Infrared, optical, and luminescent properties of quasi-one-dimensional organic charge transfer salts and polymers


Material Information

Infrared, optical, and luminescent properties of quasi-one-dimensional organic charge transfer salts and polymers
Physical Description:
ix, 206 leaves : ill. ; 29 cm.
Musfeldt, Janice Lynn, 1965-
Publication Date:


Subjects / Keywords:
Semiconductors   ( lcsh )
Polymers   ( lcsh )
Charge transfer   ( lcsh )
Matter -- properties   ( lcsh )
Chemistry thesis Ph. D
Dissertations, Academic -- Chemistry -- UF
bibliography   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1992.
Includes bibliographical references (leaves 195-205).
Statement of Responsibility:
by Janice Lynn Musfeldt.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001867907
oclc - 28997452
notis - AJU2423
System ID:

Full Text







UNV-; Y i,." rLu.'L.-n. L; ?. I -S

To David

Digilized by the Internet Aichive
in 2011 with funding from
University of Florida, George A. Smathers Libraries with support from LYRASIS and the Sloan Foundation


It is my great pleasure to thank my advisor, Professor David B. Tanner, for giving

me the opportunity to work in his lab and study the interesting problems detailed

in this dissertation. His good advice, patience, and encouragement were essential to

the success of my graduate work at the University of Florida. It was my very good

fortune to work in his group. I also thank Professors P.J. Hirshfeld, W.B. Person,

M.T. Vala, and K.B. Wagener for reading this dissertation and for their interest in

serving on my supervisory committee. Additional thanks are due to Peter Hirshfeld

for his help in solid state physics.

I would like to extend my thanks to my colleagues in my research group for

their friendship, useful conversation and cooperation during the past several years. A

special note of thanks is due to C.D. Porter for his help on various problems with the

computer software, programming and interfacing techniques. I am also indebted to

Drs. G.L. Carr, S.L. Herr, S. Jeyadev, K. Kamaris, D.B. Romero, and particularly

V. Zelezny for many interesting and useful discussions. I would also like to thank

Dr. K.F. Ferris for his encouragement and the temporary use of the diamond anvil

pressure cell. Finally, I am grateful to Dr. C.C. Homes, whose recent calculations have

greatly aided our understanding of the electron-phonon coupling in the quarter-filled

charge transfer salts.

I would like to express my gratitude to Drs. K. Kamarais, M. Almeida, and Y.

Iwasa for providing the good quality single crystals of the various organic charge

transfer salts which were essential to the completion of the bulk of this dissertation.

In addition, I am also in the debt of Dr. J. Ruiz, Prof. J.R. Reynolds, J. Wang,

and Prof. M. Pomerantz for the preparation of the various polymers used in these

studies. A special thanks is due to John Reynolds for his interest, encouragement and

support in the electro-luminescent polymer project. I have also greatly appreciated

the use of the fluorimeter in the laboratory of Prof. K. Schanze for the steady state

and time resolved photo-luminescent measurements. Nancy Thornton was a big help

with the aforementioned measurements.

I have greatly appreciated the efforts of the technical staff members in the 1 i1~ -sics

department machine shop, especially Ron Spencer and Ed Storch, the engineers in

the condensed matter group for the reliable supply of liquid helium, and the staff in

the electronics shop, particularly Larry Phelps.

I would also like to take this opportunity to thank my parents for their support,

and my brother for his encouragement and practical advice.

A very special thanks is due to my husband, David, whose encouragement and

understanding were essential to the completion of this work.

Financial support from the NSF (grant DMR 9101676) is gratefully acknowl-




ABSTRACT .... ...... .... ..... viii

I. INTRODUCTION . .... . 1
Quasi-One-Dimensional Organic Charge Transfer Salts .... .1
Luminescent Polymers . .. .. .. 5
Dissertation Outline ... .... ....... 7

II. THEORY . . . 13
Organic Charge Transfer Salts .. .... 13
Hubbard Model . ... .. 14
Peierls Transition .... ... ..... ...... 18
Models of the Electron-Phonon Interaction ... .. . 20
Luminescent Polymers .... ....... .... .... ... 28
Photo-luminescence ................ ........ 28
Electro-luminescence . . 34

Quasi-One-Dimensional Organic Charge Transfer Salts .. .. 44
General ................. . 44
Structural Considerations .. . . 45
Transport Properties ....... .. ...... ..... 47
Spectroscopic Studies .............. 48
Our Materials .. ..... ... ... .... ...51
NPrQn(TCNQ)2 .. .. ... .. ... ........ 51
DMTM (TCNQ)2 . . .. 53
K- and Rb-TCNQ ........ .. ... .. ........ .. 56
Luminescent Materials ... ... ... 61
The Field of Electro-luminescent Polymers ...... .63

Our Materials-Previous Work . . 64

Fourier Transform Infrared Spectroscopy . 73
G general . . . 73
Bruker 113V . . . .. .. 76
Optical Measurements: the Perkin-Elmer Grating Spectrometer .78
Temperature and Polarization Control . . 81
General . . .. . 81
Measurements on the Charge Transfer Salts . 82
Sample Preparation and Mounting-Organic Charge Transfer Salts 83
Analysis of Reflectance Spectra . .. . 84
The Kramers-Kronig Transform . .. 84
Method of Extrapolation ...................... 87
Luminescence Measurements . .... 88
Sample and Device Preparation .. ... .. 88
Photo-luminescent Measurements . 89
Electro-luminescence Measurements . 89



Results: NPrQn(TCNQ)2 ...........
Room Temperature Spectra . .
Temperature Dependence . .
Discussion: NPrQn(TCNQ)2 ..... .....
Electronic Features .. .. .. ..
Electron-Phonon Coupling . .
Implications for Charge Transport .
Comparison with Previous Results .
The Phase Transition ............
Results: DMTM(TCNQ)2 ...........
(010) Crystal Face: Room Temperature Spectra
(010) Face: Temperature Dependence .. ...
(110) Crystal Face . .
Discussion: DMTM(TCNQ)2, (010) Crystal Face .
Electronic Features along Rmax .

. . 102
. . 102
. . 102
. . 103
. . 105
. . 105
. . 106
. . 109
. 110
. . 113
. . 114
. . 114
. . 116
. . 117
. . 118
. . 118


Electron-Phonon Coupling Along Rmax, .. . 121
Sum Rule . . . 124
Discussion: (110) Crystal Face . .. 125

FER SALTS ............ ............ ...143
Results: K-TCNQ and RB-TCNQ .. .. .. 143
Room Temperature Spectra .. .. .... 143
Temperature Dependence ... . 146
Discussion: K-TCNQ and RB-TCNQ .. ... .. 149
Electron-Phonon Coupling . .. ..... 149
The Lattice Modes . .. ..... 153
Charge Transport ............. .. ......... 156

Phenylene Based Polymers .... .... .... 171
Optical Absorption and Steady State Photo-Luminescence .171
Electro-luminescence . .. . ... 174
Exciton Lifetimes and Luminescent Efficiencies .. .. ... .. 175
Structure-Property Relations: Phenylene Based Materials .. .176
(2-Thienyl)Phenylene Based Polymers . .... 178
Mechanism of Charge Injection in Electro-Luminescence .... 179
Future Work . ... .. .. 181

Quasi-One Dimensional Organic Charge Transfer Salts .. ... .188
Luminescent Polymers .. ..... 191

References . . . 195


Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy



Janice Lynn Musfeldt

December 1992

Chairman: David B. Tanner
Major Department: Chemistry

Polarized infrared and optical reflectance spectroscopies were used to study the

structural phase transitions in two quarter-filled semiconducting organic charge trans-

fer salts, N-propylquinolinium(TCNQ)2 and N-dimethyl thiomorpholinium(TCNQ)2;

and two half-filled salts, Potassium-TCNQ and Rubidium-TCNQ.

The spectra of NPrQn(TCNQ)2 display characteristics common to quarter-filled

charge transfer salts. Various cluster models were applied to describe the electron-

phonon coupling at 300 K. The weakly metallic transport properties above Tc in

NPrQn(TCNQ)2 are attributed to the uniform charge distribution within the tetramer

and the high degree of overlap between the intra- and inter-tetramer charge transfer

bands. Despite evidence for a larger intra-tetramer transfer integral below Tc, the

transport properties are dominated by charge localization, as evidenced by the doublet

pattern in the vibrational features.

In the DMTM(TCNQ)2 salt, measurements were made on several faces of large

single crystal samples. The electron-phonon coupling behavior was analyzed within

the framework of both an isolated dimer model and a twofold commensurate charge

density wave model. Our data, polarized close to the b crystallographic axis,

do not support the idea that the unusual transport properties of semiconducting

DMTM(TCNQ)2 are caused by low-energy inter-chain charge transfer.

The spectra of two closely related half-filled salts, K-TCNQ and Rb-TCNQ, were

also investigated. The Ag features change gradually with temperature in the potas-

sium sample, whereas in the rubidium sample little contrast was observed. The far

infrared Ag vibrational features in K-TCNQ continue to change gradually above Tc.

In contrast, the Ag vibrational modes in Rb-TCNQ vanish almost completely at Tc.

These observations suggest that while the phase transition is strongly first-order for

Rb-TCNQ, it is closer to second-order in K-TCNQ. The temperature dependence of

the lattice modes is also discussed.

Finally, optical absorption, photo- and electro-luminescence measurements were

performed on a series of semiconducting polymers based upon phenylene and thio-

phene. Visible light emission was observed from two materials with discrete emitter

units, suggesting that intra-chain mobility of the charge defect may not be essential

to radiative recombination of the exciton.


This document describes the investigation of the spectroscopic properties of quasi-

one-dimensional materials. It may be divided into two main components: the infrared

and optical properties of semiconducting organic charge transfer salts, and the absorp-

tion, photo- and electro-luminescence properties of several phenylene and thiophene

based polymers.

Quasi-One-Dimensional Organic Charge Transfer Salts

The organic molecule 7, 7', 8, 8' tetracyanoquinodimethane (TCNQ) was first syn-

thesized by researchers at E.I. du Pont de Nemours and Company in 1960.1 Neutral

TCNQ molecules stack in a columnar manner to form a semiconducting, diamagnetic

molecular crystal. However, the TCNQ molecule has a high electron affinity, and it

is often involved in the formation of organic charge transfer salts.2 These intrinsi-

cally conducting, quasi-one-dimensional organic salts have been the subject of intense

study for the past three decades.

In recent years, a great variety of TCNQ salts have been synthesized with various

donor molecules. The chemical identity of the donor cation and the stoichiometry

of the salt determines the population of the conduction band. The ratio of donor to

acceptor molecules and the degree of static charge transfer affect the details of the

crystal structure. For example, simple 1:1 salts, such as K- or Rb-TCNQ, have one

unpaired electron on each TCNQ molecule. Because of the large on-site Coulomb

repulsion (Hubbard U) between the electrons, these materials display very low dc

electrical conductivity at room temperature. Hence, they are often referred to as

"Mott-Hubbard" insulators. Electronically, they consist of either completely filled

or completely empty bands. Thus, electrical conduction can occur only by thermal

activation. For the complex salts, the most common stoichiometric ratios are 1:2 or

2:3. The behavior of these salts is more complicated than the 1:1 insulators, as they

can be insulating or conducting.

The strong anisotropy in these materials, from which the most interesting physical

properties derive, is known to result directly from their distinctive structural architec-

ture. In these materials, the TCNQ molecule acts as a monomer repeat unit, stacking

in quasi-one-dimensional columns or chains. The columns are commonly segregated

on the basis of donor or acceptor species. Within these stacks of TCNQ molecules, the

distribution of molecules can be uniform or arranged in units of dimerss," "trimers,"

"tetramers," or "n-mers."3 The structural overlap of the TCNQ molecules in the stack

determines the overlap of the electronic wavefunction. Consequently, a uniform dis-

tribution of TCNQ molecules usually results in metallic properties, while a distorted

stack has semiconducting properties.4

Many recent studies of quasi-one dimensional organic materials have concentrated

upon structural and the metal-insulator phase transitions.5-12 Detailed examination

of the systematic transport, structural and spectral changes in the neighborhood of

the critical temperature (Tc) has proved to be a useful tool in understanding various

property changes through the phase transition. Spectroscopic methods have been

used with particularly good success to probe the mechanisms and characteristics of

structural phase transitions by understanding changes in electron-phonon coupling

and the semiconducting gap.

Not only is it of fundamental scientific interest to understand the driving mecha-

nism for these metal-insulator transitions, but it is of practical interest as well. With

the understanding of the driving forces of the Peierls transition, it may become pos-

sible to inhibit such transitions, engineering conducting organic materials without

metal-insulator transitions. Repression of a metal-insulator transition may lead to

organic materials with metallic or superconducting properties at low temperature.

NPrQn(TCNQ)2, DMTM(TCNQ)2, K-TCNQ, and Rb-TCNQ have particularly

interesting structural phase transitions. The NPrQn salt undergoes a typical metal-

insulator transition at T,13,14 while the dc conductivity of DMTM salt actually in-

creases below Tc.15,16 The K- and Rb-salts are interesting because of their chemical

similarity and simplicity, combined with their strikingly different behaviors at Tc.6'17

Figure 1 displays the chemical structure of the TCNQ molecule and the various cations

used in this study. Each of these materials was available in high quality, single crystal


In order to provide further information on the nature of the high- and low-

temperature phases as well as the structural phase transitions, we have investi-

gated the infrared and optical properties of NPrQn(TCNQ)2, DMTM(TCNQ)2, K-

TCNQ, and Rb-TCNQ. Spectroscopic methods are well suited to the study of highly

anisotropic materials such as the TCNQ charge transfer salts, providing information

on the electronic charge transfer and localized excitations at high energies, as well as

the vibrational features, electron-phonon coupling, and lattice and libron modes at

lower energies.4,18,19,20,21 Infrared spectroscopy is an especially sensitive probe of the

changes which accompany a structural phase transition, as the vibrational structure is

a sensitive indicator of lattice and electronic variations. Our results, in particular the

electron-phonon coupling in the chain direction, have been treated within the frame-

work of molecular cluster models (such as the isolated dimer by Rice et al.22 and the

isolated tetramer model by Yartsev23), and in the case of DMTM(TCNQ)2, a model

for commensurate charge density waves in extended linear chains, initially devised by

Rice24 and developed further by Bozio et al.25 In these models, the totally symmetric

Ag modes of the TCNQ molecule, normally infrared inactive, become optically active

via coupling, either through a charge transfer process or with a charge density wave.

Because of this coupling, these modes show unusually large oscillator strength, with

polarization along the TCNQ stacking direction. Emphasis has been placed on the

correlation of the spectral properties above and below Tc with available structural,

transport, and band structure data, in order to obtain a greater understanding of

the characteristics and mechanism of the structural phase transitions in these four


In addition, our data on NPrQn(TCNQ)2 is compared to previous spectro-

scopic results on the closely related, quinoid based compounds such as Qn(TCNQ)2

and (NMP)z(Phen)i-_(TCNQ) by McCall et al.26 We also compare our results

for DMTM(TCNQ)2 with the well-studied N-methyl-N-c liyl morpholinium(TCNQ)2

system, NIMEM (TCNQ)2.22,27,28,29 The counterions in these two aforementioned mate-

rials are closely related, the most notable difference being the more polarizable sulfur

atom on the DMTM ring, which is known to be responsible for many interesting

two-dimensional effects in the family of ET salts.30 These comparisons illustrate the

important role of the cation in determining the transport and spectroscopic properties

of quasi-one-dimensional materials.

