Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.

com
by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

Systems
Chapter 8

253
^ ( r i , r 2 , . ■;rN)^r^iVeff(ri)
for Many—Electron
Approximation Methods
 254 Lecture notes on atomic and molecular physics; Erkoc & Uzer

8.1 Many—electron system calculations

Schrodinger's equation can not be solved exactly for two-electron atoms,
so that certain approximation methods must be used. Accurate results
can be obtained for the energy levels and wave functions of two-electron
atom by performing variational calculations. But this method becomes
increasingly tedious when the number of electrons increases in the atomic
Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com

system. Therefore some other general methods should be developed to

study many-electron atomic and molecular systems.
by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

8.2 The Thomas-Fermi model of the atoms

The Thomas-Fermi (TF) model is based on statistical and semiclassi-
cal considerations for the ground state of many-electron atoms. In this
model the N electrons of the system are treated as a Fermi electron gas
in the ground state, confined to a region of space by a central potential
V(r) which vanishes at infinity.

The aim of the TF model is to provide a method of calculating the

potential V(r) and the electron density p(r).

The total energy of the system is written as

E = EF + V(r) (8.1)

In the Fermi electron gas model Ep is given as

h2 / 9 \2/3
EF = - (*,*?) (8.2)

Combining Eqs.(8.1) and (8.2), we have

p(r) = 3/2(E y(r))3/2

i(f) - ^
p = 0 for V > E. Taking the relation between the electrostatic potential
<f>(r) and the potential energy V(r)

<Kr) = -~V(r) (8.4)

 Ch.8 Approximation methods for many-electron systems 255

and
■i

cf>o — — E (8.5)
e
one may define
$ ( r ) = (f)(r) - <f>o (8.6)
Thus,
Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com

1 / Om \ *V2
(8.7)
by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

p(r) = 0 / o r $ <0
The Poisson's equation for a given charge density is

V 2 $(r) = ep(r) (8.8)

Substituting Eq.(8.7) in Eq.(8.8), we have

V 2 $(r) = ^ ( | f ) 3 / 2 ( e $ ( r ) ) 3 / 2 for * > 0 (8.9)

V 2 $(r) = 0 for $ <0

Defining
1

X= ^ r $ ( r ) (8.10)

and re = r/& where

charge density may be expressed as

z (x\312 (8.11)
for x
p=»\S) -°
p = 0 /or x <0
Knowing that V $ —> (l/r)d [r$(r)]/dr 2 ; then one can write
2 2

rf2
X = J_x3/2 (8.12)
da;2 •v/x
This equation is known as the Thomas-Fermi equation (it is usually
solved numerically)
 256 Lecture notes on atomic and molecular physics; Erkoc & Uzei

Sx/dx2 = 0 for x < 0 and x(°) = 1- x(°°) = ° f° r neutral atom

solution.

Knowing the function x(x)i w e c a n obtain the function $(r), and

hence the electrostatic potential <j>(r), the potential energy V(r) and the
Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com

density p(r).
X(r)
by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

x
x0
Fig. 8.1. Solutions of the Thomas-Fermi equation.

(1): Neutral atom solution; a solution which is asymptotic to the x-axis.

(2): Solution for positive ion (N < Z); solutions which vanish for a finite
values x = XQ-
(3): Solution for a neutral atom under pressure; solutions which never
vanish and diverge for large x.

In the TF model the central potential V(r) is given for neutral atoms
by
Ze2 Ze2
V(r) = X in SI units V(r) = (8.13)
T (47T£ 0 )r X

where
X(x) S 1 - 1.588a: 4-... (8.14)
 Ch.8 Approximation methods for many-electron systems 257

or
V(r)^e2(-- + 1.794—) (8.15)
\ r ao )
The first term on the rihgt-hand-side of this equation represents the
nuclear attraction, and the second term represents the electron repulsion.
In SI units:
Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com

V(r) *-^-(-- + 1.794—) (8.16)

47T£0 \ r ao J
by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

In a.u. :
V(r) S - - + 1.794Z4/3 (8.17)
r

8.3 Hartree-SCF method

In this method the energy and wave function of a many-electron atom
is calculated numerically. Consider an iV-electron atom with atomic
number Z. The one-electron Hamiltonian for electron i is (in a.u.)

H[0) = -\V}-- (8.18)

Let
ij(°) = £ t f f (8.19)
i=\
Total Hamiltonian (non-relativistic) is

H = H® + ZZ— ( 8 - 2 °)

Total wave function is assumed as a Hartree-product

N
V-(l,2,...,JV) = I J ^ ( i ) (8.21)
i=\

The atomic orbitals 4>i(i) are not assumed to be the hydrogen-like or-
bitals, they are unknown.
 258 Lecture notes on atomic and molecular physics; Erkog &: Uzer

Define

v:fl=Y,<<t>m-\Hj)> (S-M)
The Schrodinger equation (the eigenvalue equation) for electron i is

( i j f + V f " ) &(*)= £*&(*) (8-23)

Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com

This equation is called as the Hartree—equation.

by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

Define the Coulomb operator

Jj(i)=<<l>M~\<i>i(J)> (8-24)
Coulomb operator is an integral operator.

^e// = E 4 W (8-25)
We rewrite Eq.(8.23) as

H? + £ Mi)) Mi) = £iMi) (8-26)

Eq.(8.26) is an integrodifferentiaJ equation. There are N such equations,

one for each electron.

V?1* may be interpreted as: The ith electron is assumed to move

in a potential field due to the average charge distribution of the other
(N — 1) electrons.

The difficulty in solving the N equations, Eq.(8.26), comes from

Vy', because V^' depend upon the unknown AOs, {<f>i(i)}, i = 1,2,..., N.
Hartree showed that these orbitals could be obtained by an iterative
technique.

As a first approximation one assumes that

jj%)=<<f>f)U)\^f)U)> (8.27)
 Ch.8 Approximation methods for many-electron systems 259

Here {<j>j (j)} are a suitable set of approximate orbitals.

