1D and 2D DC discharge in low pressure gases or mixture of gases

[Malik Tahiyat]

Is it possible to model 1 D and 2D steady DC discharges (not pulsing) in only gases (like air, helium, nitrogen, etc. or  a mixture of gases WITHOUT ANY LIQUID) in a simple axisymmetric geometry in Zapdos? 

Has anyone tried it? If os, I would like to know the starting example that can help me do so. 

[Corey DeChant]

Good evening,

You should be able to model a gas only DC discharge, but I am not sure if anyone has tried (at least there isn't a test file in Zapdos for it).  I would say an easy first step is to modify a  Lymberopolous_rf_discharge input file (the rf discharge with gas only) to have the potential BCs found in the mean_en input file (NeumannCircuitVoltageMoles_KV.). You can copy the test files and modify them in the problem directory. 

Thank you,

Corey DeChant

[Alexander Lindsay]

I’ve definitely modeled 1d argonne discharges and thought we had a test for it. I’ll have to look. If it doesn’t exist we can add it quickly 

On Apr 10, 2020, at 6:20 PM, Corey DeChant <csde...@ncsu.edu> wrote:



--
You received this message because you are subscribed to the Google Groups "zapdos-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to zapdos-users...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/zapdos-users/7a4f9776-4be6-4ea2-ad64-f9e2f9d33c43%40googlegroups.com.

[Shane Keniley]

Malik, 

Yes, you can do pure gas discharges. Liquid interfaces are not necessary.  A very simple example of pure argon (left electrode DC biased, right electrode grounded, 1D) was recently added to Zapdos here: 

https://github.com/shannon-lab/zapdos/blob/devel/tests/automatic_differentiation/ad_argon.i

If you want 2D, as Corey said, a good path forward is starting with the Lymberopolous_rf_discharge file and switching the boundary conditions. 

Putting together a reaction network is admittedly time consuming, but this is true regardless of the software you use. I'd recommend looking through literature for papers that report their reaction networks and using one of those. (Mark Kushner's group at the University of Michigan has several papers with helium-air discharges with and without liquid interfaces, for example). For electron-impact reactions you'll need to look up the cross sections yourself. Lxcat is usually what I use to do this. 

The current method for computing rate coefficients for electron-impact reactions is by pre-calculating them with a Boltzmann solver (either Bolsig+ or bolos work, although the latter does not automatically calculate superelastic reactions like de-excitation). Zapdos assumes that you tabulate rate coefficients as a function of mean electron energy, so run either of those for a range of energy values and save the results into data files.  Additionally, it is especially recommended for DC discharges that you use Townsend coefficients instead of rate coefficients for electron-impact reactions, for reasons explained in Hagelaar's paper on the subject.  (Townsend coefficients are shown in equation 67.) 

On Friday, April 10, 2020 at 11:09:46 PM UTC-5, Alexander Lindsay wrote:

I’ve definitely modeled 1d argonne discharges and thought we had a test for it. I’ll have to look. If it doesn’t exist we can add it quickly 

On Apr 10, 2020, at 6:20 PM, Corey DeChant <csde...@ncsu.edu> wrote:



Good evening,

You should be able to model a gas only DC discharge, but I am not sure if anyone has tried (at least there isn't a test file in Zapdos for it).  I would say an easy first step is to modify a  Lymberopolous_rf_discharge input file (the rf discharge with gas only) to have the potential BCs found in the mean_en input file (NeumannCircuitVoltageMoles_KV.). You can copy the test files and modify them in the problem directory. 

Thank you,

Corey DeChant

--
You received this message because you are subscribed to the Google Groups "zapdos-users" group.

To unsubscribe from this group and stop receiving emails from it, send an email to zapdos...@googlegroups.com.

[Corey DeChant]

Hi, Alex

There are some 1d argon discharge test, but they seem to be using the BC SchottkyEmissionBC or FieldEmissionBC, instead of SecondaryElectronBC (unless I over looked one). Where you talking about these tests or is there supposed to be a gas only version of mean_en?

Thank you,

Corey DeChant

[Corey DeChant]

Sorry, I did over look the "reflections" files. It looks like the test file /reflections/base/Input.i might be what you are looking for, but please correct me if I am wrong. It doesn't have any CRANE coupling, but it seems to be a basic argon discharge. Sorry for the misunderstanding.

