Lattice Parameters and Geometries

Getting the lattice parameters

So far, we do not perform geometry optimization. This means that you will start your setup from a fixed arrangement of atoms in a unit cell. Likely you already know the geometry, unit cell distance, etc. from the literature. If you do not, you can perform a literature search for this particular compound using this experimental database:

https://icsd.fiz-karlsruhe.de/search/basic.xhtml

There are several alternative databases where you can get additional information. Among them is the materials project:

https://materialsproject.org

These databases or the linked papers will provide you with the unit cell information, the space-group information, and the arrangement of atoms in the unit cell. Generically you will get three lengths (a, b, c) and three angles (α,β,γ). You will use this information to generate the three unit cell vectors as a 3x3 matrix. Check your favorite textbook for instructions; for instance in a simple cubic lattice the vectors (in units of the lattice spacing a) are $\alpha$

$$ \begin{align*} a_1=\begin{pmatrix}1&0&0\end{pmatrix} \cr a_2=\begin{pmatrix}0&1&0\end{pmatrix} \cr a_3=\begin{pmatrix}0&0&1\end{pmatrix} \end{align*} $$

In a body-centered cubic system they would be

$$ \begin{align*} a_1=\begin{pmatrix}1/2&1/2&-1/2\end{pmatrix}\cr a_2=\begin{pmatrix}-1/2&1/2&1/2\end{pmatrix}\cr a_3=\begin{pmatrix}1/2&-1/2&1/2\end{pmatrix} \end{align*}$$

…and in an fcc lattice they would be

$$\begin{align*} a_1=\begin{pmatrix}1/2&1/2&0\end{pmatrix}\cr a_2=\begin{pmatrix}0&1/2&1/2\end{pmatrix}\cr a_3=\begin{pmatrix}1/2&0&1/2\end{pmatrix} \end{align*}$$

Note that the conventional unit cell is often much larger, so make sure to take the primitive cell. Also note that anti-ferromagnetic order will enlarge the non-magnetic unit cell, since it will lead to inequivalent atoms of the same type.

The example of MnO

As an example, consider MnO. According to the materials databases the material is cubic, crystallizing in the space group Fm-3m (225). The lattice distance is

a=4.446

but for consistency with our paper we will use

a=4.445

The structure is face centered cubic. The Mn atoms are located at the origin, the oxygen atoms along the diagonal at $\begin{pmatrix}a/2&a/2&a/2\end{pmatrix}$. Because of anti-ferromagnetic ordering, we need to repeat the Mn atom along the diagonal, such that a second Mn atom sits at $\begin{pmatrix}a &a&a\end{pmatrix}$, and a second oxygen one at $\begin{pmatrix}3a/2&3a/2&3a/2\end{pmatrix}$. The code will then require the following input: a file a.dat that describes the unit cell vectors, and a file atom.dat that has the location of the atoms.

atom.dat:

Mn 0.0 0.0 0.0
O 2.22250, 2.22250, 2.22250
Mn 4.44500, 4.44500, 4.44500
O 6.66750, 6.66750, 6.66750

a.dat:

4.44500, 2.22250, 2.22250
2.22250, 4.44500, 2.22250
2.22250, 2.22250, 4.44500

Using ASE scripts to determine lattice geometry and atom positions

The Atomic Simulation Environment is a collection of useful tools for doing solid state simulations. You can obtain the information used above directly from the ASE crystal environment, using a modification of the following script. Replace the unit cell parameters and atom positions with the information you find in the materialsproject or karlsruhe databases. Make sure you enter the parameters for the conventional unit cell; the script will then generate the parameters for the primitive or conventional cell. Here is an example for LiH (materials project compound mp-23703)

import numpy as np
from ase.spacegroup import crystal
from ase.spacegroup import Spacegroup

# Unit cell parameters of conventional cell
a, b, c = 4.02, 4.02, 4.02
alpha, beta, gamma = 90, 90, 90
group = 225

'''
Construct unit cell. 
a) if primite_cell=False, a convention unit cell will be generated.
b) bases should be given in units of a, b, c 
'''

cc = crystal(symbols=['Li','H'],
             basis=[(0.0, 0.0, 0.0),(0.0, 0.0, 0.5)],
             spacegroup=group,
             cellpar=[a, b, c, alpha, beta, gamma], primitive_cell=True)

# Lattice vector
print(cc.get_cell())

# Atoms inside the unit cell
print(cc.get_chemical_symbols())

# Atom positions  
print(cc.get_positions())

# Special k points
'''
Get Bravais lattice of unit cell cc. 
Note that the Bravais lattice of a conventional unit cell is different 
from the one of a primitive unit cell. So does the special points. 
'''
lat = cc.cell.get_bravais_lattice()

# Special k points
print(lat.get_special_points())

# k-path
path = cc.cell.bandpath('GX', npoints=50)

# The kpoints are given in units of the reciprocal lattice vector.
print("Size of k-path: ", path.kpts.shape)

# Details of the space group
sp = Spacegroup(group)
print(sp)

Right-handedness of coordinate system

Beware: the coordinate system for the lattices has to be right-handed (due to some internal pySCF workings). If it is not, the code will report

Lattice are not in right-handed coordinate system. Please correct your lattice vectors

In that case please swap the order of your lattice vectors until they are in the proper ordering.

Generating High-Symmetry Path Using ASE Scripts

When creating a band structure plot, you might need to generate a high-symmetry path. To do this, you can follow the steps outlined below (you can also refer to this paper for more information about k-paths: https://doi.org/10.1016/j.commatsci.2010.05.010):

import numpy as np
from ase.dft.kpoints import get_special_points, bandpath, special_paths
# Define lattice vectors in Cartesian coordinates 
lattice_vectors = np.array([[4.1705 , 2.08525, 2.08525], 
                            [2.08525, 4.1705 , 2.08525], 
                            [2.08525, 2.08525, 4.1705 ]])

# Output high symmetry points in reciprocal lattice vector units
high_symmetry_points = get_special_points(lattice_vectors)
print("Available high symmetry points are:\n", high_symmetry_points, '\n')

# Output default high symmetry path
default_path = special_paths["rhombohedral type 1"]
print("Default high symmetry path is:\n", default_path, '\n')

# Specify high symmetry points along the desired path
# Note: ',' denotes a jump between neighboring points, meaning no 
# additional points will be sampled between them
path = "GLB1,BZGX,QFP1Z,LP" 
num_points = 200 # Define the total number of points 

# Generate high symmetry path
kpath = bandpath(path, lattice_vectors, npoints=num_points) 
path, special_points, labels = kpath.get_linear_kpoint_axis() 
print("Scaled high symmetry points along the path are:\n", kpath.kpts, '\n')

# The following postprocessing allows the path, special_points, and labels 
# to be directly used for plotting.

# Find indices where adjacent elements in special_points are equal
indices = np.where(special_points[:-1] == special_points[1:])[0]
# For each index, create a combined label and replace the original label with it
for index in indices:
    combined_label = labels[index] + "|" + labels[index+1]
    labels[index] = ""
    labels[index+1] = combined_label