A OORRELATWE STUDY OF THE UV SPECTRA OF INOOLE
AND THE ALAINDOLES (NCLUDANG PURINE;
A SYMBIOTIO COMPUTATIONAL AND

EXPERWENTAL APPROACH

Thesis for the Oe’gr'ee of Ph. D.

MEOHEGAN STATE ONWERSITY

RBCHARD WAYLANO WAGNER

31931

 

a.” LIBRARY

Michigm State
Uniw Sity

 

This is to certify that the

thesis entitled

A CORRELATIVE STUDY OF THE UV SPECTRA 0F INDOLE
AND THE AZAINDOLES INCLUDING PURINE:
A SYMBIOTIC COMPUTATIONAL AND EXPERIMENTAL APPROACH

presented by

Richard W. Wagner

has been accepted towards fulfillment
of the requirements for

Pho D. degree in BiOthSiCB

(27:, K/ ‘ jILMIIj/hyg,‘ .
Mdmmmfi (]

Date (L3 Cat {511 \

0-7839

——-__— __ . .____._ __,77‘,,7 _ , , _ - , , - __- _ _ 7,

ABSTRACT

A CORRELATIVE STUDY OF THE UV SPECTRA OF INDOLE
AND THE AZAINDOLES INCLUDING PURINE:
A SYMBIOTIC COMPUTATIONAL AND EXPERIMENTAL APPROACH
By

Richard wayland Wagner

The ultraviolet spectral characteristics of indole, the mono and
some diazaindoles and purine were compared and correlated. The
characteristics included vapor and solution absorption.measurements,
polarization measurements, fluorescence and phosphorescence. To further
aid these correlations Pariser-Parr-Pople calculations were carried out.
Agreement between the calculated observables and the experimental values
was quite good. In addition the calculated‘fihelectron charge densities
and permanent dipole moments for the ground and two lowest excited
singlet states were correlated with the absorption and fluorescence
shifts in various solvents.

The 1La<-——-1A transition was energetically lower than the other
singlet transitions for all molecules except those with an aza nitrogen
substitution in the pyrrolic ring. For these molecules the 1Lb<+-1A 1
transition was lowest. The vapor absorption spectra of indole, indazole,
benzimidazole and 7-azaindole exhibited an anomalous temperature dependent

effect.’ As the temperature was increased the 1La/lLb transition ratio

increased. This was attributed to a vibronic coupling scheme.

Richard Whyland wagner

The vapor phase 1Lb'<—-1A transition displayed less vibrational
structure as more aza nitrogens were introduced into the indole ring.
With three aza nitrogens (purine) no vibrational structure was evident.

The biological significance of these results remains obscure. The
spectral phenomena probably reflect the subtle differences in the
electronic structure these molecules possess. This structure is
biologically manifested in the chemical reactions in which these
molecules participate. Thus these results indicate that slight
differences in molecular architecture lead to quite varied biological

functions.

A CORRELATIVE STUDY OF THE UV SPECTRA OF INDOLE
AND THE AZAINDOLES INCLUDING PURINE:

A SYMBIOTIC COMPUTATIONAL AND EXPERIMENTAL APPROACH

By

Richard wayland wagner

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements

for the degree of

DOCTOR OF PHILOSOPHY

Department of Biophysics

1971

DEDICATION

To Ann, Terri, Deborah and especially Marlene
for the patience afforded the only male member of the household

during his studies for this degree

ii

ACKNOWLEDGMENTS

I wish to thank my thesis advisor, Dr. M. Ashraf El-Bayoumi, for
his guidance and encouragement throughout the performance of the
various phases leading to this dissertation. His friendship is
appreciated and his vitality and flexibility is admired.

I am also indebted to Dr. Petr Hochmann for his willingness to
lead me into the quantum chemical jungle. His computer program was
used in performing the calculations whose results are presented here.
In addition he spent many (pleasant for me!) hours discussing quantum
mechanical features underlying this program.

Other members of the department and research group must also be
recognized. Many fruitful discussions were held with Dr. Kenneth C. Ingham
and Dr. Fred watson which cemented many spectroscopic ideas. The
banter we exchanged also helped make the frustrations of this work
more bearable. The photographic facilities and aid provided by
Prof. J. I. Johnson and by John Haight are also greatly appreciated.

Finally I wish to acknowledge my gratitude to my initial major
professor, Dr. Leroy G. Augenstein, whose untimely death severed a

relationship which was particularly influential.

iii

TABLE OF CONTENTS

LIST OF TABLES
LIST OF FIGURES
CHAPTER I INTRODUCTION

CHAPTER II THEORETICAL CALCULATIONS
I. Introduction
II. Background for Theoretical Calculations
111. Format for the Calculations

CHAPTER III EXPERIMENTAL PROCEDURES
I. Vapor Absorption Spectra
II. Solution Absorption Spectra
III. Emission Spectra
IV. Solvents
V. Experimentally Studied Molecules

CHAPTER IV ANALYSIS OF SPECTRAL DATA FOR EACH MOLECULE

INDOLE
I. Vapor Absorption Spectra
II. Polarization Measurements of Fluorescence Excitation
III. Solution Absorption Spectra
IV. Fluorescence Spectra
V. Phosphorescence Spectra
VI. Theoretical Calculations
VII. Correlations and Summary

INDAZOLE
I. Vapor Absorption Spectra
II. Polarization Measurements of Fluorescence Excitation
III. Solution Absorption Spectra
IV. Fluorescence Spectra
V. Phosphorescence Spectra
VI. Correlations and Summary

BENZIMIDAZOLE
I. Vapor Absorption Spectra
II. Polarization Measurements of Fluorescence Excitation
III. Solution Absorption Spectra
IV. Fluorescence Spectra
V. Phosphorescence Spectra
VI. Theoretical Calculations

VII. Correlations and Summary

iv

vi

vii

H

48

48
48
55
57
61
72
74
78

81
81
84
84
87
88
89

9O
9O
93
94
98
100
102
102

4-AZAINDOLE 109
I. Absorption and Fluorescence Spectra 109

II. Theoretical Calculations 110

III. Correlations and Summary 110
5-AZAINDOLE 110
I. Absorption and Fluorescence Spectra 110

11. Theoretical Calculations 110

III. Correlations and Summary 111
6-AZAINDOLE 111
I. Absorption and Fluorescence Spectra 111

II. Theoretical Calculations 111

III. Correlations and Summary 112
7-AZAINDOLE 112
I. Vapor Absorption Spectra 112

II. Solution Absorption Spectra 115

III. Fluorescence Spectra 118

IV. Theoretical Calculations 120

V. Correlations and Summary 120
BENZOTRIAZOLE 125
I. Vapor Absorption Spectra 125

II. Polarization Measurements of Fluorescence Excitation 129

III. Solution Absorption Spectra 130

IV. Fluorescence Spectra 131

V. Phosphorescence Spectra 132

VI. Correlations and Summary 132
4-AZABENZIMIDAZOLE 134
I. Vapor Absorption Spectra 134

II. Absorption and Fluorescence Spectra 137

III. Theoretical Calculations 139

IV. Correlations and Summary 139
5-AZABENZIMIDAZOLE 141
I. Absorption and Fluorescence Spectra 141

II. Theoretical Calculations 142

III. Correlations and Summary 143
PURINE 143
I. Vapor Absorption Spectra 143

II. Transition Assignments 145

III. Absorption and Fluorescence Spectra 146

IV. Phosphorescence Spectra 151

V. Theoretical Calculations 152

VI. Correlations and Summary 155
CHAPTER.V INTERMOLECULAR CORRELATION OF THE SPECTRAL DATA 159

BIBLIOBRAPHY 178

Table

Table

Table

Table

Table

Table

Table

Table

Table

Table

Table

Table

3a.

3b.

10.

11.

LIST OF TABLES

Semiempirical Parameters.

Comparison of Experimental and Calculated Values
for Test Molecules.

Solvent Effects on Indole Absorption.
Indole:

Comparable Maxima in 2 Solvents.

Indole Emission in Various Solvents.

Effect of pH on Fluorescence in Indoles (Reference 40).

Calculation Results for Indole.
Solvent Effects on Azaindole Spectra.
Calculation Results for the Azaindoles.
Purine Absorption and Luminescence.
Effects of pH on Purine Absorption or Fluorescence.

Calculation Results for Purine.

vi

35

37

59

59

65

71

75

85

103

147

149

149

Figure

Figure
Figure
Figure

Figure

Figure

Figure
Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

LIST OF FIGURES

Melecules of Biological Interest Containing an
Indole or Purine Chromophore.

Molecules Investigated in this Study.
Luminescence Apparatus.
Indole Vapor Absorption Spectrum.

Indole Vapor Absorption Spectra as a Function
of Temperature.

Solvent Effects on the Absorgtion Spectrum of Indole
(after Chignell and Gratzer2 ).

Calculated Charge Densities and Bond Lengths for Indole.
Indazole Vapor Absorption Spectrum.

Indazole Vapor Absorption Spectra as a Function
of Temperature.

Benzimidazole Vapor Absorption Spectrum.

Benzimidazole Vapor Absorption Spectra as a Function
of Temperature.

Benzimidazole Absorption Spectra in Basic, Neutral
and Acidic Solution.

Benzimidazole Fluorescence Spectra in Basic, Neutral
and Acidic Solution.

Benzimidazole Phosphorescence in Ethanol at 770 K.

Calculated Charge Densities and Bond Lengths
for Benzimidazole.

7-Azaindole Vapor Absorption Spectrum.

7-Azaindole Vapor Absorption Spectra as a Function
of Temperature.

vii

43

50

53

58

77

82

83

91

92

97

99

101

107

113

114

Figure

Figure

Figure

Figure

Figure

Figure

Figure

18.

19.

20.

21.

22.

23.

24.

viii

Calculated Charge Densities and Bond Lengths
for 7-Azaindole.

Benzotriazole Vapor Absorption Spectrum.

Benzotriazole Vapor Absorption Spectra as a Function
of Temperature.

4-Azabenzimidazole Vapor Absorption Spectrum.

4-Azabenzimidazole Vapor Absorption Spectra as a Function
of Temperature.

Purine Vapor Absorption Spectrum.

Calculated Charge Densities and Bond Lengths for Purine.

122

126

127

135

136

144

154

CHAPTER I

INTRODUCTION

Indole is a unique and interesting molecule to spectroscopists,
biochemists and biophysicists. It is the chromOphore of tryptophan,
an amino acid found in most proteins. Tryptophan's maximum extinction
coefficient (at 280 nm) is five times that of tyrosine (at 278 nm) and
fifteen times that of phenylalanine (at 260 nm) which are the other
amino acids absorbing light in the UV region of the electromagentic
spectrum. Thus tryptophan accounts for most Of the energy dePOSitiOD
from UV irradiation in proteins (this statement obviously isn't true
for certain proteins which don't contain tryptophan, e.g. ribonuclease)-

The indole chromophore is also contained in several other molecules
of biological significance. A representative group of these molecules
is depicted in Figure 1. As can be seen from the figure these mole-
cules may be considered as an indole molecule with attached substitutent
groups. Serotonin is a transmitter substance found at chemical
synapses between neurons. It has also been found to block.the action
of adrenaline at chemical synapses which utilize adrenaline as a
chemical transmitter substance. Indole acetic acid (1AA) is an auxin
or plant growth hormone. Lysergic acid is an alkaloid which has achie-
ved recent prominence as the basis of the hallucinogen LSD.

If one takes indole and judiciously replaces three carbons with
three aza nitrogens purine could be produced. The nitrogen - carbon
replacement would have to occur at the 3, 5, and 7 posiitons to form
this base. Purine is the chromophoric basis of another group of

biologically important molecules. Adenine and guanine are two of the

”“2
| NH2
CHz-CH'COOH
N
\ N I \>
N K N
N I
H H
Adenine
Tryptophan
- H «- 0
CH2 C 2 NH2 H N
HO \ ‘N l \>
'1' HZNJ\N '7‘
H H
Serotonin Guanine
CHz-COOH
COOH

\\ N'CH
't‘ \ 3

H
Indole Acetic Acid

\

A
Lysergic Acid

Figure I. Milit'l'tllt'H nl' Ilinlnrjrnl lnlvrmzl (Znnminim: .In Indole
Ur I’urim- Chrunmplmrv.

3

bases found in the DNA and RNA genetic molecules. These base structures
are also shown in Figure l. Purine is the chromophoric basis of ATP
which is a very important intermediary in metabolism. The purine struc-
ture is also found in the NAD coenzyme molecule. Although an extensive
number of spectroscopic investigations have been made of the purines,
relatively few of these have been concentrated on purine itself. Most
of the effort has been directed toward the purines found in biological
systems such as adenine, guanine and hypoxanthine. However to inter-
pret the spectroscopic data amassed for these molecules, one could

begin with purine as the basis molecule and consider the other purines
as purine with various substitutents. This would assume the spectro-
szopic nature of purine was known. Obviously this approach hasn't been
taken since there is paltry data available for purine.

Spectroscopic investigations of these biologically significant
molecules can yield information which is useful in at least two respects.
First there has been much attention devoted to the effects of radiation
on biological systems. Radiation Biophysics has historically been an
important investigative avenue. The spectroscopically investigated
properties of biological molecules have been found to shed light on the
effects of radiation. Knowledge about the interactions of UV electro-
magnetic radiation with these molecules fosters an understanding of the
available pathways for irradiation deposited energy. A second use for
the spectroscopic information pertains to the structure of these mole—
cules. They are not static models composed of units of mass or elec-
trical charge. Their biological relevance recalls the concept that
they are dynamic systems. One should probe various physical aspects of

these molecules to learn about their structure and its relationship to

4

their function. UV spectroscopic investigations are probes which can
yield information about the electronic portion of the molecular struce
ture. Since the electronic portion participates in chemical reactions
it is particularly important to obtain knowledge of this molecular com-
ponent. In addition these spectroscopic studies give some insight into
the dynamic qualities of the electronic component.

Obviously there are several possible spectroscopic studies which
could be done. The approach used in this study is somewhat different
than in most previously attempted investigations. It seemed strange
that the closely related chromophores, indole and purine, are found in
such a wide variety of biological molecules. The biological function of
these molecules is also widely variant even though the chromOphoric
difference consists of only an interchange of three carbons and three
nitrogens. It is possible that the differences in ecological usage of
these chromophores is related to the presence or absence of the aza
nitrogens. There is a chromophoric family of molecules which consists
of indole, monoaza, diaza and triaza indoles. This family if formed
by successive aza nitrogen substitutions at different positions in the
indole molecule. There are a large but finite number of such substi-
tutions which can be made. If three substitutions are made at specific
positions the triazaindole called purine is formed. Thus it was decided
to undertake a systematic study of this azaindole molecule family in an
attempt to characterize the spectroscopic effects of aza nitrogen substi-
tution on indole. In this manner it was hoped that information would be
revealed about the inherent electronic differences between indole and
purine. Widiso many possible azaindoles to study, time dictated that

only a fraction of them be investigated. The choice was further

5
simplified by the commercial availability of azaindoles as well as the

accessibility of previously published data for molecules in this family.
After this paring the molecules which were investigated in this study
are shown in Figure 2.

In order to reach the objectives of this tudy it was decided that
a correlative investigation of the UV spectroscopic data for these mole-
cules should be done. All available published experimental data would
be used. This would be augmented by experimental work performed in this
laboratory. In addition it was decided that theoretical calculations
should be undertaken to add another dimension to our concept of the
electronic structure and spectroscopic properties of these molecules.

It was haped that a unified picture of the electronic structure would be
developed which could explain the spectroscOpic data. In addition it
was hoped that this spectral data unification would incorporate the
effects of the molecular environment on the spectroscopic properties of
these molecules.

One of the obvious spectroscopic features not present for indole
but which azaindoles might possess is the ndfi* transition. This transi-
tion consists of the promotion of an electron from the electron lone
pair or nonbonding orbital on an aza nitrogen to an antibonding‘fl*
orbital. It is characterized by its weak intensity resulting from the
negligible overlap of the nonbonding and antibonding orbitals. The
other transition which should occur for both indole and the azaindoles
is theflLfi* transition. It consists of an electron promotion from a
bondingiiorbital to an antibonding fl* orbital and is usually intense

due to the overlap of these two orbitals. It will become apparent in

this study that both transitions are important spectral properties of

\ u:
INDOLE [NDAZOLE
N N
H

H

.. N

N\ / \
BENZIMIDAZOLE I A‘AZAINDHIJ‘:

N \ N

H H

   

   

3— AZA I NUOLE . N 6-AzA [ NW“?

N ' N

H H

O.

l \ \N:
\ 7- AZAI NDOIE Anszm R 1.\7.m..r:

N N N

.. H H

  

A'AZAHENZIMIUAZULI‘: 5 «\7..\|>i§\l7.il‘il i).\7.\“i!
\ N \ N

,. qK

Figure 2. Molc-cnlcvc. lnvwsliunlml in this Study.

these molecules.

With this brief background we are ready to proceed to the rest of
this dissertation. The next chapter will outline the method for perform-
ing the theoretical calculations. Chapter 3 will be a discussion of the
experimental techniques and apparatus employed in this study. Chapter 4
will be compilation and discussion of all the available data for the
investigated molecules. The data for each molecule will be discussed
separately and will include results personally obtained during the course
of this study. The final chapter will be devoted to a discussion of the
trends and correlations which occur in this family of molecules. It will
be concluded with a few statements concerning the accomplishments of this

correlative undertaking.

CHAPTER II

THEORETICAL CALCULATIONS

I. Introduction

One of the powerful tools for studying the electronic structure of
molecules is the performance of correlative theoretical calculations of
certain molecular observables for a given class of molecules. An exam-
ination of trends in properties such as ionization potentials, electron
affinities, transition energies and intensities often yields valuable
information. It is also possible to then compare the calculated observ-
ables with corresponding experimental results. This gives one a feel—
ing for how reliable the calculated observables are, allows further
correlations between theoretical and experimental quantities and gives
the freedom of predicting values for which no experimental results are
available. Of course this is all predicated on the validity of the
theoretical calculations.

The following sections will be devoted to this question of validity.
Pariser-Parr-Pople type calculations were made of the molecules under
study. The general theoretical background for these calculations will
be described, followed by the rationale used here in specifically
performing them. The form, parametrization, and type of input and
output used here will also be presented.

1,2,3
11. Background for Theoretical Calculations

To date the most powerful and fruitful approach for the analytical

calculation of physically interpretable quantities pertaining to mole-

cules is that of Quantum Mechanics. By solving the Time Independent

Schrodinger Equation

A

H w-w V (1)

many molecular properties may be examined. For any molecule the

Hamilton operator H may be represented by

 

 

way": u Vang—.91. +> Elms:
Lg‘aML . ALWVL—r,‘ ldtfi‘R"-RPI

MK a
__‘/a z... (2)

 

where the K nuclei of the molecule are indexed by the values of a, with

charge Zae and mass Ma” and the N electrons are indexed by the values

of 1. However this representation has been formulated with the follow-

ing assumptions.

Assumption 1:

Assumption 2:

Assumption 3:

A non-relativistic representation is adequate for
reproducing the observables under study. That is,
relativistic corrections are of such a magnitude as

to be negligible. Thus non-electrostatic interactions
are ignored.

All molecular electrons and nuclei are regarded as
point charges. Terms which are neglected by this
assumption (such as that due to nuclear quadrupole) are
of the order of relativistic corrections.

The molecule under investigation is isolated and not
subject to any external fields. Stationary state
values of V adequately provide the observables of
interest. Thus V is a function of spatial and spin

variables only.

10

Unfortunately even with these assumptions it is impossible to
exactly solve Equation 1 except in the most elementary cases. Further
assumptions must be made to simplify the analysis. These assumptions,
as well as the preceding ones, are primarily based on our physical
intuition regarding the system being studied.

For the molecules studied here, only those properties primarily

due to the electrons are of interest. Since mi
(<1

M
Assumption 4: The separability of the electronaand nuclear motions in

the sense of the Born - Oppenheimer (Adiabatic)
Approximation is assumed. The electronic part of the

wave function is then an eigenfunction of the Hamilton-

 

ian:
NM (3)
/\ N “a. V.“ ‘ , ea ‘ K'Ni ea
H=~an~+22fiffirl~g —-2‘———,—
L=i ° Ltd, 1' «’LiRe-rbi

This eigenfunction is determined only for the nuclear configuration
which corresponds to the equilibrium geometry of the molecule in the
ground state. Therefore all properties of molecular excited states
evaluated within this approximation are those of vertical states (see
Franck - Condon Principle). The wave function now only depends
explicitly on the electronic spatial and spin coordinates., This wave
function may be represented as a linear combination of Slater deter-

minants. Let AK denote the following Slater determinant:

AK ==V‘s—7Z(~\)P’P[x.‘m,x,<am,......,xK~(~)] (4)

P

11

where ‘1 is the Ki-th spin orbital. By using this representation the
requirement of an antisymmetric wave function with respect to electron
coordinate transpositions (PauliExclusion Principle) will be satisfied.
It is apparent from the normalization factor in Equation 4 that
Assumption 5: The spin orbitals comprise an orthonormalized set.

They can be written as a simple product of MO and spin

functions n(s): ¢(r)n(s). (The n(s) are either a or B)
The molecular orbitals are the eigenfunctions of an effective one-
electron Hamiltonian which will be specified later. Since the wave
function is represented as a linear combination of Slater determinants:

w - ZKCKAK (5)

A A
Q

Since H does not contain any spin dependent operator, S2 and S

A A

commute with H. Therefore the eigenfunction of H can be simultaneously

2

made the eigenfunction of Szand S2. This condition restricts the class
of wave functions which are eigenfunctions of H. However the same
function may be used to calculate both energy and spin related observ-
ables. This is important when molecular states of a particular
multiplicity are under scrutiny.

Although molecular calculations could be made in principle with
these assumptions they would rapidly become intractable without some
further restrictions. These restrictions are made to simplify the
computations which must be performed.

Assumption 6: A one - electron effective Hamiltonian will be used. Its
form will be specified later.

This Operator will serve as a generator of electronic molecular

orbitals which form the eigenfunctions. The molecular orbitals (M0)

12
are a property of the molecule and are of the form illustrated in
Assumption 5. Each electron is assigned to an M0. When the MD are
written in the product form of Assumption 5 it is possible to place
two electrons in each spatial function ¢(r). This does not violate
the Pauli Exclusion Principle since the accompanying spin coordinates
may have either of the two allowed values, a or 8.

Certain requirements of the A in Equation 5 must be met before the

R

w defined by Equation 5 is an eigenfunction of the Hamiltonian of

Equation 3. The AK must be functions which comprise a complete ortho-

normalized set with respect to quadratically integrable antisymmetric

N - electron functions. The CK in Equation 5 are determined by the

infinite order eigenvector equation. However the N - order Slater

determinants, constructed from the complete set of orthonormalized spin

orbitals X, make such a complete orthonormalized set of N - electron

wave functions. Equation 5 is thus an eigenfunction of the stated

Hamiltonian. This equation is the basis of Configuration Interaction

which is an expansion over a complete set of N - particle functions.

In the calculations performed here

Assumption 7: A Limited Configuration Interaction (LCI) involving a
truncated set of ground and monoexcited states is used.

This assumption is made for two reasons. First experience has shown

that the ground state and monoexcited states make the largest contri-

bution to calculated properties of the ground and lowest excited

states. Second the computers used in performing these calculations have

only a finite capacity so truncation must be performed.

Obviously the function u is not analytically correct and is

completely undetermined as it has been written. It only approximates

13

the "correctusolution to the Schrodinger Equation. However to obtain
the best eigenvalue which may be compared to the experimentally deter-
mined energy, some mechanistic procedure must be found to improve
successive "guesses" of the true wavefunction. Such a procedure is

found in the Variational Theorem where for

 

 

f‘i’ *fi “V AT Variational (6)
:2: nn\n
Y Iwk \‘Y 0‘? Theorem
values of V are guessed over and over
/\
42* H .W
E = I “Y (7)
i ‘i’* ‘l’ .W
and
El" (8)

In this case W corresponds to the molecular ground state energy. The
value of V which minimizes this energy will not in principle be a good
function for calculating any other energy state or any other observable.
However these functions together with Configuration Interaction will be
used here. The parametrization which is outlined later partially over-
comes this obstacle. For the molecules studied here the Variational
Theorem is modifed to the Linear Variational Principle. This principle
is used to determine the expansion coefficients CK in Equation 5. Using

the form of Equation 5

L.
a; CT(HIJ‘ .. SITE) = O (9)

where
'i A 0
H17 = H3; "PIA: H A3. al’r(l,2.,----N) and (1 )

531, :HS;;'=QLA::llr‘TT’(h£L;'u'h0

14

There are L such linear equations and the nontrivial solution for the

coefficients requires the Secular Equation
H - S E - 0
|( IJ IJ )I

This yelds L roots for E‘ which converge to W‘ from above as more

independent A1 are added, increasing the value of L.

At this point a closer examination of the Secular Equation is

(11)

warranted. Without loss of generality the x1 and ti functions compos-

ing AI may be assumed to be orthonormal as intimated in Assumption 5 so

SlJ - 313

The Hamiltonian is now broken up

 

/\ N A . Nu“ a
HA(1.1,“°N) :. éHKM “' {TEAS-r“
where obviously
A ka 1 K K a.
. a __ V __ age
\n-‘Kb‘A _ 2m; L at \R-t‘rti

It is judicious to delineate four possible cases:

Case 1: AI and AJ are identical.

Case 2: AI and AJ differ by only one spin orbital with X
m

enteringA where X entersA.
I p J

Case 3: A3 and£§ differ by two spin orbitals with Km and Xn

entering A where X and X enters A .
I P q J

Case 4: AI andl}I differ by three or more spin orbitals.

(12)

(13)

(14)

15

Each term in the Secular Equation can be evaluated for these cases by

using the Slater Rules. In light of Equation 12

S - l for Case 1 (15)
IJ 0 otherwise
N I t N / I
%(I; +fi§(3—L*-KL’\\ for C8881
I N . . I . . I
HIJ - I"? +£1.59 LCPML—P) _._ (L "\P“)] for Case 2 (16)
O + (mni‘a%\’_ (mn‘qflfl’ for Case 3
O for Case 4

where the primes denote spin orbitals and

I;L = I; .-_ jx:(.)’HK(Aix,(a),\/r(.) W
15* = [xzcofiiK (.A x, to.) MC...)
(-. ,M ., l)’ = [x3 (.3 mun—5a X,(~3X1(B).\T(ui Am.)

 

(17)
I 0 e I
It} ‘1 (“1"“13 is the Coulomb Integral

/ . . /
Kb} :. (Cé,|*b) is the Exchange Integral

It is possible to switch from spin orbitals to space orbitals easily.

I’ - I if X and X1 have the same spin.
1

11 13
(18)
0 otherwise.
1’ . 1 . 1 19
ii ii i ( )
(ikljl)' - (ikljl) if X1 and Xj have the same spin and
Xk and X1 have the same spin. (20)

0 otherwise.

l6
- J

J13 1.
His - K1.1 if X1 and Xj have the same spin. (21)
0 otherwise
It is now possible to calculate (at least symbolically) the expec-
tation value for the ground state energy of a given molecule. Each

spatial MO used has two electrons in it with opposing spins forming a

Closed Shell Configuration. For this case

EN =Z(;LI +2 (.1317— Kiel) (22)

f-fe

where‘V- N/2.

So far it has not been apparent how to use the Variational Theorem
to best advantage in calculating the minimum E. This is accomplished
by carrying out a variational calculation using the method of Lagrangian
Undetermined Multipliers.

It is asked that

E=EN~igewsw (23)
be a minimum or 55- O. This gives
0" Zi3¢.[‘iH 4’ +Z|.,(8’CY ch.— —‘te.,¢ i] (24)

whereJ 13 I. I; CPA (“HI—Vi???“ ¢:,(.b).lar(e)ai4r(b) :K;¢:(QI1(‘)¢L@JVG)
= i .; A») W «us no.)

(25)

.. fi<«>¢.(«>=[i¢;(b),. -M mamas] 42 a.)

17

and K1j = I { mfg.) MAE-.31 41-1») qmmaom

= I we») K4.) ¢¢(.),x..(.\ = f ago) Kw.) 4w) Ava») (26>

.. 1}.) Mo.) = [rgmfi—q c». 0:) ohm] we.)

Since equation 24 is nontriviallv satisfied when the term in the bracket

is zero
/\ "’ '0
[HK(Q5 +’Z',|(1I’(G3 "’ K’IGA] ¢g(0~) = E 664. (hi, (a) (27)
without loss of generality Eij - eisij (28)
which finally yields
/\
F <a)¢1(a) - 6145“» (29)

These are the Hartree-Fock integro-differential equations. Since the
‘§~depends explicitly on the functions ¢i the tact is to guess a starting
set ¢1, evaluate‘9; solve Equation 29 and iterate until self-consistency
is attained. This is known as the Self-Consistent Field (SCF) method.
The transformation of Equation 28 is not difficult to perform and
may be done in general. However for most molecules it is difficult to
solve the Hartree-Fock formulation (Equation 29). A judicious choice
of MO would alleviate this problem. The MD may be of any form. However
one form has gained popularity because it is based on a subset of
orbitals whose properties are known.
Assumption 8: The Molecular Orbitals are constructed of a Linear
Combination of Atomic Orbitals (LCAO).

