ABINIT Discussion Forums

Dear colleagues!
I want to calculate the convergence in ecut for SiO2 in the PAW case, but the total energy do not convergence. Help me, please.
ecut runs from 10 to 30 Ha with 2 Ha step and pawecutdg is 30. I used ''014-Si-gpbe-n-campos-mod'' and ''008-O-gpbe--bm'' PAW dataset from http://www.abinit.org/downloads/PAW2
In this case, ecut and total energy are:
ecut1 1.00000000E+01 Hartree
ecut2 1.20000000E+01 Hartree
ecut3 1.40000000E+01 Hartree
ecut4 1.60000000E+01 Hartree
ecut5 1.80000000E+01 Hartree
ecut6 2.00000000E+01 Hartree
ecut7 2.20000000E+01 Hartree
ecut8 2.40000000E+01 Hartree
ecut9 2.60000000E+01 Hartree
ecut10 2.80000000E+01 Hartree
ecut11 3.00000000E+01 Hartree

etotal1 -2.8961309282E+02
etotal2 -2.9063159751E+02
etotal3 -2.9085264922E+02
etotal4 -2.9087825273E+02
etotal5 -2.9088278587E+02
etotal6 -2.9090269284E+02
etotal7 -2.9093382800E+02
etotal8 -2.9096661264E+02
etotal9 -2.9099270955E+02
etotal10 -2.9100917263E+02
etotal11 -2.9101729198E+02
My input file:

#Specific to ground state calculation

ndtset 11

#Specific to ground state calculation

ecut1 10
ecut2 12
ecut3 14
ecut4 16
ecut5 18
ecut6 20
ecut7 22
ecut8 24
ecut9 26
ecut10 28
ecut11 30

pawecutdg 30
pawmixdg 1

prtpot 0
#prtden 0
prtvhxc 0
prtwf 0
#prteig 0

toldfe 1.0D-6

# dilatmx 1.1 # Maximum scaling allowed for lattice parameters
# ionmov 3 # Use BFGS algorithm
# ntime 500 # Maximum number of optimization steps
# optcell 1 # Fully optimize unit cell geometry, keeping symmetry
# tolmxf 1.0e-4 # Convergence limit for forces as above

#######################################################################
#Common input variables

#Definition of the unit cell
acell 7.16000 7.16000 7.16000 angstrom


angdeg 90.00 90.00 90.00
spgroup 1
# chkprim 1
occopt 1
nsppol 2
# spinat 0.0 0.0 0.0
# 0.0 0.0 0.0
# 0.0 0.0 0.0
# 0.0 0.0 0.0
# 0.0 0.0 0.0
# occopt 3
# tsmear 1.0d-3
enunit 1

# kspace grid
nshiftk 1
shiftk 0.0 0.0 0.0
ngkpt 4 4 4 # Use ngkpt 4 4 6 for converged results

#Definition of the atom types
ntypat 2 # There are two types of atom
znucl 14 8 # The keyword "znucl" refers to the atomic number of the

#Definition of the atoms
natom 24 #
typat 8*1 16*2

xred
0.00000 0.00000 0.00000
0.00000 0.50000 0.50000
0.50000 0.50000 0.00000
0.50000 0.00000 0.50000
0.75000 0.25000 0.75000
0.25000 0.25000 0.25000
0.25000 0.75000 0.75000
0.75000 0.75000 0.25000
0.12500 0.12500 0.12500
0.87500 0.37500 0.62500
0.37500 0.62500 0.87500
0.62500 0.87500 0.37500
0.37500 0.87500 0.62500
0.87500 0.62500 0.37500
0.62500 0.37500 0.87500
0.12500 0.62500 0.62500
0.87500 0.87500 0.12500
0.37500 0.12500 0.37500
0.37500 0.37500 0.12500
0.87500 0.12500 0.87500
0.62500 0.12500 0.62500
0.12500 0.87500 0.87500
0.12500 0.37500 0.37500
0.62500 0.62500 0.12500

# natfix 4
# iatfix 1 2 3 4


nstep 500 # Maximal number of SCF cycles
iscf 17
# diemac 3.0
ecutsm 0.5

Have you tried to increase pawecutdg?

I have tried to increase pawecutdg up to 70, but total energy do not converge!

which polymorph of SiO2 are you studying? I don't recognize the input structure at all.

I study a SiO2 cristobalite ( Fd-3m). Then, I m going to calculate the SiO2(111) surface.

But in the cubic high temperature phase of cristobalite there is dynamic disorder of the oxygen sites, so the CIF files typically give the fractional occupancy of the O site as 0.1667. How are you taking this into account? If you ignore it, the structure probably shouldn't converge at all, I think this may be what you are seeing. The proper way to take it into accoun would be with a really large supercell.