abstract - shodhgangashodhganga.inflibnet.ac.in/bitstream/10603/21744/3/ch-1.pdf · evaluation of...
TRANSCRIPT
1
ABSTRACT
This chapter is devoted to the general discussion related to the present
thesis, the systems considered, the literature survey and the methods of calculation
are explained.
2
PART (A)
MOLECULAR POLARIZABILITY
(1A.1) INTRODUCTION :
Semi-empiricial approaches are of considerable interest for studying
molecular structure and properties of compounds. For the macroscopic
estimates of the properties of molecules, the concept of characteristic structure
is shown to be extremely helpful. Molecular polarizabilty is considered to be
one of the important properties in view of its several applications. Hasse1
evaluated the polarizabilities for the lithium ion and helium atom by using the
Lenard-Johnes potential model involving several ground state as well as
perturbed state wave functions in terms of perturbing potential. Hirschfelder2
computed the polarizability of molecular hydrogen and diatomic molecular
hydrogen ion applying the variational method of Hylleraas3 and Hasse1 and
using the eigen functions proposed by Rosen4 and Wang5. It was found that
the Kirkwood method6 is applicable to the diatomic hydrogen ion and hence
polarizabilities were obtained for many internuclear separations by using the
eigen functions developed by Guillemin and Zener7. While comparing the
values calculated by Hirshfelder2 with those of Mrowka8 and Steenshold9,
Van Vleck10 and Atanosoff11 have pointed out that the method due to Hylleraas3
is not strictly applicable to approximate eigen functions and may give values
fluctuating either high or low to the real values. Buchingham12, based on
3
Kirkwood's variational method6 , made an effort to evaluate it in the case of
heavier atoms and obtained complex self consistent field wave functions. Bell
and Long13 have calculated the polarizabilities for H2+ and H2 molecule from
six differential wave functions used. Abatt and Bolton14 on the other hand
used the polarizability of N2 molecule as a criterion for determining the
molecular wave functions of a system by SCF method. Kolker and Karplus15
used ab initio method to calculate the polarizability tensor for a series, of
diatomic molecules of first row elements.
The use of δ-function potential model was initially made by Ruedenberg
and Parr16 and Ruedenberg and Scherr17. Later, Frost18, applied the δ-function
potential model to calculate energies of various systems with the introduction
of a branching condition. This was followed by Lippincott19 with a semi
empirical δ-function potential model. Lippincott and Dayhoff20 used the semi
empirical δ-function technique in predicting vibrational frequencies,
anharmonicities, bond dissociation energies and equilibrium internuclear
separations of diatomic molecules.
Lippincott and Stutman2l applied this semi empirical model to evaluate
polarizabilities of different components (bond parallel and bond perpendicular)
in order to compute the mean molecular polarizabilites for diatomic and
polyatomic molecules.
4
Balodi and Semwal22 applied this semi empirical to determine bond
and molecular polarizabilities of Ferrocene and its derivatives and interpreted
the results in terms of observations of Waite and Papadopoulous and
others23-24. We have also applied this method for the determination of bond
and molecular polarizabilities of buckminsterfullerene and a number of
aromatic compounds and compared the calculated values with those obtained
by Semwal and Balodi25 using Miller’s method. Semwal and Uniyal26-28 have
reported the molecular polarizability of C60, C70, C84 buckminsterfullerenes.
The model assumes a potential which is infinite at the nucleus and zero
everywhere else. The integral of the potential over all space, is finite and equal
to a parameter which is called ‘ the δ-function strength’ or ‘reduced
electronegativity’ and is analogous to the effective ‘nuclear charge’ used by
Slater.
At the nucleus, a δ-function wave function is generated representing
the probability amplitude of the electron for this isolated nucleus. These δ-
function atomic orbitals are then linearly combined to form molecular orbitals
with the restriction that only two atoms may interact at a time provided if bond
exists between them. The major advantage of a δ-function model lies in its
one dimensional nature. The model assumes every bond in a molecule to be
separate entity. Since the potential has a non-zero value at only two points
5
along any given bond, the problem of calculation are trivial.
Thus careful literature survey reveals that the methods of calculating
polarizabilities can be grouped under the heads (i) Variational method
(ii) Molecular orbital Method and (iii) Semiempirical δ-potential function
method.
A molecular system may be well realised as a cluster of positive and
negative charge distributions. The force of interaction between the positive
and negative charge distributions consists of (i) attraction forces and
(ii) repulsion forces. If r be the intermolecular separation and F(r) the force of
interaction, the inter molecular potential function Φ (r) may be defined as
Φ (r) = ∫r∞ F (r) dr (1A.1)
where F(r) is the force of attraction for r>>> and the force of repulsion
for r<<<, since the nature of F(r) is dependent on the intermolecular distance,
it has been classified29 into three categories as
(i) Short range forces - applicable when molecules come close enough for
the charge distributions to overlap, hence repulsive in nature.
(ii) Intermediate forces - applicable when there is only small overlap.
(iii) Long range forces - applicable when there is no overlap, i.e. the molecules
are far apart and hence attractive in nature (dispersion forces).
6
The dispersion forces in particular can be expressed in terms of
polarizabilities of molecules. Much importance has been attached to the
evaluation of intermolecular forces to understand wide variety of phenomena
in molecular systems. In the subsequent section, some of the intermolecular
potential functions and methods for evaluating average molecular polarizability
have been discussed. Numerous spectroscopic phenomena,. like hydrogen
bonding, spectroscopy of compressed gas, have also been explained on the
basis of intermolecular forces30-31.
7
(1A.2) PRINCIPAL OF COMPUTATIONAL PHYSICS AND
QUANTUM MECHANICAL MEHTODS :
Thus, computational physics emerged in the last one decade or so due
to the growing use of computers in physical sciences12-13. In this branch of
physics starting from the study of the nature of bonds and structures to the
estimation of physico-chemical properties of molecules, including the study
of behaviour of chemical reactions is done theoretically using computational
physics methods, modeling and stimulation schemes and combinatorial physics
approaches. In fact, there are three broad classifications of methods of
computational physics.
1. Quantum Mechanics: A quantum mechanical model of electronic
structure of a molecule, which involves solving the Schrodinger equation.
Quantum mechanics can be used to predict electronic properties of molecules,
such as dipole moments, polarizabilities, spectral transitions and
spectroscopy16.
2. Semi empirical Quantum Mechanical Methods: Semi empirical
molecular orbital methods have been developed which ignore or approximate
some of the integrals used in the Schrodinger wave equation. To compensate
for neglecting the integrals, the semi empirical methods introduce parameters
8
based on molecular data. The most commonly orbital package (MOPAC)
computer programs. For example, in the computer-aided physics18 for windows
software package includes several computational physics tools including a
package of MOPAC. The MOPAC offers two of the most important semi
empirical methods (PM3 and AMI) for prediction of heats of formation, ground
state geometries and ionization potentials. Their applications to real molecular
systems have to be tested. The hydrocarbons offer very good platform to test
such theories and packages.
Other semi empirical molecular orbital method for the study of electronic
polarization and related physico-chemical properties includes the use of Dirac
delta function model of bonding. Recently developed technique like density
functional theory could also find use in predicting the properties of
hydrocarbons. Rules of molecular connectivity, and graph theoretical
approaches also find use in predicting molecular properties.
3 Molecular Mechanics: Molecular mechanics (MM) is valuable tool for
predicting geometries and heats of formation of molecules for which force
fields are available. In a molecular mechanics method, the "static energy"
calculated for the minimum energy conformation is not the energy at the bottom
of the potential well but is a good approximation for the thermally averaged
energy. This is because force field parameters obtained by fitting to
9
experimental data are thermally averaged values. Simulation and modeling,
of course, deals with the chemical engineering aspects of a reaction using
principles of mathematics and conditions of reactions, its importance is vital
in applied physics.
10
(1A.3) GENERAL THEORIES OF BOND MOMENTS AND
MOLECULAR POLARISABILITIES:
In an insulator the electrons are firmly attached to individual atoms,
molecules or ions. When such an insulator is placed in an electric field, the
Electrons are pulled slightly in one direction while the nuclei are pulled in the
opposite direction. The relative displacement of charge in matter is referred
as ‘polarization’. The differences between multipole moments of the
undistorted charge distribution are known as the induced multipole moments.
For small external fields, due to interaction with an electromagnetic radiation,
E = E0 cos (2πυt) where E0 is the amplitude and ‘t’ the time an induced dipole
moment is created. The dipole moment (p) which is induced is propotional to
the electric field, the proportionality constant ‘α’ being termed the
‘polarizability’ of the molecule. It may be represented in the form
P = α E = E0 cos2 πνt (A1.2)
The polarization of matter due to the influence of an applied electric
field is due to two sources (i) the polarizability of individual molecules and
(ii) applied electric field.
When the carbon and hydrogen bond of molecules oscillates a large charge
transfer phenomenon occurs that makes the infrared intensity complicated to
understand. Happily, several schools have contributed to our knowledge of
the nature of CH bond dipoles. In general, the dipole of a bond AB is defined
as follows:
11
If a system placed in a uniform electric field experiences a twist , measured by
turning moment , which is proportional to the field strength , then the
proportionality factor for the maximum twist in a given field, which is
characteristic of the system, is called its electric dipole moment. Such an effect
can arise if the system has poles or centers of action of its positives and negative
charges (+e and e) at fixed position separated by a vector distance, for then
when is at right angle to the field, F (see Fig. 1.1) the turning moment, is Fe.
The electric dipole moment is e and commonly written. Now the atoms in a
bond AB are allowed to perturb each other, an electron from A shifts its mean
position a distance x Å towards B, while one from B similarly shifts (1-x) Å
towards A . The net moment created will be e(l-2x) , which is zero only when
x = ½.
The order of the magnitude of electric dipole moment which may arise
in molecules may be visualized by considering the electronic charge
(4.8 x 10-10 esu), so that if an electron moves I Å i.e., a distance of the order of
magnitude of the bond distance, a moment of 4.8 x 10-10 esu is created . This
is usually called 4.8 Debye unit or 4.8 x 10-18 D (3.335640 x 10-30 c-m). Now,
when dealing with the intensities of a Vibrational spectrum of hydrocarbons,
this basic quantity, i.e., the CH bond dipole, is going to be strongly affected
and, in particular, there oscillations. Methods for infrared intensity calculation
have been published5-15 and were the subject of review by R.P. Semwal, N.K.