”-
4)‘: : 2 CL? IF where m_>_N (30)
P

Ip are the A0 and Cip are coefficients. Without loss of generality the

18

LCAO form an orthonormal set and

i
r. 31
When this substitution is made in the Hartree-Fock Equation followed
m a
by a multiplication on the left by EC 1* with integration a new
Q In; q
expansion is formed
mu 1* N\
C - — E- = (32)
E; A} gct‘t (F3? 5319 9*) O
:2 * —- (33)

and F$P=II;?IPAV -= I“; +§[a(%élF?)-(‘HH’P)] (34)

Since the A0 are presumed known it is possible to diagonalize the £1

J
matrix and to calculate the integrals.

L“, = fryqfidqmqlum (35)

and (Fears) .-= {new amp-5.1;; 1,0.) 150944.») Ads) mm

It should be noted that

M * m A
IczécafigcwfigHKIPaV (37)
3L4, ~7- % Ci; g CLP‘I: a”, IPA” (38)

M m
”4 % “figcwhs Kérr“

Performing a variational calculation on Equation 32 yields

m

% CL? (FP‘b '— SF‘b‘ 6L) : O Roothaan Equations (40)
which lead to a Secular Determinant of order m

‘1? - 5 el- 0 (41)
m M

19

Equations 37-39 show that the operatorlgxdepends explicitly on the C1D,
(see also Equations 64-66). To achieve the best MO which minimizes the
molecular energy a set of C11) is assumed, the matrix “F” is calculated
with the aid of Equations 37-39, the Secular Determinant is solved for

the N lowest eigenvalues and a new set of C is generated from the

1p
Roothaan Equations. This process is iterated until the C are self-

1?

consistent. The similarity of this process to that of the Hartree-Fock

SC? method has lead to this procedure being called the LCAO-SCF method.

If m is made large enough to form a complete set, the di generated by the

LCAO-SCF method are exactly those determined by the Hartree-Fock SCF

method. However the set m is usually truncated in practice to conform

with computer limitations. Again this loss in generality is somewhat
overcome by the subsequent parametrization.

Most of the molecular properties of interest here are related to the
ground and the lowest excited states of these molecules. The excited state
properties are primarily associated with "fl" electrons. All other elec-
trons in such molecules are called "c” electrons.

Assumption 9: The‘fi-electrons are delocalized over the whole molecule
while thetrbelectrons are associated about particular
nuclear centers.

For all molecules investigated here the‘W-electrons M0 are constructed

from 2p? AD. Physical intuition indicates that the delocalized‘fiLelec-

trons are more susceptible than the localized¢rbelectrons to physical
processes such as electronic excitation. For this reason

Assumption 10: 2:11" Separability. The wavefunction‘f’is separable into a

product of two functions (2) (W) both of which are

20

antisymmetric with respect to themselves and each other.
Each of these functions is represented by Slater determinants and is
normalized to unity.

[(zfilu, =f(1r)‘.\«r..~ = \ (42)

One of the outcomes of these conditions is that Equation 1 yields

EA _.._ E1 + E1? where (63)
E2 : [GS—V32 (2} A”): and ET!“ =f(m*qv (In Ann“. (44)

A further restriction is placed on these functions.

Assumption 11: The spin orbitals from which the fland U functions are
constructed form a disjunctive set. That is electrons are
restricted to one function or the other. Each spin orbital
belongs to one set or the other but not to both.

These assumptions become more plausible provided:

Assumption 12: The molecules investigated are considered planar.

The molecule is thus pictured as consisting of nuclei and (r-electrons

which form a molecular core of planar configuration. This core is

enveloped by a 1T-electron cloud above and below the plane. Foresight
about some of the properties of the observables to be calculated restricts
the type of A0 which form the 2 and WT functions.

Assumption 13: The I; are symmetric and the I“- are antisymmetric with
respect to reflection through the plane of the molecule.

In order to carry out the variational procedures outlined above

Assumption 14: The leavefunction must be the same for all states of the

molecule.

21

That is, to carry out SCF procedures or to calculate any observable
based only on the properties of the1TMO requires that the Z.MD contrib-
ute the same aspect to all molecular states. This means that E; in
Equation 43 is constant and the LCI SCF procedures effect only Efip

With these restrictions the Hamiltonian has become

 

xx. no /\ "‘I"“ 2
HF ("2). . ' n?) = Z. k core (L) + a. ‘r. e... V‘ \ (45)
i. L #4, " 1
/\ 513' a. K t a A
. ~ - V. ~ e _

The term 9;(r1) represents the repulsion on the ith fiLelectron due to
the average of thetrbelectrons. It is still difficult to actually per-
form calculations due to the form of the Hamiltonian. This operator
still causes analytical problems when applied to specific molecules. To
overcome this difficulty the Hamiltonian must be molded into a more
tractable form. In addition it would advantageous to build into it some
means for overcoming the lack of generality caused by the above restric-
tions!

Assumption 15: The Goeppert-Mayer-Sklar (GMS) approximation is used in

formulating It? .
core

/\
h

into a form which has correlates
core

This approximation transforms

based on physical parameters and is outlined as follows. From Equation

46

 

’v‘m =<¢z<n ’“v—Efigrm‘ M ----~*n.w>=IPz<rnr‘fr.-.H* «n

Thus V£(rL) acts as an electrostatic potential due to the a’electrons.

22

 

It is assumed that K
FEM = E pm) <48)
8° A K a
vim = Z. MUM: m A“ (49)

/\ . “a «I K e“1
and \“core 0) = -;sz Vi. + EJL/LU) " Zd SONRJ] ‘r- P“ A" (50)
The term in the brackets is essentially related to the point charge of
a given nucleus together with its U-electrons and is given the symbol
4

-z . If [0,:(r) is due to all electrons (a'and‘fi) belonging to atom at

an identity may be stated:

fl. (r) 2 PIM - Z“ ’B’Ar) (51)
with the restrictions
Rim = Par-RA)
[PHfiAr = 2., (52)
f’flUHV = I
In general‘qxr) is not spherically symmetric.

a K a
Then’h‘cmu) = — 3‘... V.“ + gums-A u... RA] ,3.“ Av
K
__ ; z‘fflmfi A.«

K K
:4?th + 2 Gm 4% 2“ 4r: m <5“

 

 

(53)

When this equation is written in terms of the basis A0

" K . K ..
(IP‘kcore\I$>=\nP%_—;TP$ +éuP$—Z‘Z 1".‘7f‘ (55)

since 7"“) : 1:0,)

20.1 = ijm 1%??- LM 4..
1‘01» :(IPU‘J 1‘“)

ca (56)

IV‘Val

 

< I? ('3‘ 71.01.)

 

 

 

I‘(VJI‘(F)> 5- 7rd?!

23
It is possible to explicitly include the effects of molecular substi-
tuents such as hydroxyl, amino, or methyl groups in Equation 55. However
the molecules investigated in this thesis do not contain substituents
other than hydrogen so these terms will not be included. The explicit
effects of the hydrogens will also be neglected here.

Intuitively it seems plausible that most of the structural features
of molecules are due to nearest neighbor interactions between the atoms
which constitute the molecule. For this reason
Assumption 16: The tight-binding approximation is applied to the Hamil-

tonian.

S ecificall h -—>h Q
p y Pq Pq Pq

whereQ -8 +14
Pq Pq Pq

and Hpq i- 1 when p and q are nearest neighbors.

- 0 otherwise.

H is known as the molecular "topological matrix." With this restric-
Pq
tion

r it '< ‘
(57)

K
‘F ‘Ir ‘.b :5 4. P4
_
where k refers to indices of atoms directly bonded to the p-th‘fl-atom.

k
The terms EU are quite negligible compared to Up and I!“ . If a
(1..) 99 PH pq
substitution is made

A -'r +11p
PP PP PP

and (58)
A - T +-UP + Uq
Pq pq Pq Pq

24

then Zuh *E a S K 4
hpq =LAu-p +(4n rm ,, 2 73"?“ P7, +[Ar‘b“§?- 7rdz¢1Mrb(59)

The term App is equivalent to the experimentally determined electron
affinity of the p-th‘fi-atom. The electron affinity in turn is related
to the p-th‘fi-atom ionization potential and the coulomb repulsion of two
electrons on atom p:

A __+ -1 + zp’a‘ (60)

PP P PP
where 2D is the effective core charge of the p-th‘fikatom. The term Ap
represents an empirical property of the bond between atoms p and q. q
Following the Huckel theory nomenclature for such a bond property it is
called the resonance integral fi;q.

Observation of the Hamiltonian, e.g. Equation 27 or 34, indicates
there are a large number of electron repulsion integrals to evaluate.
However the number and complexity of such evaluation can be considerably
reduced by recalling that A0 are being used and that the overlap charac-
ter of the 1TAO should be very small if they lie on different atoms.

Assumption 1?: The MD are constructed from A0 for which the Zero

Differential Overlap (ZDO) approximation may be used.

 

That is:
1901‘) 1‘50”” :. O for p 1‘ q (61)
Thus ea 2
<Iv('e\1.<':\\"‘“"\a - m \Imsu»=<x.<rar.tv.\ mix..<»ax.o.x>s.s.:62>
“P1.“ z‘ 71"": S!" 3155 2 7W5

With this approximation
Jh r<
P at
hpq-[-IP + Z TPP +&)urfa “Zi 7‘94] SP% +fiP$ (63)

25

The ZDO approximation may also be applied to the‘fl-electron repul-
sion terms in the Hamiltonian. To determine what integrals arise in
the actual calculations Equation 29 is the basic equation. Its terms
are delineated by multiplying on the left bqu and integrating. In

terms of the AO

«um ¢a> = §C1.§ct,<1.\$\x,> = 3C}.§Ct,[<rs\’¢....\x,>

 

+§{z<x.¢,\fi‘ I, 49) ~09» Mfiifil 4’;%>}] = (6a)
$492935” Sign: Cirgcés(<IPIr-\F§ql\lt 15>
1‘12“? Ir‘l‘gfihsIQH
If the substitution Pm - ZECLC“ (65)
is made mm m m
PM - 2% Cir, CL%[L\P% +§ Pr5(7‘rr<bs “ {3.1 lanai] (66)

The term in the brackets is indentically the qu stated in Equation 34.

When ZDO approximation is applied there are two cases:

m
Case 1: p-qF -h + P 7‘ +1/2P 1‘
pp pp 21;? tr pr pp 99
(67)
Case 2: pfq P -h -1/2P 7‘
Pq Pq pq pq
The Hamiltonian is now expressed in terms of quantities such as
ionization potentials, electron affinities and coulomb integrals which
are the properties of atoms and bonds and should be invariant between
molecules. These parameters are somewhat amenable to experimental
investigation so their values should be physically reasonable. Once
they have been determined the Hartree-Fock operator is determined and

the MD are generated. These MD are then used to calculate the observ-

ables being examined. Since the operator which generated the MO is due

26
to a particular set of parameters, any observables calculated must be
properties of the molecule under the same conditions.
Assumption 18: The parameters and calculated observables are evaluated
for molecules in an equilibrium geometry configuration.

The nuclei are thus assumed to be in their most probable position for
all calculations. A result of this assumption is that the calculated
transitions are vertical. This may not correspond to the 0-0 transition.

These are the assumptions and conditions for making the calcula-
tions. Due to the impossibility of analytically solving the Schrodinger
Equation exactly except in elementary instances, a formalism has been
derived from the quantum mechanical apparatus using a successive set
of assumptions. This formalism allows the use of semiempirical paramr
eters to generate integrals associated with molecular properties.
With these assumptions the question still remains: Are these calcula-
tions valid? The most succinct answer is that they are valid if they
work in predicting observables which correlate quite well 1
with experimental facts. This is the crux of any theoretical investi-

gation.

27

111. Format for the Calculations

The calculations were performed using the CDC 3600 computer at
Michigan State University. They were controlled with a program
conceived and written by Petr Hochmann . The program was designed to be
as general and flexible as possible. This is achieved by the program's
multiswitch character which allows a variety of one-electron effective
Hamiltonians, HO, input parameter forms and outputs to be incorporated
or formulated. New formulations for the Hamiltonian or parameters can
be inserted at will. Only those procedures and forms used in the
calculations for the molecules studied here will be presented. Other
expressions are included in reference 4, but they do not constitute a
complete set. Any user of the program may introduce procedures with a
minimum of effort.

For these calculations the set of AO used was restricted to the 2p“
orbitals of the atoms contributing to the’F-system. All molecules for
which calculations were performed were planar conjugated organic. Each
atom in these molecules contributes at least one‘W-electron (recall the
effects due to hydrogens are neglected), Thus the number of AO (m)
comprising the molecular fi-system is identical with the number of nuclei
(k) in the molecule so for the previous equations m - k.

It is also instructive to note from Equation 51 that ?Q(r) is
functionally the complement ofjf§(r) based on a spherical electronic
charge distribution for the neutral atom. Since [04(r) is due to the 2'“
charge distribution only, ?Q(r) is functionally an A0 on thect-th atom.

With these further restrictions Equation 68 becomes

k
r -h -1/2P'a‘ +g E? 7*
pqrrrrP

pq Pq vs vs (69)

28

where
k k r
h -[—1 +2”? +20 —Zza’ S
Pq P PP (k) pp r r pq + Pq
(70)
and
a)
P - 22c c (71)
Pq 1 1p 1q
The terms are defined:
I - the'fi-electron ionization potential for the p-th'fi-atom in the spzfi
p valence state.
2p - the effective charge of the core of the p-th ‘W-atom.
pp - the one-center two-electron repulsion integral on the p-th‘fi-axom.
2r - the two-center two-electron repulsion integral between‘fi-AO on
Pq the p-th and q-th‘fiLatoms.
U(k)- the two-center penetration integral between the TT—AO on the p-th

PP ’W-atom and the spherical charge distribution on the k-th‘W—atom
whose total charge is electrically neutral.

k - denotes a summation over indices of the’fi-atoms bonded directly to
( ) the p-th fi-atom.

F3 - the resonance integral between the p-th and the q-th‘fihatnms,

P - for p - q : the‘fi-electron charge density on the p-th‘fi-atom.

for p f q : the Coulson bond order for the bond between the p-th
and the q-th fLatoms.

The calculations were initiated by ascertaining the types of atoms
comprising each molecule and their relative positions. Three atomic
types were used in these calculations:

1) spzfi‘hybridized C which contributes one electron to the‘fi;system.

2
2) sp‘W hybridized N which contributes one electron to the‘fi-system
(pyridinic type).

29

3) spzfi hybridized N which contributes two electrons to the1T-system
(pyrrolic type).

From this topological arrangement a Huckel type calculation was carried
out to generate an initial set of M0 to be used in the subsequent Hartree-
Fock SCF calculations. The number of MO included in a given set was
regulated by the number of‘W-AO contributed by the given molecule. Exam-
ination of the Secular Determinant (Equation 41) indicates that the number
of HO must be identical with the number of AO comprising the‘fi-electron
system, i - k. The Hartree-Fock calculations were taken over these 1 MO,
1Jof which formed a closed shell of N - 21>e1ectrons. By the Aufbau
Principle these‘UMO were the ones of lowest energy leaving 1 minus 1)
"virtual" MO. After self-consistency was achieved the following items
were calculated:

a) Molecular ionization potentials.

b) Molecular electron affinities.

c) Ground state‘W-electron charge densities and bond orders.

d) Bond lengths in the molecular ground state.

e) Ground state molecular permanent dipole moment magnitude intensity
and orientation.

A configuration interaction calculation was then carried out using
monoexcited configurations. These configurations were formed by promo-
ting an electron from one of the-aground state orbitals to one of the i-1)
"virtual" orbitals. The following considerations were used in performing
the configuration interaction calculations. The MD are arranged in the
order 1...:D....i. Specific orbitals are labeled:

(filled) (unfilled)
1315;», m<J_<_1

(72)
lfgksfig ar<L 5:1

30

To clarify the development of configurations the Slater Determinant,

Equation 6, is rewritten:

Ax
\‘J 'AT:1¢1

(73)
/\
where A is the antisymmetrization operator given by

p/\
A ' 20-1) P (74)

For the ground state the Slater Determinant becomes

Ly - 2TH - \(o)> (75)

° i-l
Upon excitation an electron is annihilated from one of the ground state
orbitals and an electron is created in one of the heretofore virtual

orbitals. This leads to expressions of the type

|(I_J)>:. ATtd’, ¢TT¢ (76)

L3 Iol

However each MO is doubly occupied in the ground state. Since either

electron may be excited, excitation is given by the expression

‘71—:l(.11,13)>+V-J3l(a12-\,u-n>z \|,(:,:n> (77)

for the singlet excited state and

\lJ—i" 1(«11 lI)>- “MRI \ 231- \)>:-_‘3 (I I)> (78)

for the triplet excited state.(21, 2J) may be considered as denoting
spin-t and (ZI-l, 2.1-1) may be considered as denoting spin ,6. The right
hand member of these equations is expressed in the codetor notation (the
spin projection quantum number has been specifically omitted in these

formulations). In codetor notation the molecular ground state is given

31

by '1, (08> (for an explanation of codetors see reference 3, p. 204).
The matrix used in performing a configuration interaction analysis

is composed of elements with the form:

/\ figfit
<"(0fl H“’(°)>E Ho Z is “:3 PP?J(F_P‘b +L‘r3) (79)
<\,<o>\fi\s,<r,n> zeum a. so)

<5, (IJHmS’DWJ—D == {UH-3 51K ’3‘“ 5n *[UWL] <81)
+(3-S)[1KHIL] + HOSIK Sn SSS

ML] 2

(82)

3a,);
1':
gch FP‘; CL?)
stat *
,,
[NHKL] '2 5:,wa CI?) 7‘!“wa CL?)

After configuration interaction each‘fi‘electron wave function for singlet

and triplet states is

V95) = A2; 1|,(o)>g,s + EAf;j,\ s, (1,31) (83)

where
S is the multiplicity of the state.
«i is the sequence number for the state.

a
A12) and A5" are the configuration interaction coefficients.

IE denotes summation over all monoexcited configurations
CR3)included in the configuration interaction.

After this limited configuration interaction analysis the following
items were calculated:
f) Ground to singlet excited state transition energies,

oscillator strengths and transition moment magnitudes and

32
orientations.

g) Ground to triplet excited state transition energies.

h) Lowest triplet to higher triplet excited state transi-
tion energies,oscillator strengths and transition moment
magnitudes and orientations.

i) Excited state‘fi-electron charge densities and bond orders.

j) Excited state interatomic bond lengths.

k) Excited state molecular permanent dipole moment magnitudes
and orientation.

Formulas in the codetor basis for the transition momenta are given
in Reference 4. The excited state permanent dipole moments, charge
densities, and bond lengths were calculated from first and second order
density matrices. These matrices are also formulated in Reference 4
in the codetor basis.

In order to have confidence in the validity of these calculations it
was felt that as many observables should be calculated for as many mole-
cules with as little parametrization as possible. These observables
could then be checked against the corresponding experimental values. If
the calculations were of general utility there should be fairly good
agreement between these quantities. The procedure was to choose a small
set of "test" molecules, vary the parameters utilized in Equations 69 and
70 until the calculated observables agreed as well as possible with the
known experimental values and then to use these parameters in calculations
for a large group of similar molecules. This variation of parameters is
called "parameter optimization". If the calculations on the large group
evoked any major discrepancies the parameters were reoptimized until good

agreement was reached.

33

Optimization was achieved in three steps. It was first performed
for the hydrocarbons by Hochmann, et al? The next step was to optimize
the calculations for molecules containing pyridinic type nitrogens. The
test molecules for this series were pyridine, pyrazine, and 1,5 naphthy-
ridine. Finally optimization was carried out for molecules containing
pyrrolic type nitrogens using the test molecules of pyrrole, carbazole,
and imidazole. The test molecules were chosen on the basis of the amount
of experimental data available, the range a particular quantity had from
one molecule to another, and the number of different experimental
quantities available for a particular molecule. The optimized parameters
were then used for calculations of the indoles. The scheme was to use
the parameters optimized in previous procedures to perform calculations on
the molecules of interest here. Hopefully the results of these latter
calculations would not reflect any bias in parameter optimization.

Unfortunately it is evident that there is not much experimental
data available for nitrogen heterocyclics to use for optimization-espe-
cially when compared with the data available for hydrocarbons. Thus
there were fewer optimization checks which could be made for the Nitrogen
heterocyclics. When optimization had proceeded to the point that param-
eter adjustments adversely affected one observable more than another the
arbitrary choice was made to optimize the singlet transition energies as
well as possible. The other observables were optimized but their optimi-
zation was of secondary importance. Although technically feasible,
optimization was not attempted for molecules containing nearest neighbor,
i.e. contiguous nitrogens. Lack of time was the principal reason for this

omission.

34

The final values and equations used in evaluating the parameters
are listed in Table 1. It must be emphasized that this set of parameters
may be superseded at any time by an improved set which uses a different
group of constants or, more probably, new formulations. The efficacy of
this parametrization is indicated in Table 2 which compares the experi-
mental and calculated results for the nitrogen heterocyclic test molecules.
Overall the calculations vs. experimental data correlation is remarkably
good considering the breadth of comparisons being attempted. This gives
confidence for proceeding to perform correlations and predictions of the
indoles. However any calculated results must be tempered with caution

since there is no absolute guarantee for their accuracy.

35

Table 1. Semiempirical Parameters,

 

 

 

 

 

 

 

 

Note: N denotes Pyridinic type Nitrogen
NII denotes Pyrrolic type Nitrogen
TypelAtom(s) Value Units
1P C 9.84 eV
NI 12.57
NII 20.40
Zp C 1 le‘
NI 1
’rpp C 11.97 eV
NI 15.44
NII 15.44
sz§pq C-C -2.42(2.76068-1.26033 RCC) eV
0
C-NI -2.40(2.44950-1.08333 RCNI)
R =1.458-0.180 P
- - o a 0 - a
C NII 2 31(2 5 597 1 12554 RCNII)
R =1.458-0.180 P )
CNII CNII
Note: N

 

I - N
neighbofs

was not parameterized since they aren't nearest

 

 

Table 1 (cont'd.)

36

 

 

 

 

 

Type Atom(s) Value Units
"}q c-c 6.91-3.99(RCC-1.397) for n52.o ev
35e2/(R2 +2 00070§+1/(R +0 0% f R>2 0 °
CC 0 CC 0 Of a R in A
c-NI 7.51-3.99(R -1.338) for R<2.0
CN ‘-
2 2 .51 2 2?
%e /(RCN +1.68114) +1/(R +0.01389) or R>2.0
I CNI
C-NII 5.69-3.99(RCNII-1.338) for 352.0
2 2 g,
35e /(RCNH+1.68114P+1/(RCNH+0.01389) for [92.0
2 2 a, 2
NI-NI %e /(RNINI+1.17872) +1/(RNINI+0.0#5
NI-NII 35e2/(RN2N +1.17872;5+1/(R '+0.0§‘
2 12 II 5 I II a
N -N %e /( +1.17872) +1/(R +0.0)
11 II RNIINII NIINII
Note: N-N must be >2.0 A apart since they can only be nonnearest

Pq

neighbors.
R take the corresponding values shown for RP infl9pq if they are
nearest neighbors, otherwise they are t e geometrical_§qq.

 

 

(k)
qu

 

 

0.07(-183.005+131.714 RCC)
RCC=1.517-0.18O PCC

0.03(-248.760+186.667 RCN )
RCNI-1.458-0.180 PCNI)

0.03(-248.760+186.667 RCN )
RCNII=1.458-O.18O PCNII I
0.14(-52.520+40 RN 0)
RNI§1.458-0.180 PNic

0.14(-52.520+40 RN c)

R = - I
N113 1.458 0.180 iNIIC

 

6%

 

 

37

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 2. Comparison of Experimental and Calculated Values for Test
Molecules,
Pyridine T ‘3
§
Quantity Units Experimental Calc. Value
Value

Bond length c-c K 1.39, 1.40 1.40

" C-N 1.34 1.34
Ionization Potential (1) eV 9.23, 9.26 9.28

9.31, 9.28

" " (2) 9.51 ? 9.62
Transition Energy (1) K 2511, 2516 2562
" " (2) 1980, 1976 1967

" " (3) 1754, 1782 1795
Transition Angle (1) degrees from 90 90
" " (2) x axis 0 0

" " (3) 90 90
Oscillator Strength (1) 0.032 0.009
" " (2) 0.122 0.048

" " (3) 0.660 1.111

1,5 - Naphthyridine

Transition Energy (1) K 3080 3063
" " (2) 2571 2500

" " (3) 2060 2017
Oscillator Strength (1) 0.13 0.063
" " (2) 0.094 0.154

" " (3) 1.08 2.026

 

 

38

 

 

 

 

 

 

 

 

Table 2 (cont'd.)
Pyrazine 1‘ 2
'f
E a -» 2
Id
1
Quantity Units Experimental Ca1c.Va1ue
Value
Bond length C-C A 1.378 1.399
" " C-N 1.334 1.337
Ionization Potential (1) eV 9.36, 9.21 9.32
, 9.29
" " (2) 9.51 2, 10.11 10.03
Transition Energy (1) A 2601, 2610, 2580
2580, 2600
" " (2) 1943, 1965, 1975
1966, 1970
" " (3) 1811, 1846 1794
Fransition Angle (1) degrees from 90 90
" " (2) x axis 0 0
II II (3) 90 90
Oscillator Strength (1) 0.08 0.036
" " (2) 0.119 0.104
" " (3) - 1.018
4 .3
Pyrrole
5 < >52
N I
Bond Length 1-2 A 1.42, 1.383 1.354
" " 2-3 1.35, 1.371 1.388
" " 3-4 1.44, 1.429 1.408
Ionization Potential (1) eV 8.20 7.75
Transition Energy (1) A 2075, 2110, 2078
2113, 2360
" " (2) 1829 1984 week
" ” (3) 1725, 1747, 1791
1717
" " (4) 1734

 

 

 

 

 

 

Table 2 (Cont'd.)

39

 

 

 

 

 

 

 

 

 

 

 

 

Carbazole . .
N
11
Quantity Units Experimental Calc. Value
Value
Transition Energy (1) A 3397, 3367, 3357
3259, 3306,
" " (2) 2933, 2938, 2914
2890, 2907,
" " (3) 2550, 2572 2729
Oscillator Strength (1) 0.04 0.07
" (2) 0.16 0.13
" " (3) 0.25 0.18
.3 2
Imidazole
4 1
N
5H
Bond Length 1 2' A 1.33 1.33
" " 2 3 1.38 1.35
" " 3 4 1.36 1.39
" " 4 5 1.37 1.35
" " 5 1 1.35 1.35
Transition Energy (1) A 2065 2082 *

 

 

Notes:
type.

2)When more than one experimental value is listed, all values

have been reported in the literature.
3)Numbers in parentheses denote the sequence number of the

observable ordered from lower to higher energy.

1'

1)All transitions and oscillator strengths are of the singlet

CHAPTER III

EXPERIMENTAL PROCEDURES

Although there have been a variety of experimental investigations
made of the indoles there still remain a large amount of data to be
amassed. Advances in experimental techniques allow investigations of
hitherto unstudied properties as well as duplication of prior studies
made under less stringent conditions. In this investigation an assault
was made on the mountain of untapped data. Since indole and to a lesser
extent purine have been most extensively studied. Our experimental
investigations were concentrated primarily on the aza-indoles and in
particular benzimidazole. In the following paragraphs are stated the
apparatus, techniques, and materials utilized. The results of these
experimental ventures are incorporated in the appropriate discussion
section of each molecules's spectral properties.

I. Vapor Absorption Spectra.

All vapor absorption spectra were run on a Cary 15. Wavelength
calibration was achieved by observing the position of the various Hg
emission lines from a low pressure mercury arc. These lines were
recorded at the slowest scan speed and most expanded wavelength scale
(10 A717 mm) available. All vapor absorption spectra obtained for the
purpose of determining absorption peak positions were recorded under
identical conditions as the mercury emission lines. Wavelength
corrections based on these lines were then applied to the absorption
spectra. Wavelengths to five significant figures were read on the

expanded wavelength scale.

