Could "fix wall" command be a solution for my simulation?

Dear LAMMPS Users,

I am interested to simulate about annealing process in Si/SiO2/Si pillar.
Now, I’m using Tersoff interatomic potentials and annealing process was conducted with nvt ensemble.
The problem is the SiO2 layer’s O atoms flows out during annealing process and it loses its shape.

Now, I’m considering if ‘fix wall’ command could be a solution for this situation. I want the shape of pillar to be maintained.
But, I need a curved wall around surface of the pillar and it seems like there’s no applicable ‘fix wall’ command to my situation where the Tersoff interatomic potential is used.

Is there something I’m misunderstanding or missing?
Or Could “fix wall” command be a solution for my simulation?


The problem is more likely due to the potential you are using being not suitable for the system you are simulating. Have you checked that the parameters you are using are suitable for your system?

I am only aware of a one SiO2 compatible tersoff potential, and that is only suitable for bulk SiO2 and not Si/SiO2 mixtures or non-bulk systems.

Thanks for the reply
I used Si-O tersoff potential by S. Munetoh, Comp Mat Sci, 39, 334-339 (2007).
It provides parameterization about Si-Si-Si. so I thought it might be suitable for my simulation.
I found some paper about molecular dynamic simulation about Si/SiO2 interfaces with this tersoff potential. (link : Cookie Absent)

Is this approach wrong for Si/SiO2 mixtures?


Whether the potential is suitable would be evident from the paper, NOT from the parameter file itself. The Si-Si-Si entry is required by the logic of how the parameters are read, but whether is it adapted for pure Si in a Si/SiO2 hybrid system would have to be explained by the paper. I would expect that there would be different parameters for Si in bulk-Si and Si in SiO2 because of their significantly different chemical nature due to their different oxidation state.

Besides, this only covers one of the concerns. The bigger issue is that you don’t have a bulk system.
In my experience oxide systems with surfaces require some form of polarization or charge equilibration scheme to handle the polarization and reconstruction of the surface properly.

Thanks for kind explanation
Then, If I simulate a square pillar structure with periodic boundary condition with other appropriate potential, will the problem with the bulk system be solved?


I am not a prophet. If I could predict the result of simulations, then you would not need to do them.

Mind you a “square pillar structure with periodic boundary condition” would not be a “pillar” but just a 2d layer structure.

Since you placed your post in the “LAMMPS Beginners” category, here is one additional piece of advice assuming that you don’t have much experience:

Start with a simpler systems, like a bulk SiO2 cube with periodic boundary and similarly for Si itself and see if you can reproduce properties documented in the publication describing the potential. If that is causing problems, it will be easier to identify and address them. Then move on to a bulk hybrid system and again try to reproduce published simulation results. Only then I would move on to setting up a SiO2 isolated square or bar or sphere or cylinder (rectangular objects are not so likely at this size).

Only after you have been successful with those simulations and gained insight and experience in how well the parameters you are using can represent those systems, can you move on with sufficient confidence to simulate your actual target system.

1 Like

Thank you for all your assistance