Sanyal19, and others20-25.
12
It is worth mentioning that the infrared spectroscopy of hydrocarbons
and modern concept on bonding based on so called molecular orbital theory
coupled with the conceptual frame of spectral intensity prediction are readily
available for the solution of the problem of CH bond moments26-30 . We have
good reason for believing that the true CH bond moment is only about 0.4 D.
The immediate questions concern the theoretical and experimental basic for
believing the magnitudes of CH bond moment, Since 1958, when this value
was presented , the CH bond dipole has been the subject of extensive studies
by infrared spectroscopists and theoreticians , and the recent past divergent
opinions have been expressed concerning the magnitude and sense of CH
bond moments in hydrocarbons.
FIG. 1.1 The dipole in a uniform field and development of a dipole bybond formation.
As a result of this, many contradictory ideas dealing with the clarification
of a common progency, i.e., its sense and magnitude, have been presented.
+++++
—————
I
13
Since the parameters associated with a CH bond dipole represent basic
quantities in setting values of bond moments in substituted aliphatic and
aromatic molecules they are of paramount importance and, therefore , efforts
are perpetually being made to clarify and settle the situation. The development
of the problem of CH bond moments had been reviewed by many experts
including a most authoritative report by Orville-Thomas and coworkers who
have tried to reconcile the situation on the basis of infrared intensity analysis,
structural parameters, and most modern concepts of dipoles and their
interactions as encountered in the framework of molecular orbital theories.
The experimental determination of effective charges on atoms in molecules
has achieved widespread attention. The main impetus in this field, besides the
applications of NQR, ESCA, and the Stark effect, has been from infrared
intensity spectroscopy. The reviewer will confine his major coverage of the
discussion to infrared spectroscopy. Because the infrared absorption intensity
depends upon the dipole moments derivatives where Qi is the normal
coordinate of the i-th vibrational mode of the molecule, in principle it should
be possible to extract information about the polarities of bonds as well as their
magnitudes. Onces under stood, the bond moment parameters may be fruitfully
utilized in the prediction on infrared spectral intensities whose basic usefulness
is more or less established.
Result can be placed in a simple and more explicit form. The diagonal
form of αmol has two distinct components α|| and α⊥ parallel and perpendicular,
dµdx( )
14
respectively, to the bond axis. Silberstein's equations for this case are the
following which may also be derived from the above.
α = (αA + αB+4 αAαa/r²) / (1 - 4αA αa/r²) (1A.3)
α = (αA + αB+2 αA αa/r²) / (1 - 4αA αa / r²) (1A.4)
These relations illustrate certain essential features of the atom dipole interaction
model :
(i) the molecule becomes anisotropic even though the atoms are isotropic;
(ii) the predicted polarizability of a macula parallel to its long axis is generally
greater than that perpendicular to the long axis ; (iii) deviations from additivity
of polarizabilities become large as the atom polarizabilities approach r²
(1A.3.1) MILLERS AND SAVCHIK METHOD
Miller and Savchik5 proposed a new empirical approach to the calculation of
the average molecular polarizability by taking it as a square of the sum of
atomic hybrid components. In this work, we have applied the method
successfully to a number of aromatic hydrocarbons and the three fullerenes.
C60, C70, and C84. The functional form of the method is represented by the
equation.
α(abc) = [ r A ]² (Å)³ (1A.5)
15
(1A.3.2) CHARACTERISTIC C-C AND C-H BOND POLARIZATION
The polarization characteristic of (C-C) and (C-H) bonds are discussed
considering variation if bond length, bond multiplicity and state of hybridization
in case of hydrocarbon molecules. Choosing electron reshuffling in the bond
region as criterion, polarizability components are computed using Lippincott-
Stutman method. Computed values are found in good agreement with the
literature values of cyclo paraffins, isomers of benzene and some hetero
compounds.
Semi empirical approaches are of considerable interest for studying
molecular structure and properties of hydrocarbons. Bond lengths in
hydrocarbons have been the subjects of theoretical and experimental studies.
While delaing with αm values of hydrocarbons by the δfunction potential
technique, the parallel component of' polarizability (αp) are found to vary
proportional to 4th power of the (C-C) and (C-H) bond lengths. Because of
this, and particularly when bond characteristic are aggregated for total
polarizability value of molecule, it becomes important to examine fluctuation
in (αp) with respect to internuclear separation (R).
The chemical concepts of the state of hybridization, C(Sp³) - C (Sp³)
etc. ionicity of (C-X) bond and multiplicity (bond order n = 1,2,3, ....) lend
support to the concept of Bartell14. Hence while dealing with the polarizability
16
components to a first approximation, Lippincott-Stutman19 introduced the inter
nuclear separations term (R) in such a way that all the concepts of chemical
binding are automaticaly assimilated within the framework of the theory.
17
(1A.4) THE DELTA FUNCTION MODEL OF CHEMICAL BINDING
AS APPLIED TO POLARIZABILITY COMPUTATIONS
The delta (δ) function method utilises the principle of additivity of bond
polarizabilities and δ-function branching condition of Frost18. Initially
Lippincott and Dayhoff20 used semi empirical δ-function technique in predicting
vibrational frequencies, anharmonicities, bond dissociation energies and
equilibrium internuclear separations of diatomic molecules. Lippincott and
Stutman21 proposed a semi empirical model based on δ-function to evaluate
polarizabilities of different components (bond parallel and bond perpendicular)
and mean molecular polarizabilities of polyatomic molecules. The model
assumes a potential which is infinite at the nucleus and zero everywhere else.
The integral of the potential over all space is finite and equal to a parameter
which is called' the δ-function strength' or 'reduced electronegativity' and is
analogous to the effective nuclear charge used by Slater. At the nucleus a δ-
function wave function is generated representing the probability amplitude of
the electron for this isolated nucleus. The δ-function atomic orbitals are then
linearly combined to form molecular orbitals with the restriction that only two
bonded atoms may interact at a time. The major advantage of this model lies
in its one dimensional nature. Every bond in a molecule is regarded as separate
entity.
The form of assumed potential function acting on the nuclei of two
18
atoms may be written as
V = - [A1g δ(x - a/2) + A2g δ(x + a/2)] (1A.6)
Where A1, and A2 are the reduced electronegativities or δ function strengths of
the nucleus 1 and 2 respectively, x, the coordinate of motion, a, the δ-function
spacing, g, the unit δ-function strength (the case of hydrogen atom) and δ(x)
the δ-function, such that δ-function, such that δ(x) = 0 when x ≠ 0 and
δ(x) = ∞ when x = 0, and ∫0
∞ δ(x)dx = 1 for any argument of x. This implies
that the potential is zero everywhere except at the δ-function positions, i.e
x = ± a/2. The δ-function strength is related to separated atom energies, Ei, by
the relation A = ( -2Ei)½ which in turn is obtainable from the first ionization
potential or from the solution of the corresponding Schrodinger equation solved
by δ-function model. The latter yields.
Ψi = N [ exp (-Ci | Xi + a/2 |) ± exp (-Ci | Xi - a/2 |)]
N = (2/Ci) [{ 1 ± exp (-Cia) ± aCi exp (-Cia)}]½
Ci = (-2Ei)½ (1A.7)
Using Frost's δ-function branching condition18 the expression of Ci, for
homonuclear molecule is
Ci = Ag [1 ± exp (-Cia)] (1A.8)
Where plus and minus signs represent repulsive and attractive state respectively.
19
The above equations also yiels a limiting value of Ci as
Lim Ci = Ag = (-2Ei)½ a= ∞ (1A.9)
a → ∞
Lippincott and Dayhoff20 generated a 'super' one electron situation from the
corresponding n-electron situation to overcome the problem of obtaining Ci
values for individual electron. Hence the resultant, C, commonly written as
CR may be obtained as
CR= A (nN)½ for homonuclear cases)
CR12 = (CR1 CR2)½ = (n1, n2 N1 N2)
¼ (A1 A2)½
for heteronuclear cases) (1A.10)
Where A1, and A2 and δ-function strengths for atoms 1 and 2 in a bond, n1 and
n2 the principle quantum numbers for atoms 1 and 2, and N1, and N2 the number
of electrons contributing to the binding or two times the column number in the
periodical table in which the ith atom falls. The δ-function strengths for singly
bonded atoms differ from those with multiple bonded atoms due to the
difference in electronic distribution in the two cases. Lippincott and Dayhoff2O
have also found an empirical relation for, A, in terms of principal quantum
numbers.
20
(1A.5) ATOMIC POLARIZABILITY
According to the variational treatment of Hylleraas3 and Hasse1, the component
of polarizability is written as
αxx = (4n A/a0) [(x1 - <x>)² - (n-1) < (x1 - <x>) (x2 - <x>)>]² (1A.11)
where x is coordinate of any one of the n equivalence class of electrons
which falls in the first equivalence class, <x> is the average coordinate of any
one of these electrons and a0 is the radius of the first Bohr orbit for atomic
hydrogen. Under the assumptions that the δ-function potentials are
symmetrically placed and all the bonding electrons are equivalent the above
equation may be written as
αxx = (4nA/a0) (<xi2>)² (1A.12)
Using the fact that the polarizability of an atom is isotropic, i.e. <x²> =
<y²> = <z²> = (1/3) <r²>, and for a δ-function located at nucleus one may
have Ψ =Ne-A r (where N is the normalization factor = A3/2 n-½), one gets
<r²>= ∫0π
∫0 2π
∫0∞ Ψ r² Ψ∗ .r² sinθ dθ dφ dr = 3/A2 (1A.13)
so that <x²> = 1/A². Finally, the polarizability of the atom along X-axis
is given by
αxx = 4
a0³ A³ (1A.14)
21
(1A.5.1) BOND POLARIZABILITY
The mean polarizability, αM, for a diatomic molecule with an axis of
symmetry can be written as made up of bond parallel component, αII, and
bond perpendicular component, α⊥ of polarizability, i.e.
