PROJECT REPORT

DENSITY FUNCTIONAL THEORY

BY
AFNAN BAKHT
BAHROZ RASHID
WAQAS LIAQAT
NOMAN KHAN
SHEHRISH AKBAR

INSTITUTE OF PHYSICS & ELECTRONICS

UNIVERSITY OF PESHAWAR

MANY BODY PROBLEM

Many body problem is a general name given to the properties of quantum system made of
large number of interacting particle. The wave function of such a system is complicated and
holds a large amount of information. In such a quantum system, the repeated interactions
between particles create quantum co -relation or entanglement. The main problem of such
system is to find the ground state wave function of interacting electrons with an external local
potential. The solution of such system relies on approximations specific to system at hand.
EXAMPLES:

Condensed matter physics (solid-state physics, nanoscience, superconductivity)

Quantum chemistry (computational chemistry, molecular physics)

Atomic physics

Molecular physics

Nuclear physics (Nuclear structure, nuclear reactions, nuclear matter.

APPROACHES:
Out of many approaches to solve the many body problem, we will use Density
Functional Theory

DENSITY FUNCTIONAL THEORY

Density functional theory is an approach for the description of ground state properties of
metals, semiconductors, and insulators. The success of density functional theory (DFT) not only
encompasses standard bulk materials but also complex materials such as proteins and carbon
nanotubes. The main idea of DFT is to describe an interacting system of fermions via its density
and not via its many-body wave function. For N electrons in a solid, which obey the Pauli
principle and repulse each other via the Coulomb potential, this means that the basic variable
of the system depends only on three -- the spatial coordinates x, y, and z rather than 3N
degrees of freedom.
SIGNIFICANCE:
DFT is among popular and versatile methods available in condense matter physics,
computational physics and computational chemistry. The wave function of an N-electron
system includes 3N variables, while the density, no matter how large the system is, has only
three variables x, y, and z. Moving from E*+ to E*+ in computational chemistry significantly
reduces the computational effort needed to understand electronic properties of atoms,
molecules, and solids. The results of DFT calculation for solid state system are according with
the experimental data.

DERIVATION AND FORMALISM

To find ground state of N Interacting electrons in an external potential of nuclei, we solve the
Schrdingers equation

Where is Hamiltonian operator corresponding to the total energy of the system. In this case,
it is equal to the sum of total kinetic energy operator of the electrons, Coulomb interaction
between the electrons

and external potential

which is between the electrons and

nuclei. Now we apply a well know Born Oppenheimer approximation [01] on this Hamiltonian,
in which nuclei are taken at rest as compared to fast moving electrons, so the external potential
become static. The Hamiltonian becomes:

( )

Now from first equation

( (

( )

This is a second order differential equation with 3N independent co-ordinates and is extremely
difficult to handle. We use alternative methods to solve this problem.

RITZ VARIATIONAL PRINCIPAL

For solving the second order differential equation with 3N independent co-ordinates we use
Ritz variational principal [02] to find the ground state of the system. In this approach, we
minimize the ground state energy of N particles.
| |
This is done by first setting up the Hamiltonian operator of target system. From equation (1.2) it
shows that the only information that depends on the actual molecule is the number of
electrons in the system N and the external potential

determine completely the

position and charges of all nuclei in the molecule, the operator representing kinetic energy or
the electron -electron repulsion are independent of the particular molecule.
In the second step we have to find the eigen functions
eigenvalues
ground state

and the corresponding

of .There is a recipe for systematically approaching the wave function of the

, i-e, the state which delivers the lowest energy

. This is the variational

principle, which holds a very prominent place in all quantum-chemical applications. We recall
from standard quantum mechanics that the expectation value of a particular observable
represented by the appropriate operator using any, possibly complex, wave function

| |

The variational principle now states that the energy computed via equation (1-11) as the
expectation value of the Hamilton operator from any guessed

will be an upper bound

to the true energy of the ground state i-e

| |

Where the equality holds if and only if

is identical to

| |

. In this way we minimize the

ground state. This sure solves the problem but it doesnt simplify our work.