[Malik Tahiyat]

Shane,

Thanks for the reply.

1. Is there any code/software that calculates Townsend coefficients from cross-sectional data and mean energy (or E/n)? Bolsig+ probably does that for Maxwellian and 2 term Boltzmann distribution but I l also deal with other types of distribution like 'Druyvesteyn'. For that type of distribution, I created my own Matlab code for calculating Rate coefficients but calculating Townsend coefficients seems to be a tad complicated. Any suggestion (like source, formulae etc.) would be helpful
   
   
2. Can you run Bolsig+ on Ubuntu, or do you use Bolsigminus (console based one)?  If you use the Bolsigminus, can you give me instructions how to run it? I cannot seem to get the idea from the manual.

[Shane Keniley]

Is there any code/software that calculates Townsend coefficients from cross-sectional data and mean energy (or E/n)? Bolsig+ probably does that for Maxwellian and 2 term Boltzmann distribution but I l also deal with other types of distribution like 'Druyvesteyn'. For that type of distribution, I created my own Matlab code for calculating Rate coefficients but calculating Townsend coefficients seems to be a tad complicated. Any suggestion (like source, formulae etc.) would be helpful

 

I'm not sure of any software that does it automatically, but it's not complicated. Hagelaar's paper that I linked to has the formula. Any software that computes the rate coefficients will also be able to compute the electron transport coefficients, and the electric field (or reduced field) is probably used as some kind of input. If your rate coefficient is k, then your townsend coefficient is just: 

where "N" is the target species density. You just take the values of k that your code computes, divide by mobility and electric field, and tabulate the new values. (The target species will be multiplied in by Zapdos during the simulation.)

What we typically do with Zapdos is tabulate the rate coefficients (or townsend coefficients) and electron transport coefficients (mobility, diffusivity) as a function of mean electron energy, so your code will have to be run for a range of electric field/energy values. Bolsig+ has this as an input option. Another possible software is bolos: 

https://github.com/aluque/bolos

which is an open-source python script. It is also quite good; the only reason I don't use it is that it does not calculate superelastic rate coefficients (for example, if you put in a reaction like "e + Ar <-> e + Ar*", the double arrow implies that the reaction goes both ways. Bolsig+ will calculate the reverse reaction based on the cross section of the forward reaction, but bolos cannot.) If you want an example of using Bolos and calculating townsend coefficients, there are some older python scripts that Alex put together here: 

https://github.com/lindsayad/pythonForBolos

The relevant input file is "Argon.py". The townsend coefficients are "alpha", "alphaEl", and "alphaEx" for ionization, elastic, and excitation, respectively. (Note that Alex's townsend coefficients in that example include the gas density N because he divides by the reduced electric field, E/N. You should not do that anymore; you input the target species through the Zapdos input file now, so just divide by mobility and electric field.) If you have your own Matlab script, feel free to use it! You can probably tabulate the townsend coefficients the way Alex did there. If you use Bolsig+ you'll need to tabulate the rate coefficients yourself, write a python/Matlab script to read the data files, convert to townsend coefficients, and tabulate the new values.

(I know, there's a lot to do here. I've always wanted to automatically calculate townsend coefficients directly from rate coefficients in Zapdos to make this a bit easier, but when I tried I ran in to convergence issues. Maybe I did it wrong...I'll give it another shot. I've also always wanted our own Boltzmann solver written in Moose directly so all of this would be done automatically with just the cross section data and list of reactions, but that's beyond the scope of my project unfortunately. Maybe someday!)

Can you run Bolsig+ on Ubuntu, or do you use Bolsigminus (console based one)?  If you use the Bolsigminus, can you give me instructions how to run it? I cannot seem to get the idea from the manual.

I use bolsigminus on Mac and Ubuntu with no issues. I believe the website should give you an example input file somewhere? If so, simply run it as you would any other executable. In the command line: 

./bolsigminus input_file_name.dat

That should be fine. Alternatively if you just run ./bolsigminus without any input file, it should run and prompt you to type in your input file name anyway. The cross section data file should be in the same directory as your input file. Once it runs it should produce a single output file for you.