Fermi energy

Message

phuongvlpt · #1 Post by **phuongvlpt** » Fri May 07, 2010 8:31 am

Hi all users
I am a new user. I calculate about Perovskite SrTiO3.
I have a problem about fermi energyand want to ask everybody.

My KPOINTS file is:
Monkorst Pack
0
Monkorst Pack
8 8 8
0 0 0

I run VASP and have fermi energy in OUTCAR file is 3.6482(eV)
I change 8 becoming 9, 11, 12, 14 then I received other results.

Can you help me? Why do I have this results?
I think the fermi enrgy is not constanst.
But in the text about solid physics I know fermi energy is constanst per material.
Thank everybody.

Phuongvlpt

panda · #2 Post by **panda** » Fri May 07, 2010 3:47 pm

Your k-space mesh for 8x8x8 is probably too small. Try using auto k-points, with, say, 100 points along the shortest direction in real space:
http://cms.mpi.univie.ac.at/vasp/vasp/node58.htm

forsdan · #3 Post by **forsdan** » Wed May 12, 2010 11:24 am

@phuongvlpt : If you want to obtain an accurate Fermi energy you will need to have a sufficient k-point sampling of the Brillouin zone. If you step by step increase the k-point sampling you will see that the Fermi energy converges to a constant value.

Cheers,
/Dan
<span class='smallblacktext'>[ Edited Wed May 12 2010, 01:25PM ]</span>

phuongvlpt · #4 Post by **phuongvlpt** » Thu May 13, 2010 5:27 am

thank panda and forsdan.

I try to calculate with the K _ point from 8x8x8 to 18x18x18, but the fermi energy didn't converges to a constant value.

thank everybody.
phuongvlpt.

forsdan · #5 Post by **forsdan** » Thu May 13, 2010 8:51 am

How large fluctuations do you have? Because I wouldn't be concerned if you only have fluctations of the order 0.01 eV. If the flucations are higher 0.1 - 1 eV, then increase your k-point grid even more and you'll see that the fluctuations will decrease.

However, please do not use the Fermi level as a convergence criterium in your calculations, since this is a matter of absolute energy convergence. So this is in general not needed unless you want to do really accurate total energy calculations. Instead, converge the properties that you are in (e.g. lattice parameters, bulk moduli, surface energies, adsorbtion energies ...).

<span class='smallblacktext'>[ Edited Thu May 13 2010, 10:57AM ]</span>

alex · #6 Post by **alex** » Sat May 22, 2010 9:45 am

a) symmetry of the k-mesh issue?
b) you may put the mesh and the Fermi energy here.

alex

phuongvlpt · #7 Post by **phuongvlpt** » Sun May 23, 2010 3:42 pm

Thank Forsdan and Alex.

I have understand clearly.
Thank you for helping

Phuongvlpt.