DENSITY AS A BASIC VARIABLE

Until this far, we use wave function of N particle system which is dependent on 3N spatial coordinate and N spin co-ordinates, which is very difficult to solve. To simplify this problem we
use density as a basic variable which is defined by:

The operator corresponding to density is

we arrived at the conclusion that the Hamilton operator of any atomic or molecular system is
uniquely defined by N, the number of electrons,

, the position of the nuclei in space, and

the charges of the nuclei. We went on by saying that once is known, we can of course only
in principle solve the Schrdinger equation.
From the properties of electron density
I.

The density integrates to the number of electrons i-e

II.
III.

has maxima, that are actually even cusps, only at the positions

of the nuclei

the density at the position of the nucleus contains information about the nuclear charge
Z

Thus, the electron density already provides all the ingredients that we identified as being
necessary for setting up the system specific Hamiltonian and it seems at least very plausible
that in fact

suffices for a complete determination of all molecular properties.

THOMAS FERMI MODEL

Thomas and Fermi [03, 04] used the density as a basic variable and developed a theory for a
system of many particles which is known as Thomas-Fermi (TF) theory. It is the first densitybased theory for a many-electron system. This approach is only valid for infinite nuclear charge.
It only treat the kinetic energy quantum mechanically and treat and

classically.

In their model Thomas and Fermi arrive at the following, very simple expression for the kinetic
energy based on the uniform electron, a fictitious model system of constant electron density.

If this is combined with classical expression of nuclear-electron attractive potential and

electron-electron repulsive potential we will get the famous Thomas Fermi expression for
energy of an atom.

The importance of this equation is not so much how well it is able to really describe the energy
of an atom, but that the energy is given completely in terms of the electron density

. TF

model is a perception for how to map a density on to energy without any additional
information required. In particular, no recourse to the wave function is taken. After getting an
expression in which energy is expressed as a functional of density. To find the ground state, we
again employ variation principle. It is assumed that ground state is connected to the electron
density for which the energy according to which the Thomas Fermi equation is minimized under
the constraint.

For realistic systems, it yields poor quantitative predictions, even failing to reproduce some
general features of the density such as shell structure in atoms and Friedel oscillations in
solids.[05]

HOHNBERG AND KOHN THEOREMS:

TF model was not formulated on an exact physical basis and at that time expressing the energy
as a functional of electron density was not justify or whether a procedure of employing the
variational principle on density is allowed. It was replaced by density functional theory (DFT)
when a landmark paper by Hohnberg and Kohn [06] appeared in Physical Review. It is based on
2 fundamental theorems of Hohnberg and Kohn. These two theorems represent the foundation
of density functional theory (DFT) as an exact theory. The vital insight of HK-theory was the fact
that all the properties of a physical system can be determined from density of electron only.
The theorems are:
Theorem 1:
The electronic ground state density

determines the external potential

. The

density thus determines the uniquely corresponding Hamiltonian, the ground state stationary
wave function

and all the electronic properties of the system

The theorem provides the proof that a physically meaning full wave functional can be
uniquely associated with a certain density is physically justified.
Kinetic energy and energy of electron-electron interaction can be expressed by use of
density. The total energy functional for a system of electrons in external potential
expression
[ ]

[ ]

[ ]

[ ]

Where the external potential is

[ ]

And the potential due to electron-electron interaction is

[ ]

[ ]

[ ]
[ ]

We can extract the classical coulomb part from

has the

[ ]
Where

[ ] is non-classical contribution of electron-electron interaction containing all the

effects of self-interaction correction exchange and coulomb co-relation.

Where the terms [ ]

[ ] is universally valid in the sense that they are independent of

the system information i-e

Now the term [ ]

, while the term

[ ] is called HK-functional
[ ]

[ ]

[ ]is dependent of the actual system.

[ ].

[ ] we would have solved Schrdinger equation exactly not approximately. All

If we know

the intrinsic properties of electronics system are completely absorbed in HK-functional. It is

universal, independent of system, applied well to H-atom to gigantic molecule DNA.
Only ground state electronic density that contains the information about position and charges
of nuclei allow the mapping from density to external potential. The density of excited state
cannot be used.

