[lammps-users] About Tersoff potentials for GaN and BN system

Dear Steve and LAMMPS users,

I have two questions about the tersoff potential for GaN and BN system:

1.About the GaN.tersoff, according to the expressions in the paper of
(J. Nord, K. Albe, P. Erhart and K. Nordlund, J. Phys.: Condens.
Matter, 15, 5649(2003)) and the LAMMPS webpage
(http://lammps.sandia.gov/doc/pair_tersoff.html), it seems that there
should be a minus sign before the parameters B. I am confused about

2.For BN systems, what are the values of n and beta for entry (B N N)
and (N B B), respectively? For more info, please refer to Table 1 of
the paper of Albe (Comp Mater Sci, 10, 111-115 (1998).)
B N N 3.9929061 0.0000016
N B B 0.6184432 0.019251
B N N 0.364153367 0.000011134
N B B 0.364153367 0.000011134
If possible, later I may enclose the BN.Tersoff potential file I built
for BN systems.

Many thanks for any suggestions.

Best regards,

We call it pair_style tersoff, because it is based on Tersoff’s potential. The work that you cite does things a little differently and uses different notation. For example, their B is a completely different quantity to Tersoff’s B. I am fairly confident that you can recreate the BN potential in LAMMPS by choosing the corresponding Tersoff parameter values correctly, but you are going to have to work at it. We did this for GaN already (potentials/GaN.tersoff), but you are going to have to do BN yourself. If you verify that it is working correctly, we would be delighted to include it in LAMMPS.


Let me try to answer these questions. No, there is no minus sign before B. The minus sign was internally already coded in LAMMPS. Tersoff potential table is not unique so there might be different correct ways to enter data. To avoid confusing when memory fades, I always enter the data in an easy way that is correct. i.e., ignore the middle element and always enter the ij parameter in the ikj line. For instance, in your BNN NBB NNB BNN BBN NBB etc., the first and the third elements are always N and B, so you enter parameter_NB in these lines regardless the middle element. Try this and let me know if the results make sense.



According to the manual, element1 is the center atom in a 3-body interaction, element2 is the atom bonded to the center atom, and element3 is the atom influencing the 1-2 bond. So is it also correct to write the table such that BNN corresponds to B-N interaction, NNB corresponds to N-N interaction, BBN corresponds to B-B interaction, etc. That is to say ignore the 3rd element while entering the parameters in the table.



If you look at the lammps source code, it uses ijj parameters to define the pair interaction between i and j, and uses ijk parameters to add the third neighbor k’s contribution on the ij bond. So if you ignore the middle element, lammps will use the ij parameters to define ij interaction, and ik parameters for the third neighbor k’s contribution. This is the correct way. Ignoring the thord neighbor will end up using the ij parameters for the third neighbor contribution, incorrectly.