Brief Overview:

We received our license for the COMSOL plasma module around 18th June, and we started working with Dr Amir to simulate our experimental setup in COMSOL.

The objective of this computer simulation is to find out the basic plasma parameters for cold atmospheric plasma jets produced via dielectric barrier discharge, as there is a lack of knowledge in this area. (At the time of this course [late May – Aug 2018], there were scarce resources on plasma formation with the configuration [needle-electrode configuration] that we were adopting in our physical set-up.)

With this knowledge, we can further improve upon our setup and refine our final product.

 

Goal:

For COMSOL simulation, we have identified the basic parameters in which we want to input, which is as follows

Research:

We set out researching the various COMSOL examples available to us from the COMSOL application library, and reading through the research articles.

Here are some examples we identified to work on:

Atmospheric Pressure Corona Discharge in Air

Dielectric Barrier Discharge

Video links:

Simulating Low-Temperature Plasma in COMSOL

Modeling Plasmas in COMSOL

Familiarization:

We first started working on translating the 1D and 2D models we found on COMSOL into 2D-axisymmetric models to see if we could achieve consistent results.

Then, as we worked on the models, we familiarized ourselves with the interface and found out that the formation of plasma requires many input parameters, as shown below:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

For an accurate model, we had to couple the plasma physics with laminar flow.

Identifying problem areas:

As we worked more on COMSOL, we had many instances where the COMSOL solver did not converge and a complete solution was not found.

As such we set out to identify the most important aspects of plasma formation for simulation in COMSOL.

Below are some aspects we identified:

1. Geometry: Sharp tip vs Rounded tip
2. Electron Energy Distribution Function
3. Cross section of gases used(Air/Helium/Argon)
4. Reduced electric field to mean electron energy
5. Domain reactions and surface reactions
6. Initial values
7. Boundary conditions
8. Mesh used

 

Geometry:

We found that using a rounded tip for our high voltage electrode would concentrate the plasma formation towards the tip.

Electron Energy Distribution Function:

This is an important aspect of plasma formation as it determines the reaction rates for electron collision reactions which in turn affects plasma formation.

Cross section of gases used(Air/Helium/Argon)

We found that plasma chemistry for air was extremely complex. An existing COMSOL example using air (the sole one we could find) listed some generalized chemical reactions for air passing through high voltage, most of which being: (a) the loss of an electron from a general air atom giving a positive ion and a free electron (b) the reverse reaction (c) the combining of a negative ion with a positive ion to give an air atom at a metastable state (a stable state that is of higher energy than ground state). When we used these generalized reaction equations for air in our plasma model which we were building up with Dr Amir, we found that a few problems arose during solving which prevented the solution from being found in COMSOL. Iteration problems, problems with computing the Jacobian, etc. Being novices to complex plasma chemistry, we attempted to overcome those problems through making small changes and computing to study how those parameters would affect the convergence of the results.

After reading COMSOL papers and using each different gas in COMSOL, Helium was found to be the most promising of the three gases in modelling plasma since its reaction chemistry is the most simple of all the three. Additionally, certain types of data were found to dramatically affect the ability of the solver to compute results with convergence. We have included more details on this below.

Reduced electric field to mean electron energy:

In the process of resolving divergence in computation for plasma models using air as the gas, we found that the reduced electric field (the electric field divided by the concentration of neutral atoms) and mean electron energy drastically affected convergence and divergence. We attempted to investigate this using best fit curves for reduced electric field against mean electron energy instead of using discrete input values which were not available on the COMSOL application library for generalized Air gas. If the solution converged before, once we used a best fit curve the solution then diverged (ceteris paribus, all other things being equal).

Domain reactions and surface reactions:

The reactions to include and the forward rate constants to input as well as surface reactions were important as to sustain plasma formation.

Initial Values:

The solution was very sensitive to any changes to the values for initial electron density and initial mean electron energy.

Boundary Conditions:

Boundary condition used will affect the plasma formed greatly.

Mesh used:

The convergence is affected when mesh is refined, which is not consistent with general Finite Element method modeling

 

Results:

In the limited time we had COMSOL, we managed to achieve some results.

Below is a screenshot of plasma formation in point in time:

 

This is a 2D-axisymmetric model where the axis is on the left of the model at x= 0mm, and the rounded tip is the high voltage electrode.

However, this model that we came up with had two parameters skewed from our goal:

1. Temperature used in this case was 400K instead of 300K
2. Voltage applied is ac voltage of 750V.
3. Laminar flow was not coupled yet

Furthermore, we were unable to verify whether what we have achieved is indeed an accurate model.

We have been in correspondence with our COMSOL advisor, but have not received any conclusive remarks yet.

Conclusion:

Although we were unable to reach our goal in COMSOL, from this experience, we have managed to understand better the many intricacies of plasma and the physics that goes behind it. As such we were able to further refine our experiments and also our final product for our project.

Given more time, we believe that we should be able to create an accurate model of our product and maybe be able to post it in the COMSOL application library for future reference.