40

41

The spectra were taken using a modified 10 cm absorption cell with
quartz windows. The modification consisted of attaching a vacuum stop-
cock to the only opening of the absorption cell. A ball joint fitting
was installed on the exhaust end of the stopcock for attachment to a
vacuum line or to a second vacuum stopcock which in turn was attached
to a vacuum line. The modified cell was wrapped with nichrome wire
which was secured with silicone cement. The wire was also wound around
the inlet to the cell and extended to the stopcock itself. The ends of
the nichrome wire were attached to a Variac. When spectra were to be
obtained a few milligrams of the molecule to be studied were introduced
directly into the cell. A vacuum of about 3x10”6 Torr was obtained in
the vacuum line and the stopcock closed. The Variac was slowly turned
up until a spectrum could be obtained with most peaks occurring at 2/3
full scale on the normal OD scale setting. The nichrome wire was
wrapped such that there were more turns at the ends than in the center
of the cell barrel. This insured that more heat was available at the
quartz windows which retarded crystallization of the sample on the
windows. No fogging, crystallization or clouding of the windows was ever
noted during the course of these experiments. All vapor absorption
spectra were taken with ambient air in the reference compartment. The
instrument sensitivity was increased as much as possible consonant with
a tolerable signal-to-noise ratio. This allowed a very narrow split
width to be used (0.01 mm or less) which in turn increased the spectral
resolution as much as possible. With this setup it was possible to
resolve spectral features lying 3 cm.-1 from each other at 35000 cmfl.

All molecules for which vapor absorption spectra were run were also

subjected to an investigation of the temperature-dependent behavior of

42

spectral features. This was accomplished by placing a few milligrams of
the molecule under investigation in a reservoir sidearm off the exhaust
side of the cell stopcock, evacuating this sidearm, opening the cell
stopcock to evacuate the cell to 3x10- Torr, closing the stopcock
closest the vacuum line, allowing the material to equilibrate for a
few minutes at room temperature and then closing the cell stopcock.
Depending on the vapor pressure of the molecule under investigation,
there was supposedly a low but unknown concentration of molecules in
the vapor phase trapped in the absorption cell. Absorption spectra
were taken at a successive series of temperatures as controlled by the
Variac. Temperature aqlilibrstion was attained at each Variac setting
before the spectrum was run. The first spectrum in a temperature series
was run when the first spectral features appeared using the expanded OD
scale. Spectra were then recorded at successively higher temperatures
with the final run usually taken at a temperature of about 150°C.
11, Solution Absorption Spectra.

The solution absorption spectra were also run with the Cary 15. In
311 cases 1 cm pathlength cells were used and the reference cell always
contained the solvent alone. However when low temperature spectra were

recorded the reference was the solvent at room temperature.

III. Emission Spectra

 

All fluorescence and phosphorescence spectra were obtained with the
apparatus illustrated in Figure 3. A 1000 W xenon high pressure lamp

was used as the excitation light source.

.tssstmae< voodomo:w5:A .m mwswwm

 

 

sooeooox

 

 

 

 

tszzssa

s.-.osa

 

 

 

43

 

 

I
|
I
J

 

 

 

 

coeoEoEooco:

someosoJ
.4
If
IH

 

cot—esoxu

 

 

 

AL

 

 

333 3.5.5

.23.. 2s:

 

 

 

 

 

 

 

 

 

 

 

a.em:m
.osom

 

 

_
.
.
0.9.3» _
- .
_ a
m . .
1. 2.1.... s a
.11. IL
5.2320235.

 

c3325

 

 

 

44

The lamp's voltage was generated by a Christie power supply.
Excitation wavelengths were selected by a 85L 10 cm grating blazed at
3000 A in a 85L 500 mm monochrometer. UV bandpass filters were often
used to eliminate any stray light which was passed by the excitation
monochrometer. The excitation illumination was focused on the sample
whose emission was detected at a right angle relative to excitation.
The emission monochrometer was a Spex l700-II which utilized a 10 cm
BEL grating blazed at 5000 A. The emission spectrum was detected by an
EMI 955808 photomultiplier tube. The tube's high voltage was maintained
by a Fluke 4123 power supply which was normally operated at 1100 V.
Signals from the detected emission were fed to a PAR HRrB Lock-in
Amplifier whose reference was provided by a light chopper. This chopper
was located at position A when fluorescence alone was being studied or
at position B when both fluorescence and phosphorescence were to be
analyzed. Finally the amplified emission signal was displayed on a
Bristol's 12 in strip-chart recorder. All spectra reported here are
uncorrected for source, monochrometer, or photomultiplier tube response
variation with wavelength.

Some of the solution samples were degassed prior to their use in
luminescence studies. This procedure consisted of attaching the sample
tube together with a closed stopcock to the vacuum line, freezing the
tube with liquid nitrogen, opening the stopcock to evacuate the tube

above the frozen sample to a pressure of about 3x10-6

Torr, closing the
stopcock, allowing the sample to thaw, refreezing, evacuating the tube
and continuing this freeze-thaw cycling until the vacuum line ionization

gauge did not quiver when the stopcock was opened after a freeze. The

tube was then sealed off and used in the experiment.

45

IV. Solvents

Following is a list of the solvents used in this study and the
method of purification for each.
1. ‘flgtgg. Only doubly-distilled water was used.
2. Ethanol. 200 proof ethanol was placed in a flask and distilled
through a l m vacuum jacket column. The distillation rate was adjusted
such that a very slow rate (about 5 drops per minute) was maintained.
Distillation continued until the benzene-alcohol azeotrope was no longer
present as determined by an absorption spectrum of the distilled
alcohol in a 10 cm cell. That is, the characteristic benzene UV
absorption spectrum was no longer apparent. Ethanol was then distilled
and used as needed.
3. Ethyl ether. Commercial anhydrous ethyl ether was refluxed over
sodium ribbon. The ether was distilled through a 1 m column and used
as needed.
4. 3-methylgpentane. A modified version of the purification method of
Potts6 was used. Phillips Pure Grade 3-methylpentane (3MP) was shaken
for 30 minutes with a 50350 mixture of concentrated sulfuric acid and
concentrated nitric acid. It was then shaken 3 times for 30 minutes
each with concentrated sulfuric acid. This was followed with several

shakings using sodium carbonate solutions until the CO production

2
ceased. The 3MP was then shaken several times with water until the
water remained clear compared to the initial yellow color it attained
with the first shaking. After storing the 3MP overnight over anhydrous
calcium sulfate it was placed in a flask and sodium ribbon was added.

It was refluxed through a vacuum jacketed l m column and distilled for

use as needed. Passing the distillate through a 1 m column of activated

46

Silica Gel did not alter its absorption characteristics so this step was
not required in the purification process.
V. Experimentally Studied Molecules.

The means of purification is described below for each molecule.
Any changes in the UV absorption spectra after a purification step was
cause to perform another step.
1. Indole. - purchased from Aldrich was recrystallized once from an
alcohol-water mixture and subsequently recrystallized twice from petro-
leum ether. The resultant compound formed white shimmering plates.
2. 7-azaindole. - purchased from K 6 K Chemical Co. was recrystallized
twice from cyclohexane and formed white blocky crystals.
3. Benzimidazole. - was purchased from Aldrich and recrystallized
three times from hot water. It was then vacuum sublimed slowly for
two days. The material which had sublimed to the cold finger was then
analyzed by mass spectrometry. No heavier compounds were detected and
all lower molecular weight peaks could be accounted for by benzimidazole
subgroups. The purified material was white and formed needlelike
crystals when recrystallized from water.
4. 4-azabenzimidazole. - purchased from Aldrich was dissolved in
ethanol in an Erlenmeyer flask. This edution was placed on a hot plate
and the ethanol was allowed to boil off. Further heating forced sub-
limation of 4-azabenzimidazole onto the side of the flask leaving a
dark brown residue on the bottom. The sublimate was carefully scraped
from the side and the process was repeated three times. The resultant
product formed very long white needles.
5. Indazole. - purchased from Aldrich was recrystallized three times

from.water. The product formed white plates.

47
6. Benzotriazole. - purchased from Matheson Coleman & Bell was
recrystallized three times from water with the resultant product forming
long white needles.
7. 225135. - purchased from Aldrich was not subjected to any further

purification procedures.

CHAPTER IV

ANALYSIS OF SPECTRAL DATA FOR EACH MOLECULE

This chapter will be devoted to a discussion of the theoretical and
experimental UV spectral data available for indole, the azaindoles and
purine. Theoretical and experimental results performed by the author
will be incorporated in this discussion. These data will be correlated
to provide a unified picture of how the changes occurring in the elec-
tronic structure are related to spectral features. The perturbing
influence of environmental factors are powerful probes of these struc-
tures so much use will be made of them. Each molecule will be discussed
in turn with data correlation being limited to just that molecule.
Finally general predictions of spectral occurrences under heretofore
unstudied conditions will also be made based on the analysis of the
available data. The overall aim is to accumulate enough experience to
make as valid statements as possible about the electronic structures of

these molecules.

INDOLE
I. Xgpor Absorption Spectra.

The vapor absorption spectrum of indole has been obtained by Hollas7
and by El-Bayoumi, et. a1.8 Hollas assigns the 1Lb 0-0 transition at
35233.2 cm.-1 while El-Bayoumi, et. a1. assign it at 35263 cmfl. Our
observations place it at 35261 cm.l. Since Hollas' experimental
conditions were more exacting than ours we will assign the 1Lb O-O
transition at 35233.2 cmfl. Indeed, when we make this correction in

our data we can reproduce most of Hollas' data although we also see

structural features at higher and lower energies than he reports.

48

49

The lower energy features may be attributed to "hot bands" since we
obtained our data at higher temperatures than Hollas did. Observation
of the vapor spectra (Figure 4) indicates a region of absorption which
is rich in sharp vibrational structure (the long)\region). This region
seems to be superposed on a broad bandabsorption which has diffuse
character. Its maximum occurs at a shorter 21than that of the struc-
tured absorption. If one invokes knowledge about the absorption
spectrum of naphthalene which is isoelectronic with indole and uses the

9

Platt notation for absorption band monenclature, it appears that there

are two absorption bands. The ngfi—IA band has sharp vibrational

structure while the nge-IA is broad and has diffuse structure.

Although Hollas did not mention the 1

1

L;$—}A transition, El-Bayoumi,
et. a1. assigned the La 0-0 at 36000 cm"1 in the vapor by observing the
fluorescence and assuming a mirror image relationship between fluores-
cence and absorption. That is, one observes room temperature fluores-
cence in solution and locates its 0-0 position relative to its maximum.
The assumptions were made that this emission is from the 1La state and
that the separation of maximum and 0-0 are the same in absorption as

it is in fluorescence in solution. In this connection Konev108 did a
similar fluorescence-absorption analysis in cyclohexane at -196°C. He

determined the 11 0-0 to be at 34,820 enTl and the 1

b
His method of analysis consisted of finding matching peaks in absorption

La 0-0 at 34,800en'1.

and fluorescence spectra, assigning these to the lLb or 1La transitions,
and concluding that the 0-0 transition is midway between these peaks.

It should be pointed out that El-Bayoumi, et. a1. assign the 11.8 0-0
transition energetically above the 1Lb O-O transition while Konev

reverses their order but places them such that they almost coincide.

55.30QO 1839.8me uoem> mHomCH JV 3:me

.5. 253.65.

 

 

 

000 0.0 000 Os.“ 0.“ 000 0'“ 8d
4 d 4 E u a u
0.0
T \/ 1 N6
1 .91, 1335/ 1 ed .
. . .._ 1... do
I _ l 3
I __ L 3
1. . 2

 

 

 

 

 

0mm CON OFN owN

50

 

MJ002_ 3

51
These 1La O-O assignments should be considered as quite tenuous. There
is no guarantee that the vibrational contributions to the absorption and
emission transitions will be the same, i.e. the expressions for the
absorption and emission moments may not contain the same functional
vibronic character. This is especially true since these analyses were
taken from solution data and solvent-solute interactions are evident
but not accounted for in any specific manner. In addition it appears
likely that these transitions are not from single electronic states.
The discussion in the Fluorescence section for indole will amplify this
assertion. Again solvents may differentially perturb one state versus
another and attempts at correlating absorption and emission spectra may
be hazardous.

An analysis of the vibrational structure of this vapor spectrum is
difficult to carry out. A normal coordinate analysis has not been per-
formed and appears to be quite formidable since this molecule (as well
as all others included in this study) exhibits only planar symmetry.

The overlap of two distinct electronic transitions also complicates such
an analysis. Both transitions contain vibrational structure which
carries over to the neighboring transition thus forming a complex vibra-
tional pattern. However, visual clues such as groups of sharp vibra-
tional peaks offer suggestions for the assignment of vibronic
progressions or sequences. If one assumes the 1Lb O-O transition

(2834 A. on Figure 4) as the beginning of a sequence, another sequence
apparently begins at 2777 A or about 725 cm'1 above the first sequence.
Using this 725 cm"1 vibrational energy as a fundamental and successively
looking to the blue by thi amount one can observe further sequences

0
beginning at 2722, 2668, 2618, 2570 and 2522 A. Hollas also found this

52
sequence pattern and his analysis indicated this energy was most
probably attributable to a benzene ring vibration. In addition one may
observe the two diffuse peaks at 2534 and 2590 A which presumably
belong to the 1La transition. The energy separation for these peaks is
about 875 cmrl. Moving to the red by successive energy increments of
this amount one would place vibrations of the 1La of the transition at
2648, 2711, 2778, and possibly 2850. The spectrum of Figure 4 indicates
this is to be a plausible assignment. In addition another vibrational
member of this transition at 2462 A is evident and is 875 cm'1 to the
blue of the broad peak at 2534 A. Thus these assignments are consistent.
A consequence of this analysis is the indication that one member of the
1La transition (at 2850 A) lies beneath the 1Lb 0-0 transition (2834 A)
both energetically and in intensity.

To further investigate this spectrum it was decided that a "hot
band" analysis should be performed; that is, a study of the relative
vibrational peak heights as a function of temperature. The results are
shown in Figure 5 and are quite surprising. The most striking fact is

the growth of the 1

La absorption relative to the 1Lb absorption as the
temperature is increased. To our knowledge this phenomenon has not

been reported before. It cannot be explained simply on the basis of a
change in the vapor-solid equilibrium since any change in this equilib-
rium should result in similar alterations for both transitions according
to Beer's Law. It is obvious that the 1La transition is becoming more
allowed as the higher vibrational levels of the ground state become more
populated. A plausible explanation of this phenomenon is that the

nge—lA transition "borrows" its intensity from another electronic state

via vibronic coupling. The state from which it steals intensity is

0.0.

53

Temp ISO’ C

 

 

Temp ”8' C

 

 

 

 

' Temp 99‘ C
Temp 83' C
220 230 240 250 200 270 2.0 290
WAVELENGTH (um)
Figure 5. Indole Vapor Absorption Spectra as a Function

of Temperature.

54

apparently at a higher energy than the 1La state since the ngE-IA tran-
sition does not appear to lose intensity. The next higher transition
Se—lA) also exhibits an intensity increase as the temperature is
raised. If the intensity stealing is from a still higher state vib-

(18

ronic coupling may be a general trait encompassing several electronic
states. This coupling mechanism is not unlikely due to indole's lack
of symmetry.

It was also reported that naphthalene (isoelectronic with indole)
displayed an anomalous lLse—}A transition. This transition was found
to consist of two sets of bands with one set 10 times as intense as
the second. A surprising fact was that the former set was forbidden
and the latter set was allowed from symmetry arguments. Again vibronic
mixing was invoked to explain this phenomenon.11 These findings when
compared with those observed for indole led us to study the vapor
absorption spectrum of naphthalene at the temperature extremes employed

1

in the indole study. Again we found an enhancement of the La intensity

relative to that for the 1Lb transition although these results were more
ambiguous than those for indole. Thus this vibronic coupling mechanism
may be more ubiquitous than previously thought.
The hot band analysis of the (presumed) sequence beginning with

the 1Lb 0-0 transition (2834 A) indicated that indeed these 1Lb vibronic
transitions are superposed on a member of the 1La transition. This was

indicated by the fact that the whole intensity envelope of this sequence
slowly increased as the temperature was raised. This correlated quite

well with the intensity increase observed for the 1

La transition with
temperature. A possible explanation for these results is that a diffuse,
lowhintensity sequence belonging to the 1!.a 1A transition underlies

this 1Lb 0-0 sequence. The anomalous intensity behavior as a function

of temperature of the lLb O-O

55

sequence could be caused by intensity changes of hot bands belonging to
1
a diffuse 1La 0-0 sequence. This implies that the L 0-0 transition
a

l

is energetically close to the Lb O-O transition. Support for the

definite assignments of the 1La and 1

Lb 0-0 transitions would come if
a correlation of IR data with the vapor absorption spectrum were made.
Such a correlation has proven to be difficult.7
II. Polarization Measurements of Fluorescence Excitation

By analogy with naphthalene the 1Lb and 1La transitions are
expected to be long axis and short axis polarized respectively and both
transitions should occur in the plane of the molecule. Several experi-
mental observations have been carried out attempting to assess the
relative direction of transition polarizations in indole. It should be
noted that no crystal absorption measurements have been reported yet.

The first experiments were done by Weber12

on 2xlO'4M propylene
glycol solutions at -70°C. His technique was to run a polarization
excitation spectrum of indole fluorescence which was observed at 340
nm. He found two polarization maxima at 270 and 298 nm and a minimum
at 290 nm. The minimum had a positive polarization value however.
These features were ascribed to two superposed transitions in the
excitation spectrum, namely the 1La and 1Lb transitions. One of these
transitions had a "negative polarization" relative to the other and
accounted for the polarization minimum observed at 290 nm. This
transition was ascribed as 1Lb.

Zimmermann and celleagues13’14 did fluorescence polarization
excitation measurements on ethanol solutions at ~180’C. (no concen-

trations reported) with results similar to those of Weber. They gave

no interpretations of their results other than stating that two

56
transitions were responsible for the observed polarization spectra and

that the 0-0 bands for the 1La and 1Lb transitions are at 34,000 cm,"1

(294 nm) and 34,500 cm.-1 (290 nm) respectively.

10b’15’16 did another set of fluorescence polari-

Konev and coworkers
zation excitation experiments with tryptophan imbedded in stretched
polyvinyl alcohol films at room temperature. This technique was employed
to preferentially orient the molecules in-plane and along a preferred
axis in that plane. The fluorescence was observed at 330 nm. Their
results also were interpreted as a superposition of the 1Lb transition

on the 1L transition with 1

a
state that these results indicate the

Lb transition maximum at 289 nm. They also
"1Lb oscillator is oriented at
almost a right angle to the 1La oscillator".

Song and Kurtin17

also did fluorescence polarization excitation
experiments on lO'SM indole in 9:1 glycerol-methanol at 263°K. They
monitored the fluorescence at 345 nm and observed minima in the polar-
ization excitation spectrum at 240 and 293 nm. The 293 nm minimum was
atttibuted to the 1Lb O-O transition superposed on the 1La transition
with 1'La state emitting at the monitored wavelength. They interpret
their polarization excitation data as reflecting the 1La and 1LI)
transition to be 90' relative to each other. (However, I believe
their data doesn't warrant such a strong statement).

Finally Zuclich18 performed magnetophotoselection experiments on
lO-ZM akatole in ethanol solutions at 77‘K. That is, an EPR signal
was monitored as a function of polarized exciting light rotated relative
the permanent magnetic field impressed on the sample. He interpreted
the data to indicate the first two transitions were perpendicular to

each other.

57

An analysis of these polarization data reveal two features: First
the 1Lb transition is superposed on the 1La transition. Thus there are
some components of the 1La transition at lower as well as higher energies
than the 1Lb transition. However it should be remembered that these data
were taken in polar media which red-shifts the 1La transition more than
the 1Lb transition as will be shown in the following section. Second it
is premature to assert that the th and 1Lb transitions are perpendicular
to each other. If they were perpendicular some negative values for
polarization should be manifested when the 1L6$—}A transition is excited.

This effect is no doubt partially masked by the overlap of the 1La and

1Lb transitions.

III. Solugion Absorption Spectrg'

 

The most general feature of the absorption spectra in solution is
the loss of much of the vibrational structure seen in the vapor spectrum.
The general features of the spectrum remain in hydrocarbon solvents but
they are much more diffuse. In polar solvents the spectra are further

blurred. The oscillator strength also tends to decrease slightly as the

solvent polarity increases.19’2°

20

Chignell and Gratzer analyzed the absorption spectrum of indole

by adding small amounts of H-bonding solvent to isooctane solutions
(2.5xlO-4M) and observed the pattern shown in Figure 6. They separated

21 and that

this pattern into two effects: that due to H-bonding alone
due to solute-solvent interactions other than hydrogen bonding. These
results are summarized in Table 3a. Water is a notable exception to
their analysis; a blue shift of 150 cm"1 was observed. They showed

22

that all solvents listed but water follow the McRae Equation which

takes into account the dipole-induced dipole interactions between

.AONseNEesu use easeueeu ethane esteem cc Estseeem eeeesteeer see c: esseemm sesseom .e enemas

 

 

no uco>~om uuocH Noofi

 

 

Nooa ucm>~om wowmconum No

 

 

 

 

8
5

AHIEUV

COMuHmom

:oHumuOmom

z e
:LuommoucH f w I 4
r. ._ acm>~cm16u2~cm so 96 emcee—U .. 9.4 9..

ccwuumuoucH mewmccn12 so esp owcmsu 1 v;

 

9.4

59

 

 

Table 3a. Solvent Effects on Indole Absorption.
* -
A‘VH (cm-'1) A‘Os(cm-1) In")Tota1(cm 1)
Isooctane - -329.0 -329.0
Isopropyl -102 -338.0 -431.0
Alcohol
Ethyl Ether -104 -347.6 -433.0
Ethanol -112 -343.0 -441.0
Dioxane -122 -305.9 -451.0
Isobutyl -130 -327.9 -459.0
Alcohol
Water +150 ?

 

 

 

 

 

* these values are relative to isooctane

 

 

 

Table 3b: Indole: Comparable Maxims in 2 Solvents,
Methylcyclohexane EtOH
- -l
(A) cm 1 (EtOH - MCH) (A) . cm
2610.6 38305 -737 8 2661.8 37568
2662.9 37553 -677 2711.8 36876
S 2714.1 36845
S 2766.5 36147 -77 2772.4 36070
2791.8 35819 ~29 2794.1 35790
S 2804.1 35662
2871.2 34829 -50 2875.3 34779

 

 

 

 

 

 

Sflshoulder

 

 

60
solute and solvent.

Thus the excited state stabilization observed in most H-bonding
solvents is due to the sum of two effects. One effect is caused by the
inherent polar nature of such solvents and is explicitly considered in
the interactions formulated by the McRae Equation. That is, the transi-
tion moment and the excited state permanent dipole moment interact with
the solvent dipoles. If the excited state permanent dipole moment is
larger than the ground state permanent dipole moment the net effect is
a transition red-shift since the salvation energy for the solute has
increased. The data in Table 3a indicate this effect accounts for 3/4
of the stabilization energy. The other 1/4 is due to H—bonding. For
this effect the pyrrolic nitrogen's proton is hydrogen-bonded with a
solvent molecule. The electronic cnarge on the pyrrolic nitrogen
decreases as a result of excitations as will be shown later; thus the
pyrrolic hydrogen becomes more acidic and the hydrogen bond more stable
in the excited state. This will contribute to the observed red shifts.

El-Bayoumi, et.al.8 also presented data which can be analyzed for
solvent shifts. In this case an attempt was made to analyze both the
1LI) and the 1La transitions by observing the shifts of the lowest
energy vibronic 1Lb transition and the maximum in the 1La transition.
They showed that the observed red shifts for the 1La transition was
larger than for the 1Lb transition. Konev and coworkersz‘ also

observed the stronger red-shift of the 1

La transition (max at 276 nm)
than the 1Lb transition (279 and 287 nm) when n-butanol was added to
hexane.

Another possible solute-solvent interaction is between an acidic

hydrogen of the solvent (or a proton) with the‘fi-electron charge density

61
at the pyrrolic nitrogen. Such an interaction would cause a blue shift

of both the 1La and 1L transitions since the charge density at that

b
position decreases for both the 1La and 1Lb states. This explains the
blue shift observed in acidic media which will be discussed later as
well as the observed shift in water.

Table 3b lists data which I obtained in aspolar and a nonpolar
solvent. In this case the corresponding vibronic bands were tabulated
and compared. Again the bands belonging to the 1L transition are red-

a
shifted much more than those belonging to the 1L transition when one

b
shifts from a hydrocarbon to an H-bonding solvent. The spectrum in
the polar solvent also displays less vibrational structure than that in
a nonpolar solvent as was intimated in the opening paragraph of this
section. Finally the vibronic species associated with each electronic
transition are not shifted the same distance. This indicates that the
solvent differentially affects the vibronic species, i.e. the solvent-
solute interaction during excitation exhibit different couplings which
depend on the vibronic transition member.

Solvent effects on the absorption spectrum of indole lead to the
following conclusions: 1. The pyrrolic nitrogen should exhibit a
large fiFelectronic charge density in the ground state. 2. This charge
density decreases in the excited states making the pyrrolic hydrogen
more acidic, particularly in the 1La state. 3. The 1La and 1Lb state
permanent dipole moments should be greater than the ground state per-
manent dipole moment; this is specially true for the 1La state.

IV. Fluorescence Spectra.

Probably the most studied UV spectral characteristic of indole is

its fluorescence. There are a number of interesting aspects which are

in current controversy.

62

The most striking characteristics of the fluorescence is that it
appears to emanate from two excited states; that is, the 1La and 1Lb
states. The most convincing evidence for this trait comes from polari-
zation data. The first evidence and interpretation was by Zimmerman's

group.13’1‘

They found that by exciting at 275 nm, which is mostly the
ngh—CA transition, the fluorescence polarization was 0.13 at 290 nm.

The polarization spectrum then rose rather sharply to 0.30 at 340 nm where
it leveled off. Konev's group observed the same effect in polyvinyl
alcohol films15 (P - 0.10 to 0.25). Since they had measured the
fluorescence quantum yields as a function of exciting wavelength from

250 nm to 300 nm and observed no change they reasoned that a redistri-
bution of excitation energy occurred before fluorescence took place.
Subsequent experiments showed that fluorescence depolarization as an
increasing function of solute concentration was higher at longer
fluorescent wavelengths than at shorter wavelengths.25 This again is
indicative of emission from different states. Finally Song and Kurtin17
did further experiments to gain more evidence for this dual emission
process. When they excited at 285 nm, which is rich in nge-IA transi-
tion character, they observed a maximum.fluorescence polarization at
short wavelengths followed by a sharp decrease and tailing off of
polarization at longer wavelengths. Upon excitation at 265 nm the
opposite result was obtained, namely that polarization spectrum
observed by Zimmermann and by Konev. Others26 have also excited at

290 nm (lLb rich) and seen the high polarization at short wavelengths
with a subsequent tailing off at long wavelengths. These experiments

were all done in a variety of polar and nonpolar solvents at low

temperatures and indicate that the emission observed at short

63

wavelengths is due to the nge—1

l

A transition while that a longer wave-
lengths is due to the Lge—IA transition.

Another phenomenon which occurs in the fluorescence spectrum is
the extensive solvent effect on the shape and position of the spectrum
This phenomenon is somewhat related to both the analysis immediately
above and that in the absorption spectra section since the two emissions
appear to have quite different salvation properties. It was noted by

Konev24

that in n-hexane the fluorescence spectrum at room temperature
was not the anticipated mirror image of the absorption spectrum. The
relative intensities were quite dissimilar. As small amounts of
n-butanol were added the vibrational structure was lost and the 1La
transition underwent much more of a Stokes' shift than the 1L1) transi-
tion. Eventually the 1Lb transition disappeared as more butanol was
added and only a broad fluorescence band with its maximum at 330 nm was
all that remained. They reasoned 10c that solvent relaxation processes
caused the long wavelength emission and that the broad structureless
band was due to successively catching the emission at different stages
of solvent relaxation. Thus emission at the longest wavelength should
have a comparatively long T; since it was only seen after solvent

27 observed a similar shift and

relaxation had occurred. Matage, et. a1.
loss of vibrational structure when ethanol was added to n-hexane. In

this case the fluorescence maximum shift was from 300 to 335 am. They
attributed the shift to solvent "stabilization" of the 1La state rela-
tive to that of the 1Lb state. This occurred because the increase in

the dipole moment as a result of excitation was greater for the 1La

1

state than for the 1Lb state. Selective stabilization of the La state

compared with the 1Lb state occurs when a higher dielectronic constant

64

solvent is used. They reported that preliminary measurements of 7’
at different positions on the fluorescence spectrum show two decays
with that due to the lLé—+1A transition having the shortest component.
This however is opposite what one would expect from Konev's reasoning.

There have been several results of fluorescence spectra published
using various solvent conditions. Typical results are shown in Table 4.
In general it is observed that polar solvents destroy the vibrational
structure of fluorescence and cause a red shift. Several explanations of
these results have been proposed and the main ones will be outlined below.
The effects can be thought to occur as a result of three phenomena which
are not mutually exclusive: 1) the influence of the dielectric proper-
ties of the solvent on the excited state of the solute. 2) H—bonding
between solute and solvent, and 3) solvent relaxation during the lifetime
of the solute's excited state. There is no question that whatever
effect(s) cause the observed changes, such effects take place after the
molecule has been excited. This is borne out by the negligible effect
of different solvents on the absorption spectra. Moreover, viscous
polar solvents which don't have a chance to interact with the excited
solute molecules exhibit little red shift in emission.

