Cohesive energy with ReaxFF

LAMMPS LAMMPS Mailing List Mirror
1

Dear all,

I have seen some examples where people calculate cohesive energy (not necessarily with ReaxFF) as (total energy)/(number of atoms). But shouldn’t it be the difference between (total energy)/(number of atoms) and energy of a free atom?

I’m testing cohesive energy of bulk aluminum with ReaxFF. I see in the paper (Zhang, Cagin, Duin, Physical Review B 69, 2004) that the cohesive energy of bulk aluminum obtained with ReaxFF is 3.04 eV. I assume this is NOT the total energy over number of atoms because total energy shouldn’t make any sense in case of ReaxFF. Furthermore, it is not able to estimate the energy of a free atom. Unfortunately, there is no information how this cohesive energy was obtained. Does this mean I should take some reference from ab-initio for single atom? (this is not actually LAMMPS-related question but maybe someone has some experience?)

Thanks for any help,
Manana

2

Dear all,

I have seen some examples where people calculate cohesive energy (not
necessarily with ReaxFF) as (total energy)/(number of atoms). But shouldn't
it be the difference between (total energy)/(number of atoms) and energy of
a free atom?

have you considered, that the potential energy of a single atom (or two
atoms infinitely far apart) within a classical model (e.g. LJ) is zero?

I'm testing cohesive energy of bulk aluminum with ReaxFF. I see in the
paper (Zhang, Cagin, Duin, Physical Review B 69, 2004) that the cohesive
energy of bulk aluminum obtained with ReaxFF is 3.04 eV. I assume this is
NOT the total energy over number of atoms because total energy shouldn't
make any sense in case of ReaxFF. Furthermore, it is not able to estimate
the energy of a free atom. Unfortunately, there is no information how this
cohesive energy was obtained. Does this mean I should take some reference
from ab-initio for single atom? (this is not actually LAMMPS-related
question but maybe someone has some experience?)

you cannot compare ab initio and classical models.
the classical model integrates out the interaction
between the atom core and the electrons.

axel

3

I also think it should be (total potential energy of solid)/(number of atoms in the solid) rather than (total energy)/(number of atoms). Else, do people consider the Kinetic energy too and define “cohesive energy at a given temperature”?

Best Regards
Manoj

4

Hello Manoj,

I assumed 0 temperature calculations, just relaxation, so that potential energy=total energy.

Thanks Axel, I completely agree.

Cheers,
Manana

5

Hi Mañana and Manoj,

Conventionally in the field of potential development, "cohesive
energy" is regarded as the total potential energy and is usually
reported in energy per atom or energy per chemical formula. The
difference between the cohesive energy and energy of their
constituents in standard form is regarded as "heat of formation".

Best,
Ray