ABINIT Discussion Forums

Dear all users,
i have to calculate band structure with spin orbit coupling.i have studied tutorials and posted before for forums but i have not get any answer about this calculation. i use this input file to calculate band structure but when my job would be run and i plot it with abinitBandstructuremaker i have incorrect band structure.would you please look at my input file and guide me what i have done mistake?
input file for band structure with spin orbit coupling:
ndtset 2
kptopt1 4
nshiftk1 1
shiftk1 0.5 0.5 0.5
ngkpt1 20 20 20
prtden1 1
toldfe1 1.0d-6
nspinor1 2
nspden1 4
istwfk1 770*1
nsppol1 1
iscf2 -2
getden2 -1
kptopt2 -6
nkpt2 121
kptnrm2 3
nband2 20
ndivk2 20 20 20 20 20 20
kptbounds2 0.0 0.0 0.0
1.38108056 0.0 0.17174403
1.33836604 0.0 0.51523210
0.0 0.0 0.51523210
1.33836604 0.7727059 0.51523
0.66918302 1.1590589 0.51523
0.69054022 1.1960507 0.34348
tolwfr2 1.0d-12
enunit2 1
acell 3*10.45 angstrom
angdeg 3*24.13
ntypat 2
znucl 83 52
natom 5
typat 1 1 2 2 2
natrd 3
spgroup 166
spgaxor 2
brvltt -1
xred 0.399 0.399 0.399
0.000 0.000 0.000
0.792 0.792 0.792
ecut 340 ev
nstep 50
diemac 2
Best regards
Mina.

Are you doing Bi2Te3? It does not have net total spin, so why do you use nspden= 4?
nspden=4 is used when there is ferromagnetism.

nsppol 1
nspinor 2
nspden 1

btw, can you also upload the error message in the log file? It is more informative than input only.

Dear ljludwig,
Thank you for your attention to my question,i do not have any error massage in log file.my jub would be run without any error but when i plot band structure i get incorrect picture.
Best regards.
Mina.

One thing you may want to do is to tell what do you mean by "incorrect band structure".

If it is just different from literature, it is not really "incorrect".

First of all, I don't feel like ecut =340eV is enough, if I were you, I'll do two things:

1) Do a non-relativistic LDA and GGA calculation, and plot the "correct" band structure without S-O coupling, and plot the DoS.
this gives a sense about how the band looks like- the S-O coupling one should not be far away from the non-relativistic calculation.

2) Test ecut and ngkpt to see a convergence. May also need dense the K-mesh(for narrow band Semiconductor, maybe > 24*24*24, even 39*39*39 for PbTe, etc.)

Dear liladwig,
Thanks a lot for your help.my plot is different from plot in articles.i have used those things that you said and i will send my results when my job will be run.
Best regards.
Mina

Dear ljludwig,
i set nspden=1 in my input file and increased ngkpt as you said. i have band structure like this(i have attached my plot). it is better than my provious plot but would you please guide me why i have few band under fermi energy?
Best regards.
Mina.

I agree with ljludwig - test ecut and ngkpt to see a convergence.
Maybe you need better convergence criteria - use toldfe 1.0d-12 or tolwfr 1.0d-12 or tolvrs 1.0d-22 like in tpaw1_4.in, also try LDA+U using paw psps (paw tutorials) from http://www.abinit.org/downloads/PAW2 . Maybe LDA+U approach allows to fix incorrect band results

Dear Maxim,
thanks for your help,yesterday i saw in my out file Fermi (or HOMO) energy (hartree) = 0.19279 Average Vxc (hartree)= -0.3953.and i think my all band energy have shift.and i think it can make line of energy fermi in my band plot to be in incorrect place.would you please guide me about this problem?
Best regards.
Mina