There were several other observations made which weren't included
in Table 4. It was reported that there was lower fluorescence intensity
in water and in benzene than in cyclohexane, dioxane or ethanol, that
fluorescence was quenched in chloroform and in carbon tetrachloride, and
that acetone alone or as 12 in a cyclohexane or water solution quenches
fluorescence.12 When solvents having the same dielectric constant but
divergent viscosities were used (prOpylene glycol and methanol) the

fluorescence yield in the viscous solvent was 1.2 times that in the

65

 

 

 

Table 4. Indole Emission in Various Solvents.
Solvent Temp. pH Conc. Indole N-Methyl 3-diMethyl Refi
Indole Indole
n-hexane 20°C 5x1o'6u 288, 299. 5 322 24
n-pentane 20°C 5x10'5M 289 295 304 30
benzene RT 4.7x10'5M 305 28
ether 20°C 5x10'5m 303 311 30
p-dioxane 20°C 5x10'5M 310 318 30
n-butanol 20°C 5x10‘5M 326 328 345 30
E P A 770K 10-511 325 17
ethanol RT 10'411 335 27
ethylene RT 335 12
glycol
propylene RT 5x10'5M ’rf=3.9n sec 32
glycol
1120 20°C 5x1o‘6M 342 380 24
1120 RT peak-0.40 33
_3 ’Vf84.1n see
1120 RT 3x10 M ¢f=o.23 fif=0.38 34
H20 RT £82.7n sec 35
H20 RT 1.7 None None 40
H20 RT 15 400 40
<190.145
glycerol RT 346 26
1:1 ethylene 770K 310 34
glycol: H20 ¢f=0.6
1:1 ethylene RT 350 34

glycol: H20

 

 

 

 

 

 

 

 

 

66

nonviscous one. The respective lifetimes were 3.9 and 3.1 n sec32

indicating the quenching process in this case was due to internal
conversion.
Stryer had proposed an H-bonding mechanism to explain the quantum

yield ratios in D O and H20 solvents, i.e. ¢%(D20)/¢¥xH20). This ratio

2
was 1.29 for indole and 1.09 for n-methylindole‘s. He asserted that an
excited state proton exchange reaction, which was responsible for the

20, is faster in H20 than in D20. However the

proton-exchange mechanism cannot explain the data for tryptophan vs.

lower quantum yields in H

nemethyltryptophan where a sizeable isotope effect was observed in both
casesaé.

A controversy was evoked when Walker, et. al. proposed their
exciplex theory to explain the observed fluorescence shift in polar

solvents.3o’44

They observed the same red shifts in the fluorescence
spectra of indole, l-methylindole and 1,3-dimethylindole when small
quantities added were so small that the dielectric constant change could
not explain the spectral shifts. However, highly localized concentrae
tions of the polar solvent around each solute molecule could signifi-
cantly alter the local dielectric environment.20 But Lumry's group
found no correlations between the observed shift and the dielectric
constant of the polar solvent used. The only trend one could see was
that polar-protic solvents shifted the spectrum more than polar-
nonprotic solvents. They ruled out H-bonding since the red shifts
occurred for n-methyl indoles as well as indole. Using kinetic
arguments and plotting a long wavelength "second fluorescence band" vs.

concentration of polar solvent Walker, et. a1. invoked an "Exciplex"

between the excited solute molecule and polar solvent molecules to

67

explain the red shift. This was thought to be a reaction between the
excited solute and specific solvent molecules. For polar-nonprotic
solvents their exciplex stoichiometry was determined to be 1:1 and for
polar-protic solvents it was 1:2 solute:solvent molecules. This
"second fluorescence band" was their exciplex emission. It has some
charge-transfer character with the indoles acting as donors and the
solvent acting as an acceptor. Indirect evidence supporting this view
was that the observed loss of fluorescence intensity was proportional
to the dielectric constant of the polar solvent.3o Thus the loss could
be explained by a partial ionization mechanism.

Exciplex formation as an explanation of the spectral data was
challenged by Chopin and Wharton29 for two reasons: the Walker, et.a1.
data showed 1) the rate constant for exceiplex formation in polar-
protic solvents to be 100-600 times that in polor-nonprotic solvents,
and 2) the rate constant for exciplex formation for indole to be 10-15
times that for N-methyl indole in alcoholic solutions.29 Furthermore
Chopin and Wharton observed fluorescence in 1:1 isopentane:3-methylpen-
tane from room temperature down to the glass transition temperature of
the solvent. At that point the fluorescence was quenched and intense
phosphorescence appeared. However, if minute traces of water were
present the fluorescence quenching was not complete below the glass
transition temperature. If butanol were added to the IP:3MP mixture the
spectrum showed vibrational loss, red shifting and a fluorescence inten-
sity decrease at room temperature as Walker, et. a1. had reported. At
77’K both fluorescence and phosphorescence were present with both
seeming "to be superpositions of 2 emissions”. They explained their

results as well as those of Lumry and coworkers on the basis of a

68

phototautomerism to 2 H indole. Their scheme involved some intricate
kinetics with emission only occurring from an excited 2 H indole mole-
cule complexed with a solvent molecule and not from a 2 H indole mole-
cule alone. No reason was given why emission wouldn't come from an
uncomplexed molecule. In addition no rationale was given for the polar-
nonprotic solvents—induced red shift.

Another challenge to the exciplex formation came from Eisinger and
Navon who observed the fluorescence of indole and N-methylindoles as a

function of temperature in a 1:1 ethylene glycol-water solvent.34

They
found: 1) as the temperature is decreased the spectrum moves to the
blue and acquires structure with most of this change occurring around
the glass transition temperature of the solvent (this effect was also
reported in ethanol, iso-butanol, glycerol, and in polyvinyl alcohol‘é),
2) (LP decreases at higher tempratures but most of this decrease takes
place at temperatures above those at which the spectral shifts occur
and 3) a solvent isotope effect on the quantum yield is seen for all
molecules in deuterated solvents but this effect vanishes after ¢F

has leveled off as the temperature is lowered. Probably the most
important evidence against the exciplex formulation was the lack of

an isosbestic point formation as the spectral shifts were taking place
which indicates that a unique excited solute-solvent interaction
doesn't occur. Eisinger and Navon interpreted the spectral shifts as
due to progressive nondescript solvent reorientation effects. They
interpreted the quenching to be due to the potential surfaces of the
ground and excited states lying closer to each other after solvent

reorientation had occurred thus allowing tunnelling mechanisms to be

more efficient than before reorientation. Finally, using kinetic

69

arguments they reasoned that the isotope effect was due to solvent
perturbations (not including H-bonding) affecting the tunnelling

mechanisms. These mechanism are manifested by k This

non-radiative.
kinetic term.was found to be proportional to e'E‘IRT. E8 was not
found to be dependent on the isotope but the proportionality constant
was which further strengthens the perturbation argument.

All the above data show that the dipolar interaction between the
excited molecule and the solvent, as well as H-bonding, contribute to
the spectral shifts, quenching and lifetimes seen in the indoles.
However, dynamic solvent cage effects also have an important role in
the ultimate fate of the excitation energy. The magnitudes of these

effects are different for the 1 1

La and Lb states. Only after cognizance
and elucidation is made of these states' interplay with themselves and
their environment will the fundamental significance of solvent effects
be more fully understood.

An examination of Table 4 also shows that the cation and anion
species of indole exhibit different fluorescent properties than the
neutral molecule. No fluorescence has been reported from the cation,
i.e. total fluorescence quenching occurs at low pH. This is undoubt-
edly due to a change in the non-radiative transition rate, possibly
the intersystem crossing rate. The anion however does emit although
its fluorescence is weaker and at longer wavelengths than the neutral
species. Presumably the long wavelength emission is related to an
influence of the negative charge on the excitediT-electron system. For
the anion this charge resides on the pyrrolic nitrogen whose proton has

been removed. Long wavelength emission is also observed for neutral

methyl indoles. Again the methyl can exhibit an electron inductive

70

effect. Thus there is a correlation between the extra electronic
charge and the long wavelength emission.

Finally attention must be drawn to some unique results observed for
a particular set of indoles. Bridges and William did some fairly
extensive measurements on the effects of pH on the fluorescence of
indoles and substituted indoles.“o Their results are summarized in
Table 5. They noted that of all the methyl indoles, hydroxy indoles,
and indole itself only S-hydroxy indole showed fluorescence in acid
solutions. In addition Semethoxy and 5-phenoxy indole exhibited
similar fluorescence in acid solutions. However it was much weaker
than the fluorescence in neutral or basic solutions. It didn't occur
at 77'K. (It should also be noted here that indole did fluoresce at
77’K in ETOH + KOH at pH's where quenching occurred at room tempera-
ture;’va 11.) No solvent dependence was observed and the absorption
spectrum wasn't affected much by the pH change. Thus it was thought
that some excited state reaction or rearrangement was responsible for
this emission. This hypothesis was further substantiated by the
observation that the emission was at a longer wavelength than either the
neutral molecule or anion emission which indicates energy is lost in the
reaction or rearrangement.

All these fluorescence spectra data show that there is a configura-
tional change of the molecule before emission occurs. The type and amount
of change is dictated by the solvent and its effects on the excited
solute molecule. The spectral features therefore are a reflection of the
environment as well as of the uniqueness of that molecule. Vapor phase
fluorescence spectra would delineate the inherent molecular characteris-

tics. In addition these spectra would add information about the

71

Table 5. Effect of pH on Fluorescence of Indoles (Reference 40).

 

 

 

 

 

Molecule Emitter (abs) gluor)‘ Tf pKa pH range Stokes'
)max max of shift
max fluor (cmfl)
Indole and Cation 269 - - 1.7 - -
S-Me indole Neutral 270 355 0.46 14 3.3-11.0 8870
Anion 280 400 0.145 - 17.4 10710
N-methyl Dtcation 280 - - 1.6 - -
indole Cation 280 - - 2.5 - -
Neutral 282 350 0.47 - 4.3-16 7000
Cation c'293 - - - - -
5-methoxy Cation "“293 520 0.01 -O.7 15430
indole (Excited)
Neutral c"292 388 0.46 - 2.1-10.4 5070
Anion n4298 402 0.105 - 17.4 8230

 

 

 

 

 

 

 

 

 

72
apparent dual emission, i.e. whether it occurs in the vapor phase.
Also a study of methyl substituted derivatives of indole could yield
interesting information regarding dual emission.
V. Phosphorescence Spectra.

It was surprising to find that indole phosphorescence had not been
investigated too extensively. Heckman47 reported the indole phosphor-
escence spectrum in EPA and attempted to make a vibrational analysis of
it. He reported three vibrational sequences with 795 cm'l, 1235 cmfl,

l4

and 1585 cm."1 as fundamental components. Another report stated there

were vibrational sequences of 800 cmfl, 1200 cm.1 and 1600 cm’l.
Those who have reported spectral results have seen three peaks at

405, 430, and 455 mm with inflections at 417, 425, and 445 nm regardless

of solvent. The phosphorescence has been studied in EPA,17’47

alcohol finds "8

polyvinyl
ethanol,7 2:1 ethanol:ether, 9:1 methanol:ethanoll,'9
1:1 methylcyclopentane:methylcyclohexane,29 1:1 isopentance:3-methyl-
pentane,29 butanol,15 and glycerol.15 The nonspecificity of solvent
effect on phosphorescent emission is in direct contrast to the observa-
tions of fluorescence.

Reported values of phosphorescence lifetimes are: 2.4 sec in the

29

17
hydrocarbon solvents, 6.0 t 0.2 sec in the EPA, 6.3 f 0.02 sec in

the 2:1 ethanol:ether,48 7 sec in polyvinyl alcohol film.15 The phos-
phorescence to fluorescence ratio4§l¢% is also solvent sensitive.29’48
The effect of the status of the pyrrolic nitrogen's hydrogen on
phosphorescence intensity was discussed by Konev and coworkers}0d They
found that in hexane there was a large amount of phosphorescence and

little fluorescence; in methanol, ethanol, butanol, or ethylene glycol

there were about equal quantities of fluorescence and phosphorescence;

73

but in 0.1M NaOH the phosphorescence was practically nonexistent.
This indicates that the nitrogen's hydrogen is needed in the indole
structure for phosphorescence to occur but for the indole anion,
phosphorescence either does not occur or is very weak.

Probably the most interesting and informative data is that using
polarization techniques. In all reported cases both excitation and
phosphorescent emission polarization studies show that the triplet
transition moment is primarily out-of—plane as opposed to the moments

3

of singlet transitions. The emitting triplet is probably the La

52
state and there is an in-plane component of the phosphorescence tran-

sition which lies parallel to the 1Lb transition moment.

14’15’17 excited at wave-

Polarized phosphorescence spectral studies
lengths corresponding to the 1La and 1Lb transitions supports the
contention that the 3La state is the emitting triplet. Song and
Kurtin,17 using Raman and IR literature data, correlated the
650 t 100 cm-1 sequence observed in the phosphorescence spectrum‘with
either in-plane skeletal distortion of the ring, out-of-plane C-H
bending or possibly N-H bending. The 1600 t 100 cm"1 sequence could be
correlated with a ring stretching mode. They also suggested that spin

1 1

orbit coupling occurs between the 3La state and the Lb, Bb and other

states arising from of” and fink transitions.

The question arises whether the 3Lb state is involved in the
phosphorescence, i.e. that dual triplet emission is occurring similar
to dual singlet emission. This could explain the structure observed
in the phosphorescence polarization spectrum.15’l? In the phosphores-

cence polarization spectrum of indazole, two vibrational sequences with

perpendicular transition moments relative to each other was suggested.14

74

This seems to us to be a further indication for dual triplet emission.
Emission studies at very low temperatures (4’K) and as a function of
temperature could resolve this problem.
VI. Theoretical Calculations.

The results of these calculations are shown in Table 6 and Figure 7.
An assignment of the two lowest singlet transitions was made after con-
sidering the excited state's permanent dipole moments, the transition
moment orientations and the oscillator strengths. Naphthalene was

again used as the model for this assignment. Thus the 1

L state was
expected to have the lower permanent dipole moment and the 1Lb transi-
tion was expected to be preferentially oriented along the y-axis and
have the weaker oscillator strength compared with the 1La state and
transition. These three criteria all agree with the calculated
prediction that indole's first singlet transition is to the 1La state.
The first two transitions are calculated to lie energetically close to
each other. The actual calculated values do not coincide exactly with
the 0-0 transition values experimentally determined. However they do
not differ by more than 0.1 eV which is remarkable predictive agreement
for the calculations. One should recall that the calculations are for
vertical transitions and the 0-0 transitions may not belong to this
class. The transition moments are not calculated to be parallel to
either of the geometrical coordinate axes. The first two excited state
permanent dipole moments are much larger than the ground state permanent
dipole moment. Both excited states are also predicted to have molecular

internuclear distances somewhat larger than the ground state inter-

nuclear distances. The first two singlet triplet transition energies are

75

 

 

 

 

 

 

 

 

 

 

Table 6. Calculation Results for Indole.
Trans. Trans. Trans. Trans.
Energy Energy Moment Polar
(cm7 ) (nm) (Debyes) (deg. from
x-axis *)
35457 282.0 0.559 0.88 0.120
36912 270.9 0.478 -54.46 0.092
47023 212.7 0.732 -80.70 0.273
49319 202.8 0.904 63.47 0.437
Molecular Perm. Dipole Perm. Moment
State Moment Polar.
(Debyes) (deg. from
x-axis*)
Ground 3.93 -34.91
lst Excited 7.44 -68.91
2nd Excited 6.63 -47.40

 

 

 

 

* See Figure 7

 

76

Figure 7. Calculated Charge Densities and Bond Lengths for Indole.

Legend:
Numbers at each atomic position denote fihelectron
charge densities in electron units.
Numbers at each bond denote bond lengths in Angstroms.

The top number corresponds to the ground state.

The middle number corresponds to the first excited
singlet state (1L3).

The bottom number corresponds to the second excited
singlet state (lLb).

77

.n ouzmmm

mmm._
mom.~ qo~.~
o~m.H m#~.~

 

 

 

Sod
m2;
N84
5:
m: .1
e8;
moo;
mood
> 30;
X e24
m2;

wmo.c

78

calculated to be at 529 and 395 nm. However since the test molecules
were not pptimally parameterized for singlet-triplet transition energy
correlations, little credence is given to these values. The ionization
potential and electron affinity are calculated to be 7.0 and 1.8 eV
respectively. The charge density calculations show that the pyrrolic

l

hitrogen's acidity is La) ILb:>ground state.

VII. Correlations and Summary.

 

The value of our computations lies in the consistency between
calculated values and experimental data. In this section we shall come
pare our calculations with experiments.

1. The calculations perdict the vertical energies of the 1La and 1Lb

states to be very close with the 1La state energetically lower. Vapor

1L

b
transitions are close. It is difficult to choose which state is ener-

1

absorption spectra indicate that the 0-0 energies of the 1La and

getically lower. In polar media the

the lLb state. The closeness of the 1

La state lies energetically below
La and 1Lb energies is consistent
with the possibility of dual emission proposed experimentally.

2. The calculations predict the angle between the 1La and 1Lb transi-
tion moment vectors to be 55' and that neither moment parallels the x
or y axes. Experimentally these transition moments are oriented at a
relatively large acute angle to each other. The 1Lb transition moment
is predominantly oriented along the longer axis of the molecule which
is consistent with the calculations.

3. The calculations predict the first two singlet excited states have
much larger permanent dipole moments than the ground state with the

1L state having the larger of the two. This correlates very well with
a

the experimental results for absorption and emission. The polar

79

l
solvents red-shift the 1La absorption more than the Lb absorption as
predicted. These solvents also red-shift flourescent emission from the

1

1La state much more than from the L state. In fact the permanent

b

dipole moment change is so large that the fluorescence red-shift is
quite pronounced. This effect is both predicted and experimentally
observed.

4. The calculations predict that the acidity of the pyrrolic hydrogen
has the order 1La>~1Lb>1A. This is due to the decreased charge
density on the pyrrolic nitrogen as a result of excitation. The
decrease is more apparent for the 1La state than for the 1Lb state.
Thus the energy of the H-bond between the pyrrolic hydrogen and a
solvent molecule would be larger in the 1La state than in the 1Lb and
ground state. A larger red shift is therefore expected for the 1La
state due to H-bonding effects. This is consistent with the observed
red shifts in H-bonding solvents.

5. The calculated charge densities at the pyrrolic nitrogen also agree
with the ground state TT-electron stabilization mechanism proposed to
explain the effect of water on indole's absorption. The pyrrolic nitro-
gen's‘fi-electron charge density is calculated to be much larger than
that of any other atom. It also undergoes a larger decrease upon
excitation. So a solvent proton could stabilize the ground state
electron system at the pyrrolic nitrogen to a larger extent than the
1L8 or the 1Lb states. This would cause an absorption blue-shift.

The spectrum of indole in ethanol in the presence of small amounts of
H2804 exhibits a 260 cm.”1 blue-shift of the ngb—IA transition and a
140 on"1 1

blue-shift of the lLse- A transition.58 This is the expected

result if an acidic proton stabilizes the ground state and this

80

stabilization affects the 1La transition more than the 1Lb transition.
6. A further correlation of calculated charge densities and experimental
observations was provided by the ground state reactivity studies of

53 They found that in indole's ground state the 3

Hinman and Lang.
position was quite basic. This was quite evident compared with the
basicity of the 2 position. All these correlative datalend validity to
the calculated predictions of the charge densities.

Acids and bases particularly affect the excited singlet states.

In an acidic medium the fluorescence is quenched by some radiationless
transition process. In basic media the fluorescence yield is somewhat
reduced (no pun intended) audits emission is at a lower energy than in
a neutral medium. Obviously more spectroscopic data must be obtained
to understand the processes which these pH effects provoke.

Probably the most provocative aspect of the indole spectra is the
possibility of dual emission from the singlet and triplet states. Very
few molecules have been thought to exhibit such properties. Their
existence is quite unique and indicates open communication between the
excited states. If this process occurs one might expect to observe
fluorescent lifetimes with different values depending on where the life-
time was monitored in the spectrum. Although we have made some prelim-
inary attempts to find this dual fluorescent decay we have not been
successful yet. Likewise a close reexamination of phosphorescence
decay is also warranted to see if a dual decay exists.

The problem of two closely spaced excited electronic states is a
very intriguing problem both experimentally and theoretically. A study
where one can vary such spacing and look for changes in absorption and

emission properties could be quite revealing. Substituted methylindoles

81
seem to be such a potential series of molecules which should be studied.
These data show that indole is a unique molecule spectrOSCOPicallyo
The significance, if any, of this uniqueness to biology is unknown and

discussion of this significance will be deferred to the final chapter.

INDAZOLE
I. Vapor Absorption Spectra.

The vapor absorption spectrum of indazole is shown in Figure 8.
There are two electronic transition bands evident: one rich in vibra-
tional structure centered at 285 nm and the other with diffuse vibra-
tional structure centered at 245 nm. In contrast to the analogous
bands for indole, these bands are quite well separated and exhibit
little overlap. They are tentatively assigned the nge—IA and the
l

Lab-1A transitions respectively. The 1La 0-0 transition remains

unassigned but the 1L 0-0 transition is assigned at 34770 cm.-1

b
(2897 A). Other vibrational sequences in the ngF—lA transition begin
0
at 2836, 2787, and 2731 A with an average energy separation between
1

sequences of 692 cm? . Another sequence to the red of the 1Lb O-O
transition apparently begins at 2950 A and is denoted as a "primed"
sequence. Its origin remains a mystery. Again a normal coordinate
analysis of the 1Lb transition vibrational structure has not been
carried out. The 1La transition vibrational groups lie at 2570,
2516, 2446, and 2395 A with a mean energy separation between these
groups of 945 cmfl.
A hot band analysis was also performed with this molecule. The

results are shown in Figure 9. Normal results were obtained for the

lLb transition confirming the 0-0 position at 34470 cmfl. However

82

.Esuuooaw coHuQHOmn< uoam> mHonvcH .w ouswwm

.2. 5055:.

 

 

 

 

 

 

 

 

so. can can 2a 3. can ocu on. can
1 d d I 1 d d u q 1 8
/,.... I// / sf-;\ {\xllxl/ /\/\/\Im 06
z . z 1 / 1 0.0
t do
.. 3
1 so
1 0..
00m 00m OON CNN
0.0
./ Nd
/ . . .s. gage/2&1} '.0
rs“, : elf/(e; a .O . o
- . Qo
0.0

m40N<oz. o..

0.0.

83

 

 

 

210

Figure 9.

 

 

no no no m no no an ass
HVELENGTH (all

Indazole Vapor Absorption Spectra as a Function
of Temperature.

84

again the 1La transition oscillator strength grew more rapidly than

that of the lLb transition. A transition borrowing mechanism involving
vibronic coupling may be invoked to explain this phenomenon. This
molecule also has limited symmetry in the ground state so vibronic
coupling may again be significant.

II. Polarization Measurements of Fluorescence Excitation.

The only excitation polarization measurements reported were per-
formed by Schutt and Zimmermann14 in ethanol at -180°C. Their results
were quite expected. Monitoring at 310 nm they observed a negative
polarization from 240 to 260 nm. This polarization then rapidly became
positive and gradually leveled off toward a high positive value at
300 nm. The data strongly suggests that fluorescence is only emitted
from the 1Lb state. The negative polarization at the shorter wave—

lengths corresponds with excitation into the 1

L. state while positive
polarizations at the longer wavelengths corresponds with excitation
into the 1Lb state. The fact that the positive polarization does not

plateau before 300 nm indicates that there isatill some 1

La transition
character down to about 290 nm. The polarization values at their
extremes were 0.42 at 300 nm and -0.10 at 260 to 260 nm. The high
positive polarization correlates well with the statement that fluores-
cent emission is from the lLb state. However the fact that the nega-
tive polarization does not approach -0.3 suggests that the 1La and
1Lb transition oscillators are not at right angles with each other.
They probably do lie at a rather large acute angle though.
III. Solution Absorption Spectra.

Table 7 is a compilation of all the available absorption as well

as emission data for the azaindoles with the exception of purine.

E355

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.A V «unusuaouea an vouosov on. soosououw-
.3: cu vooaoumxo one onuwco~u>o3 -<
Acmeomm Asmvmmm Aemvoon booms
Aonvnom ,Acmvmsm nn< o~ousvusuumoosusun
Acmeonn Aemvo~n Aemvmsn scams
Ann-av-q.nv
AcmvmaN Asmvom~
Aaneom~.no~ Aemv~o~.~n~ Aanv~w~.¢q~ Amnv~w~.ns~ Amnvmma.am~.on~ «ac amoucvaaaucoa-un.e
Ae~0.ooe.onsncOau.o Adhvnmn A~NVOmm AsHVONM
Aeavoee.o~e “mavcnn “envoan AsmVOmm Aemvomn bosom
Annuao-n.~v
Aqmvmmw AqmemQN As_vm-.mo~ Ammvm-.mq~
azhvmam AfiNVmNN Ammvsa~.~m~ Aq~v~m~.ms~ Ammvonm.~e~.~q~ aa< odouaauuoncon
Acmvmnc Aqmvaom Asmvmam
Aamvmme Aomvoom Aomvmom Aonvoqm Ammvomn bonds
. AanHON Asmvoom Anymou AnemoN vamo~.-~ vawm~.mo~
Aomvmom.o¢~ Aomvms~.wm~ Aomvoo~.om~ Aomveo~.am~ Aamv~¢~.ow~ Ammvmm~.~w~ aa< «movcaaua-e
Agavewn Acmv~om Aemvoam toads
Aqmvo_n.ao~ Asmv~o~.oo~ na< umoccauuu-o
Aqmvmoe Asmvsos Asmv~mn bosom .
Asmcna~.wo~ Asmvnam.mem ma< «savanna--n
AcmcoNQ Asmchiq Asmvomm toads
Aqnchwn Asmv~o~ an< «ocean-un-e
AemvoNs.oo~ncoau.o Amavcan Asmvmem Asmvozm Asivso~
Annvouc.oon.ckn Ammooon Ammvoam Ammvwo~.oa~.zm~ Ammvho~.~o~.~m~ Ammvaom.~o~.mm~ Ammvmo~.~m~.~w~ noses A V
Ins-n
Ae~0o-.oo~.oe~ Aamvaa~.qe~ Aomvom~.se~.mq~
Aaquh~.om~ Aemvne~.eo~.oe~ Aqmvea~.-~.ms~ Asav¢-.ohm.~sm Aomvow~.q-.~s~
Aonv~a~.-~.~q~ Aomona~.oc~.as~ Awmva-.~e~.mem Awmvon~.~5~.es~ Ammvam~.a-.sqm Aomcow~.q-.~s~ Ammvka~.-~.mn~ na< amouauosancon
Aemvoos.omqncozu.o Aaxvomn Aaevozm AsavoNM
Aqavone.o~e Ashenon Aqmvoen Aqmv-m AemVNNM bosom
q Ann--wv
Aemvuon.~o~.nm~ Aemvaow.ew~.cn~ Ae~voo~.so~.mn~
Amhvoon Ae~v~on.oo~ Azevoou Anmvmm~.om~ Aemvno~.~o~.on~ Aan0o0~.em~.ne~ .44 «saunas»
sosoooouozooosm soon ouu< nous: “seesaw oozum cooueoouvmx uoms>

 

.suuuoam ouovsueu< so caveman uso>aom

on sun-H

 

86

Indazole does not exhibit an absorption spectrum in hydrocarbon solvents
much different than that in the vapor other than being somewhat more

diffuse. However in ethanol the spectrum is red-shifted, with the 1L

a
transition red-shifted more than the 1Lb transition (800 cm"1 vs.

400 cm-l). In water there appears to be slight blue-shift compared to
the spectrum in ethanol. Interpretation of this data follows the same
tact as for indole. The 1La state permanent dipole moment is probably

larger than that of the 1

Lb state and both are larger than that of the
ground state. Ethanol stabilizes the 1].a state more than the lLb state
through dipole-dipole interactions. In addition it can further

stabilize the 1La state more than the 1

Lb state through H-bonding pro-
vided the pyrrolic nitrogen's acidity follows the same trend as in
indole. If the latter assertion is true, water could interact via its
hydrogen proton with the‘fi-electronic charge centered on the pyrrolic
nitrogen causing a spectral blue-shift.

