Cohesion energy calculation

[antoineM]

Hi,
I ve some questions about cohesion energy calculation.
1- In some papers I saw that, for cohesion energy calculation, one must subtract the energy of the references atoms contained in the cell. But in the vasp documentation (in the description of the OSZICAR file), it is said that for the total free energy "at this point the energy of the reference atom has been subtracted". So do i must subtract it or not?
If not, is that the same if I take into account of the spin?

2- I don't understand what is the difference between the "energy without entropy" and the "energy for sigma->0". Generally, which one do I must use?
Some people use TOTEN (the total free energy), is that a good choice?
Thx

AM

[forsdan]

1. The total free energy given by VASP is with respect to the nonpolarized atom. If your atoms can have spinpolarized states you'll have to perform separate calculations allowing for the spinpolarization in order to obtain the correct cohesive energy. Please have a look at

http://cms.mpi.univie.ac.at/vasp/vasp/node231.html

to see how to perform the calculations.


2. The "energy without entropy" and "free energy TOTEN" are the internal energy and free energy, respectively, evaluated at the fictitious smearing temperature (which is determined by sigma). However, the objective is to evaluate the internal energy at T = 0 K, which is given by the "energy for sigma->0" energy. This energy is determined using the results for the internal and free energy at the finite temperature and constitutes the approximation* for the ground state energy. So for extended systems where you have applied a smearing scheme you should use the "energy for sigma->0" energy for further analysis.

* Provided that the smearing width isn't too high the approximation is correct to the first order(s) in sigma.

Best regards,
/Dan





<span class='smallblacktext'>[ Edited Tue May 19 2009, 08:16AM ]</span>

[antoineM]

Hi and thank you for the reply!
Best regards,

Antoine