This question is not an issue with LAMMPS, but more so how to calculate the so called "cohesive" or "binding" energy of a crystal.

I am trying to compare my energies calculated with LAMMPS to those obtained from ab initio DFT data. Energy comparisons only make sense if one is comparing the difference between two energies, which is why I want to compare cohesive or binding energies. In DFT, it is easy to find in literature that the cohesive or binding energy is given by the atomic energy of a single atom minus the bulk energy of the crystal.

Is this something that is usually calculated in MD simulations? If so, how is it obtained? Is it just the potential energy? Sorry if my question isn't even a valid one.

2015-07-25 7:52 GMT+02:00 Rohskopf, Andrew D <[email protected]>:

This question is not an issue with LAMMPS, but more so how to calculate

the so called "cohesive" or "binding" energy of a crystal.

I am trying to compare my energies calculated with LAMMPS to those

obtained from ab initio DFT data. Energy comparisons only make sense if one

is comparing the difference between two energies, which is why I want to

compare cohesive or binding energies. In DFT, it is easy to find in

literature that the cohesive or binding energy is given by the atomic

energy of a single atom minus the bulk energy of the crystal.

Is this something that is usually calculated in MD simulations? If so, how

is it obtained? Is it just the potential energy? Sorry if my question isn't

even a valid one.

DFT calculations are typically at 0 K, meaning only (local) minima are

considered. So if you want to compare those to predictions by classical

potentials as implementen in LAMMPS, you should look at the energies after

minimization (with our without optimization of the lattice parameters,

depending if you want to find the optimal lattice constant too, or only

want to compare energies at a given lattice constant). Cohesive energies

are computed as the differences between energy of an atom in the bulk and a

free atom; for many classical potential the latter is zero.

MD simulations can be used to obtained the cohesive energy at a nonzero

temperature, through averaging. But for this kind of ensemble properties,

I'd advice you to consult a textbook on MD and statistical thermodynamics.

Kristof