Streitz-Mintmire vacuum surface

Hey LAMMPS users,

I’m trying to simulate an alpha-alumina surface, and I need a reactive force field. I started of trying the Streitz-Mintmire potential, but as soon as I break the periodic crystal, say with a vacuum gap in the xy plane, the structure undergoes massive restructuring. This is something I have not seen referenced in any publications that look at the Streitz-Mintmire potential model of this surface, though there don’t seem to be a lot of those.

I am currently using LAMMPS version 16Mar18. Maybe the simplest way to see the problem I am experiencing is to use the SM alumina example that ships with LAMMPS. If you add a vacuum gap by expanding the box in the positive z direction, you end up with two single Al layer terminated (0001) faces. This should be stable based on my literature searching.

If I then run an NVE simulation for 10000 steps, I run into an H matrix size exceeded issue; however even before that, the beginnings of my other, more universal problem is appearing. The layers normal to the z-axis start to shift and even change order. If I run 1 step and minimize first, it still has this problem. Running NVT simulations, the same thing happens. It appears that the stable structure of this surface for the default Streitz-Mintmire-EAM potential is not behaving as one would expect based on a literature search, as no one reports such a restructuring. Is there some trick to equilibrating this surface I am unaware of?

For further notes, I have also run an NPT simulation of the bulk crystal at the temperature and pressure I am interested in, and used a snapshot close to the average xy cross section as a starting point for an NVT run, hoping that stresses in the x and y directions were affecting the structure, but the reordering still occurs. I haven’t actually calculated stress profiles, but so far, that seems an unlikely reason for this behavior.

Any ideas?


Seth Martin

The� reaxFF file is in the supporting info. Regards