DFT in Practice

Plane wave expansion & the Brillouin zone integration
DFT in practice : Part I

plane wave expansion & the Brillouin zone integration
Ersen Mete
Department of Physics
Balkesir University, Balkesir - Turkey
August 13, 2009 - NanoDFT09,

Izmir Institute of Technology,

Izmir
Ersen Mete DFT in practice : Part I
Outline
Plane wave expansion
Kohn-Sham equations for periodic systems
Potential terms in plane waves
Brillouin zone integration
k-point sampling & Monkhorst-Pack grid

Integration in irreducible Brillouin zone
Smearing methods
Kohn-Sham equations to be solved self-consistently
_
1
2
2
+ V
e
(r )
_
i
(r ) =
i
i
(r )
V
e
(r ) =
ext
(r ) +
_
dr

n(r

)
|r

r |
. .
Hartree potential
+
E
XC
[n]
n(r )
. .
Exchangecorrelation
potential
n(r ) =
i
|
i
(r )|
2
If the exchange-correlation functional is exact the total energy and the
density will be exact.
However, Kohn-sham single particle wavefunctions and eigenenergies do
not correspond to any exact physical value.
Self-consistent Kohn-Sham loop
Basis set representation of Kohn-Sham orbitals
Atomic Orbitals (AO)
(r ) =
(r )Y
m
(, ) where
(r ) =
_
e
r
2
Gaussian
e
r
Slater
molecular structures : slightly distorted atoms
small basis sets can give good results
easy to represent vacuum regions
basis functions are attached to nuclear positions
non-orthogonal
basis set superposition errors (BSSE)
Basis set representation of Kohn-Sham orbitals
Plane waves
k
(r ) = e
i
kr
orthogonal
periodic
(solids)
(nite systems) = PBC with supercell approach
practical for Fourier transfom and computation of matrix elements
atomic wave functions require large number of plane waves
nonlocalized = inecient for parallelization
Bravais lattice, primitive cell & Brillouin zone
Direct lattice : a =
_
a
1
, a
2
, a
3
Reciprocal lattice derives from connement

and Fourier analysis of periodic functions :
e
i
b
i
a
j
=
ij
=

b
i
a
j
= 2 (=integer)
Reciprocal latt. : b = 2(a)
1
=
_
b
1
,
b
2
,
b
3
Cell volume :
cell
= det(a)
BZ volume :
BZ
= det(b) = (2)
3
/
cell
Direct lattice vectors :

T(n
1
, n
2
, n
3
) = n
1
a
1
+ n
2
a
2
+ n
3
a
3

T
n
Reciprocal latt. vectors :

G(m
1
, m
2
, m
3
) = m
1
b
1
+ m
2
b
2
+ m
3
b
3

G
m
1st BZ (Wigner-Seitz cell) : boundaries are the bisecting planes of

G
vectors where Bragg scattering occurs.
Born-von Karman supercell approach
point defect slab molecule
i ,
k
(r + N
j
a
j
) =
i ,
k
(r )
Supercell must be suciently large to maintain isolation.
Plane wave expansion of the single particle wavefunctions
Blochs theorem
For a periodic Hamiltonian :
i ,
k
(r ) = e
i
kr
1
N
cell
u
i ,
k
(r )
_
i ,
k
(r +a
j
) = e
i
ka
j
i ,
k
(r ))
_
where

k is in the rst BZ and periodic u
i ,
k
(r ) can be expanded in a
Fourier series :
u
i ,
k
(r ) =
1
cell
m
c
i ,m
(
k)e
i
G
m
r
_
u
i ,
k
(r +a
j
) = u
i ,
k
(r ))
_
Equivalently,
i ,
k
(r ) =
m
c
i ,m
1
e
i (
k+
G
m
)r
q
c
i ,q
|q (q =

k +

G
m
)
Fourier representation of the Kohn-Sham equations
For a crystal,
Implications of Bloch states
i ,
k
is a superposition of PWs with q wavevectors which dier by
G
m
to maintain the PBC.
k is well-dened crystal momentum which is conserved.

