Sy n t h e s e s , C haracterizations , a n d P r o c e ssin g o f U nconventional C o n d u c t in g P o ly m ers

DISSERTATION

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

By

Xiaolin Wei, B.S., M.S.

sjc # 5(c sfe ^

The Ohio State University

1996

Dissertation Committee Approved by Professor Arthur J. Epstein Professor Terry L. Gustafson

Professor Linn D. VanWoerkom visor Chemical Physics Program UMI N um ber: 9 6 3 9 3 7 4

Copyright 1996 by Wei, Xiaolin All rights reserved.

UMI Microform 9639374 Copyright 1996, by UMI Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code.

UMI 300 North Zeeb Road Ann Arbor, MI 48103 To My Parents and My Siblings A cknowledgements

I would like to express my sincerest thanks to my advisor, Professor Arthur J. Epstein. In the course of my study and research at The Ohio State University (OSU), he has on innumerable occasions given me advice, guidance, and boundless encour­ agement which have been invaluable in facilitating the completion of my my Ph. D. at OSU. I also wish to express my appreciation to Professor Alan G. MacDiarmid at University of Pennsylvania who offered me an excellent opportunity to learn var­ ious techniques of making different kinds of conducting polymer powders and films. My gratitude is also extended to Professor Patrick Gallagher for his collaboration with my research project, and Professor Terry Gustafson and Professor Linn Van- woerkom for sharing their valuable time in carefully reviewing my dissertation and giving enlightening comments.

I have appreciated the help I received from Dr. J. Yue, a chemist graduated with his Ph. D. from Dr. Epstein’s group, who has shared his experience with me in various ways which helped me a lot to start my research in this lab, and from G. Min and Dr. Jim Shit in Dr. MacDiarmid’s group, for showing me many useful chemical processing techniques in making PANI and PPV polymer powder and films. I am grateful to my labmates Y. Wang, S. Long, and Dr. Zhiming Zhong in Dr. Patrick Gallagher’s lab for their contribution and collaboration to my research project. I also thank the undergraduate students, G. Gordon, C. Bobeczco, and S. Flagm, for their great assistance and contributions when they worked with me on certain projects.

iii Thanks also go to the previous Epstein’s group members — Drs R, McCall, Z. Wang, K. Cromack, J. Leng, Jinsoo, K. Kim, Narayan, M. Jozefowicz, Ms. K. Coplin, P. Zhou, Vasco, Shashi, Liang-Bih Lin, Will, and Z. Oblakowski; and the present group members — Gang Du, Wang, Darren Gebler, Randy, Scott, Jim, Mihai, Nadya Piskin, Ilya Lebedenko, Keith Brenneman, Alexey Saprigin, Chuck, Lee, Mats, and Fleda — who have given me a lot of support, much encouragement, and numerous useful discussions. I would like to thank Tim Lundgren in English department and the people in the graduate school who have reviewed, edited, and helped me in writing this dissertation.

I am indebted for the great amount of help and technical assistance I received from various technique support teams. I wish to express appreciation for those who have provided the high quality services and support I have received. I especially thank the following experts: G. Renkees (FTIR, EPR); R. Tucker and Joul (XPS); R. Grinsted, C. Engelman, and C. Cottrell (NMR); T. Kelch (Graduate Student Shop); Brian and his colleagues (Physics Computing Facility); B. Merritt (Liquid Helium Shop); Rita (SQUID); B. Henthorne (Glass Shop); and the secretaries of Dr. Epstein’s office — Diana Malone, Jackie Parris, Mark Baumgardner, Jenny Finnell, and Ken Hazard for their patient assistance and help.

The financial support I received from the Department of Chemistry at OSU dur­ ing my study here is greatly appreciated. Without the offer of a teaching assistantship from them I would not have been able to work here towards my Ph. D. A lot of thanks go to professor W. Mathews, D. Malone, and B. Cassity for their efforts in support of my work in the chemistry department. In addition, the financial support of the U. S. Office of Naval Research, Air Force Office of Scientific Research, and The Center of Material Research at OSU is gratefully acknowledged. I would like to thank my parents, and my brothers Qing Wei and Yong Wei, and the other members of my family for supporting and encouraging me whenever I had difficulties during my studies here. V i t a

1977-1981 ...... B.S., Department of Chemistry, Fudan University, Shanghai, P. R. China

1981-1984 ...... Instructor, University of Television and Broadcast in Shaanxi Province, Xi’an, Shaanxi, P. R. China

1984-1987 ...... M.S., Department of Chemistry, Nanjing University, Nanjing, P. R. China

1987-1988 ...... Instructor, Department of Chemistry and Chemical Engineering, Xi’an Jiaotong University, Xi’an, P. R. China

1989-1990 ...... M.S., Department of Chemistry, Northern Illinois University, DeKalb, Illinois, USA

1990-present ...... Graduate Associate, The Chemical Physics Program, The Ohio State University, Columbus, Ohio, USA P ublications

A Method of Synthesizing Highly Sulfonated Polyaniline, X.-L. Wei and A. J. Epstein, Synth. M et., 74, 123 (1995).

Synthesis and Physical Properties of Highly Sulfonated Polyaniline , X.-L. Wei, Y.-Z. Wang, S. M. Long, C. Bobeczco, and A. J. Epstein, J. Am. Chem. Soc., 118 (11), 2545 (1996).

Simulations of the in situ Cyclic Voltammetry Dependent EPR and DC Conductivity, X.-L. Wei, and A. J. Epstein, Synth. Met., xx, xxx (1996).

Highly Fluorinated Polyaniline: Synthesis and Structural Characterizations , X.-L. Wei, and A. J. Epstein, to be submitted.

XPS Study of Highly Sulfonated Polyaniline , X.-L. Wei, and A. J. Epstein, to be subm itted.

Quasi-random Oxidation Model: Important Role of the Coulomb Repulsion and the Formation of Polaron Lattice in Electrochemical Oxidation Process of Polyaniline, X.-L. Wei, and A. J. Epstein, to be submitted.

S e l e c t iv e P resentations

Highly Sulfonated Polyaniline: Synthesis and Characterization, X. Wei, Y. Wang, S. M. Long, C. Bobeczko, and A. J. Epstein, 1994 March Meeting of the American Physical Society, Pittsburgh, Pennsylvania: Bulletin of the American Physical Society B 27 7, 159 (1994).

Synthesis and Characterization of Highly Fluorinated Polyaniline, X. Wei, S. Long, Y. Wang, G. Du, S. W. Jessen, Z. Zhong, and A. J. Epstein, 1994 March Meeting of the American Physical Society, Pittsburgh, Pennsylvania: Bulletin of the American Physical Society B27 11, 160 (1994).

A New Methodology for Synthesizing Polyaniline with Improved Yield: Synthesizing Pemigraniline as an Example , X. Wei and A. J. Epstein, 1994 Great Lakes-Central Joint Regional ACS Meeting, Ann Arbor, Michigan: Bulletin of the American Chem­ ical Society 38, 311 (1993).

Quasi-random Oxidation Model: Simulations of in situ CV, DC conductivity, and EPR spectra, X. Wei and A. J. Epstein, 1996 March Meeting of the American Physical Society, St. Louise, Messuri: Electronic bulletin of the American Physical Society M23 10, xxx (1996).

Simulations of the in situ Cyclic Voltammetry Dependent EPR Spectra and DC con­ ductivity ', X. Wei and A. J. Epstein, 1996 International Conference of Synthetic Met­ als, Salt Lake City, Utah: Session P2; MSN: 229.

F ield o f St u d y

Major Field: The Chemical Physics Program

Specialized in

• Experiment in Polymer Synthesis and Characterizations

• Electrochemistry Studies of Conducting Polymer Films

• Applications such as Optical Quality Conducting Thin Film Preparation

• Electric Transport and Magnetic Studies of Conducting Polymers

Advisor: Professor Arthur J. Epstein T a b l e o f C o n t e n t s

DEDICATION ...... ii

ACKNOWLEDGEMENTS ...... iii

VITA ...... vi

LIST OF FIGURES ...... xvii

LIST OF TABLES ...... xviii

LIST OF SCHEMES ...... xix

CHAPTER PAGE

I Introduction...... 1

II Theoretical Background ...... 6

2.1 Electronic Structure, Molecular Orbital, and Band Theories 7

2.1.1 Free Electron Fermi Gas M odel...... 7 2.1.2 Hamiltonian, Energy Levels, and Wave Functions o f Free E lectron Ferm i G as S ystem ...... 9

2.1.3 Band Theory ...... 12

2.1.4 Hamiltonian of Polyaniline...... 18

2.1.5 Some Other Important Concepts...... 21

2.2 Magnetic Properties of M aterials ...... 25

2.2.1 Basic Concepts ...... 28

2.2.2 The Correlated M agnetism ...... 32

2.3 Mechanisms of Polymer Chem istry...... 34

2.3.1 General Polymerization Mechanisms...... 36

2.3.2 Polymerization Mechanism for Polyaniline Family 41

III Experimental Techniques ...... 52

3.1 Introduction ...... 52

3.2 The Four Probe Conductivity Measurement ...... 53

3.3 NMR Measurement ...... 56

3.3.1 Splitting of Energy Levels in a Magnetic Field . . 56

3.3.2 Relaxation Related Phenomenon...... 61

3.3.3 Preparation, Acquisition, and Processing of the NMR Spectrum ...... 66

3.4 EPR Measurement ...... 67

3.4.1 Analogies between EPR and NMR Spectroscopes 67

3.4.2 EPR Line Shape ...... 69

3.4.3 The Spin Exchange Phenomenon...... 70

3.4.4 Pressure Broadening ...... 71

x 3.5 Cyclic Voltammetry Analysis...... 72

3.5.1 CV Experiment Setup ...... 73

3.5.2 General Pathways of an Electrode Reaction .... 77

3.5.3 Cyclic Voltammetry and Reversal Techniques . . . 79

3.5.4 Multistep electrochemical processes ...... 84

IV Synthesis and Physical Properties of Highly Sulfonated Polyani­ lin e ...... 87

4.1 Abstract ...... 87

4.2 Introduction ...... 89

4.3 Experimental ...... 92

4.3.1 Chemical Synthesis...... 92

4.3.2 Characterization M ethods ...... 94

4.4 R esu lts and D is c u s s io n ...... 95

4.4.1 The Mechanism and Rationale for LEB Route . . 95

4.4.2 Vibrational Spectra ...... 99

4.4.3 Electronic Spectra ...... 101

4.4.4 Temperature-dependence of DC Conductivity . . . 104

4.4.5 Temperature Dependence of Electron Paramag­ netic Resonance ...... 108

4.4.6 pH-Dependent Behaviors ...... 112

4.4.7 Models for Electrochemical Redox Processes . . . 120

4.5 Summary ...... 143

4.6 Acknowledgement...... 143

xi V Highly Fluorinated Polyaniline: Synthesis and its Characteri­ zation s ...... 144

5.1 Abstract ...... 144

5.2 Introduction ...... 145

5.3 Experimental ...... 146

5.3.1 Chemical Synthesis and Sample Preparations . . . 146

5.3.2 Film preparing processes ...... 149

5.3.3 Characterizations ...... 150

5.4 Results and Discussion ...... 152

5.4.1 Mechanism and Rationale for Chemical Synthesis 152

5.4.2 Element chemical analysis ...... 156

5.4.3 Fast Atom Bombardment Mass Spectrum ..... 156

5.4.4 UV-Vis Spectra ...... 158

5.4.5 FT-IR Spectra ...... 161

5.4.6 X-ray Photoelectron Spectra ...... 172

5.4.7 Solution NMR Spectra of F P A N ...... 179

5.4.8 Thermogravimetric Analysis and Differential Ther­ mal Analysis...... 196

5.4.9 DC Conductivity measurement...... 196

5.4.10 Electron Paramagnetic Resonance Analysis .... 201

5.4.11 Cyclic Voltammetry Experim ent...... 201

5.5 Summary ...... 205

5.6 Acknowledgment...... 205

VI Conclusions...... 206

xii 6.1 Highly Sulfonated Polyaniline ...... 206

6.1.1 Synthesis...... 206

6.1.2 Physical properties...... 207

6.1.3 Quasi-random Oxidation Model ...... 208

6.1.4 Electrochemistry...... 208

6.2 Highly Fluorinated Polyaniline...... 208

6.2.1 Synthesis...... 209

6.2.2 Physical properties and characterizations ...... 209

6.2.3 The resonance model for FTIR spectra...... 210

6.2.4 The suggested future research topic in polyaniline f i e l d ...... 210

BIBLIOGRAPHY ...... 211

xiii L ist o f F ig u r e s

F ig u r e P age

2.1 Solutions of central equation in a periodic field 2Ucos(Gx) ...... 16

2.2 Diagram of the Ith unit cell of a LB c h a in ...... 19

2.3 Calculated band structure of LB in an extended zone scheme ...... 22

2.4 Polaron and bipolaron formation and inter-conversion for PANI . . . 26

2.5 VEH band structures for polaron and bipolaron lattices ...... 27

2.6 Characteristic x °f diamagnetic and paramagnetic substances .... 29

2.7 T-dependent x >n paramagnet, ferromagnet, and antiferromagnet . . 35

3.1 Conductivity measurement via four probe device technique ...... 54

3.2 NMR energy level splitting ...... 59

3.3 Classical representation of the perturbation of magnetization Mo . . . 63

3.4 Schamatic diagram of NMR spectrum aquisition ...... 65

3.5 Typical experimental setup for cyclic voltammetry ...... 74

3.6 Diagram for “three-electrode cell” setup ...... 75

xiv 3.7 An example of three-electrode cell glassw are ...... 76

3.8 Pathway of a general electrode reaction ...... 78

3.9 CV potential sweep and its resultant voltammogram ...... 80

3.10 Relationships among CV parameters of £ 1/ 2, Ep, and Ep/2 ...... 82

3.11 Parameters in CV i s characteristic ...... 83

3.12 A multi-component system and a multi-step charge transfer reaction . 85

4.1 FTIR spectra of LEB-SPAN and EB-SPAN in KBr pellet ...... 100

4.2 FTIR spectra of EB-I and ES-I in KBr pellet ...... 102

4.3 UV-Vis spectra of LEB-SPAN and EB-SPAN in 0.1 M NH 4 OH .... 103

4.4 T-dependent conductivity (

4.5 T-dependent (Tdc of LEB-SPAN, fit to 3D-VRH m odel ...... 106

4.6 T-dependent Cdc of LEB-SPAN, fit to Arrheneues expression ...... 107

4.7 T-dependent g-value of LEB-SPAN ...... 109

4.8 T-dependent EPR linewidth of LEB-SPAN (AHPP and FWHH) . . . 110

4.9 T-dependent AHfwhh to Hpp ratio of LEB-SPAN ...... I l l

4.10 Magnetic susceptibility ( x ) vs• temperature (T) plot for LEB-SPAN . 113

4.11 (x ■ T) vs. T plot for LEB-SPAN ...... 114

4.12 pH-dependent a-dc of LEB-SPAN, EB-SPAN, and PA N I ...... 116

4.13 Cyclic voltammograms of LEB-SPAN in pH = 1 and 2 buffers .... 117

4.14 Half-wave potential, E ^ , i 5 of LEB-SPAN vs. pH p lo t ...... 118

4.15 Simulation result of the in situ CV and EPR spectrum ...... 130

4.16 Simulation result of the in situ CV/DC conductivity spectra ...... 132

xv 4.17 Protonation and oxidation equilibria in polyaniline chains ...... 138

5.1 The plot of the oxidizer/monomer ratio us. the yield of FPAN syntheses 153

5.2 Fast Atom Bombardment Mass spectrum of FPAN sam ple ...... 157

5.3 Solution UV-Vis spectra of FPAN base and s a lt ...... 160

5.4 UV-Vis spectra of both base and salt form of FPAN films ...... 162

5.5 FTIR spectrum of FPAN base and salt pow ders ...... 163

5.6 Comparison of FTIR spectrum of FPAN salt with that of PAN-HC1 . 164

5.7 Intensity oscillations in various FTIR absorptions (one-scan series) . . 166

5.8 Intensity oscillations in various FTIR absorptions (64-scan series) . . 167

5.9 Time-dependent FTIR absorbances of FPAN pellet (24 hrs old) for one-scan series ...... 168

5.10 Time-dependent FTIR absorbances of FPAN pellet (24 hrs old) for 64-scan series ...... 169

5.11 Time-dependent integrated peak intensity of amine and imine-like struc­ tures for 64-scan series ...... 171

5.12 XPS N Is spectrum of FPAN base ...... 174

5.13 XPS C Is spectrum of FPAN base ...... 176

5.14 XPS F Is spectrum of FPAN base ...... 177

5.15 Comparison of 1H spectra with and without adding D 2O in FPAN so lu tio n s ...... 182

5.16 2-D H-F correlated solution NMR spectrum ...... 184

5.17 1-D ^-decoupled 19F solution NMR spectra of FPAN sample .... 186

5.18 2-D 19F COSY16 solution NMR of FPAN sam p le ...... 188

5.19 2-D solution 1BF phase sensitive NOESY NMR of FPAN sample . . . 192

xvi 5.20 TGA and DTA spectra of FPAN powder in pristine fo rm ...... 197

5.21 TGA and DTA spectra of FPAN powder in base form ...... 198

5.22 Pyrolysis gas chromatogram for the pristine form of F P A N ...... 199

5.23 Pyrolysis gas chromatogram for the base form of FPA N ...... 200

5.24 Comparison of EPR spectra of FPAN base and salt at 25° C ...... 202

5.25 Cyclic voltammogram of FPAN in 1.000 M H C 1 ...... 203

5.26 Cyclic voltammogram of FPAN made electrochemically ...... 204

xvii L ist o f T a b l e s

T a b l e P a g e

4.1 Spin-counting schemes in th simulation of in situ E P R d a t a ...... 127

4.2 Several frequently used covalent bond lengths in conducting polymer 129

5.1 The correlation between the oxidizer/monomer ratio and the yield of FPAN syntheses ...... 149

5.2 The binding energies and atomic concentrations of multi-scan XPS spectra of FPAN base ...... 173

5.3 Peak labelling for 1H-decoupled 19F NMR spectrum ...... 187

5.4 Correlation table for the COSY and the NOESY spectra of FPAN samplel90

xviii L ist o f S c h e m e s

S c h e m e P a g e

2.1 Step-reaction polymerization mechanism of Nylon 6 6 ...... 37

2.2 Example of cationic polymerization: Polymerization of isobutylene . . 38

2.3 Example of anionic polymerization: Polymerization of acrylonitrile . . 39

2.4 Mechanism of free-radical polymerization ...... 40

2.5 Polyaniline: Free-radical mechanism proposed by Mohilner et al . . . 42

2.6 Formation of aniline dimers: p-aminodiphenylamine and benzidine . . 44

2.7 Formation of aniline oligomers and polym ers ...... 45

2.8 Intermediates and compounds in mechanism proposed by Frank Lux . 46

2.9 Dimerization mechanism of PBOI to blue imine of Willstatter .... 47

2.10 Polyaniline polymerization mechanism proposed by Frank Lux .... 48

2.11 Polyaniline polymerization mechanism by Frank Lux continued (1) . . 49

2.12 Polyaniline polymerization mechanism by Frank Lux continued (2) . . 50

4.1 Chemical structures of LEB, EB, PNB, and SPAN ...... 91 4.2 The synthetic route for LEB-SPAN ...... 93

4.3 The proposed sulfonation mechanism for LEB-SPAN . 97

4.4 Interconversions among reduced and oxidized repeat units ...... 122

4.5 Visual representation of quasi-random oxidation processes ...... 124

4.6 The resonances of a polaron between two adjacent nitrogen sites . . . 136

4.7 Oxidations of the substituted LEB in less acidic m e d ia ...... 137

4.8 Oxidations of the substituted LEB in very acidic m edia ...... 140

4.9 Hydrolysis mechanisms for various substituted PANI ...... 142

5.1 The chemical structures of FPAN monomer, meta-coupled FPAN lat­ tice, ard para-coupled FPAN lattice 152

5.2 Resonance structures of amine unit in FPAN sample ...... 161

5.3 Resonance structures of partially oxidized FPAN sample ...... 178

xx C H A P T E R I

Introduction

Polymers, or macromolecules, occurring naturally (biopolymers) or synthetically, are those molecules made up of multiple repeat units joined together by covalent bonds. Their molecular weight is hundreds or thousands times higher than that of normal molecules such as water or methane, and therefore the mechanical properties of poly­ mers are unique, much different than those of other inorganic or organic materials. However, there is great variety among polymers themselves. As far as the usage of polymers (conventional polymer, per se) are concerned, they could be categorized as plastics (thermoplastic and thermosetting plastics), fibers, and elastomers or rubbers. Those differences in physical properties among different kinds of polymers are primar­ ily determined by the intermolecular and intramolecular forces and by the functional groups attached on the polymer backbone. To most of people it might not be evident that our houses, meals, cars, clothes, almost everything around us, and even more surprisingly, we ourselves are made of polymers — we are living in a polymer age today!

Since conventional polymers are insulators it was natural to think of polymers

1 2 as electrical insulators rather than conductors, until, when in 1967, the electrical transporting abilities of polyaniline (PANI), were discovered [1]. From that time on, the conventional concept about polymers has been radically changed and the doped conjugated polymers which can carry electricity, namely conducting polymers, have emerged as an important novel category of polymer. Among these relatively new conducting polymers [ 2 ] the following have become well known to scientists in various areas: polyaniline [3], polyacetylene [4, 5], poly(p-phenylene) [ 6], poly(p- phenylene sulfide) [7], poly(p-phenylene vinylene) [ 8], polypyrrole [9], polythiazyl [10], polythiophene [ 11], poly(thienylene vinylene) [ 10], poly(p-pyridine vinylene) [ 12].

The concept of doping, which converts conjugated polymer from an insulator to a conductor, has made conducting polymers a great deal of interest in many fields such as in chemistry, chemical engineering, solid state and theoretical physics, electrical engineering, and material sciences. The ease and low cost involved in the preparation and fabrication processes of conducting polymers as compared with those of metals have facilitated their uses in many applications. A well known conducting polymer to this regard is polyaniline. Polyaniline has already been used in potential industrial applications such as welding of plastics [13], catalysis of biochemical reactions [14], m em ory devices [15], electrom agnetic shielding [16], nonlinear optics [17], rechargeable batteries [18], anti-corrosion [19], light em itting diodes [20], gas selective separation membranes [21] and so on. Its broad potential applications have attracted great interest from investors. For example, in 1991, several companies, such as Allied

Signal, Americhem, the Hexel company in the USA, and the Zipperling company in Germany, announced their plans on production of polyaniline on a large scale [22], which will bring up a new era of conducting polymer technologies.

However, even for polyaniline, many research questions and problems still remain 3 until now. A few of them that we are interested in are listed below.

• T he in situ CV and EPR experiments have been performed by many groups, but the abnormal variation of the in situ EPR spectra remained to be explained until now.

• The asymmetric shape of the in situ voltage-dependent DC conductivity result has still been a focus of two contradicting conducting mechanisms, i.e. the polaron lattice versus bipolaron lattice conducting mechanisms.

• Sulfonated polyaniline (SPAN) has been made via many different ways, but the important factor, S/N ratio, repoted was at or below 50 %.

• Different electrochemical behaviors of polyaniline and its derivatives have been reported but not yet been interpreted.

• Highly fluorinated polyaniline (FPAN) has been made via electrochemical meth­ ods, but the chemistry version of the synthesis and the complete set of charac­

terizations has not been done, which limits its applications.

The excitement and the interesting open research topics in polyaniline field mo­ tivates us to chose polyaniline (its derivatives and new couplings) as a major model polymer in my dissertation. Besides, it has following advantages. (1) Polyaniline is simple in structure but rich in its physical properties and in variation of its chemical oxidation state (since the nitrogen atom, which can be easily doped or oxidized, is in­ corporated into the conjugated polymer backbone). (2) The meta-coupled polyaniline is a model for study of polymer ferromagnets. (3) It is linear so that the interchain interaction can be studied via comparison of unstretched and stretched films. (4) 4

Various substituted polyanilines are readily available for the study of the effect of substituent on band structure.

In the past of six years, many research objectives for my dissertation have been successfully achieved. The major results are listed below, and will be discussed in details in chapters that follow.

• A novel model, namely, The Quasi-random Oxidation Model , was proposed. Based on this model, the simulations on the in situ CV, DC conductivity, and EPR spectra have been very successful.

• The unexplained variation of the in situ EPR spectra has been interpreted successfully with our model.

• The nature of the asymmetry shape of the volt age-dependent DC conductivity spectrum has also been explained with our model and related simulations.

• A convincing evidence to support the polaron lattice conducting mechanism was provided with our simulation results, which will hopefully clear out the confusions in conducting mechanism in conducting polymer field.

• A new synthesis route for SPAN preparation was developed, which raises the S /N to ~ 0.75, which is % higher than the previous S/N limitation (S/N = 0.5).

• A new route for preparation of FPAN chemically was developed, and the FPAN thus made was fully characterized.

• A new route for pernigraniline with higher yield and easier preparation was developed. 5

• Many novel physical properties, such as pH-independent DC conductivity of SPAN over entire pH range and oscillating FT-IR phenomenon for FPAN ma­ terial, have been revealed and will be shown later.

In chapter II, the theoretical background of the polymer chemistry and physics for the polyaniline system, a quasi 1-D system, will be reviewed briefly. In chapter III, experimental techniques will be presented. In chapter IV, synthesis and char­ acterization of sulfonated polyaniline (SPAN) will be described. The Quasi-random Oxidation Model and the related simulations of the in situ CV, DC conductivity, and

EPR spectra will also be shown in this chapter. In chapter V, the topics of the highly fluorinated polyaniline (FPAN, a semi-transparent conductor) will be presented. In chapter VI, conclusions will be drawn on the syntheses and characterizations of con­ ducting polymers we made. References will be attached afterwards to aid the readers to access the relevant literature. C H A PT E R II

Theoretical Background

There are quite a few theories describing electromagnetic phenomena, molecular or­ bitals, and electronic band structures of metals. The significance of discussing the theorem of metals lies in the fact that metals usually are crystals (therefore with periodic structures). These theorems are therefore applicable to the conducting poly­ mers which are at least 1-D crystals. The easiest and most frequently used ones among these theorems are: (1) Huckel theory; (2) free electron Fermi gas model; (3) tight band theory; (4) SSH theory; and (5) Curie-Weiss law. In the first part of this chapter, theories about electronic and electric phenomena will be discussed briefly. In the second part of this chapter, the theorem and concepts about magnetisms will be briefly introduced and discussed. The detailed discussion can be found elsewhere [2, 23, 26, 126].

The final part of this chapter studies polymerization mechanisms of polyaniline and the related chemistry principles. The need for understanding the polymerization mechanisms lies in the improvement of the yield and selectivity of the polymerization process. It also helps to increase the crystallinity of samples, which is fundamental to

6 7 the increment of the dimensionality of the samples and therefore their conductivity. It has become more and more important to learn the details of how the polymerization of polyaniline is initialized and how it grows so that the reaction can be controlled to proceed along certain pathways (either to get rid of a suspected cancer agent, e.g. benzidine, or to make a better polymer).

2.1 Electronic Structure, Molecular Orbital, and Band The­ ories

In this section, theories and concepts related to the electronic and electric phenomena of the crystal lattice are described and discussed. We will begin with the simplest theory, i.e. the free electron Fermi gas theory.

2.1.1 Free Electron Fermi Gas Model

The free electron model can be utilized to interpret many physical properties of met­ als, e.g. thermal and electrical conductivity, heat capacity, magnetic susceptibility, and the electrodynamics of metals. Basically, this model states that the valence elec­ trons of the atoms in condensed matter are ionized to become conduction electrons which move almost freely within the whole space of the matter [23].

The free electron model is useful for understanding the physical phenomena of conducting polymers because there are more and more metallic behaviors of conduct­ ing polymers reported[24]. However, the model will be only reviewed briefly in the 8

following section since it has been discussed in great detail by many authors. The interested reader can find detailed discussion of many aspects of this model in solid state physics books such as those written by Kittel [23] and by Ashcroft [25].

To understand the basic concepts related to this model, we had better exemplify the simplest metal, alkali metal of lithium. The valence electron in a free lithium

atom is in the 2s atomic orbital while this electron in the metallic phase of lithium

(i.e. in condensed matter) becomes a conduction electron in the 2s conduction band.

We can think of this monovalent crystal as composed of N Li atoms which are ionized to give N conduction electrons and N positive ion cores.