One then defines the first approximation to the effective Hamiltonian

(called the Hartree operator) by

H f) =i? (o) +Ej

tfU=;7f)+]rjj°)W ja) ( . ) (8.28)
(8.:
Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com

m
and solving the N equations
by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

WW^^W
HP4"H )=e«#« (8-29)
to obtain a first-improved set of orbitals { $ ( » ) } . Eq.(8.29) is a pseu-
doeigenvalue equation (operator depends on the solution)

Jf to=
jU(i)==<
< t?
d>f\j)\±\<t>f\j) > (8.30)

JT«>
Hf ^Hf £ j f3((*')
= HP ++£# i) (8-31)
Hi
ifl 2)
Hl 2 )
U?# ((}) =Ww
i) = e?U?)(i) (8.32)
The process is repeated until some set of coulombic operators {•/,•(*)}
leads to Hartree operator such that (for all values of i)
H(")c*H(n+l) = HsCF t ^(») * ^ - 1 ) = ^(ij (8.33)

At this point the orbitals are such that the charge distribution arising
from them will reproduce the field. One then says that the electrons
move in a SCF. This method is known as the Hartree-SCF method.
The final orbitals are usually referred to as SCF AOs.

The total energy of an iV-electron atom in the hartree-SCF approx­

imation is given by the expectation value of the correct Hamiltonian
(not the SCF Hamiltonian). The role of the SCF Hamiltonian is just to
generate the SCF AOs.

The total energy is then

< H >=< H{0) > + V < — > (8.34)

. . Tii
 260 Lecture notes on atomic and molecular physics; Erkog & Uzei

If the AOs are normalized

< fl<» >= •£ <fcCfllflflAW>= Ee! 0) (8-35)

i=i »=i

ej is the energy of a single electron i moving in some effective potential

field provided by the nucleus and the other N — 1 electrons.
Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com

The second part of the total energy is given by

by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

E < — >= E < &(«)W)1—lfc(*)fc(j) > (8-36)

= £ < ^(i)|j,(i)|^(i) >= £ < ^-(i)|Ji(i)|^(i) >= E Ji;

i<j i<J »<;'

The total energy given by the Hartree-SCF method is

^ = E40) +E ^ (8-37)
i=l i<j

For He atom E = -2.86 a.u.

Hartree-SCF solutions satisfy the Virial theorem. The Hartree eigen­

value equation for the kth electron is

(H® + E Mk) J MQ = £kMk) (8.38)

If the kth orbital is the highest occupied orbital, then — £k is just the
first ionization potential of the atom.

8.4 Hartree—Fock method

In the Hartree-SCF method it was assumed that each electron in an
iV-electron system moved in a potential field which was an average of
that provided by the remaining N — 1 electrons. The total electronic
wave function, however, was not chosen to be antisymmetric, and thus
 Ch.8 Approximation methods for many-electron systems 261

the Pauli principle was not satisfied. If one uses the antisymmetrized
Hartree product wave function and applying the variational principle one
obtaines a new set of self-consistent-field equations, which are called as
the Hartree-Fock SCF equations.

HF equations for closed—shell atoms:

Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com

Total non-relativistic Hamiltonian of an atom with 2iV electrons is

(in a.u.)
by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

H = Y, 2JVV + E — (8-39)
where
K = ~\^l ~ | (8-40)

is the one-electron Hamiltonian.

Consider the 2iV-electron wave function is constructed as a product

of 2N spin orbitals
2N
$ = Yl S^ifi) (Hartree product) (8.41)

$ = Si(l)S 2 (2)---S2tf(2AT) (8.42)

Spin orbitals:
SM -» hitiuip) (8-43)
(f>%{p) : Spatial orbital ; U{JJL) : Spin function.

$ = fc(l)a(l)fc(2)/J(2)fc(3)a(3)&(4)0(4) (8.44)

• • • <f>N(2N - l)a(2N - l)<t>N(2N)/3(2N)

We assume that the spatial orbitals are orthonormal,

<&0x)to(ji)>=fy (8-45)

The expectation value of total Hamiltonian, Eq.(8.39), with the anti­

symmetrized wave function may be written as

< H > = < $ | H | $ > = < *H\AHA\*H > = < $H\HA2\$H > (8.46)
 262 Lecture notes on atomic and molecular physics; Erkoc <fe Uzer

= VWi < <f>„\HA\$H > = ^ ( - l f < $H\H\$H >=

p

£(-1)* < Si(l)■ ■ ■ S2N(2N)\ J2K + Y, —\pW) ■ ■ ■ S2N(2N

r iu
p n n<v t
(8.47)
2
Here we have taken $ = A$H, A = y/NlA, A = (1/y/Nl) Ep(-^yP,
Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com

and 3># = Hartree product wave function. The summations over the
identity permutations for the one-electron part of Eq.(8.47) may be
by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

written as
27V 27V

E < K >= E < W I M W > ( 8 - 48 )

y.=l M = l
Because of the assumed form of the wave function, the successive
odd- and even-subscripted spin orbitals have the forms

S^fi) = &(/*)a(/0 , 5 M + I ( ^ + 1 ) = ^(A* + 1)/9(A* + 1 ) (8.49)

Thus, after integration over spin functions Eq.(8.48) takes the form
TV N
.(0)
}
2 £ < fc(/*)|/g*<M) > = 2 Y,e? (8-50)
r
i=l i=l
Therefore
27V TV

E<>V>=2E£S0) (8-5i)
Eq.(8.51) represents the zeroth order energy of the system which corre­
sponds to the independent-particle model result.

Two-electron portion of < H >:

For identity permutations we obtain

E < S^)S»(y)\-L\S^L)Sv{v) > (8.52)

Now whenever n is odd and v = /j, + 1 we have

SpM = <kQi)a{p) , Sv{v) = 4i(v)P{u) (8.53)

 Ch.8 Approximation methods for many-electron systems 263

There are N such occurences, so this gives

N 1 N
J2 < <t>i{v)<t>i(v)\—\<t>i{v)<t>i{v) > = £ Jij (8.54) (8-54)
r
i=l M" i=l

The general expression for the Coulombic integral is

Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com

Jij =< Hy)Hv)\—\^{fj)Hv) > ( 8 - 55 )

by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

For all other values of /z < v there are four ways one can obtain a given
Coulombic integral Jj,-.