For the half-filled salts (K- and Rb-TCNQ), we concentrate our discussion on a

comparison between the two samples, once again illustrating the important effect of

the counterion identity in these salts. Such a comparison is valuable in light of the

recent and very thorough measurements of the temperature dependence of the near-

infrared and visible spectra on these same crystals, by Okamoto et al.6 Along the

TCNQ stacking direction in the low-temperature phase, they clearly resolve the "two-

oscillator" nature of the charger transfer (U-band) and follow its evolution through

Tc. The strong temperature dependence of the U band in the near infrared was antici-

pated to result in interesting temperature dependence of the electron-phonon coupling

and lattice modes in the middle and far infrared frequency regime, respectively. In

addition to their work on the near-infrared and optical regime, the aforementioned

authors6 also examine changes in the electron-phonon coupling by monitoring changes

in the CN stretching mode (Ag symmetry) as a function of temperature. To com-

plement their work, we have done systematic measurements in the complete infrared

frequency range. This is the first far-infrared work on the well studied K- compound.

We are presently unaware of any infrared measurements on single crystal Rb-TCNQ

samples. Here, our discussion will concentrate upon the far-infrared results in these

two materials.

Furthermore, the combined study of these quasi-one-dimensional salts will high-

light the importance of electron correlation effects, which are expected to be large.

In addition, I hope to briefly discuss the role of disorder in the structural phase

transitions of these materials.

Luminescent Polymers

Recently, there has been a great deal of interest in the photo-physical properties of

conjugated polymers, particularly since the discovery in 1991 of visible light emission

from a poly-phenylene-vinylene (PPV) based diode.31'32 Blue-green light emission

has also been reported for devices based upon both poly-p-phenylene (PPP) and

poly-3-alkylthiophene, suggesting that electro-luminescence of conjugated polymers

may be a more common phenomena than originally anticipated.33'34 Based upon

this supposition, researchers have begun to search for other potential light emitting

polymers. NlatCri ls which can support the polaron or bipolaron defect state, thought

to be important in the mechanism of visible light emission, are prime candidates for

such a study. In this work, we examine several materials in which the bipolaron defect

is electronically isolated.

The field of electro-luminescent polymers is relatively new, and thus many funda-

mental questions are not well understood. Many are long-term problems, and beyond

the scope of this work; however, they are still interesting to consider. They include a

concerted discussion of the mechanism of light emission, and a correlation of chemical

structure to photo- and electro-luminescence properties-most specifically, emission

frequency and efficiency.

The work detailed in this dissertation was motivated by several goals. The first

was to characterize the absorption and emission properties of several potential electro-

luminescent materials. Such characterization is a necessary preliminary for the inves-

tigation of electro-luminescence.

A second goal was the search for new electro-luminescent polymers amongst this

class of materials with discrete emitter units. It is of particular interest to find a

candidate material for use in a blue-emitting diode device. Blue-emitting materials

are rare and have been reported primarily in inorganic materials.

It is also of general interest to begin to develop qualitative ideas for what aspects

of the structure result in photo- as well as electro-luminescence. To this end, we

have studied several materials; the chemical structure of the various polymer repeat

units are displayed in Fig. 2, Fig. 3 and Fig. 4. These materials are primarily based

upon poly-meta-phenylene, poly-para-phenylene, or a symmetrically substituted (2-

thienyl)phenylene repeat unit.

The materials in Fig. 2 and Fig. 3 and allow the systematic examination of a series

of discrete emitter units, establishing a fundamental link between the conjugation of

the emitter unit and the related absorption and emission properties. It is also of

interest how these chemical considerations affect the photo- and electro-luminescent

lifetime and efficiency.

The side chains of the 1,4-bis(2-thienyl)phenylene (PBTB) based polymers Fig. 4

have been selected with a different goal in mind. Here, we seek to alter main-chain

conjugation and the band gap of the polymer with the use of side chains. Two main

effects are envisioned. First, the electron-donating or with-drawing character of the

side chains may alter the character of the main chain conjugation, thus increasing or

decreasing the semiconducting gap. Secondly, the hexyl side chains may cause steric

hindrances, which twist the main-chain backbone from its more planar configuration,

thus isolating the emitter unit; the increased ring torsion angle would also affect the

size of the band gap. In contrast, the dodecyloxy side chains produce very little steric

hindrance near the chain backbone.

It is a final goal in this study to demonstrate that electro-luminescence in polymers

is not an isolated or rare phenomena, and that within certain limitations of chemical

structure, photo- and electro-luminescence are intrinsic properties of many conjugated

organic polymers.

Dissertation Outline

There are two main thrusts to this dissertation: (1) the infrared and optical

investigations of quasi-one-dimensional organic charge transfer salts and (2) photo-

and electro- luminescence measurements on various types of semiconducting polymers.

Consequently, several Chapters (2, 3, 4, and 8) will be divided into separate sections,

to address these two topics in an organized manner.

This document is organized in the following way. Chapter 2 presents the theoreti-

cal methods which pertain to the analysis of the results contained in this dissertation.

Chapter 3 outlines previous experimental work in the areas of quasi-one-dimensional

charge transfer salts and luminescent polymers, while Chapter 4 discusses the ex-

perimental apparatus and equipment used in these investigations. For continuity,

the Results and Discussion of the quarter-filled charge transfer salts are presented in


Chapter 5, while the Results and Discussion of the half-filled charge transfer salts

are given in Chapter 6. The Results and Discussion pertaining to the luminescent

polymers are given in Chapter 7. The Conclusions are presented in Chapter 8.




y /S

Fig. 1. Chemical structure of the TCNQ molecule and the various cations used in
this study.

Fig. 2. Chemical structure of the phenylene based polymer repeat units used in this



Fig. 3. Chemical structure of the quarterphenylene polymer repeat unit used in this

R =--C6H13 or -OC12H25

Fig. 4. Chemical structure of the 1,4-bis-(2-thienyl)phenylene polymer repeat unit
and side groups used in this study.


Organic Charge Transfer Salts

Due to their quasi-one-dimensional nature, organic charge transfer salts are a class

of compounds which manifest many interesting aspects of one-dimensional physics.

Consequently, the theoretical models which describe these interactions are an impor-

tant component of this dissertation.

In these materials, interactions involving unpaired electrons in the highest oc-

cupied molecular orbital have especially important consequences.35'36'37 The tight-

binding model is often used to obtain a band structure from the overlap of TCNQ

frontier molecular orbitals, while the extended Hubbard model describes the impor-

tance of Coulomb interactions in the chain. In the case of a highly correlated lattice,

the tight-binding band is split into two bands, separated by a wide gap on the or-

der U, the on-site Coulomb repulsion energy. Finally, the Peierls model serves to

illustrate the the effects of electron-phonon coupling in these materials in two ways.

First, the physical origin of the Peierls transition is a Kohn anomaly at 2kF which

causes band splitting and introduces a semiconducting band gap between the valence

and conduction band. Secondly, the accompanying lattice distortion results in the

formation of a charge density wave, which couples to several of the TCNQ vibra-

tional modes. The final pages of this section detail several isolated cluster models

of electron-molecular vibrational coupling; the accurate description of this coupling

process for TCNQ, TMTSF, and ET based salts is still an active area of research.

Hubbard Model

The extended Hubbard Hamiltonian for a one dimensional bii '13 is

H = tZ(c!+l,c + i,+ci+,,a) + I Unj,ni,- + Ii-,ninj. (1)
ia Ia 4ij

Here, cj, and ci,, are the creation and annihilation operators for an electron with

spin cr on site i, respectively, and ni,, = c ,i,a is the number operator for site i.

The first term in Eqn. (1) is the "tight-binding" or "hopping" term. A com-

plete discussion of the tight-binding method is presented elsewhere.39 Briefly, the

tight-binding model describes the overlap of molecular orbitals and subsequent band

formation as a linear combination of atomic orbitals. As the TCNQ charge transfer

salts are essentially molecular in nature, the tight-binding approximation is a reason-

able starting point from which to think about the band structure. The tight-binding

wavefunctions have the form

(r) = ekR1 R), (2)

where O(r) is a linear combination of localized atomic wavefunctions,

E(r)= n(r), (3)

where /(r) is a localized atomic wavefunction. In these equations, R is the set of all

Bravais lattice vectors. Note that the wavefunction in Eqn. (2) satisfies the Bloch


?,(r + R) = eikR'(r).

The tight-binding energy bands take on the form

E(k) = 2tcos(ka) + const, (5)

where a is the interatomic spacing and t is the transfer integral, given by

t = Jf H i+ldr. (6)

Eqn. (5) leads to the definition of the one dimensional bandwidth as 4t. When the

overlap of the basis functions is large, the bandwidth is large, and the mean velocity

of an electron moving through the crystal is high. In this model, the higher energy

atomic levels broaden into bands, while for the deep levels, the overlap is not of

significant magnitude for band formation. Hence, the tight-binding model represents

a very localized picture of a solid while at the same time, it preserves the Bloch

condition of periodic wave functions.

It is well known that the tight-binding model is inappropriate for many systems

because of the neglect of all but the nearest neighbor overlap integrals, the indepen-

dent electron approximation, and the model's inability to treat cases in which there

is a large on-site or near-neighbor Coulomb interaction. For strongly correlated sys-

tems such as charge transfer salts, the interaction terms in the Hubbard Hamiltonian,

Eqn. (1), become important. Due to the difficulty of the fully correlated problem on

an extended lattice, it is necessary to consider these problems in terms of limiting

cases. Hence, the effect of electron correlations will be discussed with reference to the

zero bandwidth approximation and the noninteracting dimer case.38'22

Because there is a great deal of evidence that the on-site and nearest-neighbor

electrostatic interactions are much greater than the bandwidth in many TCNQ com-

pounds, the zero bandwidth approximation38 is a very important limiting case of the

full Hamiltonian (Eqn. (1)). In the zero bandwidth approximation, the ground state

of the system is determined by minimizing

H=21 Vi- nj (7)

subject to the condition

pN = n, (8)

where p is the band filling and N is the total number of sites in the chain. Here,

the on-site Coulomb term was also eliminated via the following reasoning: for p < 1,

the electrons localize as one per site, thus eliminating a substantial contribution from

the U term. For the case of p > 1, holes are the carrier of interest, and then the

aforementioned condition is satisfied.

The ground state of the system has been obtained for a particular class of poten-

tials which satisfy the following two conditions

Vn-+ as n-+oo, (9)


Vn+l + V-_1>2V, for all n > 1. (10)

Under these conditions, as well as the limitation of p being a rational fraction, the

electrons are found to have a periodic arrangement within the chain. In fact, this

arrangement can be regarded as a generalized Wigner lattice. These arrangements are

also predicted to give rise to lattice distortions, such as dimers, trimers or tetramers.

Working with the idea of a distorted chain, we can examine the physics of different

kinds of interactions. For example, Fig. 5 displays the progression from an isolated

dimer to that of a strongly interacting, large U, dimer. This diagram dilpliy's the

increased band width associated with a larger inter-dimer transfer integral, t', and

finally the effect of strong on-site electron-electron interactions, which result in a gap

of order U, the on-site Coulomb repulsion energy.

First we turn our attention to the noninteracting dimer. Rice et al.22,40 have

discussed the case of the completely isolated quarter- and half-filled dimer. In the case

of one electron per dimer, as shown in the left-most picture of Fig. 5, the Hamiltonian

is simply

H = -t(clc2 + c2) (11)

due to the absence of correlation effects and spin considerations. Eigenvalues of the

interacting system are at E = t. Note that this model contains only one transfer

integral, so it is expected that this model will not provide a completely realistic

solution for the extended chain. However, inclusion of a finite bandwidth is expected

to preserve many of the rough features, as shown for the weakly coupled picture in

Fig. 5. Here, inclusion of a finite inter-dimer overlap integral results in a broadening of

the energy levels into bands of width 2t', where t' is the inter-dimer transfer integral.

If the band motion is stronger still, the inter-dimer transfer integral increases even

more (as shown in the strongly interacting picture in Fig. 5), until finally, t = t', which

is the tight-binding picture. The electron-phonon coupling within these isolated dimer

models will be detailed in a subsequent section.

In the case of very large U (compared to t), the electrons are unlikely to doubly

occupy sites. Hence, a gap develops which separates the one electron states from the

two electron states. The gap between the lower and upper Hubbard bands is on the

order of the on-site Coulomb repulsion, U. This localizes the electrons (one on each

TCNQ molecule) in the chain. It is the tight-binding band structure modified by a

large U consideration that is depicted in the final picture in Fig. 5.

Ultimately, it is the relative importance of electron-electron correlation effects

(the size of U and also 17) relative to the bandwidth (4t) which is the important

consideration in the behavior of many quasi-one-dimensional charge transfer salts.41'37

For example, it is the relative magnitude of the aforementioned terms combined with

the band filling which determines the periodicity of the reciprocal lattice vector which

couples the highest occupied states, and thus the nature of the Peierls distortion.

This will be detailed in the next section. In metallic compounds, screening is also an

important consideration.

Peierls Transition

Although the dc conductivity for many quasi-one-dimensional charge transfer

salts increases with decreasing temperature to a certain point, the low-temperature

dc conductivity of most such materials is low due to the onset of one or more

phase transitions which culminate in a semiconducting ground state. The Peierls

transition42,35'36'43 is the most common type of metal-insulator transition.

The Peierls distortion is generally introduced by considering a one-dimensional

lattice of atoms with uniform Lr J,( iiiL, a. Peierls has shown42 that a one dimensional

electron gas at T = 0 K with finite electron-phonon coupling is unstable to a structural

distortion of the form

u, = na + uo sin(2nkFa + qf), (12)

with the constraint that the distortion, uo, is much less than the lattice constant, a.

Here, u, designates the position of the nth atom.

Because the distortion occurs at kF, the total electronic energy of the system

is reduced as the band splits. The filled states are shifted lower in energy and the

unfilled states are shifted higher in energy. Thus, a semiconducting gap is opened at

the Fermi surface. The size of the gap is given as

Egap = 2te-(), (13)

where A is the dimensionless electron-phonon coupling constant. The progression of

the band and geometric structure is diagrammed in Fig. 6 for the case of a half-filled

band with no Coulomb interactions. Here, the 2kf lattice distortion at reciprocal

lattice vector results in a doubling of the spatial periodicity and a reduction (by

one half) of the Brillouin zone.

The formation of the charge density wave ground state is one of the most im-

portant consequences of the Peierls transition. This occurs in the following way. In

the case of a half-filled band, the structural distortion of the uniform chain causes

the grouping of sites into dimeric units (Fig. 6). This pairing alters the charge den-

sity along the chain. Starting from a uniform distribution of charge, the unpaired

electrons contract to form regions of increased and decreased charge density. In an

infinite system, the charge density is given by

p(r) = Po + picos(q-r + ), (14)

where Po is the equilibrium charge density and the second term describes the varia-

tion of the charge density with position along the chain. This phenomena is called

a "charge density wave." If the periodicity of the charge density wave is an inte-

ger multiple of the periodicity of the lattice, the charge density wave is said to be

"commensurate," while it is classified as incommensurate if it cannot match a regular

lattice interval.

For the quarter-filled band, small U case, a 2kF lattice distortion at a reciprocal

lattice vector of results in a quadrupling of the spatial periodicity and a reduction

of the Brillouin zone size by a factor of four. The energy gap introduced due to

the distortion at kE has a form similar to that in Eqn. (13). As in the half-filled

case, the distorted state is characterized by a charge density wave ground state.

