3
$\begingroup$

I'm a beginner in DFT and I want to study the properties of NiO and doped-NiO materials using Quantum ESPRESSO.I found in the internet that I should introduce the spin magnetization in the Input file (or CIF file) because NiO is an antiferromagnetic material. The CIF file of conventional NiO (extracted from Materials Project website) is presented bellow:

# generated using pymatgen
data_NiO
_symmetry_space_group_name_H-M   'P 1'
_cell_length_a   4.21661958
_cell_length_b   4.21661958
_cell_length_c   4.21661958
_cell_angle_alpha   90.00000000
_cell_angle_beta   90.00000000
_cell_angle_gamma   90.00000000
_symmetry_Int_Tables_number   1
_chemical_formula_structural   NiO
_chemical_formula_sum   'Ni4 O4'
_cell_volume   74.97099306
_cell_formula_units_Z   4
loop_
 _symmetry_equiv_pos_site_id
 _symmetry_equiv_pos_as_xyz
  1  'x, y, z'
loop_
 _atom_site_type_symbol
 _atom_site_label
 _atom_site_symmetry_multiplicity
 _atom_site_fract_x
 _atom_site_fract_y
 _atom_site_fract_z
 _atom_site_occupancy
  Ni  Ni0  1  0.00000000  0.00000000  0.00000000  1
  Ni  Ni1  1  0.00000000  0.50000000  0.50000000  1
  Ni  Ni2  1  0.50000000  0.00000000  0.50000000  1
  Ni  Ni3  1  0.50000000  0.50000000  0.00000000  1
  O  O4  1  0.50000000  0.00000000  0.00000000  1
  O  O5  1  0.50000000  0.50000000  0.50000000  1
  O  O6  1  0.00000000  0.00000000  0.50000000  1
  O  O7  1  0.00000000  0.50000000  0.00000000  1

I tried to model antiferromagnetic NiO by adopting two Ni-atom positions at Ni0 (0.00000000 0.00000000 0.00000000) and Ni1 (0.00000000 0.50000000 0.50000000) for the spin up and two Ni-atom positions at Ni2 (0.50000000 0.00000000 0.50000000) and Ni3 (0.50000000 0.50000000 0.00000000) for spin down [like in the CIF file above] then I wanted to pass the file through FINDSYM website. However I found in some papers a more complicated model like the one in the picture below:

enter image description here

Is the model in the picture the one that I should use, and how to include that in my input (or CIF) file?

$\endgroup$
8
  • $\begingroup$ +1 but can you please proofread this post then edit it? For example in the first sentence alone: "I`m" should be "I'm", "Dft" should be "DFT", and "doped NiO ." should be "doped NiO." $\endgroup$ Feb 14 at 15:16
  • $\begingroup$ @NikeDattani Thank you. I did some modifications. Is it ok now? $\endgroup$
    – Camilla
    Feb 14 at 15:30
  • $\begingroup$ Please see my previous comment more carefully :) $\endgroup$ Feb 14 at 15:38
  • $\begingroup$ @NikeDattani your comment appears to me like that : "I`m" should be "I'm", "Dft" should be "DFT", and "doped NiO ." should be "doped NiO." . I corrected : I'm, DFT and some other sentences. But for doped NiO you wrote it in the same way in your comment. $\endgroup$
    – Camilla
    Feb 14 at 15:43
  • $\begingroup$ Please check it again more carefully :) $\endgroup$ Feb 14 at 15:48

1 Answer 1

3
$\begingroup$

The model in the picture is alright as far as it goes, but it is not a magnetic unit cell, it is only a structural unit cell. If the cubic frame were a unit cell, the atoms at each corner would be identical by translational symmetry; however, in this picture half of the corner atoms are "spin up" and half "spin down", so they are not simple translations of each other (e.g. look at the two atoms coupled by "J2").

The smallest magnetic unit cell for NiO is (I think) the one oriented along the (111) direction, i.e. along the diagonal of the cube. Along this direction, the spins on the Ni alternate between up and down.

You should also be aware that the self-interaction error of Ni $d$-electrons is quite large with common local or semi-local exchange-correlation functionals. I'm not aware of any LDAs which would give an AFM insulating state, they usually converge to a non-magnetic metallic state. Some GGAs also have this problem, although PBE does predict the antiferromagnetic insulating state, but the band-gap is rather small and the state is quite fragile.

The usual approach to modelling NiO, and other systems where self-interaction is a problem, is to add a small Hubbard U term to the Hamiltonian ("DFT+U") which partially counteracts the effects of self-interaction. Alternatively, the RSCAN family of meta-GGAs do fairly well, and screened non-local exchange methods can also give good results.

$\endgroup$
2
  • $\begingroup$ Does that mean if I want to model the magnetization of a structure i have always to look at the smallest unit cell and alternate it between up and down then based on that I alternate the spins of the other atoms in the cell? $\endgroup$
    – Camilla
    Feb 16 at 8:25
  • $\begingroup$ @Camilla the real answer is that you need to model the magnetic unit cell, which means the cell which is perfectly repeated (including any magnetic moments). For simple AFM materials this means alternating the atomic magnetic moments for the appropriate atoms, but there are many more expressions of magnetism including ferrimagnetism, non-collinear magnetism, AFM only between particular planes etc. $\endgroup$ Feb 20 at 0:30

You must log in to answer this question.