(8.56)
( <h(vMv) j
Identity permutations then lead to a two-electron contribution:

N N
Y^Jij+Aj^Jij (8.57)
i=i %<}

In the case of two-electron permutations (odd parity) one obtaines

- J2 < S^S^l^-lS^S^u) > (8.58)

when (/, is odd and v — fi + 1 this leads to zero integrals as a result of

spin orthogonality. For the other values of fj, < v one obtains:

< ^ ( ^ ) ^ - W I — | ^ - ( / i ) ^ M > = Kij (8.59)

where Kij are called exchange integrals. Of the four // < v combinations,
only two lead to non-zero integrals for two-electron permutations. Thus,
the total contribution due to two-electron permutations is

N
2Y,Kij (8.60)
i<j
 264 Lecture notes on atomic and molecular physics; Erkog & Uzer

Permutations involving more than two electrons lead to zero integral.

Therefore,

< H >= 2 Y, 4 0 ) + E -fc+E( 4J«; - 2 K ^ (8-61)

i=l t=l i<7

Considering
Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com

Jij = Jji , Kij = Kji , Ju = Kn (8.62)

one can write
by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

N N N
<H>= 2 5>f ) + £ JU + 2 £ ( 2 J y - «■«) (8-63)
i=l i=l i<7

N N
-Kij)
i=l
i=\ ij
i,j
This equation, Eq.(8.63), is valid only for closed-shell atoms in which
the total wave function is approximated as a single determinant of dou­
bly occupied spatial orbitals.

We now wish to find the best possible orbitals to use in a wave func­
tion restricted to the single-determinantal form. Using the variational
method, we consider the functional
N N N
J = 2 ]T £ f) + £ ( 2 ^ _ Kij) - E \j{< <t>i\4>j > " % ) (8-64)
i=l i,j i,j

where Ay are Langrangian multiphers. We require that 8J = 0 for

small variations 6<f>i of these optimum orbitals. Define the Coulomb and
exchange operators, respectively,

Ji(p)<l>j(jt) =< 4>i{»)\ \<t>i{») > <f>j(») (8.65)

Kiirihifi) =< Mv)\ — \MV) > &0") (8.66)

Then the Coulomb and exchange integrals can be rewritten as

Jij =< <tn(ji)\Jj(n)\<l>i(ji) >=< <t>i{v)\Ji{v)\<f>j{v) > (8.67)

 Ch.8 Approximation methods for many-electron systems 265

Kij =< <k<P)\Kj(f*)\<k(ji) >=< M")\K&)\MV) > ( 8 - 68 )

The first-order variation in the functional J is

6J = 2J2(< 6<t>i\hn\<f>i > + < faKlSfa >) (8.69)

i
Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com

+ E ( < SfafiJj - Kj\& > + < 4>i\2Jj - Kj\b^ >)

by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

+ E ( < 6<f>j\2Ji - Kilh > + < ^\2Ji - Ki\6fy >)

~ E ( A v < s<k\4i > + A v < &!<% >) = °

The first and second double summations are symmetric in their indices
and lead to the same final sums. Thus

8J = 2 J2i< &4*\K + £ ( 2 J ; - KiM >1 (8-7°)

i j

+2^[< &|/v + £(2J; - Jtyltyi >]

i

- E ( A i ; < 6(t>i\<f>j > +*ij < HHj >) = o

Since h^ , Jj , .Kj are hermitian, the first and second summations

are just the adjoints of each other; also < <f>j\84>i > and < 6<j>i\<j>j > a
adjoints of each other. Then Eq.(8.70) may be written as

6J = 2 £ [ < 6<k\K + E ( 2 J ^ - Ki)\k > ~ E Av < 6k\*i >] (8-71)

+2 £ [ <ff&lfc„+ £(2J,- - #,■)!& >* " E Av < W . 7 >*1 = °

» i i
This is satisfied by the conditions

[hp + YpJi - Kj)\k = E h}4>i (8.72)

[^ + E( 2 J ; - KM = E A^* (8-73)
 266 Lecture notes on atomic and molecular physics; Erkog & Uzer

These two equations, Eqs.(8.72) and (8.73), are complex conjugate of

each other and equivalent. These equations are known as Hartree—
Fock equations.

Eq.(8.72) may be written in matrix form as

. F * = A$ (8.74)
Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com

where
T = ^ + Yt2Ji ~ K>) (8-75)
by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

is called the Hartree—Fock operator (HF operator).

* = [&&••-to] (8.76)
"An • • ■ MN
A= (8.77)
. Ajvi • •• ^NN .
The eigenvalue equation for the HF operator F with the spatial orbitals
{ & } ; i = l,2,...,JV
F<j>i = Bi4>i (8.78)
where fa are HF orbitals and Ei are the corresponding HF eigenvalues.

Si =<fa\F\fa>= s\0) + Y,(2JiJ ~ Kit) (8.79)

i
Total Hartree-Fock energy, Ejjp is given by

EHF = £ > + £ ! 0) ) = 2 £ £|
0)
+ ^ ( 2 Jy - Kij) (8.80)
»=1 i=l ij

One can write the total energy of a (2N — l)-electron system as

E+ = EPF(2N - 1) = £ f c + £?) - ek (8.81)

i=i
similar to this one, for a (2N + l)-electron system
N
E~ = J5Jg(JLN + I) = Y,& + 1 40))+£m (8.82)
j=i
 Ch.8 Approximation methods for many-electron systems 267

It then follows that

E+ - E = -ek (8.83)
E~-E = +em (8.84)
where E represents the energy of the neutral system.
Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com

According to Koopman's theorem —ejt and + e m can be regarded as

an ionization potential, and the electron affinity, respectively, of the sys­
tem described by ip(2N).
by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

Hartree-SCF and Hartree-Fock methods give the He ground state

energy as —2.86 a.u. Although the variational method gives more accu­
rate than this value, it is not practical for many-electron atoms; one has
to use Hartree-SCF or Hartree-Fock methods.

