This is more targeted towards the LAMMPS user base rather than developers. I’m currently trying to make use of the Tersoff parameters developed by M.S. Daw et al. ( M.S. Daw, et al., Comp. Mat. Sci. 50, 2828 2011). Attached are the two LAMMPS potential files for both ZrB2 parameter sets reported in that work. For pure Zr I’ve gotten good agreement with the properties used for fitting (i.e. lattice parameter, cohesive energy, bulk modulus). However, for pure B and ZrB2 I’m not getting as good of an agreement, for example, the cohesive energy for ZrB2 is off by as much as ~0.2 eV and C33 is an order of magnitude off. I’ve gone over my parameter files but I can’t seem to locate any typos. My question is, has anyone implemented and confirmed these parameter sets using LAMMPS?
Thank for any help,
ZrB2_DawP1.tersoff (2.03 KB)
ZrB2_DawP2.tersoff (2.04 KB)
Well, if you are confident that you have created the potential files correctly (which I am pretty sure you have), the next logic thing to do is to compare the Tersoff functional form used in LAMMPS (1988 PRB) and the one in Daw et al.
Upon careful comparison, you will notice there are two differences in how the symmetry function (Eq. 7 in Daw) is formulated. You will have to write a new Tersoff pair_style to incorporate these two differences.
Thanks for the response. I’m not sure if I’m catching were the differences are for the zeta parameter (eq. 7) between the Tersoff implemented in LAMMPS and the one given in M.S. Daw et al. I have no problem with adding a C++ child class (I’ve done it for Stillinger-Weber), I’m just not observing the subtleties between the expressions.
Zeta is a function, not a parameter …
Anyhow, modifying the code is trivial, but what is more important is to identify the differences (two in the paper but only one in LAMMPS’ implementation).