Does anyone work with such a potential?
I’ve tried using the file that Insepov posted to http://www.ctcms.nist.gov/potentials/Mo.html#Mo in 2009, which is based on G.J. Ackland and R. Thetford, “An Improved N-Body Semiempirical Model for Body-Centered Cubic Transition-Metals,” Phil. Mag. A 56, 15-30 (1987). I have came to conclusion that something with this file or its compatibility with current LAMMPS version is corrupted.
How did you come to this conclusion? It is impossible to tell if you are describing a problem with LAMMPS, a problem with how you use the potential within LAMMPS, or a problem of whether the potential is appropriate for the problem you wish to address.
First thanks for your quick reply.
After several attempts to run MD with my specific setup (simulation cell containing dislocation dipole), with regular procedures that I’m using for other systems (fcc and bcc metals including eam/fs pair styles), I ended up trying to run MD for perfect crystal (meaning no dislocation and no modification at all. just created the box and the atoms.), skipping the energy minimization stage (was corrupted too).
MD corruption: I’m using fix npt to first equilibrate the box to some finite temperature. In the case of this Mo potential, the things get messy very quick - the temperature rises to 10^5 and the visualization confirms it (no trace to what was once solid bcc metal, just amorphous liquid). Again - this is a simple box now, filled with perfect bcc crystal, which is just by case out of Mo atoms right now. Yesterday it was bcc tungsten or iron etc.
Energy minimization: using conjugate gradient method to relax the system from an initial configuration (dislocation dipole). Usually all it does is changing a bit each atom position of course. Here, surprisingly, a surrealistic thing happens - very large ordered voids form. You get an undoubtedly unrealistic picture - there are atoms with neighbours that are far as several lattice spacings in one direction, and close as less than Burgers vector in the other direction.
Again, I’m using the same procedures for other potentials of the same kind and never encountered problems of this kind.
Thank you again
I just tried this myself with bcc and got the typical behavior I would expect when using minimize and fix box/relax. I used pair_style eam/alloy as per the notes on the potentials website. On a whim, I switched to pair_style eam/fs and saw something that probably matches your description. Which pair_style are you using?
On a whim…hmm. And I on a whim didn’t try the opposite, shame on me. Now it’s all clear. I guess I got confused because of the “.fs” file extension…
Many thanks, I deeply appreciate your help!