Atoms overlapping - Au-Ag alloy

LAMMPS LAMMPS Mailing List Mirror
1

Dear LAMMPS community

I am senior lecturer at Liverpool John Moores University (UK) and would be glad if you could help me with a LAMMPS issue.

I am trying to run a NVT simulation of a Silver-Gold alloy (86000 atoms, 20% of Au) using the respective EAM potentials (universal 3, distributed in the potentials directory of lammps). I know that when we use two different metals combined in an alloy, the EAM parameters for the alloy are automatically generated by standard mixing rules (the best approach would be to have the alloy-EAM potential already parametrized for the Ag-Au alloy, but I don’t find it). The problem is that, although I begin with a nanoalloy well centered on a cubic box (18 nm of side) and which shows a clear first RDF peak of Au-Ag at 2.9 Angstroms, after I start the simulation even at low temperature, atoms begin to strongly overlap, until they get as close as 0.1 Angstroms (revealed by the RDF analysis as well as by visual inspection using VMD of the trajectories file). Also, my simulation finishes normally without any errors. Would it be necessary to use SHAKE to prevent atoms from coming so close during the simulation? Or perhaps the repulsive part of the EAM potential only works for pure Ag or pure Au nanoparticles? Adopting the same procedure to simulate Au-Pt alloys by combining two different individual EAM potential files works fine…

I am running the calculation on a MacPro (24 logical cores) using mpirun with 24 logical cores and my input is shown below:

units metal

boundary f f f
read_data coordinates

pair_style eam
pair_coeff 1 1 Ag_u3.eam
pair_coeff 2 2 Au_u3.eam

run_style verlet

thermo_style custom step temp press pe ke etotal

min_style sd
minimize 1.0e-8 1.0e-6 10000 10000

dump 2 all xyz 100000 trajectories.xyz
dump_modify 2 element Silver Gold

timestep 0.00001

velocity all create 2 158379 mom yes rot yes dist gaussian

fix 1 all nvt temp 2 2 0.001

thermo 1000
run 10000000
unfix 1

undump 2

Thanks in avance
Rodrigo

Dr. Rodrigo Q. Albuquerque, BSc MSc PhD FHEA
Senior Lecturer in Physical Chemistry
Liverpool John Moores University

Tel: +44 (0) 151 904-6287
Web: www.labsuprachem.com
E-mail: R.Q.Albuquerque@…7203…

Address:
Byrom Street, James Parsons Building (R. 2.03b)
L3 3AF Liverpool, UK

2

This occurred because you can’t use two separate EAM potential files for this purpose. Your type 1 and type 2 atoms effectively have no interaction parameters defined. You need to use an EAM potential file with both Au and Ag parameterized, such as this one: https://www.ctcms.nist.gov/potentials/Zhou04.html

Then your pair_coeff command would be: “pair_coeff * * Au Ag”

Best regards,

Ray

3

This occurred because you can’t use two separate EAM potential files for
this purpose.

​i don't think so.​ the documentation for pair style eam says, that LAMMPS
*does* do the mixing, so only i i and j j parameters need to be provided.

the important missing piece of information (as usual) is the LAMMPS version
in use. between september 2016 and june 2017, there has been a bug in pair
style eam, that would cancel all mixed interactions.
so that is one possible explanation. another possible source of error could
be the data file, which is not provided.

axel.

4

Yes, my mistake. Thanks for the correction.

Ray

5

Thanks a lot Axel!
I was in fact using the LAMMPS (17 Nov 2016) version. I will then update it and try again.

Best regards
Rodrigo

Dr. Rodrigo Q. Albuquerque, BSc MSc PhD FHEA
Senior Lecturer in Physical Chemistry
Liverpool John Moores University

Tel: +44 (0) 151 904-6287
Web: www.labsuprachem.com
E-mail: R.Q.Albuquerque@…7203…

Address:
Byrom Street, James Parsons Building (R. 2.03b)
L3 3AF Liverpool, UK