αM = (1/3) [αII + 2α⊥ ]. (1A. 15)
The component arises due to the contributions from (a) bond region
electrons, and (b) non bonded region electrons. The contribution of bond region
electrons is obtained by using a linear combination of atomi δ-function wave
functions representing the pair of nuclei involved in the bonding and is given
by
αIIb = 4n A12 (1/a0) ( <x²>)² (1A.16)
where n is the bond order, A12 the root mean square δ-function strength
of the two nuclei, and
<x²>=R²/4+1/2C2R (1A.17)
R being the equilibrium internucler distance. In case of heteronuclear
bond, the difference in electronegativities of atoms introduces charge density
which must be accounted for by applying a polarity correction to αIIb given by
σ = exp [(-1/4) (X1 - X2) ]² (1A.18)
22
Where X1, and X2 are the electronegativities of atoms 1 and 2 on
Pauling's scale32 Thus the corrected value of the parallel component of bond
polarizability is given
αIIp = ( αIIb) σ (1A.19)
The contribution of non-bonded electrons, i.e. the electrons in valence
shell not taking part in the bonding, is calculated by using Lewis-
Langmuir33-34a Octet rule with a modification by Linnett34b in terms of double
quartet of eletrons. The general expression may be written as
ΣαIIn = Σ fj αj (1A.20)
Where fj is the fraction of the electrons not involved in bonding for the
jth atom with atomic polarizability αj.
The a⊥ component i.e. the perpendicular component of bond
polarizability as obtained by Lippincott and Stutman21takes into account the
residual atomic polarizability degrees of freedom, which for an isolated atom
is three. In case of diatomic molecule, this component is simply the sum of the
two atomic polarizabilities. But if one of the atoms is less electronegative than
the other, the average magnitude of components perpendicular to the bond
will be function of the charge separation and atomic contributions will be
considered according to the Square of the respective electronegativities.
23
Extending the principle to polyatomic molecules, the analytical expression
for all the perpendicular components is given by :
ΣXj² αj
Σ2α⊥ = ndf (1A.21)ΣXj²
where ndf is a symmetry and geometry dependent quantity and is
denoted by a factor (3N-2b), gives the idea of residual atomic polarizability
degrees of freedom. The determination of ndf is based upon some basic
assumptions. Every bond which is formed between two atoms removes two
polarizability degrees of freedom with the restriction that - (i) if two bonds
are formed from the same atom and exist in a linear configuration, three atomic
polarizability degrees of freedom are lost, and (ii) if three bonds are formed
from the same atom and exist in a plane, then five atomic polarizability degrees
of freedom are lost.
24
(1A.6) CONTRIBUTION DUE TO NONBOND-REGION ELECTRONS
TO PARALLEL COMPONENT
In the valence shell of each atom the remaining electrons also contribute
to the bond parallel component, or αII. But there must be a residual polarizability
along the internuclear axis, and this additional component is expected to be
independent of bond length. A general expression to obtain this contribution
can be written as
Σ α11n = Σ fj αj (1A.22)
where fj is the fraction of the electrons of the jth atom which are not in
the bond region and αj is the atomic polarizability of the jth atom.
(1A.6.1) PERPENDICULAR COMPONENT
In the Lippincott and Stutman method the perpendicular component is
obtained by considering the residual atomic polarizability degrees of freedom
which for an isolated atom is three. Thus α⊥ = 2αA for a nonpolar diatomic
molecule and α⊥ = αA + αB for a polar AB molecule.
Let us consider two atoms A and B, A is less electronegative than B, the
average magnitude of components perpendicular to the bond will be a function
of the charge separation and atomic contributions will be considered according
to the square of the respective electro negativites. Hence, the analytical
expression in this case will be
25
2 (χ²A αA + χ²B αB)α⊥ = (1A.23)
χ²A + χ²B
The above equation is modified in the case of polyatomic molecules,
which is written as follows :
Σ2α⊥ = ndf Σxj² αj
Σxj² (1A.24)
where ndf is calculated by 3N-2nb when N is the total number of atoms
and nb is the total number of bonds in that atom.
(1A.6.2) MEAN POLARIZABILITY (αM)
The mean polarizability or average molecular polarizability is the sum
of the parallel component, the perpendicular components of bond
polarizabilities and the contribution of nonbonding electrons. So the αM is
expressed as :
αM = 1 ΣαII pi σ + Σfiαj + ndf
Σxj²αj (1A.25)
3 Σxj²
where
αIIp = 4n Ao R² + 1 ² e -(χ
A-χ
B)
a0 4 2CR²12 (1A.26)
In the above equation A12 is the δ-function strength for a bond, a0 is the
[ ]
[ ]
26
Bohr radius, the covalent bond character as defined by Pauling and n is the
bond order. The atomic polarizabilites α (in 10-25 cm³) is taken from the work
of Lippincott et al.
(1A.6.3) AVERAGE MOLECULAR POLARIZABILITY :
The average molecular polarizability, αM is independent of bond angles
and is obtained from contributions of (i) the parallel components due to bonding
as well as non-bonding electrons, and (ii) the perpendicular components of
bond polarizability. Thus αM is expressed as
αM = 1 Σi αIIb σ + Σj fj αj + ndf
Σj Xj² αj (1A.27)
3 j αj²
Where the summation refers to i bonds and j atoms.
= 8 x 1.1964242
= 9.5713936
A12 = √A1 A2
= √.846 x 1.00
= .9197826
<x>² = R²
+ a²0
4 2CR²12
= (1.11)²
+ .279841x10-16
4 9.5713936
][
27
= .331737 x 10-16
αIIb = 4n A12 [<x>²]²
a0
= 4 x 1 x .9197826 (.331737 x 10-16)²
.529 x 10-8
αIIb = 7.65 x 10-25
For C-H bond
σ = e-¼ (x1 - x2)²
= e-¼ (2.5-2.1)²
= e-0.04
= -0.4 x .4342942
= .07371768
= 1.982628
σ = .9607916
αIIp = αIIb σ
= 7.653814 x .9607916 x 10-25
= 7.353720747x10-25
to the case of Methane, their are four C-H bonds,
= 4 x7.353720747 x 10-25
= 29.41488 x 10-25
ΣαIIp = 29.41488 x 10-25
Σfj αj = 0
28
Σ2α⊥ = ndf Σxj² αj
Σxj²
= 7 (2.5)² x 9.78 + (2.1)² x 5.92 x 4
(2.5)² x 1 + (2.1)² x 4
= 7 x 6.929836
= 48.50885
αM = 1/3 [ΣαIIp + Σfj αj + Σ2α⊥ ]
= 1/3 [29.41488 x 10-25 + 0 + 48.50885]
= 1/3 [77.92374]
αM = 25.97458x10-25
][
29
PART (B)
ATOMIC HYBRIDIZATION AND MOLECULAR POLARIZABILITY
(1B.1) INTRODUCTION
The most successful empirical approaches to the calculation of molecular
polarizabilites include the environment through atomic hybridization about
the central atom or with the atoms forming a bond or group. For this purpose
the concept of atomic hybrid component (ahc) and atomic hybrid polarizability
(ahp) was introduced. The simplicity of the ahc and ahp atomic hybrid method
is understood by nothing that the molecular polarizability cannot be written as
a sum of atomic polarizabilites, when atoms are defined only by the atomic
number.
It is well known that the atomic polarizability cannot be added simply
to get molecular polarizability. For this, each atom must be assigned a
polarizability depending on the atoms to which they are bonded. Eisenlohr35
and Vogel36 have set up an extensive system of atomic refractions which have
been supplemented by Batsanov37 in which, for example, nitrogen in primary,
secondary and tertiary amines is assigned different values. Silberstein38 pointed
out that the ever growing hierarchy of rules indicating how to treat the
exceptions to the law of additivity, although helpful to the physicist, is the
dearest confession of non additivity, unless the atomic environment is
considered in great detail. The methods that consider atoms in atomic hybrid
states eliminate the need to develop an extensive set of rules that consider the
specific atoms involved in bonding.
30
(1B.2) ATOMIC HYBRID COMPONENT
To circumvent this problem a new empirical approach to the calculation
of the average molecular polarizability as a square of a sum of atomic hybrid
components is presented by Miller &Savchick39-40. It requires fewer parameters
than other methods. and has the functional form
α(ahc) = 4/N [ΣA τA]² (Å)³ (1B.28)
Where N is the number of electrons in the molecule, and ahc refers to
the atomic hybrid components τA of α for each atom in a particular hybrid
configuration. The summation proceeds over all atoms A in the molecule. In
the present formulation τA does not depend on atoms which are bonded to A,
but it does depend on the type of bonding through the hybridization of atomic
orbitals on A.
Semwal, Uniyal and Balodi26-28 have applied this method successfully
to a number of aromatic hydrocarbons and three fullerenes C60, C70, C84. We
have applied this method to determine the polarizability of some organometallic
compounds which have recently attracted the interest of those who search for
novel and efficient non-linear optical materials41-45.
The concept of bond polarizability was introduced here and Steiger
and Smyth46(a,b) obtained bond polarizabilities directly from the atomic
31
refractions47. In contrast, Denbig48 developed a system in which a unique bond
polarizability is obtained directly for each kind of chemical bond from molar
refractions R, for example R(C-H) is 1/4R (CH4) and for the alkanes
(n-1) R (C-C) + (2n + 2) R (C-H) = R (Cn H2n+2)
from which R(C-C) follows.
The empirical formula eq. (1B.28) is presented as an additional method
for the calculation of average molecular polarizabilities. It is easy to apply.
Fewer parameters are needed in this approach than in other methods because
only hybrid components are considered and not the specific atoms to which a
given atom is bonded. For each atom, A, one valuer τA is used for each type of
hybridization. The number of and type of bonds as well as lone pairs are
automatically incorportated into a parameter τA by the hybrid state of each
atom.
32
(IB.3) RATIONALIZATION OF THE EMPIRICAL APPROACH
The functional form of the empirical formula proposed in this
investigation can be rationalized, with the variational-perturbation approach
proposed by Hylleraas3 and Hasse1 and approximated by Hirschfelder, Curtiss,
and Bird29 for the calculation of the x component of molecular polarizability.