Introducing electron-electron correlation effects splits the tight-binding band into

singly and doubly occupied states which are separated by the Hubbard gap of order

U. Thus, at quarter-filling, kF must be twice as large and it can be shown that the

Peierls distortion is 4ky in nature. The large U limit is an especially important limit

for many quarter-filled semiconducting organic charge transfer salts, including those

discussed in this dissertation.

The Peierls transition was a phenomena that was originally predicted to occur

in purely one-dimensional systems at zero temperature. The presence of this type

of metal-insulator transition in quasi-one-dimensional organic charge transfer salts

and other materials at finite temperature is primarily a consequence of the weak

three-dimensional interactions.

In a manner similar to the electronic Peierls distortion, described above, magnetic

effects can cause a spin Peierls transition, where the opening of a spin density wave

gap is accompanied by the formation of a spin density wave ground state. Several

factors have been shown to influence the behavior of spin Peierls systems; among the

most important are large Coulomb (U) interactions which essentially prohibit double

occupation of sites and, in the presence of finite spin-phonon coupling, favor the spin

density wave ground state. A good discussion of the spin Peierls transition and the

resulting implications for the transport properties has been presented elsewhere.44'36

Models of the Electron-Phonon Interaction

In the past decade, there has been an extraordinary amount of interest in the

phenomenon of electron-phonon coupling, as it has been observed in a wide vari-

ety of quasi-one-dimensional organic charge transfer salts. Several models have been

developed to aid in the understanding of the interesting manifestations of this cou-

pling process. Isolated cluster models have enjoyed particular success due to their

applicability combined with their relative simplicity. As isolated cluster models have

been used extensively in this work to understand the infrared frequency dependent

conductivity, they will be discussed here. Several good reviews of this subject are


The solution of the electron-phonon interaction for an isolated TCNQ dimer with

one electron per two TCNQ molecules has been presented by Rice et al.22 This

model is intended to mimic the behavior of a quarter-filled, infinite chain of poorly

interacting, distorted dimers. The Hamiltonian in the isolated dimer model is

H = He + Hv + EgnniQn,i Ei.-, (15)

where He and H, are the electronic and vibrational Hamiltonians, which describe

the electronic and vibrational properties of the system in the absence of the vibronic

coupling. These terms are given by

H = t(ctl,c + ct C2,i) + (Eo + Ac)ni + (Eo Ac)n2 (16)


Hv = E (P, + Q ,), (17)

respectively. Here, (Eo+Ac) denotes the energy of the 7r orbital of monomer 1 before

interaction, and (Eo-Ac) is the energy of the 7r orbital in monomer 2, also before

interaction. Eo is the energy of the 7r orbital in the absence of charge asymmetry,

while Ac refers to the site inequivalence, reflecting the fact that the unpaired electron

is not shared evenly between the two sites. The linear electron-molecular vibrational

(EMV) coupling is described by the third term in Eqn. (15). Here, the g, are

the linear EMV coupling constants and the Qn,i are dimensionless normal mode

coordinates of the Ag modes, and the Pn,i are complex conjugates. Each normal

mode is also characterized by a frequency, w,. The final term in Eqn. (15) describes

the interaction of the unpaired electron with an externally applied field; J) denotes

the electric dipole moment of the n-mer, and is given by

p= 2(ed(ni n2)) (18)

In practice, 6 is the inter-dimer spacing. In the aforementioned eciaiti .n'-. c,, and

ci,, are the usual creation and annihilation operators for unpaired Fermions on site

i, and ni = c'ci is the occupation number operator. Indices are defined as follows: i

denotes the monomer or site number, n indexes the vibrational modes, and a refers to

the electron spin. Note that only one electronic charge transfer excitation is allowed

within this formalism, its strength determined by the transfer integral, t. Note also

that the on-site Coulomb term, U, is zero in the quarter-filled case.

From Eqn. (15), the equations of motion for the dimensionless normal mode

coordinates, Qn,i,22 are

Qn,i + wn,i = -2_gnnnni. (19)

Introducing dimeric normal mode coordinates, sn and rn, the equations of motion for

the modes of interest are rewritten as

Sn+ ,Sn = -22gnwan(n1 + 712), (20)


+ Wr = -22gzwn(n, 72). (21)

Noting that (?i2 + n2) is a constant of the motion, we conclude that the symmet-

ric modes, s,, are not coupled with the motion of the radical electrons.22 These

symmetric modes are also called "amplitude modes," and can be observed in Raman

spectroscopy. It is the antisymmetric or "phase phonon modes" which are of primary

interest.24 Referring to Eqn. (18), we note that any change in the dipole moment via

interaction with the external field will excite the antisymmetric modes in the above


The frequency dependent conductivity within the quarter-filled isolated dimer

model22 is given as

(iwea2 N X(w)
w ( 4S2 1 D(w)X
k \ 4 Q ) K X ( ))' (2 2 )

where a is the dimer separation, N is the number of dimers per unit cell, and Q is

the volume of the unit cell. Here, D(w) is the phonon propagator

D(w) = ,L2- (23)
L0 Lw2 j-;'

where the An are the dimensionless electron-phonon coupling constants, w,, are the

unperturbed frequencies, and 7,, are the phonon linewidths for each Ag mode, respec-

tively. The reduced electronic polarizability, X(w), is

X( Wc, (24)
W( ) 2 2 ;o}
WCt LU e

where wct is frequency and -y is the linewidth of the electronic charge transfer exci-

tation. The dimensionless coupling constants are written as

A = X(0) 8t (25)
Wn wctwn,

where X(0) is the zero frequency limit of Eqn. (24). It is the g, which are most

commonly cited (in a generic fashion) as the electron-molecular vibrational (E.11V)

"coupling constants," and are the fundamental microscopic quantities of interest in

this analysis. The units of X(w) and X(0) are centimeters, while D(w) and A,, are

dimensionless; the g,, are in units of wavenumbers. Thus, the computed frequency

conductivity spectrum, a(w), has units of Q-1 cm-1.

For the half-filled dimer, U is no longer zero.40 Although Eqn. (22) and the

expressions for the phonon propagator remain unclia;iiged. X(0) and X(w) must be

modified to account for the effect of Coulomb repulsion.40 Thus,

X(w) (, 2 (26)

Taking the zero frequency limit of the above .c:.pressik.ii, we find

X(0) = ( (27)
A)t( 4_+ 4t2)0.5

The expression for the dimensionless coupling constants,

An = X(0).9, (28)

is modified accordingly. These results were first presented by Rice,40 as well, and saw

rapid application to half-filled, dimerized TCNQ based 5,y.1in-n.45

Interaction of the Ag phonon mode with the charge transfer excitation causes

the resonance frequency of the vibration to be shifted to lower frequency from its

unperturbed position in an isolated molecule. Within the isolated dimer model, we

can calculate

n +

Here, A and B are constants, relating to the dimensions of the unit cell and the

electronic parameters, respectively. Thus, by fitting a sum of Lorentzian oscillators

to this function, we can obtain the unperturbed frequencies of each Ag mode.28 In the

conductivity spectrum, these modes are generally observed to have Lorentzian line

shapes; however, when they overlap with the electronic band, they have a Fano typer''

(or derivative) line shape. The isolated dimer model was also used as a basis for the

first proposal as to the origin of the fine structure in the conductivity spectrum.47

In addition to presenting a model of the infrared conductivity, the dimeric cluster

models have been used to describe charge transport in semiconducting salts via an

activated hopping mechanism and to estimate the activation energy for that charge

transfer.48'49 This approach is valid for strongly localized salts.

In order to describe a wider variety of materials, the ideas of electron-phonon

coupling in isolated dimeric clusters were extended by Vartsev to include trimers,

tetramers and n-mers.23'3 The Hamiltonian for the isolated tetramer model is

H = He + H, + gnniQn,i E-I. (30)

For the case of the quarter-filled tetramer,

t t
He = t (cc2, + c2 cI1, + C3,c4,, + 4 C3,a) + t (C,c, + ,t'(ce,)

+ U Eni,,ni,_, + V(n1n2 + n2n3 + n3ln4) + A(n, + 724) + A(72 + N3) (31)


H ,i= 40(P,+ OQ,)w (32)

describe the behavior of the unpaired electrons and the symmetric intra-molecula.r

vibrations of the uncoupled material. As before, the electron-phonon coupling is

incorporated in a linear manner with the third term in Eqn. (30). In this model, the

TCNQ molecules are arranged as ABBA within the tetramer. It is assumed that

t, the transfer integral within the AB molecular pairs, is larger than t', the transfer

integral within the BB molecular pairs; a negligible transfer integral is assumed

between AA moieties. As the solution for this case of two electrons per tetramer

is similar, although more complicated, to that outlined above for the quarter-filled

dimer,23 only the final results will be presented here.

The frequency dependent conductivity for the isolated tetramer i,,,,h 1- is given


r(W) = -iwN, X1A2 + 2X12AB + X22B2 + (X2 Xn22)(A2Dt, + B D,,)
r(uj) = -twJVe"e---------- -------- .
1 Xa1Da X22Db + (X1X22 X)DaDb
The phonon propagator has a similar form as in the isolated dimer model,

Da,b 2 ---- (34)
n n;a,b '

but the subscripts a and b allow for charge differences on the A and B molecules.

The parameters A and B are defined A = (2x + x')/2 and B = x'/2, where x and

x' are the molecular overlaps for the AB and BB molecular pairs, respectively. The

reduced electronic polarizabilities are given by

I(OIni 411) 1j2w (35)
X11(w) = Z, (35)
0 W2 w^j#

X12(LJ)= lX21(w") = (nl -- 41-)(I1 2 n(36l)2w- 3)
W3 1 -/ -- ZOy3


X WI,7 -1L,03 (37)

Here, wup is the energy of transition from the excited state, /, to the ground elec-

tronic state. Mh iii: elements in the Xij are calculated as described in Ref. 23, with

the coefficients for the ground state wavefunction chosen based upon the relative

size of various parameters such as U, V, y, and -. In practice, the initial choice

of the aforementioned parameters is based upon chemical considerations, and these

parameters are varied by trial and error to obtain the best fit. The coefficients for

the excited state wavefunctions are determined from the equations presented in the

Appendix of Ref. 23. In this model, parameters are generally chosen to allow two

low energy electronic excitations in the mid-infrared, which couple strongly to the Ag

vibrational modes, and two excitations at much higher energies.

The success of the isolated cluster models22'40 and the phase phonon theory24

at explaining the physics of the electron-phonon interaction in a variety of quasi-

one-dimensional charge transfer salts has been well documented. Discrepancies have

generally been attributed to the extended nature of the compound under study. To

this end, periodic cluster models were developed.50'5, To obtain a more complete

understanding of the electron-molecular vibration interaction in less localized systems,

several additional models have been put forward. Of particular note is a two-fold

degenerate charge density wave model by Bozio et al.25 This model has been applied

to several materials with notable success, allowing an investigation as to the nature of

the gap as well.9,8,52 Finally, it has also been noted that the aforementioned models

have difficulty with regard to the intermediate Coulomb interaction limit.

However, just as important as a perfect model fit is the physical insight and in-

creased understanding of the microscopic parameters that even simple cluster models

bring to polarized infrared measurements on quasi-one-dimensional materials with

unpaired radical electrons.

Luminescent Polymers


When a material absorbs a photon of energy greater than the band gap, an elec-

tron is promoted to the excited state. Relaxation of the excited state ni'.' occur

through a number of competing processes which may be characterized as being either

radiativee" or "nonradiative." Radiative processes occur with the spontaneous emis-

sion of light, whereas nonradiative processes involve dissipation of energy through a

variety of internal conversion mechanisms, including multiphonon processes (heat), in-

tersystem crossing, Auger processes, and other complicated internal conversion mech-

anisms involving collisions or impurities.53'54'55,56

The most basic model of fluorescence considers only the radiative process. A

simplified energy diagram is shown in Fig. 7 for a spin singlet material. Absorption

of light excites an electron from its ground state, So, to a higher energy singlet surface,

S1 (or S2). After a brief lifetime in the excited state, the electron can decay back to

the ground state by emitting light (fluorescence).57'58 Emission can occur from either

an intrinsic band or an impurity level.56 Of course, transition to an excited state and

subsequent re-emission from that same state, such as a band-to-band transition, is

not generally useful, as the emitted light is simply reabsorbed. This simplified picture

in Fig. 7 is greatly affected by competition from nonradiative processes.

A more realistic picture of the energy levels of a molecule and the available

ways of decay from the excited state is displayed in Fig. 8. Note that there are two

distinctly different pathways of radiative decay: fluorescence and phosphorescence.

The fluorescence emission process becomes more complicated due to vibrational

considerations.54 This is because, in practice, the electron is promoted to an excited

vibrational level on the Si surface. Decay rarely occurs from here. Instead, the

molecule may loose energy via nonradiative vibrational processes, (.,, ..lii; down-

ward toward the lowest energy vibrational state on the excited surface.53'54'55 The

final return to the ground state electronic surface may occur via radiative or non-

radiative mechanisms. Thus, the frequency of emitted light is less than that of the

exciting source due to losses associated with vibrational motion.57'58

The alternate mechanism of radiative decay, phosphorescence, is more compli-

cated. Excitation of the electron to the S1 surface proceeds with the absorption of

light as described in the preceding paragraphs. However, before phosphorescence can

be observed, the system must convert from a singlet to a triplet spin state. This radi-

ationless process is called intersystem crossing.53'54'55 (Note that excitation directly

to TI, the excited triplet surface, is spin-forbidden unless the original ground state

is a triplet.) The electron can decay from the excited triplet state with the emission

of light. Because of Hund's rule considerations, the triplet spin state is more ener-

getically favorable than the singlet state. Consequently, phosphorescence emission is

commonly observed at a lower energy than fluorescence.

The spin selection rules of these two radiative processes are reflected in the lu-

minescent lifetimes and efficiencies. The lifetime of fluorescence is normally on the

order of 10-11 to 10-7 seconds, whereas phosphorescence occurs over a time span of

10-3 seconds to several hours.53'55,59

Finally, it is interesting to note that although the intensity of emission is com-

monly a function of excitation energy, it almost always occurs at the same en Ll'., with

similar spectral shape. This is because decay always occurs from the lowest vibra-

tional state on the S1 or T, surface (or an impurity level) after suffering nonradiative

vibrational losses-regardless of the excitation energy.53'54

In many cases, neither fluorescence nor phosphorescence is observed. This indi-

cates that the nonradiative processes are occurring more rapidly than the radiative

ones. One mechanism of nonradiative decay is "internal conversion," in which the

absorbed energy is completely converted to heat by using il,' -', -it il (i' ..- ii-Iu-. colli-

sion processes, conformational changes and vibrational losses to cascade down to the

So state.53'54,55 In this manner, an excited electron may make its way down to the

ground state without emitting any light at all! Thus, the relative rates of internal

conversion and spontaneous emission are iIy important.

Fluorescence is known to be a first-order rate process53'54 with rate constant given

as kf. This decay process may be represented as:

A* A + hv. (38)

The nonradiative dl, ..i processes (internal conversion and iiii -II in crossing) are

also described by a pseudo first order rate expression

A* -+ A + heat, (39)

with rate constant k,.r. Quenching is a third possible mechanism for molecular relax-

ation, but it will not be included in this discussion. 53'5455 The two aforementioned

competing processes can be modeled as two first-order reactions in parallel. Hence,

the overall rate expression for decay is given by:

=- d kfA* + k, A*. (40)

Separation and integration yields the usual first-order exponential decay:

A* = Ae-I, (41)

where K is the combined rate constant (kf + knr).

