I am simulating 4x4x2 triclinic Al2O3 (960 atoms) using LAMMPS 18March2018.
To minimise the system,I do
1) qeq equilibriation which gives 1.24(Al) and -0.8(O), matches with literature.
2) fix box/relax tri 0.0 which converges energy but has severe
pressure fluctuations (+/-4000 atm)
3) fix npt temp 1.0 1.0 0.02 tri 0.0 0.0 0.2 which relaxes pressure to
+/-40 atm (trace components also relaxed). I refer to this relaxed
system as R1.
I use R1 for NEB (9 replicas)
2 Al and 3 O atoms are removed to maintain charge neutrality.
Initial state is minimised and then 1 O atom is hopped to its adjacent site.
I obtain no barrier for the transition (expected ~ 1eV). When I output
the thermodynamics during minimise, I see that pressure is again
+/-4000 and doesn't reduce after the minimisation.
Stage I results in small forces and gradients.
Stage II identifies a 0.0005 barrier at replica 5.
I proceeded to again do relaxation steps 1,2,3) on R1. The partial
charges dropped to 1.17 and -0.72 (shouldn't). The system pressure was
again high for the box/relax (+/-4000) and dropped to +/-40 after npt
Tracking back, it could be that using minimise during NEB is causing
high pressure but I am not sure.
How can I interpret/resolve this?
Attached are the outputs for initial relax and NEB.
minimiseR1 (453 KB)
minimiseR2 (404 KB)
NEB (19.8 KB)