[lammps-users] MEAM


We tried to reproduce some calculations on liquid nickel performed by Cherne et al. (PRB 65 (2001) 24209) with an MEAM potential. They studied the melting of a bulk Ni sample at 1726K by using the Ni4 potential published by Baskes.

The melting temperature reported by Cherne et al. is 1570K. We used the same system : 1372 atoms, NPT relaxation followed by NVE runas. However, even at 1775K we were not able to observe the melting of the sample.

We also tried to study the self diffusion of Ni in a bulk which contains several vacancies at high temperature. We did not observe any diffusion !!! So I'm wondering if someone already find the same kind of problem with LAMMPS and the MEAM potential.

We have been trying to solve this for a long time, any suggestion is highly appreciated!


O. Politano

Note that, We used the following files :

==> library.meam provided with LAMMPS. We replaced in this file the initial value of ibar for Ni4 by 3 to agree with Baskes (Mat. Chem. Phys. 50 (1997) 152).

==> Ni.meam

re(1,1) = 2.49
Cmin(1,1,1) = 0.8
rc = 4.0
delr = 0.1

=> in.meam

units metal
boundary p p p

atom_style atomic
lattice fcc 3.52
region box block 0 7.0 0 7.0 0 7.0
create_box 1 box

create_atoms 1 box

pair_style meam
pair_coeff * * ./library.meam Ni4 ./Ni.meam Ni4

neighbor 0.3 bin
neigh_modify delay 10

thermo_style custom step temp ke pe etotal lx ly lz vol press

compute Temperature all temp
velocity all create 1775.0 87287
fix 1 all npt 1775.0 1775.0 10.0 aniso 0.0 0.0 0.0 0.0 0.0 0.0 5.0 drag 2.0
fix 3 all temp/rescale 1 1775.0 1775.0 10.0 1.0

compute 4 all ackland/atom
compute 6 all centro/atom
thermo 10

dump 2 all custom 5000 Ni-npt.xyz tag x y z type c_4 c_6
timestep 0.001
run 200000

undump 2
unfix 1
unfix 3
dump 2 all custom 10000 Ni-nve.xyz tag x y z type c_4 c_6

fix msdni all msd 500 msd-ni.out

fix 1 all nve
run 3000000

Greg Wagner (gjwagne at sandia.gov) is your best bet
for answering something specific about MEAM. I would just
say that 1300 atoms is a small sample and that melting might
be problematic with NVE which allows for no volume
change - thus overshooting the melting T would not be uncommon.