[lammps-users] about the ~/potential/library.meam file

Hi all,

I want to simulate a pure carbon system (like graphene) using MEAM
potential, but when I checked the ~/potential/library.meam file, I got
confused. Why in this file there are 6 parameter sets for 'C' element?
Actually there is 5, because the 3# and 6# parameter sets are
completely same. When I used the below script to do simulation I got
very unreasonable results:

pair_style meam
pair_coeff * * library.meam C SiC.meam C

From the manual, it seems lammps used the 1# parameter set to do

simulation. But when I just kept one parameter set and remove the rest
in the library.meam file one by one, I found the parameter set 4# can
give best result. Here I say the best result is because I compared the
forces on each atom generated by lammps with MEAM potential with those
generated by DFT calculation.

So my questions are:

(1) Why there are 6 parameter sets for C in library.meam file?
(2) What are the references (published articles) for all of the 6
parameter sets?
(3) For the pure carbon system simulation, how do I prepare the
library.meam and C.meam files?

Thanks so much!


Hi Dongshan,

The library.meam file is provided as an example, and is meant to be used with care. To understand it, you need to understand it's history: basically, it's the file that Mike Baskes provided with his original MEAM code. Mike and his collaborators edited it along the way, without providing any references; the version included with lammps is just a snapshot of the file at one point during the development of lammps. So, if there are 6 versions of the parameters for C, the reason is simply that these were all versions that Mike or someone else played with along the way. Whether any of these versions made it into the literature can only be answered by looking through the literature, I'm afraid. In fact, you'll want to do that anyway, since some of the options in lammps/meam depend on the exact form of the model used to derive the parameters (see, e.g., the augt1 parameter).

So that answers questions 1 and 2 -- sorry it's probably not the answer you want.

On question 3, for the pure carbon system you don't need any of the cross terms (like Ec(1,2)) defined for the inter-element interactions in the SiC.meam file. These may give errors (or bad behavior) if you try to define them for a single element type, I think. You may not need to specify very much at all, unless you need to override parameters like Cmin or Cmax. Or, if you really want to use the SiC.meam file, you could define a second element type for silicon and then just not create any Si atoms.

Hope that helps.