At this point, it is natural to define an emission efficiency (or "quantum ) idd")

as the number of quanta emitted by fluorescence divided by the number of quanta

initially absorbed.55 The quantum yield, 4o, is give by

4o = k A'- (42)
kfA* + kr,,A*

Thus, as the radiative emission pathway becomes more favorable, Fo increases.

Although the intrinsic fluorescence efficiency of many materials is very high, self-

absorption, surface scattering, and partial reflection at various device interfaces sig-

nificantly reduce the external device efficiency.53'56

It is an active area of research to identify the structural elements in materials

which affect this balance between radiative emission and internal conversion.54'55'59

Aromatic rings and other double bond combinations are known to facilitate fluores-

cence due to their rigidity and resonance character. Molecular rigidity diminishes

the rate of internal conversion as well as intersystem crossings-thus reducing losses

due to both internal vibrations and decaiy via phosphorescence. Electrons in con-

jugated bonds facilitate radiative recombination as they are easily promoted to the

excited state. The electron donating or withdrawing substituents affect the resonance

characteristics of a ring or double bond structure and, thereby, the emission inten-

sity. In addition, samples in the solid state have lost additional degrees of freedom,

with the result that internal conversion is less probable. Finally, impurities such as

halides, metals, residual monomer or solvent molecules may act to reduce or "quench"

fluorescence intensity via a second order rate process.

One of the most common uses of luminescence spectroscopy is to understand the

excited state geometry and electronic properties relative to those in the ground state.

The theoretical basis for such studies derives from the Franck-Condon Principle.53'54

The Franck-Condon Principle is based upon the idea that electronic motion is much

faster than nuclear motion; thus, the electronic charge density is always assumed to

be at a stationary state with respect to the nuclear skeleton. Four distinct steps

characterize the total absorption and emission process. Immediately after excitation,

the molecule is in the excited electronic state, although it is still in the ground state

geometry. This is called the Franck-Condon excited state. Within a short time,

the molecule relaxes to the lowest energy vibrational state on the 5' surface (the

"equilibrium excited state"), having acquired a more energetically favorable geome-

try which is compatible with the new distribution of electron density. Upon emission

of a photon, the molecule decays to the Franck-Condon ground state, which, imme-

diately after elii-iin. is the same as that in the equilibrium excited state. Upon

thermal relaxation, the molecule reaches the lowest energy equilibrium ground state.

characterized by the equilibrium ground state geometry.

As the radiative decay occurs from the equilibrium excited state geometry, the

emission spectra, itself is characteristic of the changes in geometry, electron and vi-

brational structure between the ground and excited state.53'55 These changes are

evaluated in two ways. First, the magnitude of the energy difference between the

absorption and emission band maxima, the Stokes shift, reflects the extent of geom-

etry relaxation between the ground and excited states. A large (greater than leV)

Stokes shift is indicative of large relaxation effects and geometry changes between

the ground and excited state. Secondly, the "mirror image" relationship is used to

compare the shape of absorption and emission spectra.53'59 Similarly shaped curves

imply a similar spacing of vibrational levels, while large differences in the emission

spectra suggest significant changes in the vibrational structure in the equilibrium ex-

cited state as compared to the ground state. These differences are most often due to

anharmonic effects in the upper energy bands. Consequently, measurements of ab-

sorption and photo-luminescence are commonly combined to access both structural,

electronic and vibrational differences between the ground and excited states.

For semiconducting polymers with non-degenerate ground st at(-. such as those in

this study, the radiative decay process is more complicated than described above due

to the role of intra-chain defects.60'61'62'63 A simplified diagram of the likely excitation

and decay process is shown in Fig. 9. The initial excitation promotes an electron from

the valence to the conduction band. This is a (7r,7r*) excitation which relaxes into

positively and negatively charged polaron sites. Due to the quasi-one-dimensional

nature of the chain and the strong electron-phonon interactions inherent in such

systems, the charge is thought to relax (by a nonradiative process) into a localized

polaron exciton (or neutral bipolaron exciton) state. The formation of such a charge

defect is also accompanied by a lattice distortion. A diagram of a neutral bipolaron (or

polaron exciton) charge defect is shown in Fig. 10 for poly-p-phenylene. This singlet

polaron exciton is thought to be the luminescent center in these materials.60'32 Note

that the neutral bipolaron exciton is lower in energy than two polaron defects because

of the Coulomb attraction of an electron and a hole.64 After a short lifetime, the

singlet exciton d'er :, s to the ground state with the emission of light. Nonradiative

relaxation lowers the energy to the equilibrium ground state. Note that radiative

decay will always occur at energies less than the band gap, thus highlighting the

importance of a large gap for the attainment of visible light emission. The depth of

the polaronic levels within the gap (with respect to the valence and conduction band)

is also an important consideration. If these levels were positioned too close to the

band edges, self-absorption would interfere with the decay process.

There are many fast nonradiative decav mechanisms which can complicate this

picture. Several, such as conjugation length and interrupting defects, multiphonon

processes and conformational effects, are particular to or more pronounced in macro-

molecular systems.60 The degree of conjugation affects the confinement and mobility

of the exciton, whereas vibrational and configurational considerations dictate the

number of available nonradiative relaxation pathways. The complexity of the poly-

meric material clearly provides an increased number of internal conversion pathways

due to additional vibrational and configurational degrees of freedom. Cliiiugated

polymers can also undergo intersystem crossing to the triplet state and subsequent

radiationless de'cavy. In addition, Friend et al.60 -iier--ted that various collisions

between polaron pairs or between polarons and bipolarons lead to the destruction

of these excitons resulting in radiationless decay via multiphonon processes. Such a

mechanism has been shown to be very efficient in highly conjugated polymers.


Photo-luminescence is a necessary condition for electro-luminescence. Thus, a

steady state photo-luminescence measurement is a useful screening process in which

to sort out potential electro-luminescent light emitters. However, not all materials

which photo-luminesce display electro-luminescence, primarily due to tie different

manner in which the excitation is achieved. In electro-luminescence, the electrons in

the excited state originate from charge injection rather than photoexcitation. Thus,

in electro-luminescence, electrical energy is converted to light.

Charge injection generally proceeds through a forward biased pn junction."5 A

simplified diagram of a pn junction (with zero applied voltage) is shown in Fig. 11. The

region of interest is the "depletion region," on either side of x = 0. In the depletion

layer, there are no free carriers-only ionized impurities or bound charges, with a

distinct concentration profile through the depletion n-'iigiii caused by the difference

in chemical potential (Fermi level) of the two materials. The spatial arrangement

of the ionized impurities produce an electric field across this region resulting in a

potential barrier, Vo, between the n and p sides of the contact (Fig. 11). This voltage

difference represents a barrier to the movement of positive charge toward the n-doped

region and vice versa. A wider depletion layer results in a larger potential barrier.

Upon application of an external voltage, the carrier acquires the necessary kinetic

and potential energy to overcome the depletion layer barrier, Vo. Hence, it is "in-

jected" into the conduction band of the other layer. I lie size of the applied voltage

modifies the voltage across the depletion layer, as well as significantly affecting the

width.39,65 Charge injection is facilitated if the electrons move in the direction of

the barrier (positive bias), while if the carriers move against the barrier, they must

overcome the potential Vo (negative bias). This is the origin of the rectifying behavior

of a pn junction.

At a sufficiently high voltage, the depletion layer becomes so wide that a new

mechanism begins to supply the charge carriers.65 Under these conditions, the cur-

rent/voltage relationship develops large non-linearities so that a small increase in the

applied voltage results in a large increase in the diode current. The two possible

"breakdown lc'-llIJiiilll.-" generally occur between 5 to 7 V, or greater than 7 V,

for the Zener and avalanche effects, respectively.65 The Zener breakdown mecha-

nism seems most applicable to our studies. It occurs when the electric field in the

depletion layer becomes so large that covalent bonds are broken, resulting in the

generation of electron-hole pairs. The large number of electrons and holes created

in this way are swept (by the field) across the junction to the opposite side, where

they may recombine.65 The avalanche breakdown is a collision process, whereby the

carriers themselves break bonds. The exciton pairs created in a. breakdown process

can recombine with the emission of light.56

For two equally doped semiconductors, the width of the depletion 1,,i.'ii is the

same on either side of x = 0. However, in cases of unequal duping, the depletion

layer will extend more deeply into the more lightly doped material.65 In the extreme

case of a metal/semiconductor contact (such as in our samples), the metal has no

depletion layer. Thus, the depletion region and potential barrier exists entirely on

the "polymer side" of the junction. This limiting case is called a Schottky diode.65

The mechanism of charge injection in polymer based diodes is less well understood

than those involving conventional semiconductors due to the increased complexity of

the exciton being formed. In the case of a conjugated polymer with a non-degenerate

ground state, the exciton is thought to be a neutral bipolaron exciton (also called a

polaron exciton). It has recently been sl.ggested that charge injection in PPV may

proceed via tunneling through the potential barrier directly into the upper polaron

gap state in PPV and a modified PPV system.32 It is not known if this is a general

mechanism for light emission in conjugated polymers. The authors32 also report that

thermal activation plays no role in the emission process in these PPV based systems.

It remains to be investigated for other materials via EL intensity vs. temperature


There is generally good correlation between the photo- and the electro-

luminescence spectrum for a wide variety of materials. Stokes shift and mirror image

similarity considerations have also been reported to be nearly identical in the PPV

based systems.31,32 It remains to extend these preliminary investigations to lumines-

cent polymers of varying chemical structure to ascertain whether this is a general phe-

nomena for these materials. The similarity between photo-and electro-luminescence

spectra has suggested that the same luminescent center and pathway of decay may

at work in both cases, although, as indicated above, the manner in which the exciton

is obtained is quite different in the two cases.56'60'32


Ef 1 Ef


t- -





Fig. 5. Evolution of the band structure with the degree of interaction. This figures
displays the progression from the zero bandwidth picture to the finite bandwidth
picture, and finally to the large U picture. Note that V = 0 here.






- --4-

Large U

Large U









- T/a

Fig. 6. Structural distortion and band structure modification due to a Peierls dis-
tortion on a half-filled uniform chain.



Fig. 7. Simplified picture of the band-to-band radiative decay process.





Fig. 8. Two pathways of radiative decay (fluorescence and phosphorescence) are
illustrated. Nonradiative decay can also take place after excitation.


* IBand



Ground Excitation Polaron
State Product Exciton


Fig. 9. Possible decay mechanism in conjugated polymers. Note that the levels
within the gap arrise due to the formation of polaronic excitations.


Fig. 10. Neutral bipolaron exciton in poly-p-phenylene polymer. Note that the
charge defect can travel along the chain due to the extended nature of the conjuga-
tion in PPP.


Holes (p) Electrons (n)
+ + + --
+ + + I- + I- -
+ + + I + -


Fig. 11. Simplified diagram of a pn junction.


Quasi-One-Dimensional Oranric: Chi.ia g' Tr;iarii S;,l1-


TCNQ charge transfer salts are the prototypical example of intrinsically con-

ducting, quasi-one-dimensional organic solids. In addition to the electrical conduc-

ti' il.', these materials have a wide range of other interesting physical properties.

These include: an unusual "stacked" crystal structure, which results in the highly

anisotropic dc conductivity, magniiti: and spectral properties, as well as electron-

phonon coupling effects, large electron-electron correlations, and numerous structural

phase transitions. Fundamental ideas from the field of one-dimensional physics (such

as the Hubbard model, Peierls distortion, charge and spin density waves) have been

applied to these materials with good success.

The dc conductivity of various TCNQ based charge transfer salts has been re-

ported to span more than eight orders of magnitude. For example, metallic con-

ductivities of up to 1000 ohm-1 cm-1 have been reported in the chain direction for

samples of TTF-TCNQ at 300 K,66 while semiconducting charge transfer salts have

conductivities less than 10-5 ohm-' cm-1. The anisotropy in the dc conductivity of

single crystal samples has been reported to be as high as 1000:1 for TTF-TCNQ,G6

and is commonly on the order of 200:1.

Resistivity measurements as a function of temperature show that in most of these

materials, the dc conductivity increases as temperature decreases. Compounds with

metallic behavior at low temperature (such as TTF-TCNQ) have historically com-

manded special interest.66 Unfortunately, most conducting TCNQ salts undergo a

metal-insulator transition at low-temperature,35 inhibiting the formation of the low

temperature metallic or superconducting state. Various magnetic phenomena also

compete with a low-temperature superconducting state in one dimension.67

For these reasons, scientists were motivated to search for other materials with ei-

ther a lower Peierls transition temperature or no structural distortion at all. Increased

dimensionality played an important role in reducing Tc.68 These efforts led to the

discovery of the BEDT-TTF (ET) based charge transfer materials, as well as the fam-

ily of Bechgaard salts, based upon TMTSF.30-70 The first ambient pressure orar-'ic

superconductor was discovered in the family of Bechgaard salts. Superconducting

phases of the ET salts have also been reported.

As synthetic metal materials have been the subject of intense study during the

past three decades, the number and diversity of experimental investigations to date

in this diverse field is extremely large. Therefore, I will attempt to present and

summarize only the major findings, concentrating on the TCNQ based compounds,

particularly those of interest in this work.

Structural Considerations

The strong anisotropy in these materials, from which the interesting physical prop-

erties derive, is known to result directly from the distinctive quasi-one-dimensional

structural architecture. In these crystals, the TCNQ molecule acts as a monomer

repeat unit, stacking regularly in quasi-one-dimensional columns or chains. The

columns are commonly segregated on the basis of donor or acceptor species. The

high electron affinity of the TCNQ molecule makes the formation of charge trans-

fer salts energetically favorable. The degree of static charge transfer is depeld''lit

both upon the nature of the cation and the ratio of the counterion to the number

of TCNQ molecules in the crystal. For example, in the 1:1 K-TCNQ salt, there is

virtually complete static charge transfer, resulting in one electron on every TCNQ

site, and a half-filled band. Incomplete static charge transfer is observed in the case of

TTF-TCNQ. The degree of static charge transfer is commonly accessed by structural

analysis or vibrational analysis.7172

Within the stacks of TCNQ molecules, the distribution of molecules can be uni-

form or arranged in clustered units of "n-mers."3 The nature of this stacking ar-

rangement has a large influence on the transport properties. For example, a uniform

distribution of TCNQ molecules usually results in metallic properties, while a ma-

terial with a distorted stack has semiconducting properties, due to the opening of

the Peierls gap, 2A at the Fermi level (Fig. 6).36 A dimeric distortion is the most

common; K-TCNQ, IMEMi(TCNQ)2 are well-known members of this class. Within

the dimeric unit, the overlap geometry is commonly a ring-bond or slightly shifted

ring-ring type overlap, while the external overlap is usually a, strongly shifted ring-ring


For distorted systems, the electron configuration within the n-mer plh. an im-

portant role in the ease of electrical transport. Compared to 1:1 salts with a half-filled

band, salts with a fractional band filling have greatly facilitated transport. This is

because electrons are less likely to hop to doubly occupied sites in these systems, re-

sulting in reduced Coulomb interactions. In addition, as lattice distortions change the

TCNQ stacking arrangement from one spacing to another, the electron configuration

determines whether the charge density wave will be commensurate or incommensurate

with the underlying lattice.