αXX = 4N/a0 [(x1-x)² - (N-1) (x1- x) (x2-x)]² (1B.29)
Where N is the number of electrons, x is the average position of an
electron in the x direction, (x1-x)² is the mean square deviation of an electron
from its average position, and (x1-x) (x2-x) is the average correlation between
two electrons in the x direction. The average value of the operator q, is
q = ∫Ψ∗qΨ dτ (1B.30)
it is calculated with the zeroth order wave function Ψ obtained from
perturbation theory. If the term in brackets in eq.(lB.29) is rewritten to include
a summation over all electrons i, j = 1, 2, 3, ...................,V, then
αxx = 4/Na0 [Σij (xi-xi) (xj-xj)]² (IB.31)
= 4/Na0 [ Lxx ]² (IB.32)
33
The term in brackets in eqs 1B.31 and 1B.32 reduces to
Lxx = Σij xi xj - (Σi xi)² (1B.33)
For closed shell systems the zeroth order wave function may be approximated
by an antisymmetrized product of molecular orbitals Ψµ, µ= 1,2,3, ................
and spin function η and η with z components equal to ± ½
Ψ = (N!)-½ A {Ψ1 (1) η (1) Ψ1 (2) η (2)............} (1B.34)
Substituting eq. (IB.34) in eq. (IB.33) yields
Lxx = - 2 ΣµΣν∫Ψ∗
x Ψν dτ ∫Ψ∗µ x Ψµ dτ + 2Σµ ∫Ψ∗
µ χ² Ψµ dτ (1B.35)
where µ and ν refer to the occupied molecular orbitals. The molecular orbitals
are expanded as a linear combination of atomic orbitals χAt
Ψµ = ΣA Σt CAtµ χAt (1B.36)
where t may refer to S, px, py and pz atomic orbitals or to hybrid atomic orbitals
on atoms A= 1,2,3----substituting eq IB.9 into eq. IB.8 and using the zero
differenntial overlap approximation yields.
Lxx = ΣA √a0 τAxx (1B.37)
34
Where √a0 τAxx =-2 ΣBΣtΣuΣvΣwΣµΣν C*Atµ CAuv C* Bvν CBwµ x ∫χ∗Atx χAu dτ
∫χ* Bv X χBw dτ + 2 ΣtΣuΣµ C*Atµ CAuµ x ∫χ*At X² χAu dτ (1B.38)
provides the link between the proposed empirical, formula, eq. 1B.28 and a
molecular orbital model.'Atomic orbitals t, u, v and w are centered on atoms
A and B in At, Au, Bv and Bw
It is convenient to assume that Ψµ, µ = 1,2,3.. are the localized molecular
orbitals studied extensively by England et al49a,b-50. Their orbitals, confined to
the region of a chemical bond, are written as a sum of hybrid atomic or π
orbitals. For acylic hydrocarbons two centre localized molecular orbitals are
found, whereas for condensed hyrbocarbons two to four center localized
molecular orbitals connected through a set of adjacent atoms are obtained.
Now, the contributions to τAxx consist of two types of terms. The first eq
(I B.38) connects atomic orbitals on a pair of atoms A and B and the second
involved only atomic orbitals on each atoms A. The first term may be viewed
as a bond contribution because the coefficients C*Atµ CBwµ and CAuν C*Bvν
connected through common localized molecular orbitals Ψµ, and Ψν are largest
if A and B are adjacent atoms. If these bond contributions, which arise for
each hybrid atomic orbital in Ψµ on atom. A connecting adjacent atoms, are
partitioned between the two atoms A and B, then one has a rationale for the
35
assignment of a unique value of τAux for each hybrid atomic state. If eq.
(1B. 38) is written as
√a0τAxx = -½ ΣBΣtΣuΣvΣw PAtBw P*AuBv ∫χ∗ AtX χAu dτ
x ∫χ∗Bv x χBw dτ + ΣtqAt ∫χ∗ AtX² χAt dτ (1B.39)
where
PAtBw = 2 Σµ C* Atµ C BWµ (1B.40)
is the bond order for doubly occupied orbitals qAt = PAt Av* which arises
because symmetry requires that t = u in the second term of eq (1B.38), is the
electronic charge, then the bond and atomic terms are more obvious. Bond
orders are small unless A and B are adjacent atoms.
This approximation ought to work well for covalently bonded systems
in which the electron pairs are nearly equally shared by two atoms. For very
polar systems, some modification of τAχχ will be required. However, in this
investigation, eq (1B.28) reproduces experimental values of average molecular
polarizabilities of organic compounds quite well.
To complete the rationalization of eq (1B.28) the component of
polarizability αXX obtained with eq (1B.29), and similarly αyy and αzz are
assumed to be calculated with the xyz coordinate system oriented along the
36
principal axis of the polarization epsoid. If the contribution of τAalong each
principal axis is assumed to be given by
τAkk.= √3 cosγkk τA (K = x,y, or = z) (1B.41)
where
Σkcos² γkk = 1 (1B.42)
then the average molecular polarizability
α = 1/3 Σkαkk (1B.43)
is obtained from
αkk = 4/N [ΣkτAkk ] ² (1B.44)
by combining eq (1B.32) with eq. (1B.37) and (1B.41)for K = x and
similarly for
K = y and z. The component τkkA are projections of √3τA onto the principal
axis K = x, y and z, or from the components of the polarizability
√αkk = cos γkk √3α (1B.45)
is the projection of 3α onto the principal axis.
In the case of condensed hydrocarbons, the localized molecular orbitals
are delocalized over more than two atoms. The contribution to each atomic
hybrid component is still obtained by partitioning the effect of terms involving
A and B over each atom. The assumption is made that there are two values τA
for the trtrtr π hyrbid states of carbon τc = 1.428 for one with a C-H bond and
37
τc = 1.800 for one in branched configuration as in graphite. This is the only
exception made in the formulation of the average molecular polarizability with
parameters τA for atoms, A, in their conventional states of hybridization and it
is required to achieve agreement between α(ahc) and α(exp).
An alternate approach to that proposed in this investigation, in which,
A represents atomic hybrid contributions, would be assumption over bond
contributions with a formula like eq 1B.28. In eq 1B.38 or 1B.39, terms
involving orbitals on atoms would be partitioned onto pairs of atoms connected
by a bond or several atoms encompassed by the localized molecular orbital in
the case of condensed hydrocarbons, and τ would be accordingly redefined.
The terms connecting pairs of atoms A and B are already cast into a form
which could be viewed as bond contribution. This approach was not explored.
It would require a set of parameters τA, for every pair of atoms with each
particular bonding type, and consequently there would be greater number of
parameters than required in the present empirical approach.
The Van der Waals radii can be correlated to the atomic polarizability
αA = 4/NA (τA)² (1B.46)
calculated with eq (1B.28) for each atomic hybrid contribution τA. Combining
eq (1B.46) with the Slater Kirkwood approximation50.
38
4 rA² ² (1B.47)αA =
a0 3
yields
√r²A = 3 (a0αA)¼ (1B.48) 2√
][
39
PART (C)
HUCKEL MOLECULAR ORBITAL METHOD
(1C.1) INTRODUCTION :
Several semi-empirical theories have emerged which are capable of
predicting the properties of molecules through mathematical equations of wave
mechanism. The most important of these theories belong to a class known as
the molecular orbital theory which is based on a set of approximation. The
simplest of these molecular methods is one developed by Huckel1 and known
as Huckel molecular orbital (HMO) method. Though this method is very
primitive and the crudest of all the molecular orbital methods, it has been
extensively and successfully used in several descriptions52-64. It has also been
argued that for a comparative study on π-electron systems, this method is as
good as any refined method65-70 and results obtained by it are if not exact are
very close to observed values.
The major advantage of the HMO method is that it is conceptually simple
and mathematically precise in the sense that it rests quite simply on Huckel’s
original assumption51 that to a first approximation only the π-electrons need
be considered in calculating the properties of such molecules as olefins,
polyenes, aromatic hydrocabons, and their derivatives including the compounds
containing cyclopentadienyl rings like chromocenes. In addition to it in this
method the diagonalization of secular determinants is relatively easier than
40
any other process involving iterative procedure, because the secular equations
has to be solved once.
Many attempts have been made71 to allow for the effects of σ-electrons
in some way or the other but all of these are not so successful, because the
inclusion of σ-electrons in MO treatment poses many problems. First, there is
the sheer magnitude of the calculations, given that the total number of valence
electrons in a molecule is very much greater than that of π-electrons in a
conjugated system of comparable size; benzene, for example contains only
six π-electrons, out of a total of thirty valence electrons, secondly, the MOs no
longer have the symmetry of those in π-system being three dimensional.
Thirdly, there is the problem of invariance in choice of coordinate axis and
finally there is the problem of predicting molecular geometry71.
In addition to these factors out of valence electrons containing σ and π
both, the σ-electrons are strongly localized in individual bonds and are
represented by bond orbitals of σ-symmetry. However, the π-electrons are
highly delocalized over carbon skeleton and chemically reactive. That is why
the π-electrons are treated in an explicit fashion and σ-electrons are regarded
as providing some sort of potential field for the π-electrons. Here every system
is divided into σ and π parts, later on we will treat all β as equal and all σ as
constant.
In the Huckel derivation, the individual MOs of a molecule are
41
eigenfunctions of the corresponding one electron Hartree-Fock operator H;
HΨµ = ∈µΨµ (1C.49)
where H is Hamiltonian operator, ∈µ is the energy of the µth state and
Ψµ is wave function of the µth molecular orbital. Instead of solving we
approximate the molecular orbitals by using the variation method with an
LCAO trial eigenfunction;
Ψµ = Σi aµi φi (1C.50)
where aµi is coefficient of the µth state for ith atom and φi is atomic
orbital for the ith atom. Assuming neglect of differential overlap, the
corresponding variation equations are
|Hij - ∈δij| = 0 (1C.51)
Σj aµj (Hij - ∈µδij) = 0, i = 1, 2, 3, --------N (1C.52)
where Hij = ∫φiHφjdτ (1C.53)
Instead of calculating the intergrals Hij, we treat them as parameters.
