VIII.

Approximate correlation energy

functionals
While the correlation energy in atoms and molecules is only a small fraction

of the total electronic energy, it is found that it is in fact a very large

percentage when one computes energy differences, such as energy of

atomization (i.e. the difference between the energy of the atomic constituents

to the energy of the molecule) or relative energies of different conformations,

as those determining the shape of the Born-Oppenheimer potential surface. In

essence, the exchange correlation energy is the chemical bonding energy. It is

therefore crucial to model this energy accurately. We describe below some of

the basic approximations for density functional theory.

A. The local density approximation (LDA)

The mapping of the interacting electron system onto the non-interacting

system, encapsulated in Eq. XX, is of formal interest only, unless we devise a

way to approximate the correlation potential. One way is to consider the

correlation energy per electron in the homogeneous electron gas of

density . This energy can be computed with relatively high precision using

Monte Carlo methods. Under this approximation we can write the correlation

energy as . However, this does not yield in practice good

enough results and is thus seldom used. A more successful ways was devised

by Kohn and Sham. They considered both the exchange and correlation per

an electron in the homogeneous gas, . In this case the correlation energy

comes out:

(8.1.1)

Electron Density Functional Theory Page 1

© Roi Baer
This approximation is called the local density approximation (LDA)2. The

functional . It leads to the following LDA approximation for the

energy functional:

(8.1.2)

The minimization of this functional, by the Kohn-Sham approach leads to the

LDA approximation of DFT. This approach is highly successful and is

considered in DFT as the basis for most of the developments of other

functionals.

Note however that in LDA the correlation energy is extremely awkward

looking because of the term – . The presence of this term has detrimental

effects which harm some of the predictions of DFT.

i. The exchange energy per electron in the HEG

In section XXX we discussed in some detail the Hartree-Fock theory of the

homogeneous electron gas. We defined a Jellium as a smeared positive

background of Volume at density together with electrons. We

showed that the Jellium self energy, the Jellium-electron attraction energy and

the electron Hartree energy all cancel exactly in the HEG. Thus, the energy

per particle is given by:

(8.1.3)

We already calculate exchange energy, using a Hartree-Fock treatment of the

HEG and we saw that (see Eq.

Error! Reference source not found.) with . In terms of the

Wigner-Sietz radius, which is a dimensionless quantity given by:

Electron Density Functional Theory Page 2

© Roi Baer
 (8.1.4)

is the radius of a Jellium sphere containing the charge of an electron. Thus:

(8.1.5)

If we have a way of computing and , we can then find from

Eq. (8.1.3). We can then also compute from Eqs. (8.1.5).

ii. Correlation energy of the HEG: the high density limit

The calculation of can presently be done analytically in two limits. One is

the high density limit where the kinetic energy dominates and the

Coulomb interaction can be treated as a perturbation. In this limit the kinetic

energy is that of non-interacting particles. Thus in the perturbative approach

one can take . The unperturbed wave function is the Slater

wave functional wave function composed of the plane wave orbitals

with taking the lowest momentum vectors . The energy of the

unperturbed state is . The first order contribution of the the e-e

Coulomb repulsions is . This quantity is equal to the direct and

exchange contribution (see Eq. Error! Reference source not found.). The

direct part is nullified by the other electrostatic interactions, so the 1st order

contribution is essentially the exchange energy of the HEG which we already

included (see Eq. (8.1.3). Thus to continue and determine the interaction

energy beyond exchange, i.e. the correlation energy, we must move to at least

second order perturbation theory. When one does this, one finds that the

usual second order perturbation theory yields infinite terms. These are

associated with low wave length excitations where a pair of electrons having

Electron Density Functional Theory Page 3

© Roi Baer
the momentum statse| and | , are excited by the Coulomb interaction

to states and (such that and

). One can show that for small this process gives a term

proportional to in the expression for the 2nd order perturbation energy.

This term is singular at low . A method of performing perturbation theory

which is non-singular and goes beyond second order, was devised in 19575.

This theory is essentially exact at the high density limit and leads to the

following relation:

(8.1.6)

Where is the Wigner-Seitz radius, namely is the radius of a

sphere in the Jelium which scoops an amount of charge equal to , where is

the elecmentary quantum of charge (the electron charge). and

. Later work refined these constants:

and .

Exercise ‎VIII-1

Using Eq. Error! Reference source not found. and the Hellman-Feynman

theorem prove that

(8.1.7)

Hints:

(a) Show that where and are, respectively, the total

electronic repulsion energy per electron and the Hartree energy per electron.

(b) From the the fact that show that the last equality is correct.

Electron Density Functional Theory Page 4

© Roi Baer
iii. Correlation energy of the HEG: the low density limit and the Wigner
crystal

The second limit is that of low density, where the electrons form a crystal. We

give the development of this limit, originally proposed by Wigner6 when he

devised a theory for the electron density in metallic sodium.

Wigner assumed that at low energy the homogeneous electron gas forms a

crystal. Now that may sound strange: how can the density be uniform and at

the same time the electrons form a crystal? Thanks to Quantum Mechanics

this is actually not a contradiction, as the following example shows.