Notice that the positive ion cores fill out less than 15 percent of the space of the Li crystal (the radii of Li+ and Li are 0.68 and 1.56 A[23]), which means that the majority of the crystal’s space is available for electrons to travel through. However, it is still difficult to understand the unusually long mean free path of the metal at low temperatures (in some cases it can be as large as centimeters) from the classical point of view. The transparency of metal to conduction electrons can be understood as

arising from ( 1) a conduction electron behaving as matter waves, which can propagate freely in the periodic crystal lattice of the ion cores so that it will not be deflected by the ion matrix, and ( 2 ) the unlikelihood of its being scattered by other electrons as the consequence of the Pauli exclusion principle (therefore this electron is said to be in a free electron Fermi gas). 9

2.1.2 Hamiltonian, Energy Levels, and Wave Functions of Free Electron Fermi Gas System

Energy levels and wave functions. As we know, the most interesting synthetic metals are linear conducting polymers so far, although several instances of 3-D electronic behavior in conducting polymers have been reported recently [27, 132]. Therefore we are more interested in a 1-D model than a 3-D model at the present time. We now turn our attention to the issues of getting wave functions and energy levels of a one dimensional electron gas. In this model, the interaction between the conduction electron and the core is considered to be weak, the potential energy is neglected in solving the Schrodinger equation below:

HV = EV (2.1)

* - a where H —^ only and p = —iftV. By applying infinite potential at the boundaries of x = 0 and x = L, we have ^&(0) = 0 and *P(Z/) — 0, and we get the energy level and its corresponding wave functions as follows:

li j n.7r >, . . £n = 2m ^ T 5 (22) ipn = A sin(^x) (2.3)

n — 1, 2, 3, ... (2.4)

Fermi energy. Fermi energy , ep, is defined as the energy of the topmost filled level in the ground state of N electron system, which is the one of the most important concepts of all in solid state chemistry and physics. For a Fermi electron gas system where the Pauli exclusion principle applies, we have 2njr = N where np denotes the topmost Riled energy level (n = 1 denotes the bottom level accordingly). The Fermi 10 energy level can therefore be obtained by substituting np into equation ( 2.2 ):

k2 TljrT 2 h2 Ni\ 2m ~L~ ~ 2m ~2L

Energy distribution of orbitals of free electron Fermi gas. The configuration of an N electron system is temperature dependent. The ground state is the state at absolute zero. When temperature increases, the kinetic energy of Fermi electron gas increases. The electrons in orbitals within ~ k sT of the Fermi level are excited thermally and therefore the energy distribution is changed. The Fermi-Dirac distribution describes the energy distribution of the equilibrium state of a Fermi electron gas system, which gives the probability of the occupation of an orbital at energy e at temperature T:

/ ( t ) = [exp[(e - ri/keT ] + 1J (2'6) where fi is the chemical potential of the electron system, a function of temperature, so chosen to meet the following conditions: (1) At absolute zero fi = ep is a divergent point at which the probability function changes discontinuously from one (filled) to zero (empty); (2) while at all other temperatures f{fi) equals one half. The other extreme is that at high energy (e — /x k s T ) the exponential term becomes much larger than one and the Fermi-Dirac distribution changes to a Boltzmann or Maxwell distribution:

/(e ) = exp [ - ( e - fi)/kBT]. (2.7)

Density of states. Another important concept is the density of state , D(e), defined as the number of orbitals per unit energy range. The total number of orbitals can be solved from equation (2.2) for 1-D system. The derivative of that is _ . . dN L ,2m ., (M> 11

3-D free electron Fermi gas. The energy level, its corresponding eigen-functions, the density of states, and the energy distribution of free electron gas in three dimensions can also be obtained similarly as those in a 1-D system. Applying the periodic boundary condition, the wave function of the system can readily be obtained as

V'k(r) = ex p (ik -r); (2.9)

^ = f - £ = 2tt 4tt kXi = 0 ; ..., Xi = x, y, z\ (2 .11) - = & = £Oi / a <2i2> (2.13)

The free electron gas model has been used to resolve quantitatively the discrep­ ancy between the experimental value and the predicted value of heat capacity of a metal. The major point is that only the electrons in orbitals with energy within

~ Ub T of the top of the energy distribution can be excited thermally at tempera­ ture T. Therefore the heat capacity of the metal is only ( T /T p ) percent of the value predicted by classical theory (IV is the Fermi temperature).

The electrical conductivity has been also interpreted using the free electron gas model. The model states that the charge transported is proportional to the charge density ne, charge to mass ratio e/m, and collision time r, which is consistent with the observed experimental law of conductivity (i.e. the Ohm’s law). 12

2.1.3 Band Theory

The free electron Fermi gas model of metals provides insight into many physical properties which depend on the kinetic properties of conduction electrons. However, it can not distinguish metals, semimetals, semiconductors, and insulators nor explain the occurrence of positive values of the Hall coefficient. Upon adding the interaction between the electrons and the periodic lattice of the positive cores into Hamiltonian for the free electron gas model, those which can not be interpreted by free electron Fermi gas model are readily resolved with the improved theory, namely, the Band Theory.

The band theory is a very powerful theory in the condensed matter field. The most important issues this model tackles are what perturbs the free electron gas system, how big the magnitude and what the origin of the energy gap is. From these concepts and quantities, almost every fundamental question related to electrons in metals can be answered quantitatively. In this section, we will again only briefly introduce some useful concepts for later use. Both the introductory and the detailed discussion can be found elsewhere [23, 28].

Energy band. Each quantum state of a free atom can be described by a given set of quantum numbers and a corresponding energy (eigenvalue). In a crystal the energy of every quantum state of an otherwise free atom spreads into a band of energies. The width of the band is proportional to the strength of the overlap interaction between neighborhood atoms. Degenerate states of these otherwise free atoms will form different bands. Bands can coincide in energy at certain values of k in the Brillouin zone.

Energy gap. The energy gap is the energy region where no wavelike electron orbital 13 exists. It is the difference between energies of the lowest point of the conduction band, the conduction band edge, and the highest point of the valence band, the valence band edge. The band gap can be obtained experimentally by measuring the threshold of continuous optical absorption [23] or from a cyclic voltammetry experiment.

In a linear crystal of N primitive cells with lattice constant of a, each primitive cell contributes one independent wavevector k, and therefore two independent orbitals (spin up or down), to each energy band. If there are an even number of valence electrons in one primitive cell, then the band is fully filled; if there is only one valence electron in the cell, then the band is half filled and the crystal will be a metal. The first Brillouin zone boundary is at k = There are two possible sources from which an energy gap can occur, one arises from the interaction of electrons with the periodic potential of the lattice, and the other arises from the Peierls instability, that is, at absolute zero, where a 1-D lattice of metal is distorted and an energy gap is created at the Fermi surface. The amplitude is determined by balancing the increased lattice elastic strain energy and the decreased electron energy below the Fermi energy (the quantitative description is given in a later section of this chapter).

There are several different approaches to the band theory, for example, the Nearly

Free Electron Model, the Kronig-Penney Model, and the general approach to the band theory. However all the approaches have one thing in common, that is, the periodic potential of ion cores is added into the free electron gas model. This periodic potential scatters the otherwise travelling wave and composes it with the reflected wave to form a standing wave at the Brillouin zone boundary. Since there are two ways of combining two eigenvectors (addition or subtraction), and the composite wave packets are different from the electron densities which are piled up differently with respect to the potential field, this results in two different energies at the boundary, 14 thereby producing an energy gap. The magnitude of the energy gap is, at the very first approximation, proportional to the Fourier component of the crystal potential.

The periodic potential of the positive ion cores can be expanded as a Fourier series:

U{x) = £ ( l / c e iC*) - 2 '£ U g cos{Gx), (2.14) G G> 0 where G is the reciprocal lattice vector and Ug=o is defined to be zero for convenience. Therefore, the leap from the free electron gas model to the band theory starts by adding this periodic potential to the Hamiltonian, which contains only a kinetic energy operator for the free electron gas model. Now the Schrodinger wave equation of an electron in a crystal becomes:

H V = EV (2.15)

+ = + (216) G The above equation can be solved by expanding tp(x) into a Fourier series:

ip = ]T C{k)eikx. (2.17) k After the expansion, the coefficient of the plane wave of a particular wavevector has to be set equal for both sides of the equation. Thus the algebraic form of the Schrodinger equation, namely, the central equation is obtained:

(A* - e)<7(A:) + £>G<7(fc - G) =0 (2.18) G where A*, is set equal to h2k2/2m. This central equation is a very useful form of the wave equation for systems with periodic potential. Its great practical use is seen when a small number of coefficients are chosen (two or four).

As an example of an application of the central equation (2.18), let’s solve this equation for a system with potential energy as UetGx + Ue~tGx at Brillouin zone 15 boundary and use a two-term approximation

${x) = C(k)eik* + C(k - G)e*(fc-C> , (2.19) therefore the central equations have the form

{\k - e)C(k) + UC(k - G) = 0; (2.20)

(A*_g - e)C(k - G) + UC(k) = 0, (2.21) with A* = h2k2/2m. The equations have a solution only when the determinant of the coefficients of C(fc)’s is vanishing. At the zone boundary, the general form of the equations above is simplified with k — | G. The energy at boundary has two roots with their corresponding eigen-functions (standing waves) as:

e = x±u = £ A g?±u' <2-22) •0(») = exp(iGa:/ 2 ) ± exp(— iGx/2). (2.23)

Therefore the periodic lattice with potential in equation (2.19) creates .an energy gap of 2 U at zone boundary The solution of the wave equation near the zone boundary can be solved similarly as above [23] and the results are shown in Fig. 2.1 below.

Though the exact form of the solutions of the Schrodinger equation for the pe­ riodic potential problem depends on the form of the periodic potential, all of these solutions have to be in a specific form, namely, the Bloch Function\

ipk(r) = u*(r)exp(zk • r), (2.24) where ujfc(r) = •u*(r + T ); a periodic function of crystal lattice translational operator T. The Bloch Theorem states that the solution of the Schrodinger equation for a system with periodic potential has to be in a form of the product of a plane wave 16

Zone boundary

2

e

0 0 0.5 1.0 1.5 (1/2 Gt)

Figure 2,1: The solutions of the central equation in a periodic potential field 2Ucos(Gx). After Kittel [23]. 17 exp(tk • r) and a function u*(r) with a periodicity of the potential of the crystal lattice system.

Tight binding approximation. The tight binding approximation, or alternatively, the linear combination of atomic orbitals approximation (LCAO), is used to calculate the band energies. Since it neglects various interactions [23], it is only a good description for the situation in which the influence of the orbitals of one atom on the orbitals of another atom is small (such as inner electrons in d bands of transition metals) but not for that of the conduction electrons.

In this approximation, the one-electron-wavefunction of a whole crystal is ap­ proximated by taking a linear combination of the atomic orbitals of all atoms in the crystal, as the name of the approximation implies,

M t ) = £ Ckj

The coefficients Ckj have to be in a form of N ~1^2etk'r> to satisfy the Bloch form of the wave function for a crystal of N atoms,

r) = J \ r 1/2 X]exp(ik-rjMr " ri)- (2.26) i The band energy of a crystal can therefore be calculated, to the first order of approx­ imation, by taking diagonal matrix elements of the Hamiltonian:

ek = < k|H|k > (2.27)

= £ £ exP[*Mrj-rm)] < >, (2.28) J m = £exp(-ik • pm) f dV (p*(r - p)H

ek = < k |H |k > = - a - 7 exp(-tk • pm), (2.30) m where we have implicitly defined

- a = J dV V*(r)Hv>(r); - 7 = J dV

The band energies for several crystal structures can be obtained by applying the equations above. For sc structure,

ek = —a — 2 7(005 (^* 0 ) + cos(fcva) + cosfA^a)); (2.32) for bcc structure,

ek = —a — 87 cos(-fcxa) cos(-A!yo) cos(-fc 2a); (2.33) 2 2 2 and for fee structure, ek = —a — 4 7 (005 ( ^ ^ 0 ) cos(^A:za) + cos(-fc*a) cos(^fc,.a) + cos(~fcj,a) cos^A ^a)). 2 2 2 2 2 2 (2.34)

The width of an energy band is entirely determined by the overlap integral 7 . The weaker the overlap, the narrower is the energy band, and greater is the effective mass

(«*• = * ’ /(& ?))■

2.1.4 Hamiltonian of Polyaniline

As is well known, the 7r-electron delocalization energy favors the phenyl rings of polyaniline in the plane of nitrogen atoms, while the substantial steric repulsion be­ tween adjacent rings forces them out of the plane. Thus, a role for ring torsional 19

21+1

21 + 3 H H

Ith unit cell

Figure 2.2: Diagram of the Ith unit cell of a LB chain. The phenyl ring is twisted at a torsional angle of $ 21+1 out of the C-N-C plane. motion has been suggested in the model Hamiltonian of polyanilines. This repulsive interaction has been summarized in two parts, the electronic and steric interactions between adjacent phenyl rings [31, 32, 33, 34, 150]:

H = Het + V't'Tic- (2.35)

A specific application of this theorem has been done for LB, leucoemeraldine base macromolecule. The diagrammatic representation of the torsional motion is shown in Fig. 2.2, where the unit cell of LB contains one N-H group (having two electrons in a pB-orbital perpendicular to the C-N-C plane) and one C 0H4 ring (having 6 pz electrons). With generalization of the treatment of the A-B polymer, the sites of Ith unit cell are labelled as follows: Site ‘A’, labelled as 2 1, is where the nitrogen atom pa-orbital is located; site ‘B’, labelled as 2 1 + 1, is where six molecular orbitals (MOs) of the phenyl ring are located; and the angle 2i+1 represents the torsional angle of the 20 phenyl ring within the Ith unit cell with respect to the C-N-C plane. The electronic part of the Hamiltonian can be approximated as a one dimensional linear combination of molecular orbitals:

I,* l,» 3 l,* 3

+ &2i + l ,i a 2f,») + *2i+l,2i+2(^2i+l,*a 2/+2,* + a 2l+ 2 ,,K l+ l ,s)} ■ (2.36)

The operator a\t J(a 2i,») creates (destroys) an electron of spin s at the nitrogen site of

Ith cell, with site energy an\ and &2H-i,*(^ 21+i,j) creates (destroys) an electron of spin

3 in the j ih MO of the Ith benzene ring (the benzene ring in Ith cell), with the site energy Cj. The transfer integral between the Ith nitrogen pz-orbital and the j th

MO of the Ith benzene ring is given by

^2/,2(+l = t c - N x cos2,+ 1 x c J i + n (2.37) where t c - N is the transfer integral between adjacent carbon and nitrogen atoms, cos is cosine of the angle between nitrogen and ring pz-orbital lobes, and c?2l+l is the coefficient of the j ih MO at a carbon atom adjacent to 21 nitrogen atom. Be­ cause of steric repulsion between adjacent phenyl rings, the ring rotation angles 2m+\ are supposed to alternate in signs: «^ 2m+i = ( — l) mV,2m+i (therefore two degenerate ground states are defined by ip2m+i = an(^ effective lattice energy is approx­ im ated as follows:

V'(V’2J-i,V’2f+i) - --^(sin2^ -! + sin 2 fal+i) + - ^ ( s i n 4 V>2(-i + sin 4 V'ji+i)

+Vi,i(sint^j+i “ sin ip2l-i)2 + constant , (2.38) where ipi and V>2 are the torsional angles of adjacent rings; V^o, ^ 4 ,0, and V \t\ are constants; the equilibrium ring angle V'o is defined in sin 2 ^0 = V2io/( 2V4 io), ~ 56° as suggested by the band structure calculations. 21

Applying second order perturbation theory to a system further away from pla­ narity, we can obtain transfer integral variation as t = tejf x cos2^. The shape of

the energy bands (effective site energy, i.e. the band center, band gap, and the band­ width) of electronic energy only are therefore modulated by the ring angle v>. Fig. 2.3 shows the calculated band structure for LB, where the ring angle was chosen as 56°

to match a ir — tt* band gap and valence bandwidth from spectroscopic studies [150].

Direct evidence of the important role of the ring torsional angle comes from the

fact that alkyl substitution on a benzene ring lowers the 7r — tt* transition energy of an unsubstituted benzene ring[36] while the UV-Vis spectra of the alkyl substituted

polyaniline, e.g. poly(o-toluidine) reveals increased transition energies over those of the parent polyaniline. This demonstrates that the steric repulsion between methyl and proton groups on adjacent rings apparently increases the ring torsional angle so

as to increase the tt — tz* transition energy [150].

2.1.5 Some Other Important Concepts

Holes. Vacant orbitals in a band are commonly called holes. In conducting polymers, charge carriers are not only electrons but also holes. In an electromagnetic field holes behave like particles with a positive charge +e.

Conductivity. Electrical conductivity comes from two contributions, the hole’s and the electron’s. The doping process of conducting polymers is different from that of semiconductors. For the former, it is an oxidation or reduction process that changes the filling status of the band structure of an otherwise insulating state; while to the latter, it is a process of deliberately adding n or p type impurities, donors or 22

8

4 CB > Q) VB 0

LU

4

8 0.0 0.5 1.0 k (p/a)

Figure 2.3: Calculated band structure of LB in an extended zone scheme. After Ginder [150]. 23

acceptors, to a semiconductor. In crystal, the ionization energy of the impurity atom is reduced by a factor of m ejm e2 (where e is the dielectric constant of the medium, m is the effective mass of a hole or an electron in crystal, and m e is the mass of an electron in vacuum) while the radius of the first Bohr orbital is increased by a factor of em /m e as compared to a free hydrogen atom. Since the radius of the impurity state increases appreciably, the impurity band can thus be formed by overlapping their orbitals considerably. In polyaniline the common doping process relates to Lewis acid doping, which creates holes, so the majority of charge carriers are therefore holes and the material is p-type. Similar to the impurity band in a semiconductor, the positively charged states (holes) can form a polaron band in emeraldine salt. The type of charge carrier can be determined by measuring the sign of the voltage of a thermoelectric power experiment [23].

Peierls instability of linear metals. Peierls suggested that at absolute zero, 1-D metal with electrons filling up orbitals to the level of kp is unstable because of lattice deformation of the wavevector at G = 2kp. The energy gap at the Fermi surface created by lattice deformation can be derived as

lAIA-WexpI-l/IVfOjK], (2.39) where A is the deformation, W — h2kp/2m, the conduction band width; ./V(O) is the density of orbitals (states) at Fermi level; V — 2A2 jC is the effective electron-electron interaction energy ( C and A are defined in elastic energy, Eeiatt jc = \C A2, and ion contribution to lattice potential, U(x) = 2AA cos(2Aifr;c), respectively). The above equation says that Peierls instability is a collective effect of all the electrons: the larger the width of the conduction band, the higher the density of orbitals at Fermi level, and the greater the electron-electron interaction, the larger is the Peierls energy gap. For polyacetylene, when the temperature goes to absolute zero, the originally 24

equivalent interatomic distances of the 7r bond dimerize into single and double bonds, which lowers the electron energy of those levels below the Fermi energy level until this

decreasing energy is balanced with the increasing of the elastic strain energy of the 1- D lattice of the metal. Therefore an energy gap is created at the Fermi surface which

transforms the 1-D metal into an insulator at absolute zero. Though polyaniline is more complicated in its chemical structure pernigraniline, the highest oxidation state of polyaniline, has indeed a double degeneracy in its ground state, and therefore has a contribution to its energy gap from Peierls dimerization [29, 150].

Polarons and bipolarons. A polaron is defined as an electron plus its associated strain field. Accompanying this phenomenon is an apparent increment in electron mass [23]. For a highly conducting polyaniline sample, models for formation and conversion of bipolarons to polarons, represented with their chemical structures, have been sug­

gested [32]. The electronic band structure calculation based on these models using MNDO (modified neglect of differential overlap) semiempirical self-consistent-field Hartree-Foch technique has been done with great success in comparing the calculated results with the optical experimental data [32]. Chemical structures of a bipolaron

and polaron lattice are shown in Fig. 2.4, while their band structures are shown in Fig. 2.5. Based on these results Bredas et al made the following statements: (1) The

polaron band is only half filled, allowing intraband absorption to occur; ( 2) direct optical transitions from band 6 to polaron band a (see Fig. 2.5 (b)) are calculated to be 1.8 eV, a result which matches well to the optical transition at ~ 1.5 eV; (3) direct optical transition from band c to band a is 2.6 eV, in good agreement with the optical transition at ~ 2.8 eV; (4) the first electronic transition involving the upper defect band and conduction band is predicted to occur at 4.1 eV, consistent with optical absorption. Further evidence of the success of the polaron band model is seen in the agreement of the calculated polaron band width with the experimental result (The 25 full width at half maximum height of the peak at 1.5 eV is ~ 1.0 eV, agreeing with

1.1 eV of the calculated bandwidth [35]).

Solitona. A soliton is a solitary wave that shows great stability in collision with other solitary waves [23]. An example of a soliton is Bloch Walls in a crystal that separate adjacent magnetized regions (domains) (see p452 in Kittel). Since pernigraniline is the only oxidation state of polyaniline which has a doubly degenerate ground state, it has two types of soliton states, namely type-I and type-II soliton states [29].

Excitons. An exciton is a bound electron-hole pair. In reflectance and absorption spectra an exciton structure can often be found just below an energy gap. The difference between the energy gap and the absorption energy of an exciton is the binding energy of the exciton, which is originated from the electrostatic attraction energy of the bound electron and hole pair. The exciton band can be found in the UV-Vis spectra of emeraldine base.

2.2 Magnetic Properties of Materials

In this section, we will only review some important concepts which are closely related to and used in this dissertation. Detailed treatment of magnetic properties of ma­ terials can be found in any solid state physics, or quantum mechanics, or statistical mechanics books [23, 37, 38]. 26

(a) H ,N, \ OJ IQ 'N' I H protonation (2x) H® (b) H I n ,N 'N' V'-V I H internal redox reaction

(c) H H I I ,N. ,N„ OJ 10100010 'N' N I I H H

polaron separation (d) H H I or* ,N, lOl XOJLQ,N„ "N" 'N' I H H

Figure 2.4: (a) EB; (b) Formation of bipolaron lattice; (c) Formation of polarons; (d) Separation of polarons into polaron lattice. After Ginder [150]. 27

o .o

X, K' cn

o 'E - 0.2 _ r -Bipolaron Band — o 1 a' a ------p O ------______—------—* k— n

LU -0.-4

- -

- 0 . 6 E—— P c3 ( a ) o .o

X, X 1tn

~ - 0.2 Polaron Band o co 2 . 6 ©V 1 .8 ©V a >- O DC LU

- 0.6 O k

C»=»>

Figure 2.5: In (a), a is HO band; a' is bipolaron lattice; there is a band gap between these two bands. In (b), b and c are HO band, while a is the lower polaron defect band, which is a half-filled band. In both (a) and (b), the x band is the LU (conduction) band and is the upper defect band (bipolaron band in (a) and polaron band in (b)) and they are flat and nearly degenerate). After Breda [35]. 28

2.2.1 Basic Concepts

Magnetic moments in materials can be correlated or non-correlated. The non-correlated magnetic moments give rise to paramagnetism and diamagnetism, while correlated (or ordered) arrays of magnetic moments give rise to ferromagnetic, or ferrimagnetic, or antiferromagnetic, or helical forms of magnetism.

The magnetic moment of a free atom or ion comes from four different sources: (1) Electron spin; (2) electron orbital momentum around a nucleus; (3) the changes in orbital moments induced by an external magnetic field; (4) a nuclear magnetic moment, which is as 10-3 orders smaller than that of the electron’s. The induced magnetic moment gives rise to a diamagnetic magnetism while others give rise to paramagnetic magnetism.

Magnetization and magnetic susceptibility. Magnetization M is defined as the total magnetic moments per unit volume, while magnetic susceptibility x Per unit volume is defined as

X = ~ (2.40) where B is the magnetic field intensity. A diagrammatic description of paramagnetic and diamagnetic susceptibility is shown in Fig. 2.6.

From Fig. 2.6, it is clearly seen that the material with negative susceptibility is defined as diamagnetic while that with positive susceptibility is as paramagnetic. We can think of a diamagnetic moment as arising from the induced current of an electron in an electron orbital within an atom, whose field is always opposing to the applied external field.

From quantum statistics [37], the Hamiltonian of an atom in a magnetic field is usacs Atr itl [23]. Kittel After substances. Figure 2.6: Characteristic magnetic susceptibilities of diamagnetic and paramagnetic paramagnetic and diamagnetic of susceptibilities magnetic Characteristic 2.6: Figure

Magnetic susceptibility 0 + t I prmagnetism param , Langevin (free spin) spin) (free Langevin Diamagnetism al prmants meas Te perature em T etals) (m agnetism param Pauli a lc prmagnetism param Vleck Van 29 30 defined as

1 , e H = (P + -A )2 - e(p (2.41) C eJ = H„ - iL ■ H + ^ £ ( H x r .)J, (2.42) where e is the amplitude of the electron charge; m the mass of the electron; A the vector potential;

fi = -(d F /d H )T.v,N, (2.43) and under the condition of (iB

fj, = N*H (2.44) where

* ^ { jl f w Q = y + Xdia, (2.46) where C is the Curie constant, and j the total angular momentum. It can be seen that the magnetic susceptibility, % — M f B, is composed of two contributions, one is 31 from the intrinsic magnetic moment, which is paramagnetic, while the other is from induced magnetic moment, which is diamagnetic in nature. In this way the classical Langevin result is confirmed. When an atom has a closed shell electron configuration with L = S = 0 (e.g. inert gas), then j is zero, and so is the intrinsic magnetic moment. Hence only Langevin diamagnetism contributes, and therefore the material is diamagnetic. When ions, or atoms, or molecules have an odd number of electrons (such as NO), or have a partially filled inner shell (such as ions or atoms of transition metal elements), or possess a biradical structure (such as the oxygen molecule), or have a metal band structure as their collective properties, the intrinsic magnetic contribution is much larger than its diamagnetic counterpart. As a consequence, the material is paramagnetic. Emeraldine salt has a structure of polaron lattice which has massive “free” electrons (and also conduction ones) in the defect band (polaron band) and therefore is paramagnetic. On the other hand, the base form of polyaniline (i.e. undoped) at any oxidation state has a filled band structure, and therefore has no intrinsic magnetic moment, i.e. its magnetic moments are from the induced diamagnetism and from the small paramagnetic contribution of isolated defect states.

Van Vleck temperature-independent paramagnetism. A system could have no mag­ netic moment in its ground state, however it can have an off-diagonal matrix ele­ ment, < s|/i 2|0 >, which connects the excited state ‘s’ to the ground state ‘ 0 ’ with energy difference of A = E t — E0. The perturbation theory is able to mix the the excited state and the ground state, which makes the perturbed ground state with a magnetic moment. The susceptibility so created has two different cases: ( 1) When energy difference A kBT, x takes a form of Curie type paramagnetism, through polarization of the states of the system (as compared to the redistribution of ions among the spin states), as x — W| < 10 > \2/kBT; (2 ) when A » kBT, the 32 paramagnetic moment such created is independent of temperature: % — ~ ~ , namely, Van Vleck paramagnetism.

Paramagnetic susceptibility of conduction electrons. It is well known that the mag­ netization of nonferromagnetic metals is independent of temperature, which is in contrast to a Curie-type temperature dependent paramagnetism. Pauli provided a sound explanation based on the Fermi-Dirac distribution: In metal, only electrons within the range ~ kgT of the top of the Fermi distribution could flip their spin states and therefore contribute to magnetic susceptibility. The temperature factor in the fraction of TjTp will be cancelled out with that in T-dependence of the other­ wise Curie-type of paramagnetism. After a Landau correction is made[23], the Pauli magnetic susceptibility of conduction electrons can be derived statistically as

which is exactly the same as the result of the “cancellation theorem” above.

The susceptibility of emeraldine salt has been found as the sum of two contri­ butions from conduction electrons and localized electrons. The contribution from conduction electrons arises from electrons in the polaron band, while the Curie con­ tribution is from the isolated polarons.

2.2.2 The Correlated Magnetism

A ferromagnet is a kind of magnetic material which has a nonzero magnetic moments even in a zero applied external field. In a ferromagnet, there exists an internal inter­ action, namely, the exchange field (B e ), which tends to line up magnetic moments 33 parallel to each other. In the mean field approximation , each atom experiences an exchange field proportional to the magnetization:

B e = AM. (2.48)

The Curie temperature , Tc, is defined as the transition temperature which sep­ arates the ordered ferromagnetic phase (T < Te) from the disordered paramagnetic phase (T > Tc). The Curie- Weiss law states that the magnetic susceptibility at temperature highers than Curie temperature follows

X = T - Te' ^2'49^ where C is the Curie constant and Tc the Curie temperature which is related to A in the mean field approximation as Tc = CX.

In the Heisenberg model the exchange energy is defined as

U = — 2JSj • Sj, (2.50) where J is the exchange integral which is related to the overlap integral of the charge distributions of atoms i and j, while Sj and Sj are the electron spin on atom i and j, respectively. The mean field theory also relates the exchange integral to the Curie temperature Te as M b Tc v (2.51) 2zS(S + 1) ’ where z is the number of nearest neighbors.

Magnons. The elementary excitations of a spin system having a wavelike form are called magnons. Magnons (spin waves) are oscillations in the relative orientations of spins on a lattice whereas phonons (lattice vibration) axe oscillations in the relative positions of atoms on a lattice. 34

Ferrimagnetism. The term ferrimagnetic is used to describe any compound in which some ions have a moment anti-parallel to other ions. The analytical expression of the susceptibility of ferrimagnetism is more complicated than that of ferromagnetism.

Antiferromagnetism. In anti-ferromagnetism the spins are ordered in an anti-parallel arrangement with zero net moment at temperatures below the ordering phase (which is defined as Neel temperature , Tjf). The susceptibility of antiferromagnetic materials at T > Tjv is 2 C X = T + Tn ' *2'52) Antiferromagnetism can be distinguished from ferromagnetism in terms of an ex­ change integral in the Heisenburg model, U — —2 JSj • Sj. If J > 0 then it is ferro­ magnetic, while if J < 0 it is antiferromagnetic.

The temperature dependence of susceptibility for paramagnetism, ferromagnetism, and antiferromagnetism are presented diagrammatically in Fig. 2.7.

