Relaxation of metals: what to do with entropy term?

[lahaye]

Hi,

I have a metalic Carbon Nanotube with 32 atoms in the unit cell.

The handbook states that I should use ISMEAR=1 with a high value of
SIGMA, but such that the entropy per atom is less than 1 meV per atom.

When I do this for my Carbon Nanotube with a certain SIGMA, I get
typically this in the OUTCAR file:

Code: Select all

entropy T*S    EENTRO =         0.00805847

Which is about 8.058 meV, right?

Should I aim for this number to be less than 1 meV?
Or should 0.00805847 / 32 = 0.25 meV/atom be less than 1 meV?

I guess the last case applies, but I want to be sure.
Can somebody confirm or correct this?

Thanks,
Rob.

[admin]

your entropy term should be less than 32 meV if you have a 32-atoms cell.