Theorem 2:
The functional E[n] of total energy satisfies a variational with respect to density
The total energy E[n] reaches its minimum value E0 for correct ground state density.
[ ]
Where

indicates a variation over all ground state densities of arbitrary N electrons

system. Ground state density is sufficient to obtain all properties of the system but how to find
that a certain density is really a ground state density that we are looking for.
In plane words this theorem states that

[ ] , the functional that delivers the ground state

density of the system, delivers the lowest energy if and only if the input is true ground state
energy of the system. This is nothing but variational principle.

[ ]

[ ]

[ ]

[ ]

Now for any trial density n(r) which satisfy the necessary boundary condition such as
and

, and which is subjected to some external potential

. The energy

obtained from above equation represent an upper bound to true ground state energy

result if and only if the exact ground state density is inserted in the above equation. Now the
trial density defines its own Hamiltonian

and hence its own wave function

. This wave

function can now be taken as trial wave function for the Hamiltonian generated from true
external potential

| |

.Thus, we arrive at
[ ]

[ ]

[ ]

| |

Constraints-Search Formulation
HK-theorems imply that for any arbitrary density

there exists an external potential

which is called V-representable. Due to this fact it is possible to construct such densities which
are not V-representable. This implication obviously leads to formal problems which can
circumvent by more general formulation of HK-theorem called constrained-search formulation
[07].
The densities which are V-representable are called N-representable if it can be constructed
from anti-symmetric N-particles wave function. Based on so called N-representable densities, a
more general and modified scheme for
to

[ ] minimization shows no explicit connection

. Then the total energy functional is

[ ]|

[ ]
[ ]
The notation
to density

[ ]

| [ ]

indicates now a minimization over all the wave functions

leading

. Now [ ] is defined via the constraint search

[ ]|

[ ]

| [ ]

Now as Ritz Variational Principle comprises over all allowed wave function within its search
while constrained-search is limited to wave functions that generates

. Now the

minimization runs without requiring V-representability over all wave functions. A relation of the
Ritz variational principle in with variational principle of the 2nd HK-Theorem can be established
by simply separating the search over all wave functions

as it is performed in the Ritz

variation, into two separate searches that lead to a variation performed using the second HKTheorems.
|

10

|
|

[ ]
The 1st minimum search runs over all wave functions resulting the density

. The second

search after that lifts the constraint to a particular density and extends the new search

over all the densities. Though the HK-theorems and the constrained-search

formulation create a strict mathematical framework and now the existence of the total energy
functional is ensured, but one still has no practically feasible scheme to treat the N interacting
electrons at hand at this point which is solved by Kohn and Sham [08].

11

Kohn-Sham Approach
The main idea of the Kohn-sham (KS) method is to map the interacting electrons system to a
fictitious system of non-interacting electrons that produce exactly the same ground state
density and ground state energy as the interacting system. In this system, they behave as a
charge fewer particles which do not interact with each other but each of the electrons felt the
field produced by all other electrons of the system. In other words, the KS scheme represents a
mean-field theory.
The wave function

[n] of the KS system is defined as the wave function that minimizes the

constrained search minimization

[ ]| | [ ]

[ ]
Where

| |

[ ] is the kinetic energy functional of the KS system. The exchange energy by

definition is
[ ]| | [ ]

[ ]

The correlation energy is

[ ]

[ ]|

| [

[ ]|

| [ ]

Then total energy can be written as

[ ]
[ ]

[ ]
[ ]

[ ]
[ ]

[ ]

Where, exchange-correlation functional is

[ ]

[ ]

[ ]

which contains difference of the kinetic energies plus difference of the expectation values of
the electron-electron interaction of the systems.
[ ]

[ ]

[ ]

12

[ ]

[ ]

Here the problem is in reformulating in a way that main part of the energy is treated in an exact
manner and all other unknown contributions are absorbed in the exchange-correlation
functional

.This feature can be seen one of major merits of the KS-DFT.

Up till now, the original problem of the solution of Schrodingers Equation

only transferred onto the problem of finding the exact expression for the function

has
[ ] of

exchange-correlation, which is unknown yet.

