# How to make ferro, feri and antiferro magnetic structure for QE

I have below &SYSTME information from my input file

      &CONTROL
calculation = 'scf'
etot_conv_thr =   1.0000000000d-04
forc_conv_thr =   1.0000000000d-04
outdir = './tmp/'
prefix = 'pwscf'
pseudo_dir = './'
tprnfor = .true.
tstress = .true.
verbosity = 'high'
/
&SYSTEM
degauss =   1.4699723600d-02
ecutrho =   3.6000000000d+02
ecutwfc =   4.5000000000d+01
ibrav = 0
nat = 12
nosym = .false.
ntyp = 2
occupations = 'smearing'
smearing = 'cold'
/
&ELECTRONS
conv_thr =   1.0000000000d-09
electron_maxstep = 100
mixing_beta =   3.0000000000d-01
/
ATOMIC_SPECIES
Ta       ta_pbe_v1.4.uspp.F.UPF
Ni       ni_pbe_v1.4.uspp.F.UPF
ATOMIC_POSITIONS crystal
Ta            0.0000000000       0.0000000000       0.2500000000
Ta            0.0000000000       0.0000000000       0.7500000000
Ta            0.5000000000       0.5000000000       0.7500000000
Ta            0.5000000000       0.5000000000       0.2500000000
Ni           0.1677000000       0.6677000000       0.0000000000
Ni           0.3323000000       0.1677000000       0.0000000000
Ni           0.8323000000       0.3323000000       0.0000000000
Ni           0.6677000000       0.8323000000       0.0000000000
Ni           0.1677000000       0.3323000000       0.5000000000
Ni           0.6677000000       0.1677000000       0.5000000000
Ni           0.8323000000       0.6677000000       0.5000000000
Ni           0.3323000000       0.8323000000       0.5000000000
K_POINTS automatic
7 7 8 0 0 0
CELL_PARAMETERS angstrom
4.9910000000       0.0000000000       0.0000000000
0.0000000000       4.9910000000       0.0000000000
0.0000000000       0.0000000000       4.2470000000


How can I make ferromagnetic, anti-ferromagnetic, and ferrimagnetic structures from it?

• I believe, that answer only covers the anti-ferromagnetic structure. How can we do anti-ferri and ferromagnetic in this case? Sep 9 '21 at 2:37
• It is ovious that for ferro magnetic start with equal magnetization on all species and same way you modify for ferri Sep 9 '21 at 7:10
• For AFM, I took degauss =2.0d-02 ecutrho =1.0d+03 ecutwfc= 9.0d+01 ibrav=0 nat=12 nosym=.false. nspin=2 ntyp=3 occupations = 'smearing' smearing = 'cold' starting_magnetization(1)=1 starting_magnetization(2)= -1 starting_magnetization(3) =0 ATOMIC_SPECIES Ta ta_pbe_v1.4.uspp.F.UPF Ni1 ni_pbe_v1.4.uspp.F.UPF Ni2 ni_pbe_v1.4.uspp.F.UPF ATOMIC_POSITIONS crystal Ta 0.00 0.00 0.25  Ta 0.00 0.00 0.75  Ta 0.50 0.50 0.75  Ta 0.50 0.50 0.25  Ni1 0.16 0.66 0.00  Ni2 0.33 0.16 0.00  and so no Sep 9 '21 at 8:34
• For FM, I took degauss =2.0d-02 ecutrho =1.0d+03 ecutwfc= 9.0d+01 ibrav=0 nat=12 nosym=.false. nspin=2 ntyp=2 occupations = 'smearing' smearing = 'cold' starting_magnetization=1 ATOMIC_SPECIES Ta ta_pbe_v1.4.uspp.F.UPF Ni ni_pbe_v1.4.uspp.F.UPF ATOMIC_POSITIONS crystal Ta 0.00 0.00 0.25  Ta 0.00 0.00 0.75  Ta 0.50 0.50 0.75  Ta 0.50 0.50 0.25  Ni 0.16 0.66 0.00  Ni 0.33 0.16 0.00  and so no Sep 9 '21 at 8:35

Since the question concerns only magnetic order, I am going to address only that, leaving out details about smearing and other parameters.

Let's take a simple A-type magnetic system from this illustration.

Note that there are primarily two types of magnetism: Intra-layer and inter-layer. I would advise the following procedure for creating your input file:

1. Determine what atoms in your system are magnetic - Is it the Ni, Ta etc..
2. Determine what kind of spin arrangement you require - pay attention to both intra-layer and inter-layer channels.

As an example, let me work with the system in your question. Let's assume the nickel atoms are magnetic. You can clearly notice that 4 nickel atoms are in the same plane, and they are separated by half a lattice constant along 'z'. I can outline some ideas so that you get the general gist.

a) A type antiferromagnet:

Ni1           0.1677000000       0.6677000000       0.0000000000
Ni1           0.3323000000       0.1677000000       0.0000000000
Ni1           0.8323000000       0.3323000000       0.0000000000
Ni1          0.6677000000       0.8323000000       0.0000000000
Ni2           0.1677000000       0.3323000000       0.5000000000
Ni2           0.6677000000       0.1677000000       0.5000000000
Ni2           0.8323000000       0.6677000000       0.5000000000
Ni2           0.3323000000       0.8323000000       0.5000000000


You would set starting_magnetization of Ni1 as 1 and starting_magnetization of Ni2 as -1.

b) C-type antiferromagnet:

Ni1           0.1677000000       0.6677000000       0.0000000000
Ni2           0.3323000000       0.1677000000       0.0000000000
Ni1           0.8323000000       0.3323000000       0.0000000000
Ni2          0.6677000000       0.8323000000       0.0000000000
Ni1           0.1677000000       0.3323000000       0.5000000000
Ni2           0.6677000000       0.1677000000       0.5000000000
Ni1           0.8323000000       0.6677000000       0.5000000000
Ni2           0.3323000000       0.8323000000       0.5000000000


You would set starting_magnetization of Ni1 as 1 and starting_magnetization of Ni2 as -1.

I hope you get the idea. For ferrimagnets, you would set the magnetization values so that the net value (whether intra or inter, depending on your system) is non-zero. Also keep in mind that for some systems the magnetic ordering of spins is in a specific direction - for example NiO. In these cases, I would advise you to thoroughly research the arrangement of spins before formulating your input file.

• Thank you Sir for your detailed response. How can I decide the type of antiferromagnetism if the magnetic system is not in a perfect plane? I have added a picture where Ni atoms are not in a perfect layer. My question may be silly but I really confused. Sep 14 '21 at 10:08
• 'How can I decide the type of antiferromagnetism' - Unless this is a completely fictional system, you can read up some papers that report the magnetic structure of your system. You should find your answer there. Sep 14 '21 at 19:48
• Thanks Sir, I will do that. Sep 15 '21 at 8:14