The diagonal element Hij is called a Coulomb integral and is supposed to have
a value αi characteristic of the AO φi and the atom of which is an AO, and
independent of the rest of the molecule, while the off-diagonal element Hij,
written as βij and called a resonance integral, is likewise assumed to have a
42
value characteristic only of the AOs φi, φj. It is usually assumed that
βij = 0 Unless φi, φj are AOs of two atoms (1C.54)
that are linked by covalent bond
on the basis of various intuitive arguments, it is also assumed that if φi
is an AO of an atom p, then
αi = Ionization potential of an electron occupying
the AO φi of an isolated atom p [ (≡W(φi)] (1C.55)
The ground state of a closed-shell molecule is supposed to correspond
to a situation where the 2N electrons are placed in pairs in the N MOs of
lowest energy; the total binding energy ET of the molecule is then equated to
the total orbital energy of the electrons, i.e.
ET = 2 Σµocc ∈µ (1C.56)
The energy Ep of an atom p is likewise equated to the total orbital energy
of its electrons; i.e.
Ep = Σip ni α (φi) (1C.57)
where the sum is over AOs φi of atom p, and where ni is the number of
electrons occupying the AOφi. If one is interested in the comparison of ground
43
state properties of a series of structurally related molecules, the MO theory is
the most suitable approach, however, a more convenient route is provided by
the Valence Bond (VB) method, which owes its origin to the pioneering work
of Heitler and London72 on the hydrogen molecule. The VB method was
extended in a more general fashion by Pauling73 and Slater74. A very
comprehensive and systematic presentation of the basic ideas and applications
of the VB method was made by Pauling75 in 1960.
In contrast to the MO theory which is based on the concept of electronic
configurations, VB method is directly linked with the common chemical
concepts like bonds, lone pairs, Lewis structures, resonance etc. In the MO
theory one works with the Slater determinants constructed from an orthonormal
set of MOs, while in the VB theory the building units for such determinants
are the AOs, which are not mutualy orthogonal. These determinants form a set
of non-orthogonal many electron functions corresponding to different pairing
schemes of electron spins. Starting from a given set of AOs the MO and VB
theories will lead to identical results provided all possible electronic
configurations are taken into account in the former and all possible pairing
schemes are considered in the later. Such a complete treatment being not feasible
for large molecules, one is forced to work with approximate wavefunctions,
such as the single-determinant wavefunction in the MO theory. There is no
such simple prescription for the approximate wave function in the VB theory.
44
Thus it is the mathematical simplicity for which the MO theory has become so
attractive to the theoretical physicists. However, the MOs are many-centre
functions and they do not represent as such the valence structures with which
physicist are so familiar with. Even a staunch protagonist of the MO theory
cannot undermine the utility of structural formulas in physics .
Wheland76 has shown that the inclusion of overlap between nearest
neighbour atoms in the HMO approximation does not lead to serious
mathematical complications provided all the overlap integrals are considered
equal. Maslen and Coulson77 have extended the Coulson and Longuet-Higgins
perturbation theory to Wheland’s modification. The secular determinant in
the Wheland modification now has the matrix elements (for alternant
hydrocarbons) α−∈ and β−∈∆ for diagonal elements and nearest neighbour
off-diagonal elements, respectively. One can define a quantity x by
x = α−∈ (1C.58)
β−∈∆
so that the Wheland secular determinant has the same form as in the
HMO approximation with the overlap omitted. The MO energies are thus given
by
45
∈j = α-xjβ
= α - xj (β−∆α)
(1C.59)
1-xj∆ 1-xj∆
If one defines the quantities yi and β’ by
yi = xj
β’ = β-∆α(1C.60)
1- xj∆
the MO energies have the form
∈j = α- yjβ’ (1C.61)
this is the same form as in the simple HMO approximation except that
the resonance parameter β is replaced with a new empirical parameter β’ to
obtain the MO energy in the Wheland modification one simply uses the roots
of the simple HMO approximation to compute the yj’s. The overlap integrals
can be calculated theoretically using Slater-type 2pπ AOs of carbon78. For
linear molecules one usually uses ∆ = 0.27 and for cyclic molecules ∆ = 0.25.
In general, one finds that the inclusion of overlap tends to crowd bonding
levels closer together and to spread antibonding levels farther apart. Also the
difference between the highest occupied MO and the lowest unoccupied MO
is increased (assuming β and β’ have the same value). This has the effect of
giving somewhat better agreement with spectroscopic data. Nevertheless, the
failure to elucidate the electronic interactions means that singlet and triplet
46
states are not differentiated, just as in the original HMO approximation.
Inclusion of overlap removes the pairing properties of the MO energies and
the MO coefficients. The new coefficients are given in terms of the simple
HMO coefficients by
brj = crj (1C.62)
(1-∆xj)½
Chirgwin and Coulson79. have shown that charge densities and bond
orders must be redefined in the Wheland modification. The proper definitions
are
qr = Σj n j (b²rj + Σrbrj br’j ∆ rr’) (1C.63)
prs = Σj nj [brj bsj + ½ (Σs’ ∆ss’ brjbs’j + Σr’ ∆rr’ bsj br’j)] (1C.64)
where r’ represents atoms adjacent to r (with the exception of s in the
case of prs) and s’ represents atoms adjacent to s (with the exception of r). It is
found that the above formulas lead to precisely the same charge densities and
bond orders as for the HMO approximation without overlap. Nevertheless,
the two methods do not generally lead to the same results. Ruedenberg80 has
carried out an extensive study of the Wheland treatment (which he calls the
tight-binding approximation) to show that this is a bonafide first approximation
to an accurate quantum-mechanical treatment, whereas the HMO
approximation in which nearest-neighbour overlap is not neglected. This
47
conclusion does not appear to be necessarily valid if the basis set used in the
simple HMO approximation can be justifiably, assumed to consist of
orthogonalized AOs. Amos and Crispin81 have also redefined polarizabilities
as per above modification.
Hoffman82 has investigated an interesting extension of the HMO
approximation in which the total electronic wave function is separated, not
into a sigma-elecron portion and a pi-electron portion, but rather into an inner-
shell nonbonding portion and a valence-shell portion. For a hydrocarbon of th
general formula CnHm and a total of 6n+m electrons, the total wave function
is written as
Ψ (1,2, ------, 6n+m) = a ΨN (1,2, -------, 2N) x ΨB (2n + 1, 2n + 2, --------,
6n + m) (1C.65)
where ΨN and ΨB are separately antisymmetrized and well behaved.
The function ΨN describes 2n electrons localized in ls AOs of the carbon atoms,
and ΨB describes the remaining 4n+m electrons, i.e., the valence electrons,
the function ΨB now plays the same role in the theory as Ψπ does in the σ−π
separation approximation. Each MO used to construct ΨB is given by the LCAO
form
Ψi = Σp φp cpi (1C.66)
48
The basis set used by Hoffmann consisted of STOs; m ls AOs of
hydrogen, n 2s AOs of carbon, and 3n 2p AOs of carbon, this is of course, the
minimal basis set. The effective Hamiltonian of the system then is
KB = Σµhµcore + Σµ<ν
1 (1C.67)
rµν
which sums over the electrons in the valence shell. The inner shell
nonbonding electrons now play the same role as sigma electrons in the HMO
method. As in the HMO method, one can define a suitable Hartree-Fock
operator and solve the corresponding Roothan equations. As in the HMO
approximation, the matrix elements of the Hartree-Fock operator are guessed
without elucidating the core hamiltonian, but now all the Fpq and ∆pq elements
are given explicit nonzero values, i.e. one does not invoke the nearest-neighbour
approximation for either Fpq or ∆pq. Hoffman choose the diagonal elements of
the Hartree-Fock operator equal to the negatives of the valence state Ionization
potentials associated with the relevant basis AO. For calculations on molecules
containing sp³ carbon atoms, Hoffmann used the values given by
-11.4 eV for 2pc AOs
Fpp = -21.4 eV for 2Sc AOs (1C.68)
-13.6 eV for lSH AOs
Hinze and Jaffee83 have also calculated valence-state energies which
should be useful in this context. Hoffman has appoximated the off diagonal
49
matrix elements by the relationship
Fpq = 0.5K (Epp + Fqq) ∆pq (1C.69)
As justified by Hoffman, the constant K must be chosen so as to satisfy
l<K<α (where α can be regarded as an average value of Fpp) in order to avoid
certain trivial or absurd results, Hoffman chose K=1.75. The overlap integrals
are calculated analytically. Since these depend upon the molecular atomic
coordinates, one can carry out the calculations as a function of the molecular
geometry. Since the treatment includes all the sigma electrons, which play a
major role in the geometry of the molecule, one might expect that the
approximation would predict molecular geometries quite well.
The total electronic energy can be written in the same form as for the
HMO approximation, namely,
EB = Σi ni ∈i + G (1C.70)
where G is the interelectronic term. The molecular energy is given by
E(R) = EB + Σk < 1
ZK Z1 (1C.71)
rkl
= Σini ∈i + Σk<l
ZkZl + G
(1C.72)
rkl
where R represents the collection of internuclear coordinates. Slater84
50
has observed that the sum of the nuclear-repulsion energies and the
interelectronic term is small and is roughly constant as the nuclear coordinates
change. This implies that the sum of the orbital energies behaves approximately
as the molecular energy does. Thus it should be possible to predict qualitatively
acceptable molecular geometries without elucidating the inter-electronic terms.
Failure to elucidate this term, however, will lead to serious quantitative and
qualitative difficulties in attempts to carry out spectroscopic correlations, just
as in the pi-electron HMO approximation.
Hoffman’s method will have to be investigated more thoroughly before
its advantages and disadvantages are fully revealed. The approach is certainly
a useful extension of the simple HMO approximation. It is applicable to
saturated as well as unsaturated hydrocarbons. In fact, the method should be
applicable to virtually any type of molecule. The calculations of this kind only
recently been attempted in a serious manner, and the best of these still fall
short of the success achieved in the pi-calculations. Nevertheless the results
already obtained have been encouraging enough to virtually ensure ultimate
success, and progress in this field is likely to be rapid in near future.
An MO method similar to that of Hoffman has been discussed by Pople
and Santry85 and applied to saturated hydrocarbons. Hartman86 has discussed
a modification of the HMO approximation which is not so far-reaching as that
of Hoffman. It is quite similar to ordinary approximation except that the
51
pi MOs are approximated by an extended basis set, that is, 2p and 3p AOs.
