Incorporating 2 MEAM files to define pair_coeff between 3+ elements

Hi All!
Hope you’re doing well.
I’m trying to simulate the forcefields for Ti-6Al-4V on lammps, and I discovered the absence of potential files that directly correspond to these atoms. But I did find AlTi.meam and VNiTi.meam from the NIST portal.

These were the 2 ways I tried using them:
pair_style meam/c
pair_coeff * * library.meam Al Ti AlTi.meam Ti Al NULL
pair_coeff * * library.meam V Ni Ti VNiTi.meam Ti NULL V

The above code gave an error stating: "ERROR: All pair coeffs are not set "

Another way I tried working around this was trying hybrid:
pair_style hybrid meam/c
pair_coeff * * meam/c library.meam Al Ti AlTi.meam Ti Al NULL
pair_coeff * * meam/c library.meam V Ni Ti VNiTi.meam Ti NULL V

this section gave an error: “ERROR: Incorrect args for pair coefficients”

My only reference paper (doi:10.17222/mit.2019.172) used EAM and morse potential, and I can’t seem to find either of them online, so I assume them to be generated through DFT calculations by the researchers. I do not understand that as I’m relatively new to LAMMPS.

I do not know how to model this forcefield, any advice would help.

There are a number of issues to take into account here.
As a starting point, the paper you are referencing does not quote the actual parameters used. That makes it not a very reliable source, but more importantly its model is flawed.

To understand that, you have to recall how embedded atom model calculations work (both MEAM and the EAM potentials used in the reference fall in that category). The basic EAM approach works by computing a pairwise additive contribution to the potential energy (i.e. similar to Lennard-Jones or Mores) and an embedding term that depends on first summing the electron density contributions from neighboring atoms. The second terms makes this a many-body interaction. MEAM then differs from EAM because it has additional, directional pairwise and manybody terms that allow to address shortcomings of the (much simpler) EAM model. Because of the embedding term, any model that tries to simulate an alloy with a hybrid/mixed potential approach has to be flawed since the embedding term is incomplete because some terms are missing because of the non-EAM potentials or computed separately for the EAM subsystems.

So you have to find some way to do all elements in the same potential or you will have a bad model.
Using a hybrid pair style for many-body potentials is only acceptable if you have two different subsystems, e.g. a metal and a silicon or diamond cutting tool, each completely described by a many-body potential and then you can approximate the mixed interaction with an LJ or Morse potential with sufficient accuracy. For your system this can work if you find single element EAM potentials for each element (they have to be compatible with pair style eam, not eam/alloy or eam/fs), then LAMMPS can do the mixing for you. You can get an equivalent behavior when using MEAM without a second meam parameter file (use NULL) and the you will only use the generic parameters in the MEAM library file. However, in both cases, you won’t have any corrections where the material behaves differently from the generic properties of each individual element in an alloy (which is quite common).
In short, you will have to do more research on suitable parameters in the published literature. The combination of elements doesn’t look too unusual, so there are good chances that something is out there.

Now to the two errors that you have observed. The first happens because the second “pair_coeff” statement will wipe out the settings and parameters of the first. Pair style meam/c may only have one pair_coeff statement. The second error happens because the order of arguments is incorrect, but then you would get the same error as the first, since - again - the second pair_coeff would wipe out the effect of the first. Now that can be avoided by specifying meam/c as hybrid sub-style twice, but then you still have the issue that there must not be any overlap of elements between the two MEAM substyles and you must have a pair-wise additive pair style to describe the interactions between the two subs-styles. But this is all moot, because it would be a flawed model as I have explained at the beginning.

1 Like

Thank you very much for the extremely quick response!
I do not entirely understand how EAM and MEAM differs, and I’ll do a bit of reading on my part to understand your answer. I will try to scour the internet to either find eam files for the individual potentials, and take a good look at the current model I have assumed.

Hi @bavesh.rajaraman
Were you able to find a solution to this problem? I am also working on modeling Ti6Al4V in LAMMPS.

Hey @Kishore_A_Nair, unfortunately, my project did not proceed further because I did not find the right potential files to simulate the system. If you take a look at the files I did manage to find, you will find that they correspond either to TiAl alloy systems, AlV systems, or to the VNiTi system. Which really do not help with our case.

Instead I restricted my project to just TiAl systems and tried to work around that. I do not have experience in actually generating interatomic potentials using other software. Do let me know if you have any idea regarding the same!