ABINIT Discussion Forums

hello all:
i have abinit 5.3.4
i want to surface relaxation for iron
input:

#Definition of occupation numbers
occopt 3
tsmear 0.01 Ha

#Definition of the k-point grids
ngkpt 6 6 1


# difinition of the atom types
ntypat 1
znucl 26

#difinition of the atoms
natom 20
typat 20*1
xred
0.0000000000000000 0.0000000000000000 0.0000000000000000
0.5000000000000000 0.0000000000000000 0.0000000000000000
0.0000000000000000 0.5000000000000000 0.0000000000000000
0.5000000000000000 0.5000000000000000 0.0000000000000000
0.2500000000000000 0.2500000000000000 0.0799582481304313
0.2500000000000000 0.7500000000000000 0.0799529537486308
0.7500000000000000 0.2500000000000000 0.0799610167292499
0.7500000000000000 0.7500000000000000 0.0799574340619102
0.0001863529520918 0.9999519332296671 0.1653586580251343
0.5000815207698253 0.0000422783932663 0.1653848704285323
0.0001865098933962 0.5000481415744105 0.1653587030743893
0.5000815511605836 0.4999577312334793 0.1653848244311575
0.2500792188342871 0.2500000123809158 0.2508257731625037
0.2502450761203718 0.7500000140879308 0.2508094734834501
0.7503051129573635 0.2500000514714813 0.2507400699971427
0.7502400378438878 0.7499999848301826 0.2507280635475695
0.0001551838329945 0.9998148745562929 0.3306609755962810
0.5003873863229913 0.0000687621433144 0.3305954374870907
0.0001551455089910 0.5001851091057361 0.3306606635116438
0.5003873995145156 0.4999312455191074 0.3305957187603212
# dififnition of unit cell
acell 3*5.4292 Bohr
rprim 2.0 0.0 0.0
0.0 2.0 0.0
0.0 0.0 6.0
chkprim 0

#Definition of the SCF procedure
nstep 500
toldfe 1.0d-6

ecut 50 Ha

ixc 11
iscf 7

ionmov 3
tolmxf 5.0d-4

output file is very good but it has a warning:
scprqt: WARNING -
nstep= 500 was not enough SCF cycles to converge;
maximum energy difference= 1.554E+00 exceeds toldfe= 1.000E-06

how i can fix it, please help me.

nobody ? no idea?
thanks

Hi!

It's hard to tell what's exactly the source of bad convergence of Etot for your case. It could be several things, like pseudopotential, scf option,... Did you check the convergence wrt ecut, ngkpt and vacuum slab?
There are also some scf convergence options which can be increased to improve it, like nnsclo, nline and npulayit. They help somtimes.

Best regards,

Igor