Coercivity. The coercivity is the magnetic field Hc required to reduce the magnetiza­ tion or the induction B to zero from saturation.

Anisotropy Energy. In a ferromagnetic crystal the energy that directs magnetization along certain crystallographic axes (called directions of easy magnetization) is defined as magnetocrystalline or anisotropy energy.

2.3 Mechanisms of Polymer Chemistry

The traditional polymerization methods include condensation polymerization, ionic- chain-reaction polymerization, free-radical-chain polymerization, and copolymeriza- 35

Exchange intergal > 0 Exchange intergal < 0

HUHUH

Ferromagnet i sm Ant i f erromagnet i sm

(a) Spin ordering in ferromagnets (J > 0) and antiferromagnets (J < 0)

Paramagnetism Ferroagnetism Antiferromagnetism

Complex behavior

>.

QJCti «O CO -q o tn c = _C

Curie law Curie-Weiss law (T > TN ) (T > Tc)

(b) Several different types of temperature dependence of magnetic susceptibilities.

Figure 2.7: Temperature dependent magnetic susceptibility in paramagnet, ferromag­ net, and anti-ferromagnet. After Kittel [23]. 36

tion. The methods can also be classified as step-reaction polymerization and addition polymerization. Some new polymerization methods have been developed in recent

years, e.g. ring-opening metathesis polymerization of cyclic alkenes, group transfer polymerizations, living anionic and carbocationic polymerization, etc. [40], which have major advantages of producing a lower index of polydispersity product. In this sec­ tion, we briefly introduce the concept of these methods to facilitate a mechanistic discussion of the polymerization of polyaniline.

2.3.1 General Polymerization Mechanisms

General Step-Reaction Polymerization. The equation for stepwise polymerization is shown as follows:

nA + n B — > A (B A )n„iB -f BP (2.53) where A and B are bifunctional reactants, A {B A )n-\B is a polymerization product, and BP is a byproduct. An example of step-reaction polymerization is the polymer­ ization reaction of polyamides, which is shown in Scheme 2.1.

Ionic-Chain-Reaction Polymerization. Ionic-chain-reaction polymerization belongs to addition polymerization. There are two types of ionic-chain polymerization, i.e. cationic (an example of cationic polymerization is shown in Scheme 2.2) and anionic polymerization (see Scheme 2.3).

Free-Radical-Chain Polymerization. This type of polymerization also belongs to Ad­ dition Polymerization. The polymerization needs to be initialized with decomposition of initiators. The polymerization usually takes four steps: ( 1) Initialization; (2) ad­ dition reaction; (3) chain growth; and (4) a termination step. The mechanism is seen 37

H 02C(CH2)4C 0 2H + H2N(CH2)6NH2 ------*■ [■02C(CH2)4CO-][H3§(CH2)6NHj

heat J ? C(CH2)4CfXCH2)4CNH(CH2)6NH— + h 2o n

heat Adipic acid + 1,6-Hexanediamine Nylon 66 + Water

Scheme 2.1: Polymerization mechanism of Nylon 6 6: An example of step-reaction polym erization. A fter C arraher [39].

in Scheme 2.4.

Copolymerization. The copolymerization is different from homopolymerization in that the former has more than one kind of monomers in the feed, while the latter has only one kind of monomers in the feed. Notice that the reactivity of monomers may differ when more than one reactant is fed, therefore the form of the copolymer may differ greatly but it is not a mixture or a blend ([Mi]„ + [M 2]n)- A copolymer may be random copolymer, i.e. its sequence of M 2 and M 2 is randomly arranged as

M iM 1M 2M iM 2M2...; it may be alternating polymer, in which case the order of and M 2 is alternating in polymer chain; it may be graft copolymer, in which case the main chain is composed of one kind of monomers while the branches are composed of the other kind of monomers; it may also be a block copolymer, which has a formula as (M 1)n(M 2)n. 38

(1) BFi + H20 + (BF3o f )

,CHi (2) H2C=c' + P(BF3ofl —Ifa - HCC^, BF3OtP, Rj = kilCIIMI \:Hc h 3 ^CH3

c h 3 CHt P " 3 {=) (3) H3CO(+) , BF301T-y + n H2C =C v c h 3

h 3H 1 ,CH3 V b f 3oip , Rp = kp[ M | | p c h 3 H Ch J n t

H C (4) H- - i —c(+) , b f 3o P + M kTr 1I c h 3 H C H

CH3 b f 3oiP + P , Rnj = kxrlM IlP c h 3 4fFfH CH3 1

Scheme 2.2: Example of cationic polymerization: Polymerization of isobutylene. Af­ ter Carraher [39]. 39

ki (1) + Hh 202c =JNc h H2N H H

Rj = ki(M|[\P|

N CN H CN H CN Lo (2) H2N' P + n H2C=CH —^ H2N--(j:—C— U ■ H 9 H H H H H H

Rp = kp[Np][NH3]

kTr (3) H2N- + NH2 ( f f H H H H

H CN H CN ,0 H2N — H + :NH2

H H H H

RTr = kTr|NH 3 ] [ \ ^ |

Scheme 2.3: Example of anionic polymerization: Polymerization of acrylonitrile. Af­ ter Carraher [39]. 40

(1) l 2R*

Rd = - = kdUl

(2) R- + M — - — ► RM'

Ri = J ™ ! ± = kj[M|[R-

(3) + nM *■*"' M— M —n-l :-- M *

kt (4) r m * + -MR L RMMR

Rt = - = 2kt[M-|2

Scheme 2.4: Mechanism of free-radical polymerization. After Carraher [39]. 41

2.3.2 Polymerization Mechanism for Polyaniline Family

Polymerization mechanism for polyaniline and its derivatives has been subjected to extensive studies since 1910’s. Up till now dozens of polymerization mechanisms have been proposed already [41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55).

In 1910’s, Willstatter and coworkers denoted that PANI was a 1,4-oxidation polymerization product of aniline, which meant that the polymerization product is para-coupled anilines [55, 56].

In 1962, based on the result of the electrochemical kinetics (a Tafel plot resulted in two electrons as the number of electrons involved in the initial charge transfer step), Mohilner et al proposed a free radical mechanism for the electrochemical poly­ merization mechanism as seen in Scheme 2.5.

In the first step of Mohilner’s mechanism, aniline is equilibrated with proton while in the second step the protonated aniline is oxidized to a cationic radical. In step (3), the cationic radical is combined with a resonant structure of the cationic radical to give rise to a aniline dimer. The dimers are further oxidized and coupled together in step (4) and (5). This process continues until octamers are formed. We can see that this mechanism involves second order homomolecular recombinations (dimerization) which may put limit on the growth of the chain since the probability of such a special alignment of the oligomers are not so high on electrode. Further more, a 4ra + 2 type of proton migration, which needs a fairly high activation energy to start with, has to occur after the recombination step (dimerization).

In 1973, M. Breitenbach and K.-H Heckner also thought that the polymerization reaction of aniline was a kind of coupling reaction and found that the initialization Or^" * CXa.„— OrN'®L, * > ® <-’»

0 ^ - © ^ X X

C rTX BH • © rtX , - O XyO tto^ *»

+ 2H + 2 e (6) OrXkJOr^DL.- QXk&TXNH

O X K ^ X , * ® X k & X \

(7)

Scheme 2.5: Free-radical polymerization mechanism proposed by Mohilner et al for electro-oxidative polymerization of aniline. After Mohilner [44]. 43 process was only possible if the respective aniline molecules were non-protonated. They proposed that the polymerization reaction was initialized from coupling of doubly oxidized cation with rearranged aniline. With their proposed scheme (see Scheme 2.6), the formation of p-aminodiphenylamine and benzidine could be ex­ plained [57, 58, 59].

In 1975, disagreed with the mechanism proposed by Breitenbach and Heckner, Dunsch thought that it was the anilinium radical reacted with anilinium cations (protonated aniline but not the free aniline itself) to initiate the formation of ani­ line oligomers and polymers [60]. The proposed chain growth process is similar to that proposed by Mohilner, that is, the higher order of oligomers are produced via dimerization of lower order of oligomers. Their complete mechanism is shown in Scheme 2.7.

In 1994, Frank Lux reviewed the results of both chemical and physical properties of polyaniline from other research groups and his own studies, and proposed his own polymerization mechanism [61]. In his mechanism, he disputed the widely accepted proposition that the chain growth process is via dimerization processes of lower order oligomers, and believed that it was the coupling processes of anilinium radical with oxidized oligomers under the catalysis of hydrochloric acid (see Scheme 2.8), which is based on the fact that ( 1) the higher order of oligomers are easier to be oxidized than the aniline itself and (2) the oxidized form of p-aminodiphenyl amine (ADPA), PBQI (see Scheme 2.8), is susceptible to dimerization if it is in an acidic medium of a sufficient low pH, even without oxidizing agent (see his dimerization mechanism in Scheme 2.9). It is well known that the tendency of delocalization of protonated imine sequence will in increase with increased polymer molecular weight [62], therefore the fact that the molecular weight of PANI is limited, weather PANI is produced 44

:KlH NH,

<" ♦ NH. 6 0 ~ NHh" Ill

+B N H -< } NH, O- nh - < 0 > - nh -BH IV (/

Scheme 2.6: Formation of aniline dimers: p-aminodiphenylamine and benzidine. The mechanism was proposed by Breitenbach et al. After Breitenbach [57]. 45

NH CMij, ads 2, ads

- H ^ NH; a[Js ^ J - N H ^ i ,

pH < 1 dimerization O H i

H,N -hQ ^^>-nh, <0MH^ C M H‘ads

+ 2e - 2e

H,N NH, + 2 H €H NH2 solv + 2H [H] pH = 7 Dimerization

Tetramer (EB)

- 2 e * 2 d Q ^ m , - 5 H - 4 e -2H*

Tetramer (PNB) NH

Octormer (PNB) ■ N o - n e Aniline black

Scheme 2.7: Formation of aniline oligomers and polymers. The mechanism was pro­ posed by Dunsch. After Dunsch [60]. (a) (b) p-benzoquinone (BQ) benzidine

NH < O ~ nh - < 0 ^ nh-

(c) (d ) p-aminodiphenylamine (PADA) N-phenyl-1,4-benzoquinonediimine

NH

(e) (0 protonated N-phenyl-1,4-benzoquinonediimine anilinium cation

NH, NH.

(g) (h) anilinium radical cation anilinium dication

(i) anilinium radical

Scheme 2.8: Chemical intermediates and compounds used in the mechanism proposed by Frank Lux. After Lux [61] 47

(II <(^N-<^>-NH ------

<2) + <0 ^ N ^ ^ n h

Scheme 2.9: Dimerization mechanism of PBOI to form the blue imine of Willstatter. After Lux [61], from usual polymerization method or from the one with aniline dimer ADPA as the starting reagent [61], could be explained with his proposition. Frank Lux’s rather comprehensive mechanism, 12 steps in total, is shown in Scheme 2.10 to Scheme 2.12.

Lux’s mechanism focuses on how various anilinium cations and radicals, Will- statter’s blue and red imine, oligomers and subsequent polymer, and some byproducts such as benzidine and p-benzoquinone, are generated. His mechanism is special at how the growth of oligomers eventually leads to aniline black, the polymerization product. It is essential in his mechanism that oligomers have to be oxidized which then transforms into a radical form under the catalysis of hydrochloric acid before it 48

Step 1: Formation of the anilinium cation 1

+ HgO ------< ^ - N H , + HaO

(1.D (1. r)

Step 2: Formation of the anilinium radical cation 1

2 < © ^ NHa + o - to ' 2 ©-M h. - asor (1, I) (2. r)

Step 3: Formation of the anilinium radical 1

Step 4; Formation of the anilinium radical 2

CM* * ■

Step 5: Generationof benzidine via the anilinium radical dication 1 and its resonance structure, i.e. the anilinium radical dication 2

_ h _ .q h n h , ' Q - "."' \ , * \ s . (5, m) (5, r1 > (2, 0 (5. r1)

, + H e ° - H , N — ( ( ))-----( ( ) ) — NH., (5. r) - 2 H

Step 6: Generation of p-benzoqulnone from the anilinium cation

2 NH, + i - s o * ' 2 HSO* + 2 HO— NHa (6, m)

(1. r) H20 —------O—( )—O (6, r)

Step 7: Coupling of anilinium radicals 1 and 2 (7a) Generation of p-amlnodiphenylamine

< ^ > - nh + Q - nh ------

(3. r) (4, r) (7a, r)

Scheme 2.10: Polymerization reaction mechanism proposed for polyaniline by Frank Lux. After Lux [61]. 49

(7b) Generation of 1,2-diphenylhydrazine

© O (7b, r) NH ■ H,NNH, (3. r) (7b. m)

(7c) Generation of benzidine

H,N NH, O" (5, r2)

Step 8: Generation of N-phenyl-1,2-benzoquinone diimine from p-aminodiphenylamine

* £ £ j ^ — N— y ~NH + 2 HSO* (7a. r) (0. r)

Step 9: Growth of aniline oligomers and polymers

(9a) Growth via p-aminodiphenylamine and anilinium radical 2 . < ^ NH * 2 H S ° * (4. r) (9a, r1) (7a, r)

<9c, m) L J n+1 (da, r2) (9b) Growth via N-phenyl-1,4-benzoquinone diimine and the anilinium radical 2, catalysed by the acid “ H)C

O-NH HX X (4, r) (0, r) (9b, rl)I) O— (4, r) (9b, r2) (9b. r3) O-SO, oI -so < — <^>—N—<^>-N —<^>-N —( )—NH + 2 HSOa*

(9b, m)

Scheme 2.11: Polyaniline polymerization mechanism proposed by Lux continued (1). A fter Lux [61], 50

HX; (n-1) S 20 B —- < 0 ^ _< 0 ^ _'<© ^ nh 2n NH (9b, m) • < o (9b, r4)

Step 10: Generation ol Wlllstatter's blue Imine

(10a) Generation via N-phenyl-1,4-benzoquinone dllmlne

< ( ^ 5 ) — N— \ y ~ NH ■ ■ HX t n nh (8, r) C ^ H H O ^ ; C ^ ^ Z ^ (10b, m)n) ■ HX

(9b, m)

(10b) Generation via N-phenyl-1,4-benzoquinoe diimine and p-aminodiphenylamine H* -NH 0 ^ n_C ^ (7a, r) (8,( 8 , r)

- HX

(10b. m) (9b, r3) o-so; ° ~ SO ‘ < Q ) ~ N — 'N—< ^ ) ~ N —( )~NH + 2 HSO°

(9b, m)

Step 11: Generation of Wlllstatter’s red imine from Wlllstatter’s blue imine

o h q - o - h

(9b, m) (11. r)

Step 12: Generation ol p-benzoqulnone from Wlllstatter's red imine

O-SO, -NH + q _ s o * - H,0 <0 ^ N"^(Zyl— N—CZ)-r <11, r)

NO + O o + NH,' + 2 SO* + X*

(12. r) (6, r)

Scheme 2.12: Polyaniline polymerization mechanism proposed by Lux continued (2). A fter Lux [61]. 51 can be coupled with anilinium radical to yield a higher order oligomer. Repetition of this process eventually results in a polymer chain. By this route, the possibility of chain growth increases greatly as compared to the homo-coupling of two nth order oligomers into a Snth order of oligomer. This rationalization is supported by exper­ imental fact , i.e. the polymerization of aniline using p-aminodiphenylamine as the starting material [61]. However, the earlier stage of his polymerization mechanism involves too much speculations. Some species involved in mechanism are not detected physically. It is well known that weather a mechanism correct or not could not be judged from the fact that it can interpret some experimental facts since even a wrong presumption could coincidently lead to a plausible explanation of some experimental results. For example, benzidine can be formed in step 5, step (7b), and in step (7c) in Lux mechanism, therefore it can not be a supporting evidence that the polymer­ ization starts from anilinium radical dication( 2 , r), or from anilinium radical 1 (3 , r), or from its resonant state of anilinium radical 2 (4, r). A successful mechanism should be based on some direct evidence instead of on speculation referenced from some experimental results. The best way of gathering such evidence is from NMR or EPR experiment and possibly from their in situ CV and/or conductivity and/or optical experiments. These experiments could provide direct measure of the concen­ tration of radicals and their structural information. We will follow this idea to build our own models for the mechanistic studies of electrochemical oxidation processes of polyaniline and its derivatives. CHAPTER III

Experimental Techniques

3.1 Introduction

Synthesis of polyaniline by conventional oxidative polymerization has been reported since the 1910’s [63]. Only recently has a method which makes analytical pure emeral- dine been reported by MacDiarmid et al, Off course a great number of other new synthetic routes have now been reported for producing polyaniline and its deriva­ tives [65]. Meanwhile, we ourselves have developed some new synthetic methods to produce new forms of polyaniline and its derivatives. These synthetic methods as well as the characterization results of the polymers made with these methods will be discussed in detail later in this dissertation. Certainly the polyaniline and its deriv­ atives produced via these new routes need to be carefully characterized, since they may have a different physics and chemistry in many respects as compared to those made via conventional methods. In this chapter, we will focus on introducing some important characterization methods and techniques, i.e. the four probe technique for

DC conductivity measurement, NMR, EPR, and CV methods.

52 53

It is of benefit for us to discuss NMR technique first as a basis for other reso­ nance experiments such as EPR. NMR technology has had great impact in chemistry, biochemistry, and medical sciences, where NMR is used as an essential tool for deter­ mining the structures of very complicated molecules such as proteins. This is because NMR provides high resolution in diamagnetic liquids and a ‘safe’ 3D tomography. After a discussion of NMR we will introduce EPR technology by analogy.

3.2 The Four Probe Conductivity Measurement

The DC conductivity measurement is essential for characterizations of conducting materials. Recently, a method of measuring DC conductivity via the four probe technique has been developed in this group [125], The experimental setup contains a current source, a multimeter, a sample to be measured, and four gold wires composing a closed circuit (see the Fig. 3.1). These gold wires are attached onto the sample surface via “Silver Paint” or “Electric Dag.” The contact area of adhesive material to the sample surface should be as thin as possible to reduce relative errors associated with the measurement. In the design of the device shown in Fig. 3.1, the inner two wires are connected to a multimeter, e.g. a Keithley Model 181 multimeter, while the outer ones are connected to a current source, e.g. a Keithley Model 220 current source. Since the current source provides a constant current across the cross sectional area of the sample, the conductivity of the sample can be obtained from calculations based on the result of both the voltameter reading, and the current source reading in addition to the dimension parameters such as I, h, and w. The DC conductivity is 54

gold wires

sample strip

gold wires

Figure 3.1: Schematic diagram of the four probe technique for measuring DC conduc­ tivity on a sample strip. A current source provides a constant current to the sample strip. A multimeter measures the voltage drop across a segment of the sample strip of length /, width w, and height h. 55 then calculated based on the following equation:

a — l/(w x h x R) (3.1)

R = V /I (3.2) where k, I, and w in equation (3.1) are the height, the length, and the width of the sample strip, respectively. The parameters in equation (3.2) have their usual meaning. Usually / is ~ 0.5 cm long and w is ~ 0.2 to 0.4 cm wide, while h depends on the preparation of the sample pellet being measured, that is, it is equal to the sample thickness.

The four probe technique is used to reduce the possible contact resistance. Usu­ ally tests are done via alternating the polarities of the input current. If the voltages thus obtained are not equal, then the average voltage value is taken to calculate R.

Ohm’s law, equation (3.2), should be obeyed when the magnitude of the current is changed (in other words, R should be a constant).

Sometimes some materials may interact with “Silver Paint” or “Electric Dag,” then a “press-contact” method is needed instead of using adhesive to maintain the electrical contact. The “press-contact” method essentially requires some hydro-static pressure to build the electric contact directly between the sample surface and gold wires. The conductivity of the thin film of the fluorinated polyaniline sample was measured via this alternative method. 56

3.3 NMR Measurement

As has already been mentioned, modern NMR technology is the most powerful and versatile analytical tool for scientists working in various areas, such as chemistry, physics, pharmaceutical, medical science, ... etc. With today’s rapid advancement of computer technology and the development of complicated pulse sequences for NMR spectroscopy, most NMR spectrometers are now able to perform the powerful two dimensional and higher NMR experiments needed to resolve a huge range of structure and property determination problems. This certainly brings new excitement to the research field of polyaniline. There have been many reported 1-D and 2-D NMR studies on polyaniline and its derivatives so far [67, 68, 69, 70, 71, 72, 73, 74, 75, 76,

77, 78].

In this section, we will briefly introduce the principle of the NMR technique for the reader’s convenience in understanding the applications of NMR technique for structure determination of polyanilines in this dissertation.

3.3.1 Splitting of Energy Levels in a Magnetic Field

Just as an electron has a spin one half, some nuclei have non-zero nuclear spins. Therefore they will have magnetic moment and energy level splitting in a magnetic field. As a consequence, those with a non-zero magnetic moment will precess about the applied external magnetic field with its Larmor frequency u ?0 as

du 1. = wq x ft, (3-3) 57

where p and are the nuclear magnetic moment and Larmor precession frequency, respectively. They are defined as

P = 'frfhl, (3.4)

uj0 = 7JvH0, (3.5)

where I is the nuclear magnetic quantum number; 7^ is the magnetogyric ratio for a nucleus, a constant being defined as the ratio of the nuclear magnetic moment to nuclear spin and having an inverse dependence on its mass as shown below

7JV = , (3.6) zmjvc and it is correlated with quantity 0 as

7 Nh = g0, (3.7)

^ = 2mc" <3'8> In the above equations, c is the speed of light, m the mass of the nucleus, q the charge of the nucleus, and g the p-factor of the nucleus. From equation (3.6) and equation (3.7), we see that every nucleus has its own unique Larmor frequency depending upon its mass. Therefore their (NMR’s) excitation energy goes into radio frequency range as compared to the microwave of EPR’s. The energy levels of a nucleus with nuclear magnetic quantum number I in an applied field H 0 is represented as

E m, — —p ■ H0 (3.9)

= 'ypfhMjHo, M i = —I, —7 + 1,...,/— 1,/, (3.10) where Ho is the applied external magnetic field. For a nucleus of spin one half, the energy difference between the states of spin-up and spin-down is

A E = 7ivftHo, (3.11)

= 90 H0. (3.12) 58

Therefore the excitation frequency of a nucleus will be equal to its resonance Larmor frequency

vex = Vo = 7atH0, (3.13) being proportional to the external magnetic field strength. As a consequence, the resolution of NMR usually increases with increasing applied magnetic field strength because that non-overlapped spectrum width is usually fixed at the order of Hertz.

Fig. 3.2 shows the energy level splitting of a nucleus in the applied external magnetic field.

The energy separation between Mj = | and M/ = -| states, the A E as defined in equation (3.12), increases linearly with an increased magnetic field as shown in the diagram. Therefore the excitation frequency also varies linearly with the applied magnetic field strength. It should be noted that the energy term in equation (3.12) could not differentiate one particular nucleus in a different environment, which cer­ tainly contradicts the fact that the excitation frequency of one particular nucleus may be different in a different solvent. This is because the electron-nucleus interaction, the electron screening effect as it is usually referred to, is missing in equation (3.12) (We may recall that nuclei are always associated with surrounding electrons one way or another, which makes them unique from each other). The consequence of the ab­ sence of this interaction (i.e. no shielding effect exists) would be a disaster for NMR spectroscopy: every nucleus in different samples would have the same set of nuclear energy and therefore the same Larmor frequency, which would result in the NMR technique being of no use for chemical structure determination at all. Fortunately, for a given nucleus, the diamagnetic electron-shielding effect, or the chemical environ­ ment effect, does exist and is unique to a nucleus in certain environment. Only when this is taken into consideration does the application of NMR spectroscopy to structure 59

E

O

Figure 3.2: NMR energy level splitting. 60 determination become possible. The diamagnetic shielding effect of electrons on the nucleus will reduce the magnetic field strength the nucleus feels,

Hioc = (1 -

E — —^/Hloc, (3.15)

E — -

oj = 7jv(1 - cr)H0 = o;0(l -

Although the shielding constant itself is independent of the field strength, the Larmor frequency and therefore the excitation frequency, along with the contribution of the diamagnetic shielding effect from surrounding electrons to the nuclear Hamiltonian, —o^TAr^-LHo), is dependent upon the external magnetic field strength and therefore is machine-dependent. To perform a machine independent measurement, a general scheme of measuring the electron shielding effect is needed. Usually a reference mate­ rial, e.g. TMS, tetra-methyl silane, is needed to define the relative shielding constant, the chemical shift constant, as

6 =

u = 7jv(1 — o")H0, (3.19)

u>T = 7jv(1 - ov)H0. (3.20) 61

Therefore the chemical shift constant, a representation of the interaction of a nucleus with its chemical environment, being defined in terms of the difference between two machine-independent quantities, is machine-independent. In this way, the correlation of NMR peaks with their chemical shift constants is meaningful to the structural determination of the chemical compounds.

In addition to a diamagnetic shielding effect, which always exists in any molecule containing a cr bond, a paramagnetic shielding effect is also familiar to researchers. It is originated in materials such as acetylene and aromatic compounds containing delo­ calized 7r-electrons. The delocalization of these 7r-electrons will induce an anisotropic magnetic field over the nuclei in the vicinity. Therefore, depending on orientations, some nuclei may experience an increased local field while others may feel a reduced field, which can be quite useful for the collection of stereo-conformational information about complicated molecules.

3.3.2 Relaxation Related Phenomenon

In addition to the chemical shift effect, the concept of the relaxation of the perturbed

magnetization vector has its unique role in NMR technique. The calculation shows [79] that when a nucleus of spin one-half is in a magnetic field, the level of spin- up state has slightly more population than that of the spin-down state ((lV,pin_up —

N,pin-down)I{Nspin- up + N tpin^down) ~ 10-5 at room tem perature [80]). To effectively flip the spin state (i.e. to revert the spin population) the excitation pulse must meet the following two requirements: (1) its frequency must be the same as the Larmor frequency of the nucleus; ( 2 ) the electric oscillator must be in the same direction 62 as that of the applied external magnetic field (or equivalently, the magnetic field component of the electromagnetic wave of the perturbation must be perpendicular to the external field).

The classical analogy about the quantum representation of the excitation dis­ cussed above builds on the tilting of the precession of the nuclear spin vector. When an excitation pulse satisfies the two requirements mentioned above, its effect on the nuclear spin considered can be illustrated in the vector diagram shown in Fig. 3.3, where i, j, and k are the unit vectors for x, y, and z axes of the laboratory frame, respectively; z', j', and k! (same as k ) are those for x ’r y \ and z’axes of the rotating frame, respectively; the transformations of the unit vectors from the laboratory frame to the rotating frame are given in Fig. 3.3 (B); z and z’axes are coincident and both are aligned to be parallel to the applied external field; Ho is the static external field vector; Mo is the magnetization vector of the total spin in the sample studied (it is the sum of two vectors in the example shown in the diagram); u?o is the Larmor frequency of the nucleus under investigation, being parallel to z or z' axis. Fig. 3.3 (A) and (C) are the representations of two-spin systems in a laboratory coordinate system without and with the radio-frequency perturbation while Fig. 3.3 (B) and (D) are those in the corresponding rotating frame, respectively. In the laboratory frame, spins precess around the z axis (Ho direction), while in the rotating frame x’ and y' axes rotate around the z' axis, while spins don’t precess at all (here the effective total magnetic field strength becomes zero because of the relativistic effect of the rotating frame). Notice that a magnetic oscillator in x axis can be represented as two rotating vectors in x-y plane, shown in Fig. 3.3 (C). One rotates clockwise and the other anti-clockwise, with the same frequency as that of the Larmor frequency of the nuclei under consideration. Study shows [79] that only the one which rotates in the same direction as the Larmor frequency vector will flip (tilt) the spin significantly, so 63

/ ( k ) /.' ( k )

H„ = IH(|I k

A y Uj >)

j* = -fsin (tq,l) + i/'cos(cn>t)

A x ( i ) A x’ ( ( ’ ) ?*= ? cos (cq,t)+ *sin (tHiO

(A) (B)

(H)=yH

B = 2B|Cos(ftY]t) i

A A x ( / ) x '( i' )

(C) (13)

Figure 3.3: Classical representation of a perturbation (introduced by a radio frequency electromagnetic wave pulse) on the magnetization vector M0 in a static external magnetic field. 64 the contribution of the other to the tilting can be ignored. Therefore, the magnetic oscillator along the x axis in the laboratory frame, Fig. 3.3 (C), can be represented as a static magnetic field vector (EM') in the rotating frame, Fig. 3.3 (D). Notice that the field strength of the static magnetic field in Fig. 3.3 (D) is only one half of the magnitude of the oscillator in Fig. 3.3 (C). Since the magnitude of the magnetic field strength of the excitation pulse is much smaller compared to that of the super­ conducting magnet of NMR spectrometer, the Larmor frequency of the perturbation pulse is much smaller too.

When an excitation pulse is introduced to a system, the magnetization vector is tilted away from its equilibrium state (which is originally aligned with the applied external magnetic field) at an angle 8 that is proportional to both the field strength of the excitation pulse introduced and the duration of the pulse. When 8 is equal to

7T (i.e. the pulse tilts the magnetization vector to an angle of 7r and then is removed), the population of the spin system is reverted, i.e. the higher energy level has more spin than the lower one, which is not a thermal equilibrium state and will be relaxed back to a Boltzmann equilibrium state.

