Question about pair_style buck/coul/cut

Dear Steve,

I am using LAMMPS for oxide systems, e.g. La-Ti-O systems, in literature they published the parameters as:
La-O: Aij=1159.23 eV rhoij=0.351884 Angstrom qi=2.76e
Ti-O: Aij=14567.4 eV rhoij=0.197584 Angstrom qi=2.36e
O-O: Aij=6249.17 eV rhoij=0.231472 Angstrom qi=-1.4e

Aij=0 for cation-cation interactions.

The potential part in LAMMPS script is as:

units metal
boundary p p p
atom_style charge

neighbor 2 bin
neigh_modify delay 5
read_data LaTiO101010.lammps

mass 1 15.999
mass 2 47.867
mass 3 138.9

group oxygen type 1
group Titanium type 2
group Latium type 3

pair_style buck/coul/cut 12 15
pair_coeff 2 * 0.0 10 0 15
pair_coeff 3 * 0.0 10 0 15
pair_coeff 1 1 6249.17 0.231472 0 12
pair_coeff 1 2 14567.4 0.197584 0 12
pair_coeff 1 3 1159.23 0.351884 0 12

#dielectric 0.2

set group oxygen charge -1.4
set group Titanium charge +2.36
set group Latium charge +2.76

After relaxation, it can not get the equilibrium lattice parameter and stress free.

Could you please give some suggestion if the potential format above is correct for LAMMPS?



nothing jumps out as being obviously wrong. But I got a suggestion and a question:

Suggestion: try out a long range coulomb solver (use the buck/coul/long style + pppm).

Questions: choice of cutoffs can make a difference in particular for stuff like stress. So did they use the same cutoff in the literature (or did they even use long range coulomb solver?). Also a third question: many of these type of potentials are not actually giving the correct density at stree zero, or the other way around at the correct density the stress can be way too large or too small (huge positiv or negative stress).
So did the literature on that particular setting specifically said they have zero stress at the correct density/temperature?.


-------- Original-Nachricht --------