When absorption is done in an acidic solution the resultant cation
spectrum has a demonstrable red-shift. Although the acidic proton is
stabilizing the pyrrolic nitrogen's‘fi-electron charge it also must be
interacting elsewhere in the molecule to cause this red-shift. The
most likely place would be at the‘W-electron charge on the pyridinic
nitrogen. This stabilization would occur if the excited states?
charge density exceeded that of the ground state. Since the 1Lb
transition is red-shifted more than the 1La transition, our reasoning
predicts that the pyridinic nitrogen's charge density is more in the
lLb state than in the 1La state, The charge density changes at the

pyridinic nitrogen must be more than those at the pyrrolic nitrogen

since the effects are antagonistic and a red-shift is observed. In

87

a basic medium the resultant anion also shows a red-shift. This is not
surprising since the Coulombic attraction between the pyrrolic hydrogen's
proton and the/fi-electron charge and the residual negative charge after
the proton is extracted. It then takes less energy to excite the
electronic system. The anion also exhibits an apparent merging of the

1

two absorption transition bands. The La transition loses intensity and

appears to form a high energy shoulder to the now broad 1Lb transition.
Thus the anion has a different spectrum than either the neutral or
cation species.

These absorption spectra indicate that the replacement of a carbon
with a pyridinic nitrogen introduces new factors which must be considered
when one interprets these spectra,particularly solvent perturbations.

IV. Fluorescence Spectra.

The relative red and blue solvent shifts of the fluorescence
spectra parallel those solvent shifts observed in the absorption spectra.
Fluorescence with some vibrational structure is exhibited by the neutral
molecule. The Stokes' shift of this emission in polar solvents is not
as large as that for indole. This indicates that the lowest excited
state permanent dipole moment doesn't differ much from the ground state
permanent dipole moment. In addition the escited state stabilization
due to H-bonding does not manifest itself greatly in a Stokes' shift
contribution. Fluorescence polarization shows that emission is from

only the 1Lb state.1“ This is an expected result since the 1

L and
a
1Lb transitions are well separated.

In contrast with the results for indole, the cation of indazole

fluoresces. Its emission is somewhat broader with less vibrational

structure than the neutral species. The red-shift of the cation

88

emission from the neutral molecule correlates well with the supposition

that the acidic proton stabilizes the 1

Lb excited state at the pyridinic
nitrogen position. Only the 1].b state emits and our absorption spectra
analysis predicted the pyridinic nitrogen charge density to be greatest
for that state. It seems reasonable to conclude that the cation is
formed through an interaction of the acidic proton with the lone pair
electrons of the pyridinic nitrogen.

The anion also fluoresces. Its emission is slightly to the red
of the cation's fluorescence and is broad and vibrationally structure-
less. As for indole this emission is probably due to an influence of
the pyrrolic nitrogen's residual negative charge on the excited 1?
electron system.

V. Phosphorescence Spectra.

The only phosphorescence spectra published for any of the azain-
doles excluding purine has been that of Schutt and Zimmermann14 in
ethanol at -180‘C. For indazole they obtained spectra for both the
neutral and cation species. The neutral molecule displays peaks at
422, 452, and 481 nm and inflections at 437, 467, and 500 nm. The
cation's spectra is red shifted and displays peaks at 446, 481, and
508 nm and inflections at 461 and 560 nm. Phosphorescence excitation
1

L1) and 1La excitations. This

yields the expected interpretation that the phosphorescent transition(s)

polarizations were negative for both

is polarized out-of-plane. The emission polarization measurements,
though expectedly negative, surprisingly showed spectmfl structure. As
stated before, Schutt and Zimmermann interpreted this to indicate

phosphorescent emission from two different triplet states. These are

presumably the 3La and the 3L states. Since excitation was to the

b

89

1Lb state the emission polarization spectra could be interpretédhas
showing that the phosphorescence peaks corresponded with emission from
the alb state and the inflections with emission from the 31a state (the
more negative polarizations corresponded with the phosphorescence peaks
and the more positive polarizations with the inflections). If phosphor-
escence is indeed emitted simultaneously from two states the 31a and
3Lb state overlap is quite coincidental. This is especially true when
one considers the separation between the 1La and 1Lb states.

The vibrational structure of the cation phosphorescence was more
diffuse than that of the neutral species but the essential features,

polarizations and data interpretations were the same.

VI. ‘gorrelations and Summary.

 

Indazole exhibits an observable energy separation of the 1La and

l

the Lb state. Apparently these two lowest transition moments form

an acute angle with each other and lie in‘the.plane of the molecular

1A transition has the lowest energy. Fluorescence

nuclei. The 1Lb
then only emanates from this lower state. The fluorescence spectra

has vibronic structure which is difficult to correlate with the cor-
responding absorption spectrum. Thus one must view correlations between
absorption and fluorescence vibronic structure as somewhat temerarious.
Since the anion absorption shows a merging of the 1La and 1Lb bands,
emission polarization studies may show dual fluorescent emission similar
to that noted for indole.

The absorption and fluorescence data also show that solvent per-

turbations can yield knowledge about the electronic structure of this

9O

molecule. For indazole it appears that the acidity of the pyrrolic

nitrogen is 1La>1Lt§>ground state and the basicity of the pyridinic
l

nitrogen is 1Lb2> Lg>ground state. In addition the permanent dipole

moment magnitudes appear to follow the trend 1L8>1Lb>ground state.

BENZIMIDAZOLE
I. Vapor Absorption Spectra.

One report has been published of the benzimidazole vapor absorption
spectrum.23 Our results were quite similar to those reported by Gordon
and Yang with the exception that we obtained more structure to the red
and to the blue of their spectrum. We also scanned the 1!.a region of
the spectrum. Our spectrum is shown in Figure 10. The ngh-IA and
1LJP-1A transitions are well separated and are centered at 275 and 240
nm respectively. The 1Lb 0-0 transition occurs at 36023 cmf1 (2772 A)
and the 1!.a 0-0 transition remains unassigned. Again the 1Lb transition
is characterized by the richness of vibrational structure while the 1].a
transition is much broader and has very diffuse vibrational contribu-

1

tions. The mean separation of vibrational groups in the La transition

is 1064 cmfl. These groups occur at 2266, 2320, 2381, and 2439 X.
Vibrational sequences other than the 0—0 sequence for the lLb transition
begin at 2561, 2610, 2661, and 2715 A. The mean separation between
these sequences is 743 cmfl. An interesting phenomenon was observed in
the two lowest energy sequences of the 1Lb transition. The higher

energy members of these sequences are actually doublets. There is about

1

an 11 emf separation between the individual peaks comprising each

doublet (the higher energy peak of the first member in the 1Lb transi-

tion lowest energy sequence was chosen as the 1L 0-0). The origin of

b

 

91

.Esuuomam cowuauOmn< uomm> mHonmoefiszmm .oH muswwm

.e: 2.93.353

 

 

 

8. can can oz can on» 3a 8a 8a
q 4 q ....... a ...... . a q a q

l _ 06
j j a l a
I |/ ‘52 l 0.0

6

I 1 .6
I l 3
f 1 o.

 

 

 

 

 

00m 00m Ohm omN
0.0

m6

7. .3
. .o.o

so

no

w40N<o_2_szm o._

0.0.

92

 

Temp IN“ C

Temp ISO’ C

 

 

Temp IO.“ 0

 

'70” 00‘ C

220 230 840 260 200 270 ”O 2.0

Figure 11.

HVELEIGTH (hm)

Benzimidazole Vapor Absorption Spectra as a Function
of Temperature.

93
these doublets remains unexplained.

Vapor spectra of benzimidazole at different temperatures were
obtained. The results are shown in Figure 11. Again the oscillator
strength of the 1La transition increases relative to that of the 1Lb
transition. A vibronic coupling mechanism.mey be invoked to cause the
presumed transition stealing. In addition another curious effect was
noted. The first sequence to appear in this analysis occurs 1377 cm?1
to the red of the 1Lb 0-0 transition. This sequence is subsequently
overwhelmed by the oscillator strength of the 1Lb transition as the
temperature is increased. It is also present in the"concentrated"
vapor absorption spectra but only appears as minor peaks in the 1!.b 0—0
transition sequence. However in polar and nonpolar solvents all
vestiges of this sequence are absent. Thus it does not seem that this
sequence is due to a solute impurity since this substance should also
absorb in dilute solution.

II. Polarization Measurements of Fluorescence Excitation.

Excitation polarization measurements for benzimidazole were
reported by Schutt and Zimmermann11 in ethanol at -180'C and reproduced
in our laboratory. The polarization values were slightly negative from
240 to 250 nm, rapidly became increasingly positive to 270 nm and then
gradually became more positive to 280 nm. The final polarization value
was 0.45. These data show that fluorescent emission emanates from the
1Lb state and that the 1La transition moment is oriented at an acute
angle to the 1Lb transition moment. This angle is not 90'. In fact
our investigations indicate that the angle between transition moments

is about 56'. The gradual increase in polarization values from 270 to

280 mm is indicative of some 1La transition character in this

94

predominantly 1

Lb transition portion of the spectrum.

Fluorescence excitation polarization measurements were also made
for the benzimidazole cation.14 Results similar to those described
above were obtained so the same interpretations as stated above are
sufficient to explain these data.

III. Solution Absorption Spectra.

The data in Table 7 are again useful for this discussion. Hydro-
carbon solvents tend to diffuse the absorption spectrum of benzimidazole
relative to the vapor absorption spectrum. In addition there is a
slight red-shift in these solvents which can be attributed to dispersive
forces.22 The spectrum in ether shows a further slight red-shift with
the 1La transition red-shifted more than the 1!.b transition. Since ether
can H-bond with the hydrogen on the pyrrolic nitrogen this observation
probably means that the pyrrolic nitrogen's 1?-electronic charge density
follows the pattern ground state >1Lb> lLa'

In both methanol and ethanol there is a slight blue-shift of the
1Lb transition relative to the spectrum in ether. A slight blue-shift
for the 1La transition was observed in ethanol compared to ether. These
shifts cannot be interpreted in terms of the dipole-dipole interaction
since the permanent dipole moment increases to some extent as a result
of excitation to either the 1La or lLb states. This is supported by
the red shift observed for both transition bands in going from vapor to
hydrocarbon solvent. This shift is due to dipole-induced dipole inter-
actions. The increase of the 1La state permanent dipole moment in
benzimidazole is probably much smaller than that in indole. The
observed blue-shift must be due to another mechanism which more than

compensates for the small dipolar interaction red-shift. In addition

95

the mechanism must compensate for the red-shift produced by solvent
H-bonding with the pyrrolic nitrogen's hydrogen which should not differ
appreciably from that observed in ether. The mechanism probably

involves the influence of the solvent proton on the pyridinic nitrogen's
TT-electronic charge density. The pyridinic nitrogen's charge density
must therefore be smaller in the excited states than in the ground state.
This would lead to solvent stabilization of the ground state compared to
the 1Lb or 1La states giving rise to a blue-shift. When water was

utilized as a solvent a further slight blue-shift of both the 1

La and
the 1!.b transitions was observed when compared with the spectra obtained
in the alcohols. This further strengthens the contention that there is
a stabilizing interaction between a solvent proton and the solute
pyridinic 1T-electronic charge. The further blue-shift only reflects
the relative acidity of the water proton compared with the alcoholic
proton.

Although it is difficult to make an exact assessment of the rela-

tive shifts it appears that the 1

Lb transition blue-shifts more than

the 1La transition when varying the solvent from ether to alcohol to
water. One must recall that the shifts are the result of several

modes of interaction, namely, dipole-dipole interaction, H-bonding
involving the pyrrolic hydrogen and that involving the pyridine
nitrogen. Individual effects can be cause antagonistic shifts. Thus
only qualitative statements can be made: The dipole-dipole interactions
cause red shifts of both bands with the 1La transition shifted to a
slightly larger extent (3MP vs. vapor spectra). H-bonding with the

pyrrolic hydrogen causes a red shift (ether vs. 3MP spectra) indicating

that the hydrogen is slightly more acidic in the 1La and 11b states.

96

The acidity is greater in the 1

La state. H-bonding with the pyridinic
nitrogen causes a blue shift (alcohol vs. ether) indicating s smaller
charge density at the pyridinic nitrogen in both the 1La and 1Lb states.
The smaller blue shifts for water vs. alcohol (relative to alcohol vs.
ether) is a reflection of a greater dipole-dipole interaction red shift
in the more polar solvent. The smaller blue shifts (relative to ether)
for the 1!.a state is a reflection of a larger dipole moment for that

1

state compared with the moment of the Lb state. One may conclude then

that the pyrrolic nitrogen charge density pattern is 1A:>1Lb:>1La and

the ‘fi-electron pattern for the pyridinic nitrogen is 1A:>1La:>-1Lb.
In acidic media a still further blue-shift is observed for the

1 1

1La and Lb transitions. The Lb transition again shifts more than the

1!.a transition. The same explanations as proposed above fit these data.
The benzimidazole cation also exhibits an oscillator strength diminution

of the 1

La transition compared with the results obtained in the pre-
viously discussed solvents. The 1Lb transition oscillator strength
remained visibly unaffected. It is tempting to link this result with
the anomalous vapor phase 1La transition oscillator strength alterations
which were temperature induced. Such spectral similarities should not

go unnoticed and a vibronic stealing mechanism should at least be con-
templated. However it must be pointed out that the cation is in some
respects a different species from the neutral molecule.

The anion of benzimidazole exhibits a radically different absorption
spectrum than the spectra previously discussed. This spectrum together
with those of the cation and neutral species are shown in Figure 12.

The anion spectrum is quite similar to that reported for indazole. The

1L

1La transition is less intense and apparently merges with the b

0.0.

 

ANION

 

q
q
q
d
as
d
1
q

 

 

 

NEUTRAL
1
l 1 1 : 1 1 1 i 1
CATION
i 1 1 1 J J 1

 

220 230 840 200 800 870 2.0 ”O 300
WAVELENGTH (nm)

Figure 12. Benzimidazole Absorption Spectra in Basic,

Neutral and Acidic Solution.

 

98
transition. In turn the 1Lb transition vibrational structure differs
from its analogues in the other media. Both transitions exhibit the
expected red-shift which has been purported for the previous molecules
to be a result of the anion negative charge on the‘fi-electron structure.
The resultant Coulombic repulsion presumably causes some destabilization
of the‘fi-electron structure which is manifested as a red-shift. The
different spectral profile of the anion probably reflects the fact that
this species differs somewhat from the neutral molecule and cation.
IV. Fluorescence_§pectra.

Benzimidazole fluorescence exhibits some vibrational structure.

It also exhibits a relatively small Stokes' shift which indicates that
the emitting excited state is not solvent-stabilized to any extent
compared with the ground state. Thus the emitting state (probably 1Lb)
permanent dipole moment is not much larger than the ground state's
moment. This correlates well with the interpretation proffered for the
relative shifts observed in various solvents.

The solvent-induced shifts in the fluorescence spectra parallel
those shifts observed in the absorption spectra for 3-methylpentane,
ethergalcohol and water. The spectrum in water at room temperature is
shown in Figure 13 together with spectra obtained in basic and acidic
media. Again it is noted that the fluorescence is species specific.

14

Emission polarization data in alcohol at -180°C verified that

l

the L state fluoresces in the neutral molecule. A slightly negative

b
flat polarization resulted when excitation was monitored at 240 nm
which is in the 1La transition band. However the results were some-

what different for the cation. Excitation at 265 nm, which is pre-

1
dominantly comprised of the Lb transition, resulted in a polarization

99

 

 

 

 

 

 

I I I I I I I I I I I
' ANION
" NEUTRAL
__ ‘4
4); - -
p— —
_ CATION _
P
4 I L l l l l l L

 

 

 

zeo 300 no 340 sec sec 400 no «0 «0 «so
WAVELENGTH (nm)

Figure 13. Benzimidazole Fluorescence Spectra in Basic, Neutral and
Acidic Solution.

100
spectrum which was initially flat and positive at shorter fluorescent
wavelengths but became less positive toward longer wavelengths. My
data at room temperature show only a broad fluorescence peak at 370 nm
for the cation. This suggests that Schutt and Zimmermann have not
obtained spectra for the cation at all! They probably were making
observations with the neutral species.

V. Phosphorescence Spectra.

 

The phosphorescence spectrum in ethanol at 77°K was obtained in
our laboratory is shown in Figure 14. It differs from the spectrum
reported by Schutt and Zimmermann.la The phosphorescence 0-0 occurs
at 374 nm. Other peaks appear at 386; 394, 398, and 420 nm. Inflec-
tions appear at 407, 414, 425, and 440 nm. Polarization data were
obtained by Schutt and Zimmermann and show that the phosphorescence is
polarized predominantly out-of—plane relative to the molecular topology.
When they excited at 279 nm (the 1Lb transition) the emission polariza-
tion was slightly negative from 250 to 280 nm. These data strongly
suggest that the phosphorescence transition oscillator is not oriented
exactly perpendicular to the molecular plane but instead lies at some
acute angle to this plane.

The cation emission reported by Schutt and Zimmermann was remark-
ably similar to phosphorescence from the neutral molecule. The general
appearance,peak and inflection positions, emission polarization and
excitation polarization were hardly distinguishable between these spe-
cies. This is surprising since both the absorption and fluorescence
spectra exhibited a blue-shift. They probably were performing measure-

ments on the neutral molecule so no differences should be seen.

101

one

owe

.x Own um Hosmcum cw mucmommpooamogm mHommowEwNCmm

A55
oee

Ihozw4m><3

0N?

one

on»

.qH moswwm

 

 

_

fl

a

_

_

 

 

102

VI. Theoretical Calculations.

 

Results of these calculations are shown in Table 8 and Figure 15.
Using the criteria of permanent dipole moment magnitudes, transition
moment orientations and oscillator strengths the first excited state
was assigned as 1Lb and the second excited state as 1La. Of these
criteria the transition moment orientation was of little use since both
moments were oriented at approximately the same angle from the principal
axes. The orientation of these transition moment axes is neither paral—
lel to the principal axes nor at right angles to each other. These two
lowest singlet transitions are calculated to be energetically close to
each other. However it should be recalled that accuracy to within 0.1
eV is considered quite good. The ionization potential and electron
affinity are calculated to be 7.2 eV and 1.8 eV respectively. Both
excited states are predicted to have larger internuclear distances than
the ground state. Charge density results predict the pyrrolic nitrogen's
acidity is 1Lé>leg>ground state and the pyridinic nitrogen's basicity is
calculated to be ground state > lLb> 1L a .

VII. “Correlations and Summary.
The solution and vapor absorption spectra analysis, polarization

data and the computational results all correlate to assign the ngP-lA

transition below the 1'L<----1A transition. The energy separation between

a

these transitions is not definitely established. Certainly the main
spectral features of each transition are well separated except possibly
for the anion. However the polarization data suggest there is some 1La

transition character at least down to 275 nm in solution. The calcula-

tions also predict 3 small transition energy separation.

103

Table 8. Calculation Results for the Azaindoles.

 

 

 

 

 

 

 

 

 

 

 

Benzimidazole
Trans. Trans. Trans. Trans. f
Energy Energy Moment Polar
(cm. ) (nm) (Debyes) (deg. from
x-axis*)

36094 277.0 0.420 32 0.069
37424 267.2 0.620 -40 0.156
48019 208.2 1.048 89 0.572
49637 201.5 0.539 31 0.156

4-azaindole
34788 287.4 0.513 13 0.099
36418 274.6 0.535 -62 0.113
49394 202.4 0.746 48 0.298

5-azaindole
35790 279.4 0.579 - 9 0.130
36914 270.9 0.406 -53 0.066
47065 212.5 0.670 -88 0.229
49082 203.7 0.941 62 0.472

6-azaindole
35245 283.7 0.585 0.10 0.131
37373 267.6 0.420 -45 0.071
46975 212.9 0.601 -74 0.184
49994 200.2 0.512 41 0.142

7-azaindole
34955 286.1 0.551 - 9 0.115
36268 275.7 0.483 -77 0.092
47936 208.6 0.813 -69 0.344
49078 203.8 0.705 53 0.264

 

 

 

* The x-axis parallels the short axis of the molecule.

Table 8 cont'd.)

104

 

 

 

 

 

 

 

Trans. Trans. Trans. Trans. f
Energy Energy Moment Polar
(cm‘l) (nm) (Debyes) (deg. from
x-axis*)
4-azabenzimidazole
35568 281.2 0.468 52 0.085
37131 269.3 0.590 -43 0.140
48483 206.3 1.177 -78 0.729
49701 201.2 0.614 30 0.203
5-azabenzimidazole
36565 273.5 0.318 34 0.040
37330 267.9 0.639 -35 0.165
47972 208.4 0.991 78 0.511
49727 201.1 0.485 34 0.127
* The x—axis parallels the short axis of the molecule.

 

105

Table 8 eont'd.)

 

 

 

 

 

Benzimidazole
Molecular Perm. Dipole Perm. Moment
State Moment Polar.
(deg. from
x-axis*)
Ground ' 4.14 ‘ ~34
lst Excited 5.50 ~58
2nd Excited 8.28 ~54
7~azaiadole
Ground 3.57 ~36
lst Excited 8.92 ~71
2nd Excited 5.25 ~46

 

 

 

*The x~axis parallels the short axis of the molecule.

106

Figure 15. Calculated Charge Densities and Bond Lengths for
Benzimidazole.

Legend:
Numbers at each atomic position denote‘fibelectron
charge densities in electron units.
Numbers at each bond denote bond lengths in Angstroms.

The top number corresponds to the ground state.

The middle number corresponds to the first excited
singlet state (lLb).

The bottom number corresponds to the second excited
singlet state ( La)'

107

.nu on:ufim

mum.~
mmm.~

moo.H
oq~.~
o~o.~

 

 

 

._
a

.5;
sec; .
weed

108

The two lowest singlet transition moments are not oriented at right
angles to each other and neither one lies parallel to either of the prin~
cipal axes of Figure 15. The calculated angle between these moments is
70' and an analysis of-the polarization data indicates the angle is 56'.
This is a remarkably good correlation considering the difficulties
involved in obtaining each value. These results are not surprising when
one considers the lack of symmetry this molecule possesses. The polar-
ization data indicate that the phosphorescence transition moment is pri-
marily out-of-plane but not perpendicular to it.

Fluorescence emanates from the 1Lb state for the neutral molecule.
This fact is exemplified by the polarization data. For the charged
species it is difficult to make such positive statements. The acidity
of the pyrrolic hydrogen follow the order 1La>>1Lb2>1l.which is consis-
tent with the‘fi-electron charge density values calculated at the pyrro-
lic nitrogen. In addition the calculated prediction of the pyridinic
nitrogen basicity being more in the ground state than in the excited
states correlates well with the solvent shifts. The basicity follows
the order ground state >1Lb>1La'

The calculated permanent dipole moments also correlate well with
the observed solvent shifts. The permanent dipole moment magnitudes
are 1La>1Lb>ground state. Thus in polar solvents the 1Latransi-
tion is predicted to red-shift more than the 1Lb transition. The
observed red-shift in going from vapor to 3MP and in going from 3MP to
ether correlates with thepyrrolic nitrogenflfi~electron charge density
being lower in th excited states than in the ground state. The observed
absorption and fluorescence shifts in alcohol, water and acidic media

correlate with the calculations when the dipolar interaction vs. proton~

109

11~electron interaction are both taken into consideration. Thus the
solvent protons interact with the pyridinic and possibly the pyrrolic
nitrogens"fl~electronic charge to cause a blue-shift. This is Opposed

1

by the dipole-dipole interactions which tend to red-shift the La tran-

sition more than the 1Lb transition. The result is a blue-shift of
both transitions with the 1Lb transition blue-shifted more than the
1La transition. Since the pyridinic nitrogen contains a lone pair of
electrons it seems reasonable to assume that most of the protonéfiL
electron interaction occurs at this site.

Benzimidazole also exhibited the fact that vibronic energy spacings
were not identical between homologous members in absorption and emission
spectra. This again supports the view that quantitative correlations
between absorption and fluorescence spectra are tenuous. The relatively
small Stokes' shift between absorption and fluorescence correlates with
the small difference between the permanent dipole moments of the ground
and emitting 1Lb state. This adds credence to our ability to predict
Stokes' shifts based on calculated permanent dipole moments. Finally
the anomalous behavior of the vapor absorption data as a function of

temperature evoke our ubiquitous vibronic coupling scheme as a possible

explanation.
4-AZAINDOLE

1. Absorption and Fluorescence Spectra.

The few data which have been reported for these Spectra are tabu-
lated in Table 7. One can see that the absorption spectrum of the
cation is red-shifted compared with the spectrum of the neutral molecule
in water. The fluorescence data exhibit a definite red-shift when
comparing the water spectrum with that in ethanol. There also seems to

be a further slight red-shift when the cation is formed. All these

110
fluorescence spectra are at longer wavelengths than those of the
azaindoles previously discussed. These spectral positions are reminis-
cent of those observed in indole.
II. Theoretical Calculations.

The results of these calculations are shown in Table 8. Based on
the transition moment orientations and the oscillator strengths the first
and second excited states remain unassigned. The first two transitions
lie energetically fairly close and their transition moments are mutually
oriented about 90' to each other according to the predictions. Neither
moment is calculated to be parallel to a primary molecular axis. Perman-
ent dipole moments and charge densities were not computed for this mole~
cule except for the ground state. The ionization potential and electron
affinity were calculated to be 7.1 eV and 1.7 eV respectively.

III. Correlations and Summagx.
Since the amount of experimental data is so scarce any correlations

would be more speculative than even I would consider.

5-AZA1NDOLE
I. Absorpgion and Fluorescence Spectra.

Table 7 again shows a paucity of data for this molecule. There
seems to be a red-shift of the absorption spectrum of the cation relative
to the absorption spectrum of the neutral species. A.moderate Stokes'
shift is noted in ethanol, water, and acid. The water fluorescence is
red-shifted relative to the ethanol emission and the cation is apparen-
tly slightly further red-shifted.

II. Theoretical Calculations.

The results are again shown in Table 8. The transition moment

orientations and the oscillator strengths both agree with the assignment

111

of 1L8 as the first singlet excited state and 1Lb as the second singlet
excited state. These two states are predicted to lie energetically
quite close to each other. Their transition moments are calculated to
lie about 45' from each other and neither moment parallels a molecular
principal Qxis. The ionization potential and electron affinity are
predicted to be 7.1 eV and 1.8 eV respectively. Neither the excited
state permanent dipole moments nor‘fi-electron charge densities were
calculated.
III. Correlations and Summary.

Again the scarcity of experimental data eliminates any correlative

conclusions.

6-AZAINDOLE
I. Absorption and Fluorescence Spectra.

Table 7 depressingly again shows a dreadful dearth of data.' There
is an apparent red-shift of the cation absorption when compared with
absorption by the neutral species. A very large Stokes' shift is noted
for this molecule in polar solvents. There is a slight absorption spec-
trum red-shift from ethanol to water and a slight blue-shift from water
to acid. The large Stokes' shift is very suggestive of a quite large
emitting state permanent dipole moment.

II. Theoretical Calculations.

Table 8 again shows the results of these calculations. Based on

the transition moment orientations and the oscillator strengths the

energetically lowest singlet excited state is assigned as 1

next excited state as 1Lb. Their energetic separation is not large.

L and the
a

The two transition moments are predicted to lie about 45' relative to

each other. Neither moment is calculated to lie parallel to a

112

primary molecular axis. The excited state permanent dipole moments and
their‘fiLelectron charge densities were not calculated. The ionization

potential and electron affinity are respectively predicted to be 7.1 eV
and 1.8 eV.

III. Correlations and Summary.

Due to the absence of data again correlations will not be attempted.

7~AZAINDOLE
I. Vapor Absorption Spectra.

Figure 16 shows the vapor absorption spectrum of 7~azaindole. This
spectrum is quite similar to the vapor spectrum of indole. The 1L:-1A
and 1L5e—1A transition envelopes significantly overlap each other. Rich
vibrational structure characterizes the 1Lb transition while the 1L8
transition has broad diffuse structure. The transition overlap appears
to be worse for 7~azaindole than for indole; that is a visual separation
of these two transitions is more difficult to perform for 7~azaindole
than for indole. Due to this more apparent transition merging it seems
likely that the 1La 0~0 transition is energetically below the 1Lb 0-0
1Lb 0-0 at

34676 cm‘1 (2883 A) and the 11.8 0-0 at 34414 em‘1 (2905 i).8 The

transition. The only report in the literature assigned the

method used by these authors to assign the 1La O~0 was discussed in the

indole Vapor Absorption Spectra section together with my comments about

the applicability of this method. Our analysis places the 1Lb 0-0 at

34636 cm-1 (2888 A) and leaves the 1La 0~0 unassigned since I believe

no suitable procedure has been discovered for making the latter assignment.
The 7~azaindole vapor spectrum exhfiflted an effect similar to one

1
noted for benzimidazole: the three lowest energy sequences in the L1)

113

.Esuuoomm sewuqo0m2< uomm> oHoochN<In .oH opsmwm

.e: F8333.

00m 000 Om“ 0.0 on“ 0'0 and

q q -. q q u

 

 

 

 

e . . ...f/ 7 .24 (3115 file‘s-(1.31}

 

 

0mm 0mm CNN

mqooz.<N<Ih

o.o
~.o
to

d .o
ed
so

0..

