AIREBO force field for graphene

LAMMPS LAMMPS Mailing List Mirror
1

Dear Axel,

I want to use AIREBO potential in Lammps to model graphene sheets in my simulation setup. May you please let me know if that is required to define bond, angle and dihedral styles and coefficients for carbon atoms of graphene sheets or those are defined in AIREBO potential by default?

Thank you.

Best.

2

Hi
No you should not define bonds with AIREBO, REBO and reaxFF

3

Hi Pascal,

Thank you. What about angle and dihedral?

Also is that correct to define pair coefficient between carbon atoms as “pair_coeff 3 3 airebo” or epsilon and sigma parameters for LJ interaction needs to be implemented too? \

Best

4

Many-body potentials like AIREBO, REBO, and ReaxFF do not use explicit bonds, angles, dihedrals, or impropers.

I did a tiny bit of research a few years ago with AIREBO for CNTs. Here’s what I had in my input script:
pair_style airebo 3.0 0 0
pair_coeff * * CH.airebo C C

Hope this helps,
Will

5

Same as for bond
See the manual for pair_coeff :wink:
It is rather simple
Best
pascal

Hi Pascal,

Thank you. What about angle and dihedral?

Also is that correct to define pair coefficient between carbon atoms as “pair_coeff 3 3 airebo” or epsilon and sigma parameters for LJ interaction needs to be implemented too? \

Best

6

Hi Will,

Thank you. I have water molecules with graphene sheets contain only carbon atoms in my simulation setup. May you please let me know if “pair_coeff * * CH.airebo C C” definition works fine in my simulation since I have interactions between other atoms in my simulation?

Thank.

Best

7

pair_coeff * * CH.airebo C C will NOT work for your system. The asterisks mean all pairs. You will need to set the AIREBO pair coefficients for just your carbon atoms (as far as I know AIREBO only works for carbon and hydrogen). So it would be pair_coeff 1 1 CH.airebo C (if your carbon atoms are atom type 1). But your system also includes water molecules which can’t use the AIREBO pair style so you’ll need to use a hybrid system. I recommend checking out the documentation for pair_style hybrid here: https://lammps.sandia.gov/doc/pair_hybrid.html.

Best Regards,
Will

8

Dear Will,

Thank you very much. In my simulation setup, carbon atoms are type 3, so the pair coefficient for carbon atoms must be "pair_coeff 3 3 CH.airebo C ". Right?

I am going to equiliberate graphene sheets under NVT constrain with considering AIREBO potential. In my simulation setup, I have also vacuum. Is that required to also fix the position of graphene sheets using “fix spring/self” to hinder displacement of graphene sheets into vacuum region?

Thank you,
Best.

9

Dear Steve,

I am using AIREBO force field with hybrid style with TIP4P water model. To define pair coefficient between carbon atoms I used " pair_coeff * * airebo CH.airebo C NULL" in my Lammps script file which was mentioned by you through the provided following link.
https://sourceforge.net/p/lammps/mailman/message/27964017/

But, it still doesn’t work and showed me the error message of “ERROR: Incorrect args for pair coefficients (…/pair_airebo.cpp:178)”

May you please let me know what wrong with this and how the pair coefficient between only carbon atoms needs to be implemented?

Thank you,
Best

10

Without knowing your full pair_style and pair_coeff commands no one can answer you.

Ray

11

That might be right, but I'm not sure.

If you don't want the graphene to drift into the vacuum region, then
it looks like fix spring/self would work. I usually use fix momentum
for this.

12

Hi Ray,

Below, I provided the commands I used to define define pair_style and pair_coeff for all atoms. In my simulation, H is type 1, O is type 2 and Carbon is type 3.

13

Thank you Will.

The correct definition of pair coefficient in my work would be " pair_coeff * * airebo CH.airebo NULL NULL C". In my data file, C atoms are type 3.

It shows me the error of " ERROR on proc 0: Cannot open AIREBO potential file CH.airebo (…/pair_airebo.cpp:3389)", however, I implemented the file name correctly based on Lammps documentation. Do you know what this issue comes from principally?

Thank you,
Best

14

Hi
I think that 1 3 and 2 3 pair_coeff are missing. 1 1 and 1 3 can be treated by airebo.
I think also taht for airebo pair_coeff should be * * not restricted to 3 3. And thus should appear as the first instance of pair_coeff
Thus
pair_style hybrid …
pair_coeff * * airebo H NULL C
pair_coeff 1 2 …
pair_coeff 2 2 …
pair_coeff 2 3 …

Best
Pascal

Pascal Brault
DR CNRS
GREMI UMR7344
CNRS-Université d’Orléans

15

Hi Pascal,

Thank you for your comments. I modified pair coefficients based on what you have just pointed, as below:

16

some clarifications, since this thread has many guesses and people copying LAMMPS commands without checking the literature or the LAMMPS documentation.

  • pair styles like AIREBO only support the pair_coeff * * form. thus they cannot be limited to pairs of types, since they are not pair-wise additive. order does not matter, though, unless there is an overlap, which would be an error leading to a bad model.
  • it is a mistake to include the water hydrogen in the AIREBO substyle. with a hybrid pair style that includes manybody potentials, each manybody potential may only describe a whole subset.
  • if a TIP4P water potential is to be used, it is important to apply the force field completely, that is with Lennard-Jones and Coulomb (through assigning charges). AIREBO does not use explicit (partial) charges all interactions are included in the formulation using the parameter file. for a CNT or graphene sheet, that is simple, however, as those would have charge zero. of course, polarization can only be modeled in a mean field approach, same as for the water.
  • the missing interactions between the water atoms and the carbon have to be consistent with the water model used.
  • a lot of these issues have been discussed on this mailing list repeatedly. i also strongly recommend to read up on the discussions in the LAMMPS manual. using a hybrid pair style should be considered an advanced simulation procedure. it is a powerful tool, but also opens many doors to make mistakes with the result of bogus simulations. in most cases, those errors are of conceptual nature and thus cannot be detected by the simulation software. it is even possible, that the simulation will run without crashing or producing obviously incorrect results. thus one needs to study the literature, make tests, verify choices, compare to and reproduce results from published studies. there must be a gazillion of publications of simulations of systems of carbon structures in water. asking on a mailing list is no substitute for that.

axel.

17

I think TIP4P water potential have real unit and AIREBO in metal.
How to do simulation with two different units

Thanks

18

Dear Axel,

Thanks. Yes, that is correct. I am still in the step of running the simulation then compared the results with the ones published in the literature. I used the hybrid option for defining the water and graphene potentials, as below:

19

I think the only difference between real and metal units in terms of force field is the energy values. The energy value in TIP4P model is in unit of Kcal/mol compared to the AIREBO force field which has a unit of eV.

Axel, please correct me.

Best.

20

potentials have no units, they can be used with any units, you just need to convert the parameters.
this is, of course, much easier for the case of lennard-jones with two parameters than for AIREBO, where there are lots of parameters and also tabulated functions used. thus parameter files that have been generated for a specific choice of units will need to be used with that.

running a simulation with two sets of units makes no sense.

axel.