[lammps-users] NEB method slightly different energy vacancy


I am having an issue in the reproduction of a vacancy migration energy barrier in pure Ni (FCC). I am using a FeNiCr eam potential (https://doi.org/10.1088/0965-0393/21/8/085004) in which it is said that the migration energy barrier of a vacancy in pure nickel is 1.09 eV. Also, another paper confirmed that using the LAMMPS neb method with the same potential the energy barrier is about 1.09 eV (https://doi.org/10.1016/j.actamat.2017.12.064). Also, other literature says the vacancy migration barrier is about 1.04-1.09 eV. However, after several attempts with different settings I get an energy of 0.93 which does not change very much by changing settings. What I do is:

Set a box of size 10x10x10 lattice constants

Relax it for thousands of timesteps with the fix npt @ 300 K and 0 bar


Get the actual lattice constant

Get the data file that I will use as initial.Nivac file for the neb method

Set another input file as follows:

Read the initial.Nivac data file

Remove one atom in the bulk of the box

Generate a final.Nivac file in which I move 1 nearest neighbor of the removed atom to the position of the removed atom

Set the box temperature to 300 K (tried also different Temperatures)

Relax the box with the fix npt


Run the neb method as follows:

Fix neb all 1.0 (I tried with neb “smallergroupofatoms” and also different settings, like perp, parallel, different kspring values etc)

Timestep 0.01

Minimization style quickmin (tried also fire)

Neb 0 0.01 1000 500 10 final final.Nivac (tried with several different values of etol ftol N1 N2 Nevery parameters)

(I usually run partitions like this: 10x1, 8x2, 8x4)

I then plot all the “PEN” values of the final line of the log.lammps file making the difference (PEN – min(PEx))

I always get the maximum value of such difference between 0.9 and 0.94 eV. Still, literature says I should get higher values, especially with the potential I’m using.

(I also dumped the files and I actually see the chosen NN of the vacancy moving to the vacancy spot leaving the new vacancy where it was.)

I would like to know if there is something wrong in what I’m doing or if there are some settings I am missing.

I can attach the in. initial. And final. Files if it can be useful.

Thank you.


Stefano Segantin