The latest extension of the HMO theory is by Cyvin87-91. Using two-
dimensional approach to HMO theory he has extended the use of the approach
to sp-hybridized carbon atoms in sigma-system in addition to the sp² hybridized
ones. Hence the extended theory covers systems with acetylenic C≡C bonds
and cumulated C=C bonds, in addition to conjugated C=C bonds and aromatics.
In the case of heteromolecules all the atoms and bonds are not same
and the difficulty is in choosing the values for coulomb and resonance
parameter. This is a difficult problem of pi-electron theory and the prospects
of solution appear none too bright41. We know that the coulombic parameter
of the xth atom is given by
αx = α + hxβ (1C.73)
where hx is electronegativity parameter and α, β are the reference
integrals in terms of which any coulombic integral can be expressed. One
might then hope that hx could be related to the ionization potentials of the
reference atom of the heteroatom42. Specifically one might hope to express the
electronegativity parameter as
hx = Ix - I
β (1C.74)
52
Where I is the ionization potential of the reference atom and Ix is the
ionization potential of the xth atom. The problem in using is that one must
choose valence state ionization potentials (not always easy to estimate) and
must assume an explicit value for β. Mulliken92 and Laforgue93 have suggested
a more effective and easy to use relationship
hx = χx - χc (1C.75)
where χx and χc is the electronegativity of the xth and carbon atom
respectively. The obvious disadvantage of using is that it implies that hx is the
same for a given atom in any molecule containing the same reference atom.
For example the value of hN would be used in pyridine as in pyrrole. One
should evaluate hx in terms of atom surrounding it in a given molecule, but we
have used this relationship in chapter 5. What appears to be a very sensible
way to evalutate the electronegativity parameter is through the method of
Streitwieser and Nair94 given by equation
hx(n) = (1-qx
(n-1)) ω (1C.76)
where qx(n-1) is charge density of the xth atom resulting from a calculation
based on a previously assumed value of hr. The new value of hx is then used to
recalculate qx and thereby to lead to yet another value of hx. The iterative
procedure is carried out until consecutive values of hx agree within previously
53
chosen limits. The quantity ω is itself an empirical parameter and is usually
chosen to have the value 1.4. This method has been reformulated by Doggett95,
by avoiding some of the convergence difficulties commonly encountered in
the original method. The resonance parameter is represented by the relationship
βcx = kcxβ (1C.77)
where kcx is called the bond parameter. We have calculated this parameter
by the following Lennard-Jones96 relationship
kcx = βcc
= E(c=x) - E(c-x) (1C.78)
βcx E(c = c) - E (c-c)
where the E’s are empirical bond energies for the bonds indicated and
obtained from the literature such as those given by Pauling75.
Another widely used relationship for kcx is
∆ cx (1C.79) kcx =
∆ cc
which was proposed by Wheland76. This later relationship assumes a
direct proportionality between the resonance integral and the corresponding
overlap integral. Ruedenberg has suggested that a more correct relationship
is80
54
kcx = ∆cx (1 - ∆cx) (1C.80)
∆cc (1-∆cc)
The difficulty with either (1C.79) or (1C.80) is that one assumes
∆cx = δrs in the HMO approximation. If the basis AOs are carbon 2pπ AOs then
∆cx ≠ 0, and the above relationships are satisfactory. However, it is then
inconsistent to neglect overlap for nearest neighbours. On the other hand, if
the basis AOs are orthogonalized, so that overlap formally disappears, the
above relationships lead to trivial reuslts.
One could also choose resonance parameters by an iterative technique
having some resemblence to the Streitwieser’s ω-technique for coulombic
parameters. Following Longuet-Higgins and Salem97, one could write
kcx = e-a(dcx -- d) (1C.81)
where dcx is the length of the cx bond, d is the length of a standard C-C
bond (the one in whcih β is based), and a is an empirical constant (usually
a-1 = 0.3106A). An interative scheme for obtaining self-consistent values of
the bond parameters could be set up by relating dcx to the bond order pcx.
Longuet-Higgins and Salem used the empirical relationship
dcx = 1.50 - 0.15pcx (1C.82)
which was based on alternate hydrocarbon calculations.
55
The simplifying assumption is usally made that all the atoms in a
heteromolecule are equivalent except for the heteroatom (s) and that all bonds
not involving heteroatoms are also equivalent. The atoms and bonds not
involving the heteroatom (s) then define the reference atom and the reference
bond. This assumption has the advantage of greatly reducing the number of
parameters which must be chosen empirically. It is rather pointless to argue
the justifiability of such an assumption on theoretical grounds, since the HMO
approximation is so overwhelmingly empirical. The simpler one can keep an
empirical scheme and still have it work the better.
Recently Gupta98-99 has tried to rationalize the parameters for heteroatoms
in HMO theory by the help of the concept of molecular connectivity given by
Kier and Hall61. He has determined various physical properties like ionization
potential, electron affinity, bond order and bond length of some heteroaromatic
systems by the help of classical values of parameters obtained empirically and
by using the molecular connectivity indices, on comparing both the results
with the available experimental values he concluded that there is better
parameterization by use of molecular connectivity than by the empirical
method. In chapter 5, we have also done the same kind of study for one of the
parameter i.e. h by taking different kind of systems and more in number, we
found that h(δ) and k lead better agreement with the observed values than
those by use of empirical h and k. Even after this investigation no final
56
conclusion can be drawn at this stage about which method gives better values
of both the parametes, till the extensive study on the various kind of systems is
done, the study to be carried out in future.
The vast literature of the HMO approximation is replete with examples
of different ways of choosing the electronegativity and bond parameters for
heteromolecules. There are those who argue for the use of auxiliary
electronegativity and bond parameteres i.e. the heteroatom is regarded as
polarizing other atoms and bonds in proportion to their proximities to the
heteroatom. For example, in pyridine one could use the additional parameter
h2=h6 to describe an inductive effect of the nitrogen atom on its carbon atom
neighbours. Similarly one could use the parameter k23 = k56 to describe a similar
effect on the bonds next to the C-N bonds. Unfortunately, the more empirical
parameters one is forced to choose, the harder it becomes to attach even a
remote physical significance to the choices, and the less meaningful the results
become. It has been found that the SCF solutions of Roothan’s equations for
heteromolecules are capable of shedding some light on the HMO parameter
problem. Lykos100 has discussed in greater detail the choice of parameters for
the generalized HMO method and Ohno101 has carried out a critical study of
this method and has showed that the usual methods of evaluating the one-
centre coulombic integrals and atomic core energies are quite good. Ohno has
also investigated semiempirical (as opposed to empirical) methods for
estimating core resonance integrals. Such a method would be valuable for
choosing core resonance integrals involving heteroatoms.
57
PART (D)
THE MOLECULAR CONNECTIVITY
(1D.1) INTRODUCTION
Several attempts have been made at a level below the quantum
mechanical to provide quantitative structure descriptions. Some of these are
classified as topological methods. Most of these approaches attempt to assign
numbers to parts of molecule and to sum these numbers in accordance with
molecular topology to yield indices. The values assigned to molecular fragments
may be calibrated against a physical property or even a biological activity,
these include methods due to Wiener102, Smolenski103, Franklin104, and Klages105.
Free and Wilson106 gave a method in which a set of dummy variables is used in
an attempt to capture the pattern of substituents. The most common of these in
terms of its wide spread application is the use of a physical property as a
surrogate for the structure107. One may call these approaches empirical, they
are lacking in a fundamental respect as structural descriptors. What is needed
is a universal, non-empirical weighting and counting of molecular fragments
to give numerical descriptors which reflect the structure, as evidenced by the
ability to relate physical or biological properties.
Such a method has been suggested in embryonic form by Randic108 and
developed mainly by Kier, Hall and Murray109-113. Kier and Hall has treated
the method in depth in a book and called this method molecular connectivity.
58
In the molecular connectivity method we seek direct relationships between
molecular structure and activity. The molecular connectivity indices are unit
less quantitative descriptors of molecular structure characteristics which are
basically geometric and topological in character. Hence no physicochemical
method or mechanism is presupposed in this approach.
The molecular connectivity method may be considered as a de novo
approach. The information content of the connectivity descriptors is the
structure of the molecule, that is the set of connections between the atoms
expressed in topological terms. Hence the significance to be attached to the
connectivity terms is similar to that given to any fundamental structure
description. The significance is based on topology, not on forces, properties
or other physicochemical type quantities.
(1D.2) MOLECULAR CONNECTIVITY IN STRUCTURAL
INFORMATION
The place of molecular connectivity in the hierarchy of structural
information can be illustrated as in Fig 1.2. At the most primitive level A the
information is limited to the elements present in the molecule. Hence we can
draw only certain limited generalizations derived from physical and chemical
experience with other molecules of the same chemical composition. We may
discuss the molecule in terms of hydrocarbon chemistry, in this case.
59
More information is conveyed at the level B description. The atom
combining ratio as well as the atom types in the molecule are presented in the
empirical formula. Such information indicates that the molecule is an alkane.
Other generalizations about properties, beyond that related level A, can now
be stated.
At level C the information contained in the molecular formula is greatly
increased. However a specific molecule is not identified because of the isomeric
possibilities. Nonetheless, we can now state approximate values of properties
such as solubility, density, molecular refraction as well as discuss chemical
reactivity.
At information level D there is presented for a first time the manner in
which the atoms are arranged or connected within the molecule. Thus, level D
reveals the molecular topology of two structural isomers, structural information
not apparent at level C. The chemist’s physical and chemical experience can
now be used to derive much information concerning the properties of these
molecules. It is at this level that much chemical intuition used by scientists is
focussed.
All of the information about the molecule is contained, in principle, in
the quantum mechanical description level E. This information takes the form
of numerical values and mathematical relations and includes electron
60
distribution probabilities and energetics.
However, at level D, the level of common chemical intuition, a very
large and useful amount of physical and chemical information can be obtained.
At this level it is clear that some properties are directly related to the number
of atoms in a series of molecules, whereas others follow a different pattern. It
is observed that some properties depend upon the branching pattern in a regular
fashion. Other topological features such as cyclization and heteroatom position
have well known influences on properties. Topological methods are based
upon these relationships. Molecular connectivity is a method for developing
correlation based on a formal developments of molecular topology using
elementary aspects of graph theory.