The functions chosen to represent AOs (4>neme(T,9,(f>)) are usually

combinations of the spherical harmonics (Yem((Q, <j>)) and one of the fol­
lowing radial functions (Rni(r)):

1. Hydrogen-like functions (orthogonal)

Rrj{r) = reL™+el (—\ e~Zr/n ; L : Laguerre functions (8.85)

2. Slater functions (not orthogonal)

Rnt(r)=rn-1e-ar ; » = 1,2,... (8.86)

3. Gaussian functions (orthogonal)

i^(r) = r ^ e " " 3 ; n = 0,l,2,... (8.87)

< W , * <S>) = Rni(r)Yemi(0,4>) (8-88)

8.5 The electron correlation energy

The HF energy is usually within about 1% of the experimental value (to­
tal energy). However, one is usually interested in energy differences, e.g.,
the energy difference between two spectroscopic states. Unfortunately,
 268 Lecture notes on atomic and molecular physics; Erkoc & Uzer

these energy differences themselves are often no larger than about 1%

of the total energy of either state. Thus, small absolute errors in the
total energies may easily lead to large relative errors in their differences.
For this reason there is a tremendous amount of interest in quantum-
mechanical calculations which give better energies than the HF method
does. However, this method is used as a sort of reference point for more
accurate calculations. Let us call EJJF the Hartree-Fock energy of a
Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com

given state and ipjjF the corresponding wave function. Both EJJF and
xpHF axe only approximations to the exact energy Eexact and the exact
by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

wave function ipexact of the non-relativistic Hamiltonian.

P.O. Lowdin (in 1959) denned the electron correlation energy as

E„r =<H> (exact)- <H> (HF) = E ^ - ERF (8.89)

Correlation energy in HF method reflects the fact that Coulombic

interaction between pairs of electrons, especially electrons with antipar-
allel spins, is not properly accounted for. Electrons of parallel spins are
kept apart by the Pauli principle, an effect which overrides the Coulom­
bic repulsion, and are thus described somewhat better than electrons of
antiparallel spin.

The quantity E^aa is the exact energy of the non—relativistic Hamil­

tonian and hence is not quite the same as experimental energy. A certain
amount of correlation is already included in the ipjjF because of the fact
that it is totaly antisymmetric. Ecorr represents the correlation effects
not included in the tpjjF-

8.6 Many—electron theory of Sinanoglu

It is well known that the HF method gives a fairly reasonable represen­
tation of the shell structures and electron densities of atoms. Relying
partly on this observation, O. Sinanoglu (in 1961) presented an analysis
of the correlation problem by writing the exact wave function in the
form
^ = <h + x (8.90)
 Ch.8 Approximation methods for many-electron systems 269

where 4>o is the HF wave function and x is a wave function representing

the correlation correction. Furthermore,

<<£o|x>=0 , < i>\4> > = 1+ < x\x > (8.91)

The HF energy in terms of spin orbitals can be written, in general,

Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com

2N 2N
EHF =< <i>o\H\<i>o > = ] £ > + E ( J v ~ KH) (8-92)
i i<j
by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

using the wave function, Eq.(8.90), the exact energy can be written

Eexact { ]
~ <viv>
= p 2 < <j>0\H - EHF\X > + <x\H- EHF\X >
HF
1+<X|%>
Sinanoglu finds it convenient to introduce the following operators

ei = T - ei (8.94)

mi; = — + Ji} - Kij - Gi(j) - Gj(i) (8.95)

Gi(j) = Ji(j) ~ KM) (8.96)

The correlation energy is then given exactly by

Ecorr = E^act ~ EjJF (8.97)

_ 2 < 4>o| Ei<j mjlx > + < x\ Eie» + Ei<j mj\x >
1+ < xlx >
The two-electron operator rriij is called a fluctuation potential.
The problem now is to find x such that the total energy is minimized.
The function x c a n be represented exactly by an expansion in terms
of functions which correspond to successively higher-excited configura­
tions.

Sinanoglu has shown that the first-order wave function V of an

iV-electron system can be broken down into terms involving only pair
 270 Lecture notes on atomic and molecular physics; Erkog & Uzer

functions. These pair functions satisfy equations just like those of an

actual two-electron system expect that now each electron moves in the
HF field of the entire system of particles.

The exact energy is approximated by

E<EHF + Y,£v ( 8 - 98 )
Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com

i<j

where ey are pair correlation energies determined by separate mini­

by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

mizations for each pair of electrons. The advantage of the Sinanoglu

formulation is that it reduces the solution of one iV-body problem to
the solution of ( ) = ^N(N — 1) two-body problems.

8.7 Slater Atomic Orbitals and screening con­

stant rules
The simplest and the most practical method of calculating the energy
levels of many-electron atoms is the Hartree-SCF procedure. In this
method the atomic orbitals come out as numerical tables. Since such
wave functions are not practical to use, J.C. Slater (in 1960) has pro­
posed analytical functions,

Kn = rn'-le-z^ln*Ylmt{e, <f>) (8.99)

which approximate the atomic orbitals, where Yimt(6, cf>) is a spherical

harmonic in real form. These functions (STOs) are wave functions de­
scribing the motion of a single electron in a central field in which the
potential energy is given by

Tr. , Z-ni n*(n* — 1) .

V(r) = - ^ + \r2 I (8.100)

The quantity n* is an effective principal quantum number and Z^i is an

effective nuclear charge. Both of these quantities are treated as empiri­
cal parameters. The Slater orbitals have no nodes in the radial portion
whereas the hydrogen-like AOs have n — I — 1 such nodes.
 Ch.8 Approximation methods for many-electron systems 271

Slater has shown that satisfactory values of n* and Znt may be cho­
sen by use of some simple rules. For K, L, and M shells, one uses
n* = n. For JV, O, and P shells n* equals 3.7, 4.0, and 4.2, respectively.

The effective charge Z^i may be written

Zni = Z-Sn£ (8.101)

Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com

where Z is the atomic number and S„£ is a screening constant depending

by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

upon the orbital cf)^ and the particular electronic configuration.

The screening constants are evaluated according to the following

empirical rules:
1. Divide the orbitals into the following groups, each group having a
different shielding constant:
(Is) , (2s, 2p) , (3s, 3p) , (3d) , (4s, 4p) , (4d) , (4/) , (5s, 5p)

2. The shielding constant S,u is the sum of the following contribu­

tions:
a) Zero from any shell outside the one considered.
b) 0.35 from each other electron in the group considered, except
that 0.3 is used for the Is orbital.
c) For s and p orbitals, 0.85 is subtracted for each electron in the
next inner shell and 0.1 for all electrons still further in. If the
orbital is d or / , 1.0 is subtracted for every electron within this
orbital.