The variational problem presented in the second HK-theorem, the ground-state energy of a
many-electron system can be obtained by minimizing the energy functional, subject to the
constraint that the number of electrons is conserved, which leads to
[ [ ]

By varying the expression (3.20) with respect to density yields the Euler-Lagrange equation of
the KS theory
[ ]

[ ]

[ ]

Now, defining the potential of KS in this Equation as

[

[ ]

And the Lagrange multiplier in above Equation becomes

[ ]

[ ]

With the Hartree potential

And the exchange-correlation potential

[ ]

13

To construct such energy functional there are no simple methods but it is certainly possible to
separate the exchange part from the correlation part. The same holds for the exchange and
correlation potentials as the functional derivative of the energy functional with respect to the
density
[ ]
[

[ ]

In DFT the exchange energy functional is defined by the relation [09]

[ ]

[ ]| | [ ]

[ ]

[ ]

The term ( [ ]| | [ ] is expectation value of electron-electron interaction of KS State not

the actual electron-electron interaction energy. the difference between true electron-electron
interaction energy and expectation value ( [ ]| | [ ]) is absorbed in correlation energy,
also it contains kinetic energy differences of real and KS non-interacting systems of electrons
The correlation energy then can be
As
But

[ ]

[ ]

[ ]

[ ]

[ ]

[ ]

[ ]

[ ]

So
[ ]
Now put the value of

[ ]

[ ]

[ ]

[ ]

[ ]

[ ]

[ ]| | [ ]

[ ]

[ ]

[ ]

[ ]

[ ]

[ ]

Cancel U[n] both sides and Rearranging we get,

[ ]

[ ]

[ ]

[ ]

[ ]| | [ ]

So by knowing the values of the four terms on the right side of this equation we can easily find
the correlation functional.

14

Approximate Exchange Co-relation Functionals

Local Density Approximation
Kohn-Sham formalism which allows an exact treatment of most of the contributions to the
electronic energy of an atomic or molecular system, including the major fraction of the kinetic
energy. All remaining unknown parts are collectively folded into the exchange-correlation
functional

[ ]. These include the non-classical portion of the electron-electron interaction

along with the correction for the self-interaction and the component of the kinetic energy not
covered by the non-interacting reference system. One has to approximate the exchange corelation functional. One of the oldest and most known approximation (LDA), which is proposed
by Kohn and Sham themselves and it lies at the roots of DFT and KS-formalism.
The original physical system for LDA is homogeneous electron gas, means a system of electrons
which moves on a uniform background of positive charge in such a way that the entire system
becomes electrically neutral. The electrons and the volume of gas V approach to infinity, while
the electron density

reaches a constant value, which remains same for each

spatial point . Inhomogeneous system of electrons such as atom, molecule or solid are treated
as homogenous system with the corresponding exchange and correlation energy
per electron at point by the associated homogenous electron gas with density at each point .
Hence, LDA approximation is most suitable for slowly varying density systems. However in solid
state physics LDA is the most employed scheme while in chemistry it has no comparable impact
in which the atoms and molecules are parameterized by a rapidly varying density. As the
homogenous electron gas is determined by its density, energy due to the exchange-correlation
within LDA approximation is calculated by integrating the parameter
the local density

at each point in space

[ ]

This energy is again separated between its exchange and correlation term
[ ]

15

weighted with

And LDA exchangecorrelation potential is determined by the functional derivative of above

equation
[ ]

The exchange part of the energy per particle

)
was originally derived by Dirac [10] and given

as

Where

is the radius of a sphere with effective volume of an electron. However, there is no

explicit expression for the co-relation energy term. Ceperley and Alder used the quantum based
Monte Carlo scheme to calculate the total energy for a homogenous electron gas and attained
the correlation energy by subtracting the corresponding exchange and kinetic energies [11].
Various authors have presented analytical expressions of
interpolation schemes. The most widely used representations of

based on sophisticated
are the ones developed

by Vosko, Wilk, and Nusair (VWN) [12], 1980, while the most recent and probably also most
accurate one has been given by Perdew and Wang [13], 1992. VWN interpolated this data and
formed the LDA correlation

Where

,
,

and for zero spin polarization

)]]

Generalized Gradient Approximation

In the early eighties, first successful extension to the local approximation was made. The logical first
step in that direction was the suggestion of using not only the information about density

at a

particular point , but to supplement the density with information about the gradient of the charged
density

in order to account for the non-homogeneity of the true electron density which is varying

fastly. So the exchange co-relation term become

16

[ ]