The distance between the TCNQ molecules in the stack and the mode of overlap

between the TCNQ molecules in the stack determines the overlap of the electronic

wavefunction, and thus, the transfer integral, t. The transfer integral is commonly

on the order of 0.1-0.3 eV for TCNQ based materials.4'35 Ultimately, it is the value

of 4t (the bandwidth) relative to U (the on-site Coulomb repulsion) which is the

important quantity in these salts, as the ratio plays an important role in determining

the periodicity of the Fermi wavevector. The electronic interactions perpendicular to

the stacking direction are less favorable than along the chain direction.

A phase transition is almost always accompanied by a change in the lattice peri-

odicity, and the nature of these structural changes can be ascertained by X-ray and

neutron scattering measurements. As such changes alter the size of the Brillouin

zone, they strongly affect the distribution of charge density in the material. Hence,

the nature of a structural distortion is also reflected in physical methods which probe

quantities related to the electronic structure, such as transport, magnetic and spec-

troscopic properties.

Transport Properties

Structural phase transitions almost always result in a change in the dc conduc-

tivity as a function of temperature. In fact, it is usually through such transport

measurements that a phase transition is identified. For example, one interesting phe-

nomena that is easily observed in the dc conductivity is the "metal-insulator" tran-

sition. As the name implies, the dc conductivity drops significantly due to opening

of an energy gap at the Fermi surface. The size of the gap is commonly temperature

dependent. Thus, the low temperature metallic state remains unrealized for most

compounds. The Peierls transition is a type of metal-insulator transition. A spin

density wave transition can also affect the transport properties, opening a gap at the

Fermi level. Here, the energy gap represents the difference between the single and

triplet spin states. A third, and less well-studied, type of structural phase transition is

one driven by order-disorder considerations; it also results in a small dc condu I i. ii'..

The onset of a structural phase transition is often evident in the magnetic prop-

erties of these quasi-one-dimensional materials, and information regarding the nature

and driving forces of the structural distortion can be gleaned from these measurements

as well.73'7 For example, in 1:2 compounds with large U, the magnetic susceptibil-

ity is unaffected at a 4kF distortion, whereas a 2ky system is characterized by a

sudden drop in Xm at Tc.20 Thus, magnetic susceptibility measurements can distin-

guish between a magnetically driven structural phase transition, such as a. spin Peierls

distortion, and a phase transition which is dominated by other considerations.

Spectroscopic St ,il,.-

The unique structural architecture and various charge distributions of the various

TCNQ charge transfer salts are strongly manifested in the spectroscopic properties.

As the major thrust of this dissertation, this subject will command special attention

in this review.

The polarized infrared and optical properties of numerous TCNQ charge transfer

salts have been reported at room temperature. In the many good review articles

in print,4'18'2019'30 the authors note the extreme anisotropy between various crys-

tal directions and clearly demonstrate the spectroscopic uniqueness of the TCNQ

stacking direction. The measurements by Brau et al. on single crystal samples of

TEA(TCNQ)2 make this point particularly well.75 (NMe3H)(I)(TCNQ) is another

good example, as the authors are able to obtain data in three orthogonal polariza-

tions. In the chain direction, the spectrum displays metallic characteristics, while in

the other two polarizations, the spectra are characteristic of a semiconductor with a

few obvious non-Ag vibrational features on a relatively flat background.76 Identifica-

tion of the electronic charge transfer features and their corresponding polarizations

underscored the importance of localized effects and lead to a basic understanding

of how charge is transported in these materials.77'78'75'79 Intra-molecular excitations

(and the polarization dependence of such features in the solid state) were also iden-

In the semiconducting materials, ten vibrational features were observed to dom-

inate the infrared spectrum.75,45,78,77'80 (In metallic compounds, these features are

screened.) Observation of unusually strong Ag vibrational modes polarized in the

chain direction was the impetus for the development of various models of electron-

phonon coupling.24,81,22,23 Three of these models were detailed in the previous Chap-

ter. These models, based upon the interaction between the Ag molecular vibrations

and either a charge transfer excitation or an oscillating charge density wave were suc-

cessful in explaining the basic physics of these coupling processes in many materials.

Electron-molecular-vibrational coupling constants were determined for a wide variety

of materials, and it was concluded that the degree of coupling was similar in many

different TCNQ salts. To this end, Painelli et al.82 developed a set of "universal

coupling constants" for a wide variety of materials, with small differences based only

upon the degree of static charge transfer to the TCNQ from the counterion. Lipari

et al.83 calculated the unperturbed phonon frequencies for the TCNQ- anion. These

were a useful comparison for the frequencies obtained in the aforementioned exper-

iments. Finally, Lipari et al. also presented a modified valence force field model

calculation of the vibrational normal modes of Ag symmetry.

Later studies characterize the individual nature of the various phases in these

materials, which are separated by phase transition of various kinds. Changes in

the electron-phonon coupling between the high and low temperature phases of these

salts commanded special attention, as this interaction is fundamentally related to

changes in the lattice distortion and electronic configuration.22'27'84 For ,,iiiph.

in the well-studied MEM(TCNQ)2 system,22'27 the authors show that the electron-

phonon coupling constants are fundamental material properties, and not a function of

temperature. However, Graja et al. arrive at a different conclusion, and they report

strongly temperature dependent coupling constants for IH-TCNQ.5 The authors

also suggest that the intensity of the Ag phonon modes should be related to a product

of three functions, which describe changes in the electronic interaction, ground state

occupation, and geometry with temperature, respectively.85 Hence, the temperature

dependence of the coupling constants is still a matter of investigation. These studies

demonstrated the particular utility of infrared techniques, and provided the basis for

the more recent spectroscopic work as a mechanistic probe to follow structural phase


Many of the most interesting spectroscopic studies of quasi-one dimensional or-

ganic materials have concentrated upon structural phase transitions and the associ-

ated metal to insulator transitions.5-12 Detailed examination of the systematic spec-

tral changes in the neighborhood of Tc has proved to be a useful tool in understanding

changes in electron-phonon coupling through the phase transition and the changes in

the gap below Tc. Swietlik and Graja84 report the temperature dependence of the

infrared spectra for the MTPP(TCNQ)2 semiconducting charge transfer salt. Below

Tc, the spectra di(1p1L,: doublets of Ag origin and a strong charge transfer excitation

centered at 2750 cm-1. The reflectance changes discontinuously at Tc. Because

of the first order nature of the phase transition, the temperature dependence of the

reflectance displays a hysterisis. Above Tc, the vibrational features are significantly

broadened and reduced in intensity; the doublet character is still present, but greatly

reduced. The charge transfer excitation is also significantly weaker. Tanner et al.5

presented a particularly interesting study on TTF-TCNQ in which they measured

the far infrared spectra above and below the Peierls transition temperature. Above

Tc, the spectra display metallic behavior, with the ac conductivity extrapolating to a

finite value of the dc conductivity. Below Tc, a semiconducting gap develops and the

spectra displays a sharp pinning mode of the charge density wave at low energy.

Our Materials

NPrQn(TCNQ)2 and DMTM(TCNQ)2 are two "quarter-filled" materials which

have particularly interesting structural phase transitions. The NPrQn salt under-

goes a second-order metal-insulator transition at Tc,13,14 while the dc conductivity of

DMTM salt actually increases sharply below Tc.15'16 In NPrQn(TCNQ)2, the high-

temperature phase is weakly metallic, which is unusual for a 1:2 quarter-filled salt;

the dc conductivity drops sharply at Tc. For DMITNI(TCNQ)2, it is thought that

the unusual transport properties below Tc are due to a crystal field distortion, which

results in the narrowing of the semiconducting band gap.86 The role of disorder is

certainly important in both of these materials, although it is not well understood. Fi-

nally, K- and Rb-TCNQ are two half-filled charge transfer salts in which the TCNQ

stacks go from a uniform to a dimerized molecular spacing at Tc. These sharply first

order structural phase transitions have been characterized as "spin Peierls" in nature.

This study will highlight the important role of the cation on the nature of the phase


The following three sub-sections present a brief background specific to each of the

aforementioned materials. These transport, magnetic and structural studies were the

basis and impetus for our work.


N-propylquinolinium bis-7,7',8,8' ditetracyanoquinodimethane, NPrQn(TCNQ)2,

is a 1:2 organic charge transfer salt containing segregated chains of TCNQ tetramers

and NPrQn counterions. The stoichiometric ratio of the donor to acceptor entities

is such that NPrQn(TCNQ)2 is referred to as a "quarter-fi led" charge transfer salt.

It undergoes a second-order structural metal-semiconductor transition at a critical

temperature (T,) of 220 K.13'14

NPrQn(TCNQ)2 is different from most other quarter-filled 1:2 charge trans-

fer salts, which are generally semiconductors as a result of a structural distortion

that distorts the uniform intermolecular spacing and introduces gaps in the metallic

band structure.19-18 The temperature dependence of the dc conductivity is shown

in Fig. 12.13 The transport properties of NPrQn(TCNQ)2 in the high-temperature

phase have been described as weakly metallic, with a value"1 for the de conductiv-

ity of approximately 1 ohm-lcm-1. Below Tc, the dc conductivity drops sharply

by about four orders of magnitude.13 In the low-temperature phase, conduction in

NPrQn(TCNQ)2 has been shown to proceed by excitation over a temperature de-

pendent gap,14 as in an ordinary semiconductor. An explanation for this unusual

metallic conductivity above Tc was proposed by Jainossy et al.,14 who 'iiu.' -ted that

the Coulomb interaction of the donor cations with the TCNQ chains could perturb

the band structure in the TCNQ stack, resulting in a closed gap and enhanced con-


Structural studies87'88 of this material reveal that the intra-tetramer distances

change significantly between the high temperature and low-temperature phases. At

300 K, intra-tetramer distances of 3.24 A and 3.28 A are reported, while at 100 K,

these distances decrease to a uniform 3.15 A. Thus, at low temperature, the TCNQ

chains become more uniformly tetramerized. In contrast, the inter-tetramer distances

are temperature independent, remaining constant at 3.43 A, resulting in relatively

well isolated tetramers within the TCNQ stack. The Pi space group, the mode of

overlap within the tetramer, and the intramolecular bond lengths and angles remain

unaffected by the phase transition. Based upon the structural data, it has been con-

cluded that the crystal structure becomes more uniform at low-temperature.88 How-

ever, this conclusion is in direct contradiction with low-temperature measurements

of other physical properties,13"14 which indicate a less uniform charge distribution

within the tetramer below Tc.

The phase transition causes important changes in the magnetic state of N-

PrQn(TCNQ)2.1314 Paramagnetic susceptibility (X) measurements show that X is

large and constant in the high-temperature phase, corresponding to 0.8% free spins

per formula unit. Between 200 and 80 K, there is a strong decrease in the suscepti-

bility, which is characteristic of a singlet ground state with a low lying triplet spin

excitation. EPR measurements also confirm the formation of triplet states below 100

K. The energy gap between the singlet and triplet states increases with decreasing

temperature. At low-temperature, X is small; below 80 K a Curie tail appears. On

the basis of these observations, it has been suggested that magnetic interactions may

be the driving force for the structural phase transition.13'14'73'74

The dielectric constant, measured at 9.1 GHz, is constant at approximately 6 in

the high-temperature phase. Between 240 K and 180 K, it gradually decreases to a.

final value of 2 in the low-temperature phase.13


N-N'-dimethyl-thiomorpholinium bis-7,7',8,8',

DMTM(TCNQ)2, is also a "quarter-filled" semiconducting organic charge transfer

salt containing segregated chains of TCNQ dimers and DMTM counterions. This

material undergoes a first order structural phase transition at a critical temperature

(Tc) of 272 K, with the unusual feature that the dc conductivity is three orders of

magnitude higher below the transition temperature.'5 More specifically, adc = 10-2

S-'cm-1 and EA=0.26 eV just above the phase transition; adc = 3 -lcm'-1 and

EA = 0.03 eV just below it.1516 This unusual behavior (shown in Fig. 13) has been

described as an "inverted Peierls transition," and has tentatively been explained in

terms of a crystal field distortion.86

Structural studies86'89"90 reveal that the high-temperature phase is monoclinic.

with the TCNQ molecules stacked in a dimerized fashion along the c direction. The

intra- and inter-dimer distances are 3.25 A and 3.29 A, respectively. The DMTM

cations, which are located on mirror planes, are disordered. The mirror planes result

in the crystallographic equivalence of the two TCNQ sheets (as defined by the a/c

plane) within the unit cell.

In the low-temperature phase, the crystal symmetry is reduced to a triclinic space

group. At Tc, the b axis rotates slightly (about 100 away from the normal) into the

a/c plane. It is likely that ordering of the DMTM counterions results in the loss of the

mirror plane symmetry. The reduced crystal symmetry results in the i -I 1ll, graphic

inequivalence of the two TCNQ sheets within the unit cell. The structure of the two

TCNQ sheets in the unit cell (defined by the a/c plane) is unaffected by the phase


The unusual transport properties of DMTMI(TCNQ)2 have tentatively been ex-

plained in terms of a crystal field distortion.86 The band structure which is thought

to result from this distortion is diagrammed in Fig. 14. The proposed band struc-

ture of quarter-filled DMTM(TCNQ)2 in the high-temperature phase (monoclinic) is

shown on the left. A crystal field distortion along the b axis at the phase transition

temperature reduces the crystal symmetry (to triclinic), causing a splitting of the

bands, as shown on the right. Four other bands occupy a similar configuration at

higher energy, separated from those shown in Fig. 14 by the Hubbard gap.

In terms of transport, the net effect of the distortion is to reduce the energy gap

between the valence and conduction bands. Based upon this argument, one would

predict a large decrease in the activation energy for transport in the low-temperature

phase, thus facilitating thermal excitation of the carriers. Indeed, the activation

energy for conduction (EA) has been measured to be 0.26 eV and 0.03 eV in the

high and low-temperature phases, respectively.15,16 In the high-temperature phase,

the small value of the dc conductivity and the large value of the activation energy

suggests that the dc conductivity is due to hopping.48,49 Despite the "inverted Peierls"

nature of the low-temperature phase, the dc conductivity is relatively low, slg, IIpg li'

both a modest number of carriers and a limited mobility.

Thermopower measurements16 on DMTM(TCNQ)2 also show clear evidence of a,

first order structural phase transition, in agreement with previous results. In the low-

temperature phase, these data are consistent with that expected for a transition to

a small band gap semiconductor with a gap of 0.06 eV. Thus, thermopower provides

further evidence for a reduction of the band gap at Tc, in support of the crystal field

distortion theory by Visser et al.86

Magnetic susceptibility (X) measurements15 display the "inverted Peierls" behav-

ior as well. In the high temperature phase, X rises slightly from its room temperature

value of w 1.0 x 10-3 emu/mol as temperature is lowered from 300 K. At Tc, X drops

sharply from a value of 1.2 x 10-3 to 1.0 x 10-3 emu/mol and then rises with

a further decrease in temperature to a value of 2.2 x 10-3 emu/mol; below 50 K,

X drops sharply, reminiscent of a spin-Peierls transition. From an extrapolation of

these results to zero temperature, it has been suggested that the value of X may be

consistent with two different TCNQ chains in the low-temperature phase, possibly

one with and one without a 2ky distortion.