The Slater rules tend to become unreliable when applied to orbitals of

principle quantum number 4 or greater. The above rulas yield Z\s — 1.7
for the ground state of He. For the K and L electrons of the ground
state of the carbon atom one obtains 5.70 and 3.25, respectively.

5 is = 0.3

S^ = 0.35a: + 0.85y + z ; n > 1 , £ = 0,1 (for 2s, 2p, 3s, 3p)

x = # of remaining electrons in the same shell

 272 Lecture notes on atomic and molecular physics; Erkog <fe Uzer

y = # of electrons in the shell with p.q.n. (n — 1)

z = # of electrons with p.q.n. < (n — 2)

SM = 0.35a; + y

x = # of remaining 3d electrons
y = # of electrons with n < 3 and ^ < 2
Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com

Fore more accurate calculations the screening constants may be de­

by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

termined variationaly. The equations obtained by E. Clementi and D.L.

Raimondi (in 1963) are (valid for Z = 2 — 36 and ground state config­
urations) :

Su = O.S[N(ls) - 1] + 0.0072[iV(2s) + N(2p)] 8.102)

+0.0158[iV(3s) + N(Zp) + N(4a) + N(5d) + N{Ap)\

S2s = 1.7208 + 0.3601 [N(2s) + N(2p) - 1] 8.103)
+0.2062[iV(3s) + N(3p) + N(4s) + JV(3d) + N(4p)}
S2p = 2.5787 + 0.3326[iV(2p) - 1] - 0.0773iV(3s) 8.104)
-0.0161 [JV(3p) +N(4a)] - 0.0048iV(3d) + 0.0085AT(4p)
S3s = 8.4927 + 0.2501 [N(3s) + N(Sp) - 1] + 0.0778iV(4s) 8.105)
+0.3382iV(3d) + 0.1978iV(4p)
S3p = 9.3345 + 0.3803[iV(3p) - 1] + 0.0526iV(4s) 8.106)
+0.3289iV(3d) + 0.1558iV(4p)
S4s = 15.1505 + 0.0971[iV(4s) - 1] 8.107)
+0.8433iV(3d) + 0.0687iV(4p)
S3d = 13.5894 + 0.2693[iV(3d) - 1] - 0.1065iV(4p) 8.108)
S4s = 24.7782 + 0.2905[iV(4p) - 1] 8.109)
 Ch.8 Approximation methods for many-electron systems 273

Table 8.1: Values of Z^ in Slater AOs:

z Atom Zrd
Is (2«,2p) (3*,3p)
1 H 1.00
2 He 1.70
Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com

3 Li 2.70 1.30
4 Be 3.70 1.95
by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

5 B 4.70 2.60
6 C 5.70 3.25
7 N 6.70 3.90
8 0 7.70 4.55
9 F 8.70 5.20
10 Ne 9.70 5.85
11 Na 10.70 6.85 2.20
12 Mg 11.70 7.85 2.85
13 Al 12.70 8.85 3.50
14 Si 13.70 9.85 4.15
15 P 14.70 10.85 4.80

8.8 Roothaan method; H F method

for molecules

The basic approach to the calculation of molecular orbitals consists of

efforts to solve the Hartree-Fock equations which are formally identical
for atoms and molecules. We have seen that the problem is already quite
complicated for atoms; in molecules, the absence of three-dimensional
rotational symmetry and the occurrence of multicenter integrals increase
the complexity to an extent that direct, iterative, solutions of the HF
equations are almost impossible. Instead, one postulates that molecular
orbitals are to be expressed as linear combinations of atomic orbitals
(LCAO). This is known as the Roothaan method. We will consider the
closed shell molecules, actually many molecules in their ground state
have close shell configurations. Since all the electrons in a closed shell
have their spins paired (S — 0) the total antisymmetric wave function
 274 Lecture notes on atomic and molecular physics; Erkog & Uzer

can be represented by a single Slater determinant

V-i(i) &(i) VJV(1) ^JV(I)

^(2) for(2)
y = (8.110)
V-i(2) &(2)
VWV-
^(2iV) ^(2tf)
Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com

*h(2N) fa(2N)
As in the atomic case, the objective is to find the orbitals that will
minimize the energy
by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

£=<#|iJ|tf> (8.111)

where H is the complete non-relativistic electronic Hamiltonian of the

molecule.

The molecular orbitals that satisfy the variational principle are ob­
tained from solutions of the Hartree-Fock equations; for a closed shell
they are of the form
F(l)Vi(l) = fiilfo(l) (8.112)
where
F(l) = h0(l) + '£[2Jh(l)-Kk(l)] (8.113)
k

2m (8.114)
y %
Jit and Kk are the Coulomb and exchange operators, respectively, which
depend on the molecular orbitals themselves and are denned as

J fc (l)Vi(l) = < ^ ( 2 ) | — |Vfc(2) > Vi(l) (8.115)

Kk(l)il>i(l) =< ^ ( 2 ) | — |^(2) > -0fc(l) (8.116)

T\2
The total energy is

(8.117)
 Ch.8 Approximation methods for many-electron systems 275

We now introduce the constraint, due to Roothaan (1951), that the

molecular orbitals are to be expressed as linear combinations of atomic
orbitals:
c
v-i = E ^ (8-118)
or in matrix form
Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com

* = $C (8.119)
where * and <& are now matrices and C is a square matrix. The <£M are
by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

the atomic orbitals and they are centered on various nuclei. The <^ are
not necessarily orthogonal. Hence we define

-V =< 4M» > (8-12°)

as the overlap integral for the orbitals cj>^ and 4>v . The molecular or­
bitals, on the other hand, are required to satisfy the orthogonality con­
dition
<ipi\ipj>=6ij (8.121)
so that in the LCAO approximation,

Y, C;iCuj < ^ „ > = £ C^CjS^ = 6ij (8.122)

which in matrix form

C+SC = 1 (8.123)
The charge density at the point r is defined by
occ
p(r) = 2 j > ? ( r ) i k ( r ) (8.124)
i

the factor of 2 appearing because of the double occupancy of the molec­

ular orbitals by two electrons with opposite spin. In terms of atomic
orbitals

p(r) = 2 £ Y, C*C«*1{T)M*) = £ Pr*W<Mr) (8-125)

where occ
p^ = 2^c;ia i (8.126)
 276 Lecture notes on atomic and molecular physics; Erkoc &: Uzer

is the density matrix.