114

'70.» 090’¢37

 

 

Temp "0' C

 

0.0.

 

Tesw» OF’ C

 

0.

Temp 00' C

 

 

 

220 230 no zoo sea no see zoo zoo
HVELENGTH (In)

Figure 17. 7-Azaindole Vapor Absorption Spectra as a Function
of Temperature.

115

transition contained doublets for the high energy members of these
sequences. The separation between individual peaks comprising each

doublet varied but for the 1Lb 0-0 sequence the average separation was

5 cm71. For this sequence the higher energy peak of the highest energy

member was chosen as the 1

Lb O-O transition. Again the origin of these
doublets remains unexplained. A reasonably consistent sequence progres-
sion (separation of sequences) was noted for the 1Lb and 1La transitions.
For the 1Lb transition the energy separation is 706 cm"1 which places
the beginning of sequences at 2719, 2772, 2806 A as well as 2888 A for

the 0~0 sequence. The 1

La transition vibrational group separation is
723 en'l with groups occurring at 2545, 2591, 2641, and 2693 3. There
undoubtedly are other groups at lower energies but their topography is
masked by the superimposed 1Lb transition. Thus it is difficult to

judge that a 1La transition sequence lies energetically below the 1

0-0 sequence.
A series of vapor spectra were taken as a function of temperature
and are shown in Figure 17. A hot band analysis confirmed the 1!.b 0-0

1

transition at 34636 emf . In addition the 1La transition oscillator

strength increased with temperature relative to the 1Lb transition
oscillator strength. Again we are confronted with this anomalous tran-
sition behavior. It seems natural to invoke the coupling mechanism
between vibronic states to explain the anomaly. The mechanism could
cause the 1La transition to become more "allowed" through intensity
borrowing from anothhr transition.

II. Solution Absorptiong§pectra.

The data in Table 7 provide the information used in this discus~

sion. In 3~methylpentane the absorption spectrum is again more diffuse

116

and has lost much of the vibrational character evident in the vapor
phase. There is also a slight red-shift of both the 1La and 1Lb transi-
tion in this solvent. This shift is probably due to the dispersive for-
ces manifested in solvents.22 In ether both transitions are further
red-shifted. This is indicative of an effect caused by H-bonding,
probably at the pyrrolic nitrogen. The red-shifts indicate that this
nitrogen is more acidic in the excited than in the ground states.

The spectral shifts observed in ethanol, water, and acidic media
are much more puzzling. In ethanol both transitions are further red-
shifted. The 1La transition is shifted more than the 1!.b transition
but both shifts are comparatively slight. There are now three different
interactions which must be taken into account. The first interaction
is H-bonding which occurs at the pyridinic as well as the pyrrolic
nitrogen. The H-bonding effect involving the pyrrolic hydrogen should
not differ significantly from that observed in ether selution. The
further red shift observed when comparing the spectra in ethanol to
that in ether is therefore attributed to the pyridinic nitrogen
~solvent H—bonding. A second possible interaction is the effect of an
acidic proton on the TT-electron charge which resides at the pyrrolic
and pyridinic nitrogens. Although this interaction may also occur at
other nuclear centers the effects of these interactions will not be as
pronounced as the effects at the nitrogens. The spectral shifts pro~
duced by this interaction will be the same at the pyridinic nitrogen as
that produced by H-bonding at this nitrogen. However at the pyrrolic
nitrogen the solvent proton Jfi~electron interaction will produce a
spectral shift opposite that due to H-bonding at this nitrogen. The

third interaction consists of the solvent dipole's effect on the

117

permanent dipole of the solute molecule's ground and excited states.
The red shifts observed in going from vapor to hydrocarbon solvents
indicate that the permanent dipole moment increases as a result of
excitation to the 1La and 1Lb states. This is particularly true for

excitation to the 1

La state. One would therefore expect that the
dipole-dipole interaction in polar solvents causes a red shift. The
observed shifts in ethanol are probably predominantly due to the
dipolar interactions. Since the 1La transition is shifted more than
the 1Lb transition the permanent dipole moments would then follow the

trend 1

La)'1Lg>ground state. The small observed shifts in going from
other to alcohol can be explained in terms of an increasedclpole~dipole
interaction in the more polar ethanol.
1 1
In water the La and Lb transitions are blue-shifted relative to the
1L.

transition. These shifts are probably due to the higher acidity of the

spectrum in alcohol. The 1La transition is shifted more than the

water proton compared with the acidity of ethanol. This causes a blue
shift as a result of a solvent hydrogen interaction with the‘fi-election
charge density at the pyrrolic nitrogen. Dipole-dipole interactions
would cause a larger red shift in the more polar water compared with the
spectrum in ethanol. Thus the magnitude of the blue shift must exceed
such red shifts.

Although the water proton induced a blue-shift the spectrum in an
acid medium displayed a red-shift. It seems to indicate there are two
loci for solvent proton interactions which provide antagonistic spectra
shifts. The likely position for the interaction causing the blue-shift
is at the pyrrolic nitrogen. The 1T~electronic charge there was hYPO‘

thesized to be less in the ground than in the excited states to explain

118

the ether induced red-shift via H-bonding. Thus a solvent proton~fi~
electron interaction at this nitrogen would be manifested as a blue-
shift. The red-shift observed in an acidic medium could be due to an
interaction at the pyridinic nitrogen. This scheme nessitates the
excited states fi~electron density to exceed the ground state density
at the pyridinic nitrogen.

A curious aspect of this analysis is the blue-shift observation in
water compared with ethanol and acid. Similar shifts were observed for
indazole. The small red-shift observed in ethanol relative to ether
is probably due to the dominance of the dipolar interactions as well as
aza-nitrogen H-bonding over the blue-shift produced by the proton~fi~
electron interaction at the pyrrolic nitrogen. In water the slightly
more acidic proton interacts preferentially with the‘W-electronic charge
at the pyrrolic nitrogen to cause the blue-shift. This effect dominates
the other effects causing a blue shift relative to ethanol. When the
medium becomes acidic the protons interact with the pyridinic nitrogen
‘fi~electronic charge. In addition there are dipole-dipole interactions
with the net result of a red-shift.

The fact that subtle changes in solvent acidity cause drastic
differences in the solvent induced spectral shifts indicates that this
molecule possesses an unusual spectral instability toward solvent pro~
tons. Further evidence of the increased acidity of the pyrrolic hydro-
gen as a result of excitation is manifested by the excited state proton
tautomerization which was reported to occur for 7~azaindole.59
III. Fluorescence Spgctra.

In all solvents the 7~azaindole fluorescence was broad band and

exhibited some vibrational struc re especially at low temperatures.

119

There was a large Stokes' shift observed in all solvents which indicates
that the emitting state permanent dipole moment is much larger than the
ground state permanent dipole moment. The emitting state is assigned as
1L‘ on the basis of the vapor absorption data. The fluorescence is
progressively red-shifted through the solvent sequence 3~methylpentane,
ether, ethanol, water and acid. Although fluorescence was sought in a
basic solvent none was observed. This means that efficient quenching
via some radiationless route occurs for the anion. The emission in acid
is probably due to fluorescence from the cation whose emitting state is
stabilized at the pyridinic nitrogen by the protons. However the other
solvents are ordered by increasing dipole moment. This indicates that
the progressive Stokes' shifts are primarily induced by dipolar inter-
actions. The emission in water is red-shifted relative to the ethanol
spectrum.while the absorption is slightly blue-shifted. This dissimi-
larity further emphasizes’ the predominant importance of the dipole-
dipole relaxation effects in the Stokes' shifts. This reflects the
large disparity between the emitting and ground state permanent dipole
moments in 7~azaindole.

As mentioned in the prior section 7~azaindole exhibits a unique
photovinduced proton tautomerization which leads to emission from a
new species. This fluorescence occurs at 480 nm in 3-methylpentane or
ethanol and is broad band. It exhibits some vibrational structure at
low temperatures. This emission is not observed in dilute ether solu~
tions nor in water. The lack of fluorescence in water remains unex~
plained. The green emission presumably occurs when a hydrogen atom is

attached to pyridinic nitrogen and one is removed from the pyrrolic

120

nitrogen leading to an electronic rearrangement in the molecule. This
phenomenon is quite unusual and demonstrates the unique electronic
structure and reactivity of this molecule's excited state.
IV. Theoretical Calculations.

Results of these calculations are shown in Table 8 and Figure 18.
The permanent dipole moment magnitudes, transition moment orientations
and oscillator strengths all predict the lowest singlet excited state is
1

Lb. These transition moments do not parallel a primary molecular axis

and are predicted to lie at a 70' angle to each other. The permanent

dipole moment of the 1

La state is calculated to have a much larger mag-
nitude than the moment of the ground state. The two transitions are
predicted to be energetically close to each other. For the pyrrolic
nitrogen the acidity follows the trend 1La>>1Lg>ground state. For the
pyridinic nitrogen the basicity follows the trend 1La>1Lb>-ground
state. The charge density difference is quite sizeable between the
ground and 1La states at the pyrrolic nitrogen. This difference is
much larger than between the ground and 1Lb states at the pyrrolic

nitrogen as well as between the ground and either the 1L3 or the 1L

b
states at the pyridinic nitrogen. The internuclear distances do not
vary appreciably between the ground and either excited state. Finally
the ionization potential and electron affinity are calculated to be

7.1 eV and 1.8 eV respectively.

V. Correlations and Summary.

 

The calézlations predict and the vapor absorption data allude to
the nge-IA transition lying energetically below the 1Lgt-1A transition.

Fluorescence of the solvated molecule emanates from the 1La state. The

1

transition moments of the La and 1Lb states are probably oriented at an

121

Figure 18. Calculated Charge Densities and Bond Lengths for 7-Azaindole.

Legend:
Numbers at each atomic position denote’fiielectron
charge densities in electron units.
Numbers at each bond denote bond lengths in Angstroms.

The top number corresponds to the ground state.

The middle number corresponds to the first excited
singlet state ( Lh)'

The bottom number corresponds to the second excited
singlet state (lLb).

122

be;
2.14
moss;

 

mom
0mm

_
_

.mu Uhimm

 

 

«mod
mg;
30.0

omoé
«mo;
emoé

123
acute angle relative to each other and neither moment is expected to
parallel a molecular primary axis. The calculated permanent dipole
moments of the ground and excited states correlate well with the
observed Stokes' shifts. The penmanent dipole moments appear from the
observed solvent shifts to follow the sequence 1La>-1Lb>ground state.
This sequence is also predicted by the calculations. Thus the solvent
dipole moment interacts strongly with the 1La state moment to produce
the experimentally noted Stokes' shifts. This interaction seems to be
the dominant influence governing the shifts in various solvents.

The calculated charge densities for the ground and excited states
also correlates quite well with the absorption spectra shifts observed
between solvents. It appears that H-bonding alone at either the pyrrolic
or pyridinic nitrogen will induce a red-shift. However a solvent proton~
fi~ electron interaction at the pyrrolic nitrogen should induce a blue~
shift while a similar interaction at the pyridinic nitrogen should
induce a red-shift. The observed blue—shift when changing solvents from
ethanol to water appears to be due to an interaction between the more
acidic water proton and the pyrrolic nitrogen‘fibelectronic charge. The
preferential interaction at this center indicates a wider change between
the ground and excited state‘fi-electron densities there than at the
pyridinic nitrogen. This hypothesis correlates with the calculated
charge density values.

Both the critical solvent dependency of the absorption spectra and
the presence of emission from a proton tautomerized excited state
indicate that 7~azaindole possesses an unusual electron structure. One
of the unexplained mysteries of this structure is the absence of emis-

sion from the tautomer in water solution. The explanations of this

124

excited state proton tautomerization would not preclude this process
from occurring in water; but it does not appear to occur or at least is
not manifested as emission. Anyhow the uniqueness of the spectral pro~
perties of this molecule has prompted much work to be done at this labor-
atory in an attempt to further elucidate the electronic mechanisms
involved in provoking the observed spectra. Due to the energetic close-
ness of the 1La and 1Lb states a set of polarization experiments should
be performed to determine which state emits. Since the Stokes? shifts
for 7~azaindole are quite large one would be tempted to look again for
dual emission. A superposition of the 1La and 1Lb transitions and large
Stokes' shifts also characterized the indole spectra and fluorescence
apparently emanates from both states in that molecule. The polarization
measurements should also yield useful information about the tautomer
emission. At present there is no information available about the tauto-
mer transition moment relative to the transition moments of the non-
tautomerized species. The tautomer transition moment may not lie in the
same direction as the moment of the "normal" molecule.

Finally the vapor absorption spectra again show an increase of the
1La transition relative to the 1Lb transition as the temperature is
increased. This fact coincides with the vapor spectra observed for the
previous molecules. Thus again the vibronic coupling mechanism may be

invoked as a possible explanation of these results.

 

The

rich

trax

lat.

bro

ref

pa:

802

10‘

as

125

BENZOTRIAZOLE

I. Vapor Absorption Spectra.

Figure 19 shows the vapor absorption spectrum of benzotriazole.
The most striking feature of this spectrum is the noticeable lack of
rich, sharp, vibrational structure which characterized the 1Lgt-1A
transition of the previously examined molecules. There are some undu~
lations present but they are relegated to minor perturbations of the
broad lowest energy transition. This new spectral feature undoubtedly
reflects the additional replacement of a carbon with a nitrogen. This
particular replacement results in three continguous nitrogens in this
molecule. Two reasonably well separated transitions are apparent. The
lowest energy transition is centered at 275 nm and is tentatively
assigned as 1L6G-1A. The next transition centered at 242 nm is tenta-

1Lge-1A. The separation and assignment of these

tively assigned as
transitions coincides with the results obtained for indazole and
benzimidazole. However the 1Lb transition observed for benzotriazole is
much broader and more intense relative to the 1La transition. This is
unique compared with transitions observed for the previously discussed
azaindoles. A factor which must always be considered when interpreting
this spectrum is the possibility of an n-fi* transition. There are now
two nitrogens which could contribute to such a transition. Since this
transition is usually weak, broad and contains very diffuse structure it
could underlie the 1Lb transition. This would lead one to erroneously
think the 1Lb transition was intense. It is thus possible that an n ~fi*
trensition centered approximately at 275 nm occurs in benzotriazole.

Although the vibronic structure was slight, an attempt was made to

assign the 1Lb 0-0 transition and to locate vibrational sequences. The

126

 

 

 

 

 

 

mJONdEbOszm

.Bsuuoomm ooHumuomn< noom> oHouowuuonsom .mH omowwm
.e: 5330:.
o.- oo. 8a 8a 8a o8 68 one 8.
q 1 1 1 3 1 1 4 1 -
VII 1 m 8
r as
I ,/I, ed
// as
' /.. I} a
. /. \I\\ I/ /
I I/.. I... .I»IIII\. . a
f 2/1). 3
00m 00m 05m 00m
0.0
«.0
ed
0.0
0.0

0..

0.0.

127

 

 
   

 

 

 

 

 

 

 

 

 

\\ Mw_\\m ‘
— ,J/l/ ‘x\
w" ‘ Temp I6I° C
__ '4
fl
—
_ 4
.4
—
Temp IZB' C
_. j
_ —
_ -
Temp 99‘ C
__ .1
— -
P d
— —
Temp 66' C
p- -u
K "'
- fl ;1“'““1 ‘~51 v"-""."""\n_~nud~u_i -
l l l l l l l l I
220 230 240 250 260 270 200 290 300
WAVELENGTH (nm)
Figure 20. Benzotriazole Vapor Absorption Spectra as a Function

of Temperature.

128

1Lb 0~0 transition was assigned at 35826 cm.-1 (2791 A) and other

sequences in this transition appeared to occur at 266 and 272 nm. The
average separation between these 1Lb transition sequences is 820 cmf1.
The 1La transition vibrational groups appear to occur at 238, 242, and

249 nm with an average separation of 937 cm“1

. The 1La transition 0~0
remains unassigned. It was again observed that the higher energy meme
bers of the 1Lb transition sequences again consisted of doublets with
an average individual peak separation of 16 cm.1 within each doublet.
It must be emphasized thatthe sequence and 0-0 assignments are quite
tentative and based primarily on intuition fostered by similar assign-
ments for the previous molecules.

As for the previous molecules a hot band analysis was carried out
for benzotriazole. The results are depicted in Figure 20. Contrary to
the results obtained for the previous molecules it is apparently
impossible to discern a temperature dependent growth of the 1La transi-
tion relative to the 1Lb transition. Both transitions seem to grow
simultaneously by the same amount. It seems strange that this hereto~
fore expected result is an exception rather than the rule for these
molecules. The broadness of the 1Lb transition probably contributes to
the temperature dependent simultaneous growth phenomenon observed for
this molecule. Thus either this result is an optical illusion or the
1Lb transition contains some character which allows its intensity to
vary in the same manner as the 1La transition.

Confirmation of the 1Lb 0-0 transition was impossible to achieve
with this hot band analysis. The vibronic structure was so prohibitively

broad that such a task was incapable of being performed. However an

effect was noted which was similar to one observed for benzimidazole. The

129
first benzotriazole vibronic peak to appear as the temperature was
slowly increased occurred at 290 um. Its origin remains unexplained.
II. Polarization Measurements of Fluorescence Excitation.

The only reported measurements of this type were those of Schutt

14

and Zimmermann in ethanol at ~180°C. Monitoring the fluorescence at

330 nm they found an increase in polarization from 0.15 to 0.35 as the

excitation wavelength was increased from 255 to 300 nm. These data

1

indicate that the Lb state emits and that there is some 1L transition
a

character in the excitation spectrum down to 300 nm. Since the polari-

zation values never reach 0.5 one may posit some character other than

1Lb in the lowest excitation band. This character may be either 1

or possibly n-fl* in nature. The absorption spectrum in ethanol does
1

La
exhibit a general diffuseness and merging of the 1La and Lb bands so
there is probably character of both transitions in varying proportions
from 250 to 300 nm. The two fi=fl* transition moments would not be
expected to be parallel to each other and the nonnegative polarization
values in the predominantly 1La transition region of the spectrum
indicates these moments are also not perpendicular.

When Schutt and Zimmermann made excitation polarization measurer
ments for the cation the observations differed from those seen with the

neutral molecule. The 1

La and 1Lb transitions had merged into one
broad band whose peak occurred midway between the respective transition
maxima of the neutral molecule. The polarization was 0.3 at 240 nm,
decreased to about zero at the transition maximum and then rapidly
increased to approximately 0.5 at 300 nm. These results show that the

cation and neutral species have quite different spectral features. In

fact these are the most drastic differences yet observed for this group

130

of molecules when comparing a charged with a neutral species. It appears
that fluorescence again emanates from the 1Lb state in the cation. How-
ever the transition shifts and polarization data strongly suggest an n~fl*
transition underlies the 1Lb transition of the neutral molecule. This
nrfl* transition is blue shifted in the cation as would be expected.
Together with the 1La transition this blue-shifted n~fl* transition could
give the appearance of a 1Lb and 1La transition merger as well as the
observed polarization minimum. The increase in polarization at higher
energies is probably due to the more pprallel orientation of the 1La and
1Lb transition moments compared with the more perpendicular orientation

of the 1

Lb and n~fi* transition moments. In addition the 18b transition
moment which probably is closely parallel to the 1Lb moment could con-
tribute to the more positive polarization at the short wavelengths. It
must be iterated that this interpretation is very tentative and more
experimental data must be obtained for support of this hypothesis. The
strict merger of the 1La and 1Lb transitions does not explain the polar~
ization data unless one postulates a very broad 1Lb transition upon
which a narrow 1La transition is superposed. This picture is exactly
opposite the one constructed for all the previously discussed molecules
which had much supportive data.

III. Solution Absorption Spectra.

The data in Table 7 show the characteristic broadening of the
vibronic structure which occurs in hydrocarbon solvents compared with
the vapor spectra. However as the solvents are changed in the sequence
hydrocarbon, ethanol and water the 1Lb transition blue-shifts and the
1

La transition red-shifts until they become one broad band in the latter

solvent. Antagonistic dipole-dipole and H-bonding interactions are

131

probably responsible for this merger. The lower acidity and smaller
dipole moment of ethanol compared with those of water explains the smal-

1

ler shifts of the bands in ethanol. An n~fl* band underlying the L

b
transition band would undergo a blue shift in changing solvents from
hydrocarbon to ethanol to water. This may contribute to the appearance
of a single absorption band in water.

The blue shift of the 1Lb band and the red shift of the 1La band
in going from hydrocarbon to ethanol is unique and can be explained in
terms of an increase in the charge densities at one nitrogen upon
excitation and a simultaneous decrease of the charge densities at
another nitrogen. This situation is not unlikely. However the magni-
tude of these changes must be of such proportions that H~bonding at
either site are equally probably in order to promote a concurrent red
and blue shift of the two transitions. In addition one site must have
the interaction primarily affecting the 1Lb state while at the other
site the interaction primarily affects the 1La state.

IV. Fluorescence Spectra.

The fluorescence of benzotriazole is quite broad and exhibits
vibronic structure. In addition a sizeable Stokes' shift is evident in
polar solvents. This indicates that the emitting excited state perman-
ent dipole moment is muchdarger than the ground state's moment. The
fluorescence band is red-shifted with the solvent sequence ethanol,
water and acidic medium. This reflects excited state stabilization
via dipolar and H~bonding interactions. The relative importance of
each of these interactions in promoting the stabilization cannot be

assessed from these data.

132

Fluorescence polarization measurements in ethanol were slightly
positive and flat throughout the emission spectrum.14 Since excita-
tion was monitored midway between the 1La and 1Lb transition bands the
low polarization value indicates the mutual overlap of these bands.
The flatness of this polarization spectrum shows that emission
emanates from a single state, probably the 1Lb state. The same fluo-
rescence polarization results were also obtained for the cation.

Again the transition overlap could explain these observations.
V. Phosphorescence Spectra.

The data of Schutt and Zimmermann show phosphorescent emission for
both the neutral molecule and the cation. Both emissions display some
vibronic structure. The phosphorescence of the cation is somewhat red-
shifted relative to the neutral molecule's emission. This red-shift
probably reflects the excited state stabilization due to H~bonding.

Phosphorescence excitation and emission polarization measurements
were performed by Schutt and Zimmermann for the benzotriazole cation.
The excitation polarization was negative indicating that the phosphor—
escence is predominantly polarized out-of—plane. The fact that the
polarization was not ~0.33 probably means either that the triplet
transition is not exactly perpendicular to the plane of the molecular

nuclei or that there is some n-fi* transition contribution in the region

of this excitation polarization spectrum.

VI. lggrrelations and Summary.

Benzotriazole exhibits two distinct transition bands in the vapor
phase and in hydrocarbon solution. The lowest energy transition is
assigned 1L1'-----1A and the next higher transition is assigned 1L<s~1A.

b a
The vapor spectrum does not show the well resolved vibronic structure

133

observed with the previously discussed molecules. The 1Lb transition
for this molecule appears to be more intense than previously observed.
The breadth of this transition makes a distinction between it and the
1La transition more difficult to perceive when compared with a similar
distinction for indazole or benzimidazole. In polar solvents these
transitions merge such that they become almost indistinguishable. In
fact the cation's absorption spectrum exhibits only a single peak whose
position is midway between the peaks observed in hydrocarbon solvents.
One of the difficulties fostored by the lack of vibronic resolution
in the vapor spectra is the difficulty in assigning the 1Lb 0-0 transi-
tion. It was tentatively assigned at 2791 A on the basis of the 1L

b
transition band shape in comparison with the 1L bands of the previously

b
discussed molecules.
Another "unique" feature of the benzotriazole vapor spectra was the
lack of a disproportionate intensity growth for the 1La transition com~

1

pared with the Lb transition as the temperature was increased. This

result adds further support for the 1La transition intensity growth
observed for the previously discussed molecules not being an experi-
mental artifact. Using the same interpretation as previously discussed
vibronic coupling apparently does not occur for benzotriazole. The
reason for the lack (or even presence) of this vibronic coupling remains
obscure.

Since a number of spectral features for benzotriazole differed
from those of previously discussed molecules it is obvious that more
spectral work should be concentrated on this molecule. It is evident

that the introduction of a second aza nitrogen has a noticeable effect

on the spectral properties and characteristics of the indoles.

134

4-AZABENZIMIDAZOLE

l. Vapor Absorption Spectra.

 

The vapor absorption spectrum of 4~azabenzimidazole is shown in
Figure 21. This spectrum exhibits two well-separated bands centered at
235 and 280 nm. The lower energy transition displays some resolved
vibronic structure although not as much as was observed for indole and
the mono-azaindoles. The higher energy transition is quite broad and
displays very diffuse vibronic structure. Comparing this spectrum with
those of the previously discussed molecules the higher energy transi-

tion is assigned 1LgL-1A and the lower energy transition is assigned

ngé—lA. For this molecule the 1Lb transition intensity is larger than
that of the 1La transition. This observation is Opposite the results
obtained for indole and the mono-azaindoles. Even for benzotriazole

the two transition were at most of equal intensity. A possible explan-
ation of this large 1Lb transition intensity is that an n-fi* transi—
tion underlies thisfi—fi* transition. If so one would expect the presumed

fl

n~u* transition to significantly blue-shift in H~bonding solvents.

The 1

La 0-0 transition remains unassigned but for the 1Lb transi~
tion the 0~0 is assigned at 35201 cm-1 (2833 A). Since the higher
energy members of the 1Lb 0-0 transition sequence were actually doublets
the highest energy peak of the first doublet was chosen as the 1Lb 0-0.
Other vibrational sequences in the 1Lb transition occur at 272 and 277
nm with an average sequence separation of 720 cm-1. Vibrational groups
in the 1La transition occur at 232, 236 and 239 nm with an average
sparation of 800 cm-1.

The results of the hot-band "analysis" for this molecule are shown

in Figure 22. These results indicate that the above 1Lb 0~0 transition

.Eouuooom sewuouOmo< oomm> oHomooHEmNcoomN<Io .HN ooswmm

.m: 35:30:-

 

 

 

 

 

 

 

OON 000 2a 0.“ 000 00¢ 80 80
q 8 4 1 o 1 J o 1 06
I /ll\\l/L a
1 2\ 1 e0
. as
. l 0.0
r .. 3
f- 1 0..
00m 00m 0km 00m
U 0.0
«.0
0.0
dd
0.0
0.0
0..

m40N<o=§Nzwm<N<Iv

136

'Temp ICI’ c

 

Temp ISO‘ C

0.0.

 

 

Temp I00“ C

 

' Temp 26' C

 

280 zoo no no no see no soo
IVELEIGTH (as)

Figure 22. 4~Azabenzimidazole Vapor Absorption Spectra as a Function
of Temperature.

137
assignment was correct. Another observation is that the 1La and 1Lb
transitions do not exhibit any intensity growth disprOportion between
them. This result is similar to that observed for benzotriazole. Thus
it does not appear that any intensity borrowing occurs for the 1La and
1Lb transitions in 4~azabenzimidazole.

However the temperature dependent spectra do exhibit the existence
of a vibronic series of transitions at 288 nm before the 1Lb 0-0 transi-
tion sequence appears. This phenomenon was also noticed with benzimi-
dazole and benzotriazole and remains unexplained.

II. Absorption and Fluorescence Spectra

 

The solution absorption data in Table 7 indicate very little or no
shifts when comparing Spectra obtained in cyclohexane with those obtained
in water. There appears to be a red-shift for the cation relative to
the spectrum in water. The lack of spectral shifts is probably due to
different specific solvent-solute interactions which induce antagonistic
shifts. It would be expected that the excited state permanent dipole
moment magnitudes exceed the ground state moment magnitude. This would
lead to a demonstrable red-shift in water relative to the spectrum in
hydrocarbon solvents. The lack of any shift indicates that there is an
H~bonding interaction at a site whose excited state charge density is
less than the ground state charge density. The resulting blue-shift
would tend to cancel the dipolar interaction induced red-shift. The
net result would be little, if any, spectral shift.

The apparent red-shift which occurs in acidic media is probably
caused by an interaction between the "excess" protons and a'fi-electron

site in the molecule whose excited state charge density exceeds its ground

state density. The most likely position for this interaction is one of

138

the pyridinic nitrogens. Obviously this site must differ from the one
described above which was the center for a prOposed blue-shift inter-
action. That center could be the other pyridinic nitrogen but more
probably would be the pyrrolic nitrogen.

In a basic solvent there is a red-shift of both hands. This shift
probably reflects the effect of the negative charge at the pyrrolic
nitrogen on the fi-electron system. The Coulombic repulsion presumably
destabilizes the fi~electronic structure. Thus less energy is required
to excite this structure.