High-resolution 3"C NM studies91'92 on DMTM(TCNQ)2 show unusually large

chemical shifts of lines assigned to CN groups. Below Tc, these lines split due to the

inequivalence of the two TCNQ molecules in the unit cell. Because of the unusually

large chemical shift of these lines, it is suggested that they are Knight shifts, and thus,

related to the unpaired electron density on the dimer. The Knight shift increases with

decreasing temperature, in accord with the magnetic susceptibility measurements.

EPR studies on single crystal samples93 also confirm the inequivalence of the

TCNQ chains below the phase transition temperature as proposed by the crystal

field distortion model.

K- and Rb-TCNQ

Potassium and rubidium bis-7,7',8,8'-tetracyanoquinodimnethane, K-TCNQ and

Rb-TCNQ, are "half-filled" organic charge transfer salts containing segregated chains

of TCNQ dimers and alkali metal counterions. Both compounds undergo first-order

structural phase transitions at a critical temperature (Tc) of .3'' and 381 K, re-

spectively. Due to the large on-site Coulomb repulsion (U), these 1:1 materials are

commonly referred to as Mott insulators.94'41

In the low-temperature phase (25 OC), K-TCNQ has monoclinic symmetry, with

a space group of P21/n and eight TCNQ molecules per unit cell.95 The TCNQ

molecules are stacked in a dimerized fashion in two mutually orthogonal, incquivalent

columns along the a direction. Average intra- and inter-dimer distances are 3.237 A

and 3.567 A, respectively. Above Tc (140 oC), K-TCNQ is also monoclinic, but with

a space group of P21/c and only two TCNQ molecules per unit cell.95 The TCNQ

molecules are stacked along the a axis with a uniform interplanar spacing of 3.479

A. The two stacks per unit cell remain mutually orthogonal, as before, but are now


In the low-temperature phase (-160 OC), Rb-TCNQ (I) has monoclinic ''li ry,

and a space group of P21/c.96 There are four molecules per unit cell. The TCNQ

molecules are also stacked in a dimerized fashion along the a direction. Intra- and

inter-dimer distances are 3.159 A and 3.484 A, respectively. The dimeric units in

the Rb- compound are more isolated than those in the K- material because of the

unusually short intra-dimer distance as well as the greatly reduced inter-dimer overlap

along a. In addition, the TCNQ molecule itself is not planar, but "boat-shaped." The

crystal structure of Rb-TCNQ has not been reported in the high temperature phase.

Given the similarity of the rubidium system in the low-temperature phase to other

that of alkali-metal TCNQ salts, it is not an unreasonable expectation that the lattice

dimerization may also be either sharply reduced or absent entirely above To.95-98

Despite the similar nature of the crystal structure in these two materials in the

low- (and perhaps, the high-) temperature phases, the structural changes near to Tc

differ substantially. Figure 15 illustrates the temperature dependence of the superlat-

tice x-ray reflection intensity, as reported by Terauchi.17 The superlattice reflection

intensity is directly related to the square of the lattice dimerization.17 For K-TC'N-Q,

the data show a gradual decrease in the lattice dimerization upon approach to Tc, a

discontinuity and weak decrease at Tc, and evidence of a slight residual lattice dimer-

ization in the high-temperature phase. The behavior of the rubidium system presents

a strong contrast. In Rb-TCNQ, the data displays little temperature dependence in

the low-temperature phase, a sharp discontinuity and very abrupt decrease at Tc,

and no evidence of residual lattice dimerization above Tc. These data offer a striking

illustration of the changes wrought by cation substitution on the solid state geometry,

principally the difference in ionic radius, and the resulting behavior of the lattice at

the structural phase transition temperature.

The transport properties of half-filled 1:1 quasi-one-dimensional materials have

been well studied.99,100 K-TCNQ and Rb-TCNQ are both semiconductors above and

below Tc, with conductivities on the order of 10-4-10-5 in the stacking direction. At

Tc, the dc conductivity increases in a discontinuous manner to ;1.4 times its value in

the low-temperature phase. A pressure induced phase transition has been reported

in a variety of alkali-mnetal-TCNQ salts as well.100

Magnetic susceptibility (X) measurements on both K- and Rb-TCNQ display simi-

lar behavior.99'101 The high-temperature phase is paramagnetic, with X having values

- 0.77x 10-4 emu and 1.82x 10-4 emu, for the K- and Rb- compounds, respectively.

At Tc, X exhibits a discontinuity, indicative of a first order phase transition. Below

Tc, X drops sharply (by two orders of magnitude), reminiscent of a spin Pcierls tran-

sition. The low-temperature phase is diamagnetic in both materials. Residual free

spins have been attributed to the existence of spin-solitons.102

Vetger et al.99 also report the heat of transition, as measured by DSC, for several

alkali-metal TCNQ salts. They find AH=60 cal/mol for the potassium cjiiuliiil.

and AH=1010 cal/mol for the rubidium material. The violent nature of the phase

transition in the Rb- compound has been reported by numerous researchers, and is

well understood to reduce the accuracy and reproducibility of measurements above

Tc, leading, in many cases, to the destruction of the crystal.

The microwave dielectric constant of K-TCNQ is temperature independent below

Tc, having a value of el t61.9 We are unaware of any such data for the rubidium


K-TCNQ has been the subject of numerous spectral investigate i'i-. while the Rb-

salt has been much less well-studied. Electronic spectroscopies on the potassium salt

have concentrated on the two oscillator nature of the charge transfer band in the

chain direction, the localized excitation of (7r,t7*) origin, centered at 17000 cm-1, and

the importance of Coulomb interactions.78,13''105'105'106'94'6 Low-temperature mea-

surements clearly resolve the doublet nature of the charge transfer band105'6 splitting

it into a sharp lower-energy excitation at 7500 cm-1 and a weak, broader excitation

near 11000 cm-1. These two absorptions are commonly attributed to intra-dimer

and inter-dimer charge transfer excitations along the chain, respectively, 10 although

various other interpretations have been suggested. It is still a matter of investigation

as to the exact nature of the 10000 cm-1 absorption.6'106'78 The single particle gap.

2A, has been estimated to be 0.7 eV.94 Several groups have also estimated the on-site

Coulomb interaction, U, for the potassium salt as -1 eV, indiiiating the importance

of electron correlations in these materials.94 An estimate of t and U for the Rb-

salt obtains similar values as the K- compound.106 Recently, Okamoto et al.6 have

completed a detailed investigation of the temperature dependence of the U bands

through the phase transition. They observe a systematic red-shifting and merging

of the two charge transfer excitations with increasing temperature. This is in good

agreement with the earlier but less detailed studies by Yakushi et al.105 The authors6

report that the optical spectra of the Rb- salt is analogous to that of the K- salt in

the low-temperature phase. However, above Tc, the results are quite different; in the

Rb- compound, the inter-dimer charge transfer excitation is completely disappeared.6

The vibrational spectra of K-TCNQ has also been extensively investigated.107-78

In these studies, the authors note the unusually strong intensity and polarization

dependence of the Ag vibrational modes. Various models of the interaction between

the Ag vibrational modes and the charge transfer of charge density wave have been

used to understand the electron-phonon coupling in the potassium salt. These range

from the application of the relatively simple isolated dimer model78 to complex charge

density wave formalisms to explain this coupling.20 Due to its relative simplicity,

the 300 K spectrum of K-TCNQ has been used extensively as a model compound

to test these the applicability of many methods. We are presently unaware of any

temperature dependent infrared spectra on single crystal samples. In addition, no far

infrared data is available for the K- compound. A room temperature middle infrared

measurement of the Rb-TCNQ salt in mull form has been reported in transmission

at 300 K.111 However, we are also unaware of any complete infrared measurements

on Rb-TCNQ single crystal samples.

A few studies have compared the behavior of the electron-phonon coupling in-

duced modes in these two salts. Bozio and Pecile11o show that, for powder samples of

a variety of 1:1 charge transfer salts (including both K- and Rb-TCNQ) the intensity

of the Ag modes decreases with increasing temperature in the low-temperature phase.

More recently, Okamoto et al.6 have reported the temperature and polarization de-

pendence of the CN stretching mode on single crystals of both materials.

Raman spectroscopy has been used extensively to study the neutral TCNQ molec-

ular crystal as well as the K- and Rb-TCNQ charge transfer salts. Because of the

different selection rule, it has been used to identify the unperturbed frequencies of

the Ag vibrational modes in both the K- and Rb- compounds. 12-114 The Ag phonon

modes in the potassium salt have been reported to change substantially through the

phase transition, increasing in intensity upon approach to Tc and reducing again above

it.114 Raman data for Rb-TCNQ is only available only for pressed salt-KBr pellets

and for single crystal samples above 350 cm-in the low-temperature phase.112111

However, it is reasonable to expect that the symmetry assignments of the various low

energy modes in Rb-TCNQ may be similar to other 1:1 alkali TCNQ salts.

Various lattice and libron modes have been identified in the low frequency

studies as well.21'113,114 For the TCNQ molecular crystal, Girlando et. al.21 report

lattice modes at 41, 76, 97, 105, 133, and 144 cm-1. The lattice mode frequencies in

the dueterated crystal are almost identical. For the K- salt, Truong et a1113 report

lattice mode frequencies of 70, 81, 110, 170, 229, and 295 cm-', with the lowest

three assigned as librations. The scattering intensity of several of the lattice modes

in the K- salt has been shown to decrease with increasing temperature and broaden

significantly above C.-113,114 The frequencies and,.I- ry designations of these

modes will be discussed in detail in Chapter 6.

Work has also focused upon the nature of the various structural phase transi-

tions in Mott insulators. The alkali-metal TCNQ charge transfer salts are gerin-l ally

considered to be spin Peierls systems,115'116'117 with a 2kF structural distortion at

T,. Experimental evidence for this conclusion derives primarily from the temperature

dependence of the magnetic susceptibility, where the sharp decrease of X at Tc is evi-

dence for the opening of a spin-wave gap.99,101 The diamagnetic phase is accompanied

by the formation of spin-soliton defects.102

I i i'i iirits: ie t Mlati.i itl

Heterogeneous (III-V and II-VI) inorganic semiconductors have historically re-

ceived the majority of attention in the field of luminescent materials technology. Some

of the materials currently under intense study for possible use in opto-electronic ap-

plications include GaAs, ZnS, and ZnSe.118-124 However, the widespread use of these

materials has been hampered by fabrication problems and low EL efficiency. Re-

searchers have implanted dopants in these semiconductors in an effort to circumvent

these shortcomings.121-120 Porous silicon is another inorganic material which has also

generated a great deal of excitement recently. It exhibits visible red-orange photo-

luminescence, and research is currently in progress to demonstrate reliable electro-

luminescence and develop high efficiency devices from this technology.125

Organic materials have also been successfully fabricated into light emitting diodes.

For example, anthracene molecule is a blue emitter, although the efficiency is not

very high.126 Additionally, devices fabricated from low molecular weight organic

materials have had stability problems.127 A more promising organic material for use

in opto-electronic applications is buckmisterfullerine (C60). Recently, a broad near-

infrared and visible EL peak was reported in an undoped C60 diode device.128 Copper

phthalocyanine based diode devices are also under study.129

Photo-luminescence of conjugated polymeric materials has been a well known

phenomena for several years. There are numerous studies in the literature

evaluating the optical absorption and steady state photo-emission properties,

the Stokes shift, the photoinduced absorption effect, as well as time resolved

measurements.60,130,62,131'132,133 The studies on poly-phenylenevinylene (PPV) and

related materials by the Cambridge group have been the most extensive.60,134,135 The

nature of the various excitons and the mechanism of radiative decay (as well as the

nature of the competing non-radiative processes) dominate much of their work. Their

analysis of the photo-excitation and recombination process is as follows. After photo-

generation of polaron lattice defects, the quasi-one-dimensional nature of the lattice

acts to local the charges in polaron-exciton or (neutral bipolaron exciton) defects.

Photo-induced absorption experiments have provided the most important substan-

tiating evidence for this proposal.60,134'131,136 Further evidence for the formation of

such defects also comes from the absence of ESR and photoinduced ESR signal, as

these singlet excitons have no net charge.60'130 Subsequent recombination of the po-

laron exciton leads to light emission. Evidence for the neutral bipolaron exciton ihas

been observed in other materials as well.130'137'138

The use of conjugated polymers as a diode material is a relatively recent devel-

opment. In fact, the first reports of electro-luminescence in pi' 1i i I i appeared in

late 1990! Undoped polymers are ideal materials for diode applications because of

the large, tunable (7r,7r*) band gap139'140'141 and the quasi-one-dimensionality of the

chain. Polymeric materials offer significant advantages over the aforementioned con-

ventional semiconductors, the most important being the ease of processability. This

is partially due to the fact that good luminescence properties can be obtained from

oligomeric chains.

In the next section, I will present a review of the field of electro-luminescent

polymers. In the following section, I will briefly discuss experimental results which

are directly relevant to the materials used in this study.

The Field of Elctlco-llnmine-.cInt Pol, ymers

Burroughes et al.3 were the first to demonstrate yellowish light emission (at

600 nm) from poly(p-phenylene vinylene) (PPV) diodes in late 1990. The authors

reported the current-voltage characteristics, and showed that there is a. 14 V threshold

for substantial charge injection in this material. They also demonstrated that the

integrated EL iiitc-risity was approximately linear with current. Device efficiencies on

the order of 0.05% were reported.

The results of Burroughes et al. were confirmed by Braun and Heeger32 for both

PPV and a chemically modified PPV system. Here, the authors suggest a plausible

mechanism for light emission in these materials, in which the electrons tunnel from the

rectifying metal contact into the upper polaron state in the gap. Structural relaxation

to the excited state of the polaron exciton (neutral bipolaron exciton) proceeds by a

non-radiative process, after which, radiative decay occurs in the characteristic manner

described previously by the Cambridge group.60 It is not known if the tunneling

mechanism is a general phenomena.

Ohmori et al.34 have studied the effects of the length of an alkyl side chain on

the electro-luminescence of poly(3-alkylthiophene) diodes. They report red-orange

emission at room temperature, and increased emission intensity from the diodes with

the longer alkyl side chains.

Recently, Grem et al.33 have reported the photo- and electro-luminescence of

poly-(para-phenylene) (PPP) made by a precursor method. PPP is a wide band

gap semiconductor, and due to the non-degenerate of the ground state, supports

the formation of polaron charge carriers.137,138 Grem et al. report a Stokes shift of

about 1.1 eV between the absorption maximum and the photo-emission maximum

in PPP. They also observe greenish-blue emission properties from the diode devices;

efficiencies between 0.01 and 0.05 % are reported.

Finally, copolymers have also been successfully incorporated into diode devices.141

This process can potentially allow device patterning, and hence, frequency tuning of

the emitted light at various points on the diode. These devices display improved

emission efficiencies, as well.

On the more practical side, there have been several reports on the impor-

tance of hole and electron transport layers in fabricating devices with improved EL

performance.142,143,144'34'145 The transport layers are thought to facilitate control of

the recombination process, resulting in increased EL efficiencies.142'143 In recent ex-

periments, the chemical identity of the metal contact has been shown to significantly

influence the intensity of diode light emission.32 This is most likely due to the combi-

nation of work function and Fermi level matching considerations which act to reduce

the width of the Schottky barrier.32 For example, it is thought that the calcium elec-

trode facilitates charge injection and reduces heating near the anode. Indeed, Braun

and Heeger32 report improved light emission properties for a calcium anodic contact

rather than an indium one. In addition, researchers have recently reported the fabri-

cation of an "all-polymer" light emitting diode."46 Such a flexible, large area diode is

expected to find many uses in the electronics industry. Finally, polymers with very

high luminescent efficiencies are also finding applications as laser dye materials.147

Our Materials-Previous Work

The chemical structure of the polymers used in this study are diagramed in

Fig. 2, Fig. 3 and Fig. 4. The structures displayed in Fig. 2 are based upon nmcia-

linked phenylene, the structure in Fig. 3 is poly(2,1':4',1":4",1"'-quarterphenylene-

1,4) (PQP), while the structures shown in Fig. 4 are based upon 1,4-bis(2-

thienyl)phenylene (PBTP) with symmetric side groups.