If N is the number of occupied (spatial) orbitals, the closed shell will

contain 2N electrons; therefore the integral of the charge density over
all space must be equal to 2N:

l p(r)dr = £ P^ f ^ ( r ) ^ ( r ) d r = £ P^S^ = 2N (8.127)

Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com
by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

With these relations

/i 0 (l)Vi(l) = E fto(i)c^(i) ( 8 -l 2 8 )
v

J fc (l)Vi(l) = E C*\kCokCvi < <}>x{2)\ —|6r(2) > 6,(1) (8.129)

K f c (l)^(l) = J2 CikCokCvi < 4>x{2)\ —16,(2) > 6.(1) (8.130)

F(1)V<(1) = E ^ ( l ) C ^ ^ ( l ) + E ChP*kC* (8.131)

" fcAcrt,
2 2
x[2 < ^(2)1 — 16,(2) > MV- < MV\ — \M2) > <Mi)l
r\2 J"i2
= Ee*c^«'(i)
This equation is the HF equation in LCAO form. Let us introduce the
notation
e2
< 6.(l)fo(2)| — | 6 , ( l ) M 2 ) > = < H ^ > (8-132)
ri2
This definition implies

< iJ,v\\a > = < \a\nv > , < /J,V\\C >*=< ufj,\a\ >* (8.133)

Multiplying both sides of Eq.(8.131) by <£* (1) and integrating over the
coordinates of electron (1) we obtain

E CviH^ + E CUC0kCyi[2 < fw\\ff > - < /JUT\XU >] (8.134)

C kXrru
 Ch.8 Approximation methods for many-electron systems 277

= ^iCviSpv
v
in which
#^=<(^(l)Ml)|<Ml)> (8-135)
inserting the density matrix,
Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com

£ C
^ 2 CviH^ + ^ P\a[< f*v\\(T > - X < / ^ l A f >] = 5 Z i <nSfj.u

(8.136)
by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

Finally, by writing

F,», = Hp, + J2P\A< HA^ > - r < H A ^ >] (8.137)

2
A<r

the HF equations assume the compact form

£ ( F ^ - EiS^Cvi = 0 (8.138)

or in matrix form,
(F-eS)C = 0 (8.139)
where e is a diagonal matrix whose diagonal elements £j are the or­
bital energies. The set of homogeneous equations will have nontrivial
solutions only when
\F^ - Si«V| = 0 (8.140)
The LCAO approximation has converted the original partial differential
equations (Hartree-Fock) into algebraic equations (Roothaan).

The total energy of a closed shell was shown to be

E = 2 J2 k+5>J* - K^ ( 8 - 141 )
i i,k

In the LCAO appproximation we have

C
£f) = < ^(1)1/10(1)1^(1) > = £ UC^H^ ( 8 - 142 )
)1V

e2
Jik =< ^ ( 1 ) ^ ( 2 ) 1 — 1^(1)^(2) > (8-143)
T-12
 278 Lecture notes on atomic and molecular physics; Erkog <fe Uzer

= E cifi\kcvicak < HA* >

Kik =< ^ ( 1 ) ^ ( 2 ) 1 — 1^(1)^(2) > (8-144)

ri2
= E C^filkCuiC.k < /MT\XU >
Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com

fiXva

Substituting in total energy, E, the expression for the energy becomes

by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

E P
=E »»H^ +2 E P
^PA«T[< H ^ > - - < H A i / >] (8-145)

p
= - E ii"(H^ + F^u)

8.9 Computational methods for molecules

The Roothaan equations into which the HF equations have evolved as
a result of the LCAO approximation still require an iterative procedure
because the matrix elements JFM„ depend on unknown coefficients (or
density matrices). A typical calculation involves the following steps:

1. Specify a set of atomic orbitals (the basis set). Among the popular
choices are the Slater orbitals (STO) and Gaussian orbitals.

2. Compute the overlap integrals SM1/.

3. Compute the core Hamiltonian matrix H^.

4. Compute the two-electron integrals < /j,v\\a >.

5. Compute the eigenvectors of HM„. This gives a starting set of

LCAO coefficients (and the initial set of molecular orbitals).

6. Assign electrons in pairs to the lowest molecular orbitals until all

the electrons have been assigned.

7. Compute the density matrix P,

 Ch.8 Approximation methods for many-electron systems 279

8. Compute the total electronic energy

i 1
E
= L P^H^ + - £ P^PxA< pwfi" >~o< vAv° >]
(8.146)
9. Compute the Fock Hamiltonian
Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com

F^ =Hp, + Y, P\*[< H A C T > - o < ^X\ua >1 (8-147)

by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

10. Compute the eigenvectors of F^ to obtain a second set of LCAO

coefficients.
11. Continue until the total energy E remains constant to the required
accuracy.
Despite the simplification in the HF formalism resulting from the
introduction of the LCAO approximation, the most series limitation in
MO calculations is the large number of integrals that must be evalu­
ated. Many integrals are of the two-electron, multicenter type which
consume most of the computer time. There are some approximations
which reduce the number of integrals without lossing the accuracy of
the calculations. We will summarize some of these approaches here.

1- Neglect of differential overlap (NDO). This consists of setting

^ M (l)^„(l) = 0 for fj, y£ v in the electron repulsion integrals, and has the
effect of eliminating all three- and four-center integrals.

2- Neglect of all but valence electrons. Atomic orbitals, chosen ex­

clusively from the valence shell of each atom, constitute the basis set
for the LCAO expansion. Inner shell electrons, because they are more
tightly bound, do not participate to any significant extent in the forma­
tion of molecular orbitals. Empirically, it is known that chemical effects
are largely due to the valence electrons.

