Hello LAMMPS mailing list,
I have been attempting to simulate a CuSn system, and have recently learned that LAMMPS is capable of auto-generating alloy interactions from single-element EAM potential files, as performed via:
pair_style eam
pair_coeff 1 1 Cu.eam
pair_coeff 2 2 Sn.eam
However in searching for Sn potential files, I have only found MEAM potentials.
Q: Can the MEAM interaction (i.e. pair_style meam/c) be implemented to “auto-alloy” with single-element MEAM files, in a similar manner as the above EAM interaction? Perhaps via:
pair_coeff 1 1 library.meam Cu Sn Cu.meam Cu Sn
pair_coeff 2 2 library.meam Cu Sn Sn.meam Cu Sn
(…assuming library.meam has been updated with Sn data. I am aware there are issues with this syntax but thought I’d include it to illustrate my thought process.)
The LAMMPS documentation is normally very good but seems inconclusive/non-specific on this matter. The only related question I was able to find (https://lammps.sandia.gov/threads/msg37289.html) seems to suggest it is possible, though offers no specifics on implementation syntax.
i disagree with this statement. but to understand the manual you have to understand the process of how MEAM potential parameters are applied, and this is explained in great detail in the documentation for the meam/c pair style.
first off, with pair style meam/c you are use a single pair coeff * * command to set parameters for all atom types and map the (numerical) LAMMPS atom types to the (text) potential types in the meam “library” potential file.
then you have to be aware that MEAM has two kinds of parameter selections and thus uses two parameter files. the first file is the “library” potential file which contains a library (sic) of atom types and (generic) per type parameters. then you provide a second file (or use NULL) which will tweak the generic parameters for specific pairs or triples of atom types. while the library file is generic the second is very specific and indexes atom types numerically by the order in which they are extracted from the library.
if you look at the library.meam file bundled with LAMMPS there are two entries for Cu (Cu and CuS) and two entries for Sn (Sn and Sn5). assuming the the Cu and Sn entries would be the ones to use, your initial pair_coeff command would be (assuming that you have Cu as atom type 1 and Sn as atom type 2):
pair_coeff * * library.meam Cu Sn NULL Cu Sn
or:
pair_coeff * * library.meam Sn Cu NULL Cu Sn
both should give the exact same results since you are not using any specialization parameters in a second potential file. however, with this you are missing any specialization for CuSn alloys, so your resulting model may be less accurate than it could be. Parameters for those you would have to look up separately in the published literature (they may also come with a different set of library parameters) and then translate those to the format LAMMPS uses which is described in the pair_style meam/c documentation.
axel.