The fluorescence of this molecule probably emanates form the lLb
state. It exhibits a moderate Stokes' shift in polar solvents. This
indicates that the emitting state permanent dipole moment is larger
than the ground state moment. This coincides with the above assumption
concerning the dipolar interaction effects on the absorption spectra.
The fluorescence maximum in water is blue-shifted compared with that in
ethanol. This ihift could be caused by the solvent proton~fi~ electron
interaction at a site whose ground state fl-electron charge density
exceeds the emitting state charge density. This mechanism corresponds
with one of those proposed above to explain the lack of absorption spec-
trum shift between hydrocarbon and water solvents. Thus either a pyri-
dinic nitrogen or the pyrrolic nitrogen is a candidate for the interac-
tion site. The most likely choice is the pyrrolic nitrogen. Fluores-
cence from the cation is slightly red-shifted compared with the spectrum
in water. This corresponds with the observed absorption spectrum shift

of the cation vs. the neutral species.

139

III. Theoretical Calculations.

 

Using the calculated oscillator strengths of Table 8 the first
excited state was assigned 1Lb and the second excited state was assigned
1La° It should be recalled that normally the 1La transition oscillator
strength exceeds that of the 1Lb transition. The calculated transition
moment orientations were of little use in making this assignment since
they were oriented at approximately the same angle from the principal
axes of the molecule. However these calculated orientations do not
parallel a principal axis. Their mutual orientation, though, is pre~
dicted to be almost perpendicular. The predicted energies of these two
lowest transitions are not too disparate. The calculated ionization
potential and electron affinity of 4~azabenzimfldaeole are respectively
7.3 eV and 1.8 eV. Calculations were not performed for either the
excited state permanent dipole moments or their‘fi-electIOn charge densi-
ties.

IV. Correlations and Summary.

The first (lLb) and second (1L3) singlet transitions are energeti~
cally quite well separated. Although this fact is well established in
the vapor absorption spectra it is not predicated by the calculations.
This demonstrates the possible amount of error inherent in these calcu-
lations. However the error in predicting the second singlet transition
is less than 0.5 eV so it is not too unsatisfactory. The vapor absorp-

1

tion spectra also show that the Lb transition oscillator strength

apparently exceeds the 1La transition oscillator strength. This was
not observed for any of the previously discussed molecules. A possible
explanation of this occurrence is that one or more n~fl* transitions

1

underlietthe Lb transition. Thus the lowest energy transition band

140
would be composed of several transitions. Although no supportive evi-
dence for the existence of an ndfl* transition was found, its occurrence
still cannot be eliminated since very little spectral data has been
obtained for this molecule.

Fluorescence probably emanates from the 1L1) state. Spectral shifts
of both fluorescence and absorption indicate there are a number of pos~
sible solvent-solute interactions. The moderate Stokes' shifts imply
that the emitting state permanent dipole moment exceeds the ground state
moment. Thus polar solvent-solute interactions must occur and cause
absorption spectra red-shifts when compared with the spectra in hydro-
carbon solvents. However no such shift was evident for water solutions.
Some antagonistic blue-shift must be occurring in thissolvent. This
shift could be caused by a solvent proton~W~ electron interaction at a
site whose‘fi-electronic charge density in the ground state is larger
than in the excited states. This site is probably the pyrrolic nitrogen
or possibly a pyridinic nitrogen. This scheme also could explain the
blue-shift of the fluorescence in water compared with the emission in
ethanol. In acidic media the fluorescence and absorption spectra are
red-shifted relative to the spectra in neutral media. Apparently the
additional protons interact at another ~electron site whose excited charge
density exceeds its ground state charge density. This site is probably
a pyridinic nitrogen. The anion absorption spectrum is red-shifted
relative to the spectrum of the neutral species. This reflects the
destabilization of the‘fi-electron structure which is a result of the
Coulombic repulsion between the'fi~electron system and the negative

charge at the pyrrolic nitrogen.

141

The vapor absorption spectra obtained as function of temperature
indicate that there is no intensity stealing which occurs for this mole~
cule. That is, both 1L8 and 1Lb transition oscillator strengths main~
tain the same relative proportion throughout the range of utilized tem-
peratures. This proportion constancy further supports the experimental
evidence of relative oscillator strength prOportionality changes which
were obtained for many of the previously discussed molecules.

Due to the lack of symmetry the lowest two singlet transition
moments are not expected to be oriented parallel to the principal axes
of the molecule. This correlates well with the calculated moment orien-

tations. However the calculated moments for the two lowest transitions

are approximately perpendicular. This occurrence may be coincidental.

S-AZABENZIMIDAZOLE
I. Absorption and Fluorescence Spectra.

Again Table 7 shows there are scanty data available for 5~azabenzi~
midazole. There appears to be a blue-shift in the absorption spectrum
when the solvent is changed from water to an acidic medium. There is a
progressive blue-shift in the fluorescence spectrum when the solvent
sequence of ethanol, water and acid is followed. A moderate Stokes'
shift is also observed in polar solvents. This indicates that the
emitting state's permanent dipole moment is somewhat larger than the
ground state's moment. Thus one would expect dipolar interaction
induced red-shifts of the absorption and fluorescence spectra in more
polar solvents. The fact that these red-shifts are not reported means
that some other interaction is occurring which produces the observed
blue-shifts. This interaction is probably between solvent protons and

the'fi-electronic charge at certain sites on the molecule. All three

142

nitrogens interact with the solvent protons to produce these shifts.
Their ground state/fi-electronic charge densities must exceed their
excited state densities to cause these blue-shifts. The difference
between the ground and excited state‘fl-electron densities probably
differs for these three nitrogens. Thus as the acidity of the solvent
increases, more nitrogens become involved in this blue-shift producing
interaction. For example in acidic media the additional proton could
predominately interact at a nitrogen which had hardly been involved in
the interactions with water and ethanol. The feebleness of interaction
for this nitrogen in these latter solvents would probably be due to the
relatively small’fi-electron density change between the ground and excited
states compared with the change at the other two nitrogens.

11. Theoretical Calculations.

 

Based on the calculated results in Table 8 the first singlet excited
state is assigned as 1Lb and the second singlet excited state is assigned
as 1L3. This assignment is based on the calculated oscillator strengths
only since the first two transition moment orientations were at approxi-
mately the same angle but at opposite directions to the principal axes
of the molecule. These transition moment orientations are predicted to
be neither parallel to a principal axis nor perpendicular to each other.
The two lowest singlet transitions are predicted to lie energetically
close to each other. The excited state permanent dipole moments and
'fi-electron charge densities were not calculated for this molecule. The
ionization potential and electron affinity were predicted to be 7.3 eV

and 1.8 eV respectively.

143

III. Correlations and Summary.

The sequence of blue-shifts in both the fluorescence and absorption
spectra which was reported from 5~azabenzimidazole has not been observed
for any of.the previously discussed molecules. This result may indicate
that all three nitrogens have the same‘fi-electronic charge density pattern
between the ground and excited state. That is, the ground state charge
density exceeds that in the excited states. This pattern explains the
observed blue-shifts as the medium becomes more acidic provided these
shifts are caused by solvent protondfi-electron interactions. If this
interpretation becomes verified, 5~azabenzimidazole is a unique indole
since it contains three nitrogens with the same fihelectron density change
pattern between the ground and excited states.

The first singlet excited state is assigned as 1Lb and the second
singlet excited state is assigned as 1La. Their transition moments are
not expected to parallel a molecular principal axis or to be perpendicu-
lar to each other. The fluorescence probably emanates from the 1Lb
state. No n~fl* transition has been reported for this molecule. However
on the basis of the data analyses for benzotriazole and 4~azabenzimi-

dazole an n~fl* transition may occur in 5-azabenzimidazole and probably

underlies the more intense fi~fi* transitions.

PURINE

I. Vgpor Absorption Spectra.
The vapor absorption spectrum of purine has been reported by El-

Bayoumi, et. a1.8

It is identical with the spectrum obtained in this
laboratory and shown in Figure 23. The most remarkable characteristic

of this spectrum is the complete lack of sharp vibrational structure

144

00m

00m

00m

2::
oz

00m

.Esuuoomm sceoouomo< ooom> oseusm

Ihozw4w><>>

nxuN

Agog“

0mm

.mm spamma

ONN

 

TuPII ~II

 

_

a

szDA.

.

_

_

—

_

_ III.

 

.0 .0

145
which was evident in indole and the mono and di-azaindoles. That is,
in this triazaindole the existing vibrational structure has very broad,
overlapping envelopes. Overall the features of this absorption spectrum
resemble those of the azaindoles recorded in a polar solvent. However,
there is even less evident vibronic structure in the purine vapor spec-
trum than was often observed in the solution spectra of the previously
discussed molecules. For this reason transition assignments and spectral
analyses based on solvent shifts are difficult to perform and were not
attempted here. In addition a hot band analysis was not attempted. One
of the factors which thwarted this type of analysis was the low vapor
pressure of purine. Even when the solid was introduced directly into
the 10 cm absorption cell a relatively high temperature was needed to
obtain a vapor absorption spectrum.
II. Transition Assiggments.
Assignment of the transitions was first made by Clark and Tinoco60
from a methylcyclohexane solution absorption spectrum. The first‘fi~fi*
transition had three broad peaks at 259, 262, and 267 nm. A less pro-
nounced, weak peak at about 240 nmeas ascribed to the second fi~fi*
transition by comparing purine's absorption spectrum with that of benzi-
midazole. Purine also exhibited a long absorption tail from 280 ~ 310
nm which was ascribed as an n-fi* transition. When these authors used
trimethylphosphate as a solvent they noticed that the firstfi~fi* transi~
tion became more rounded with a loss of vibrational structure and was
slightly red-shifted. The secondfi~fi* transition was more red-shifted
and lost absorptivity. In addition a third'fi-fi* transition was found

with peaks at 190 and 200 nm.

146

A different assignment was made on the basis of polarized reflec-

61 It should be noted

tance measurements of a single purine crystal.
that these assignments were based on a numerical analysis of the
reflectance measurements and are thus subject to any assumptions made
when performing this analysis. The transitions at 294 and 250 nm were
assigned as out-of-plane n~fl* transitions originating from nitrogens
at positions 5 and 7 (using the same numbering system as that for
indole). Then-0* transition at 263 nm was found to lie in—plane and
had its moment lying 48° counterclockcwise from the short molecular
axis-~that axis formed by the line from C8 to C9. The'fi-fi* transition
at 200 nm was also in—plane and had its moment 51° counterclockwise
from the short axis. The strong transition at 190 nm had its transi-
tion moment out-of-plane and was unassigned regarding its character.
These two analyses show a major discrepancy in the assignment of
the transition at 245 nm. This discrepancy is more remarkable since
Clark was involved in both assignment attempts. However both analyses
do show an n~fi* transition at lower energies than the lowest energy
fibfl* transition. Thus an n~fi* transition becomes readily apparent when
there are three aza nitrogens in the indole skeleton. A feature of the
crystal polarization measurements worth noting is that the long wave-
length.‘fi~fi* transition moment (at 263 nm) was not found to be polar-
ized parallel to one of the principal axes of the molecule. This is

a result which has been expected for all the indoles.

III. Absorption and Fluorescence Spectra.

 

The available absorption and emission data for purine in various
solvents are shown in Table 9. Another series of experiments were

performed on purine which monitored absorption or fluorescence as a

147

Table 9. Purine Absorption and Luminescence.

 

F

 

 

 

 

 

 

 

 

 

 

 

 

 

Solvent Temp . )s abs f luor X phos RBf .
Me—Cyclohex. RT 240, 262, 280 60
Me-Cyclohex. RT none 62, 63
Me-Cyclohex. 77°K 351, 370, 395 62

Dichloroethane RT 377 62
Ispropanol RT 263 only 62
Methanol RT 362 62

Ethanol RT 310 350 54

E P A 77°K 263 364, 380, 400 64
Glycerol 77°K 351, 370, 395 62
Water RT 263 387 62
Water 77°K 351, 370, 395 62
Water RT 281 370 54
Water RT 380 65

 

 

 

 

 

 

 

 

148

function of pH. These results are listed in Table 10. The existence

of conflicting data as well as the breadth and lack of vibronic struc-
ture in the absorption spectra do not allow an analysis based on
solvent shifts. Although no fluorescence spectrum has been published
the data from tables indicate that this emission is also broad band.
Some error in reporting the fluorescence maxima would be expected.

The data do show that a moderate Stokes' shift occurs in polar solvents.
This indicates that the emitting state permanent dipole moment exceeds
the ground state moment. Thus, in general, dipolar interactions should
produce red-shifts.

Although some of thhse data are conflicting certain general trends
seem to be evident and are used to hypothesize what is occurring in the
molecule which lead to these spectroscopic results. In hydrocarbon
non-hydrogen bonding solvents the lone pair electrons are unperturbed
and an n~fi* transition has lowest energy. Since its oscillator strength
is very low its fluorescence lifetime is quite long. Thus radiationless
processes may easily occur before fluorescence can take place. In this
case very little if any fluorescence would be expected and little occurs
according to the data. However if purine is put in a hydrogen bonding
solvent these lone pair electrons are stabilized which in turn shifts
the n~fi* transition to higher energies. Now a‘fi-W* transition is of
lowest energy and fluorescence is predicted and observed. Dipolar and
solvent proton ~fi~ electron interactions also probably occur in the po-
lar solvents which alter the energy of fluorescence. These perturba-
tions would be manifested as solvent shifts. The data show that such

shifts occur.

149

 

 

 

 

 

 

 

Table 10. Effects of pH on Purine Absorption or Fluorescence.
A or F )‘max pKa 4’:
(220, 260 @ pH 0.28 2.39
A .(220’ 263 @ pH 5.70 8.93
219, 271 @ pH 11.0
400 @ pH ‘2 3.2 0.008 @ pH 2.46
F 380 @ pH 5.2 9.2 0.002 @ pH 8
370 @ pH 11 0.045 @ pH 10.6

 

 

 

 

 

 

 

 

 

 

TabIe 11. Calculation Results for Purine.

Trans. Trans. Trans. Trans. f

Energy Energy Moment Polar

(cmT ) (nm) (Debyes) (deg. from

x~axis*)

36425 274.5 0.384 81 0.058
36542 273.7 0.598 ‘33 0.142
48889 204.5 0.982 72 0.512
49958 200.2 0.564 ~43 0.172

 

 

*See Figure 24

150

On the basis of this scheme Borreson analyzed the pH effect on the
purine fluorescence.65 From pH 3.2 to pH 9.2 the n-fi‘* transition has
lowest energy and little fluorescence is expected or observed. Between
pH 2.5 and pH 3.2 protons are stabilizing the lone pair electrons causing
the ndfl* transition to be stabilized relative to the lowest energyT~fl*
transition and fluorescence is predicted and observed. Below pH 2.5 an
n~fi* transition is again the lowest energy transition due to a stabi-
lising of the electrons by the additional protons and fluorescence is
not expected nor observed. Above pH 9.2 the pyrrolic nitrogen has a
negative charge since the base has pulled a proton away from it.
Borreson believes this negative charge is pulled up into the system.
He states that further evidence for this attraction is provided by
the experimental result that this pKa (9.2) is lower than that of
other azaindoles such as benzimidazole. This additional charge in the
‘fisystem repels the lone pair electrons thus stabilizing them. The‘fi4fi*
transition now has lowest energy and fluorescence is predicted and
observed. Borreson's interpretation of the anion fluorescence is vague
and does not explain the anion red shift in absorption. An alternative
explanation for the existence of anion fluorescence is that the
Coulombic repulsion due to the negative charge on the pyrrolic nitrogen
causes a rearrangement of the fi-electron system. This rearrangement
could have such a destabilizing effect that the fi-fi* transition energy
would be b ow the n~fl* transition energy. Fluorescence would be
expected and is observed. This explanation would explain the observed
absorption spectra shifts between the neutral and anion species infi

Table 10.

151

It should be noted at this point that some investigators report fluores-
cence occurrence for purine under certain experimental conditions while
other do not. This is due to the relative sensitivity and signal-to—
noise figures for various experimental setups. Purine has a low fluores-
cent quantum yield, especially at neutral pH's. In a 50:50 propane-1,2-
diol: 0.01M phosphate buffer solution at pH 7 the quantum yield is
0.0263 (see also Table 10).

IV. Phosphorescence Spectra.

In contrast to these phenomena regarding fluorescence, ph08phores-
cence is readily observed in most solvents (see Table 9). For example
in propane-dial phosphate buffer the phosphorescence to fluorescence
ratio is 1.963 so phosphorescence becomes more readily observed. From
excitation polarization studies the phosphorescence is concluded to be
emitted from the‘fi~fi* triplet state. That is, when excitation was via
the lowest energy fi~fi* transition the excitation polarization was nega~
tive while n~fi* excitation gave positive.polarization.64 This is con.-
sistent with the notion thati%fi* phosphorescence is out-of-plane
polarized making it perpendicular t0‘fi-fi* singlet transitions and
parallel to n-fi* transitions. In addition the relatively long lifetimes
of observed phosphorescence (1.6 sec) and large singlet-triplet separa-
tion energies are consistent with the fi-fi* phosphorescence emission
conclusion.

Another set of phosphorescence excitation polarization experiments
confirmed the negative polarization when the lowest fi-fi* transition was
excited.66 In addition these investigators found that excitation via
the second‘fi~fi* transition (the one at 240 nm in Clark and Tinoco's

assignment scheme) also gave negative polarizations. This further

152

indicated that phosphorescence emission is out-of-plane. Their results
also indicated the two in—plane‘fizfih'transitions were not parallel but
they could not say these transitions were perpendicular with any certi-
tude. They also noted some structure in their phosphorescence emission
polarization measurements which maintained position and magnitude when
excitation was either the first or second‘fiLiF transition. The struc~
tural features consisted of relative maxima and minima of the negative
polarization which were quite well aligned with structural features on
the phosphorescence emission itself. The explanation given for this
structure was that the more negative minima were due to direct spin-
orbit interaction or coupling of the n~fi* singlet and the emittingfiiflk
triplet while the less negative maxima were due to vibronic interactions
between the n-TI‘" triplet and the 1011): triplet.

V. Theoretical Calculations.

The results of these calculations are shown in Table 11 and Figure
24. Provided one may still use the same excited state nomenclature as
was used for the other molecules the first singlet excited state is
assigned as 1Lb and the second singlet excited state is assigned as 1La.
The transition moment orientations, transition oscillator strengths and
the ground and excited state permanent dipole moments all agree with
this assignment. Neither transition is predicted to parallel a prin-
cipal axis of the molecule. Also the two transitions are not predicted
to be mutually perpendicular.

The two lowest singlet transitions are calculated to lie energe-
tically very close to each other. The close proximity of these transi-
tions does not allow an assignment of the fluorescence emitting state

1

to be made. However the permanent dipole moment of the La state is

much lower than the ground state dipole moment. This correlates with the

153

Figure 24. Calculated Charge Densities and Bond Lengths for Purine.

Legend:
Numbers at each atomic position denote‘flbelectron
charge densities in electron units.
Numbers at each bond denote bond lengths in Angstrom.

The top number corresponds to the ground state.

The middle number corresponds to the first excited
singlet state (lLb).

The bottom number corresponds to the second excited
singlet state (1L3).

154

com

alw—

 

 

eqm ooswmm

omm..
cow.—
om...

 

mom
o—slnsf
I—IOv-i
.......-3

$00..
oqo.0

155

experimentally observed Stokes' shifts in polar solvents. The correla-
tion suggests that fluorescence emanates from the 1La state in polar
media.

The calculated bond lengths do not vary drastically between the
ground and excited states. At the pyrrolic nitrogen position the
calculated‘fi~ electronic charge density does not noticeably change
between the ground and 1Lb state. However this nitrogen is predicted

1

to be much more acidiic in the La state than in either the ground or

1Lb states. The calculated basicities at the pyridinic nitrogen posi-

l
tions are: 3 position ~ ground >1Lb> La; 5 position ~ 11's) ground>

1L'7'ground. It is interesting to note that

b
the magnitude and direction of solvent induced shift as a result of

1L ; 7 position - 1L a»
b a

H~bonding with aza nitrogens depends on the position of that nitrogen
and the excited state (1La or 1Lb). The ionization potential is cal-
culated to be 7.4 eV and the electron affinity is calculated to be
1.8 eV.

VI. Correlations and Summgry.

Purine exhibits two features which were not readily apparent for
the rest of the azaindoles. First there is a defineable n-fi* transi-
tion which occurs at longer wavelengths than the longest energy€>fi*
transition. The existence of this transition has been verified by the
vapor absorption spectra, polarization data and the solution absorp-
tion experiments. Second, very little vibronic structure is evident in
the‘fi-fi'* transitions. This structure is even absent in the vapor
absorption spectrum. It is quite possible that these two features are

a manifestation of the incorporation of a third aza nitrogen into the

indole molecule.

156

Some controversy has been raised concerning the type of transition
which occurs at 245 um. I believe this is the second (1L8)fi~€* transi-
tion. This belief is based on anumber of facts. First there have been
twoi=fi* transitions in the range 220 to 300 nm for every molecule in
this study. The second transition for indazole, benzimidazole, benzo-
triazole, and the azabenzimidazoles have occured in the region of 240
nm. Purine may be considered as a diazabenzbmidazole so it also should
have its second‘fi~fi* transition in this region. In addition the phos-
phorescence excitation and emission spectra“,66 indicate this is a
‘fi~fi* transition. Also if this transition were n~fi* it should be blue-
shifted in polarprotic solvents. The opposite result was experimentally
observed for the only published spectra.62 In fact the observed red~
shift was so large that it partially merged the twoi~17* transitions.
This red-shift would be expected and large if the appropriate excited
state (1L8)permanent dipole moment significantly exceeded the ground
state moment. This trend is predicted by the calculations. The only
data which refutes this assignment of the 245 nm transition is that of
Chen and Clark,61 Their data analysis is contingent on certain assump~
tions and is somewhat subjective. Perhaps some of these assumptions
aren't fulfilled.

The calculations position the 1Lb transition moment at 82' and
the 1La transition moment at ~34' relative to the axis formed by C8 and
09 in the molecule. Chen and Clarks' data place the lowest energy
fl=fl* transition moment at ~48“. Although the accuracy of calculated
angles is not good this discrepancy is somewhat unsettling and warrants

further investigation. Possible Chen«and Clark were observing the 1La

transition but this seems unlikely. Anyhow the two singlet transition

157

moments are not predicted to parallel a molecular principal axis nor
be perpendicular to each other. This is an unexpected result.

The phosphorescence emission polarization reported by Drobnik, et.
al.66 exhibited some spectral structure. These authors attributed
this structure to the relative importance, across the phosphorescence
spectrum, of the vibronic mixing of the triplet (n,fl*) and (17.1%) states
and the direct spin-orbit coupling of the singlet (n;fl*) state and the
triplet (fl,fi*) state. An alternative explanation for this structure
is dual emission polarization structure is dual emission from the 3La
and 3Lb states. This phosphorescent emission polarization structure is
reminiscent of that observed for indole and an interpretation of that
polarization structure was dual emission. In polar solvents, at least,
the two lowest singlet transitions apparently are energetically close.
The lowest singlet transition was assigned as 1Lb and the second transi-
tion was assigned as 1La. However the fluorescence emittimgstate was
assigned as 1La on the basis of the large Stokes' shifts exhibited in
polar solvents. There is no guarantee that such shifts would be present
in nonpolar solvents. Thus one would be tempted to investigate the
existence of dual emission.

The calculated charge densities are also in accord with the
experimental observation of electrophilic attack of ethanol at the 4
position when purine is in the excited state.70 This carbon's cal-
culated‘fiBelectron density changes more than any other carbon atom in
the molecule upon excitation. In addition its excited state density
is only rivaled by the‘n~electron excited state density of a carbon at
the 2 position. Experimentally it was found that excitation of purine

9
in ethanol at 2537 A caused a decrease in absorption at 260 nm and an

increase in absorption at 290 nm with an isosbestic point at 270 nm.

158

This photolysis reaction was claimed to be an alcohol adduct at the

70 On the basis of the cal-

4 position as analyzed by NMR spectroscopy.
culated charge densities it would appear that there should also be elec~
trophilic attack at the 2 position. However electrophilic attack.may
occur only at the site which exhibits the maximum charge density change
between the ground and excited states.

It must be noted that no vibronic coupling mechanism was contem-

plated for this molecule. This mabee another facet of the addition of

a third aza nitrogen.

CHAPTER V

INTERMOLECULAR CORRELATION OF THE SPECTRAL DATA

There are certain trends, similarities and dissimilarities which
characterize this group of molecules. This chapter will discuss the
evident aspects which wed these molecules into a family. The discus-
sion will be general with specific examples used only for illustrative
purposes.

Consistent structural patterns were exhibited in the vapor absorp-
tion spectra of these molecules. The features which formed these
patterns by their predictable presence became the diagnostic crutches

l

for assigning the singlet transitions. The 1L$e—-A transition exhibi-

ted much more vibronic structure than the 1L56—1A transition. This

structure however became more diffuse as more aza nitrogens were
incorporated into the molecule. The number of vibronic sequences also
diminished with the number of aza nitrogens. This diffusivity and
diminution was climaxed in the purine spectrum which exhibited no
vibronic structure in the 1Lb transition. For most molecules the 1Lb
transition oscillator strength was less than that of the 1La transition.
Again as the number of incorporated aza nitrogens increased the relative
transition intensities also changed. For benzotriazole the two transi-
tions appeared to have equal intensities and for 4~azabenzimidazole the

1Lb transition oscillator strength surpassed that of the 1

La transition.
Purine's vapor spectrum was more difficult to decipher but the lower
energy (supposedly the ng transition intensity appeared to be decidedly
larger than the higher energy (1Lg)transition intensity. Thus the 1Lb

transition vibronic structure vanishes and the transition intensity ratio

159

160

1Lb/1La progresses from less than unity to greater than unity as the
aza nitrogen population increases. These are major trends which occur.
Minor variations of these trends are exhibited which depend on the posis
tions of these nitrogens as illustrated above with benzotriazole and 4-
azabenzimidazole.

Another consistent feature for all the molecules but purine was
the relative separation between vibronic sequences or groups in the lLb

and 1

La transitions. The average sequence separation in the 1Lb transi-
tion was always less than the vibronic group separation in the 1La
transition for each molecule. Presumably the 1La transition groups are
reasonable representatives of vibronic sequences so group separations
should reflect sequence separations. There did not seem to be any pat-
tern which correlated the sequence or group separations with the number
of aza nitrogens or their positions. In addition the 1Lb transition
sequence separation for some molecules exceeded the 1La transition
group separation for other molecules. The only predictable pattern

was that the average 1La transition group separation would exceed the
average 1Lb transition sequence separation for a given molecule. The
variation between sequence or group separations was nonzero but nominal
for each molecule. However this variation increased as the number of
aza nitrogens increased. There were also fewer sequences or groups for
molecules with more aza nitrogens. For example the diazaindoles had
only two sequences and two groups. Thus this increased variation with
increased aza nitrogen population undoubtedly reflects the diffuseness
of vibronic structure and the concomitant imprecision in vibronic band
position.

The average sequence separation in the 1Lb transition ranged from

1 1

629 cm7 to 820 cm71. The average group separation in the La

161

1 to 1064 cmfl. These energies probably

transition ranged from 723 cm-
represent whole molecule breathing modes. This assertion is based on
the little reported correlation that has been made between IR spectra
and these energies.

When aza nitrogens were introduced into the pyrrole ring the 1La

1

and Lb transitions were quite well separated. But for aza nitrogen

substitution into the benzene ring only or for no aza nitrogen substi-
tution the two transitions were merged. Distinction between the two
transitions was difficult and was based on the recognizable vibronic
sequence or group contours. For molecules with aza nitrogen substitu-
tion in both rings the transition separation was again distinct. This
transition separation vs. aza nitrogen position phenomenon was first

57

recognized by Mason and was verified in this study. One may conceptu-

alize these two lowest singlet transitions as normally overlapping such
that they are merged. Aza nitrogen substitution into the benzene ring
does not destroy the normalcy of this merger. However aza nitrogen

substitution into the pyrrole ring does alter the transition positions

by blue-shifting the 1La transition approximately 5000 cm71. Further

aza nitrogen substitution has no effect on the transition positions.

Parenthetically there was no apparent pattern of either 1L8 or 1L

b
transition position as a function of actual aza nitrogen position

other than the gross effect just described.

A measure of the drastic amount of 1L transition shift is found

a
in the relative positions of the lowest energy sequences of the 1La and

1
Lb transitions. For indole and 7~azaindole the lowest energy sequence

1

of the La transitidn is apparently at longer wavelengths than the

1
lowest energy sequence of the Lb transition. The vapor spectra for

162

the other molecules all exhibited well separated transitions; thus the
lowest energy sequence of the 1La transition must be at shorter wave~
lengths than the lowest energy sequence of the 1Lb transition. One of
the difficulties in assigning the 1La transition 0—0 sequence was the
inability to locate this sequence. This inability was attributed to

two factors. First the broad, diffuse, overlapping nature of the
vibronic groups made distinguishability between these groups very
difficult. Second the pattern of these groups indicated that the 0~0
sequence must have low intensity. This is in contrast with the observa-
tions of the 1Lb transition 0~0 sequence. This transition intensity

was strong enough to make the individual members in the 1Lb O~0

sequence easily discernable. Thus it appears that the 1Lb transition
O-O sequence is Franck-Condon allowed and the 1La transition 0-0
sequence is Franck-Condon forbidden. In other words the ground state
and first excited singlet state potential surfaces have similar contours
and their minima occur at approximately the same general coordinate
positions. However the 1La state potential surface probably has a
different contour and its minimum occurs at a different position than
the ground or 1Lb state.

