if i am wrong.
links: e-collection.library.ethz.ch/view/eth:8438
thanks
kiran
Please login with a confirmed email address before reporting spam
Hi
I am trying to solve a similar problem and I am using PDE (c) interface. Anyway, how did you try to solve it?
Thanks!
Thank you kiran. The thesis of Mr Dold was helpful to me understanding, how to introduce the heat input of the laser. I tried to follow the model of Verburg et al. (link.springer.com/article/10.1007%2Fs00339-013-7668-5), but had serious convergence problems with that.
I had to set the absolute tolerances for each dependend variable withing the time-dependent solver. Here is a little quote from the report I wrote about these issues which might be helpful to you.
However, I would put a large warning sign at this point. Before you try to do such complicated things, ask people how have done this before (with success). COMSOL may not even be the best choice here, since you may have to dig deeper than it is possible with this suite. Additionally, it happend more than once to me, that I got along some error messages, that weren't even my faults, but some internal COMSOL failures. And you could hardly distinguish these two. Be careful.
Meshing, tolerances and convergence
Meshing is very important for finite-element problems, as it influences convergence and the quality of the solution. For smaller time steps you have to chose a finer mesh to prevent unphysical oscillations of the solution in-between the nodes of the mesh, but you should keep the mesh as simple as possible as the computational complexity grows with the number and type of mesh elements.
Convergence is usually defined by the relative and the absolute tolerances. The smaller they are set, the more accurate the solution will probably be, but also the longer the calculation of each time step may take. If this tolerance criterion cannot be reached within a certain number of Newton iterations, the solver throws an error. To reach it anyway while maintaining a physically correct and accurate solution it is very important to set these values correctly. Sometimes it can be very hard to make a good compromise here.
Additionally, for better computational performance, the dependent variables are scaled so that their values are near to one. The tolerances are usually applied to the scaled variables. Although COMSOL tries to guess the scaling values, it fails especially for complex problems with equations of your own, as we deal with here. Therefore it is very important for the smooth convergence of the problem to manually choose the right scaling value for each variable. This is typically the highest order of magnitude you would expect the variable to reach during simulation.
At the end I got it running (although very unstable), but physical meaning...hm...
image-upload.de/file/WN2Ybq/1e5e6d146d.png
Best regards,
Henning
Please login with a confirmed email address before reporting spam
Since the Verburg model yields (at least for me) a very unstable model, I would not recommend that as a start.
The thesis of Dold was already mentioned: dx.doi.org/10.3929/ethz-a-010114797
Also www.comsol.de/model/laser-heat...g-a-self-guided-tutorial-12317 may be useful to you.
You may have a look at the numerous papers there are for laser-induced damage in glass (SiO2), using a sub-bandgap high intense laser, e.g. Sun et al. www.opticsinfobase.org/oe/abstract.cfm?uri=oe-21-7-7858
This thesis also deals with finite element modeling and lasers
urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-121209
If you go into detail, you soon realize, that you HAVE to simplify, since very many parameters are actually temperature dependent : dx.doi.org/10.1103/PhysRevB.77.075133
best regards,
Henning Hollermann
Please login with a confirmed email address before reporting spam
Thanks for your support.
How did you interface the two Coupled PDEs in TTM ?
I am not able to interface these equations.
please help me to solve the problem.
thanks
kiran
Please login with a confirmed email address before reporting spam
Dear Henning,
Thanks for your support.
How did you interface the two Coupled PDEs in TTM ?
I am not able to interface these equations.
please help me to solve the problem.
thanks
kiran
Hello kiran,
In general, you would have to introduce some input energy (intensity of the laser beam I(t)), a heat equation for the electrons (energy source from the laser) and a second heat equation for the lattice (energy source from electron-phonon-coupling) which would give you the lattice temperature. As source term for the second heat equation, you generally use some coupling function, which depends on the electron temperature usually (simple case)
Q_lattice = G*( T_electron - T_lattice ).
with material dependent electron phonon coupling constant G.
best regards,
Henning
Please login with a confirmed email address before reporting spam
Thanks for your support.
I need to simulate the ultra-short pulse laser interaction model. (time step: 100e-15)
can i run this model in laptop or i need a workstation to run the model.
can you suggest me the system configuration to run my model.
thanks
kiran
Please login with a confirmed email address before reporting spam
Why not just try it?
You may experience a very unstable simulation (difficult to reach convergence criteria), but this is not due to calculation power.
Best,
Henning
Please login with a confirmed email address before reporting spam
Thanking you very much
Please login with a confirmed email address before reporting spam
One new thing you may be interested in with version 5.1 (released in April) is the "Local Thermal Non-Equilibrium Multiphysics Interface" available within the Heat Transfer Module. Although it was designed to simulate heat transfer in porous media on the macro scale, where the temperatures in the porous matrix and the fluid are not in equilibrium, it would likely be quite easy to adapt this interface for your modeling needs. This interface solves for two temperature fields, and there is a pre-built coupling between the two.
Best Regards,
Please login with a confirmed email address before reporting spam
Could you share your model with me ?
Thank you for your kind consideration of this request.
Thanks for your help! Email:lin_qing2008@126.com
Best wishes!
Qing
Please login with a confirmed email address before reporting spam
Anybody tried two-temperature-model (electronic/lattice) for fast laser-excitation with 5.1 version?
Please login with a confirmed email address before reporting spam
Anybody worked with "Local Thermal Non-Equilibrium Multiphysics Interface" in 5.1 version? I wish to see the two-temperature-model (electronic/lattice) for fast laser-excitation of a thinfilm.
Please login with a confirmed email address before reporting spam
Thank you
Please login with a confirmed email address before reporting spam
Hi, I was also trying to implement TTM to predict ablation in thin film. However, one thing I don't understand is how to removel the material in thin films which would definately change the thermal condition during the process. Is LTNE helpful?
Please login with a confirmed email address before reporting spam
Hello Tong,
If you're wanting to model material removal, you may also find this article helpful: https://www.comsol.com/blogs/modeling-thermal-ablation-for-material-removal/
Best Regards,
Please login with a confirmed email address before reporting spam
Hello,
I am trying right now to implement a e-p thermal coupling in a solid throught the LTNE physics. So far I haven't manage to go through since I want to set the coupling between two solids and not a solid and a fluid.
The problem is that when I set up a "solid" node in the "Heat transfer in Fluid" physics, all the domains in "domain selection" of LTNE are set as "non applicable".
I gonna contact Comsol and let you know the answer if I have one.
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
- KNOWLEDGE BASE Solving Models with Step Changes to Loads in Time
- FORUM Semiconductor laser
- BLOG Implementing a Simple Temperature Controller with a Component Coupling
- FORUM Implementing a refractive index profile matrix into comsol
- KNOWLEDGE BASE Solving electromagnetic heating models with high-frequency loads