Exercise: Show that for 2 particle in a 3D box of volume with periodic

boundary conditions, if the Hamiltonian is:

(8.1.8)

Then

1) The eigenstates have a homogeneous 1-particle density

2) The pair correlation function has structure. (
In the low density regime the electron kinetic energy (per electron) can be

neglected since, as seen in XXX it is proportional to while the repulsion

energy between the electrons per electron is proportional to . At lkow

density the Pauli exclusion priniciple is non-operative, since electrons do not

overlap. Thus the quantum nature of the electron is gone at this limit and we

can think of the electron as a classical particle that localizes. This is because

non-localization of particles in quantum mechanics arises only from their

need to reduce kinetic energy. The electrons will arrange themselves in the

lowest energy state by forming a close packed crystal. Each electron is then as

far as possible from each other electron, while still filling 3D space with

Electron Density Functional Theory Page 5

© Roi Baer
average density n . Let us calculate the energy of such a crystal. Consider one

of the electrons in the crystal. We imagine it together with a cell containing 1

unit of positive charge. This cell shape depends on the crystal symmetry.

Following Wigner, we neglect the crystal structure and assume each electron

is surrounded by a sphere of positive charge completely neutralizing it. Our

approximation then neglects the volume of the space between the spheres.

The radius of the sphere is and it is filled with smeared positive charge and

with one negative charged point-electron at its center. The spheres do not

interact since they are neutral and have no electric moments.

The total energy per electron is the energy to assemble the Jellium

sphere and the energy needed to bring the electron from infinity into

the center of the sphere.

Let be the energy to assemble a sphere of charge density and radius

. Suppose we now enlarge it by adding a shell of radius . The electric

potential at distance outside the sphere is where is the

charge in the sphere. The charge in the shell is and bringing it

from infinity, where the potential is zero to its place on top of the existing

sphere involves the energy . Thus, by integration

from to , we find: . And so at :

(8.1.9)

Next, we want to calculate the energy to bring an electron from infinity to the

center of the sphere. This will be done in two stages, first bringing the electron

from infinity to the rim of the sphere, a distance from its center and then

from the rim to the center. Accordingly write The first part

Electron Density Functional Theory Page 6

© Roi Baer
is easy since we already know the potential, and it is negative since energy is

released by this process, so:

(8.1.10)

Inside the sphere, at a distance from the center there exists an electric field

due to the Jellium, which according to Gauss’ is . This force

is a Harmonic force, with force constant . The work to move

an electron in this field to the center is: The

energy for the second stage is therefore and the total energy per

electron in the crystal is:

(8.1.11)

This then is the exchange-correlation energy for low density. We neglected the

volume between the spheres. The exchange energy we already know from

(8.1.5), is . Thus, Wigner's approximation for the

correlation energy in the low density limit is:

(8.1.12)

Wigner also considered the correction due to the finite kinetic energy when

is finite. Since we saw that the electron inside the spherical Jellium drop is a

Harmonic potential, one can reduce the correlation energy by the 3D

Harmonic zero-point potential, . The correlation energy is then:

(8.1.13)

Electron Density Functional Theory Page 7

© Roi Baer
iv. Monte-Carlo determination of the correlation energy for the HEG

Between these high and low limits there is no analytical theory, in general and

a numerical computation can be made based on quantum Monte Carlo

methods. The results of the calculation are then fitted to an analytical form

which respects the limits

v. The polarized HEG; local spin-density approximation (LSDA)

Up to now we have assumed that the electron gas is unpolarized, i.e. the total

z component of spin Sz per electron is zero. However, Sz is a good quantum

number and it can vary continuously from to . The extreme case is the

fully polarized case. In general one may define the density of spin-up electron

and that of spin down . Then:

(8.1.14)

For a fully polarized gas and the difference is first of all in the Fermi

energy. For the HEG, since every momentum state can populate only one

electron, we find by a similar analysis as in the unpolarized case:

(8.1.15)

The total kinetic energy us the sum of up and down contributions:

, which is evaluated to be using the expressions for

we obtain

Electron Density Functional Theory Page 8

© Roi Baer
 (8.1.16)

As for exchange energy, since exchange interaction occurs only between like

spins, we XXXXX

vi. Successes and failures of LSDA

vii. Plausible reasons for the success of LSDA

One of the uses of Eq. Error! Reference source not found. expression is to

explain the success of a simple theory such as LDA8. TO see this, let us

expand the XC hole in terms of moments around

(8.1.17)

where: and

(8.1.18)

Then consider the XC energy, it can be written as:

(8.1.19)

Were:

(8.1.20)

And:

Electron Density Functional Theory Page 9

© Roi Baer
 (8.1.21)

Thus, only the 00 moment of the dependent XC holes enters the expression.

Therefore, in a sense the angular shape of the average XC hole gets averaged

over and only the radial dependence affects the XC energy. This is used to

explain some of the success of LDA.

We see that the XC anisotropy of the XC hole around r is averaged over. Only

this average enters the XC energy formula. In LDA we use the homogeneous

electron gas to compute the HEG. Of course this leads to an isotropic XC hole.

Yet, since only the spherical average of the hole enters into the XC energy, this

drastic approximation gives a reasonably good XC energy.

Exercise: Calculate the spherically averaged X-hole for a 1-electron system (H

atom for example)

Solution: The orbital is , the density is and the DM is

thus:

(8.1.22)

The hole is independent of the reference point .

One of the important results shown below is that only the spherically

averaged hole enters the XC energy. Thus we only need the spherically

averaged hole:

(8.1.23)

Which becomes:

(8.1.24)

Electron Density Functional Theory Page 10

© Roi Baer
For the H-atom , defining therefore

and :

(8.1.25)

The form of this spherically averaged hole function as function of s and r is

shown here:

The hole has a cusp at the r-s origin.

B. Semilocal functionals and the generalized

gradient approximation

Electron Density Functional Theory Page 11

© Roi Baer