SiC Tersoff/ZBL Incorrect Lattice Parameter after minimization


I was playing around with elastic constants of SiC using potentials available in lammps potentials directory using the examples/ELASTIC formulation for doing so.

Essentially I believe there is a bug in the SiC.tersoff/zbl parameter file somewhere.

  1. The tersoff/zbl potential gives incorrect lattice parameter of 4.2796A after minimization.
  2. SiC_Erhart/Albe.tersoff potential tested gave a lattice parameter of 4.3593A after minimization

Is there anything amiss with that may be causing this issue?


(Attachment elasticDOTtgz.jpg is missing)

Aidan can comment. In general, there is no guarantee
that a potential will produce an exact match to a lattice
constant. It is fit to a variety of properties, and none
of them will be matched exactly. So assumuing
there is not some typo in the potential file, it is a "feature"
not a bug.



An unexpected result is not a sufficient condition for the existence of a
bug, nor is it even necessary. Rather than comparing with experiment, you
should compare your calculations with the results published by the authors
of the potentials. Note that authors often publish more than one variant
of a potential. Note also that every calculation has a numerical
uncertainty due a variety of factors. In the case of lattice constants for
fully relaxed cells, that uncertainty should be less than 1%.


I agree with Steve. Potentials are often fitted to reproduce some combination of lattice constants, elastic constants, melting temperature, and defect properties. Even the well-known Tersoff potential for SiC has several different parameterizations to optimize different properties. The Tersoff potentials slightly underestimate the lattice constant of beta-SiC. This is not a failure or drawback of this potential. Empirical potentials are approximate representations of reality.

The published values of the lattice constant of beta-SiC with the Albe and Tersoff potentials are 4.359 and 4.280, respectively. [See Erhart and Albe, PHYSICAL REVIEW B 71, 035211 (2005)]. The lattice constant you got is not "incorrect" and this is not a sign of a bug in the SiC.tersoff/zbl parameter file or in LAMMPS.
If anything, this indicates that the parameterization is correct and LAMMPS is working as it is supposed to.