Potential has the lattice periodicity as u
i ,
k
(r ).
V
e
(r ) =
m
V
e
(
G
m
)e
i
G
m
r
Only

G
m
vectors are allowed in the Fourier expansion.
The non-zero matrix elements are,
q
|V
e
|q =
m
V(
G
m
)q
|e
i
G
m
r
|q =
m
V
e
(
G
m
)
q
q,
G
m
Substituting these Bloch states into the Kohn-Sham equations,
multiplying by q
| from the left and integrating over r gives a set of

matrix equations for any given

k.
_
dr q
|
_
_
_
_
1
2
2
+ V
e
(r )
_
q
c
i ,m
|q =
i
q
c
i ,m
|q
_
_
_
_
1
2
|
k +

G
m
|
2
m,m
+ V
e
(
G
m
G
m
)
_
c
i ,m
=
i
c
i ,m
Matrix diagonalization needed, but still easier to solve.
Eigenvectors and eigenenergies for each

k are independent in the 1st
BZ.
In the large = N
cell
cell
limit,

k points become a dense continuum
and
i
(
k) become continuous bands.

Hartree potential in plane waves
V
H
(
G) =
1
cell
_
dr e
i
Gr
_
_
dr

n(
)e
i
|r r
|
_
=
1
cell
n(
)
_
dr e
i (
G)r
_
du
e
i
u
|u|
=
1
cell
n(
)
cell
G,
2
_

0
_
1
1
e
iG
u cos
u
u
2
d(cos )du
= 2n(
G)
_

0
e
iGu
e
iGu
iG
du = 4
n(
G)
G
_

0
sin(Gu)du
= 4
n(
G)
G
lim
0
_

0
e
u
sin(Gu)du
= 4
n(
G)
G
lim
0
G
2
+ G
2
= 4
n(
G)
G
2
(G = 0)
Exchange-correlation potential in plane waves
V
XC
(
G) =
1
cell
_
dr e
i
Gr

n
_
n(r )
xc
([n],r )
_
=
1
cell
_
dr e
i
Gr
_
_
_
xc
([n],
)e
i
r
+
G,
G
n(
G)e
i
Gr

xc
([n],
)
n
e
i
r
_
_
_
=
1
cell
_
_
_
xc
([n],
)
_
dr e
i (
G)r
+
G,
G
n(
G)
xc
([n],
)
n
_
dr e
i (
G+
)r
_
_
_
=
1
cell
_
_
_
xc
([n],
)
cell
G,
G,
G
n(
G)
xc
([n],
)
n

cell
G,
_
_
_
=
xc
([n],
G) +
n(
xc
([n],
)
n
Fourier representation of E
H
and E
XC
terms
By similar arguments,
E
H
=
1
2
_ _
dr dr

n(r )n(r
)
|r r
|
= 2
cell
G=0
n(
G)
2
G
2
E
XC
=
_
dr n(r )
xc
(r ) =
cell
G
n(
G)
xc
(
G)
External potential in terms of structure and form factors
Ionic potential as a superposition of isolated atomic potentials
ext
(r ) =
n
sp
=1
n
j =1
T
V
(r
,j

T)
n
sp
species, n
atoms at
,j
for , the set of translation vectors

T
V
ext
(
G) =
1
dr
ext
(r )e
i
Gr
=
1
n
sp
=1
n
j =1
_
duV
(u)e
i
G(u+
,j
)
T
e
i
T
. .
N
cell
=
1
cell
n
sp
=1
_
duV
(u)e
i
Gu
. .
G)
n
j =1
e
i
G
,j
. .
S
G)
n
sp
=1
cell
S
G)V
G)
This form is particularly useful when V
(r ) = V
(|r |)!
Form factor for a spherically symmetric ionic potential
V
G) =
1
_

0
_
1
1
_
2
0
V
(r )e
i
Gr
r
2
dd(cos ) dr
=
2
_
V
(r )
e
iGr (cos )
iGr
1
-1
r
2
dr
=
2
_