During the relaxation, the magnetization vector will pass by the x-y plane at some moment. If a detector (solenoid) is placed at the x or y axis in the laboratory frame, then an induced voltage will be detected so that the magnetization vector can be measured as a function of temperature. In addition, the relaxation itself can be measured from time dependent results. We can also see that the duration of a pulse determines the angle of tilt of the magnetization vector. Therefore a pulse can be assigned as the angle of tilt of the magnetization vector. Almost all the manipulations of a n-D (n = 1, 2, ...) NMR experiment are built on the variation of the pulse angle, the relaxation time, and other factors such as saturation time, etc. 65

Sample

Magnet

Magnetization

Perturbation

Response

Detection

Data

Fourier Transformation

Spectrum

Storage

Figure 3.4: Schematic diagram of preparation, acquisition, and processing of the NMR spectrum. After Sanders [80]. 66

3.3.3 Preparation, Acquisition, and Processing of the NMR S p ectru m

Figure 3.4 diagrammatically describes how a NMR spectrometer operates. After a sample (solid or liquid) is prepared, it is placed into a sample chamber usually located inside of a superconducting magnet surrounded by a liquid helium dewar. The sample is then magnetized and a sequence of radio frequency pulses of particular widths is introduced to perturb the equilibrium system. We know from the Heisenburg uncertainty principle that the shorter the width of the pulse in the time domain the larger the spread of the radio wave will be in the frequency domain. Therefore all kinds of protons in the sample could be excited simultaneously. The excited spins are then allowed to relax back to their equilibrium states. The relaxation of the magnetization vector then induces a voltage signal in the detector solenoid as a function of time. This analog signal in the time domain is digitized, recorded in the computer’s memory, and then Fourier transformed into a spectrum in the frequency domain. Such a spectrum can be plotted out or stored along with the digitized time domain signal in the computer’s hard disk so that it can be further manipulated. 67

3.4 EPR Measurement

3.4.1 Analogies between EPR and NMR Spectroscopes

An EPR experiment, as its name implies, measures the electron paramagnetic sus­ ceptibility of a sample at the resonance field. Therefore it is a useful tool to char­ acterize PANI and its derivatives, having a considerable amount of free spin (a high concentration of free radicals, or polarons). Recently a great deal of in situ EPR, DC conductivity, and cyclic voltammetry (CV) experiments have been per­ formed to study the electrochemical oxidation processes of the PANI family members

[85, 8 6, 15, 87, 8 8, 89, 90]. These have revealed new features of electron paramagnetic resonance as a function of the applied potential of a CV scan in a controlled manner. In this section, we will briefly introduce the basic principles and simple applications of EPR spectroscopy by making analogy with those already described in the NMR spectroscopy section. Detailed discussions can be found in an EPR textbook [92].

The energy of an electron spin in a magnetic field has a similar expression as that in equation (3.9) and equation (3.10) for a nucleus in an applied external field,

EMs = - f l H0 (3.21)

= 7 fiM 5 H0, M s = - S ,- S + 1,...,S - 1,5. (3.22)

The definitions are analogues to those for the corresponding quantities in the NMR technique. We will omit the subscript ‘e’ for 7 hereafter for clarity (it is the electronic magnetogyric ratio). The difference is that the magnetic moments for electrons de­ pend on the magnetogyric ratio of electrons and on the electron spin instead of those 68 corresponding values for the nucleus,

p. — jh S , (3.23) where

’ “ - s h - (3-24» a negative value opposite to that of the nucleus because the sign of the electron charge is opposite to that of the nucleus. In addition, the electron mass is usually on the order of several thousandth times smaller than that of nucleus so that the magnetogyric ratio for the electron is several thousand times bigger than those for nuclei. A similar relationship exists between the quantities of 7 and 0 for the electron as that for the nucleus,

7 h = -g 0 , (3.25) where 0 is defined as 0 = (3.26) 2mec v ’ and has a special name, Bohn magneton, after the name of the famous physicist. Again, the opposite signs in the corresponding equations arise from the opposite signs of the charges of electron and nucleus. The definitions of the quantities involved in the above equations are as follows: m e is the electron mass, and e the magnitude of the electron charge, g the (/-factor for the electron (for free electron, ge ~ 2.00232).

As has been already mentioned, 7 and 0 apply to both electron and nucleus and have different magnitude, therefore we see why the Larmor frequency for an electron is ~ m //m e times (usually a thousand times) larger than that of nucleus’, and the excitation energy for electron spin is thus in the micro-wave region as compared to that in the radio-wave region for nuclei. From this point of view, EPR is an analogue to NMR. 69

In addition, EPR has exactly the same energy dependence on the variation of the magnetic field strength as that of NMR, except with the opposite sign,

AE = (3.27)

- - g p Hr, (3.28) and the resonance of the excitation energy has the same form as equation (3.13) for the nucleus,

uiex = ur — —7 Hr. (3.29)

Therefore the g-value for the electron spin can be experimentally determined as hv 9 = m - (3.30)

3.4.2 EPR Line Shape

The EPR line shape is determined by factors such as the lifetime of the electron spin state and the uncertainty principles. The line shape is usually characterized by comparing it with a Lorentzian or Gaussian type of line shape. In terms of measurable experimental parameters, the analytical expressions for these two line-shapes can be described as follows:

r 2 Lorentzian Y* —= Y* m (3.31) r 2 + (h - H r)2 ’ (—/n2)(H - H r)2 Gaussian Y■* = -*Y TTl cexp (3.32) T2 where for a Lorentzian line shape, Y- = ^ while for a Gaussian line shape, Ym

( W iO - 70

3.4.3 The Spin Exchange Phenomenon

Exchange broadening and exchange narrowing are related phenomena of exchange of spins in their effect on the spectrum line width and line position. An electron spin can exist in many distinct forms because of the inherent chemical processes related to the molecule where the spins reside. For example, electron spins in polyaniline can exist at nitrogen sites or at carbon sites because the electron spins can resonate between these different sites. When spins are localized the lifetime r is long enough that the line can be assigned to a specific species. However, when the exchange rate increases, the lifetime of individual sites is shortened and the uncertainty of the energy of the spin state increases, which results in a broadened line width T. For an electron spin which has two distinct spin states, A and B, the line width T becomes

F = r 0 4- (3.33) where F and To are the line width in Gauss in the presence and absence of intercon­ version, respectively; 7e is the electronic magnetogyric ratio; and r is defined as the mean lifetime of species A and B, viz.,

t = TATB- ■ (3.34) t a + tb

When the exchange rate becomes equal to the difference of the resonance frequencies of the distinct species, the lines are further broadened and begin to shift towards the weighted midpoint (see definition below) of their field positions. The shift can be correlated to the lifetime as

((*H0)2 - (£He)2)^ = — , (3.35) 7eT where 5Ho is the line separation in the absence of the interconversion, and 6He is the line separation when the conversion is taking place. Coalescence of the peaks of 71 different species will eventually occur if the interconversion rate increases to a value big enough, which can be readily explained by the uncertainty principle. If the lifetime is smaller than St, during which the different lines can be distinguished, SiSv ~ 1/2, only one line is observed. For a very fast interconversion, the line width is given by

r ~ r 0 + ‘1'T P A ‘PB < (£H0)2 > (3.36) and the line position converges to the weighted mean

Ha + Pb < H > = (3 .37) Pa + Pb

3.4.4 Pressure Broadening

The discussion of pressure broadening is parallel to that already given in the chemical exchange related broadening section. Therefore here in this section the pressure (concentration) related line broadening phenomenon will be introduced very briefly. Interested readers should refer to aore detailed discussion in a fundamental EPR textbook [92].

In pressure broadening the exchange rate is proportional to the concentration of free radicals. The second-order rate constant can be obtained as

1 (3.38) 2 2t [R] ’ where [R] is the free radical concentration.

When the solution of free radicals is dilute, the hyperfine peaks are well resolved. When the concentration increases, the lines begin to broaden and at certain concen­ tration the rate approaches to the field difference between the different species, and 72 the line positions begin to shift. When the concentration is high enough, the lines coalesce to a single line, which becomes narrower at even higher concentrations. From this concentration on, the electron spins exchange so fast that the time averaged hy- perfine field is nearly zero. It is said that the EPR line is exchange-narrowed. It is well known that most of the pure solid spectra of free radicals are exchange-narrowed. The rationale for this is that the concentration of free radicals is so high that the ex­ change is very fast because of the greater overlap of the free radicals’ molecular wave functions of (or the greater bandwidth of the free radical band, e.g. the polaron defect band in emeraldine salt).

3.5 Cyclic Voltammetry Analysis

Cyclic voltammetry (CV) is an i — E characteristic of the reversal technique for a controlled potential sweep process. CV has now become a very popular characteriza­ tion method for chemists working on the synthesis of polyaniline and its derivatives [93, 94, 95, 96, 97, 98, 99, 100, 101, 102). It has been proven to be very useful in obtaining information from fairly complicated electrode processes. In recent years, CV has been used as a vital component in the in situ CV, EPR, DC conductivity, and optical studies[91, 85, 15, 87, 88, 89, 90], In a later chapter of this dissertation, we will present the Quasi Random Oxidation Model and the associated computer simulation results based on the experimental data of the reported in situ CV, DC conductivity, and EPR studies. Therefore a brief introduction of CV technique is important for understanding the later illustrations. A more comprehensive discussion can be found in a standard electrochemistry book [139]. 73

3.5.1 CV Experiment Setup

An experimental setup for cyclic voltammetry is shown in Fig. 3.5. In this particular setup, a function generator (HB series), a logarithmic converter (HG series), and an X-Y recorder are combined with a potentiostat/galvanostat (HA series) [103]. The proper wire connections are clearly shown in this graph. Note that, however, the logarithmic converter is not used in most of our CV experiments. The input cable is routed out to a three-electrode cell (WE2 has a very high input impedance and is used to reduce the ohmic loss in the connection wires). The block-diagram of this “three-electrode cell” setup is shown in Fig. 3.6. In the block-diagram, the working

(or indicator) electrode is the electrode system of interest, which is coupled with an electrode, such as AgCl/Ag or SCE electrode, that approaches an ideal nonpolarizable electrode of known potential, usually defined as the reference electrode. The current passes between the working electrode and the auxiliary (or counter) electrode. In order to avoid any interfering reactions on the working-electrode surface caused by the side reaction on the counter electrode, the counter electrode is frequently placed in a compartment separated from the working-electrode compartment by a sintered-glass disk or other separator. Therefore, in this case the choice of an auxiliary electrode is rather flexible, since its electrochemical properties do not affect the behavior of the working electrode. The three-electrode cell arrangement is used preferably when the ohmic potential drop in solution, iR ,, is high (e.g., in the cases of a nonaqueous solution or insulating film system with low conductivities). A well designed three- electrode cell setup is shown in Fig. 3.7. Notice that the auxiliary and the reference electrodes are separated from the main compartment (electrode compartment) by a 74

Inunction G »n«rator l.ogari tlunlc C o n v e rto r (H G l0 5 ,e tc .)

XCfoTCMTlAL}

PATUtlAl W U N T

X-Y liacordsr

i’o ten tio a to t RE (lU -s«rlm ) WEfV.CB WE2

J

Figure 3.5: Typical experimental setup for cyclic voltammetry. 75

Power supply

Working Auxiliary electrode electrode

Ewk vs. ref

Reference electrode

In cell notation Working or ? Indicator Reference

Auxiliary or T counter electrode

Figure 3.6: Block-diagram for a “three-electrode cell” setup. 76

Vacuum

1 2 /3 0

9 /2

2 9 /2 6

Reference Auxiliary electrode electrode

Solution level

Medium frit

Stirring bar

Figure 3.7: An example of three-electrode cell glassware. 77

medium frit (a sintered-glass disk) to prevent possible side effects on the working electrode. The setup has provisions for stirring and evacuating the solution system. The working electrode shown is a platinum-disk electrode, but a platinum-plate elec­ trode could be used instead. This cell setup is suitable for studying of a nonaqueous electrolyte as well as for an aqueous one.

3.5.2 General Pathways of an Electrode Reaction

Just like a heterogeneous catalytic reaction on a catalyst surface, an electrochemical reaction on an electrode surface can be very complicated (see Fig. 3.8). For a general electrode reaction 0 + ne ^ R, it may involve several possible processes:

• Mass transfer reaction (e.g., R transfers from the bulk solution to the electrode surface region).

• Electron transfer at the electrode surface.

• Chemical reactions preceding or following the electron transfer. For instance, protonation, or dimerization, or catalytic decomposition on the electrode sur­ face.

• Other surface reactions, such as adsorption, desorption, or crystallization (elec­ trodeposition).

In the case of the electrochemical synthesis of polyaniline, the processes are asymmetric to the polymerization reaction on the electrode surface as compared to those shown in Fig. 3.8, because it involves the electrodeposition (polymerization) of 78

Electrode surface region Bulk solution region Electrode

Chemical Mass trans fer reactions

«i VVVV VJ y-% ► tlsurf Obulk

ne Electron trans fer

R ads

d.- Chemical reactions

W * VV VT *▼ to R surf w w s ^ ^bulk

Pathway of a general electrode reaction

Figure 3.8: The pathways of a general electrode reaction. 79 aniline monomers. The processes can be described as follows: aniline monomers are diffused from a bulk solution into the electrode surface region (mass transfer), which may experience protonation either in the electrode surface region (chemical reaction) or in bulk solution region. Then they are adsorbed onto the electrode surface where they are oxidized to their cation-radical form (adsorption and electron transfer). The aniline monomers in cation radical form may be polymerized (electrodeposition) on the electrode surface, or desorbed into the electrode surface region (desorption) where further side reaction may occur (chemical reaction). The desorbed oxidized monomers and all of their byproducts are diffused into the bulk solution region (mass transfer).

The electrochemical redox reaction of deposited polymer film on the working electrode appears to be less complicated than the electrochemical synthesis process (assuming that the polymer film is stable, i.e. not subject to degradation or hydrolysis reaction) since it is stationary. However, the electron transfer process on the electrode surface area involves protonation and deprotonation reactions, which require diffusion and other processes of protons and their counterions. The processes of protons and their counterions can also be described by the general processes shown in Fig. 3.8. Therefore the above discussion about the electrode reactions applies to both the electrochemical synthesis and the electrochemical redox processes.

3.5.3 Cyclic Voltammetry and Reversal Techniques

As we have mentioned, CV is an * — E representation of a reversal technique combined with linear sweep voltammetry. The other representation of the same technique is the i — t characteristic, recorded on a strip-chart recorder instead of X — Y recorder 80

Faradaic A + Forward h baseline .--'scan for reverse scan

0 Switching time, X Reverse scan A'-f-* A

Figure 3.9: Cyclic voltammetry: (a) The cyclic potential sweep; (b) the resultant cyclic voltammogram.

used for recording the i — E characteristic. Fig. 3.9 shows the most commonly used potential sweep input signal and the resulting cyclic voltammogram.

Fig. 3.9 (a) shows that the repeat cycle of the potential sweep is simply a trian­ gular wave, the analytical form of which is shown below:

(0 < t < A) E = E i - v t ; (3.39)

(t> X) E = Ei- 2At + vt. (3.40)

Fig. 3.9 (b) shows the resultant CV spectrum, where the forward potential sweep starts at Ei (corresponding to the zero time in Fig. 3.9 (a)) and passes through E°\ the standard reduction potential, where the A + e —» A' process starts. While the 81

reverse scan invokes the oxidation process, A' — e —> A, it is a reverse of the forward process. These two scans are switched at E\ corresponding to the switching time A in Fig. 3.9 (a). Note that the measure of the peak intensity of the reverse scan is from the Faradaic baseline indicated in Fig. 3.11 and Fig. 3.9, not from the horizontal potential axis as might be mistakenly done [139]. Note that the standard reduction

potential, E 0' or £ 1/2, is not equal to the peak potential, Ep (Fig. 3.10), which is the consequence of the diffusion control and the depletion of the “O” species in the electrode surface region. The relationship among the standard reduction potential E 0' (equivalent to the half-wave potential E1/2), the peak potential Ep, and the half­ peak potential Ep/2 is shown in Fig. 3.10. The analytical relation among these three is shown as follows:

RT Bp/2 ~ /2 + 1.09—— — E \/2 T 28.0/n mV at 25°C; (3.41) n r RT |Ep - Ep/2I = 2-2^ = 56-5/™ mV at 25°c- (3-42)

The other most frequently encountered parameters in the CV experiment are shown in Fig. 3.11. For a nernstian system with E 0' — Ex > 35 m V, Ex in Fig. 3.9 is equivalent to E 1 in Fig. 3.11. Two important relationships can be extracted from

Fig. 3.11: (1) ipajipci the ratio of the anodic and cathodic peak currents is unity regardless of the scan rate v , the reversal potential E\ , and the diffusion coefficients; (2) the difference between Ep,, and Epc (A E) is always close to 2.3 R T jn F (59/n mV at 25°C). Therefore these two relations can be used as a diagnostic test for a nernstian system. It can be seen that the anodic peak current ipa is measured from the decaying cathodic current as the baseline. Current (arbitrary unit) Figure 3.10: T he relationship among CV param eters eters param CV among relationship he T 3.10: Figure + + 100 ( - 12 / ) mV (E 1/2 E - n El/2 0 EXj2, Ep, -100 and and E /2. p/ 82 Figure 3.11: The most frequently encountered param eters in CV CV in eters param encountered frequently most The 3.11: Figure Current function 2 2 4 3 1 1 0 3 4 c p a p a p c p 1 - -200

-300 i —i E -400 characteristic. 83 84

3.5.4 Multistep electrochemical processes

The electrode processes of a multicomponent system involving multiple-step charge transfers are illustrated in Fig. 3.12. The two consecutive reductions (non-stepwise) of two substances 0 and O' are shown below:

O + ne R; (3.43)

O' + n e -» R'. (3.44)

Fig. 3.8 shows that O and O', originally in the bulk region, diffuse into the electrode surface region and finally onto the electrode surface. If the diffusions of O and O' occur independently, then their current flux are additive, and therefore the i — E curve (curve 3) of the system is the sum of the two individual i — E curves of O and O' (curves 1 and 2). It is important to note that the second curve builds up on the root (the decaying current of the first one) of the first one (usually decays as t-1/2). There are several ways to establish the baseline for the second wave to allow a reliable measurement of its peak current. The most commonly used method is to stop the scan at the peak potential of the first wave, allowing the current to decay to a small value and then continuing to scan and measure the second peak current from the potential axis as the baseline of the second one [139].

An important application of the stepwise reduction of a substance 0 is the step­ wise redox reaction:

O + 7iie —> Ri {E°); (3.45)

R\ + 712 e —y Rz [E2 )• (3.46)

The shape of the i — E curve depends on A E° (= E \ — E °)y the reversibility of each step, and rii and 713. Detailed discussion can be found elsewhere [139]. 85

1

100 -100 200 -300 -400 E / mV

Figure 3.12: A multi-component system and a multi-step charge transfer reaction with (1) n = n', (2) the same surface concentration, and (3) the same diffusion coefficient for both O and O': (1) O alone; (2) O' alone; and (3) the mixture of O and O' 86

There is an important difference among the cases discussed above and the case of the redox reaction of polyaniline film. For the latter case R is the benzenoid ring while 0 is the quinoid ring unit. The second electrode reaction is the same as the first one except that the Coulomb repulsion in the second reaction is higher than that of the first one. We will discuss the redox reaction of polyaniline film again, based on our own quasi-random oxidation model, later in chapter IV. C H A PT E R IV

Synthesis and Physical Properties of Highly Sulfonated Polyaniline

4.1 Abstract

Sulfonated polyaniline (SPAN) is a self-doped conducting polymer. It has a high water solubility and a novel pH-dependent DC conductivity that is of interest for fundamental science and also for applications in such areas as rechargeable battery and pH control technologies. We report here the extensive characterization and details of synthesis of a new form of sulfonated polyaniline (LEB-SPAN) which shows novel or significantly improved chemical and physical properties. LEB-SPAN has a much high sulfur to nitrogen ratio (S/N) of ~ 0.75, 50 % larger than that previously reported for

EB-SPAN, S/N ~ 0.50. This change in composition leads to significant alteration of the properties including an order of magnitude increase in the room temperature DC conductivity to ~ 1 S-cm_1, nearly double the solubility in water, and a completely different pH-dependence of the oxidation potential (E i / 2)- For LEB-SPAN the DC conductivity is unaffected by pH over the range 0 < pH < 14, strikingly different from

87 88 the behavior of both parent polyaniline and EB-SPAN which become insulating for pH > 3 and pH > 7.5, respectively. Temperature-dependent DC conductivity and EPR measurements for LEB-SPAN reveal a lower activation energy for the conductivity and a higher density of states at the Fermi energy level as compared with those of EB-SPAN. The dramatic differences in the pH-dependence of the DC conductivity, cyclic voltammetry (CV), FTIR, and UV-Vis results for LEB-SPAN and EB-SPAN are shown to be a consequence of the much higher S/N ratio in LEB-SPAN.

We propose and describe a novel quasi-random oxidation model for the electro­ chemical oxidation of polyaniline and its derivatives at the microscopic level. This model quantitatively describes many of the phenomena and physical properties found in the polyanilines including the origin of the defect states, the asymmetric voltage dependent in situ DC conductivity peak, and the variation of the in situ EPR sig­ nal during CV potential scans. Also the statistical nature of this model suggests its general applicability to the oxidation processes of other conducting polymers. Com­ puter simulations based on this model are presented and show good agreement with the in situ CV/EPR data as well as the in situ CV/DC conductivity data reported earlier. From the simulation result, the evidence of the polaron lattice formation and the proportionality of the experimental DC conductivity data to the simulated Pauli spin density were revealed. In addition other models are proposed to interpret the reported experimental differences in the pH-dependence of Ex/2 among LEB-SPAN, EB-SPAN, and its parent polyaniline samples. Mechanisms for the new sulfonation route are proposed. 89

4.2 Introduction

Sulfonated polyaniline (SPAN, the chemical structure is in Scheme 4.1) is of interest because of its unusual electroactive physical properties, improved processability, and potential industrial applications [105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115]. SPAN is the first reported self-doped water soluble conducting polyaniline derivative and a prime model for dopant and secondary dopant induced processability [116, 117] in addition to self-doping [118]. It has been shown that SPAN has better thermal stability than its parent polyaniline doped with HC1 [108]. It has been found that SPAN has potential use in rechargeable batteries with a higher charge density [109, 110] as compared to that obtainable utilizing the parent polyaniline [119, 120]. It has been reported that SPAN was used in fabricating a multilayer heterostructure light emitting diode devices [111]. SPAN also has potential for application in the electrochemical control of electrolyte acidity and enzyme activity [112]. In addition, SPAN has been proposed for use in patterning by coating a SPAN containing resist on a wafer [114].

A key control factor for the electroactive phenomena, processing, and potential applications is the degree of sulfonation, that is, the sulphur to nitrogen (S/N) ratio. A number of different synthetic routes earlier have been developed to achieve significant S/N ratios. Emeraldine base (EB) and pernigraniline base (PNB) forms of polyaniline (the chemical structures are in Scheme 4.1) have been used as starting materials for the preparation of SPAN (defined here as EB-SPAN and PNB-SPAN, respectively). Also fuming sulphuric acid, chlorosulfonic acid, and sulfur trioxide/triethyl phosphate complex have been reported as sulfonation agents in the synthesis of SPAN [107]. However, all of these earlier methods resulted in a maximum S/N ratio of 0.5. 90

The origins of some of the electroactive phenomena related to SPAN and parent polyaniline until now have remained unresolved. For example, the earlier reported

[106] strong pH dependence of SPAN’s first oxidation potential, Ei/ 2pi, differs from the pH independent behavor of E1/Z)1 of the parent polyaniline [121, 91, 85, 86]. Also, cyclic voltammetry (CV) experiments and the correlated in situ electron paramagnetic resonance (EPR) as well as the in situ voltage dependent DC conductivity studies have been performed [91, 85, 15, 87, 88, 89, 90], however the variation of the in situ EPR intensity with the potential of the CV scan have until now remained unexplained and the confusions about conducting mechanism originated from the result of in situ CV/DC conductivity data [90] still needs to be resolved. In addition, the origin of the asymmetric voltage dependent DC conductivity curve [86, 88, 90, 122] is waiting to be fully accounted for.

We report the detailed synthesis and characterization of very highly sulfonated polyaniline, using the most reduced form of polyaniline, leucoemeraldine base (LEB), as the starting material (the final product is therefore termed LEB-SPAN). The LEB- SPAN has a higher S/N ratio, an order of magnitude greater DC conductivity, a dif­ ferent pH dependent oxidation potential, and other novel properties as compared with the parent polyaniline, EB-SPAN, and PNB-SPAN. A comparison of the results of FTIR, UV-Vis, pH dependence of CV and DC conductivity, temperature dependence of EPR and DC conductivity, and elemental chemical analysis of LEB-SPAN with earlier results for parent polyaniline, EB-SPAN, and PNB-SPAN is given, showing important ramifications of the higher S/N ratio for LEB-SPAN. A novel quasi-random oxidation model and mechanisms are proposed to account for electroactive phenomena of the polyanilines. The simulation result for the in situ CV, EPR, and DC conductiv­ ity spectra are presented to fully interpret the in situ experimental result and resolve the confusions in conducting mechanisms. The different pH-dependent E i/2 of LEB- 91

(1 a)

H I -N- N- (1b)

(1c)

SO

(Id)

Scheme 4.1: Chemical structures of LEB, EB, PNB, and SPAN polymers. 92

SPAN, EB-SPAN, and parent polyaniline (PANI) are quantitatively interpreted based on the quasi-random oxidation model.

4.3 Experimental

4.3.1 Chemical Synthesis

In this section, the synthetic route for LEB-SPAN is outlined briefly; a schematic diagram is given first in Scheme 4.2. The detailed description of this scheme can be found elsewhere [123].

The common procedure is as follows: ~ 0.5 g of pre-prepared EB (EB prepared via the standard method [124]) was placed in a glass mortar. Then 2.5 ml of phenyl hydrazine was added and the mixture was pressed with a glass pestle for 5 minutes followed by stirring for one hour to allow EB to be reduced to LEB. Meanwhile, 10 ml of fuming sulfuric acid for a later step was pre-cooled to ~ 5° C. After the reduction reaction was completed, the mixture was diluted with 75 ml ethyl ether, stirred for 15 minutes, then filtered and washed with three portions of 50 ml of ethyl ether, and suction dried.

The dried LEB was then sulfonated in the 10 ml of pre-cooled fuming sulfuric acid for one hour. The reaction mixture was subsequently introduced into 0.75 L of 75:25 ice-water mixture to cause the SPAN product to precipitate out. It then was washed with three portions of 250 ml of cold water. SPAN powder was dried with a common procedure [124] in a vacuum oven at room temperature. The dried product reduced in phenyl hydrazine

washed with ethyl ether

LEB

H2 SO 4

LEB SPAN

■SO 3,

Scheme 4.2: The synthetic route for LEB-SPAN. 94 was weighed and the yield was calculated to be ~ 70%.

Samples used in the study of pH-dependence of conductivity were prepared by dissolving LEB-SPAN powder in pH = 1 through pH = 12 buffers (50 mg/20 ml) (made in The Ohio State University Reagent Laboratory on Oct. 10, 1994). The resulting solutions were stirred for one day and then pH were measured with a pH meter. Subsequently, these solutions were casted onto glass plates to form films which were dried overnight. Then these films were peeled off the glass substrate and further dried in a vacuum oven at 40° C for two days. Subsequently the thin films were pressed into pellets for four probe conductivity measurements.

4.3.2 Characterization Methods

Elemental chemical analyses (MHW Lab, Arizona) of a typical preparation of SPAN m ade via the above method yield a S/N ratio of ~ 3/4. Both the room temperature value and the temperature dependence of the DC conductivity of pressed powder pellets were carried out via a four probe technique [125]. Electron paramagnetic resonance experiments utilized a Bruker ESP 300 spectrometer. UV-Vis spectra for the base form of LEB-SPAN were taken in 0.1 M aqueous NH4OH solvent with a Perkin Elmer A-19 spectrometer. FTIR spectra of LEB-SPAN powder (in KBr pellet) were acquired with a Mattson CYGNUS 100 spectrometer. Cyclic voltammograms were obtained with Hokto Corp. (HC) potentiostat/galvanostat (Model HA-301) combined with a HC function generator (Model HB-111). All of the studies of LEB- SPAN presented here were performed on samples from the same synthetic batch. Small variations of the physical properties occur among different batches of sample, 95 presumably reflecting small variations in the degree of sulfonation.

4.4 Results and Discussion

4.4.1 The Mechanism and Rationale for LEB Route

The chemical structures of the various oxidation states of base form parent polyaniline and that of SPAN are shown in Scheme 4.1, where x and y are the degree of poly­ merization and the sulfonation level (equivalent to S/N ratio) of SPAN, respectively. The same definition is used in Scheme 4.2.

LEB is the most reduced form and PNB is the most oxidized form of the polyani­ line family members while EB is intermediate. Based on the characterization results, a plausible sulfonation mechanism is proposed in Scheme 4.3 below. For simplicity, we will only write out the sulfonation route for the unprotonated repeat unit while omitting the similar route for the protonated repeat unit.

