TERSOFF potential parameters


I have some questions about the form of the Tersoff potential parameters used in GULP. I have seen a couple of different form of the equations in literature for Tersoff potential (please see the attached images). Now, let’s say we have two atom types (Si and O for example) in the system. In one of the form of the equations, the parameters such as n, beta,c,d, and h are not symmetric for cross types of atoms (for example: will be different for Si-O vs O-Si terms). My questions are:

  1. Which form does GULP exactly use among these two?

  2. Looks like, if I use the “combine sub-option” in the " botwobody option", then symmetric values of the parameters will be generated using the mixing rule. However, if I do not want to use that and still want to keep the symmetry, do I need to explicitly list the different cross body terms as equal numeric values? (In the previous example: Should I list the Si-O and O-Si terms separately? or just listing one of them like Si-O will suffice?)

Any suggestions will be really helpful.

Thanks in advance.

Shoieb Ahmed Chowdhury


  1. The equations that GULP use are specified in the help text in case you want to check the details. Broadly they are like the second image in that the f_A(r) is defined as negative and so B is input as positive, but you should check the other details.
  2. Cross terms for Tersoff can be different for Si-O and O-Si as for bond order models (and embedded atom etc), though could be specified as symmetric in some cases as well. See the examples with tersoff_zrl library where the potentials aren’t symmetric.