E Ψi = C1i φ1 + C2i φ2 + -----
Ψn = C1n φ1 + C2n φ2 + ----
D CH2 CH3
H3C CH2
CH3
CH
CH3 CH3
→
→
61
C C4H10
B CnH2n+2
A C, H
Fig 1.2 hierarchy of structural descriptions from the most primitive level
A, the elemental constituents of a molecule, to the most complete level E, the
quantum mechanical level D represents the topology of the molecule and is
the basis for numerical indices of structure.
62
(1D.3) GRAPH THEORETICAL AND QUANTUM MECHANICAL
APPROACHES
The distinction between a graph theoretical approach to the molecular
structure and the quantum chemical approach is in the nature of the parameters.
Both schemes aim at the same goal, which is a search for corrections between
the theoretical numbers and selected molecular properties, but the both are not
necessarily concerned with the same properties, and that the emphasis in graph
theory is on conceptual development while the quantum mechanical
calculations tend to be pragmatic, not fundamental. There is complementarily
between graph theoretical and quantum mechanical approaches to molecular
structure. Since molecular structural description is the quantification of
information pertaining to the atomic composition of the molecule such
information includes the number and kinds of atoms together with their
formally-bonded and spatial relationships, where as the most complete
quantitative representation of molecular structure is obtained by the use of
quantum mechanics. Several approaches based in quantum mechanism can be
applied to the graph theory and vice versa62-64, 114-116. In chapter 5 we determined
some physicochemical properties of conjugated aromatic hydrocarbons by
both the methods.
63
PART (E)
ORGANOMETALLIC COMPOUNDS
(1E.1) INTRODUCTION
The compounds in which organic groups are linked directly to the metal
through at least one carbon atom either directly (σ bond) or through a special
(π) type of bonding are called organometallic, in specific case when such
complexation is made by a arene ring; they are called π-arene metal complexes
when metals are transition metals. These compounds are π-arene transition
metal complex such as ferrocene, cobaltocene, chromocene, etc. The suffix
metallic in the terms organometallic is rather loosely interpreted to include
non-metallic elements such as boron, silicon, phosphorous, arsenic, selenium
and tellurium.
As in other areas of chemistry, there are some obvious exceptions to
this simple definition of organo-metallic compounds, for example, inspite of
the species like CaC2 Hg (CNO)2 and Fe(CN)6-4 having metal carbon bond,
these are not considered organo-metallics for traditional or/and some other
obvious reasons.
(1E.2)HISTORICAL ACCOUNT
‘Cacodyl’ (tetra methly diarsine, As2 (CH3)4 was probably the first
organometallic dertivative isolated as early as 1760 by a Prussian military
apothecary. Cadet de Glaussicourt.
64
In 1842 Robert Bunsen discovered an ‘organic-element’ (now called
radical) in the form of methyl group, which forms a compound with arsenic.
Edward Frankland (a British student of Bunsen) attempted to isolate
ethyl-radicals in 1849. This attempt of Frankland led to the synthesis of the
first organozinc compound diethylzinc.
Frankland was soon able to lay the foundations of organometallic
chemistry by extending his studies to other elements like mercury, cadmium,
tin-lead and silicon. These studies added much not only to the initial
development of organo-metallic chemistry, but incidentally a study of their
composition and empirical formula led to the first reliable assignments of atomic
weights of these elements.
The most notable use of organo-metallic (mainly organo aluminium)
compounds as polymerization catalyst for polyethylene and other hydrocarbon
polymers was crowned by the award of Nobel Prize to K. Ziegler and G. Natta
in 1963.
In 1900 Grignard reagents (organo magnessium halides) are discovered.
Lithium alkyls were discovered in 1917 by W. Schlenk and Holtz, by
now organolithium reagents have almost out classed Grignard reagents both
on synthetic and catalytic (industrial) fronts.
A signal recognization of the highly useful role of organometallics in
synthetic organic chemistry has been in the award Nobel prize to Prof. H.C.
Brown in 1977 for the discovery of now well known ‘hydroboration reactions’.
65
(1E.3) GENERAL IDEA OF ORGANOMETALLIC COMPOUNDS
Broadly organometallic compounds can be classified in two ways :
(1) Based on hapticity (from the Greekword haptien : meaning to fasten);
this can be indicated by the number of carbon atoms of the organic moiety,
which are within the bonding distance of the metal such as monohapto
(h1 or η1), dihapto (h² or η²) penta hapto (h5 or η5) organometallics etc.
(2) Based on the nature (extent of polarity) of the metal-carbon bond, on the
basis of which the structure and reactivity of specific organometallic
compound are often explained.
(1E.4) POLARITY OF THE BOND
In all the organo-metallic compound, the bond Mδ+-- Cδ- unlike the truly
covalent c-c bond) is polarized generally in the direction shown in the figure
1.3 and 1.4.
Calculating the percentage ionic character of M-C bond by the simple
Pauling formula : % Ionic character = 100 [1-exp {1/4 (χc-χm)}], where χc and
χm represent the electro negativities of carbon and metal respectively, the Si-c,
Al-C, Mg-C and Na-C bonds should have roughly 12, 22, 34 and 45% ionic
character respectively. For contrast, it may be pertinent to recall in this
connection that the C-C1 bond which has only 6% ionic character on Paulings
formulation, is considered to be quite polar in organic chemistry and many
66
reaction of chloro compounds are explained on that basis.
Similary a synchronous four centre (Fig. 1.3) or six centre (Fig. 1.4)
transition state of the type :
(1E.5) CHROMOCENE
(C5H5)2 Cr chromocene is obtained by reacting with anhydrous
chromium (III) Chloride with sodium cyclopentadienide in tetra hydrofuran
solution gives dark-red air-sensitive volatile biscyclopentadienyl-chromium
(chromocene), melting point 173°C. This chromium complex (C5H5)2Cr has a
sixteen-electron configuration, which is two less than the rare-gas configuration,
and two holes are unpaired, resulting in two unpaired electrons in accord with
the observed para magnetism.
(1E.6) FERROCENE
(C5H5)2 Fe is an orange, crystalline, diamagnetic solid, melting point
173º-174° C. Ferrocene is very volatile and can be purified by steam distillation
or sublimation. It has a high thermal stability decomposing only around 500.
Ferrocene was the first π-cyclopentadienyl derivative to be prepared. It
was discovered by kealy and Pauson as an unexpected product of the reaction
Mδ Rδ
yδ- xδ+
Mδ Rδ
Zδ C- C-
Cδ+
>
>
>
>
Fig. 1.3 Fig. 1.4
67
between cyclopentadienyl magnesium bromide and ferric chloride in dethyl
has been suggested for carbon-metal bond cleavage in their interaction with
polar substrates, X+-Y-. With the above mechanism, the influence of bond
polarity on reactivity is evident.
In this portion four organometallic compound viz cobaltocene,
chromocene, Ferrocene & Nickelocene are studied, their average moleuclar
polarizabilties are calcuated by the δ-function potential model. The structure
and general idea of five organometallic compound are as follows :
(1E.7) COBALTOCENE
Cobaltocene (C5H5)2 Co is formed by reacting with cobalt (II) chloride
with sodium cyclopentadienide in tetrahydro furan solution gives violet black
biscyclopentadienyl cobalt (cobaltocene), metling point is 173º-174º C,
cobaltocene has a nineteen - electron configuration (one in excess of the next
rare gas) and hence exhibits the expected paramagnetism for a complex with
one unpaired electron.
Ether solution in an unsuccessful attempt to prepare the still unknown
dihydrofulvalene C5H5-C5H5 by a coupling reaction. At about the same time,
ferrocene was independently prepared by Miller, Tebboth, and Tremaine in
low yield by passing cyclopentadiene vapour over a special iron-molybdenum
catalyst at elevated temperatures.
Wilkinson, Rosenblum, whiting and woodward postulated the
68
pentagonal antiprismatic structure for ferrocene, which was subsequently
confirmed by X-Rays crystallography.
(1E.8)NICKELOCENE
(C5H5)2 Ni, dark-green in colour, volatile crystalline solid. Nickelocene
has a twenty electron configuration, which is two greater than that of the next
rare gas.
69
REFERENCE
1. Hasse H.R. Proc. Cambridge Phil Soc, 26 (1930) 542; 27 (1931) 66.
2. Hirschfelder J.O., J. Chem Phy, 3 (1935) 555.
3. Hylleraas E.A., Z Phys, 65 (1930) 209.
4. Rosen N., Phys Rev, 38 (1931) 2099.
5. Wang S., Phys Rev, 31 (1928) 579.
6. Kirkwood J.G., J Chem Phys, 5 (1937) 479; Z Phys, 33(1932) 57.
7. Guillemin V. & Zener C, Proc Natl Acad Sci, USA, 15 (1928) 314.
8. MrowkaB.; Z Phys 76 (1932) 300.
9. Steenshold G; Z Phys, 93 (1935) 620, 94 (1935) 770.
10. Van Vleck J.H., Electric and Magnetic Susceptibilities, Oxford Univ
Press, London (1932)
11. Atanosoff J.V., Phys Rev, 36 (1930) 1242.
12. Buchingham R.A., Proc Roy Soc, 160A (1937) 93.
13. Bell RP & Long D.A., Proc Roy Soc, 203A (1950) 364.
14. Abatt JA & Boltan H.C.; J Chem Phys, 20 (1952) 762; Proc Roy Soc,
216A (1952) 477.
15. Kolker H.J. & Karplus M; J Chem Phys, 39 (1963) 2011.
16. Ruedenberg K. & Parr R.G; J Chem Phys, 19 (1951) 1268.
17. Ruedenberg K & Scherr C.W., J Chem Phys, 21 (1953) 1565.
18. Frost A.A; J chem Phys, 22 (1954) 1613; 23 (1955) 985.
19. Lippincott E.R., J Chem Phys, 23 (1955) 603; 26 (1957) 1678.
70
20. Lippincott E.R. & Dayhoff MO, Spectrochim Acta, 16 (1960) 807.
21. Lippincott E.R. & Stutman J.M., J Phys Chem, 68 (1964) 2926
22. Semwal R.P. and Balodi J.P., Ind. J. chem 34A (1995) 206-209
23. Waite J. & papadopoulos M.G., J. Phys. chem,. 94 (1990) 6224
24. Ghosal S., Samoc M., Prasad P.N. & Tufariello J., J. Phys chem, 94
(1990) 2847.