In Scheme 4.3 (I), two molecules of sulfuric acid interact to yield a solvated proton, an anion of sulfuric acid, and a molecule of sulfur trioxide. In Scheme 4.3

(II), one aniline unit is oxidized by a SO 3 under catalysis of a proton to yield an intermediate, (a). In Scheme 4.3 (HI), this intermediate is rearranged into its more stable counterpart, (b), the sulfonated leucoemeraldine base. In Scheme 4.3 (IV), one repeat unit (in the form of (b)) is oxidized by a strong oxidant, fuming sulfuric acid, to its quinoid counterpart, (c). In Scheme 4.3 (V), the quinoid form (c) is subsequently protonated either by the sulfonic acid attached to the backbone or the protons from 96 solution to yield a bipolaron, (d). This bipolaron is then relaxed into its semi-quinoid counterpart, the polarons, (e) in Scheme 4.3 (VI), which can subsequently separate further. Notice that we have delibrately used a different notation in (e) for the Cg ring to emphasize the electronic structure difference between (e) and (d). In (d) the

7T—electrons are paired up, i.e., the originally six equivalent carbon-carbon bonds of the unperturbed benzene ring dimerize into single and double bonds so that there will be no unpaired spins exist in the resultant bipolaron structure. However, in structure (e), the spin localized schematically on the nitrogen atom is unpaired and it will have an EPR signal.

The sulfonation of the phenyl rings is an electrophilic substitution reaction, there­ fore higher electron density on the phenyl ring will result in higher reaction rate, and as a consequence, a higher S/N ratio and higher yield. When EB or PNB is mixed with fuming sulfuric acid, nitrogen atoms at quinoid sites are protonated, and the positive charges effectively delocalize into the quinoid ring units due to conjugation of the pa-orbital at the nitrogen site with the 7r-orbital within the Cg ring (therefore pos­ itive charges can resonate into protonated quinoid ring units). The protonated imine repeat units are therefore essentially deactivated for the subsequent electrophilic aro­ matic substitution reaction, i.e., the sulfonation reaction, which may be the origin of the typical S/N ratio for the other routes being 0.5. When the amine repeat units are protonated in sulfuric acid, the newly formed H-N bonds are of sp3 type so that the positive charges are mainly localized at the nitrogen sites due to reduced conjuga­ tion between 7r-orbitals of phenyl ring and (r-orbitals of the amine cations (resonance of positive charges into benzene rings are therefore excluded). The electron density within the benzene rings of LEB in fuming sulfuric acid is higher so that electrophilic substitution on the ring occurs more easily compared to those of EB and PNB. In addition to the resonance effect already considered, the field effect of positive charges 97

(I) 2 H . S 0 4 Hj O+ + HS04 + SOj

SO

H\cat) (n ) (a) \

(m) (a) N

s o 3h

Hj S04 (S03) \ (1/) (b) . N N C S 0 2, +I2 O

(v) (C) N (d)

(vi) (d)

Scheme 4.3: The proposed sulfonation mechanism for LEB-SPAN. 98 at nitrogen sites also results in a higher positive charge density on the carbon atoms adjacent to the protonated inline repeat units in EB or PNB because of shorter bond length of sp2 orbital compared to that of sp3 orbital, therefore the protonated imine unit is subject to hydrolysis to a greater extent than that of amine unit, which will result in shorter polymer chain length for EB-SPAN and PNB-SPAN. Hence it is not difficult to rationalize the higher S/N ratio and higher conductivity of typical LEB-SPAN than those of EB-SPAN or PNB-SPAN. Yue et al observed that the sul­ fonation level in EB-SPAN was higher than that of PNB-SPAN [107], which supports our assumptions that the greater the electron deficiency on the ring unit, the lower the S/N ratio of the sulfonated product, and the lower the conductivity.

Oxidation of amine atoms and sulfonation of phenyl rings could be thought as two competitive reactions for the current route. However the higher S/N ratio of

LEB-SPAN eliminates the possibility for the current route that the oxidation occurs before the sulfonation reaction (if the oxidation occurred first in the current route then S/N would be the same as that for the previous EB sulfonation route). The proposal of sulfonation occurring before the oxidation reaction in Scheme 4.3 is based on this experimental implication. The rationale follows. Protonation of amine atoms in the concentrated sulfuric acid has effect of protecting the amine atoms from being oxidized [126]. However, after the aniline repeat units are sulfonated the protonation equilibrium shifts to the deprotonated side somewhat because of (1) the presence of oxygen atoms or anions of the sulfonic acid group in the vicinity of the protonated amine atoms (lone electron pairs on oxygen are competitors for protons) and (2) the electron withdrawing (EW) nature of sulfonic acid (which makes the amine a even weaker base). Therefore the oxidation of the amine atoms more likely occurs after the sulfonation of the adjacent phenyl ring. As a consequence, the oxidation state of the LEB-SPAN can be either lower or higher than emeraldine salt oxidation state 99

depending on sulfonation level, the reaction time, and other reaction parameters. Recent XPS study of LEB-SPAN [127] revealed that the reduced ring unit can vary from much below 50 % (only ~ 25 % of amine moiety is present) to significantly more than 50 % (about ~ 65 % of amine moiety has been observed). Efforts to elucidate this complicated variation of the oxidation state of LEB-SPAN are in progress.

4.4.2 Vibrational Spectra

The FTIR spectra of both LEB-SPAN and EB-SPAN are compared in Fig. 4.1. The number of peaks and the peak positions for LEB-SPAN are essentially the same as those for EB-SPAN while the relative intensities of some peaks vary appreciably. The bands at 1070 and 1040 cm"1 are usually assigned as aryl-S linkage (aromatic ring vibration having some C-S stretching characteristic); they overlap a broad intense absorption band in the region of ~ 850 cm"1 to ~ 1200cm"1. For the sake of identi­ fication of this intense peak, FTIR spectra of emeraldine base (EB-I, solid line) and emeraldine hydrochloric salt (ES-I, dashed line) are compared in Fig. 4.2. It is evident that this broad band is a common feature of the emeraldine salt and not a signature for aryl-S linkage of SPAN sample. After subtracting an assumed smoothly varying broad band (850 cm"1 to 1200 cm"1) for the LEB-SPAN spectrum, the ratio of the areas of the two bands at 1070 and at 1040 cm"1 to that of the sum of the two bands at 1,500 and 1,600 cm"1 (stretches within the Ce rings) is considerably larger than

that for EB-SPAN. In addition, the peak at 610 cm"1, the C-S stretching vibrational band [107, 128] is also more intense for LEB-SPAN. Therefore two conclusions may be drawn: First, the same number of peaks and essentially the same peak positions indicate that SPAN has been made via the LEB method. Secondly, the FTIR result rmrf [107]. ref. from self doped form of EB-SPAN in KBr pellet. The spectrum of EB-SPAN was adapted adapted was EB-SPAN of spectrum The pellet. KBr in EB-SPAN of form doped self Figure 4.1: FTIR spectra: spectra: FTIR 4.1: Figure Absorbance (arbitrary unit) 1800 X -

1600

Wavenumbers, Wavenumbers, ------1400 sl oe fr o E-PN n B pelet , t; lle e p KBr in LEB-SPAN of form doped self ,

1200

1000 cnrr

1 800

600

400 100 101 is consistent with that of the element chemical analysis showing higher S/N ratio for LEB-SPAN.

4.4.3 Electronic Spectra

It is seen (Fig. 4.3) that the UV-Vis absorption bands of a dilute solution (0.3 mg/ml) of LEB-SPAN in aqueous 0.1 M NH 4OH are blue shifted relative to those of EB-SPAN in similar solution (312 nm vs. 320 nm for the benzenoid it — 7r* transition [129, 130] and 541 nm vs. 566 nm for the exciton band transition [131] (compared to 339 nm and 635 nm respectively, for emeraldine base). This might be understood as the result of additional sulfonic acid groups, a very strong electron withdrawing group, substituted on benzene rings, increasing the absorption band gap. Higher sulfonation level also further increases the torsional angle between adjacent rings to reduce the larger steric strain [106, 129, 130], which also will decrease the intrachain interaction leading to a larger band gap. Therefore, the UV-Vis results also are supporting evidence for higher sulfonation level of the LEB-SPAN. One potential argument for the blue-shift is that the spectrum is influenced in some manner by the oxidation that occurs in the course of sulfonation. However as the reactant, LEB, is in a more reduced form and the present route utilizes a much smaller ratio of fuming sulfuric acid to the phenyl ring unit, the interference by increased oxidation is unlikely. In fact, approximately the same ratio of benzenoid to quinoid band intensity was observed in the FTIR spectra of self doped EB-SPAN and LEB-SPAN (Fig. 4.1). curve is that of ES-I in KBr pellet. KBr in ES-I of that is curve Figure 4.2: FTIR spectra: upper curve is FTIR spectrum of EB-I in KBr pellet; lower lower pellet; KBr in EB-I of spectrum FTIR is curve upper spectra: FTIR 4.2: Figure

Absorbance (arbitrary unit) 701300 1700 1500 Wavenumber (cm') 1100 900 700 500 102 103

400 600 800 Wavelength, nm

Figure 4.3: Solution UV-Vis spectra for SPAN base: — ---- , LEB-SPAN in aqueous 0.1 M NH4OH, maxima at 312 nm (eV) and 541 nm (eV); ---- , EB-SPAN in aqueous 0.1 M NH 4O H , maxima at 320 nm and 563 nm. Cuvettes used for all of above spectra are from Aldrich Chemical Company ( 10-mm light path and transparent from 165 nm to 2600 nm). The spectrum of EB-SPAN was adapted from ref. [107], 104

4.4.4 Temperature-dependence of DC Conductivity

The temperature dependence of DC conductivity (< 7ac(T)) was measured via four- probe technique. The room temperature Ode of pressed pellets of freshly prepared and well ground samples were measured to be as high as ~ 1 S/cm. The temperature dependence can be fit into the quasi-one dimensional variable range hopping (quasi- ID-VRH) model [132]:

LEB-SPAN.

For comparison, Fig. 4.5 shows the poorer fit to 3D-VRH model [133], i.e., log(T 1/2 -{TdcfT)) vs. T ' 1/4 plot,

T 1/2crdc(T) = AexV[-B jT xl\ (4.2) where A and B are constants, doesn’t provide a good fit to the data. Fig. 4.6 shows that the experimental data also deviates considerably from the behavior of Arrhenius expression,

frdc(T) = Cexp[-D/T], (4.3) where C and D are constants. From these plots we conclude that the quasi-lD-VRH model [132] best describes the conductivity of LEB-SPAN. This can be understood in that interchain charge transport to nearest neighbor chains is necessary as the intrachain transport is limited by polymer chain length and structural defects. 105

1.0E+00

1.0E 01

1.0E-02

Si 1.QE-03

1 0E-04

10E -05

1 0E-06 0.050 0.060 0.070 0.080 0.090 0.100 0.110 0.120 T '/ K '1

Figure 4.4: Temperature dependence of DC conductivity of LEB-SPAN sample plot­ ted as log10(r va. T -1/2 (fit to 1D-VRH model). 106

1.0E+01

1.0E+00

1 0E-01

1 .OE-02

1.0E-03

1.0E-04

t.OE-05 0.25 0.270.23 0.29 0.31 0.33 0.35

Figure 4.5: Temperature dependence of DC conductivity for LEB-SPAN, plotted as log(o’rfc) ■ T 1/2 vs. T "1/4 (data fit to 3D-VRH model). 107 0.014

0.012

0.010

0.008

T 1 / K-1 1 T 0.006

0.004

0.002

0.000 1.0 6.0 2.0 3.0 0.0 4.0 5.0 ------T-1 (data fit to the Arrheneues expression).

(wo/S) / (3Po)6o| vs. Figure 4.6: Temperature dependence of DC conductivity for LEB-SPAN, plotted as log((Tjc) log((Tjc) 108

4.4.5 Temperature Dependence of Electron Paramagnetic Res­ on an ce

Electron paramagnetic resonance signal intensity was calibrated by comparing the doable integral of EPR signals of LEB-SPAN with that of l,l’-diphenyl-2-picrylhydrazyl (DHHP) (diluted with KBr powder), g-value analysis of LEB-SPAN samples (see Fig. 4.7: g-center, defined as g-value at the center field of the first derivative spec­ trum (dispersion spectrum) of EPR signal, is marked with “o” ) shows that at room temperature g is ~ 2.0026, a typical value of 7r radical of carbon atom, supporting that spins are delocalized into phenyl rings [134, 135]. When the temperature decreases to ~ 70 K , g increases to ~ 2.0030, a typical value of spin at hetero-atom [136, 137], indicating that spins tend to become more localized at the nitrogen sites. This is in agreement with the decreased conductivity with decreased temperature, indicating greater localization at lower temperatures. The peak-to-peak linewidth ( o ) of the first derivative spectrum, AHPP, and the full width at half maximum height linewidth (o) of the absorption spectrum, A H fw h h j decrease monotonically with increasing tem­ perature (see Fig. 4.8). This trend also correlates with the increase of conductivity with increasing temperature, as the narrowed linewidth with increased temperature is ascribed to motional (exchange) narrowing effect. The ratio of A H fw h h to AH pp as function of temperature is plotted in Fig. 4.9. It is nearly constant (~ 1.8) and is close to that for a Lorentzian line shape (1.732, the dashed line in Fig. 4.9). The solid line in Fig. 4.9 is the ratio expected for a Gaussian fine shape, 1.177 [138]. The Lorentzian type line shape of LEB-SPAN indicates that spins have generally motionally narrowed the local hyperfine fields throughout the temperature range studied.

Magnetic susceptibility as a function of temperature, x(T), obtained by double integration of EPR derivative curve, is plotted in Fig. 4.10; a Pauli component of 109

2.00310

2.00305

o o

2.00300

2 00295 o o

2.00290

2.00285

2.00280

2.00275 70 120 170 270 320 T/K

Figure 4.7: Temperature dependence of g-value for LEB-SPAN sample: g-value at center field of EPR signal of derivative spectrum (o). The solid line is the linear fit to the data. ieit AHP () ad ulwdha-afmaia ieit F M (o). HM FW linewidth axima full-width-at-half-m and PP (o), H A linewidth Figure 4.8: Tem perature dependence of EPR linewidth for LEB-SPAN: peak-to-peak peak-to-peak LEB-SPAN: for linewidth EPR of dependence perature Tem 4.8: Figure

Linewidth / Gauss 0.00 2.00 3.00 6,00 5.00 1.00 [,00 % 50 100 8 $ T/K 150 Oo o O o 200 ooo0o 0 o o °o 250 OO OoA 300 110 Ill

3.00

2.50

& 2 00

1.50

0.50

0.00 0 50 100 150 200 250 300 T/K

Figure 4.9: The ratio of AH fwhh to Hpp for LEB-SPAN as function of temperature (•); the reference line at 1.732 is for Lorentzian type of lineshape (■••); the reference line at 1.177 is for Gaussian lineshape ( ------). 112 magnetic susceptibility (invariant with temperature) is present. From (x(T) ■ T) vs. T plot (Fig. 4.11), we separate the Curie and the Pauli components and obtain the density of states at Fermi level as D(cf) ~ 1.0 states/(eV two-rings) for each sign of spin and 0.022 spins/two-rings for Curie spin concentration, compared to those of ~ 0.8 states/(eV two-rings) and 0.02 spin/two-rings for Pauli and Curie component, respectively, for EB-SPAN via Faraday balance m easurem ent [106].

The EPR results are consistent with the DC conductivity data within the quasi

1-D VRH model. For quasi 1-D VRH model To equals 8a/(zD(eF)ks) (where a -1 is the localization length; z is the number of nearest neighbor chains and is set equal to 4). TqEB_span/ToB_span is determined to be 0.65 from conductivity data and D(eF)EB-SPAN/D (eF)LBB-SPAN is 0.78 from EPR data. Therefore the localiza­ tion lengths for LEB-SPAN and EB-SPAN are self-consistent, i.e., (a- ijleb-span j (a-l)EB-SPAN ^ 1Q

4.4.6 pH-Dependent Behaviors pH-dependence of conductivity: As LEB-SPAN has a higher S/N ratio than EB-SPAN, it is of interest to compare the pH-dependence of their conductivities. We plot pH > 7.5 and for PAN-HCI c r drops more than ten orders of magnitude in the region of 4 > pH > 1. This striking difference arises from the greater concentration of sulfonic acid groups attached to polyaniline Figure 4.10: Magnetic susceptibility (from double integral of E P R derivative curve) curve) derivative R P E of integral double (from susceptibility Magnetic 4.10: Figure vs. e eaue plot. perature tem

X / (omu / (mole-2 rings)) 0.0000 0.0002 4 0 0 .0 0 6 0 0 0 . 0 0.0010 0.0010 0008 8 0 0 .0 0 0.0012 • 4 1 0 0 . 0 6 1 0 0 . 0 0 0 3 0 5 2 0 0 2 0 5 1 0 0 1 0 5 0 T/K 113 114

0.025

• •

„ 0.020 «CT>

■g 0.015 E • • ¥ • • • 3 | ° ° 1° • •

eX 0.005

0.000 I 50 100 150 200 250 300

T/K

Figure 4.11: Magnetic susceptibility times temperature versus temperature plot. D(ep) was obtained from linear fit to the data (■ • •) in the temperature region of (2 K, 250 K). 115 backbone for LEB-SPAN, and may be enhanced by the formation of C 2NHOS —” 6-member ring complexes. As a consequence, the bonding strength of these protons are raised by the additional chemical bonding effect (complexation) so that they are more difficult to be dedoped (even when the protons are exchanged with cations such as Li+, 6-member-ring conformation may still exist and the imines may still be doped by those weaker metal cation Lewis acids). Therefore the samples treated with alkaline aqueous solution are still conducting on a comparable level. The size of different cations (or equivalently in this case, the strength of the Lewis acid) used in the buffers certainly has some effect on conductivity, but it appears to be insignificant. This may result from: (1) the acid strength of small cations being closer to that of a proton, due to the complexation; (2) the equilibrium position for cation exchange reaction being shifted to much higher pH when an increasing concentration of sulfonic acid groups is attached to polymer backbone and when larger cations are used in the buffer solution. pH-dependence o f E \/2 in cyclic voltammogram : The cyclic voltammograms of LEB- SPAN in pH = 1 and pH = 2 buffer solutions are shown in Fig. 4.13. E^tpH ), the half-wave potentials [139] as a function of pH, are plotted in Fig. 4.14 (two parallel solid lines with slopes of -59 mV/pH in Fig. 4.14 represent the reversible behavior of electrochemical oxidation processes while the line with slope of -118 mV/pH is for the hydrolysis process). A similar plot for EB-SPAN also showed a linear relationship between E ^ and pH for both of the first and the second oxidation potentials [106]:

£1/ 2, i = £1/ 2, * “ k pH , i = 1, 2 (4.4) where E i/ 2, i and E"/2 j are the half-wave potential at arbitrary pH and at pH = 0, respectively; i = 1 and 2 denote the first and second oxidation waves, respectively; 116

1.0E+01 1.0E+00 A 1.0E-01 - O A 1.0E-02 1.0E-03 1.0E-04 E 1.0E-05 iJS 1.0E-06 ■ 1.0E-07 • 1.0E-08 1.0E-09 1.0E-10 A A 1.0E-11 • 1.0E-12 1.0E-13 10 12 14 PH

Figure 4.12: pH-dependence of DC conductivity at room temperature for LEB-SPAN (o), EB-SPAN (o), and PAN-HC1 (A). The data of EB-SPAN and PAN-HC1 were adapted from ref. [107]. iue .3 Cci vla ga o E-PN n H . ( 1.0 = pH in LEB-SPAN of s ogram m voltam Cyclic 4.13: Figure ( -----

bfe slto. h sa rt i 5 mV/ ad urn rne s 0 mA. 100 is range current and /s V m 50 is rate scan The solution. buffer ) Current / mA O TD < O cc O C O - . 0.0 0.2 Potential (vs. Ag/AgCI) / Volts / Ag/AgCI) (vs. Potential 0.2 0.4 0.6 ------0.8 ad H 2.0 = pH and ) 117 118

0.800

0 . 7 0 0

0 . 6 0 0

0 . 5 0 0 A 0 . 4 0 0 *! o A > 0 . 3 0 0 6 e 6 UJ A 0.200 A 6 0.100 a 0.000

- 0.100

- 0.200 -1.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0

pH

Figure 4.14: Half-wave potential, E^a, i, of LEB-SPAN sample vs. pH plot: The first oxidation potential E 1/2, 1 was fit into a straight line with slope = -59 mV/pH while the second oxidation potential Ei/ 2, 2 was fit into two segments of straight lines (the slope of line from pH = 1 to pH =5 is -59 mV/pH while that from pH = 5 to pH = 7 is -118 mV / pH), (o) and (•) represent data for one electrode and so do “open” and “filled” (A) for another. The electrodes were prepared with powders from the same batch of SPAN. 119 ki = 59 mV/pH and k2 — 118 mV/pH [106] for EB-SPAN. The pH dependence of El/2 for the first and the second oxidation waves of parent polyaniline with ki ~ 0 mV/pH and k2 ~ 118 mV/pH [121, 91, 85, 86] differs dramatically from that of LEB-SPAN (both fci and k2 ~ 60 mV/pH).

The comparison of the Ei/2(pH) vs pH plots for both LEB-SPAN and EB-SPAN reveals some new features for LEB-SPAN. At pH = 0 the Ej/ 2, i value is considerably larger while E i/2, 2 value is smaller than those of corresponding values of EB-SPAN [106], indicating a higher S/N ratio, in accord with the suggestion by Yue et al th at the reduced separation of the two peaks could be a consequence of steric effects of the bulky sulfonic acid substituent [106]. An alternative interpretation is that the sulfonic acid groups, being strong electron withdrawing groups (EWG), when attached to phenyl rings, lower the energy level of the valence band (or the highest occupied molecular orbital (HOMO) of the polymer) and increase the band gap between the conduction and the valence band [126]. Therefore more energy is needed to take off (ionize) an electron from the polymer chain. As a consequence, Ej/2j j increases. The reduced Ei/2> 2 value could be attributed to the effect of the lowered energy of transition state (similar to Anomeric Effect) [126] : In the transition state, the non­ bonding atomic orbital (where a lone pair of electrons resides in) on oxygen atom of sulfonic acid group could overlap with the empty p-orbital (one electron is being taken off) on nitrogen atom to form a 5-member ring complex of lower energy than that of otherwise an open chain transition state. In other words, the energy of the transition state is lowered to yield a smaller Ei/2t 2. Therefore the closer set of two peaks could be again a positive sign of more sulfonic acid groups attached onto phenyl rings for LEB-SPAN. It is noted, however, that recent quantum chemical calculation of EB- SPAN compared to emeraldine salt (in polaron lattice form) indicate little difference in their HOM O-LUM O gap [140]. 120

There is apparent deviation of the vs. pH from a straight line behavior at pH — 5 and above, which may be attributed to the irreversible degradation (hydrolysis) of polymer chain. When amine moieties are oxidized to the iminium cation, some positive charge density will be induced on the adjacent carbon atom which is readily attacked by hydroxide anions and subsequently hydrolyzed [112, 121, 141, 142]. The higher the pH (therefore the higher concentration of hydroxide anion) and / or the higher the scan potential, the more severe this type of nucleophilic substitution reac­ tion will be. The mechanism for hydrolysis reaction at the second oxidation wave is schematically illustrated in Scheme 4.9 infra. The quantitative representation will be given below.

4.4.7 Models for Electrochemical Redox Processes

We first present a simplified picture, namely, the quasi-random, oxidation model, for the electrochemical oxidation processes of polyaniline backbone (a similar statistical mechanics treatment of a chain of polarons and bipolarons, namely, the “Box Model”, was proposed for polypyrrole by Genoud et al [144, 145]). This model is suggested as a basis for understanding pH-dependent CV experimental results and the in situ CV, EPR, and DC conductivity data (i.e. hydrolysis effects and also the large ratio (2 ~ 3) of the EPR signal that occurs during the first CV oxidation wave (LEB —► ES) to that which occurs during the second CV oxidation wave (ES —> PNB)) [91, 85, 15, 87, 88, 89, 90]. In addition, it is also expected to interpret the origin of the asymmetric shape of the voltage dependent in situ CV/DC conductivity data because within this model the Pauli and Curie spins are separable. This model assumes that the oxidation of amine sites along LEB polymer backbone occurs quasi-randomly. The effects of 121 including Coulomb repulsion between the (assumed localized) charged oxidation sites are examined. In addition, the role of a polaron lattice (Pauli susceptibility vs. Curie susceptibility) on the simulation of the reported in situ EPR signal is studied. Based on this schematic microscopic picture, we propose a set of half electrode reactions for pH-dependent electrochemical oxidation processes, predict the slopes of their Eay2, i vs. pH plots from the Nernst equation, and compare those predicted values with the experimentally determined slopes.

Quasi-random Oxidation Model

Several assumptions are made in this model: (1) At the early stage of the oxida­ tion process the polymer repeat units are oxidized to form polarons; (2) coulomb repulsion favors the configurations with the largest separation of charged polarons (i.e., the formation of doubly charged bipolarons is not favored at the early stage);

(3) adjacent pairs of polarons when they do occur have zero magnetic susceptibility (caused by strong anti-ferromagnetic coupling, as expected for the spins on nitrogen sites interacting through the para positions of benzene rings [151], or by formation of a bipolaron-like doubly charged spinless quinoid units, or by formation of spinless neutral quinoid units as the result of deprotonation of charged bipolarons).

The mutual relationships between the Reduced Substituted Aniline ring unit (RSA) and the Oxidized Substituted Aniline ring unit (OSA) (including its cation radical form (OSA+) and its deprotonated neutral radical form (OSAn)) are given in Scheme 4.4 (I). These relations are represented in their symbolic forms in Scheme 4.4 (II) to enable an easier description of Scheme 4.5 that follows. 122

(n) Equivalently, reaction (1) can be symbolized as

o+

O

Or

w here R (RSA);

o+ (OSA+);

0 (OSAn); 0 - (OSA), either (OSA+) or (OSAn)

Scheme 4.4: Symbolized interconversion among the reduced and the various oxidized polymer repeat units. 123

In Scheme 4.5, we illustrate the oxidation processes schematically in terms of the symbolic polymer repeat units defined in Scheme 4.4 and at the bottom of Scheme 4.5. It is impossible to represent all of the possible configurations and paths in Scheme 4.5, therefore we illustrate only the random oxidation of RSAs and the origin of the large ratio of the intensity of the first peak over the second one in the in situ EPR spectra. We note where paths and configurations are omitted with symbol “...” in Scheme 4.5. The simulation presented below does include all possible configurations.

In Scheme 4.5, RSAs in a substituted LEB polymer chain (I), represented as N repeat segments composed of 8 RSA units each, is shown being oxidized to the charged radical states (OSA+s) along configuration (II) via path 1. Here we explicitly write out configuration (II) which has the smallest Coulomb repulsion.

Potentially there are three paths by which the polymer chain (II) can be further oxidized to polymer chain (III X) (X = A, B, C) (note that (III A) and (III C) are equivalent). The shape of the in situ EPR and CV spectra will be shown to be dependent upon the paths taken. First it is assumed that (II) is oxidized only through the “conventional” paths, i. e., (II) — ► (III A) or (II) ----► (III C). In the model of Huang et al [121], adjacent amine nitrogen atoms always are paired up during the oxidation reaction, which implies that the two nitrogen atoms are oxidized either simultaneously or consecutively in the oxidation process. Thus the oxidized polymer ((III A) or (III C)) obtained through this path contains only doubly charged spinless bipolarons, resulting in a zero intensity in the in situ EPR/CV spectra, contradicting the reported nonzero minimum in situ EPR intensity [91, 85, 15, 87, 88, 89, 90], Therefore other paths contributing to spin signal are present.

We now discuss the oxidation of polymer from configuration (II) to (III B) through path 2b. The oxidation via path 2b occurs readily because (1) the amines 124

1 _A_

other paths RR RR R R RR to path 1 -2Ne" CD o.| R RR o,| R RR

palh2a/i2Ne path 2b - 2N e path 2c \ . 2N s (III A) ("I C) R 0 O RR o, R 0 R

path 3a' - 2N epath 3 a \: 2N a " Pa,h 3b palh3c/^2Ne path 3c' - 2N e

(IV B) (IV A) (IV A) path 4 a \ ; 2N o path 4b - 2N e

other paths

(V)

Notation: (RSA);

(OSA) with one spin up;

(OSA+) with one spin up;

(OSA) with one spin down.

Scheme 4.5: Visual representation of quasi-random oxidation processes. 125

undergoing oxidation are less affected by the induction effect of the positive charge already present (the cations are further separated compared to those produced follow­ ing the “conventional” path) and (2) the formation of a polaron lattice further lowers the energy. In fact, the thermodynamic stability of configuration (III B) over that of configuration (III A) or (III C) (a bipolaron lattice) has been implied in the proposed bipolaron-polaron phase transition [146]. More importantly, this configuration would contribute to the magnetic susceptibility, supporting its central role in the in situ

EPR signal recorded during a CV potential scan.

The configuration (III B) in Scheme 4.5, the oxidation product of polymer chain (II) along path 2b, is a representative of configurations which, differing from those of (III A) and (III C), have a positive contribution to the magnetic susceptibility. In other words, at this oxidation stage the Pauli and the Curie spins are mixed together

with their weighting factors as a function of the scan potential and the number of polarons centered on alternating sites.

Paths 3x and 4x (x = a, b, c) show stepwise oxidation of the substituted ES to PNB. The experimental EPR peak intensity during the second oxidation wave is smaller than that during the first one. Based on the quasi-random oxidation model we are able to roughly estimate the ratio of intensities of the two EPR peaks.

