Bulk Lattice constant:by Fitting or Relaxtion ?

[gpkmohan]

Dear List

I am using VASP to calculate Bulk-Lattice constant for the Ni(111), with two different approaches, using the same KPOINT file (please see below) and other non-dynamic parameters.

Approaches:

1. Performing single point calculations: i.e. by varying different lattice constants in the f.c.c. unit cell, and fitting the curve of energy vs. lattice constant by means of a cubic-spine; and takes the minimum domain as the lattice constant. [pls see : www.nsc.liu.se/lcsc2005/vasp/handson_excercises_4.ps].

2. from the volume relaxation: as depicted in http://cms.mpi.univie.ac.at/vasp/vasp/node211.html.

note: Since I found the cell-shape is changed, and force is converged, I did not perform the “second relaxation from previous CONTCAR file (re-relaxation).�

Results:
I obtained almost similar experimental lattice parameter by approach-1 (3.5225 Angstroms) and approach-2 is found only 3.516 Angstroms !

parameters:

KPOINTS
::::::::::::::
K-points
0
Gamma
11 11 11
0. 0. 0.

INCAR// for the volume relaxation
::::::::::::::
SYSTEM = Ni,fcc
IBRION=1
ISIF=7
NSW=100
EDIFF = 1E-6
GGA=91
VOSKOWN=1
ISPIN=2
MAGMOM=1
LREAL=.FALSE.
ADDGRID = .TRUE.
PREC= Accurate
Electronic minimisation
ENCUT =400 eV
DOS related values
ISMEAR = 1;

INCAR//single point calcs
::::::::::::::
SYSTEM = NI-fcc
EDIFF = 1E-6
GGA=91
VOSKOWN=1
ISPIN=2
MAGMOM=1
LREAL=.FALSE.
ADDGRID = .TRUE.
PREC= Accurate
Electronic minimisation
ENCUT =400 eV
DOS related values
ISMEAR = -5;

Interestingly when I tighten-up the KPOINTS to 15 15 15 on the KPOINT file[all-else is same] and EDIFF = EDIFFG = 1E-6, and ENCUT=600, in INCAR[volume relaxation INCAR file] file I got 3.5225 as the lattice constant which agrees with approach-1 value.

Question:
I wonder why we need to setup higher settings in approach-2 in order to get acomparable value from approach-1 [or experimental].

Any help would be GREATLY appreciated!!

With best regards and thanks for reading,

Krishna Mohan GP.
THEOR CHEM
University Of Leiden.

[forsdan]

The issues with approach 1 vs. 2 are pointed out in http://cms.mpi.univie.ac.at/vasp/vasp/node209.html . The stress tensor will be incorrect if your cut-off energy is insufficient and thus the equilibrium volume will be off. Your results supports this, since when you increase the cutoff to 600 eV the results coincide, pointing to that the stress tensor is more accurate in this case.

[gpkmohan]

Thanks forsdan for your quick response. Its really helpful