3- Except in the very simplest cases, some use is made of exper­

imental information such as atomic ionization potentials and electron
affinities. These are used to assign numerical values to a number of the­
oretical parameters. A popular version of MO theory is known as the
 280 Lecture notes on atomic and molecular physics; Erkoc & Uzer

CNDO method (complete neglect of differential overlap) in which con­

dition ^(1)^^(1) = 0, (/i 7^ v) is used for all two-electron interaction
integrals. In this scheme, using the convention that (f>M is on atom A and
cj>u on atom B,

Fm = U^ + \P^1W + £(F- - Zrftr (8.148)

Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com

F^ = H^ - -P^^ (ji ^ v) (8.149)

by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

In these expressions Z7M/X is a core integral which is approximated by

Upn = —Ifj,, where 1^ is the atomic ionization potential for the orbital
<j>fi. The condition of zero differential overlap is employed so that
< fj,X\va >—< fj,fj,\up > 8^x8^0- (8.150)
7^„ is the two-electron repulsion integral, 7M„ = < MJ\VV >. When
fj, = v, the integrals are approximated by
7W =Ifl-All (8.151)
where A^ is the electron affinity in the orbital </>M. H^ is a core reso­
nance integral which is often approximated by setting

H^v = 0 for fi and v not nearest neighbors,

R^v — empirical constant when fj, and v are bonded nearest neighbors.

Zv is the core charge of atom B and is equal to the nuclear charge

minus the number of inner shell electrons.

Another approximation which is worth to mention is the Extended

Hiickel Method (EHM) in which all, or almost all, integrals are replaced
by empirical parameters thereby eliminating the self-consistent, itera­
tive procedure. If the secular equation {F^ — eiS^l = 0 is written as
\Hfiv — £iS^\ = 0, the diagonal elements are assigned empirical values
depending on the orbitals and the type of molecule. The off-diagonal
elements are approximated by

H^ = -K{H^ + H^S^ (8.152)

 Ch.8 Approximation methods for many-electron systems 281

in which if is an empirical parameter and S^ the overlap integral for

atomic orbitals <^ and (j>u centered on the appropriate atoms. Having
made the numerical assignments; solutions to \H,u, — BiS^l = 0 yield
the eigenvalues £i and hence the coefficients C^ in the molecular orbitals
Ipi = L M CpiVr-
Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com

8.10 Worked examples

by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

Example - 8.1 :

Using Slater rules calculate (a) the ionization potentials of lithium atom,
(b) the 2p orbital radius of carbon atom to estimate the atomic size.

Solution :

a) Li(ls22s) : Zls = 2.7 , Z2s = 1.3 ; Eu = 2Els + E2s

Eu = [-2(2.7/l) 2 - (1.3/2)2] x 13.6 eV = -204.034 eV

Li+(ls2) : Zls = 2.7 ; ELi+ = 2Els

EIA+ = - 2 ( 2 . 7 / l ) 2 x 13.6 eV = -198.288 eV

Li++(ls) : Zls = 3.0 ; ELi++ = -(3.0/1) 2 x 13.6 eV = -122.4 eV

I Pi = ELi+ - ELi = 5.746 eV ; {IPi)exP = 5.39 eV

IP2 = Eu++ - ELi+ = 75.888 eV ; (IP2)exP = 75.7 eV

IPZ = -ELi++ = 122.4 eV ; ( J P 3 W = 122.4 eV

b) C(ls 2 2s 2 2p 2 ) , Z2p = 3.25

n2 4
r =
e// = T~"o ^9?°° = L23a
° = °"65 A
 282 Lecture notes on atomic and molecular physics; Erkoc & Uzer

(reff)ezp = 0.62 A

It is a good approximation.

Example - 8.2 :
Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com

Calculate the ionization potentials of Li atom using the screening con­

stants determined by Clementi and Raimondi.
by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

Solution :

Li(ls22s):
Sls = 0.3 x (2 - 1) + 0.0072 x (1) = 0.3072
Zis = Z - Su - 3 - 0.3072 = 2.6928
S2s = 1-7208 -(- 0.3601 x (1 - 1) = 1.7208
Z2s = Z - S2s = 3 - 1.7208 = 1.2792
Eu = 2Els + E2s = [-2 x (2.6928/1)2 - (1.2792/2)2] x 13.6 eV =
-202.795 eV

Li+(ls2):
Su = 0.3 x (2 - 1) = 0.3
Zu = Z - Sis = 3 - 0.3 = 2.7
Eu+ = 2Eis = - 2 x (2.7/1) 2 x 13.6 eV = -198.288 eV

Li++(ls):
Sis = 0.3 x (1 - 1) = 0
Zis = Z — Sis = 3.0
EU++ = Eu = -(3.0/1) 2 x 13.6 eV = -122.4 eV

IPi = Eu* ~ Eu = 4.507 eV ; (IPi)^ = 5.39 eV

IP2 = Eu++ ~ Eu+ = 75.888 eV ; {IP2)^ = 75.7 eV

IP3 = -ELi++ = 122.4 eV ; (/P3)exJ, = 122.4 eV

Example - 8.3 :
 Ch.8 Approximation methods for many-election, systems 283

Using Slater rules calculate (a) the screening constants, (b) the first
ionozation potential, and (c) the atomic size for silicon atom.

Solution :

a) Si configuration: ls 2 2s 2 2p 6 3s 2 3p 2 , Z = 14
Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com

Su = 0.3 ,Zls = Z- Su = 14 - 0.3 = 13.7

S2s = S2p = 0.35a: + 0.85y + z = 0.35 x 7 + 0.85 x 2 + 0 = 4.15
by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

Z2s = Z2p = Z- S2s = 14 - 4.15 = 9.85

S3s = S3p = 0.35a: + 0.85y + z = 0.35 x 3 + 0.85 x 8 + 2 = 9.85
Zzs = Z3p = Z- S3s = 14 - 9.85 = 4.15

b) Si+ configuration: ls 2 2s 2 2p 6 3s 2 3p , Z = 14

Sls{Si+) = Sls(Si) , S2s(Si+) = S2s(Si) , S2p(Si+) = S2p(Si)

S3s(Si+) ± S3s(Si) , S3p(Si+) / S3p(Si)