When estimating the spin concentration for oxidation stages indicated in Scheme 4.5, we assume, for simplicity, equal probability for all possible parallel paths. Also note that we use a localized description rather than a delocalized picture to represent a polaron lattice [146, 147]. If the EPR signal at the first wave is assumed to arise from configuration (II), then the total spin will be N (i.e. 2N contributions of spin one-half), and the other similar configurations will also have a total spin of ~ N. In contrast, if the EPR signal at the second oxidation wave is assumed to derive from 126 the configurations shown in (IV), weighted by the probability of their occurrences, then the total spin will be reduced. The average spin is determined as (5/9)N, from 8 configurations of spin zero (four (IV A)s from path 3a' and 3c' each) and 10 con­ figurations of spin one (6 (IV B)s from path 3b and two from path 3a and 3c each), i.e. (8x0 + 10xl)N/18 = (5/9)N. This yields an EPR integrated intensity ratio be­ tween the first and the second oxidation wave of approximately N/[(5/9)N] = 1.8, in agreement with the reported experimental ratio ~ 2.

Simulations [152] based on the quasi-random oxidation model assumptions have been carried out for chain lengths up to 1000 CgN units. Simulation results have been averaged over 500 polymer chains. The experimental CV spectrum of PANI was normalized to the chain length used in the simulation (for example, 1000 repeat units). Then the normalized oxidation density distribution function, D oxi{ V) (the number of amine atoms oxidized per unit voltage as a function of CV potential scan), was calculated, according to the formula

D - * y ' = (4-5> where V is the voltage of the CV scan potential, v = i>(V) the scan rate of the in situ CV experiment, I(V ) the corrected anodic current (taking the charging effect and solution ohmic polarization into consideration), V{ and V/ the initial and the final scan potential, respectively, and N the chain length chosen in our simulations. This distribution function was subsequently used in the voltage-dependent EPR spin density and DC conductivity simulations. The important contribution for a polaron lattice is included as described below. The polymer spin concentration contributed by configuration (III B) would be over-estimated if the polaron lattice were not con­ sidered. The Pauli susceptibility is /igN(eF) (the density of states, N(^f), is measured from both the conductivity and the magnetic susceptibility experiments). To pre- 127

Table 4.1: Spin-counting schemes used in the simulation of the in situ CV and EPR data.

Configuration of Oxidized Polymer Chain Type Spins

A 0 R 0 R 0 R 0 R O S3 O R O R O R Pauli (rk +0.001) •"*1 w B ORRRRR R |R o R O R R R 0 R Curie X COOR RRR R |R 0 O R R R R O O Bipolaron 3 x 0 D 0 0 R R O R OR o ROR RRR 0 Mixed £ + 0.501

vent the over-estimation of the spins from configuration (III B) and in recognition of the zero spin contribution of doubly oxidized tetramers of emeraldine salt [148] and the Pauli susceptibility of octamers [149] and longer lengths [150] of emeraldine salt, we adopt the following counting schemes. The spin contribution of lengths of six or more polarons centered on alternating sites (i.e. hexamers or longer) are counted as Pauli spin and set its value to ^ + 0.001, where lV is the length of “Pauli segment”.

For example, we count the spin of the 8-ring segment in configuration (III B) of Scheme 4.5, as 0.001 + 1/82. While we count isolated polarons (separated by more than two ring units) as Curie spins (e.g. 2N times spin one-half as in the configu­ ration (II) of Scheme 4.5) and count two adjacent polarons as spin zero (e.g. zero spin for configuration (V) in Scheme 4.5). These counting schemes are exemplified in Table 4.1.

In the simulation, we varied the Coulomb repulsion between two polarons on the nearest neighbor sites U\ (therefore U\ = 2£/2, see Eqn (4.6) below), and found the best fit for the ratio of the probabilities of the second nearest neighbor amine sites of a polaron to be oxidized versus those of the first nearest sites, P2/P i, ~ 200. This 128 ratio yields a dielectric constant ~ 10, a reasonable averaged value between those for non-conducting and conducting powder samples, as evaluated in the paragraphs that follow.

As we know, the Coulomb interaction energy U is inversely proportional to the inter-particle distance r, u = (M !M , (4.6) er where e is the dielectric constant of the media, and Z;e is the charge of i’th particle [153]. For a bound state of a hydrogen-like atom at its ground state (quantum number n = 1), the total energy is

E = < T > + < U >= - < T > (4.7)

- -13.6Z2 (eF), (4.8) where < T > and < U > are the expectation value of kinetic energy and potential energy of the hydrogen-like atom, respectively. Therefore we can have a simplified expression for Eqn (4.6) as (J = 2 W, (en (49) where r' is a dimensionless number, multiple of Bohr radius a0 (0.529 A). From Maxwell distribution, we have the ratio of the probabilities of the second nearest neighbor amine sites of a polaron being oxidized over those of the first nearest neighbor sites as

P2/P 1 = exp (-(U2-U,)/kBT) (4.10) . -27.2. 1 1, 1 , , = (4 -n > .27.2 1.1605 x 104. = ^ )• (4-12) 129

Table 4.2: Several frequently used covalent bond lengths in conducting polymer [156].

covalent hond bond length (A) chemical bond bond length (A) C-C 1.54 C=C 1.35 C-F 1.38 C-N 1.30 C-H 1.09 CC bond (benzene) 1.40 C-N 1.47 N-H 1.00

Notice that in the transformation of Eqn (4.11) to Eqn (4.12), a conversion factor = 1.1605 x 104 has been implicitly applied. Now using the covalent bond lengths listed in Table 4.2, r\ and r'2 can be estimated as 10.8 a0 (5.7 A) and 21.6 a0 (11.4 A), respectively. For a poly aniline sample, the dielectric constant (e) varies [157]: undoped ~ 3-4; powder sample ~ 20; stretched film ~ 200; “metallic” polymer film ~ 105 or negative. Therefore P2jP \ is “106” and “12” for undoped (e = 3.5) and conducting powder (e — 20) sample, respectively. While the simulation results in a ratio of P2 to P\ ~ 200, yielding a dielectric constant ~ 10. This is a reasonable averaged value of non-conducting and conducting powder samples. Therefore the simulation result is consistent with dielectric constant data.

The simulation result for the in situ EPR spectrum is plotted in Fig. 4.15. The ratio of the peaks of the spin concentration in Fig. 4.15 converges to ~ 2.2 with a minimum susceptibility at ~ 1/10 of the first peak value, similar to the experimental value.

The in situ potential dependent DC conductivity experiments have also been performed by several research groups [86, 88, 90]. However, confusion still remained. First, there is an issue related to the conducting mechanism. Kruszka et al [90] cn oeta. r te opttoa dt vrgd o 0 hnrd times. hundred 500 to averaged data computational the are potential. scan Figure 4.15: Simulation result of of result Simulation 4.15: Figure

Spirts / [1000 Rings] 120 100 20 40 80 - 0.2 . 0.2 0.0 n situ in EPR signal intensity as a function of CV CV of function a as intensity signal EPR Voltage/V 0.4 0.6 0.8 130 131 noticed that the increase in conductivity seemed to be correlated with the decrease in spin concentration, and therefore proposed two alternatives of explanation:44 (a) The decrease of the spin susceptibility corresponds to a cross-over from localized polaron to a polaron band with Pauli susceptibility [150], or (b) the spin pairing is originated in the polaron to bipolaron conversion upon doping.” It is also found that the shape of the potential dependent DC conductivity curve is asymmetric [86, 88, 90]. Within the quasi-random, oxidation model the Pauli spins and the Curie spins are separable from the total simulated spin density function. Therefore with our model it is possible to resolve these confusions and to reveal some hidden physics from the correlation of the simulated Pauli spin density with the reported DC conductivity data.

Fig. 4.16 shows the simulated Pauli spin density curves of various P2 jP \ ratios.

It is seen that the Pauli spin curves converge to an asymmetric asymptote with P 2/P 1 ~ 200, which fits well to the experimental DC conductivity data (labeled with 44o”). This provides evidence for the formation of a polaron lattice. In addition, some other important conclusions can also be drawn for conducting mechanism studies:

(1) the DC conductivity is proportional to the number of Pauli spin-1; anJ ( 2) the conduction of polyaniline is via Pauli spins defined as the spins in polaron lattice, it therefore disagrees with the bipolaron conduction model for polyaniline. These conclusions are expected but have not yet been reported, therefore providing further evidence of the success and significance for our quasi-random oxidation model .

Based on Scheme 4.5 and the simulation result in Fig. 4.15, we present a qual­ itative description of the reported in situ EPR signals [91, 85, 15, 87, 88, 89, 90] as a function of the CV potential. Configuration (I) represents a polymer chain in LEB oxidation state, which has zero spin, corresponding to the stage before the threshold of the first oxidation wave. Through path 1, LEB is oxidized to yield configuration 132

1 6 0 cv Pauli (22) Pauli (82) 140 Pauli (192) O DC data - 5.0 d .\ o- 120 - 3 4.0 u> 2 CO c +- 100 - ._ «- ♦3 3 2 O CO O 60 3 Q . O TJ 2.0 £*■ O o a. o 40

20 -

o o

- 0.20 0.00 0.20 0.40 0.60 0.80 1.00 Potential / Volts

Figure 4.16: The simulation results of the in situ DC conductivity as a function of CV potential scan. Various curves are labeled in the legend table inside Fig. 4.16: CV (—); DC conductivity (o) are adapted from Kruszka et al [90]; and Pauli spin densities for various ratios (broken lines with corresponding Pg/P\ ratio labeled in the legend table). The computational results are averaged over 500 polymer chains. 133

(II), a representative of configurations with predominantly isolated polarons (Curie spins) along the chain (namely, protoemeraldine [63]), which has the largest possi­ ble EPR signal within this model. Upon oxidation of the protoemeraldine, spinless emeraldine (paths 2a and 2c, yielding either doubly charged spinless bipolarons, or, after deprotonation, spinless inline units) and spin containing emeraldine salt (the Curie spin as (III B) literally indicated and the Pauli spin, both implied through path 2b) are formed. This oxidation stage corresponds to a non zero EPR intensity as reported in the region between the peaks of the first and the second oxidation waves. As the oxidation continues through path 3x, the polymer chain is oxidized further to nigraniline [63], which is suggested to correspond to the second peak of the in situ EPR signal. At this potential the EPR intensity is one-half of that at the first maximum. Further oxidation of the polymer leads to the highest oxidation state of polyaniline, pernigraniline, which is spinless. It is noted that neither protoe­ meraldine nor nigraniline are a separable oxidation state of the chemically oxidized polymer (which only have leucoemeraldine, emeraldine, and pernigraniline as discrete oxidation states [154]).

The other evidence which supports this model comes from the consistency of the predicted and the experimentally determined ratio of two in situ EPR peaks of hy­ drolyzed samples. For example, if hydrolysis occurs during the multiple potential scan in the in situ CV and associated EPR experiment, and the sample film is continuously degraded into smaller and smaller fragments, say, segments with lengths less than 8 ring units, then the ratio of the intensities of the in situ EPR peaks corresponding to the first and the second oxidation waves of CV experiment will increase dramatically. Our model shows that oligomers with lengths less than 5 do not contribute to the second peak of the in situ EPR spectrum. Indeed, the ratio of the EPR intensities at the first and at the second oxidation waves reported by Lapkowski ei al [89] is ~ 134

5 when the “central” peak (signature of hydrolysis [141]) is present. It is worthwhile to note that the presence of the defect states of the base forms of polyaniline and its derivatives is a natural outcome of the quasi-random oxidation model.

In summary, the statistical quasi-random oxidation model of the electrochemi­ cal oxidation of LEB to PNB via multiple paths accounts for the variation of EPR intensity and the asymmetric shape of DC conductivity data, in contrast to the con­ ventional (consecutive or simultaneous) oxidation model [121].

Model for pH-dependence of Oxidation Reactions

The pH-dependence of the oxidation potential, E, of a reversible half-electrode reac­ tion,

R —> O + ne + m H +, (4-13) can be described by Nernst equation,

E = E° + ~ l n [ H +]m n F 771 = E ° - 0.059— pH, at 25° C, (4.14) n where R represents the species at lower oxidation state and O the species at higher oxidation state, m and n are the coefficients for protons and electrons involved in the reaction (4.13), respectively. The standard potential, E° in equation (4.14), is equivalent to E°/2, the half wave potential [139] in a cyclic voltammogram.

In order to interpret the complex phenomena related to the electrochemical redox processes of polyaniline and its derivatives, some general electrochemical processes are proposed in Schemes 4.6 — 4.9. It is assumed that these proposed electrochemical 135 processes are reversible enabling application of the Nernst equation. Because the time scale of the redox process (scan rate of 10 mV/sec was typical) in cyclic voltammetry is

~ 50 s, quasi equilibrium conditions are assumed and hence a steady state treatment [87] is valid. As a consequence, we can calculate and compare the calculated slopes with the experimental ones.

In Scheme 4.6 (I), the symbols used in Schemes 4.6—4.9 are defined. A f repre­ sents a substituent, which could be a hydrogen atom, or a “ — SO 3H ” group, or any other substituted group. The subscript “f” denotes a number (a fraction or a whole number, taking values from 0 to 4, representing the number of “A” attached to a phenyl ring statistically). M + can be a proton, or a metal cation (such as K+, from the supporting electrolyte, say, 1 M KC1 in a pH > 1 solution). The “spring-like” symbol defined in Scheme 4.6 (I) represents a substituted phenyl ring that does not change during electrochemical processes.

Scheme 4.6 (II) schematically illustrates the resonance of a polaron between two adjacent nitrogen sites, (4C) and (4D), through an intermediate state, (4CD), via an internal redox reaction. Though positive polarons are thought to be more delocalized than indicated here, their resonance between two adjacent sites will proceed in the same manner.

Fig. 4.17 describes general equilibria of redox reactions and their related pro­ tonation and deprotonation reactions. Kp_ ; and Kd ; are equilibrium constant of protonation and deprotonation reaction for i’th reagent at j’th oxidation level, re­ spectively. kJ,roto. and kJd(.proto. are protonation and deprotonation rate constant for i’th reagent at j ’th oxidation level, respectively. In Fig. 4.17 (I), a general description of sequential protonation equilibria of the polymer backbone at arbitrary oxidation level (i.e., substituted LEB, or EB, or PNB) is given. In Fig. 4.17 (II), general oxida- 136

(4A) <4B)

A

d -

Scheme 4.6: schematic illustration of the resonances of a polaron between two adjacent nitrogen sites. Scheme 4.7: Oxidations of the substituted LEB in less acidic media.

tion processes, where radical cations are formed followed by subsequent deprotonation and protonation equilibria, are shown. Off course the effect of the Coulomb repulsion will modify the individual equilibrium constant.

Scheme 4.7 shows the proposed oxidation processes in less acidic media, where the deprotonation of oxidation product occurs completely. In Scheme 4.7 (I) and (II), the oxidation of the substituted LEB (4A) to the substituted EB (7E), and then to the substituted PNB (7F) are given. As each of these steps involves the simultaneous removal of one electron and one proton, the slope of the half-wave potential vs pH for each of these processes is therefore ~ -59 mV/pH. 138

kproto) (i) P + HA (PH) A , kdeprotoi kproto2 , 2 + . 2 - (PH) A + HA kdeproto2 ^ A ’

(n-l)+ (n-1)- + W o „ . (PH n_-|) A + H A (PH n)n+ + A n ' ^deproto „

(n-1 ) h ’'p ro to , or simply, (PHn.i) H*' (PHn)n+ kdeproto „

’'proto i Kn, j - P " “ k,deproto i

n+ (n+1)+ (n+2) + (H) (PHn) + Og + 2e

k l kproton kdeproton kprotOn deproton kprotOn | d e p ro to n

(n-1) + + i+ . .+ (n+1)+ + (P H n .!) + H , + h , 0 2 + H , * *

k2 'protOn -1 kdeproton-l kprot0n-i kdeproton -1 proton -1 kdeproton -1

u deprotoi I d, i = “ j — = —— V o t o i Kp j

Figure 4.17: Equilibria of protonation and oxidations among various oxidation states of polymers. 139

Scheme 4.8, in contrast to Scheme 4.7, illustrates how the most possible oxida­ tion processes occur in very acidic media where both the starting polymer and the oxidation products are protonated to a considerable extent. For the emeraldine ox­ idation state, the deprotonated form is in equilibrium with the protonated form (or the usual emeraldine salt form, etc.), with their ratio affected by pH. Scheme 4.8 (I) shows that a protonated amine is oxidized first. Then the product (8H) is deproto­ nated because that the two positive charges experience a strong Coulomb repulsion. Therefore the half-wave potential vs pH is ~ -59 mV/pH. On the other hand, the slope of the oxidation potential vs pH for Scheme 4.8 (II) is likely between 0 and -59

mV/pH, which can be rationalized as that at higher potential and lower pH (say, 1 M HC1), the accumulating polarons centered on adjacent sites combine to yield doubly charged polaron ‘pairs’ (81), with strong anti-ferromagnetic coupling (suppressing the

measured spin) [87, 151], These polaron ‘pair’ sites are eventually deprotonated to form the quinoid counterpart (7F) in the presence of an applied electric field (see Scheme 4.8 (III)).

Scheme 4.9 shows the mechanism for electrochemical oxidation-hydrolysis reac­

tion. In Scheme 4.9 (I), the oxidation-hydrolysis of EB in less acidic media is shown and slope is ~ -118 mV/pH. While in Scheme 4.9 (II) the oxidation-hydrolysis of ES in very acidic media is described and it is seen that the slope is only ~ -59 mV/pH.

In short, three types of slopes are obtained: (1) For the processes involving equilibria of protonation-deprotonation of the species during the oxidation process, the slope is between zero and -59 mV/pH, which may corresponds to the slightly decreased slope at the beginning portion of our plot (see Fig. 4.14); (2) for the process with m /n = 1 (see equation (4.14)) the slope is -59 mV/pH, which is well observed in the middle portion of our plot; (3) for the processes where oxidation-hydrolysis occurs 140

e (i)

(8G) (8H)

^deprou ^proton

( 4 0

Af \

Af kdepro| + e + H .N-

(8J)

Af \ (m ) + 2H \

(81) <7F)

Scheme 4.8: Oxidations of the substituted LEB in very acidic media. 141

(m/n = 2), the slope is -118 mV/pH. The portion of our data at higher pH yield a slope of ~ -118 mV / pH, providing an evidence for hydrolysis. This is supported with the observation of the irreversibility associated with the first and the second oxidation wave: we observed a third set of peaks, the so called “central peak” [155] (signature of hydrolysis [141]), at ~ 0.4 volts (us Ag/AgCl) on the CV spectrum of an LEB-SPAN film coated electrode in pH — 1 solvent for scan potential < 0.7 V. This electrode had delibrately been subjected to CV earlier (for scan potential < 0.35 V) in electrolyte of pH = 6. As shown in Scheme 4.9 (II), at low pH, hydrolysis could also occur with a slope of -59 mV/pH, the same slope as observed during the usual normal redox process of PANI and its derivatives. We have observed evidence of the occurrence of hydrolysis at low pH with the first and the second set of oxidation waves of the CV rapidly merging into a new set of “central” peaks when the upper limit of the scan potential was set to 1.4 volts for a LEB-SPAN electrode at pH = 1.

The differences in electrochemical behavior between LEB-SPAN and other forms of polyaniline can be understood as following. First, the reduced hydrolysis for LEB- SPAN at pH < 5 is suggested to be the consequence of having more substituents such as sulfonic acid groups; these groups could partially shield the carbine [126] from being attacked by anions, e.g. hydroxide ions). A second note concerns the difference in the slope of oxidation potential vs pH for the SPAN samples as compared with the parent PANI. For parent PANI, the corresponding two half-electrode processes are: (1) (4A) — ► (4C), with Jbi ~ 0; (4C) — ► (81) — ► (7F), yielding k2 ~ 118 m V /pH . On the other hand, LEB-SPAN has bulky substituent which are EWG in nature, resulting in: (1) EWGs are withdrawing some electron density from the ring therefore destabilizing the polaron or doubly charged bipolaron (in other words, deprotonation occur more readily); (2) Bulky substituents open more and larger channels for ions to diffuse in or out, which makes the local pH closer to the bulk pH. Therefore LEB-SPAN will have 142

(i) + M + H2O

(7E)

M - H+ o-

(9K) (9L)

M \ V. / H (9N)

H+ (cat.) (I) + 2 H2O • 2 e ' - 2 H 1

(90)

H V V N- I'

/ H/ {9P> (9Q)

Scheme 4.9: Hydrolysis mechanism for various oxidation states of substituted PANI. 143

an improved dynamic response to bulk pH, yielding both ki and k 2 ~ 59 mV/pH.

4.5 Summary

The element chemical analyses, consistent with XPS analyses, of LEB-SPAN show sulfonation levels as high as S/N ~ 0.75 with high yield (~ 0.7). FTIR and UV-Vis, CV spectra, and pH-dependence of conductivity are consistent with the higher S/N ratio. EPR and DC conductivity data are self-consistent, showing that LEB-SPAN has a higher conductivity with a weaker temperature dependence as compared to EB-SPAN. The conductivity of LEB-SPAN persists to much higher pH than that of parent emeraldine salt and even than that of EB-SPAN.

We have proposed a statistical quasi-random oxidation model for the electro­ chemical redox processes of PANI and its derivatives. Based on this model, we have successfully simulated the reported in situ CV, EPR, and DC conductivity spectra, found evidence for the polaron lattice formation, discovered that the proportionality of the reported DC conductivity data to the simulated Pauli spin density therefore disagreed with the bipolaron conducting mechanism, rationalized the different behav­ iors between the pH-dependent CV experiments of SPAN and PANI, and interpreted the hydrolysis related phenomena during the electrochemical redox reactions.

4.6 Acknowledgement

This work has been supported in part by the US Office of Naval Research. C H A P T E R V

Highly Fluorinated Polyaniline: Synthesis and its Characterizations

5.1 Abstract

We report here the synthesis and characterization of the chemically made highly fluorinated polyaniline. Elementary chemical analysis, along with XPS, mass spec­ troscopy, and solution NMR, suggests that the final polymerization product is meta- tri-fluorinated polyaniline (FPAN), which is rationalized by the mechanism proposed in this paper. The thermogravimetric analysis was carried out in Ar environment, showing good thermal stability of FPAN samples. Smooth transparent polymer thin films (ca. 0.25 fim) were spin-coated on quartz substrates and have DC conductiv­ ity as high as ~ 0.2 S-cm-1. FT-IR spectra show a novel oscillating phenomenon, which is rationalized by proposing a resonance mechanism of amine-like and imine-like structures. The resonance nature of some solution NMR signals are consistent with the observations made in FT-IR experiment. UV-Vis spectra of the doped polymer films reveal a nearly frequency independent free carrier tail up to 2,400 nm. Cyclic

144 145 voltammagrams of FPAN films coated on Pt electrodes show higher oxidation poten­ tials as compared to those of parent polymers, consistent with the expected strong electron withdrawing effect of multi-fluorine substitution. The existence of two dif­ ferent kinds of EPR signals was suggested from the EPR data of undoped and H 2SO 4 doped FPAN samples. The origin of these signals is discussed. The DC conductivity is also reported.

5.2 Introduction

Recently, polyaniline and its derivatives have aroused a great deal of attention. Thou­ sands of papers have been published in this area [159]. Among them only few papers have dealt with fluorination [160] despite the fact that there are a great number of ref­ erences in fluorination of polymers in general [159], It is recognized that the fluorine substituent is a very strong electron withdrawing group and a component of hydro­ gen bonding. Therefore, fluorinated polyaniline could have a high half-wave oxidation potential and a blue shift of the optical absorption maxima of its electronic spectrum [164]. As a consequence, highly fluorinated polyaniline has some advantages over its parent polyaniline and other polyaniline derivatives: (1) It has widely opened visible window for optical applications, e.g. possible usage as a positive charge ejection layer for light emitting diode; (2) It could be used as a protection layer against corrosion because of its dense structure and high oxidation potential; (3) It can be used as a novel electrode material because of its conducting properties [160]. However the fluo­ rination of polyaniline has been accomplished only with an electrochemistry method that limits the amount of FPAN produced. Besides, the characterizations reported [160] are rather incomplete. All of this motivated us to develop the current chemical 146 polymerization method and to perform various spectroscopic characterizations on our chemically produced FPAN.

5.3 Experimental

5.3.1 Chemical Synthesis and Sample Preparations

Synthesis of the pristine form of FPAN can be accomplished with either Route I or Route II described below.

Synthetic route I: Synthesis with an ultrasonic bath

A typical procedure of the synthesis of the tri-fluorinated polyaniline is as follows:

1.500 g tetra-fluorinated aniline (TFA) was weighed and placed in a 250 ml Erlenmeyer flask. Then 150 ml of 0.100 M HC1 was added, resulting in a solid-liquid TFA-HC1 mixture (I). At the mean time, 1.500 g of ammonia persulfate, (NH^SjOg, was dissolved into 50 ml of 0.100 M HC1 in a 150 ml beaker. This solution (II) was pre-cooled with an ice-water mixture at ~ 5° C. The TFA-HC1 mixture (I) was then heated to and then kept at room temperature until TFA solid melted (two immiscible liquid phases were observed). Then it was introduced into an operating ultrasonic bath filled with ~ 5° C water for about 30 minutes to allow the liquid phase to break up into a milk-like colloidal suspension. Then 50 ml of ammonia persulfate solution (II) was poured into the colloidal suspension. At this time, the 147 suspension became a clear solution (III) and gradually its color changed into light yellow. Ice was added frequently into the ultrasonic bath to keep the temperature of the reaction system ~ 5° C. Then solution (III) was stirred in the ultrasonic bath for about 2 hours. During this time period, the color of the reaction system became darker and some precipitate immerged. The Erlenmeyer flask containing solution (III) was subsequently transferred into an ice-box and cooled with ice (to slow down the reaction rate) for 24 hours. After then, the flask was removed from the ice-box and some brown powders were found. Then the reaction system was stirred under room temperature for one day to allow further polymerization. The product was filtered with a water-aspirator and was washed with four portions of 50 ml of aqueous 0.100 M HC1 acid. Effort was made to keep solvent level higher than the surface of the cake. Then cake was further dried until it was cracked (about 15 minutes). It was then filtered and the collected powders were transferred into a vacuum oven and dried at room temperature for five hours (the color of the cake became yellow-green). Then it was ground into fine powders, collected, and further dried at ~ 80° C for three days.

Synthesis route II: Synthesis without an ultrasonic bath

To facilitate the polymerization, a scheme aimed at increasing the solubility of TFA in aqueous solvent was developed as follows. Acetic acid (50.0 ml) was gradually added into a two phase mixture of 5.000 g of TFA in 150 ml of 0.100 M HC1. Then 5.000 g of ammonia persulfate in 100 ml of 0.100 M of HC1 was poured in. The mixture was put in an ice-water mixture in a patrick dish and stirred for two hours. The rest of the synthesis is the same as route / described above. This method results in a greater 148 control of synthesis, higher yield, and less reaction time. Since it eliminates using the ultrasonic bath, this synthesis can be utilized to synthesize a large quantity of FPAN. Up till now it was found that the FTIR and UV-Vis spectra of samples made via route I and route II are identical. Further studies were undertaken on samples produced via route II.

Synthetic scheme for increasing the yield of polymerization reaction

Since the commonly used scheme for polyaniline (PANI) synthesis only results in a 5 % or less yield of FPAN, efforts to develop a new scheme with higher yield have been made. We found out that the use of higher molar ratios of oxidizer to TFA could increase the yield of FPAN polymerization, and we have therefore adopted this strategy in the syntheses. The correlation of the yield with the molar ratio for several batches of FPAN syntheses is listed in Table 5.1. It can be seen that the higher the ratio of ammonia persulfate to TFA, the higher the yield of FPAN polymerization reaction product (we have discovered the same correlation in PANI synthesis [161]). The oxidation states and chemical compositions of the products of these batches were found to be roughly the same via FT-IR and UV-Vis analyses.

Basification of fluorinated polyaniline

Basification of the fluorinated polyaniline proceeded using following procedure: 1.00 g of the fluorinated polyaniline was placed in a 250 ml Erlenmeyer flask and then 200 ml of 0.100 M NH 4 OH were added. The mixture was stirred with a magnetic bar at room 149

Table 5.1: The correlation between the ratio of oxidizer to monomer and the yield of FPAN polymerization reaction

Sample # Oxidizer/Monomer Ratio Yield (%) 1 0.0712 3.55 2 0.1800 8.96 3 0.4759 20.03 4 0.5700 31.25 5 0.835 41.45 6 0.8600 42.28 temperature for one hour and then the acidity of the mixture was checked to make sure that pH > 9 (if not, the mixture should be filtered and another portion of 200 ml of 0.100 M NH 4O H should be added again). Afterwards, this system was stirred for 24 hours (the color of the mixture became brownish). The brown powder was then filtered and washed with four portions of 50 ml of 0.100 M NH 4O H . Then the FPAN cake was dried in a vacuum oven at room temperature for five hours, ground into fine powders with an agate mortar and pestle, and then dried at an elevated temperature at ~ 80° C for two days. The color of the final product was brownish.