Phenylene and thiophene groups (as well as unsaturated double bonds) are well

known components of other electro-luminescent polymers, such as PPV, PPP and

P3AT. This is because the conjugated character of the ring can stabilize excess photo-

or chemically induced charge by forming a quinoid based resonance structure.64 De-

pending on the spin and charge, these defects can be polaronic or bipolaronic in

nature.64 Due to their non-degenerate ground state, the phenylene and PBTP based

materials are expected to form polarons upon photo-injection. Due to the quasi-

one-dimensionality of the chain and strong electron-phonon coupling effects, these

polarons are thought to combine to form the polaron exciton (or neutral bipolaron).

The effect of this neutral charge defect on the chemical structures of PMP and PQP

is shown in Fig. 16.

It is anticipated that the phenylene components of the materials shown in Fig. 2

and Fig. 3 will function as active emitter centers. By linking the various emitter

groups in a meta fashion (Fig. 2), each group is electronically isolated from one

another, but can individually support a charge defect of the type discussed above. In

the case of the the quarterphenylene (PQP) polymer (lower structure of Fig. 16), steric

hindrances act to twist the main-chain out of its planar configuration. This destroys

the long range conjugation which is inherent in the unsubstituted PPP system, thus

isolating the charge defects on the pendant side chains. Although these materials can

support the formation of a neutral bipolaron exciton, it can not travel along the chain

due to the aforementioned broken conjugation.

This idea of a discrete EL emitter center is the most important fundamental idea

that we investigate in this work. In a discrete emitter unit, the charge defect as well

as the lattice distortion are localized on one portion of the chain. All of the 1ph i',. ]i'w

based polymers in Fig. 2 and Fig. 3 may be thought of as containing a series of discrete

emitting units, without any long-range conjugation. This set of materials allows us

to study the influence of emitter unit structure on the electronic properties.

Solution phase absorption and photo-luminescence data of the lII, I.-1.,I. based

materials (shown in Fig. 2) has been reported by Ruiz and Reynolds.148 For poly-m-

phenylene, they find the absorbance maximum and the semiconducting energy gap

to be 260 and 350 nm, respectively. These absorbance features are a relatively weak

function of the emitter unit conjugation length of the material. For example, upon

increasing the number of ,ij .iit'l1 units from two to five, a 20 nm red shift of the

absorbance maxima was observed. The photo-emission data display a strong, narrow

emission spectra in the blue frequency regime. With increasing conjugation length

of the discrete emitter center, the photo-luminescence maximum is red-shifted by

approximately 40 nm. For the PQP, the optical absorption maximum is observed at

300 nm, while the photo-luminescence maximum is found at 401 nm.149

Our measurements on the PMP and PQP materials can be also be compared

with similar measurements by Grem et al.33 and ctlii- -',1' on PPP. A detailed

discussion of this data will be presented in Chapter 7. At least one result may be

anticipated in advance. It is well known that upon reduction of the conjugation

and bond length alternation, the gap, 2A, decreases. Thus, it is expected that the

semiconducting gap of PMP will be larger than that of PPP. Solution measurements

of these quantities in PMP bear out this supposition.148

Promising results have already been obtained on the poly-3-alkylthiophene sys-

tem, and thus, this material has enjoyed significant attention recently. The optical

absorption has been well-studied,s50 and the poly-thiophene and alkyl-substituted

poly-thiophenes have been shown to support polaronic structure.' 51'136'13 The ther-

mochromic properties of this material have also engendered special interest.150 In ad-

edition, visible light emission has recently been observed from thiophene based diode

devices.34 It is a natural extension of this work to combine the thiophene emitter

unit with a phenylene emitter unit, as we have begun to do here.

The (2-thienyl)phenylene (PBTP) materials are another class of potential light

emitting polymers. The side chains (shown in Fig. 4) have been carefully selected to

alter the band gap of the main-chain polymer, due to their electron-withdrawing or

electron-donating behavior. The long side chains are also expected to result in steric

hindrances, which may twist the main-chain backbone from a more planar configura-

tion, thus isolating the potential emitter units. As in the plhelny.lel:, based polymers,

these materials are expected to support a bipolaron defect, but their mobility min,:

be limited due to out of plane twisting. Hence, the PBTP system is another material

in which we believe that the polaron exciton will be localized in a discrete unit.

Recently, the PBTP system (substituted with a wide variety of electron-donating

and withdrawing substituents) has been the subject of an extensive investigation.152

Absorption spectra identified these materials as wide band gap semiconductors. Upon

oxidative doping, the spectra of the PBTP compound with dudc~kylox'y substituents

display the characteristic inter-gap features indicative of polaron, and finally bipo-

laron, formation. With the hexyl side group, the PBTP materials do not dope. This

is because local steric considerations hinder the formation of a planar main-chain

geometry, which is necessary for charge defect stabilization. In addition, symmetri-

cally substituted PBTP materials, such as those in Fig. 4 were observed to exhibit

significant long range ordering, as evidenced by X-ray and DSC.

Together, the various polymers allow us the opportunity to systematically assess

the idea of a discrete emitter center, as well as to ascertain the effect of variations in

the chemical structure upon the light emitting properties of two potentially interesting

families of materials.




2 3

4 5
1 03/T (K-1

Fig. 12. Log of dc conductivity vs inverse temperature along the TCNQ chain
direction for NPrQn(TCNQ)2.













Temperature (K)

Fig. 13. Log of DC conductivity vs temperature along the c* direction for

III I l Il i lI i


_ L

I -



Fig. 14. Band structure modifications of the proposed crystal field distortion, which
is thought to occur in the b crystallographic direction.




40 80




Fig. 15. Temperature dependence of the superlattice X-ray intensity.7 Solid line:
K-TCNQ; Dashed line: Rb-TCNQ.

Fig. 16. Neutral bipolaron charge defects. Upper panel: PMP, lower panel: PQP.


This chapter will detail the experimental methods and apparati used in this work.

It is divided into two sections. The first section presents a general discussion of

the spectrometers, the temperature and polarization control, the sample mounting

procedures and the data analysis techniques used in the investigations of the organic

charge transfer salts. The second section describes, more briefly, the spectrometers,

the electro-luminescence circuit and apparatus, and the device preparation techniques

used in the investigations of the luminescent polymers.

Fourier Tr;in ftrr In tnfrni -d Spectroscop,


Fourier transform spectroscopy has several advantages over conventional grating-

based spectroscopy.153 These include: the multiplex advantage, the stability, the

ability to easily average multiple scans, as well as the increased speed at which data

can be obtained. These advantages result in an increased signal to noise ratio as

compared to a conventional grating machine.

A simple, two-armed interferometer (Fig. 17) will be used to discuss the general

principles of interferometry. In this diagram, the wave traveling from the light :-i "..i .

is split into two components at the central beamsplitter. When the reflected beams

are recombined, they have a phase difference, 0 = 27rwx, due to the path difference

of the moving mirror. Thus, the signal (or intft :iogiram) is eventually detected as a

function of this optical path difference. The interferogram is written as

I(x) = j S(w)exp-i2rwz dw + o, (43)

where x is the path difference, S(w) is the spectral distribution function, and coJ is the

intensity at infinite path difference (the average value of the interferogra.m). In the

ideal case, I(z) is symmetric about x = 0, x runs from -oo +oo, and the sampling

is carried out continuously as a function of path difference. Thus, the spectrum may

be obtained directly by an inverse Fourier transform of the collected interferogram:

(w) 00(x) ex x dx

The advent of fast computers and the development of the fast Fourier transform

algorithm has made such a. computation routine.

In practice, several approximations and assumptions are essential to the act of

measuring the interferogram as well as to the computation of S(w) using Eqn. (44).

The most common limitations to be addressed in measuring the interferogram include:

1. the finite path of the movable mirror.

2. accurate determination of the zero path difference position of the movable mirror.

3. the discrete nature of I(x) due to non-continuous sampling.

These limitations cause various aberations in I(z), which result in a distortion

of the true spectrum. The nature of these non-idealities are well documented.153-156

Each of these limitations and their common corrections will be discussed briefly below.

The finite path difference, A, of the movable mirror is probably the single most

important limitation in Fourier transform spectroscopy, as it results in a truncation

of the complete interferogram at x = A, rather than x = o00. This truncation

introduces large side lobes (or "feet") into S(w). Apodization of the interferogram

removes these lobes.153 At the same time, convoluting an apodization function with

the interferogram reduces the spectral resolution. A wide variety of apodization

functions are commonly used.153 The choice of apodization function is critical-

not only for the removal of the lobes but for the degree of broadening and spectral

distortion which is incurred in the procedure. Thus, we see that the resolution of a

Fourier transform spectrometer is limited in part by the maximum path difference.

The accurate determination of the zero path difference position of the movable

mirror is critical to the collection of the interferogram, I(x). An accurate determi-

nation of the x = 0 position is essential to properly locate I(x) on the x axis, thus

preserving the symmetry of the interferogram. The location of the zero path differ-

ence of the movable mirror is also used for other purposes. For example, the x = 0

position is needed to center the truncation and apodization functions. Improper posi-

tioning of these functions with respect to the x = 0 position will result in asymmetry

of the interferogram.

Phase errors result from the asymmetry of the interferogram, when I(x)#I(-x).

In terms of the spectral distribution function, this translates into S(w) having an

imaginary part, which is physically unreasonable. Commonly, phase error is the result

of an inaccurate determination of the zero path difference of the movable mirror, but

it can also result from misalignment of the spectrometer. A method for the correction

of phase errors has been discussed in detail elsewhere.24

Eqn. (44) also assumes that the user has collected an interferogram which is a

continuous function of path difference, x. In practice, the interferogram is sampled

at small intervals along the positive path difference z = kb, where k either zero or an

integer. The Nyquist sampling theorem puts an upper limit on the frequency spectrum

as wmax= ,. This limit defines the largest sampling interval, 6, that can be used to

prevent aliasing in the frequency spectrum. The finite sampling consideration also

introduces a high frequency limit to the spectrum. Electronic filtering of the signal

removes frequencies which are higher than the maximum cut-off frequency.

Bruker 113V

A Bruker 113V Fourier transform infrared spectrometer (FTIR) was used to mea-

sure the polarized reflectance of the various samples in the 30-4000 cm-1 frequency

regime. Figure 18 displays a ray and block diagram of this instrument. For discussion

purposes, we will divide the spectrometer into four chambers: the source housing, the

optical bench, the sample area and, finally, the detection area.

Our FTIR is equipped with two sources: a mercury arc lamp and a, globa.r, for

use in the far and mid-infrared, respectively. An ;ljlltl,,1 aperture allows the light

into the optical bench.

The main components of the interferometer are a two-sided, moveable mirror,

the various beam splitters, and two independent interferometers. The beam splitters

are mounted on a rotating wheel, allowing beam splitter changes without a break

in the vacuum. The various beam splitter materials, thicknesses, and efficiencies for

each spectral region, as well as other operating parameters, are shown in Table 1.

As in Fig. 17, the divided beam returns to the beam splitter with a known path

difference caused by the moving mirror. Two smaller interferometers occupy less

visible positions on the optical bench. A H,'Nr- laser reference interferometer monitors

the path difference of the movable mirror. A white light source is used to determine

the zero path difference position of the moveable mirror. Optical filters can also be

rotated into the beam path, depending on the frequency region. For example, a black

polyethylene filter is used to block the near-infrared and ultra-violet radiation from

the mercury arc source, whereas a wire "mesh" can be used to partially block the

beam, thereby reducing its intensity. Concave spherical mirrors focus light onto the

beamsplitter and onto the adjustable exit aperture.

There are two sample chambers in our FTIR. For transmission measurements,

the reference and sample are placed directly at the focal point. For reflectance mea-

surements, we use the reflectance stage, diagramed in Fig. 19. Alignment of the stage

makes reflectance measurements more difficult. A series of concave and flat mirrors

direct the light through the polarizer and into the detection chamber.

In the far infrared, a liquid helium cooled Si bolometer (Infrared Laboratories) was

used as the detector. In the mid-infrared, a room temperature pyroelectric deuterated

triglycine sulfate (DTGS) detector or helium cooled (4.2 K) Si:B photocell, also from

Infrared Laboratories, was used to collect the signal. The helium cooled detectors

were used for most work due to their high sensitivity. A diagram of the helium cooled

detector is shown in Fig. 20.

Because the two-sided mirror moves at a constant velocity, v, the optical path

difference may be re-written as x = 4vt, where t is the time that the mirror has

traveled from the zero path difference position. In this way, the infrared frequency

radiation is converted to a radio frequency signal. The aforementioned detectors

actually produce an audio (rather than infrared) frequency signal.

After amplification, electronic filtering and digitization, the signal is sent to an

Aspect computer. Here various apodization and phase corrections are applied, and

the interferogram is Fourier transformed into the single beam spectrum.

The spectra are taken under vacuum. Sufficient time is allowed for thermal equi-

librium of the detector, beamsplitters and other optical bench components before

commencing measurements.

Optical lMeasurements: the Perkin-Elmer Grating Spectrometer

A modified Perkin-Elmer 16U grating spectrometer was used to obtain spectra

from the mid-infrared through the ultra-violet (800-30000 cm-1). Unlike FTIR, each

frequency is sampled sequentially; signal averaging is done by taking multiple readings

at each frequency.

Figure 21 displays a ray and block diagram of the instrument. The operation

and calibration of this instrument has been described in detail elsewhere,157 so here.

I will only briefly outline the main components and operating principles for this


Three different light sources are positioned in a water cooled housing. These are

the globar, quartz tungsten, and dueterium arc lamp, respectively. Upon leaving

the source housing, the beam passes through a chopper and a series of filters. The

chopper creates a square wave signal, which is cIailv amplified by a lock-in amplifier.

The long-pass and band pass filter eliminate higher (unwanted) orders of diffraction.

Next, the beam is directed through the monochromator entrance slit towards the

grating. Diffraction of the light at the grating is the most critical element of this

process, as the grating angle and grating constant uniquely determine the frequency of

light which is allowed to pass out of the exit slit and into the rest of the spectrometer.

The relationship between grating angle and the frequency of diffracted light can

be derived as follows.158 The general equation relating the angle of incidence and

diffraction to the wavelength of diffracted light is

mA = d(sin i + sin D), (45)

where m is the spectral order, A is the wavelength, d is the spacing between grating

lines, and i and D are the angles of incident and diffracted light, respectively. As

shown in Fig. 22, both angles, i and D, are measured with respect to grating surface

normal; the grating angle, 0, is measured from the zero position. Writing i and D in

terms of 0, we obtain

S & 6
mA = d(sin(O + ) + sin(q )) = 2d sin cos (46)

where 6 is the angular separation of the entrance and exit slits. Rearranging, setting

m = 1 and substituting for wavenumbers in terms of frequency, we obtain

2wd cos = csc q. (46)

For our monochromator, b=40, so cos 1. Thus, the angle of the grating is then

related to the energy of the diffracted light (in wavenumbers) as

2wd = csc (47)

The quantity
-1 (48)
2d cos

is usually defined as the grating constant.

