Tersoff potential taken from paper does not converge

Hello all,
I’m trying to implement Tersoff potential files for Ga-In-As heterostructures using data taken from this paper Atomic-scale modeling of self-positioning nanostructures (Computer Modeling in Engineering and Sciences, vol.26, no.2, pp.91-106, 2008).
I started with the Ga-As pair, for which there already is a reference Tersoff file in the Atomic Potentials Repository. The file I generated using the formulas in the paper and the LAMMPS documentation does not converge for an equilibrium configuration (Zincblende crystal with the right constant), as the reference one does.
This is the content of the potential file, I’ve checked multiple times for the formulas. Am I doing something wrong?

|Ga|Ga|Ga|1|1|0|0.07629773|19.796474|7.1459174|3.4729041|0.23586237|1.490824|136.123032|3.5|0.1|2.50842747|993.888094|
|As|As|As|1|1|0|0.60879133|0.75102662|0.15292354|0.60879133|0.00748809|1.7287263|546.4316579|3.5|0.1|2.384132239|1571.86084|
|As|As|Ga|1|1|0|0.60879133|0.75102662|0.15292354|0.60879133|0.00748809|1.7287263|546.4316579|3.5|0.1|2.384132239|1571.86084|
|As|Ga|As|1|1|0|6.31741|0.790396|−0.518489|6.31741|0.357192|1.72301158|314.45966|3.5|0.1|2.82809263|2543.2972|
|As|Ga|Ga|1|1|0|6.31741|0.790396|−0.518489|6.31741|0.357192|1.72301158|314.45966|3.5|0.1|2.82809263|2543.2972|
|Ga|As|As|1|1|0|6.31741|0.790396|−0.518489|6.31741|0.357192|1.72301158|314.45966|3.5|0.1|2.82809263|2543.2972|
|Ga|As|Ga|1|1|0|6.31741|0.790396|−0.518489|6.31741|0.357192|1.72301158|314.45966|3.5|0.1|2.82809263|2543.2972|
|Ga|Ga|As|1|1|0|0.07629773|19.796474|7.1459174|3.4729041|0.23586237|1.490824|136.123032|3.5|0.1|2.50842747|993.888094|

Nobody can tell from just a bunch of numbers.

Which repository? Which reference paper?

The best path forward for you is probably the following.

  • get hold of the paper describing the GaAs potential file and try to re-generate that file. That way you can see, if there are problems with how you translate the parameters into a LAMMPS .tersoff file. Different researchers may use different conventions for naming some of the constants, so you need to compare what the paper uses to what LAMMPS uses.
  • compare the two papers and make sure you identify equivalent parameters and then repeat the recreation but now with the new parameters and see if that file differs from what you currently have.
  • from here on it depends on whether creating the GaAs potential part is successful or leads to the same outcome (or exact same parameters) as your previous attempt. Then either you have the confirmation that the potential parameters have a problem with your test system, or you have corrected and confirmed parameters and then can proceed with translating the remaining settings to complete the ternary .tersoff file.