0% found this document useful (0 votes)
224 views12 pages

Introduction To Density Functional Theory (2022)

This document provides an introduction to density functional theory (DFT). It discusses how DFT can be used to model solid materials by representing them as interacting electrons and nuclei. Solving the many-body Schrodinger equation for such systems is computationally intractable. DFT overcomes this by replacing interacting electrons with non-interacting electrons under an effective exchange-correlation potential derived from a universal functional. This allows properties of solids to be predicted using only electron density as input rather than complex wavefunctions. Approximations like the Born-Oppenheimer approximation are also discussed which further simplify the modeling.

Uploaded by

APURI SIMON B
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
224 views12 pages

Introduction To Density Functional Theory (2022)

This document provides an introduction to density functional theory (DFT). It discusses how DFT can be used to model solid materials by representing them as interacting electrons and nuclei. Solving the many-body Schrodinger equation for such systems is computationally intractable. DFT overcomes this by replacing interacting electrons with non-interacting electrons under an effective exchange-correlation potential derived from a universal functional. This allows properties of solids to be predicted using only electron density as input rather than complex wavefunctions. Approximations like the Born-Oppenheimer approximation are also discussed which further simplify the modeling.

Uploaded by

APURI SIMON B
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 12

INTRODUCTION TO DENSITY FUNCTIONAL THEORY

SIMON BAJONGDO APURI (MPHIL- PHYSICS)

University Of Ghana, School of Basic and Applied Sciences, Department of Physics

Post Office Box LG25, Legon


Ghana- Greater Accra Region, www.ug.edu.gh
(20/JNUARY/2022)

1.1 INTRODUCTION

In our quest to identify materials with desirable properties, we have to understand the
structural composition of these materials. Identifying and controlling the behaviour of the
elementary particles (atoms, molecules and ions) in these materials is very useful in science
and engineering. Scientists strived over the centuries to formulate refined methods that can
model and predict the properties of these solid materials around us. Equations like the
Schrodinger’s equation and the Dirac’s equation are very useful for solving problems
pertaining to these solid materials but these equations become almost impossible to solve
where the number of particles is of the order of Avogadro’s number( N A  6.022 1023 ). In
the last few decades, an alternative approach to solving these equations was developed and it
is called the density functional theory (DFT).

The density functional theory (DFT) is a quantum mechanical computational modelling


method mostly used in solid state physics, solid state chemistry and material science for
describing the properties and the electronic structure of these materials. The focus of this text
is to give a conceptual explanation of the density functional theory and how this self-
consistent approach can be used as a powerful tool for predicting the behaviour of solids
which can be measured in the laboratory.

1.2 THE PROBLEM TO BE SOLVED:

The problem that we want to solve is a solid. A solid which is a collection of nuclei indicated
in the red colour (Figure1.1) and a collection of electrons indicated in blue colour
 
(Figure1.1). The position vectors of the nuclei are represented by Ri and R j and the position
 
vectors of the electrons are represented by ri and rj . The electrons are coulomb interacting

1
with each other, the nuclei are coulomb interacting with each other and the nuclei are
coulomb interacting with the electrons. This is the solid we want to solve and predict all
properties of the solid that can be measured in the laboratory.

Figure1.1: The coulomb interactions of particles in a solid.

The behaviour of that solid, the solution to that solid can be predicted by solving for the many
– body wave function in the Schrodinger’s equation:

Hˆ   E [ 1.1]

The Hamiltonian such a system has five components which can be written in the following
form:
 
 
2 2
R  2 2 ri Zi e2 e2 Zi Z j e2
H  
ˆ i
             
1 1 [1.2]
i 2M i i 2mi i, j ri  R j 2 i j r  r 2 i j R  R
i j i j

       
Hˆ  (r1..........rN , R1........RN )  E (r1.........rN , R1.......RN ) [1.3]

(The many body Schrodinger’s equation)

Starting from the left hand side of equation [1.2 ], we have the kinetic energy of the nuclei,
the kinetic energy of the electrons, the potential due to electron - nuclei interactions, the
potential due to electron-electron interactions and the potential due to nuclei-nuclei
interactions. Solving the Schrodinger’s equation with this many body Hamiltonian seems a
formidable task. If it were a single particle or two body problem, this equation can be solved
exactly. Three or more body problems have no analytical solution; we have to involve some
approximations and tricks to obtain numerical solutions which agree with what experiments
measure.