5.3.2 Film preparing processes

To cast a smooth film from solution, 200 mg of the base form or pristine form of

FPAN powders were weighed and subsequently dissolved into 2.0 ml of 1.0 M of H 2SO4 alcoholic solution. The color of the filtered solution became dark-green. This solution was stirred for 30 minutes and then filtered with a 0.2 /im Sterile Acrodisc filter on a 10-cc syringe (brand name: LUER LOK) to get rid of any dust or undissolved 150 powders. The FPAN films were prepared subsequently via spin-coating method. It is very important to pre-treat thoroughly the substrate on which the film is to be coated. The typical procedure is as follows: the quartz pieces to be used are treated with 50-

50 H2O2-H2SO4 solution for one hour. Upon finishing they are rinsed with distilled water three times, then transferred into a beaker containing anhydrous MeOH, and subsequently stirred in an ultrasonic bath for 30 minutes. Then they are rinsed with double distilled water and dried in an oven.

5.3.3 Characterizations

All samples made via above methods were sent to M-H-W analytical lab (in Arizona) for element chemical analysis. Ion bombardment mass spectra were taken at the center of analytical facility at The Ohio State University (OSU). FT-IR experiments on solution and solid samples were conducted with a Mattson spectrometer in the chemistry department at OSU. UV-Vis spectra of FPAN in alcoholic solvent and on quartz substrates were obtained using a Perkin Elmer A-19 spectrometer in our lab. EPR experiments were carried out with a Bruker ESP300 spectrometer in our lab. DC conductivity measurements were carried out via a press-contact four-probe technique [125] with a Keithly multimeter and a current source on the films spin- coated on quartz substrates. Cyclic voltammograms were obtained with a Hokto potentiostat/galvanostat (Model HA-301) combined with a HC function generator (Model HB-111).

The thickness of the films coated on quartz substrates were measured with an Alpha-step instrument at professor J. C. Garland’s lab and with a Nikon Profile 151

Projector V-12 (made in Nippon Kogaku, Japan) at professor Harris Kagan’s lab.

A sample pellet for the Resonance FT1R Phenomenon experiment was prepared as a normal FTIR sample pellet. The concentration is 3 mg FPAN per 500 mg KBr. The pressure used was 15000 pounds. The instrument resolution was chosen as 4 cm-1 wave numbers. Some spectra were collected for 64 scans while the others are 1 scan each.

Thermal gravimetric analyses were carried out in pure Argon atmosphere from

40.0° C to 70° at rate of 5° C/m in and then from 70.0° C to 140° at rate of 3° C/min at professor P. K. Gallagher’s lab in the chemistry department at OSU.

An XPS experiment was done at the Shared Analytical Instrumental Laboratory in the chemistry department at OSU. FPAN sample pellets for XPS were prepared by pressing FPAN powders with a clean FUR pellet press. The pellets were mounted onto sample stub via double sided adhesive tape and grounded with DuPont silver paste. The XPS spectra were collected on a VG SIMS-ESCA-ME system (series No.: SIM-ESCA-ME 1448) with a MgKa X-ray source (1253.6 eV photons). The X-ray source was operated at 14 kV and 20 mA. The polymer powder samples were mounted onto standard VG sample studs with double-sided adhesive tapes and pumped in the preparation chamber to 10-8 mBar before being sent into ESCA chamber. The pressure of the ESCA analysis chamber was maintained at 10-9 mBar or lower during the sampling. For optimal acquisition, the line adjustments on the coordinate system and on the tilt angle (20° for small area XPS) were made to maximize the signal of spectrum. 152

(B)

NHt

(C)

Scheme 5.1: The chemical structures of FPAN monomer and FPAN lattices: (A), mefa-coupled FPAN lattice; (B), para-coupled FPAN lattice; and (C), the monomer of both FPANs, tetra-fluorinated aniline.

5.4 Results and Discussion

5.4.1 Mechanism and Rationale for Chemical Synthesis

To prevent the possible defluorination of monomer, oligomers or polymer from occur­ ring under catalysis of a stronger acid, 0.1 M HC1 was used in the synthesis instead of 1.0 M HC1 used in the standard polyaniline synthesis [121]. il o PN oyeiain reaction. polymerization FPAN of yield Figure 5.1: The correlation plot between the ratio of oxidizer to monomer and and monomer to oxidizer of ratio the between plot correlation The 5.1: Figure

Yield 0.00 0.05 0.05 0.10 0.15 0.20 0.25 0.25 0.30 0.35 0.40 0.45 0.45 .0 .0 .0 .0 0.80 0.60 0.40 0.20 0.00

Oxidizer/Monomer Ratio Oxidizer/Monomer 1.00 154

The monomer, TFAN, has very strong multiple inter-molecular interactions (hy­ drogen bonding such as “-F-H-N-”). As a consequence it does not dissolve into any aqueous solvent. Therefore, an ultrasonic bath was used to suspend the monomer into an aqueous solvent. Adding a suitable amount of acetic acid (with the pH of the solution maintained at ~ 1) was found to substantially increase the solubility of TFA.

The effect of the oxidizer/monomer ratio on the yield of the polymerization re­ action has been studied on both PAN and FPAN synthesis [161]. As tetra-fluorinated aniline has much higher half-wave oxidation potential the yield was found to be unac- ceptably low when the ratio used was the same as that used in the standard polyani­ line synthesis. A series syntheses with variation of the ratio was carried out. The correlation between the yield and the ratio was plotted in Fig. 5.1.

It is seen that the yield is proportional to the ratio used in syntheses,

Y = 0.4988[iZ], (5.1) where Y is the yield of the polymerization product and [iZ] is the oxidizer to monomer ratio, respectively. This linear relationship implies that the polymerization is probably via a stepwise polycondensation route.

The issue of whether the FPAN polymerization product is lm eta J or ‘para ’ or ‘ortho’ coupled has been studied and discussed for a long time. Based on the prin­ ciples of organic chemistry, the meta-coupling is the most likely linkage for polymer backbone. The rationale follows: In an oxidative and acidic polymerization media, a cation-radical would attack on a ring-carbon atom, of higher electron density, attached with a fluorine atom and result in a “meta-coupled” or less possibly a “‘ortho-coupled” structure, having a consequence of losing one fluorine atom from each ring unit. The 155 ortfio-coupled structure is less possible because it would cause a huge steric hindrance in its transition state. Therefore the meia-coupled structure is expected to be the major coupling scheme for the polymerization process. As a result, every ring should have three fluorine, one hydrogen, and two nitrogen atoms attached.

Assuming the oxidation state of the FPAN sample is similar to the emeraldine ox­ idation state, chemical structures of two different linkages are proposed in Scheme 5.1.

In Scheme 5.1 (A), meta-linkage is shown. It is seen that fluorine atoms at position 6 and 2’ are chemically equivalent, so are the ones at 3 and 5’, while fluorine at 2 and 6’ are not. Therefore four major clusters of fluorine signals at four different chemical shift locations will show up if the polymerization product is mefa-coupled.

In Scheme 5.1 (B), one of the possible para-linkage is shown (conventional polaron lattice structure for polyaniline derivative). If polymerization follows a para-linkage path, then at the condition where no fluorine is defluorinated from ring, only two groups of 19F signals at different chemical shift values will be present. While if polymerization follows a para-coupling route with one fluorine atom randomly lost from every ring [see structure (B)J, then there can be six chemically different major clusters of 19F peaks (peaks at six different chemical shift values).

In reality, localized representations of structures (A) and (B) in Scheme 5.1 are not accurate. As presented later in Scheme 5.2 and Scheme 5.3, both positive and negative charges can delocalize into ring units. If the delocalization rate is slow enough, the chemical shift value of each peak will change from the expected values from structure (A) or (B). However, since the positions where the charges can resonate onto are interconnected by some symmetry operations such as C2 and cr2 referenced to the axis through the ‘N’ and ‘H’ atoms, the changes will have to follow some 156 regulations. In other words, the “four main peak configuration” may still hold for meta-coupled linkage.

The various characterization results, favoring the “meta-coupling route” pro­ posed above, will be presented in later sections.

5.4.2 Element chemical analysis

Element chemical analyses resulted in a fluorine to nitrogen ratio (F/N) ~ 3.0. Anal.

Found: C, 41.41; H, 1.16; N, 8.01; F, 32.50; S 0 4, 16.21; total, 99.29 % (no ash was found). Calc, for (CeHiFajNHfH^SO^o.as^ total 99.18 %; oxygen is determined by difference. This implies that tri-fluorinated polyaniline was obtained as the polymer­ ization product. The ratio of nitrogen to aromatic ring (N/Ce) ~ 0.97 yields a rough estimation of average chain length ~ 18.

5.4.3 Fast Atom Bombardment Mass Spectrum

The strongest peaks in the Fast Atom Bombardment Mass Spectrum (FABMS) were found at m/e = 452, 863, 988, 1264, 1411, etc. (see Fig. 5.2), assigned to 3, 6, 7,

9, and 10 repeat unit fragments, respectively. The basic ‘reduced’ repeat unit is “[(CeHFajNH]” having a formal molecular weight of 145. The peak located at 452 is assumed to have one extra fluorine or nitrogen atom on the end ring unit. The peak located at m/e — 988 is assigned to a 7 repeat unit fragment with one fluorine missing while peaks at m/e = 1264 and 1411 are assumed to correspond to 9 and 10 157

Figure 5.2: Fast Atom Bombardment Mass Spectrum for FPAN 158 repeat unit fragments with two fluorine atoms missing during ionization processes. The outstanding difference (though small) in formal molecular weights between the calculated ‘reduced’ fragment and the experimental one is probably attributed to the extraction of protons or hydrogen atoms. Based on these assignments most of the fragments are considered as being highly oxidized (oxidation states are higher than that of EB state) during ion bombardment. The possible rationale follows: First, the fluorine substituent stabilizes the free radical or anion on the adjacent carbon atom so that reconfiguration to imine-like structure proceeds easier; second, the possible multiple hydrogen bonding makes vaporization of the ionized fragments more diffi­ cult so that it takes longer for hydrogen atoms to be stripped off from the polymer chains before being vaporized and moving to the detector. A radical recombination mechanism is suggested for possible loss of fluorine and/or hydrogen atoms. The rest of the peaks are weaker, and are not assigned due to the complication of possible side reaction (some para-coupling may exist) and multiple fragmentation and/or subse­ quent radical recombination processes. It is found that the assigned fragments have some features such that the number of rings are 3, 6, and 9, which might imply that the polymer chain (mefa-coupling) may take some special configuration (for example, coil-like against rod-like configuration). Notice, however, that MS result is consistent with that of the element chemical analysis results reported above.

5.4.4 UV-Vis Spectra

The UV-Vis spectra of FPAN base and salt solutions are presented in Fig. 5.3 while the spectra of FPAN base and salt films are shown in Fig. 5.4. The common feature seen in the solution spectra is the widely opened transmittance window in the UV-Vis 159

region. This is attributed to the multiple substitutions of EW groups (EWGs) on the benzene ring, which not only shifts the r — ir* absorption to the blue region but also reduces the intensity of the polaron (in salt) or exciton peak (in base). The blue-shift in the absorption spectrum is related to the reduced electron density in the valence band, so that the band gap between the conduction band and the valence band is increased. The reduction in the absorption of polaron or exciton bands relative to that of 7r — 7T* transition has the following implications: (1) The oxidation state of FPAN is well below that of emeraldine salt or base, a consequence of much higher oxidation potential in TFAN compared to that in aniline (see cyclic voltammogram (CV) shown later); (2) the quinoid structures are reduced greatly because the polymerization probably proceeds mainly through a meta-coupled mechanism. The XPS and FTIR data shown later are more favorable to the latter assumption.

The spectra of FPAN films coated on quartz substrates show some interesting features. First, the transition between benzene ir — 7r* bands dwindled. Secondly, the absorption spectra has a long free carrier tail extended to the near IR region. The possible explanation for this is that there exists a strong interaction between polymer chains as a result of the multiple hydrogen bonds which pull polymer chains close together. In other words the stronger inter-chain coupling reduces the band gap and activates the low energy excitation. The resonances between the amine­ like and imine-like structure proposed in section 5.4.7 might also provide plausible explanations for this, since an imine-like resonance state is not an aromatic structure anymore (see Scheme 5.2) [128]. Instead, it resembles the structure of polyacetylene. In fact, the UV-Vis spectra of the solid and base forms of the FPAN sample appear as plateau-shaped, which resemble those of the polyacetylene spectra [162], providing further evidence for the meta-coupling of polymerization product. curve is the salt form of FPAN; the dashed curve is the base form of FPAN. of solid form The base the isFPAN. of curve forms dashed salt and the base FPAN; of both of form salt spectra the is solution curve UV-Vis 5.3: Figure

Absorbance 0.2 4 . 0 0.6 0.8 0 1 0 5 9 0 5 8 0 5 7 0 5 6 0 5 5 0 5 4 0 5 3 0 5 2 Wavelength / nm / Wavelength 160 161

6 - 6-6

(a) (b) (c)

Scheme 5.2: Resonance structures of amine unit in FPAN sample.

As a consequence, the aromatic 7r — 7r* transition decreases. At the same time the intraband transition may be more activated (in the defect band) than the inter- band transition. Therefore their total contribution will result in a flat absorption throughout the UV-Vis and near IR regions.

5.4.5 FT-IR Spectra

FT-IR spectra of the base and the salt form of FPAN are shown in Fig. 5.5 together for the purpose of comparison. The spectrum of FPAN salt is also compared with PANI-HC1 in Fig. 5.6. The following conclusions are obtained as the result of the comparisons. First, the ratio of the intensity of the peak at 1,600 cm-1 to that at 1,500 cm-1 is slightly lower in FPAN salt than in FPAN base, consistent with the conventional explanation that doping converts the imine nitrogen to iminium structure. However, this characterization might be a risky one, and will be discussed curve is the salt form of FPAN film; the bottom curve is the base form of FPAN film. FPAN of form base the is curve bottom the film; FPAN of form salt the is curve Figure 5.4: UV-Vis solid spectra of both base and salt form FPAN films: the top top the films: FPAN form salt and base both of spectra solid UV-Vis 5.4: Figure

Absorbance 4 . 0 0.8 200 0 0 6 1000 Wavelength / nm / Wavelength 1400 0 0 8 1 2200 0 0 6 2 162 h dse crei fr PN at te oi crei fr PN base. FPAN for is curve solid the salt; FPAN for is curve dashed the Figure 5.5: FT-IR spectrum for FPAN base powder along with th at for FPAN salt: salt: FPAN for at th with along powder base FPAN for spectrum FT-IR 5.5: Figure

A b so rb an ce 2000 1600 s/cm'1 m c / rs e b m u n e v a W 1200 800 400 163 h ahd uv i fr PN at te oi crei fr AI salt. PANI for is curve solid the salt; FPAN for is curve dashed the Figure 5.6: FT-IR spectrum for FPAN salt powder along with that of salt form PANI: PANI: form salt of that with along powder salt FPAN for spectrum FT-IR 5.6: Figure

Absorbance (arbitrary unit) 2000 1600 W avenum bers / cm / bers avenum W *i i» i J* \J * \J 20400 1200 1 ' \ 800 164 165 under the heading of “observation of resonance phenomenon in FTIR” that follows. Second, the broad absorption (from 1,000 to 1,300 cm-1) increases considerably in FPAN salt compared to that in FPAN base, suggesting that this broad band is related to the vibration of the doped imine groups. Thirdly, the presence of significantly more vibrational bands in FPAN spectra compared to those in parent polyaniline spectra is consistent with the asymmetry of the FPAN sample, in the sense of its odd number of fluorine substituents on Ca ring and its meia-linkage of the polymer backbone.

The observation of the resonance phenomenon in FTIR spectroscopy

The oscillations of the resonance structures of amine and its counterpart, iminium, have been repeatedly implied in FTIR spectra in tri-fluorinated polyaniline (the phe­ nomenon lasted for days in freshly pressed pellets of freshly prepared samples). In Fig. 5.7, a stack of one-scan spectra of undoped FPAN is shown for tri-fluorinated polyaniline in KBr powders. A series of 64-scan spectra of the same sample pellet used in one-scan experiment reveals the same phenomenon (Fig. 5.8).

In Fig. 5.7, the spectrum taken at t = 0.0 min. (the top spectrum) shows a PNB-like spectrum; at t = 0.11 min the spectrum looks like EB’s; while at t = 0.32 min the spectrum is abnormal: the absorption of the imine-like structure is suddenly intensified for more than 5 times the normal PNB like structure, while that of the amine-like structure remains almost unchanged; at t = 3.2 min the ratio of absorptions of the two bands goes down a little bit but it is still “abnormal.” A similar situation can be seen in Fig. 5.8. However, on the second day one-scan spectra still oscillated while 64-scan spectra remained essentially static (see Fig. 5.9 and Fig. 5.10 for one-scan and 64-scan series, respectively). c a e b o n s r a > iminium, and amine bands. amine and iminium, Figure 5.7: One-scan FT IR spectra shows oscillations of the absorptions of imine, imine, of absorptions the of oscillations shows spectra IR FT One-scan 5.7: Figure 0 0 8 1 u i ...

0 0 0 1 0 0 2 1 0 0 4 1 0 0 6 1 \ s, cm /c 1 , rs e b m u n e v a W A. / . A A A \ i =2, min 8 ,0 2 = e tim tim e - 0 .0 0 min 0 .0 0 - e tim tim e = 0 .3 3 min 3 .3 0 = e tim i = .2 min 0.62 = e tim 0 0 6 0 0 8 166 I 167 6 0 0 A A i ' "■ ' r' r' A AA„. -A J\f\, \ 8 0 0 ■" V _ ^ ■" V V / time = 14.0 min time = 12.0 min time = 4.0 min time = 26.0 min tim e = 0.0 min / / \ / I \ y \ /'-A A A..

Wavenumbers, /am 1 V / v A i i V I I i / N 1800 1600 1400 1200 1000 Figure 5.8: 64-scan FTIR spectra shows oscillations of the absorptions of inline, iminium, and amine bands. r s n fc> c e o < n a 168

time = 13

A -A A Ah ' ,

time = t2

A b s o r \ /* A A I' b a n c e I I

A . - A A

time = tO

sa A.J\ A, . A A A . A , I "T---- 1800 1600 1400 1200 1000 800 600 Wavenumbers, 1/cm

Figure 5.9: Time-dependent variations of absorbances of amine and imine-like struc­ tures of PFAN pellet (24 hrs old) for one-scan series. Time tO = 0.0 min.; tl = 0.32 min.; t2 = 0.71 min.; t3 = 1.20 min. 169

A'VA

tim e = 12 A b s o / \A r \. \ / b a n c time = 11 e

V y ^ I / \ AA„ /I'V A

time = to

M i 1800 1600 1400 1200 1000 800 600 Wavenumbers, 1 /cm

Figure 5.10: Time-dependent FTIR absorbances of FPAN pellet (24 hrs old) for 64-scan series. Time tO = 4 min.; tl = 8 min.; t2 = 12 min.; t3 = 16 min. 170

In order to correlate the two oscillating peaks, the peak positions and intensities of the two bands were carefully studied. It was found that the peak positions roughly remained the same for both bands at all times while the integrated peak area of the iminium like structure varies greatly. The time dependent integrated peak areas of the amine-like and iminium-like bands for the 64-scan series are plotted in Fig. 5.11. Several conclusions can be drawn from the plot: (1) the trend of the variation of the integrated peak intensity of amine-like band is opposite from that of imine-like band, reflecting the complexation nature of the two resonance structures, i.e. amine and iminium resonance structures (secondary aromatic amine absorbs strongly at 1342-1320 and 1315-1250 cm-1; aromatic Schiff base absorbs near 1,630 cm-1; while charge separation stiffens both amines and imines). However, the magnitude of the two changes are not equal; (2) the lifetime of the resonance state is long enough to be observed in IR spectroscopy. (3) The ratio of two bands can not be used to characterize the oxidation state of the FPAN sample.

The rationale for this phenomenon follows the Scheme 5.2 [128]. EWG at ortho or para position will stabilize this resonance (b) or (c), respectively. For FPAN, it is just the case. Therefore the imine-like structure would be dominant which is in accordance with the observations made. The unequal changes in intensities of imine- like and amine-like structures may be explained as follows. It is well known that the intensity of the infrared band will be proportional to the square of the change in the dipole moment with respect to the normal coordinate. The charge separation in (b) and (c) will result in larger changes in the dipole moment when vibrating as compared to (a) without charge separation. As a consequence, the same changes in number of (a) will not result in the same changes in intensity as compared to that of (b) or (c). This might explain why the trends of the changes in intensity for amine-like and imine-like modes are opposite, but the magnitude of the former is quite a bit smaller ue fr 4sa series. 64-scan for tures iue .1 Tmedpnetitgae ekitniyo mieadiielk struc­ imine-like and ine am of intensity peak integrated e-dependent Tim 5.11: Figure Integrated Peak Area / (Absorbance.Min) 200 250 300 100 150 350 400 450 0 - 50 0 —o- 10 - 3 C 20 Time / Min / Time 30 40 50 160 - 150 152 154 156 158 162 164 171 172 than that of latter.

It was also observed that all of the spectra show that their oxidation states are close to that of PNB’s while UV-Vis shows it is lower than EB’s. This could possibly be the consequences of the meta-coupled structure, since m-disubstituted benzene has stronger IR absorption around 1,600 cm-1 than that around 1,500 cm-1 [128] due to the averaging of the IR absorptions of the resonances between the amine-like and imine-like structures. While in UV-Vis the polaron or exciton structure is not possible because of the meta-coupling of the FPAN sample.

5.4.6 X-ray Photoelectron Spectra

In the data analysis, the reference binding energy (BE) was set to the core level of C Is neutral component peak of graphite at 284.5 eV in order to compensate for the surface charging effect [163]. The Shirley background was subtracted and satellite peaks were removed for all element peaks before peak synthesis. The vendor-provided iterative curve fit software [163] was used to fit the experimental spectra into components of the Gaussian line shape with Lorentzian broadening function. The surface elemental compositions were determined by the ratios of peak areas corrected with empirical sensitivity factors. All the individual spectra were smoothed by the vendor-provided three-point averaging routine.

X-ray photoelectron spectroscopy (XPS) analyses are consistent with the element chemical analysis and MS results. A typical XPS analysis of the experimental atomic concentrations of FPAN base is as follows: Anal. Found: C, 47.0; N, 8.71; F, 34.3; O,

9.96; total, 99.97 %. Calc, for [(C6HF3)NHx](CH20)o.3: total, 99.88 %; hydrogen is 173

Table 5.2: The binding energies (unit in ‘eV’) and the atomic concentration in paren­ theses (conc., unit in ‘area %’) of the charge corrected XPS components of C Is, N Is, and F Is peaks in FPAN sample.

O rbital Component 1 / conc. Component 2 / conc. Com ponent 3 / conc. N Is 399.6 (34) 398.1 (60) 396.3 (6) C Is 291.0 (29) 285.4 (48) 284.5 (23) F Is 687.9 (66) 685.5 (34) added to balance the oxidation state of each element. A small excess of carbon and oxygen, (CH20)o.3, is assumed to be carbohydrate deposited on the sample surface during XPS sample preparation [163]. The discrepancy of the carbon element between XPS and element chemical analyses occurs for the following reason: XPS is only sensitive to surface composition, while the element chemical analysis is also sensitive to bulk composition.

In addition to the atomic ratio reported above, the XPS multi-scan spectra pro­ vide us some other insight into the structure of FPAN. The charge-corrected binding energies for the components fit in each element as listed in Table 5.2 below.

The charge correction is -7.3 eV, obtained from the bonding energy difference between those of the reference peak (graphic C Is neutral component) and the lower C Is component peak of FPAN. This correction is much bigger than that of EB-I (~ -2 eV) and SPAN base (~ -3.4 eV), showing the very strong electron withdrawing (EW) effect of fluorine atoms. It was found that the N Is multiplet (Fig. 5.12) has three component peaks, which is reminiscent of the N Is spectrum of EB. The main C Is peak is decomposed into two component peaks (see Fig. 5.13) while its satellite peak, found at much higher bonding energy (6.1 eV higher), is assigned as the tt bond shake- up peak. As a comparison, the 7r bond satellite for C Is in polystyrene is above 6.7 eV iue .2 XS I setu fFA bs i ft no he cmpnn peaks. ponent com three into fit is base FPAN of spectrum Is N XPS 5.12: Figure

k Counts 10 12 4 0 2 6 8 404

400 Binding Energy / eV / Energy Binding

396

392 174 175 higher than the main peak [163]. The shift of the N Is peak of FPAN would be mainly related to the static charge effect because FPAN base is essentially an insulator, while a bigger shift of the carbon peak was anticipated because there existed not only the static charge effect, but also the EW effect of the attached fluorine atoms. The fact that these two elements have essentially the same displacement in their binding energies (the N Is referenced against the ES’s while the C Is referenced against the graphite’s) suggests that not only nitrogen cations could resonate its positive charge to fluorines [see Scheme 5.3 (a), (b), and (c)] but also fluorines could propagate their EW effect onto nitrogen cations [see Scheme 5.3 (d)] via their adjacent carbon double bond. As a consequence, the amine nitrogen atoms are imine-like. Therefore the fact that the imine-like nitrogen takes 66 % while amine-like nitrogen takes only 34 % could be readily explained. The explanation follows the imine-like assumption above.

This result is consistent with FABMS results. In the FABMS experiment, the high energy particle hit on N-H bonds and extract hydrogen atoms away. Since fluorinated carbon radicals are more stable than unfluorinated ones, the hydrogen extraction is more competitive with electron ionization of a nitrogen atom for FPAN than that for PANI during FABMS data collection. As a consequence, amine may be oxidized to imine moities during the data collection.

Another important feature comes from F Is multiplet spectra. It can be seen from Fig. 5.14 that F Is peak can be decomposed into two components. The ratio of the area of the peak with the higher binding energy to that with the lower binding energy is ~ 2:1. Therefore, from the fact that three fluorines are attached to each of the

Cfl ring units, we could assign two fluorines to the higher binding energy component and one to the lower one. This 2:1 ratio implies that the polymerization product is a mefo-coupled FPAN rather than a paro-coupled FPAN. The rationale follows the proposed Scheme 5.3 below. component peaks while the shake-up peak is fit into one component. one into fit is peak shake-up the while peaks component Figure 5.13: XPS C Is spectrum of FPAN base. The main peak is fit into two two into fit is peak main The base. FPAN of spectrum Is C XPS 5.13: Figure

k Counts 20 25 0 0 3 0 9 2 5 9 2 Binding Energy / eV Energy Binding 5 8 2 5 7 2 0 8 2 176 rto faes 2:1. ~ areas of ratio a Figure 5.14: XPS F Is spectrum of FPAN base is fit into two component peaks having having peaks component two into fit is base FPAN of spectrum Is F XPS 5.14: Figure

k Counts 0 4 20 5 4 15 4 25 30 0 0 7 5 9 6 Binding Energy / EnergyeV Binding/ 0 9 6 5 8 6 680 5 7 6 177 178

H -r/ f

(a) H H

(IB) H ------f / F

' f t * (ID)

(b) nn

(IE) (IF)

(1G) (1H)

c F F

(ID (1J)

+F—H

Scheme 5.3: Resonance structures of partially oxidized FPAN sample. 179

It is seen from Scheme 5.3 (a) that the positive charge l+ ’ on nitrogen atom site ‘71 can resonate onto carbon sites ‘2’, ‘4’, and ‘6’ but never onto sites ‘1’, ‘3’, and

‘5’ within aromatic ring. The same holds true when a positive charge resonates from nitrogen atom site ‘8’. This positive charge on the carbon site can resonate onto its adjacent fluorine atom as shown in Scheme 5.3 (c). It is obvious that this result is independent on the positioning of the double bonds within the Kekule ring. Therefore meto-coupled fluorinated polyaniline results in a 2:1 fluorine population as long as the positive charge delocalizes fast enough along the partially oxidized polymer backbone. On the other hand, the para-structured one, assuming the random loss of one fluorine atom on a repeat unit, results in a 1:1 fluorine population as shown in Scheme 5.3 (b).

It is interesting to compare the charge-corrected binding energy of the F Is com­ ponent at higher binding energy in FPAN, ~ 687.9 eV, with that of p — (CF = CF), 688.7 ~ 689.2 eV. These values are rather close, suggesting that the fluorines in the two polymers have similar chemical environments, consistent with the mechanism pro­ posed above (even the small energy difference could be explained if the incorporation of electron rich nitrogen atoms into the polymer conjugation chain, which certainly moves binding energy downward, is considered).

5.4.7 Solution NMR Spectra of FPAN

Both 1-D and 2-D solution NMR experiments have been performed on an FPAN sample for structural determination purposes. Unless stated explicitly, the resolution for all the 1-D spectra is ~ 0.5 Hz/pt, while the resolution for the projections on 2-D 180

NMR spectra are usually lower (about several Hz/pt). From 1-D lH-decoupled 19F NMR it is found that the chemical shifts of the monomer, tetra-fluorinated aniline

[see Scheme 5.1 (C) for its chemical structure], is at 8 = -144 (8 peaks) and at

8 = -164 (6 peaks) with coupling constant J f^f* — ~24 Hz, J f7fs = ~12 Hz, and JFiFa = ~7 Hz (we will omit the symbol “5” for chemical shift and the phrase “peaks” in parentheses thereafter). Therefore for a fluorine substituted benzene ring with less than four fluorine atoms, the multiplicity of a substituted fluorine atom should be less or equal to four (22!) under our controlled experiment conditions where the long distant coupling is not emphasized [80]. While 19F-coupled *H spectrum of the monomer has an aromatic peak at 6.5 (9; a triplet of triplets with J haf 3 — 10.8 and

Jha Fi = 7.3), a broad singlet amine proton peak at 5.4 whose integral intensity is equal to twice as much as that of the multiplet at 6.5, consistent with the ratio of the amine protons to that attached to ring unit. Solvent peaks are at 2.49 (DMSO) and

3.45 (H2O). All of these set up a criteria for the further analysis of tri-fluorinated polyaniline.