The most puzzling aspect of the vapor absorption spectra was the
temperature dependence of the lLa/lLb transition oscillator strength
ratio for some of the studied molecules. This phenomenon became appar-
1

Lb 0-0

transition position (verification was accomplished in most cases).

ent when hot-band analyses were being performed to verify the

The nonexistence of this phenomenon for benzotriazole and 4~azabenzi-
midazole further strengthens the contention that this phenomenon is

not experimental artifact. The actual mechanism for explaining its

163

occurrence is not known. Certainly the mechanism must incorporate the
fact that populating higher vibrational and rotational levels in the
ground state increases the nge-IA transition probability. This was
exemplified by the experimental observation that there was a 1La tran-
sition intensity increase relative to the 1Lb transition intensity

as the temperature was increased. In other words the value of the 1La
transition moment, R- (Vifi TL), was changed as the temperature changed.

In this expression 91 is the initial state wave function, fi is the
transition dipole operator and W?. is the final state wave function.
For any molecule W’can be expressed as a product of functions, ng;q;,
which respectively represent the purely electronic, vibrational and
rotational components of the given wavefunction. As the temperature is
increased the number of molecules in a given WV in W; can change. If
this change is serendipitous it could increase the value of R. Usually
the intensity of one transition cannot change without affecting the
intensity of some other transition within a given transition manifold.

Hence this 1

La/lLb transition intensity ratio variation may involve
intensity borrowing through vibronic coupling. That is the‘i’i,v at
higher temperatures may readily couple with a vibrational mode of a
higher excited state and consequently the 1La transition becomes more
allowed. The excited states generally exhibit less symmetry than the
ground state. ’

Another puzzling result was the appearance at lower temperatures'
1

vibronic bands to the red of the Lb 0-0 transition sequence in the

1Lb transition band. The appearance of these vibronic bands preceded

the 1Lb 0-0 transition sequence as the temperature was increased. This

occurred for benzimidazole, benzotriazole and é-azabenzimidazole.

164

Another fact which must be considered is that the solution spectra
of all the molecules, which exhibited these anomalous transition inten-
sity changes, displayed a static 1La/lLb transition intensity ratio -
even at different temperatures. The static ratio displayed in the
solution spectra was the same as observed in the "concentrated" vapors.
This ratio also appeared to be asymptotically reached at higher temper-

atures in the 'equilibrium concentration" vapor spectra. Again the
vibronic coupling mechanism could explain the stability of the 1La/lLb
transition intensity ratio. Stability would be accomplished if one
postulated that the solvent or other colliding solute molecules inter-
acted with a solute molecule such that the 1La transition was maximally
allowed. The interaction could be either through solute-solvent (or
solute-solutebvibronic coupling or via a solute-solvent perturbation
which enhances transition borrowing in the solute molecule.

If the proposed vibronic coupling mechanism is true there are
still some unanswered questions: What is the specific vibrational
coupling? Why does this coupling occur for some molecules but not
for others? Why does the 1La/lLb transition intensity ratio always
increase with increasing temperature and not decrease? It seems that
some of the answers to these questions are inherently related to the
role which the aza nitrogens have in formulating the‘fi-electron
structure. This is evident from the fact that the 1La/lLb transition
intensity ratio doesn't appear to vary when at least two aza nitrogens
are present. In addition the premature appearance of vibronic bands
at energies below the 1Lb 0-0 transition sequence only occurred when an

aza nitrogen was present at the 3 position. Obviously the answers to

these and other questions must await further experiments;including

165
observations with different molecules.

One might wonder why these vapor absorption spectra anomalies only
occurred in this study, i.e. why this is the first report of such anoma-
lies. The answer is twofold and involve the fact that our spectra were
obtained photometrically while most vapor spectra are obtained photo-
graphically. The latter method is used to more accurately make vibronic
band assignments. Usually, though, only a specific portion of the spec-
trum is scrutinized since resoludon is lost as more of the spectrum is
encompassed. With the photometric method resolution is maintained no
matter what wavelength limits are used. This is due to the fact that
the photometric scan is displayed on a chart with optional length while
the photographic scan is displayed on a plate with specified length.
Thus the relative transition intensities were easily discernable photo—
metrically since both the 1La and 1Lb transitions were easily scanned
at any temperature. The second reason for our observation of this
phenomenon is that its presence requires accurate determinations of
relative transition intensities. The photometric method is designed
to accomplish these determinations and the photographic method isn't so
designed. The emulsion change versus light intensity is usually not
linear or logarithmic. Thus our experimental procedures were luckily
optimized for the chance observance of this anomalous phenomenon.

Turning attention now to the solution absorption and fluorescence
spectra, it would seem to be quite difficult to ascribe solvent induced
spectral shifts from the data in Table 7. The trick was to focus atten-
tion on the spectral data presented by only one author at a time. When
an assessment was made of one author's data the spectra of another

author was analyzed. Usually these data verified each other. Often

166

the data of several authors were so well interrelated that the solvent
shifts became established beyond reasonable doubt.

It was possible to correlate the solvent-induced shifts for those
molecules which displayed sufficient absorption and fluorescence data.
The fact that this correlation was accomplished is probably the single
most positive contribution of this dissertation. The correlation con-
sisted of explanations of the solvent shifts using specific known
solvent-solute interactions. These explanations were based somewhat on
physical intuition about the types of expected interactions. However
the explanations were also fashioned in a logical pattern. That is
specific interactions were used to explain only a specific experimental
observation. Each interaction could have only one effect. There was
no ambivalence about specific cause and effect. The observed shifts
were thus explained as a result of a combination of specific inter-
actions. Further aid in forming these explanations was provided by the
‘fi-electronic charge density calculations. These calculations were used
to gain some insight into the charge pattern differences between the
ground and excited states. These calculations also were helpful in
verifying which solvent-solute interactions were pertinent for explain-
ing the observed shifts. Thus the experimental observations and the
calculated results were meshed to provide a consistent explanation of
the solvent-induced spectral shifts.

An example of the verificative correlation between the calculations
and the observed spectra was the amount of Stokes' shift exhibited by
various molecules in polar solvents. Those molecules which exhibited
large Stokes' shifts were also calculatedto have much larger excited

state permanent dipole moments than ground state moments. Those

167

molecules with small Stokes' shifts also had smaller calculated differ-
ences between their ground and excited state permanent dipole moments.
In fact the relative amount of Stokes' shift was easily predicted from
the calculated magnitude of the change between the ground and excited
state moments for all molecules where this comparison was made. This
is a strong correlation because the Stokes' shifts are caused by
dipolar interactions between the solvent and solute molecules. Large
increases in the solute molecule permanent dipole moment magnitudes
upon excitation cause a stabilizing rearrangement of polar solvent
molecules. This stabilization is manifested as a large Stokes' shift.
The confidence fostered by the Stokes' shift correlation with calcula-
ted permanent dipole moments allowed predictions to be made for those
molecules whose permanent dipole moments hadn't been calculated. That
is,a prediction was made concerning the relative difference between

the ground and fluorescence-emitting state permanent dipole moment
magnitudes for those molecules whose moments weren't calculated. Pre-
dictions could also be made via calculations for molecules whose
spectra had not been examined. However this was not attempted here.
Indazole, benzimidazole and A-azabenzimidazole exhibit small Stokes'
shifts. Indole, benzotriazole and S-azabenzimidazole exhibit moderate
Stokes' shifts. Those molecules with a single aza nitrogen substitu-
tion in the benzene ring and purine exhibit large Stokes' shifts.

Thus the position of the aza nitrogens has a profound effect on the
permanent dipole moment magnitude differences between the ground and
excited states. These differences occur in the‘fi-electron structure so
the aza nitrogen positions must have strong influences on the‘fi-electron

rearrangement which occurs upon excitation.

168

The dipolar interactions which were responsible for the Stokes'
shifts are also involved in the shifts observed between a solute's
electronic transitions in different solvents. However, the observed
solvent induced shift effects were not as pronounced as the Stokes'
shift effects. Other effects which contribute to these shifts are H-
bonding and solvent protonefi-electron interactions. These three inter-
actions were thus combinatorally used to explain the observed absorp-
tion and fluorescence shifts. At an aza nitrogen the solvent proton-
‘fi-electron interactions and H-bonding were postulated to manifest the
same effect. However, at the pyrrolic nitrogen these two interactions
would lead to shifts in opposite directions. Thus at a pyrrolic nitro-
gen the proton-i%electron interactions would cause a blue-shift and H-
bonding would cause a red-shift if the‘fi-electron charge densities at
this nitrogen decreased upon excitation. The two effects would be
reversed if the charge densities increased upon excitation. At
pyridinic nitrogens both interactions would cause a red-shift if the
‘fi-electron densities at these nitrogens increased upon excitation.

The interactions would cause a blue-shift if the densities decreased
upon excitation. It was expected that solvent protons would preferen-
tially interact at the pyridinic nitrogens since these species possessed
lone pair electrons. Dipolar interactions would cause a red-shift if
the excited state permanent dipole moment magnitude exceeded the ground
state moment magnitude. A blue-shift would occur if the excited state
moment magnitude was less than the ground state moment magnitude.

These three interactions were correlated with the excited and
ground state‘fi-electron charge density calculations to explain the

observed solvent induced shifts in the absorption and fluorescence

169

spectra. Subtle shifts and reversals in shift trends were explained as
due to combinations of these interactions or as interactions at differ-
ent sites in the molecule. The nitrogens were chosen as the interaction
sites because intuitively they should have more‘fi-electron density and/
or undergo larger‘fi-electron charge density alterations upon excitation
than should occur at the carbons. This supposition was verified with
the charge density calculations. There are obviously other possible
explanations for the solvent-induced spectral shifts. However, the
correlations found here adequately explain these shifts and are reason-
able. Based on physical intuition the explanations are sound.

The experimental data seemed to indicate that there were solute
dependent differences in the amount of vibronic structure present in a
given solvent. Apparently the differences depended on which solute
molecule was being investigated. These differences seemed to occur
for the absorption as well as the fluorescence spectra. Benzimidazole
displayed sharp vibronic structure in its fluorescence spectra even
in polar solvents. Indole in nonpolar solvents and indazole and 7-
azaindole in both polar and nonpolar solvents displayed¢fiffuse vibronic
structure in their fluorescence spectra. These four molecules as well
as purine also exhibited some vibronic structure in their solution
absorption spectra. No absorption or fluorescence structure was
reported for the other molecules. The presence or absence Of this
vibronic structure is probably ndehenomenologically significant.
Instead the structure probably reflects the resolving ability of the
spectrometers used in making the spectral measurements. Thus there
is presently little information pertaining to the electronic struc-

tures of these molecules which is derivable from the presence or absence

170
of vibronic structure in solution. Experiments must be performed with
instruments possessing good resolution capabilities before any analysis
of this structure is undertaken.

The pH of a solution had demonstrable effects on the spectral
properties of these molecules. These effects were correlated with the
data obtained in other solvents to give a consistent causal picture for
the observed spectral shifts. A somewhat disturbing aspect of these
observations is the fact that anions or cation supposedly were not
formed unless a spectral change had occurred compared with that in
neutral solutions. This presumes that the anion or cations species
have different spectral characteristics than the neutral species.

For most molecules this presumption is probably quite valid. However
this statement may not be true for all molecules. One might be
observing spectral changes for dications or dianions rather than
cations or anions. This could occur if the doubly charged species was
formed as a result of an interaction which strongly perturbed the
molecule's fi-electron structure and the singly charged species was
formed without perturbing the‘W-electron structure.

The species in acidic media whose spectra differed from those in
neutral solutions did not exhibit a consistent shift relative to the
spectra obtained in neutral solutions. It was impossible to predict
whether a red-shift or a blue-shift would be observed. This was
probably due to the type of protonJfi-electron interaction which occur-
red in a specific molecule. On the other hand the "anion” regularly
displayed a red-shift relative to the spectra in neutral solutions.
The only reported exception of this statement for these molecules was

the anion fluorescence of benzotriazole. The red-shift was attributed

171

to a’fi-electron structure destabilization caused by the Coulombic
repulsion between this structure and the negative charge at the pyrrolic
nitrogen. This residual charge was the hypothesized result of a proton
extraction at this nitrogen's hydrogen by the basic solvent. This
hypothesis could be easily tested by replacing the pyrrolic nitrogen's
hydrogen with a methyl group and again observing the spectra in basic
solution. The spectral changes should not occur at the same pH as the
pKa of the normal nonmethylated molecule if this hypothesis is correct.
Another consistency noted for the anion was that the pKa of the molecule
in the excited state was lower than that molecular pKa in the ground
state. This supposedly reflects the relative acidity of the pyrrolic
nitrogen in the ground and excited states. However very scanty data
were available to verify this hypothesis. Again the hypothesis could be
checked with molecules methylated at the pyrrolic nitrogen. In summary
it would be fruitful if concentrated experimental effort were devoted
to studying the effects of pH on the spectral properties of molecules.
The amount of available phosphorescence data was remarkably scarce.
Thus very little comparison of such data between molecules was possible.
A disappointing feature of the phosphorescence data analyses was the
poor correlation between the calculated and the experimentally deter-
mined transition energies. This reflects the inadequacy of the calcu-
lations. The inadequacy could probably be corrected with better
parametrization. Phosphorescent emission from the "cation" was red-
shifted relative to emission from the neutral species for the few
molecules whose emission was monitored in the two media. This red-shift
paralleled the fluorescence shifts for these molecules between the

cation and neutral species. Thus the same proton-fibelectron

172

interaction probably stabilizes the emitting state for both emissions.
In contrast with this parallel between the cation and neutral species'
emissions there was no apparent correlation between the observed Stokes'
shifts and the phosphorescence position for these molecules. This
indicates that dipolar interactionndo not play an important role in
determining the phosphorescence position. It should be noted that all
phosphorescence spectra exhibited a moderate amount of vibronic struc-
ture. An analysis of this structure was not attempted. However this
analysis could be performed and might yield information about the
vibrational modes in the lowest triplet state for these molecules.

Two other experimental aspects were noted for these molecules.
These were dual emission and ndfifi transitions. The dual emission
explanation was hypothesized to occur for indole, purine, benzotriazole
and indazole. The bases for this type of emission was discussed in
appropriate sections in the discussion for these molecules. Purine
displayed an n-fl* transition. In addition benzotriazole and 4-azaben-
zimidazole were hypothesized to exhibit this transition. These di-
and triazaindoles were the only instances where this type of transition
seemed to occur.

The results of the theoretical calculations were reasonably corre-
lative with the spectral prOperties of the studied molecules. This
heartening correlation fostered trust in most of the calculated quanti-
ties whose corresponding experimental parameters had not been measured
(excluding the calculated quantities associated with phosphorescence as
noted above). This included properties of singlet transitions for those
molecules which had not been extensively experimentally examined.

Unfortunately the calculated singlet transition energies were not

173

discriminatory for the first two transitions. That is the separation
between states was about the same for all molecules whose transitions
were calculated. Separation of the first two transitions was not pre-
dicted for benzimidazole and 4-azabenzimidazole although this fact was
experimentally observed. Likewise merging of these transitions was
not predicted for indole and 7~azaindole. The lack of these predictions
was not considered an inconsistency for the calculations. The calcula-
tions and observed energy of the first transition agreed within 0.2 eV
for all molecules where this comparison could be made. Such a compar-
ison for the second transition yielded agreement within 0.3 eV. Such
agreement is all that can be reasonably expected. The separation or
merger of these two transitions are subtle differences which can be
experimentally observed but not easily discernable with calculations.

Assignment of the two lowest calculated singlettransitions was
based on three criteria: transition moment orientation, oscillator
strength and excited state permanent dipole moment magnitude. The

1La state was postulated to possess the higher dipole moment magnitude

and the 1

La transition was presumed to be preferentially oriented along
the short axis of the molecule and have the larger oscillator strength.
The 1Lb state was postulated to have the lower moment magnitude and the
corresponding transition was presumed to be primarily oriented along
the long axis and possess the smaller oscillator strength." These
assignments were based on experimental observations of indole and
related molecules. The permanent dipole moment criterion was derived
from polar solvent induced shifts. Indole's and 7-azaindole's 1La

transition shifted more than the 1Lb transition when a solvent-solute

dipolar interaction was involved. The transition orientation criterion

174

was based on the observed orientations in naphthalene which is iso-
electronic with indole. The oscillator strength criterion was based

on experimental observations for naphthalene and indole. This criterion
does not appear to be correct for é-azabenzimidazole and possibly benzo-
triazole or purine. However, further experimental data must be obtained
before a final judgement of this criterion's applicability is made.
Other than for these three possible exceptions the experimental observa-
tions and the assigned transitions based on these criteria agreed per-
fectly. That is the criteria for the assignments of the calculated
transitions and the experimentally observed spectral parameters coincided
for all molecules where a comparison could be made. Such a remarkable
amount of correlation was surprising and lent credence to the validity
of the calculations.

For these molecules the calculated singlettransition moment orien-
tations did not parallel a molecular primary axis and were not predicted
to be perpendicular to each other. Although the criterion was used that
they preferentially possess a major component along a particular axis
it was not experimentally expected that the moments parallel an axis.
Thus the calculated results reinforced our naive physical conception of
these transitions. There was also general agreement between the calcu-
lated angle between the two lowest transitions and the available polar-
ization data. However, this statement must be tempered with the fact
that small adjustments in the parameters used for the calculations
significantly altered the resultant transition moment orientation. Thus
these calculated orientations must not be interpreted literally. There

is still negligible chance that the moments would be predicted to parallel a

175

primary axis of the molecule.

The good correlation between experimental observations and calcu-
lated observables for the singlet transitions and states engendered
trust in the plausibility of the‘W-electron charge density calculations
for these states. This trust was further enhanced by the correlation
between these density calculations and the relative reactivities of
the atomic centers in the molecules. The correlation was quite good
although little reactivity data was cited. More data is probably
available and could be used to further substantiate these density cal-
culations. It was also noted above that the calculated densities
provided realistic values for the interactions involved in the solvent
shift interpretations.

All this evidence for the believability of the calculated“?-
electronic charge densities provides confidence in the picture of the
molecular electronic structure which these densities portray. Thus
the relative shifts in/W-electronic charge between the ground and
excited states appears to be an accurate picture of the actual physical
processes. In additunathese calculated densities appear to accurately
reflect the relative position of the charge in the ground and excited
states. This pictorial representation of the electronic charge is
the most powerful and perhaps the most useful result of the calcula-
tions. Experimental observations alone could not develop such a picture.
This conceptualization of the electronic structure is the culmination of
the previous paths of correlations. It is the essence and was the goal
of this thesis.

One of the conclusions from this study is that minor variations in

the relative‘fi-electron charge densities at various positions can be

176

effected by or can affect the spectral properties of these molecules.
For example the absorption and fluorescence spectra of benzimidazole
in neutral and basic solutions are quite different. The difference
occurs between the two processes as well as between the two solvents.
This large variation occurs as a result of the solvent change which
presents only a small perturbation to the fi-electron structure. A
second example is the differences in values of the permanent dipole
moments for the ground and excited states. An electronic transition
can drastically alter the value of the molecular dipole moment.
Another example is presented by the charge density calculations for
any two molecules. Although the charge density variations between'
molecules are small at any position the spectral characteristics are
quite different. These spectral characteristics are so discernably
unique that they provide identifying features of these molecules.
These spectral differences were not unexpected. However, the minor
rearrangement of‘fi-electron charge accompanying these differences was
surprising and enlightening.

There were two spectroscopic phenomena discussed in this study
which appear to be relatively unique properties of these molecules.
These phenomena are the hypothesized vibronic coupling which occurs
«hring excitation and the dual emission from two excited states which
occurs for some molecules. The question arises concerning the biologi-
cal significance of these phenomena. These processes signify close
communication between electronic states. Is this biologically relevant?
Since the normal biological function of these molecules does not depend
on interactions of the molecules with light there does not seem to be

any relevance of these processes. However, the natural environment of

177

these molecules includes electromagnetic radiation. UV light can induce
mutations in DNA. UV irradiation can severely disrupt the function of
enzymes. Also the chemical reactions of these molecules must involve
the‘W-electron system since these electrons are an integral part of

the molecules. But there still does not seem to be any biological
function in which these molecules participate that depends upon these
unique electronic properties. Similar chromophores which did not
exhibit these phenomena could probably participate in the biological
activity. Yet the nagging unanswered question still remains that these
specific chromophores are the actual ones involved in biological via-
bility. Perhaps these electronic properties are important but their
relevance remains unknown. A related unanswered question regards the
properties of these molecules which dictate their use in different
biological functions. It is still unknown why these molecules which
form a chromophoric family display diverse biological roles. The
position and/or type of substituent as well as the aza nitrogen posi-
tion may be implicated. These position differences were manifested in
the spectral data so they are important.

Finally it is hoped that this study has shown that a coordination
of spectroscopic data is feasible and can yield information about the
electronic structure of molecules. In addition the data can be used
to compare properties within a family of molecules. The exhibited
spectroscopic information isn't composed of isolated features. There
are aspects of this information which are interrelated. Coordinating

the information can reveal these interrelationships.

B IB LI OGRAPHY

10.

11.

12.

13.

14.

15.

16.

17.

18.

BIBLIOGRAPHY

I. N. Levine, Quantum Chemistry, Allyn and Bacon (1970).
F. L. Pilar, Elementary Quantum Chemistry, McGraw-Hill (1968).

R. G. Parr, Quantum Theory of Molecular Electronic Structure,
W. A. Benjamin (1963).

P. Hochman and R. Wagner, Description of the TE DEUM Program,
internal publication (1970).

P. Hochmann, V. Kleinwachter, and M. A. El-Bayoumi, to be published.
W. J. Potts, Jr., J. Chem. Phys. 29, 809 (1952).

J. M. Hollas, Spectrochim. Acta 12, 753 (1963).

M. A. El-Bayoumi, C. A. Taylor, M. Kasha, to be published.

J. R. Platt, J. Chem. Phys. 12, 101 (1951).

S. V. Konev, Fluorescence and Phosphorescence of Proteins and Nucleic
Acids, Plenum Press 1967, a) p. 16-17 b) p. 29-35 c) p. 19

d) pa 39-53e

D. P. Craig, J. M. Hollas, M. F. Redies and S. C. wait, Jr., Phil.
Tran. Roy. Soc. 253AI 543 (1961).

G. Weber, Biochem J. 12, 335 (1960).

H. Zimmermann and N. Joop, Z. Elektrochem., Ber. Bunsenges Physik.
Chem. 62, 61 (1961).

ll-U. Schutt and H. Zimmermann, Z. E1ektrochem., Ber Bunsenges.
Physik. Chem. 61, 54 (1963).

E. A. Chernitskii, S. V. Konev, and V. P. Bobrovich, Dokl. Akad.
Nauk BSSR 1; 628 (1963).

A. N. Sevchenko, S. V. Konev, and M. A. Katibnikov, Dokl. Akad.
Nauk SSSR 153, 875 (1963).

B-S. Song and W. E. Kurtin, J. Am. Chem. Soc. 21, 4892 (1969).

J. Zuclich, J. Chem. Phys. 22, 3586 (1970).

178

19.
20.
21.
22.
23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.
35.
36.
37.

38.

39.
40.

41.

179
B. L. Van Duuren, J. Org. Chem. 16, 2954 (1961).
D. A. Chignell and W. B. Gratzer, J. Phys. Chem. 12, 2934 (1968).
S. Nagakura, J. A. Chem. Soc. 16, 3070 (1954).
E. G. McRae, J. Phys. Chem..§1, 562 (1957).
R. D. Gordon and R. F. Yang, Can. J. Chem. 481 1722 (1970).

E. A. Chernitskii and S. V. Konev, Dokl. Akad. Nauk BSSR 8‘ 258
(1964).

S. V. Konev, V. P. Bobrovich and E. A. Chernitskii, Biofizika 1Q,
42 (1965).

V. P. Bobrovich, G. S. Kembrovskii, and N. I. Marenka, Dokl. Akad.
Nauk BSSR 19, 936 (1966).

N. Matage, Y. Torihashi, and K. Ezumi, Theor. Chim. Acta. 2, 158
(1964).

B. L. VanDuuren, Anal. Chem. 31, 1436 (1960).
C. M. Chopin and J. H. Wharton, Chem. Phys. Lett.Ԥ, 552 (1969).

M. S. walker, T. W. Bednar, and R. Lumry, J. Chem. Phys. 41, 1020
(1967).

G. S. KembrovSkii, V. P. Bobrovich and S. V. Konev, Zh. Prikl.
Spektrosk. 2, 695 (1966).

I. Wéinryb, Biochem. Biophys. Res. Comm.‘34, 865 (1969).

L. F. Gladchenko, M. Y. Kostko, and L. G. Pikulik, A. N. Sevchenko,
Dolk. Akad. Nauk. BSSR.2L 647 (1965).

J. Eisinger and G. Navon, J. Chem. Phys. 29, 2069 (1969).

R. F. Chen, G. G. Vurek and N. Alexander, Science-126, 949 (1967).
H. Sprince, G. R. Rowley, and D. Jameson, Science 122, 442 (1957).
R. W. Cowgill, Arch. Biochem. Biophys. 199, 36 (1963).

H. V. Drushel, A. L. Sommers and R. L. Cox, Anal. Chem. 32, 2166
(1963).

G. Weber and F. W. J. Teale, Trans. Faraday Soc. 53, 646 (1957).
J. W. Bridges and R. T. Williams, Biochem. J. 107, 225 (1968).

M. G. M. Balemans and F. C. G. Van De Veerdonk, Experientia 23,
906 (1967).

42.

43.

44.

45.

46.

47.
48.
49.

50.

51.
52.
53.
54.

55.

56.

57.
58.
59.
60.
61.
62.
63.
64.

65.

180

S. V. Konev and E. A. Chernitskii, Dokl. Akad. Nauk. BSSR 9, 328
(1965). 7

S. V. Konev and E. A. Chernitskii, Biofizika 2; 520 (1964).

M. S. walker, T. W. Bednar and R. Lumry, J. Chem. Phys. 45, 3455
(1966).

L. Stryer, J. A. Chem. Soc. 88, 5708 (1966).

L. G. Pikulik, L. F. Cladchenko and M. Y. Kostko, Zh. Prikl.
Spektrosk. 6, 210 (1967).

R. C. Heckman, J. Mol. Spect. 1, 27 (1958).
V. L. Ermolaev, Optika i Spektroskopiya 11, 492 (1961).
S. Freed and W. Salmre, Science 128, 1341 (1958).

C. Cailly and A. Bourkhors, C. R. Acad. Sci., Paris, Ser. C264,
480 (1967).

G. M. Barenboim, Biofizika Z, 227 (1962).

R. Pariser, J. Chem. Phys. 84, 250 (1956).

R. L. Hinman and J. Lang, Tetrahedron Lett. 11, 12 (1960).
T. K. Adler, Anal. Chem. 84, 685 (1962).

J. Derkosch, O. E. Polansky, E. Rieger and G. Derflinger, Monatsh
Chem. 81, 1131 (1961).

G. H. Beaven, E. R. Holiday, E. A. Johnson, Spectrochim. Acta 4,
338 (1951).

S. F. Mason, J. Chem. Soc. 1884, 2071.

Present work.

K. C. Ingham, this laboratory.

L. B. Clark and I. Tinoco, J. Am. Chem. Soc. 8], 11 (1965).

H. H. Chen‘and L. B. Clark, J. Chem. Phys. 81, 1862 (1969).

J. Drobnik and L. Augenstein, Photochem. Photobiol. 2; 13 (1966).
J. W. Longworth, Biochem. J. 83, 104p (1962).

B. J. Cohen and L. Goodman, J. Am. Chem. Soc. 81, 5487 (1965).

H. C. Borreson, Acta Chem. Scand. 11, 921 (1963).

66.

67.

68.
69.
70.

71.

181

J. Drobnik, V. Kleinwachter and L. Augenstein, Photochem. Photobiol.
81 l47 (1967).

R. Guermonprez, C. Helene and M. Ptak, J. Chim. Phys. 84, 1376
(1967).

C. Helene, Biochem. Biophys. Res. Comm. 18, 237 (1966).
M. A. Slifkin, Biochim. Biophys. Acta 103, 365 (1965).
H. Linschitz and J. S. Connelly, J. Am. Chem. Soc. 88. 2979 (1968).

J. W. Longworth, R. 0. Rahn and R. G. Shulman, J. Chem. Phys. 48,
2930 (1966).

J

‘ n1&meMimymnflngulwu '