It is similar to the many – body problem in classical mechanics – our solar system. The
planets are interacting with each other and their interacting with the sun. What makes this
problem mathematical so much less convenient to solve is the planet – planet interactions.
That is what makes it almost impossible to be solved.

2
Figure1.2: Our solar system has interactions which are similar to the interactions in a real
solid.

If we can develop a techique which allows us to switch off the planet – planet interactions(of
course we cannot physically do this unless we assume that such a solar system exist) then
we can reduce the problem to a two – body problem which can be solve exactly. The biggest
challenge about this idealization is that we must find an external gravitational field
somewhere that can be used to correct for the omitted interactions. If such an external field
exist and we have a universal generator of such a field then we can solve any solar system
that we want and not only our solar system.Density functional theory gurantees that such a
universal generator exist for crystals and so we can confidentlly carry out such
approximations.

1.3 THE EXCHANGE CORRELATION FUNCTIONAL(XC)

If we apply the density functional theory to the complicated solid above, DFT tells us to
switch off the electron-electron interactions - that is what is making this problem so much
difficult to solve. Of course that solid will no longer be a real solid and we want to describe
the real solid and so we need to add an external field to correct for the omitted field due to the
electron-electron interactions. Density functional theory guarantees that such an external field
exists and there is a universal generator called the exchange correlation functional (XCF)
for generating such a field. That functional (the exchange correlation functional) when we
feed it with the information of the solid, we get the field we need to correct for the omitted
interactions. In summary, what the DFT is telling us is the following:

 Replace the electrons with non-interacting ideal electrons - they only interact with the
nuclei but not with each other.
 Apply an exchange correlation field to correct for the omitted interactions
 The exchange correlation field can be derived from a universal exchange correlation
functional. If you know the functional, then you can make any prediction about the
solid without accuracy loss.

3
1.4 BORN-OPPENHEIMER APPROXIMATION
Since exact solutions in quantum mechanics are hard to come by, most of the times
approximations are needed. The basic idea of the Born-Oppenheimer approximation is very
simple; a nucleus is many times heavier than an electron. As a result, electrons easily respond
to changes in their surroundings as compared to nuclei. If the nuclei are too heavy and do not
easily respond to changes easily, then the uncertainty in their positions can be ignored and
that is equivalent to taking their kinetic energies out of the equation. We can then fix the
positions of the nuclei so that they have a constant potential. If they have a constant potential,
then they do not affect our solution; we can calculate it once and add it to our calculated
energies later.

Recall that the complex machine that describes the electrons in the solid is our many – body
Schrodinger’s equation which can be written in the following form:

H m m  Em m [1.4]

If we apply the Born-Oppenheimer approximation to the solid, we fix the positions of the
nuclei and reduce the complicated problem to electronic constrained problem:

H e G  EG G [1.5]

In this form, the ground state electronic Hamiltonian has the following form:
2 2 Z e2 1 e2
H e   i   J   [1.6]
2me i , j ri  R j 2 i j ri  rj
H e  T  Vext  Vc [1.7]

The first term is the total kinetic energy of the electrons, the second term (also called the
external potential) is the potential feel by the electrons due to the nuclei and the last term is
the coulomb potential due to electron – electron interactions.