This approach is known as generalized gradient approximations (GGA) and generally is given by a
functional which explicitly contains density gradient

. In this case the exchange and correlation

terms can also be treated separately

[ ]

[ ]

[ ]

In most of the cases GGA functionals for the exchange part can be written as
[ ]
Where the argument of

is a dimensionless parameter

Where F(s) is a scaling function that can have in practice a pretty complicated form. Here,

for the Fermi wave vector given by

stands

GGA functional yields often better results than the LDA. However, some of these functionals may not
necessarily based on the new physical ideas, while some are of the semi empirical character; contains
fitting parameters and sometimes are solely constructed in order to satisfy required boundary
conditions and gain fine results in an adequate computing time. Oftenly, these are optimized in order to
gain good energy value results and do not give bearable results for the related potentials. Some of the
most important and most widespread GGAs are

Beckes exchange functional (B88) *14].

Perdew, Burke and Ernzerhofs exchange-correlation functional (PBE) [15].

Lee, Yang and Parrs correlation functional (LYP) [16].

Perdews 1986 correlation functional (P86) *17, 18].

Perdew and Wangs correlation functional (PW91) *13].

Engel and Voskos (EV) correlation functional *19].

Wu and Cohen density gradient functional (WC) [20].

17

The functional derivative of the general GGA exchange correlation functional is given by [21]
[

)]

)]

With two dimensionless parameters

From the above expressions it can be seen that GGA potentials contain even higher density
derivatives than the related energy functionals and together with the already some complicated
functionals of energy, the analytical equations for the potentials can become highly complicated

18

Full Potential LAPW Method

The LAPW method is a very accurate scheme for the calculations of electronic structure. The
linearized augmented plane wave (LAPW) method can be implemented through WIEN2k [22]. In this
approach a special basis set is adopted to solve the problem. Basically this method is derived from the
approach of augmented-plane-wave (APW). Slater constructed APW and was motivated towards it by
the fact that close to an atomic nucleus potential and wave functions were alike to those in an atom;
nearly spherical and strongly varying. While, between the atoms both potential and wave functions
were smoother. In view of that, unit cell space is divided into two regions; non overlapping atomic
spheres and interstitial regions. The different basis expansions are applied in these regions: radial
solutions of Schrdingers equation for the atomic spheres region and plane-waves for the remaining
interstitial regions.

Where
Here

is the wave function for the system,

and

energy and

is the cell volume and

are the expansion co-efficients. If

is the regular solution.

is a parameter which is equal to the band

is a spherical component of the potential inside the sphere, then

[

The particular selection of the functions was motivated by Slater who noted that plane waves are
solutions of the Schrdinger equation for constant potential, while radial functions are solutions for a
spherical potential, for

equal to the eigenvalues. This approximation is known as muffin-tin

approximation and is very good for close-packed (e.g. fcc and hcp) materials. The double
representation defined by the above equation is not certain to be continuous on sphere boundaries,
as it is supposed for the kinetic energy to be well defined. Thus a constraint on the APW method is
imposed by defining

in terms of the

by the spherical harmonic expansion of plane waves. While

the coefficient of each lm component is matched at the sphere boundary

19

Here the origin is taken at the center of the sphere and R is sphere radius. Thus
determined by the plane-wave coefficients

and the energy parameters

20

are completely

The LAPW Basis and Its Properties

In LAPW method, the basis functions are the linear combinations of a radial function
well as their energy derivatives inside the spheres,

are fixed for the

as

. The energy derivative

satisfies
[

In the non-relativistic case, on sphere boundaries these functions are matched to values and
derivatives of plane waves. Plane waves augmented in this way are called as the LAPW basis
functions. In the form of this basis, the wave functions
[

Where

are the coefficients for energy derivatives analogous to

. In this scheme shape

approximations are not made and consequently such a method is often called as Full-Potential
LAPW (FP-LAPW). LAPW method is better than the APW. In the former, at a given k-point accurate
energy bands are obtained with single diagonalization, while for APW method the diagonalization is
required for each band. Moreover, LAPW basis is more flexible inside the spheres because it has two
radial functions instead of one. However, there is a price to be paid for the flexibility. This arises due
to the requirement that the basis functions must have continuous derivatives. Higher value of plane
wave cut-off is needed to achieve a given level of convergence. In the WIEN2K code, the total energy
is calculated according to the Weinert scheme [23]. The cutoff parameter
convergence of the basis set and it is usually takes between 6 to 9. The

controls the

is known as the muffin-tin

radii and is the smallest of all of the atomic sphere radii in the unit cell and

is magnitude of the

largest K vector. Much higher angular momenta in the wave functions are required (typically
to 12 for accurate calculations) to satisfy the continuity conditions accurately.