S3s = S3p = 0.35x + 0.85y + z = 0.35 x 2 + 0.85 x 8 + 2 = 9.54

Z3s = Z3p = Z- S3s = 14 - 9.54 = 4.46
Z\s = 13.7 , Z2s = Z2p = 9.85

IPX = (2Els + 8E2s + 3 ^ s ) S i + - (2Els + 8E2s + 4E3s)Si

= 3^s(5i+) - 4E3s(Si)

- 3 M U - A ( - ^ - \ __3X(4.46) 2 4X(4.15)2
2 X 3 2
\ L V 2x32JSi- 2x3* +
2x3* *"•
S -3.3153 + 3.8272 = 0.5119 a.u. = 0.5119 x 27.21 eV Si 13.93 eV
{IP\)exP = 8.149 eV

c)
2 2 2
reeffff = —ao = —ao = — x 0.529 A S 1.147 A
Ze 4.15 4.15
Tori, = 1.068 A (quantum mechanical calculation).
Copyrighted Material
 284 Lecture notes on atomic and molecular physics; Erkog <fe Uzer

Example - 8.4 :

Consider the ground state of a neutral atom with nuclear charge Z.

Assume that all subshells with the same p.q.n. n are filled; that is, we
are considering closed shell atoms like He, Ne, Ar, etc. Further, assume
that the electrons are non-interacting. The orbital energies of the atoms
then become hydrogenic.
Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com

a) Show that the forgoing assumptions lead to the relation

Z = no(no + l)(2no + l ) / 3 , where no is the p.q.n. of the highest occupied
by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

atomic orbital.
b) Determine the ground state energy in terms of no-
c) The Hartree-Fock ground state energies for He and Ne are —2.86 and
—128.5 a.u., respectively. Calculate the ground state energy for He and
Ne, using the energy expression from (b). Comment on the discrepancy
between those energies and the HF energies.
d) The effect of electronic screening is often approximately taken into
account by using screened hydrogenic orbitals rather than exact hydro­
genic orbitals. In a screened hydrogenic orbital the orbital exponent
is Zni — Z — Sni, where S^ is the screening constant. Use Slater's
screening constants to calculate the ground state energies of He and
Ne. Comment on the accuracy of this approach relative to HF and the
non-screened approach.

Solution :

a) The number of electrons is

"0 n—1 £ no n - 1 710 /„ .. ■,
Z
= 2z2z2 E l = 2 £ X > m ) = 2 $ > 2 = no(no + l ) ^ ^
n=l £=0 mt=-£ n = l £=0 71= 1 "*

b) The number of electrons with p.q.n. n is Nn = 2n 2 . The orbital

energy of such an electron is en = -Z2/(2n2). The ground state energy
becomes

E = JT enNn = -n0Z2 = - n g f o + l ) 2 ( 2 n ° + 1)2

9
71=1

c) Ejje = - 4 a.u. , no = 1 , Ejjp = -2.86 a.u.

 Ch.8 Approximation methods for many-electron systems 285

ENe = -200 a.u. , 7io = 2 , EJJF = -128.5 a.u.

Both the HF and the present hydrogenic approximation are independent-

particle models. In the former, an electron moves in the average poten­
tial from all the other electrons. This average e — e interaction, which
is absent in the hydrogenic approximation, is thus very important for
obtaining a realistic independent-particle model description.
Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com

d) Zu{He) = 1.70 -> EHe ~ -2.89 a.u.

by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

Zls(Ne) = 9.7 , Z2s(Ne) = Z2p(Ne) = 5.85 -* ENe = -128.31 a.u.

The screened results are in very close agreement with the HF energies.

Example - 8.5 :

In semiempirical molecular orbital methods, such as Hiickel and ex­

tended Hiickel methods, the ground-state total energy is usually esti­
mated as £?o = E r n*-£r i where nr is the occupation number of molecular
spin orbital ij)r; that is, nr — 1 if ipr is an occupied molecular orbital
and nT — 0 if ipr is an unoccupied molecular spin orbital. er is the
molecular orbital energy. Let Vnn and Vee denote the average value of
the nuclear and electron repolsion operator with respect to the exact
Hartree—Fock ground state. If we assume that the Hartree-Fock total
energy can be approximated as above, where e denotes the Hartree-Fock
orbital energies, show then that this assumption implies that Vnn = Vee .

Solution :

The exact Hartree-Fock total energy can be written as

EHF = 2 E r ^ - Ezri2(fcfc|M) - mm+E„>, ^

Or EHF — E r nr£r + K n ~ Ke

and Vnn = Vee follows straightforward when EQ = EHF

 286 Lecture notes on atomic and molecular physics; Erkog <fe Uzer

Example - 8.6 :

Assuming that the Born-Oppenheimer approximation is valid, deter­

mine the conditions under which the potential energy curves for two
different electronic states of a diatomic molecule can cross. (Hint: As­
sume that all of the wave functions are known except those associated
Lecture Notes on Atomic and Molecular Physics Downloaded from www.worldscientific.com

with the two states in question. Then expand these two states in terms
of the functions xpi and tfo which are orthogonal to each other and to all
by UNIVERSITY OF BIRMINGHAM on 01/02/16. For personal use only.

of the known wave functions.)

Solution :

As suggested we expand the unknown states in term of il>\ and ip2-

u = ciipi + C2ip2- The coefficients can be determined by solving the
2 x 2 secular determinant:

.ffii - E H\2
= 0
H\2 H.22 — E

The solutions are u = -75 001 ± V^) and

E± = \(Hn + H22) ± y(H22-Hn)2+4Hf2

Assuming that [2Hu/(H22 - Hu)]2 < 1 and taking H22 > i?n, E± can
be expanded to obtain:

H H
E\ — i?n — — h ——I.... , w.£?2_ =nr_
H22.• +. u
H22-H11 H22 — H11
In order for the levels to cross for some value of R so that E\ = E2
we must have Hn - H22 = 0 and H\2 = 0. Otherwise there will be no
simultaneous solutions for those equations. If ^1 and i/'2 have different
symmetry properties, if 12 = 0 since H is totaly symmetric. In general, if
ipi and ip2 have the same symmetry properties H\2 ^ 0 and no crossing
can occur. States having different multiplicities have different symmetry
properties and thus may cross.