0
e
iGr
e
-iGr
iGr
r
2
dr
=
4
_

0
j
0
(|
G|r )V
(r )r
2
dr
= V
(|
G|)
Cuto : nite basis set
Computationally, a complete expansion in terms of innitely many plane
waves is not possible.
The coecients, c
m
(
k), for the lowest orbitals decrease exponentially

with increasing PW kinetic energy (
k +

G
m
)
2
/2.
A cuto energy value, E
cut
,
determines the number of PWs (N
pw
)
in the expansion, satisfying,
(
k+

G
m
)
2
2
< E
cut
(PW sphere)
N
pw
is a discontinuous function of the PW kinetic energy cuto.
Basis set size depends only on the computational cell size and the cuto
energy value.
Electron density in the plane wave basis
n(r ) =
1
N
k
occ
k,i
n
i ,
k
(r ) where n
i ,
k
(r ) = |
i ,
k
(r )|
2
Then, in terms of Bloch states,
n
i ,
k
(r ) =
1
m,m
i ,m
(
k)c
i ,m
(
k)e
i (
G
m

G
m
)r
and
n
i ,
G
(r ) =
1
m
c
i ,m
(
k)c
i ,m
(
k)
where

G
m
=

G
m
+

G = sphere of n(
G) has a double radius.

Brillouin zone integration
Properties like the electron density, total energy, etc. can be evaluated by
intergration over

k inside the BZ.
for the i
th
band of a function f
i
(
k)
1
BZ
_
BZ
d
k f
i
(
k) =

f
i
=
1
N
k
k
f
i
(
k)
Example :
a 2D square lattice
25

k-points in the BZ
f
i
= f
i
(
k
1
) + . . . + f
i
(
k
25
)
k-point sampling : uniform vs non-uniform

Moreno and Soler [Phys.Rev.B 45,24] :
A mesh with uniformly distributed

k-points is preferred.
Example : A rectangular lattice
They have nearly the same number of

k-points.
(a) : Isotropic sampling
(b) : ner sampling vertically
poor sampling horizontally
Monkhorst-Pack grid
A uniform mesh of

k-points can be generated by Monkhorst-Pack
procedure
k
n
1
,n
2
,n
3
=
3
1
2n
i
N
i
1
2N
i
G
i
where N
i
is the number of

k-points in each direction and n
i
= 1, . . . , N
i
.
Example : 441 MP grid for 2D square lattice
G
1
=

b
1
=

a
k
x
,

G
2
=

b
2
=

a
k
y
,

G
3
= 0
k
1,1
=
_
-3
8
,
-3
8
_

k
1,3
=
_
-3
8
,
1
8
_
k
1,2
=
_
-3
8
,
-1
8
_

k
1,4
=
_
-3
8
,
3
8
_
Shifted-grids
Example : a 2D square lattice
Centered on Centered around
=
25

k-points
[shifted by (1/8,1/8,0)]
16

k-points
They both have the same

k-point density in the reciprocal space!
Time-reversal symmetry
Kramers theorem
Let the hamiltonian, H, be invariant under time-reversal,
THT
1
= H (in the absence of magnetic elds!)
then
H =
TH = HT = T
T
(r , t) is also a solution of the same Schrodinger equation with

the same eigenvalue.
The solutions and T are orthogonal, |T = 0
Bloch states,
i ,-
k
and
i ,
k
satisfy the condition (r +

T
n
) = e
i
T
n
(r )
=
i ,-
k
=
i ,
k
Kramers theorem implies inversion symmetry in the reciprocal space!
Lattice symmetry and the Irreducible Brillouin Zone
We use symmetry of the Bravais lattice to reduce the calculation to a
summation over

k inside the IBZ.
Example : IBZ of the 2D square lattice
Special

k-points :
(0, 0) : full symmetry of the point group.
X
_
1
2
, 0
_
: E, C
2
,
x
,
y
M
_
1
2
,
1
2
_
: full symmetry of the point group.
Then, we can unfold the IBZ by the symmetry operations to get the
solution for the full BZ.
Preserve symmetry
Example : Hexagonal cell
Shift the

k-point mesh to preserve hexagonal symmetry!
In this case, even meshes break the symmetry
A mesh centered on preserves the symmetry.
Integration in the IBZ
Generate a uniform mesh in reciprocal space
Shift the mesh if required
Employ all symmetry operations of the Bravais lattice to each of the
k-points
Select the

