How to converge acell, kpt, and vacuum for NbSe2 Monolayer?

Total energy, geometry optimization, DFT+U, spin....

Moderator: bguster

Locked
nimblepanda
Posts: 1
Joined: Wed Jul 19, 2017 12:40 am

How to converge acell, kpt, and vacuum for NbSe2 Monolayer?

Post by nimblepanda » Wed Jul 19, 2017 12:42 am

For reference, my ecut convergence input file is this:

#Dataset
kptopt 1
ngkpt 9 9 1
nshiftk 1
shiftk 0.0 0.0 0.5
toldfe 1.0d-7

#Definition of Unit Cell
acell 2*6.80301 18.8973
angdeg 90 90 60

#Definition of the Atom Types
ntypat 2
znucl 41 34

#Definition of the Atoms
natom 3
typat 1 2 2

xcart 0.0 3.92772 0.0
0.0 7.85544 -3.325918
0.0 7.85544 3.325918

ndtset 13
ecut: 10
ecut+ 2

pawecutdg 40

#Exchange Correlation
ixc 11

#Definition of SCF Procedure
nstep 50
diemac 12.0

TriSugiono
Posts: 10
Joined: Fri Jun 09, 2017 10:50 am

Re: How to converge acell, kpt, and vacuum for NbSe2 Monolay

Post by TriSugiono » Mon Oct 30, 2017 7:53 am

Hello nimblepanda

convergence acelll kpt and other is same way with ecut. same syntax with ecut.
it mean u need to change variable same like ecut. ex acell: 2 2 2 acell+ 1 1 1

is that work?

NPike
Posts: 34
Joined: Fri Mar 18, 2016 1:15 pm

Re: How to converge acell, kpt, and vacuum for NbSe2 Monolay

Post by NPike » Mon Nov 06, 2017 9:10 am

Hello NimblePanda,

During any DFT calculation, for any system you wish to investigate, you need to first determine the energy cutoff (Ecut) and k-point mesh for your particular system. As you have shown in your example, you are currently running 13 calculations, starting from an ecut of 10 Hartree and ending at an ecut of 36 Hartree as shown in your input file, copied below: Note that, you might have some difficulties because NbS_2 is metallic as a monolayer. You will need to add smearing.
#Dataset
kptopt 1
ngkpt 9 9 1
nshiftk 1
shiftk 0.0 0.0 0.5
toldfe 1.0d-7

#Definition of Unit Cell
acell 2*6.80301 18.8973
angdeg 90 90 60

#Definition of the Atom Types
ntypat 2
znucl 41 34

#Definition of the Atoms
natom 3
typat 1 2 2

xcart 0.0 3.92772 0.0
0.0 7.85544 -3.325918
0.0 7.85544 3.325918

ndtset 13
ecut: 10
ecut+ 2

pawecutdg 40

#Exchange Correlation
ixc 11

#Definition of SCF Procedure
nstep 50
diemac 12.0


Additional information about how to determine the Ecut for your system, and how to set up the input files, can be found in the tutorial found here. As displayed in the link, you can determine which Ecut you should use by plotting Total energy vs Ecut.

After this calculation, or in parallel with this calculation, you need to run another calculation in which you determine the k-point mesh for your system as outlined here. Since you are doing a two-dimensional slab of NbS2 you will need to determine the optimal k-point mesh in-plane only (ask yourself why).

After determining your Ecut and k-point mesh then you can do a relaxation calculation which will relax the atomic positions and lattice parameters using an input file similar to the one below:


Code: Select all

#Dataset
kptopt 1
ngkpt  !Enter your optimized grid
nshiftk 1
shiftk 0.0 0.0 0.5
toldfe 1.0d-7

#Definition of Unit Cell
acell 2*6.80301 18.8973
angdeg 90 90 60

#Definition of the Atom Types
ntypat 2
znucl 41 34

#Definition of the Atoms
natom 3
typat 1 2 2

xcart 0.0 3.92772 0.0
0.0 7.85544 -3.325918
0.0 7.85544 3.32591

ecut !Enter your optimized Ecut
pawecutdg 40

#Exchange Correlation
ixc 11

#Definition of SCF Procedure
nstep 50
diemac 12.0

#The Relaxation
#**************
ionmov 2
tolmxf 1.0d-7
ntime 100
optcell 2  ! Note this value will need to be changed to reflect the two-dimensionality of your cell
dilatmx 1.05
ecutsm 0.5



As mentioned above, see the definition of optcell here.

Please let us know if you have any further questions!
Nicholas A. Pike
Research Scientist

Locked