21

References
[01] M. Born and R. Oppenheimer, Zur Quantentheorie der Molekeln, Ann.Phys. (Leipzig) 84, 457484 (1927).
[02+ W. Ritz, Ueber eine neue Methode zur Lsung gewisser Variations probleme der mathematischen
Physik, J. Reine Angew. Math. 135, 161 (1908).
[03] L. Thomas, The calculation of atomic fields, Proc. Cambridge Phil.Soc. 23, 542 (1927)
[04+ E. Fermi, Un metodo statistice per la determinazione di alcune proprieta dellatomo, Rend.
Accad. Lincei 6, 602 (1927).
[05+ H. Hohenberg and W. Kohn, Inhomogeneous electron gas, Phys. Rev.136, 864 (1964).
[06+ P. Hohenberg and W. Kohn, Inhomogeneous electron gas,. Phys. Rev.136, 864 (1964).
[07+ M. Levy, Electron densities in search of Hamiltonians, Phys. Rev. A26, 1200 (1982).
[08+ W. Kohn and L. Sham, Self-consistent equations including exchange and correlation effects,
Phys. Rev. 140, 1133 (1965).
[09] R. M. Dreizler and E. K. U. Gross, Density-Functional Theory,Springer, Berlin (1990)
[10+ P. A. M. Dirac, Note on exchange phenomena in the Thomas atom,Proc. Camb. Phil. Soc. 26,
376 (1930).
[11+ D. M. Ceperly and B. J. Alder, Ground state of the electron gas by a stochastic method, Phys.
Rev. Lett. 45, 566 (1980).
[12+ S. J. Vosko, L. Wilk, and M. Nusair, Accurate spin-dependent electron liquid correlation energies
for local spin densitycalculations: A critical analysis, Can. J. Phys. 58, 1200 (1980).
[13+ J. P. Perdew, Electronic Structure of Solids, Akademie Verlag, Berlin(1991).
[14+ A. D. Becke, Density-functional exchange-energy approximation with correct asymptotic
behavior, Phys. Rev. A 38, 3098 (1988).
[15] J. P. Perdew, K. Burke, and M. Ernzerhof, Generalized gradient approximation made simple,
Phys. Rev. Lett. 77, 3865 (1996).
[16+ C. Lee, W. Yang, and R. G. Parr, Development of the Colle-Salvetti correlation-energy formula
into a functional of the electron density, Phys.Rev. B 37, 785 (1988).

22

[17+ J. P. Perdew, Density-functional approximation for the correlation energy of the inhomogeneous
electron gas, Phys. Rev. B 33, 8822 (1986).
[18+ J. P. Perdew, Erratum: Density-functional approximation for the correlation energy of the
inhomogeneous electron gas, Phys. Rev. B 34, 7406(1986).
[19+ E. Engle and S. H. Vosko, Exact exchange-only potentials and the virial relation as microscopic
criteria for generalized gradient approximations,Phys. Rev. B 47, 13164 (1993).
[20+ Z. Wu and R. E. Cohen, More accurate generalized gradient approximation for solids, Phy. Rev.
B 73, 235116 (2006).
[21+ J. P. Perdew and W. Yue, Accurate and simple density functional for the electronic exchange
energy: Generalized gradient approximation,Phys. Rev. B 33, 8800 (1986).
[22+ P. Blaha, K. Schwarz, and J. Luitz, A Full Potential Linearized Augmented Planewave Package for
Calculating Crystal Properties, (TechnicalUniversity Wien, Vienna, ISBN 3-9501031-0-4, (2001).
[23] M. Weinert, E. Wimmer, and A. J. Freeman, Total-energy all-electron density functional method
for bulk solids and surfaces, Phys. Rev. B 26,4571 (1982).

23