Coupling DSMC and PIC

[hamidmoezzi1988]

Hi everyone,

I want to use PICLas for simulating the interaction of ions, electrons, and neutrals in a hot plasma where I want to investigate the interaction of these particles, possible chemical reactions (eg. ion-electron recombination), space charge effects (considering the repulsive electric forces between the ions), and etc. First of all, Is PICLas capable of doing these types of simulations? and second, if yes how can I couple DSMC and PIC (electric and magnetic fields) modules of PICLas? Is there any documentation (publication, user guide, etc.) to start with?

Thank you very much for your time and help. Kind regards, Hamid

[scopplestone]

Hi Hamid,

can you substantiate your use case a litte bit more as your description is very general? In general piclas is able to simulate that what you described.

If you have not yet found the user guide, it is linked on the front page of the piclas github: https://piclas.readthedocs.io/

Furthermore, we have some tutorials that can be found under https://github.com/piclas-framework/piclas/tree/master/tutorials and also a lot of regression test cases under https://github.com/piclas-framework/piclas/tree/master/regressioncheck (the regression test cases are not documented in detail, you can look into the readme.md files for a description of what the test does and the builds.ini should show you the compiler settings).

Kind regards Stephen

[hamidmoezzi1988]

Hi Stephen,

Thank you so much for your reply. I am reading the user guide and learning how the code works from the tutorials. I want to study the flow of argon carrying different ions, perhaps electrons, and other atoms or molecules through an Inductively Coupled Plasma Mass Spectrometry (ICP-MS) interface. There exists a Coulumbic repulsion force between ions or there might be chemical reactions between different species including the ions (e.g. ion-electron or ion-ion recombination). I don't know how much I am correct, but I found that I should use a combination of DSMC and PIC to be able to model this kind of flow and that's why I started to work on PICLas. However, just to make sure, is PICLas able to consider the effects mentioned above?

Thank you. Hamid

[scopplestone]

Hi Hamid,

You can use PIC-DSMC to simulate your setup with piclas, but this will be limited by computational ressources. Depending on your gas/plasma properties, you may be able to consider the neutral gas as an inter background gas. But only if the plasma density is much smaller than your neutral gas density. In this case you can use PIC-MCC, which would be much faster than PIC-DSMC (where neutral gas particles are kinetically modelled). What is the range of your expected plasma density versus the neutral gas density? Chemical reactions between charged particles is an ongoing development in piclas and Coulomb interaction has been investigated in the past, but currently no production-ready model is included.

Kind regards Stephen

[hamidmoezzi1988]

Hi Stephen, I appreciate your clarification. The plasma density is about 0.1 % so, based on your explanation, I should use PIC-DSMC and consider the main gas as background, however, I am confused about the concept of PIC-DSMC. If Coulomb interactions are not considered in piclas, then what is the role of PIC here? I imagine PIC as a tool for applying an external voltage to boundaries or in my case to calculate the attractive or repulsive forces between charged particles. Then, in my opinion, PIC-DSMC and piclas should be able to consider a phenomenon such as space charge effect, am I right?

Also, I searched through the tutorials and the piclas github and found that for simulations in which chemical reactions are included, the parameter "omega" is the same for all species (why?), and second its value is different from what we find in the literature ("omega = omega_bird - 0.5"). What is the philosophy behind it? and how much do these discrepancies affect the results?

Thank you very much Hamid

[scopplestone]

Hi Hamid,

I appreciate your clarification. The plasma density is about 0.1 % so, based on your explanation, I should use PIC-DSMC and consider the main gas as background, however, I am confused about the concept of PIC-DSMC. If Coulomb interactions are not considered in piclas, then what is the role of PIC here? I imagine PIC as a tool for applying an external voltage to boundaries or in my case to calculate the attractive or repulsive forces between charged particles. Then, in my opinion, PIC-DSMC and piclas should be able to consider a phenomenon such as space charge effect, am I right?

I assume you mean the degree of ionization is 1/1000 and I guess that in that case, the PIC-MCC modelling might be applicable. The two fundamental works in the literature that describe PIC and the capabilities and drawback are the following

• Plasma Physics via Computer Simulation (By C.K. Birdsall, A.B Langdon)
• Computer Simulation Using Particles (By R.W Hockney, J.W Eastwood)

PIC started out and is still widely used for the simulation of collisionless plasma flows. Although collisional interaction has been of interest for a long time, the development of such models in the PIC context is still an ongoing task, especially the interaction of charged particles with each other.

In the case of Coulomb collisions, this is the case and the image below shows the general problem in a very simplified manner:

"Although PIC simulations proved to be a powerful tool for studying dusty plasmas, the PIC method has a considerable drawback. The space resolution in the PIC scheme is limited by the size of the grid which is typically of the order of the Debye length (fraction of a millimeter for low temperature plasmas). The size of the dust grains is typically in the micrometer range and thus much smaller than the grid size. In PIC algorithms, the particles are represented by charged clouds of the grid size and are able to penetrate each other [49]. This leads to inaccuracies for inter-particle interaction when the distance becomes smaller than the cell size (see Fig.1). The interaction force is strongly deviating from the Coulomb force for small distances and tends to approach zero as the inter-particle distance decreases."

To read a bit more on the matter, see the original paper https://www.sciencedirect.com/science/article/abs/pii/S0022311507000906 Usually, in PIC models, the smallest spatial scale that is resolved is the Debye length therefore the space charge effects if charged particle species.

Also, I searched through the tutorials and the piclas github and found that for simulations in which chemical reactions are included, the parameter "omega" is the same for all species (why?), and second its value is different from what we find in the literature ("omega = omega_bird - 0.5"). What is the philosophy behind it? and how much do these discrepancies affect the results?

The parameter $omega$ comes from the Variable Hard Sphere (VHS) collision model in the DSMC method, see the groundwork

• Molecular Gas Dynamics and the Direct Simulation of Gas Flows (By Graeme Austin Bird)

and the different values stems from the different definitions.

This, however, does not yield any kind of difference in the simulation results.

Kind regards Stephen

[hamidmoezzi1988]

Hi Stephen,

Thank you very much for your time and patience in explaining the concept. I'll go through the materials you mentioned to learn more about PIC. I started to simulate the main gas (Ar in my case) together with a few species undergoing chemical reactions, but it seems that considering the main gas as the background gas is not applicable in 2D axisymmetric geometries. Am I right?

Thank you. Kind regards, Hamid

[scopplestone]

Hi Hamid,

Thank you very much for your time and patience in explaining the concept. I'll go through the materials you mentioned to learn more about PIC.

no problem, I hope that I could help you in some way. I will close this issue for now then.

I started to simulate the main gas (Ar in my case) together with a few species undergoing chemical reactions, but it seems that considering the main gas as the background gas is not applicable in 2D axisymmetric geometries. Am I right?

Actually, I am currently simulating an Ar plasma configuration with 2D axisymmetric geometries and background gas, but this model is still under development and therefore not published yet.

Kind regards Stephen