k-points which fall into the IBZ
Calculate weights,
k
, for each of the selected

k-points
k
=
number of symmetry connected

k-points
total number of

k-points in the BZ
IBZ integration is then given by
f
i
=
IBZ
k
f
i
(
k)
Example : 441 MP grid for 2D square lattice
441 MP grid = 16

k-points in the BZ
4 equivalent

k
4,4
=
_
3
8
,
3
8
_
= w
k
=
1
4
4 equivalent

k
3,3
=
_
1
8
,
3
8
_
= w
k
=
1
4
8 equivalent

k
4,3
=
_
3
8
,
1
8
_
= w
k
=
1
2
Then, the BZ integration reduces to,
f
i
=
1
4
f
i
(
k
4,4
) +
1
4
f
i
(
k
3,3
) +
1
2
f
i
(
k
4,3
)
Smearing methods : fractional occupation numbers
Fermi level : the energy of the highest occupied band.
IBZ integration over the lled states :

f
i
=
IBZ
k
f
i
(
k)(
i
(
k)
F
)
For insulators and semiconductors, DOS goes to zero smoothly before the gap.
For metals, the resolution of the step function at the Fermi level is very dicult
in plane waves.
Trick is to replace sharp (
i
(
k)
F
) function with a smoother
f ({
i
(
k)}) function allowing partial occupancies at the Fermi level.

Fermi-Dirac smearing
In order to overcome the discontinuity of the functions at the Fermi level,
f {
i
(
k)} =
1
e
(
i
(
k)
F
)/
+ 1
where = k
B
T
T 0 : Fermi-Dirac distribution approaches to the step function.
Finite temperature T introduces entropy to the system of non-interacting
particles,
S(f ) = [f ln f + (1 f ) ln(1 f )]
and the Free energy is given by,
F = E
i
S(f
i
)
Drawback : reduced occupancies below
F
are not compensated new
occupancies above
F
.
Gaussian smearing
An approximate step function is obtained by integration of a
Gaussian-approximated delta function.
f {(
i
k
)} =
1
2
_
1 erf
_
k

F
__
Smearing parameter, , has no physical interpretation.
Entropy and the free energy cannot be written in terms of f .
S
_

F
_
=
1
2
exp
_
_

F
_
2
_
Method of Methfessel-Paxton
To overcome the drawback introduced by the Fermi-Dirac smearing,
Expand the step function in a complete set of orthogonal Hermite
functions. (Hermite polynomials multiplied by Gaussians)
(x) D
N
=
N
n
A
n
H
2n
e
x
2
(H
2n
: (x) is even)
(x) S
N
= 1
_

D
N
(t)dt
S
0
=
1
2
(1 erf(x)) (Gaussian smearing)
S
N
= S
0
(x) +
N
n=1
A
n
H
2n1
(x)e
x
2
Yields negative occupation numbers!
Marzari-Vanderbilt : cold smearing
Amends the negative occupation numbers introduced by
Methfessel-Paxton.
The delta function is approximated by a Gaussian multiplied by a rst
order polynomial,
(x) =
1
e
[x(1/
2]
2
(2
2x)
where
x =

i
k

F

Linear tetrahedron method
Divide the BZ into tetrahedra
Interpolate the function X
i
within these tetrahedra
Integrate the interpolated function
to obtain

X
i

DFT in Practice

Transféré par

Informations du document

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

DFT in Practice

Transféré par

Droits d'auteur :

Formats disponibles

Plane wave expansion & the Brillouin zone integration

DFT in practice : Part I

k-point sampling & Monkhorst-Pack grid

Reciprocal lattice derives from connement

k is well-dened crystal momentum which is conserved.

| from the left and integrating over r gives a set of

k) become continuous bands.

k), for the lowest orbitals decrease exponentially

G) has a double radius.

k-point sampling : uniform vs non-uniform

(r , t) is also a solution of the same Schrodinger equation with

k)}) function allowing partial occupancies at the Fermi level.

Ersen Mete DFT in practice : Part I

Vous aimerez peut-être aussi