25. Semwal R.P. and Balodi J.P., Ind. J.Chem. 40 A(2001) 687-692
26. Semwal R.P. & Uniyal R.C. and Balodi J.P., J. Ind. Chem Sco. 81 (2004)
31.
27. Semwal R.P. & Uniyal R.C. and Balodi J.P., J. Ind. Chem Sco. 81 (2004)
68.
28. Semwal R.P. & Uniyal R.C. and Balodi J.P., Ph.D. thesis in Physics
submitted to HNB Garhwal Univ. (2002)
29 Hirschfelder J.O., Curtiss C.F., and Bird R.B., Molecular Theory of
Gases.
30. Dixit L. and Semwal RP. J. Sci Ind. Res., 45 (1986) 389.
31. Deb B.M., Ranbir Singh and Semwal R.P., J. Phys. B 26 (1993) 3447.
32. Pauling L., ‘Nature of Chemical Bond’, Cornell Univ Press, Ithaca,
N.Y. (1960)
33. Lewis G.N., J.Am. Chem. Soc., 38 (1916) 762.
34.a. Langmuir I., J.Am. Chem. Soc., 38 (1916) 2221.
34b Linnet J.W., J.Am. Chem. Soc., 83 (1961) 2643.
71
35. Eisenlohr F., Z.Phys. Chem (Leiqzig), 75 (1911) 585, 79 (1912) 129
(1911) 585 & 79 (1912) 129.
36. Vogel A.J. Chem. Soc. (1948) 1833.
37. Batsanov S.S., “Refractometry and Chemical Structure”, Consultants
Bureau, New York, 1961.
38. Silberstein L., Philos, Mag 33 (1917) 92.
39. Miller K.J. & Savchik J.A., J. Am. chem Soc., 101 (1979) 7206
40. Miller K.J., J. Am. chem. Soc., 112 (1990) 6533
41. Winter G.S., Oliver S.N. & Rush J.D., Opt. commun. 69 (1988) 45
42. Green MLH, Marder S.R., Thompson M.E., Bandy J.A., Bloor D.,
Kolinsky P.V. & Jones R.J., Nature 330 (1987) 360.
43. Tam W. & Eaton D.F., calabrese J.C., Williams L.D., Wang Y. &
Anderson A.G., Materials 1 (1989) 128
44. Tam W. & Calabrese J.C.; chem. Phys lett, 144 (1988) 79
45. Jean-Marie A & Joseph D., chem Rev, 91 (1991) 843
46a Von Steiger A.L., Ber., 54 (1921) 1381.
46b Smyth C., Philos. Mag., 50 (1925) 361
47. The molar refraction R (cm³) is related to the mean Polarizability by (3/
4 πN) R = α=0.3964 R (ų)
48. Denbigh K.G., Trans. Faraday Soc., 36 (1940) 936
49. (a) England W., Gordan M.S., Ruedenberg K. Theor. chim. Acta, 37
(1975) 117
72
(b) England W. and Ruedenberg K., theor. chim. Acta 22 (1971) 196
50. Slater J.C., & Kirkwood J.G., phys. Rev. 37 (1971) 196
51. Huckel E, Z Phys, 70 (1931) 204; 72 (1931) 310; 76 (1932) 628.
52. Streitwieser (Jr) A, Molecular Orbital Theory for Organic Chemists,
Wiley, New York (1961)
53. Heilbronner E & Bock H, The HMO Model and its Application, Wiley,
London (1976).
54. Dewar MJS, The Molecular Orbital Theory of Organic Chemistry,
McGraw-Hill, New York (1969).
55. Pilar FL, Elementary Quantum Chemistry, McGraw-Hill, New York
(1968).
56. Coulson CA & Strietwieser (Jr) A, Dictionary of π-Electron Calculations,
WH Freeman & Company, San Francisco (1965).
57. Salem L, The Molecular Orbital Theory of Conjugated Systems,
Benjamin, New York (1966)
58. Coulson CA, Valence, Oxford University Press, Glasgow (1961).
59. Pullman B & Pullman A, Quantum Biochemistry, Wiley, New York
(1961).
60. Kier LB, Molecular Orbital Theory in Drug Research, Academic Press,
New York (1971).
61. Kier LB & Hall LH, Molecular Connectivity in Chemistry and Drug
Research, Academic Press, New York (1976).
73
62. Wilson R, Introduction to Graph Theory, Academic Press, New York
(1971).
63. Ariens EJ, Drug Design, Vol 1, Academic Press Inc., New York (1971).
64. Purcei WP, Bass GE & Clayton JM, Strategy of Drug Design : A guide
to biologicial activity, Wiley, New York (1971).
65. Murrel JN, Schmidt W & Taylor R, J Chem Soc, Per2, (1973) 179.
66. Dewar MJS, Frotschr Chem Forsch, 23 (1971) 1.
67. Lowdin PO in Molecular Association in Biology, Pullman B, Ed
Academic Press, New York (1968) p. 539.
68. Pullman B, Int J Quant Chem, 35 (1969) 83.
69. Castro EA & Sorrarain OM, J Chim Phys, Physiochim Biol, 69 (1972)
513
70. Green JP, Johnson CL & Karg S, Ann Rev Pharmacol, 14 (1974) 319.
71. Dewar MJS & Schmeising HN, Tetrahedron, 5 (1969) 166, 7(1960) 96.
Mulliken RS, Tetrahedron, 5 (1969) 253; 6 (1959) 68.
72. Heitler W & London F, Z Phys, 44 (1927) 455.
73. Pauling L, J Am Chem Soc, 53 (1931) 1367, 3225; 54 (1932) 998,
3570.
74. Slater JC, Phys Rev, 37 (1931) 481; 38 (1931) 1109.
75. Pauling L, The Nature of the Chemical Bond, Cornell Univ Press, Ithaca,
New York (1960).
76. Wheland GW, J Am Chem Soc, 63 (1941) 2025, 64 (1942) 900.
74
77. Maslen VW & Coulson CA, J Am Chem Soc, 79 (1957) 4041.
78. Mulliken RS, Rieke CA, Orloff D & Orloff H, J Chem Phys, 17 (1949)
1248.
79. Chirgwin BH & Coulson CA, Proc Roy Soc, 201A (1950) 197 also
Lowdin PO, J Chem Phys, 18 (1950) 365.
80. Ruedenberg K, J Chem Phys, 34 (1961) 1861, 1878, 1884, 1892.
81. Amos AT & Crispin RJ, J Chem Phys, 63 (1975) 1890.
82. Hoffmann R, J Chem Phys, 39 (1963) 1397; 40 (1964) 245, 2474, 2480.
83. Hinze J & Jaffe HH, J Am Chem Soc, 84 (1962) 540.
84. Slater JC, Quantum Theory of Molecules and Solids, Vol 1, McGraw-
Hill, New York (1963) p107-8.
85. Pople JA & Santry DP, Mol Phys, 7 (1963) 269; 9 (1965) 301, 311.
86. Hartmann H, Advan Chem Phys, 5 (1963) 1.
87. Cyvin SJ, J of Mol strtuct, 86 (1982) 315.
88. Cyvin SJ, Kgl Norske Videnskab Selskabs Skr, 4 (1982) 315.
89. Cyvin SJ, Tetrahedron Letters, 22 (1981) 2709.
90. Cyvin SJ, Cyvin BN & Brunvoll J, Kgl Norske Videnskab Selskabs
Skr, 5 (1982) 1.
91. Moseley (JR) WD, Ladik J & Martension O, Theoret Chim Acta,
8 (1967) 18.
92. Mulliken RS, J Chem Phys, 46 (1949) 497.
93. Laforgue A, J Chem Phys, 46 (1949) 568.
75
94. Streitwieser (Jr) A, Nair PM, Tetrahedron, 5 (1959) 149.
95. Dogget G, Mol Phys, 10 (1966) 225.
96. Lennard-Jones JE, Proc Roy Soc, 158A (1937 280.
97. Longuet-Higgins HC & Salem L, Proc Roy Soc, 251A (1959) 172.
98. Gupta SP, Tayal P & Singh P, Match, 19 (1986) 53.
99. Gupta SP & Sharma MK, Match, 21 (1986) 123.
100. Lykos PG, J Chem Phys, 35 (1961) 1249, Advan Quant Chem, 1 (1964)
171.
101. Ohno, Theoret Chim Acta, 2 (1964) 219.
102. Wiener H, J Amer Chem Soc, 69 (1947) 17, 2639; J Phys Chem, 52
(1948) 425, 1082.
103. Smolenski EA, Russ J Phys Chem, 38 (1964) 700.
104. Franklin JL, Ind Eng Chem, 41 (1949) 1070.
105. Klages F, Chem Ber,m 82 (1949) 358.
106. Free (Jr) SM & Wilson JW, J Med Chem, 7 (1964) 395.
107. Hansch C & Dunn WJ, J Pharm Sci, 61 (1972) 1.
108. Randic M, J Am Chem Soc, 97 (1975) 6609.
109. Kier LB, Hall LH, Murray WJ & Randic M J Pharm Sci, 64 (1975)
1971, 65 (1976) 1226.
110. Kier LB, Hall LH & Murray WJ, ibid, 64 (1975) 1974, 1978.
111. Kier LB & Hall LH, ibid, 66 (1977) 1, 642; 70 (1981) 583; 72 (1983)
1170.
76
112. Kier LB & Hall LH & Murray WJ, J Med Chem, 18 (1975) 1272; 19
(1976) 573.
113. Kier LB, Hall LH & Dipaolo T, Mol Pharm, 13 (1977) 31.
114. Kier LB & Holtje HD, J Theoret Biol, 49 (1975) 401.
115. Kier LB & Hall LH, ibid, 0 (1976) 177.
116. Kier LB, George JM & Holtze HD, J Pharm Sci, 63 (1974) 1435.