5
$\begingroup$

I need to get the optimal number of layers of (110) niobium that is a good representative of the behavior of the material, but I am facing two problems using Quantum Espresso,

I Keep getting the following warning when relaxing the bulk material: SCF correction compared to forces is large: reduce conv_thr to get better values

I tried reducing conv_thr, increasing ecutwfc and/or K-mesh, and occupations but no success yet.

The surface energy I get from the (110) slab I cleave from this bulk structure is continuously increasing with increasing number of layers, I am confused whether I am having a problem with the bulk or my surface calculations or even the PP I am using (I need to keep using NCPP as I am going to look into phonon bands), I also examined using dipole cancellation but it does not make a significant difference.

input file for bulk :

&CONTROL
 calculation=  'vc-relax',
 title='ml_gr',
 prefix='ml_gr',
 verbosity='high',
 restart_mode=   'from_scratch',
 nstep=200,
 iprint=1,
 tprnfor=.true.,
 outdir='./tmp',
 disk_io='default',
 pseudo_dir='../pp_sg15_oncv/',  ! vanderbilt  NC PBE,
 forc_conv_thr=1.0d-6,
/


&SYSTEM
ibrav=0,
nat = 2 ,                            
ntyp = 1,
ecutwfc =  55.0 ,   
ecutrho =  440.0,
nosym=.true.,
occupations='smearing', smearing='cold', degauss=0.02,


vdw_corr='grimme-d3',
input_dft='PBE',

/
&ELECTRONS
startingwfc='random',
diagonalization='cg',
conv_thr = 1.0e-12,
mixing_beta = 0.2,
electron_maxstep=200,

/
&IONS
ion_dynamics='bfgs',
upscale=20.0,
/

&CELL
press_conv_thr = 0.5D0,
press = 0.D0,
cell_dynamics = 'bfgs',
cell_dofree = 'all',
cell_factor = 1.5D0,
/

ATOMIC_SPECIES
Nb 92.906  Nb_ONCV_PBE-1.2.upf  ! [Kr] 4d4 5s1,


K_POINTS automatic
12 12 12 0 0 0

CELL_PARAMETERS (angstrom)
        3.3205199242         0.0000000000         0.0000000000
        0.0000000000         3.3205199242         0.0000000000
        0.0000000000         0.0000000000         3.3205199242

ATOMIC_POSITIONS (crystal)
Nb     0.000000000         0.000000000         0.000000000
Nb     0.500000000         0.500000000         0.500000000

input script for slab :

&CONTROL
 calculation=  'relax'
 title='ml_gr'
 prefix='ml_gr'
 verbosity='high'
 restart_mode=    'from_scratch'
 nstep=60
 iprint=1
 tprnfor=.true.
 outdir='./tmp'
 disk_io='default'
 pseudo_dir='../pp_sg15_oncv/'  ! vanderbilt  NC PBE
 forc_conv_thr=1.0d-6,
 tefield=.true.
 dipfield=.true.
/


&SYSTEM
eamp = 0.001  ! default = 0.001 Hartree
edir = 3
emaxpos= 0.9
eopreg= 0.05

ibrav=0,
nat = 10 ,                             
ntyp = 1,
ecutwfc = 50   
ecutrho = 440  
nosym=.true.,
occupations='smearing', smearing='cold', degauss=0.02


!vdw_corr='grimme-d3'
!input_dft='PBE'

/
&ELECTRONS
startingwfc='random'
diagonalization='cg'
conv_thr = 1.0e-12
mixing_beta = 0.2
electron_maxstep=200
mixing_mode= 'local-TF'
/
&IONS
ion_dynamics='bfgs'
upscale=20.0
/


ATOMIC_SPECIES
Nb 92.906  Nb_ONCV_PBE-1.2.upf  ! [Kr] 4d4 5s1


K_POINTS automatic
12 12 1 0 0 0


CELL_PARAMETERS (angstrom)
        4.6764001846         0.0000000000         0.0000000000
        0.0000000000         3.3066999912         0.0000000000
        0.0000000000         0.0000000000        24.3547992706


ATOMIC_POSITIONS (crystal)
Nb           -0.0000000048       -0.0000000013        0.7645284914
Nb            0.4999999930        0.4999999972        0.7645331978
Nb           -0.0000000009       -0.0000000008        0.5760377744
Nb            0.5000000008       -0.0000000057        0.6725692685
Nb            0.0000000027        0.5000000010        0.6725616708
Nb            0.5000000005        0.4999999955        0.5760366470
Nb            0.0000000061        0.0000000081        0.3875391377
Nb            0.4999999975       -0.0000000019        0.4795087546
Nb           -0.0000000008        0.5000000095        0.4795079850
Nb            0.5000000059        0.4999999984        0.3875406089
Improve this question
$\endgroup$
6
  • 2
    $\begingroup$ few suggestion: You convergence value is too tight forc_conv_thr=1D-3 and conv_thr=1D-5 is enough. you can start startingwfc= atomic. You can check without vdw correction as well. kindly put your surface setting as well with question $\endgroup$
    – Pranav kumar
    Apr 4, 2022 at 8:47
  • $\begingroup$ Hi Pranav, I tried different combinations of what you suggested for bulk and unfortunately it did not help with the convergence, I will add the script I am using for the slab to the above question, thanks $\endgroup$
    – Roozbeh
    Apr 19, 2022 at 19:57
  • $\begingroup$ Roozbeh: It's possible that @Pranavkumar didn't see your response because you didn't use the @ symbol to ping him. Users don't get notifications if the @ symbol is not used (unless it's a comment on someone's own post, so I don't need to use the @ symbol to ping you right now!). Pranav, did Roozbeh provide the surface settings that you were hoping to see? $\endgroup$
    – Nike Dattani
    Sep 7, 2022 at 14:25
  • $\begingroup$ @NikeDattani I can see surface setting, I can only suggest to change starting wfc to be atomic and if it doesn't work try another ultrasoft pseudo potential to check $\endgroup$
    – Pranav kumar
    Sep 8, 2022 at 3:54
  • $\begingroup$ Roozbeh, did Pranav's suggestion work? $\endgroup$
    – Nike Dattani
    Oct 26, 2022 at 3:35

0

Reset to default

Browse other questions tagged .