The ground state wave function  G d , z, r still poses a big challenge because the
electrons have 3N dimensional coordinates accessible to them. Suppose you have 105
electrons in a computer, the wave function will become very difficult to represent in the
computer. The situation is even worse if you have Avogadro’s number of electrons in your
computer – no way can you handle such a wave function no matter how fast your computer
may be. The goal of density functional theory is to solve the ground state problem stated in
equation [1.5] above by skipping the wave function and work with the electron density
instead. The density of the electrons has only three variables; n(nx , ny , nz ) which is much
easier to represent in a computer than wave functions.

From equation [1.7], if we take the derivative of the electron energy with respect to a
parameter (say the coordinate  ), then the kinetic energy term vanishes since it depends on

4
derivative  but not  itself. This results to only the derivative of the potential which is
essentially the force on the electron.
dH e dV 
  Fe [1.8]
d d

We can therefore conclude that if we treat atoms and molecules in the ground state the force
on them is only a function of density, then we can calculate the forces on the molecules or
atoms from their surrounding electron densities accurately and we do not need to worry about
wave functions. This idea was proposed by Hellmann and Feynman and was later developed
by Hohenberg and Walter Kohn as summarized in the following theories.

1.5 HOHENBERG AND KOHN THEORY

The complications associated with the many body wave functions in the Schrodinger’s
equation can be frustrating. Hohenberg and Kohn proved that instead of dealing directly with
the wave functions, we can instead use the ground state electron densities to predict all the
ground state properties of the solid. Hohenberg and Kohn formulated two fundamental
theorems which are the corner stones for the density functional theory:

 Hohenberg and Kohn (First theory)

For non-degenerate ground states, there exist a one-to- one correspondence between the
electron density and the wave functions:

Figure1.3: One –to- one mapping of the electron density onto the wave functions

This mapping enable us to express the ground state energy in the Shrodinger’s equation as a
unique functional of the ground state electron density: E[n(r )]

Thus:

o For a non-degrade ground states, two different Hamiltonians cannot have the same
ground state electron density
o If you know the Hamiltonian of the system you know there is a density corresponding
to it.
o The knowledge of the particle density implies the knowledge of the wave function
and the potential and hence of all variables.

5
It follows from this theory that, the many-body wave functions for all states (both ground and
excited) can be determined from the ground state particle density no (r ) . Therefore, all
properties of the solid are completely determined given only the ground state density no (r ) .

A schematic representation of Theorem I is given in Figure1.4 below:

Figure 1. 4: Schematic representation of Hohenberg-Khan theorem (I).

 Hohenberg and Kohn (Second theory)

The ground state energy functional is minimal when no (r ) is the actual ground state electron
density among all the possible electron densities.

If the true functional E[n(r )] is known, we can vary the electron density n(r ) until the
energy from the functional is minimized.

1.6 THE KOHN-SHAM EQUATIONS


Imagine there is a protocol that can transform a very large machine like the large hadron
collider into a very simple machine like the voltmeter such that the transformed voltmeter
gives the same information as the large hadron collider without accuracy loss. First, such a
transformation is economical since it will cause only some few dollars to make a voltmeter as
compared to constructing the large hadron collider. Secondly, it is much easier to read the
voltmeter than the large hadron collider. The density functional theory operates the same
way. The very complex machine is the many – electron Schrodinger’s equation. The protocol
that converts the complex machine (the many –electron Schrödinger’s equation) is the density
functional theory (DFT). It transforms the complex equation into a set of single particle
equations called the Kohn –Sham equations.

The set of Kohn-Sham equations can be solved by many numerical methods like: LAPW
(Linearized Augmented Plane Wave) method, PAW (Projected Augmented Wave) method
and Grids method.

These numerical methods are implemented in computer codes like; the VASP codes,QE
codes and WIEN2k codes.

6
Figure1.5: DFT transforms the many-electrons problem into a set of Kohn Cham equations.

It follows therefore that, if we take any solid and feed it into the DFT machineary we can
convert the Shrodinger’s equation to Kohn- Sham equations and then solve. We can then
make what ever prediction we want to make about that solid. This sounds very simple but
there is a price to pay. We must find the exchange correlation functional .Without that
functional, we cannot make the transformation.The density functional theory guranttees that
such a magical functional exist to make the exact transformation but it does not tell us what
exactly that functional is.

