Roberto Veiga wrote:
Hello,
in PRB 214103 (2003), for the Fe16C1 system (BCC iron with a carbon atom
in an octahedral site), the solution enthalpy at 0 K is defined as
H=E(Fe16C1)-E(Fe16)-E(C),
being E(C) the total energy per atom in graphite and the meaning of the
other ones is straightforward. They have obtained, with VASP, H=0.58 eV.
By using Siesta with a DZP-DZP basis for iron (with which I obtained a
very good agreement of the lattice parameter, vacancy formation energy,
and cohesion energy with respect to the literature) and DZP-DZ for
carbon (with which the cohesion energy of graphite is in agreement with
the above mentioned reference and the lattice parameter with the
experimental one), and applying the same equation above to the same
system, I obtained H=0.30 eV. I guess the difference between VASP's H
and Siesta's one is due to the BSSE. Nevertheless, I did not figure out
how to apply the correction. If I replace E(Fe16) by E(Fe16+C_ghost) and
E(C) by E(C+Fe16_ghost), things become worst. Does anyone have any idea
on how apply a BSSE correction in this case?
Hi,
Are E(Fe16C1) and E(Fe16) referenced to the same zero of energy? Were
they in the PRB paper you refer to? Phys. Rev. B 53, 3813 (1996), near
Eq (2.5) talk about this (they talk about charged defects, but I think
the problem might also be there for neutral ones).
Also, I would only do the E(Fe16) by E(Fe16+C_ghost) replacement when
trying to correct for BSSE, but not the other one.
Regards,
Xavier