During normal operation, a stepping motor moves the grating (at predetermined

intervals consistent with the necessary spectral resolution) from one position to the

next. The angle of rotation together with the exit slit-width determine the resolution

of the monochromator.158 For our imiclhiiii, 0.1%.157

After exiting the monochromator, the beam is focused onto position ., using

a large spherical mirror. A second large spherical mirror focuses the light onto the

detector. Three detectors were used in these measurements: a thermocouple with

a KBr window, a room temperature lead sulfide detector, and a Si photocell. The

chopped signal is amplified by an Ithaco Model 393 lock-in amplifier, and then sent

to a Fluke Model 8520A digital voltmeter. The signal average processes occur at

the Fluke. The various operating parameters of the Perkin-Elmer are summarized in

Table 2.

Note that the Perkin-Elmer spectrometer can be used in reflection or transmission

mode. In reflection mode, the reference mirror and sample are both positioned at

position A. A single beam spectra is obtained for the aluminum reference mirror;

then the sample is rotated into position, and the single beam spectra of the sample

is recorded. The ratio of these two single beam spectra define the sample reflectance.

It is common practice to correct for sample roughness and the reflectivity of the Al

reference mirror by coating the sample with 2000 A of Al, remeasuring the single

beam spectra, and computing the following ratio:

R(w) = ( s mp ) RAt, (49)

where Smirror and Ssanl, are the single beam spectra of the mirror and sample, re-

spectively, and RAI is the absolute reflectance of aluminum. In transmission mode,

an aluminum reference mirror is placed at position 2, and the reference and sample

are placed at the focal point in position 9.

The entire spectrometer can be evacuated to pressures below 150 millitorr. This is

especially important in the mid- and near-infrared frequency regime, where absorption

of water or CO2 is readily evident in the spectra.

Tenmp rat rn tie anild Polriza tion Coniti l


The copper sample holders, upon which the crystals are mounted, consist of a

small hole surrounded by a conical face; these sample holders were mounted on an

adjustable copper frame. The frame was attached to the cryostat tip.

A Hansen and Associates High-Tran cryostat, transfer line and Model 8000 tem-

perature controller were used for the non-ambient temperature experiments. The

temperature sensor itself is a calibrated Si diode, which is embedded in the cryostat

tip. The cryostat also contains a heating element. The cryostat tip is surrounded by a

vacuum shroud, which is evacuated with a diffusion pump to prevent ice from forming

on the sample. The shroud has either a polyethylene or KCI window to transmit far

or mid-infrared light, respectively. A transfer line is used to move the liquid helium

from the storage dewar through the cryostat. The helium gas is recovered for re-use.

Flow meters control the flow of helium through the system.

To correct for the effect of the shroud window and any slight alignment differences

caused by insertion of the transfer line, we normalize our temperature dependent

spectra with respect to the room temperature spectrum taken without a window and

corrected for the finite reflectance of the aluminum reference mirror, as previously

discussed. This renormalization is given as:

R(w, T, win)
R(w, T) = R(w,Twin) R(w, 300). (50)
R(w,300, win)

The polarizers used in the Bruker FTIR are wire grid polaii.,,i/ mounted in

oriented polyethylene and AgBr for the far and mid-infrared, respectively. Polarizers

rotate in or out of the beam path, and the orientation is easily changed u-iilg a gear


The polarizers used in the Perkin-Elmer spectrometer are wire grid, dichroic, and

plastic polarizers, for the infrared, near infrared and visible frequency regions, respec-

tively. These polarizers may be moved in and out of the beam path using external

(to the vacuum tank) spring-loaded cables. The position of the polarizer monitored

using a potentiometer readout. The orientation of the polarizer is determined from a

calibration of the potentiometer with respect to a polarizer of known orientation.

Measurements on the Charge Transfer Salts

Near normal polarized reflectance measurements were made on the large face

of single crystal samples of NPrQn(TCNQ)2, K- and Rb-TCNQ. Experiments were

performed in two polarizations on the large crystal faces: with light polarized parallel

and perpendicular to the TCNQ chain axis.

DMITM(TCNQ)2 represents a special case, as several of the i i. -. 1l facets were

large enough to permit investigation. Figure 23 depicts the typical crystal shape and

labels the cr.-t;, llographic faces of DMTM(TCNQ)2. Note that the a/c plane defines

the (010) face of the sample. Here, we report our measurements on t e (010) and (110)

faces. Experiments were performed with light polarized parallel and perpendicular to

the maximum reflectance on each unique crystal face. Symmetry requires that two of

the principal axes of the high temperature phase lie on the (010) crystal face. Once

the optic axes were found at 300 K, the polarizer position remained unchanged for

all temperatures studied.

For each polarization, infrared measurements (30-4000 cm-1) were made at sev-

eral temperatures ranging from 10 K to 420 K, concentrated around Tc. For the

quarter-filled salts, data in the near-infrared region (4000-15000 cmn-), were col-

lected at 100 K and 300 K. Reflectance measurements for the visible frequency

region (15000-30000 cm-1) were obtained at room temperature. Thus, the total

frequency range of data collection ran from 40 to 30000 cm-1. For the K- and

Rb-TCNQ samples, complementary data for the near-infrared and visible frequency

region (3000-20000 cm-'), were obtained from various literature sources, and used

as a high frequency extrapolation in this study. Thus, the total frequency range of

reflectance data available ran from 30 to 20000 cm-" for the half-filled charge transfer


Sample Prep, rIat ion and Mountitni-Org taic Cl a ri'" TrAin-f i ;ilt.-

The NPrQn(TCNQ)2 single crystal samples were prepared by the method of

Melby et al.2 by Dr. Katalin Kamaras at the Central Institute for Research in Bu-

dapest, Hungary. These crycfIala were precipitated from an acetonitrile solution by

slow cooling. Typical sample dimensions were w 4 x .5 x .5 mm .

The DMTM(TCNQ)2 single crystals were prepared by Dr. Manni.l Almeida at

the National Laboratory of Engineering and Industrial Technology, Portugal. In this

reaction, TCNQ was reacted with the cation iodide in boiling acetonitrile under a

nitrogen atmosphere.2 Large single crystals, up to 5 x 5 x 1.5 mm3 were grown by

slow cooling of deaerated saturated and seeded solutions under dry nitrogen, using a

technique similar to that previously described for other complex TCNQ salts.159'160

Under a microscope, both the NPrQn(TCNQ)2 and DMTM(TCNQ)2 samples

were smooth, so no attempt was made to coat the surface with aluminum to correct

for scattering loss. Such losses are expected to be more important at high energy.

Needlelike single cri-tal- of K-TCNQ and Rb-TCNQ(I) were prepared by the

reaction of TCNQ with the cation iodide in acetonitrile by Dr. Yoshihiro Iwasa at

the University of Tokyo, Japan.2 Typical sample dimensions were 8 x .4 x .4 mm

for the potassium salt, and a 7 x 1 x 1 mm for the rubidium salt.

Because of the small sample size, two K-TCNQ crystals were mounted adjacently

on the sample holder to form a mosaic, thus increasing the effective sample size. The

absolute level of the reflectance was obtained by coating the surface of the crystals

with aluminum to correct for scattering loss and the area between crystals of the


Due to the violent first-order nature of the structural phase transition, a different

approach was taken with the Rb- compound. In this case, a single Rb-TCNQ crystal

was mounted on two thin copper wires, which were secured to the copper sample

holder. The flexibility of the copper wires allowed for the thermal expansion of

the crystal, thus preventing complete breakage at Tc. It is important to note that

the sample did break and fissure at Tc due to internal (and external) stresses. The

absolute level of the reflectance for these spectra was determined as follows. First, the

absolute level of the reflectance was determined at room temperature with the sample

the same size as the reference mirror. Here, the Rb-TCNQ sample was smooth, so no

aluminum correction was used. In the low temperature phase, spectra were ratioed

to the 300 K spectrum in the customary manner. Due to the surface cracking and

sample breakage, we do not know the true absolute level of reflectance in the high

temperature phase. It would have been highly desirable to coat the crystal with

Al, and thus, renormalize for these effects. Unfortunately, due to the delicacy of

the crystal, this was impossible. Making an educated guess, we renormalized these

spectra above Tc to be consistent with the level and trend below Tc.

Analysis of Reflectance Spectra

The Kramers-Kronig Tr;,ii-fr n,1

Optical constants presented in this dissertation were obtained by Kramers-Kronig

analysis of the power reflectance spectrum.161'162 This analysis gives the complex

dielectric function

e(Wo) = eI(W) + ie2(w) = ei(W) + --o-(CW), (51)

where eq is the real part of the dielectric function and oa is the frequency depen-

dent conductivity. These properties are fundamental material characteristics, as they

effectively define its optical response to light.

The Kramers-Kronig relations are based upon the requirement of causality and

the fact that the real and imaginary parts of a response function are alw
through a dispersion relation.161'162 For reflectance measurements, the response func-

tion is
r(w) = = p(w) exp iO(w). (52)

The measured reflectance is

R(w) -re 2 = 2(w), (53)

and the phase shift upon reflection is calculated from

( w In R(w') In R(w) (54)
O(w) = d_ (54)
71r 0 W W12

Although our data was collected over a wide frequency range, extrapolation be-

yond the frequency interval of interest is necessary, as the integral in Eqn. (54) extends

from zero to infinity. The details of these extrapolations and the effect upon the re-

sulting optical constants are the subject of the next brief sections.

(n 1)+i
r () = -- ) + i (5 ,)
(n+1) + i

one can obtain the real and imaginary parts of the complex refractive index. The

refractive index, n, is given as

n(w) = -R( (56)
1 + R(w) 2 V/R(w)cos O(w)

and the extinction coefficient, K, is given as

2 R(w)- 2sinc (w)
1 + R(w) 1^IR(w) cos O(w)


These relationships can be re-cast in terms of the dielectric function, C(w). Since

N=VE for a non-magnetic material,

c = N2 = (n + i)2.


The real part and imaginary parts of e(w) are

](W) = 72 K,~

62(w) = 2nK.



The real part of the dielectric function describes the dispersion of the carriers, while

the imaginary part of c(w) describes the attenuation of the carriers.

The absorption coefficient, c, can be obtained as

a(w) 47raw.


Thus, by Kramers-Kronig analysis of reflectance data, we can obtain the optical

constants of a material, which effectively characterize its response to light.

Method of Extrapolation

The high frequency extrapolation is actually comprised of two distinctly differ-

ent regimes: the interband region, where there is some (unknown) contribution to

the overall oscillator strength from interband transitions, and the higher energy free

electron region. In the interband region, from 30000-1.0x106 cm-1, the reflectance

is modeled as

R(w) = Rf(-), (62)

where Rf and wf are the reflectance and frequency of the last measured data point,

and s is an adjustable parameter typically taking on a value between zero and four.

Above 1.0x106 cm-1, free electron behavior, given by

R() = Rfp(l) ', (63)

is assumed.161'162 Hence, the high frequency extrapolation of the Kramers Kronig

analysis has two adjustable parameters: s, the interband exponent, and w/j,, the

free electron frequency. A reasonable choice of these two parameters is critical to the

success of this method. The high frequency extrapolation naturally affects the optical

constants in the visible and near-ultra violet the most.

The method of extrapolation at low energy in Eqn. (54) must also be addressed.

For semiconducting materials, it is common to approximate R(w) as a. constant to

zero frequency,161,162 as in such samples, there are no low lying excitations with large

oscillator strength.

Luminescence Measurements

Sample and Device Preparation

The molecular structure of the polymers used in this study are di-Il.', '1 in Fig. 2

and Fig. 4. The synthesis of these materials was done by Dr. Jose Ruiz and .J. \\'n ,i.

under the direction of Profs. J. Reynolds and M. Pomeranz, at the University of

Arlington, TX. The meta-linked phenylene based oligomers had molecular weights of

6600-2000 g/mole were obtained, with polydispersity ratios on the order of 1.2-1.7.148

In addition, they are stable in air, and remain so at elevated temperatures.1'18 PBTP

based samples were synthesized via oxidative polymerization, and yielded oligomers

with an average degree of polymerization of about six repeat units. The polydis-

persity ratios were 1.69 and 2.46 for the symetrically substituted dodecyloxy- and

hexyl- PBTP samples, respectively. These materials were also stable at elevated

temperatures.152 Finally, the PQP sample was obtained as an oligomeric material,

and treated to remove the Cl end-groups. The details of the various synthetic meth-

ods are reported elsewhere.148152'149 These materials have an advantage over other

previously reported potential EL polymers in that they are directly soluble in com-

mon organic solvents and thus, do not need to be converted from a precursor stage.

as do PPV and some forms of PPP.31,33

The diode fabrication process proceeded as follows. Films were solution cast or

spin coated onto indium/tin oxide (ITO) covered glass substrates. We estimate the

film thicknesses to be less than 1 Inm. Good quality thin films were obtained in this

manner. Next, 2000 A of aluminum was evaporated on top of the polymer film using

a micro-etch. The aluminum served as a counter electrode contact. Silver paste Cwas

used to attach thin wire leads to both the ITO and Al contacts.

Photo-luminescent Measurements

The steady state photo-luminescence measurements were carried out on a mod-

ified SPEX Model 1681 fluorimeter in the laboratory of Dr. K. Schanze at the

University of Florida. The spectrometer had a Xenon lamp source, two monochro-

mators and a photomultiplier detector. Experiments were run in emission mode,

using Aezcit=330 nm for the phenylene based polymers, and 330 and 500 nm for

the thiophene-phenylene-thiophene based materials with hexane and dodecyloxy side

groups, respectively.

Time resolved PL measurements were used to determine exciton lifetimes. These

measurements were done using a Photochemical Research Associates instrument

(Model 1551) and a photomultiplier detector. Excitation energies were identical to

those used for the steady state emission measurements. Decay profiles were mea-

sured at the emission maximum, or as close to the maximum as possible, given the

availability of filters.

Electro-luminescence Measurements

The overall design showing the various components of the electroluminescence

experiment in the light tight box is diagramed in Fig. 24. lMea.iirements were run in

a light-tight box that was purged with nitrogen to prevent oxidation or degradation

of the polymer films under high voltage. The sample holder is mounted on an x/y/z

stage inside the light tight box. A computer was interfaced to automatically step the

power supply and simultaneously record the resultant current-voltage characteristics

and photomultiplier light intensity (as measured by Keithley digital multimeters) in

a data file.

The circuit diagram for the electroluminescence experiment is shown in Fig. 25.

The indium-tin-oxide (ITO) electrode was run at positive bias with respect to the Al

electrode. Thus, holes were injected at the ITO electrode and electrons were injected


at the Al counter electrode, presumably combining in the film with the emission of

light. Active areas were on the order of 0.4 cm2.

Table 1. Bruker FTIR Operating Parameters

Range Beam Split. Opt. Filt. Source Pol. Detect.

cm-1 MI t ( i al Material Material

35 90 Myllar Black PE Hg arc 1 bolometer

80 400 Mylar Black PE Hg arc 1 bolometer

100 600 My\lar Black PE Hg arc 1 bolometer

450 4000 Germanium on KBr none Globar 2 DTGS, photocell

PE = polyethylene. Polarizer 1 = wire grid on oriented polyethylene; polarizer 2 =

wire grid on AgBr.