1.7 DENSITY FUNCTIONAL THEORY CALCULATIONS

The actual calculations with density functional theory depend on three different aspects .
These three aspects are:

 The choice of the exchange correlation functional


 The choice of the numerical method for solving the Kohn-Cham equations
 The choice of the computer code for implementing the numerical method

1.7.1 THE CHOICE OF THE EXCHANGE CORRELATION FUNCTIONAL(XCF)


Density functional theory is an exact transformation provided you know the exchange
correlation functional which is behaving well. Since DFT does not provide a clue on how to
find the exhange correlation functional, the possible way to go is to make a guess to figure
out a well behaved functional( The functional which predicts the experimental values as
close as possible) to excute your calculations. The method disussed in this text is the one
formulated by Kohn and Cham , it is called the local density approximation(LDA).

The concept of the local density approximation is very ideal.You start up with something that
does not exist in the real world and then you make a leap of faith by comparism to a real

7
crystal structure.To do this we imagine a single non-degerate system that can be solved
analytically. A very special system that does not exist in the real world but from which we
can derive an external potential for it. In that special system, the nuclei are evenly smeared
over all space,then we add electrons to neutralize the sytem. Since there is no structure in the
nuclei, there is no structure in the electrons and that reduces the system to only three degrees
of freedom( number of electrons per cubic meter). Such an ideal system is called a jellium or
homogeneous electron gas and figure1.5 shows such a model.

Figure1.6: Homogeneous electron gas system

Eventhough such a system does not exist but we can do calculations with it because we can
determine the contribution of the unit cell of such a sytem to the exchange correlation energy.

The trick here is to make a comparism of the ideal system to a real solid. By scanning the
entire real solid we identify the unit cell of that solid and the value of its electron density.This
point in space would make a contribution to the exchange correlation energy as much as the
unit cell of a homogenous electron gas with the same density would do. With this trick, we
can now construct the total exchange correlation energy for the entire cystal system.

1.7.2 THE CHOICE OF NUMERICAL METHOD


Once you have been able to construct the exchange correlation energy of your crystal
structure, you now make a choice on which numerical method to use for the DFT
calculations.As I have mensioned early on, there are many numerical methods available for
DFT calculations. Some of these methods include: LAPW (Linearized Augmented Plane
Wave) method, PAW (Projected Augmented Wave) method and Grids method.

1.7.3 THE CHOICE OF COMPUTER CODES


The chosen numerical method has to be implemented in computer codes. There are
numerious computer codes which can be used for the same DFT calculation and produce
almost the same results. Some of these codes include: the VASP codes (The Vienna Ab initio
Simulation Package) ,QE (Quantum expresso) codes and WIEN2k codes.

1.8 SOLVING THE KOHN CHAM EQUATIONS


With the approximations discussed earlier on, we now know that the total energy functional
in the Schrodinger’s equation ( see equation1.1) can be written as a functional of the electron

density n(r ) . After applying the Born-Oppenheimer approximation and the Kohn-Cham

8
homogenous electron gas model, we can write the total energy functional in the
Schrodinger’s equation in terms of the kinetic energy of the non-interacting electron gas and
the external potential. These terms however come with additional terms. Thus, the total
energy functional can be written as:

E[n(r )]  K .EKS  K .EC  Vext  Exc [1.9 ]

Where:
The first term is the total kinetic energy of the non-interacting fermions or the Kohn Cham

kinetic energy ( K.EKS ) and can be squeezed out of the Kohn-cham wave functions i (r ) :

2 
K .Ee   1 i (r ) 2i (r )dr [1.10]
i

The corrections to this term due to the many-body interaction of the electrons are included in
the last term, the exchange-correlation term.

