SL and ML graphene tensile tests with the LcBOP potential.

Dear lammps Users,

Im sorry for the long post but ive got to a deadend and i really need your advise.

Ive done some strain-stress tests in single layer and multilayer graphene using the LcBOP potential in the isothermal-isobaric ensemble (NPT) but ive got some results that do not agree well with the literature.

Periodic boundary conditions were used and the strain was applied by varying the size of the box (Fix Deform) along the strain axis. In the axis perpendicular to the strain axis the box was relaxing at a pressure of 1 bar. Graphenes with 5000-10000 atoms were used and the strain rate was extensively checked. The stress was computed with the stress per atom algorithm in lammps were the pressure* volume quantity was divided with the volume of the structure which was dynamically computed at every step.

For the monolayer the young modulus (Y) and Poisson ratio (v) were calculated as Y~910 GPa and v=0.22 for T=1K and Y~810GPa and v=0.15+/- 4 for T=300K. The young modulus seems a bit low since its expected to be ~1000GPa.

This does not agree with REF: Fasolino et al. PRL 102, 046808 (2009) where they performed Monte Carlo calculations, since in their case the Young moduli was slightly increased from 0K to 300K and was found to be above 1000GPa. The poisson ratio is in very good agreement though bot 1K and 300K.

  1. Could it be that there is a problem with the usage of stress per atom algorithm in NPT conditions?

Heres how i invoke it:

compute peratom2 GMidLay stress/atom
compute p2 GMidLay reduce sum c_peratom2[1] c_peratom2[2] c_peratom2[3]
variable px2 equal c_p2[1]
variable py2 equal c_p2[2]
variable pz2 equal c_p2[3]

  1. About the stress per atom formula:

i noticed it contains kinetic energy terms, pairwise energy terms, plus bond, angles, dihedral, improper and Kspace terms.

Since LcBOP is a pair style potential, which of these terms are actually used for the calculation of the pressure? (for example the bond terms should not be used since the pairwise energy should do the same thing.)

Im really grateful for any reply and again im sorry for the long post.

A. P. Sgouros

The first thing to check is that whether the LCBOP potential and parameterization you used can actually give you values that agree with experiments. Keep in mind that this is an empirical potential and its transferability is often limited. If Youngs modulus was not included in the fitting, then chances are great that the potential can miss it.