The second term corresponds to the electron Coulomb interactions and can be written as:
n( r ) n( r ' )
2  r  r '
K .Ec  1 drdr ' [1.11]

The Coulomb interactions between electrons and nuclei are included in the external potential
which has the following form:
Vext   Vext (r , R)n(r )dr [1.12]

The exchange correlation energy is a functional of the electron density ( Exc[n(r )] ) and so
putting all these terms back into equation [1.4 ], the total energy functional from the many
body Schrodinger’s equation has the following form:
n( r ) n( r ' )
2 i
E[n(r )]   1  * (r ) 2i (r )dr  1 2  drdr '   Vext (r , R)n(r )dr  Exc[n(r )]
i rr '

[1.13]

The electron charge density n(r ) is determined from the single electron Kohn - cham wave

functions i (r ) according to the expression:
N 2

n ( r )   i ( r ) [1.13]
i


With N the number of occupied states for all atoms of the system. The electron density n(r )
is then varied until the energy functional is minimized in accordance with the Kohn-Cham
second theory of DFT. This energy functional relative to the electron density yields an
eigenvalue problem, called the Kohn-Sham Equation, which has the form of a singleparticle
Schrödinger Equation:
 2 2 
    Veff (r ) i (r )   ii (r ) [1.14]
 2m 

9
The effective potential is also called the Kohn Cham potential VKS (r )
 2 2 
    VKS (r ) i (r )   ii (r )
 2m  [1.15]
This equation[1.11] is a one electron Schrodinger’s equation that can be solved for Kohn
Cham molecular orbitals.
The Kohn Cham potential VKS (r ) is given by:
n( r ' ) '
Veff (r )  Vext (r , R)   dr  Vxc (r ) [1.16]
r  r'
The Kohn Cham potential generates the fictitious system of non-interacting electrons with
density equal to the density of the interacting electrons.

With Vxc , the exchange-correlation potential defined by:


Exc
Vxc (r )  [1.17]
n(r )

Various approximations can be used for exchange correlation potentials and they usually
determine the accuracy of the calculations. The simplest one is the local density
approximation (LDA) explained earlier on. Another approximation which is more accurate is
called the generalized gradient approximation (GGA).

ALGORITHM FOR DENSITY FUNCTIONAL THEORY


The general algorithm for executing the density functional theory calculations of the Kohn
Cham equations for any metallic material is presented in the flow chart below; we start by
making an initial guess for the electron density which could be opted by just summing the
densities of the individual atoms in a molecule. The density has spin dependence associated
with it and you can get spin density as well. The next step is to create an effective Kohn-
Cham potential which has three terms; the external potential, the Hartree or the coulomb
potential and the exchange correlation potential. The next step on the flow chart is to solve
the Kohn-Cham equations which are Eigen value problems. Finally, you construct a new
density and check whether it is consistent with the density you have started with.

If there is self-consistency in your first guess, then you have solved the Kohn-Cham
equations. Mostly your first trial would not create that self-consistency and you have to go
back and make a new guess for the density. Once your guess happens to be the right guess,
and then you have the ground state density and consequently get access to the ground state
energy, forces and other properties from your solution.

10
Figure1.7: Self – consistent Kohn – Cham solution

11
REFERENCES
[i] Richard M.Martin, electronic structure: basic theory and practical methods , Cambridge
University press,119-187(2004).

[ii] David S.Sholl, Janice A.Steckel,Density functional theory ,A practical introduction,


John Willey and sons Inc.publications,1-47(2009).

[iii] Jorge Mario Osorio Gullen, Density functional theory in computational material

science,Upsala University, Sweden ,5 - 15 (2004).

[iv] N D Woods , M C Payne and P J Hasnip, Computing the self-consistent field in Kohn–

Sham density functional theory, J.phys: condens.matter ,20-30(2019).

[v] Stefaan Cottenier , Density functional theory,Univesity Gent, centre for molecular

modelling, Lect, Monday April (2015)

[v] Vincent Meunier ,Density functional theory,Rensselaer Polytechnic Institute,Lect (2019)

12

You might also like