In the sections that follow, we will establish several objectives. First we will try to provide the evidence that supports the existence of an imine-like resonance structure. Second, we will demonstrate that the fluorine-to-ring ratio is 3:1 to confirm our XPS and the elemental chemical analysis results. Third, we will try to present the evidence for meta-coupled structure. Finally, we will estimate the oxidation level from the ratio of the imine and amine protons. In the imine-like resonance structure, the aromatic ring maps into olefinic structure. Therefore our first objective can be accomplished by showing the presence of olefinic proton or fluorine peaks. On the other hand, the proof of meta-coupled structure will be relatively difficult: it is necessary to identify the four major clusters of 19F peaks for the corresponding four groups of chemically equivalent fluorine atoms, as pointed out in section 5.4.1. This will be accomplished 181 by analyzing many 1-D and 2-D NMR spectra in the section that follows.

Assignment of Amine and Imine Proton Peaks

Deuterium, like any other nuclei with nucleus spin N > b is quadrupolar. Its re­ laxation depends on the interaction of the quadrupole moment with the electric field gradient. Though the quadrupole moment of 2H is small, it still relaxes rapidly in a non-cubic environment where the electric field gradient is quite large. As a conse­ quence, all the otherwise resolved couplings to it are lost. This effect, in combination with its fast chemical exchange with imine or amine protons as well as with other effects (see discussion below), makes the linewidth of proton signals too broad to be observed.

The quadrupole relaxation is also sensitive to J-time scale, rc. If the exchange rate is greater than the frequency difference between two resonant peaks, kex > 7tAi//\/2 = 2.22J, then its multiplet will coalesce to a singlet. It is then said to be exchange decoupled. We will apply this concept in the assignment of l9F NMR spectra.

The line-width broadening of protons adjacent to quadrupolar nuclei is not ex­ clusively a result of exchange decoupling. The other possible interpretation is that the scalar relaxation of the second kind, i.e. the rapid spin-flips of the quadrupo­ lar nucleus, may provide an efficient T2 relaxation mechanism for the incoherently broadened lines (due to smaller T2 or shorter lifetimes).

In Fig. 5.15, the spectrum of FPAN in deuterated DMSO with and without adding a few drops of D20 are shown in (a) and (b), respectively. The disappeared 182

(a)

p-,------,------,------1 , 1 | T HO 7 5 7 0 6 5 60 5 5 50 PPM

I 11 I ...... I 1 1 1 'I1 > 11 | i i, «„r BOO 8.50 H.00 7.50 7.00 6.50 E.OOPPM 5 00

Figure 5.15: Comparison of 1-D solution NMR spectra of FPAN sample: (a) with and (b) without addition of DzO into polymer solutions. The spectrum (b) has been taken with higher resolution. 183 peaks at 6.25 and 6.60 in spectrum 5.15 (b) can therefore be assigned to amine protons while the peaks in the range of 8.60~9.40 are assigned to those connected to imine atoms. It can be seen that (1) the integrated peak areas of those imine and amine atoms are apparently comparable and (2) the proton connected to the amine or imine atom has more than one peaks with different chemical shift values. This state that (1) the oxidation state of the pristine form of FPAN is apparently comparable to that of EB and (2) there exists some sort of chemical exchange which results in multiple nitrogen proton peaks at different chemical shift locations.

Imine-like to Amine-like Resonance Structure and m-coupling

It is quite characteristic for 19F NMR [81] that (1) there is a very extensive coupling between most of the 19F atoms, e.g. the aromatic fluorine atoms and the fluorine atoms on the side chain, and (2) the shift range of olefinic fluorine is quite consid­ erable, i.e. from -126 to -162, as compared with that of the aromatic fluorine from -132 to -174. It has also been observed that the size of the coupling constant usually decreases in the order “ortho > para > mefa.” On the other hand, the chemical shift range of olefinic protons is within a fairly characteristic range (5 = 4.6 ~ 5.9 which differs from 8 = 6 ~ 8.5 for aromatic protons [169]). It is also true for par­ tially fluorinated aromatics that the coupling constant between hydrogen and fluorine atoms is remarkably constant, being in the following range: J hf ,ortho = 8 ~ 12 Hz,

J hF,meta — 5 ~ 8 Hz, and J hf ,para = 1.5 ~ 2.5 Hz. Therefore it is a better idea to ascertain the presence of olefinic structure by 1-D *H NMR or by 2-D NMR instead of by 1-D 19F NMR alone. Both the 1-D *H NMR spectrum (in Fig. 5.15) and the 2-D 1H-19F correlated NMR spectrum (in Fig. 5.16) show two proton peaks within the 184

Figure 5.16: 2-D 1H-19F correlated solution NMR of FPAN sample. 185 olefinic proton shift range (4.6, 5.9), which indicates the existence of olefinic struc­ ture and therefore the imine-like resonance structure. The existence of two olefinic proton peaks states that the proton environment is asymmetric, possibly suggesting the meta-coupling of the polymer backbone. It is also found that the sum of the integrated intensities of the aromatic and the olefinic proton peaks is more or less the same as that of the amine and imine atoms, consistent with the chemistry stoichiom­ etry of m eta or otfior-coupled fluorinated polyaniline. It is also interesting to note that the “imine-like" proton (~ 8.8) is about 2.3 ppm higher than the “amine-like” proton (~ 6.5), being about the same as the difference between chemical shifts of aromatic proton (~ 7.6) and olefinic proton (~ 5.4). This implies the possible cou­ pling between the resonance of aromatic proton and that of protons attached on the nitrogen atom, that is, the coherence of the changes on both the nitrogen protons and the ring protons, a further evidence for amine-like to imine-like resonances.

F/N Ratio and meta-coupling

The 1H-decoupled 19F spectra are shown in Fig. 5.17. Spectrum (b) is the whole spectrum while spectrum (a) is a section plot of an expanded region of spectrum (b) around 5 = 139 to 142. The spectra were taken in a deuterated DMSO solvent as usual, with the addition of a small amount of deuterated H20 to eliminate the couplings through the nitrogen proton bridges. The peaks shown in (b) are labeled in Table 5.3.

It is found that most of the peaks in the 1H-decoupled 19F spectrum are quartets, while in the 1H-coupled spectrum the number for the multiplet is increased because of the H — F coupling. This fact indicates that most of the fluorine atoms are coupled 186

Figure 5.17: 1-D 'H-decoupled 19F solution NMR of FPAN sample: the spectrum (a) is an expanded region of the spectrum (b), a whole spectrum of 1H-decoupled 18F spectrum. 187

Table 5.3: Peak labelling for ^-decoupled 19F NMR spectrum.

Peak label Chemical shift Multiplet Peak label Chemical shift M ultiplet 1 -138.3 4 10 -145.6 4 2 -138.8 4 11 -146.1 4 3 -139.3 4 12 -147.8 4 4 -139.9 4 13 -148.8 1 5 -140.2 4 14 -150.5 4 6 -140.7 4 15 -151.1 1 7 -141.4 4 16 -152.6 4 8 -142.6 4 17 -154,2 4 9 -144.2 4

with aromatic hydrogen atoms on the same ring (as only a short distance interaction exists due to the lack of amine proton bridge when adding deuteroxide), and every fluorine atom interacts with two other magnetically non-equivalent fluorine atoms so that it results a quartet (22). It is also found that there are four major groups of 19F peaks. Therefore all of these observations provide essential supporting evidence for the polymerization mechanism proposed in section 5.4.1, i.e. the polymer backbone is mefa-coupled, and on every ring unit there are three fluorine, one hydrogen, and two nitrogen atoms. Further assignment of these four major groups of peaks and the other minor signals will be addressed below.

Further Structural Determination with Solution 2-D NMR Spectra

COSY spectra of FPAN sample

As is well known, the chemical bond connectivity information can be obtained 188

JL

■ H* •

i*

•

9 a

o ■

g • ■* y I *

y ■ • %

■ II m

p * ■" Q r ■ 1 f a . ■ / • m

m

- 1 39.0 - 1 4 0 .0 - 1 49.0 - 190.0 - 199.0 - 100.0 PPM

Figure 5.18: 2-D 1BF COSY16 solution NMR of FPAN sample. 189 from COSY type of spectra [84]. From the correlations found in the COSY 1H-19F spectrum (Fig. 5.16), it is seen that the major 19F peaks are divided into different categories: the first subset (peak ‘4’ and ‘10’) is correlated with the 1H peak at 7.9 ppm (higher oxidation state), the second subset (peaks ‘6’ and ‘16’) is with the peak at 7.6 ppm (lower oxidation state), while the peaks ‘13’ and ‘15’ are with the 1H peaks at 5.2 and 5.7 ppm (the signals from the olefinic protons), respectively. Therefore from the 1H peak assignment above, the first subset of the 19F peaks is related to the fluorine atoms on the ring where *H is down field, the second subset is to the ring where the is relatively up field, and the other peaks are related to resonance (non-aromatic ring) structures. However, from XPS and the elemental chemical analysis results, it is known that three peaks are co-existent on every single ring unit. The interpretation for the 1H-19F NMR observation based on the known fact of 3:1 F/H ratio, as explained below, provides further supporting evidence for the m eta -coupling of the polymer backbone of the FPAN sample. In the meta­ coupling assumption (see Scheme 5.3), two fluorine atoms are vicinal to each other, and one of them is adjacent to the proton atom. On the other hand, the third one is isolated from the two vicinal fluorine atoms and from the proton by two of their adjacent nitrogen atoms. Therefore the coupling of the third fluorine to the proton (4-bond distance) is much weakened by the inactive nucleus surrounding of the third proton as compared with those between the other two fluorine atoms and the proton

(the interaction is either direct or indirect via the spin diffusion mechanism). As a consequence, only two fluorine signals can be found to correlate with proton peak in the 1H-19F spectrum under the assumption of meta-coupling. On the other hand, in ortho-coupling environment, no fluorine atoms are isolated by two nitrogen atoms, therefore three fluorine peaks are expected to correlate with the peak, which contradicts the experiment result. Therefore, the minor possibility of ortho-coupling 190

Table 5.4: Correlation table for the COSY and the NOESY spectra of PPAN sample.

Peak 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 1 C 2 C C 3 N c 4(m) N Cn 5 C 6(m) N Cn 7 N C 8 C 9 C 10(m) Cn N C 11 C N 12 c 13(m) N 14 C 15 N 16(m) Cn C N 17 CN is eliminated. This argument will be reinforced by the analysis of the 19F-19F COSY spectrum and NOESY spectrum in later sections.

For the sake of the visualization of the correlations in the COSY and NOESY spectra, we label the correlations (i.e. the cross peaks in both the NOESY and the COSY spectra shown) among the 19F signals in Table 5.4. In the table, we used symbol “C” and “N” (or “n” if the cross peak appearing in the NOESY spectrum is a ‘smaller’ type) to denote the appearance of a cross peak in the COSY and the NOESY spectra, respectively.

In the 19F-19F COSY spectrum (Fig. 5.18), both major peaks ‘4’ and ‘10’ and 191 major peaks ‘6’ and ‘16’ are correlated in pairs, consistent with the aH-19F COSY spectrum discussed above. In contrast, the correlations among three fluorine atoms (correlated peaks ‘10’ and ‘16’ are served as bridges) do show up due to the stronger 19F-19F interaction (the greater the 19F-19F coupling constant), i.e. the couplings of the isolated fluorine atom with other two fluorine atoms showed up as the cross peaks in the 19F-19F COSY spectrum. Therefore this fact might also provide convincing evidence for meta-linkage of the polymer backbone.

NOESY spectrum of FPAN sample

The efforts of acquiring the stereo-conformation (spacial correlation) and the chemical exchange information have been made from analysis of NOESY type of spectra [83]. In the case of chemical exchange related NOESY peaks, the lifetimes of the exchange-related states may be different. One resonance may be more stable

(lower energy, longer lived) than the other’s. The ratio of their intensities varies (it could be any positive value) depending on the difference of their lifetimes. On the other hand, for the spatial related NOESY peaks, the intensity ratio of the correlated peaks must agree with the ratio of the number of equivalent fluorine atoms in the correlated peak clusters. As a consequence the ratio of the spatial correlated NOESY peaks is limited by the covalent bonding capacity of a carbon atom, i.e. 1/4 to 1/1. Therefore the resonance related peaks and spatial distance related peaks could be possibly distinguished within the NOESY type of NMR spectra by the intensity ratio of the correlated peaks (e.g. Fig. 5.19).

There are many possible types of chemical exchanges. In our case, they are suggested as following: (1) Imine-amine resonance related chemical exchange; (2) charge or free radical transport related chemical exchange; (3) regular-to-irregular chain configuration transition related chemical exchange (similar to ‘cis-trans’ transi- 192

A,

0 - 154.0

- 152.0

- 150.0

- 145.0

- 142.0

PPM T T T T - 140.0 - 142.0 - 144.0 - 146.0 - 140.0 - 150.0 - 152.0 -154 0 PPM

Figure 5.19: 2-D solution 19F phase sensitive NOESY NMR of the FPAN sample. 193

tion). Many experiments need to be designed for making an unambiguous assignment on this matter.

It is seen from phase sensitive NOESY spectrum that every major peak is as­ sociated with a smaller peak about one-tenth the size of the major one (in terms of a cross peak, labelled with ‘N’ in Table 5.4). Therefore, based on the discussion above, it is understood that the major peaks and their corresponding mirror peaks are chemical exchange related. This probably originated from resonance structure related chemical exchange. It is also found that major peaks are correlated (in terms of smaller cross peaks, labelled with ‘n’ in Table 5.4), coincident with the correlations in the COSY spectrum (Fig. 5.18), therefore we can conclude that peaks ‘4’ and ’10’ are correlated spatially therefore they are from adjacent pair of fluorine atoms, and so are the peaks ‘6’ and ‘16’.

In addition, two shape singlets are correlated in the NOESY spectrum, in contrast with the situation in the 19F-19F COSY spectrum where they are not correlated and with the situation in the 1H-19F COSY spectrum where two shape peaks are not correlated with a single 1H peak. This fact clearly states that these two singlets are neither from two fluorine atoms on the same ring nor on the separate ring unit, when the ratio of the intensities of the two signals and when above contradiction situations are considered. On the other hand, in both the COSY (the 1H-19F and the 19F-19F COSY) spectra and the NOESY spectrum, the major peaks ‘4’ and ‘6’ are not correlated. Therefore neither peaks ‘4’ and ‘6’ nor peaks ‘10’ and ‘16’ should be on the same ring unit. This can also be seen from the fact that peaks ‘4’ and TO’ and the peaks ‘6* and ‘16’ are grouped by two different protons in the 1H-19F COSY spectrum, respectively.

Now question arises; How to explain that the peaks ‘10’ and ‘16’ are correlated 194

in 19F-19F COSY spectrum if they are not on the same ring? To account for this, we must make assumptions that each of the two peaks ‘10’ and ‘16’ is a composite peak composed of two magnetically equivalent l9F signals. In fact, this assumption is consistent with the proposed meto-coupled structure (A) in Scheme 5.1. Therefore the interpretation above also provides further supporting evidence for the meia-coupling

mechanism.

Now let us discuss the correlations related to the minor peaks. It is seen that the minor peaks, like their mirror images (the major peaks), are also correlated among themselves, i.e. peaks ‘3’ and ‘11’ are correlated, so are peaks ‘7’ and ‘17’. This forms a one-to-one mapping of correlations between major and minor peaks. Therefore three

of our 2-D NMR spectra are self-consistent. All of these provide further supporting evidence for a resonance related peak segregation. The rationale follows: when a

polymer chain experiences a resonance (an internal redox reaction) as suggested in section 5.4.6, the chemical shifts of all the major peaks are altered. Positive charge

can resonate to m eta position while negative charge will delocalize to para and ortho positions; therefore the signals of fluorine atoms at m eta position will shift to lower field while the peaks at ortho and para positions will move to higher field. It is what we have seen in the NOESY spectrum (Fig. 5.19).

Comment on further evidence of the resonance state and on impurity peaks

It is worthwhile to comment on the fact that the two sharp 19F peaks are corre­ lated with two olefinic peaks in 1H-19F correlated NMR spectrum. Further observa­ tion reveals that when major 19F peak resonates to up-field the correlated JH peak moves to down field. This suggests that the proton and the fluorine atoms are vicinal on a double carbon-carbon bond. The rationale follows: when a polymer backbone dimerizes into a hetero-polyacetylene-like conjugated system, the partially fluorinated 195 double bond is polarized because of the greater electronegativity of the fluorine atom. Therefore the 1H atom feels somewhat decreased electron screening while the 19F atom feels somewhat increased electron screening. As a consequence, the *H is down field while the 19F is up field.

From the correlation table (Table 5.4), it is seen that the minor peaks ‘1’, ‘2’, ‘5’, ‘8’, ’9’, ‘12’, and ‘14’ are correlated among themselves but not correlated with any of the major peaks in both the COSY and the NOESY spectra. Therefore these peaks are minor impurities arising from other polymerization pathways, or from having a different oxidation state, or from being a defluorinated FPAN sample, etc. However, these signals are negligible because their total intensity is up to only a couple of percent of the total fluorine signal of the major FPAN component.

In summary, the analyses of 1-D and 2-D solution NMR spectra provide convinc­ ing evidence for the assumption of meta-coupling of polymer backbone, the 3:1 ratio of fluorine to ring unit, polaron-like (~ 60 % oxidized) lattice structure, the resonance assumption, and the other rich structural information. All of these conclusions are consistent with other spectroscopic and the elemental chemical analysis results. How­ ever, since the NMR analysis is complicated by the resonances of chemical structures and possibly other changes, more work (e.g. simulations or more 2-D NMR spectra) needs to be done for further analysis and complete assignment on all the 10F signals. In other words, more work needs to be done to reveal the “regulations” mentioned in section 5.4.1 196

5.4.8 Thermogravimetric Analysis and Differential Thermal A n a ly sis

Both the pristine form and the base form FPAN samples have been subjected to thermal gravimetric analyses and differential thermal analyses (the spectrum for the pristine FPAN is in Fig. 5.20, while that for base is in Fig. 5.21). Good thermal stability is shown in the temperature range of 40° — 140°C. The weight loss is ~ 4 % and 2 % for the pristine and the base form powders, respectively, which are probably due to desorption of the absorbed water and other absorbent. Consistent with TGA and DTA data, the pyrolysis gas chromatography (PGC) do not show any noticeable decomposition (see Fig. 5.22 and Fig. 5.23). The only noise is seen in the FT-IR spectra of the final fluent.

5.4.9 DC Conductivity measurement

DC conductivity of the spin-coated films on quartz substrates was estimated as ~ 0.2 S/cm. It was found that the contact resistance was less when the gold wires were pressed onto the film surface compared to when they were glued onto it with either “silver paste” or “electric dye.” A plausible explanation could be that FPAN might react with these adhesive materials. Further electrical transport study is undertaken on FPAN material. uv s o T hl te oi crei fr DTA. for is curve solid the while A TG for is curve Figure 5.20: TGA and DTA spectra for FPAN powder in pristine form: the dashed dashed the form: pristine in powder FPAN for spectra DTA and TGA 5.20: Figure

DTG %/min 3 2.0 -< 4 6 5 0.5 3.5 5.0 35 60 ep° (Heating) °C Temp 85 110 100.2 96.8 98.0 99.1

TG % 197 Figure 5.21: TGA and DTA spectra for FPAN base powder: the dashed curve is for for is curve dashed the powder: base DTA. FPAN for is for curve solid spectra the DTA while and TGA TGA 5.21: Figure

DTG %/min Q -1.25 -0.45 0.35 1.15 585 35 60 Temp °C (Heating) °C Temp 110 100.31.95 97.9 98.5 99.1 99.7

TG % 198 Figure 5.22: Pyrolysis gas chromatogram for the pristine form of FPAN. of form pristine the for chromatogram gas Pyrolysis 5.22: Figure Chromatogram - 3 1 0 0 . 0 4 1 0 0 . 0 5 1 0 0 . 0 8 1 0 0 . 0 9 1 0 0 . 0 0 0.0021 0 . . 0020 0022 - . - 0 2 5 1 0 1 Time (min) Time 199 Chromatogram 0.0012 - 4 1 0 0 . 0 “ 6 1 0 0 . 0 - 8 1 0 0 . 0 0.0022 “I 0.0022 Figure 5.23: Pyrolysis gas chromatogram for the base form of FPAN. of form base the for chromatogram gas Pyrolysis 5.23: Figure 0 . 0020 - - Time(min) 200 201

5.4.10 Electron Paramagnetic Resonance Analysis

The room temperature EPR experiment has been performed on both the salt and the base forms of FPAN. Two overlapped signals, the broad one with AH ~ 15 Gauss and the sharp one with AH ~ 0.4 Gauss, were observed in an FPAN salt sample. The sharp signal of FPAN salt disappeared after deprotonation, which is demonstrated in Fig. 5.24. The signal intensity was calibrated against DHPP and the total signal intensity of FPAN salt was estimated as one spin per 666 ring units at room temperature (the intensity of the sharp peak is only ~ 1/10 of the broad one).

It is suggested that the sharp signal is from polaron that is responsible for electric transport, while the broad one is assumed to be an intrinsic signal arising from the meta-coupled units.

5.4.11 Cyclic Voltammetry Experiment

The cyclic voltammegram (CV) of FPAN (see Fig. 5.25) is similar in shape to that of the electrochemically made sample (see Fig. 5.26) [160]. However, the two major sets of redox peaks occur at somewhat higher oxidation potentials, i.e. 0.1 V higher than those in the CV spectrum of the electrochemically made sample. This could be explained by differences in the morphology of the films, as one was spin-coated onto the electrode (more amorphous) while the other was made in situ (more crystalline). salt. Figure 5.24: Comparison of room temperature EPR spectrums of FPAN base and and base FPAN of spectrums EPR temperature room of Comparison 5.24: Figure Derivative Signal (arb) 0 9 3 3 0 8 3 3 0 7 3 3 0 6 3 3 0 5 3 3 0 4 3 3 0 3 3 3 Gauss) (G H FPAN Salt Salt FPAN PNBase B FPAN 202 203

20 mA

o

0.0 0.5 1.0 1.5 Potential / Volts

Figure 5.25: Cyclic voltammogram of FPAN in 1.000 M HC1. The potential scan rate is 50 mV/sec. 204

2 : LU 20yA QC OC 3 o

00 0-5 10 POTENTIAL / VOLTS

Figure 5.26: Cyclic voltammogram of FPAN made electrochemically (after Cassity). 205

5.5 Summary

In this chapter the synthesis of highly fluorinated polyaniline via chemistry route was reported. Many physical properties such as thermal property, electrochemistry, electric transport, and magnetic resonance have been reported. The oscillation phe­ nomena in FTIR and temperature dependent EPR experiments have been reported and a model for them was proposed to interpret this topic well. Besides, the same model has been used successfully to resolve seemingly contradiction between UV-Vis and FTIR together with XPS. The success of this model itself provides a convince assignment of FPAN as meta-coupled polymerization product. The solution 1-D and 2-D NMR analyses further support this assignment and provide more insight to the chemical structural and physical property knowledge about FPAN sample. More op­ tical and transport studies are now being undertaken to gain further knowledge of this new material, with aims toward future practical applications.

5.6 Acknowledgment

This work has been supported in part by the US Office of Naval Research. CH APTER VI

Conclusions

In this dissertation, studies on some new conducting polyanilines have been reported. It covers the syntheses, characterizations, and the physical property studies of the highly fluorinated and the highly sulfonated polyaniline, as well as the successful proposal of the novel quasi-random oxidation model and the related computer simu­ lations. These are summarized in the sections those follow.

6.1 Highly Sulfonated Polyaniline

6.1.1 Synthesis

A new synthetic route of SPAN has been described as a way to make SPAN with higher molecular weight and better physical properties. A sulfur to nitrogen ratio as high as 50 % greater than those of EB-SPAN has been achieved.

206 207

Some important improvements on the reaction conditions are: (1) the solvent of

the sulfonation reaction is changed from MeOH to H 2O; (2) less fuming sulfuric acid (relative to starting material) is used; (3) a reduced form of PANI is used instead of EB, reducing the possible hydrolysis and other side reactions. All of these changes warrantee our synthesis to produce a polymer with higher molecular weight, and therefore a longer conjugation length. Additionally, the synthesis proceeds more easily, produces more in a single batch, and needs less post-treatment. Notice that

the elimination of the potential for contamination of the environment methanol would greatly reduce the cost of potential industry applications.

6.1.2 Physical properties

The high S/N ratio results in various important consequences. They are the following: (1) higher DC conductivity as high as ~ 1 S/cm was obtained; (2) higher density of states at Fermi energy level was analyzed as a self-consistent result; (3) better temperature dependence of DC conductivity with lower electric transport activation energy (small To) was achieved; (4) novel pH dependence of half-wave potential is shown as a result of less hydrolysis during the electrochemical oxidation process; (5) novel pH dependence conductivity was observed as nearly pH independent in a broader range than that for traditional SPAN and that for PANI; (6) different FTIR and UV-Vis spectra are consistent with higher S/N ratio. 208

6.1.3 Quasi-random Oxidation Model

This proposal of the Quasi-random, Oxidation Model aims to interpret the different CV behavior among EB-SPAN, LEB-SPAN, and PANI samples as well as the earlier extensively reported but unexplained in situ CV-dependent DC conductivity and EPR spectra of the PANI sample. The successful simulations of the in situ spectra based on our model provide plausible interpretations for this fascinating mystery that has puzzled scientists world wide. Discovery of the proportionality between the simulated Pauli spin density and the CV-dependent DC conductivity provides convincing evidence for the formation of the polaron spin lattice. As a consequence, it is expected to effectively end the standoff between the polaron lattice and the bipolaron conducting mechanisms. This model also explains the origin of the defect states and the solvent as well as hydrolysis effects on the in situ spectra.

6.1.4 Electrochemistry

The different pH-dependence of CV spectra for PANI, EB-SPAN, and LEB-SPAN has been explained based on the Quasi-random Oxidation Model. Oxidation-hydrolysis was found to have considerable influence on in situ CV spectra. 209

6.2 Highly Fluorinated Polyaniline

6.2.1 Synthesis

The chemical synthesis of the highly fluorinated polyaniline has been developed dur­ ing my studying at OSU. The first synthetic route intelligently utilizes an ultrasonic bath to break immiscible monomer into aqueous solvent while the second route is based on the increased solubility of the monomer in spectral usage of aqueous sol­ vent. It was found that the higher oxidizer to monomer ratio could be used to raise the polymerization yield of FPAN and other PANI derivatives, which makes FPAN polymerization possible.

6.2.2 Physical properties and characterizations

The following achievements have been made: (1) the thin and transparent conducting FPAN film was cast from various organic solvents (as a matter of fact, it can be cast from most of the organic solvent), whose UV-Vis spectrum showed a long free carrier tail up to 2,400 nm; (2) the FTIR spectrum of the FPAN sample shows some oscillatory phenomenon, which has been interpreted by a proposed mechanism; (3) the powder EPR spectrum shows two overlapped peaks, identified as neutral defect and charged polaron absorptions; (4) the XPS result, consistent with the CV result, shows a much bigger blue shift of binding energies as a consequence of multi-fluorination 210 of the polymer repeat unit; (5) extensive 1-D and 2-D solution NMR studies of the FPAN sample provides supporting evidence for resonance structures, meta-linkage of polymer backbone, and the F/N ratio discovered by various methods listed below.

The proposal of meta-coupled FPAN backbone has been supported by many convincing evidences from the results of the elemental chemical analysis, FABMS, solution 1-D and 2-D NMR, UV-Vis, FTIR, and XPS techniques.

6.2.3 The resonance model for FTIR spectra

The novel phenomena of osculating FTIR has been revealed, studied, and interpreted by the resonance model proposed in this dissertation. Based on this model, a novel model for synthetic metal was shown.

6.2.4 The suggested future research topic in polyaniline field

During the course of my study, many strange phenomena, such as oscillating IR intensity, repeatedly occured, and therefore discussed in this dissertation. However many were not repeatable (therefore not discussed), for example, a solution EPR signal of ‘PNB’ made via the route developed by myself [161] was seen as a five-peak signal spectrum for the first field scan. But in the second run a three-peak signal spectrum appeared instead. The resonating phenomenon is irreversible and has not been repeated since then. Therefore many rich but hidden messages are waiting to be revealed. I believe the future of the polyaniline research field is still bright and 211 attractive. I am looking forward to hearing exciting news from this field, especially from the lab I am currently working in. Bibliography

[1] M. Jozefowicz, L. T. Yu, G. Belorgey and R. Buvet, J. Polym. Sci.(C), 16, 2931 (1967). [2] Handbook of Conducting Polymers, vol. I and II, ed. by T. A. Skothem (Dekker, New York, 1990). [3] A. G. MacDiarmid, J. C. Chiang, M. Halpern, W. S. Huang, S. L. Mu, N. L. D. Somasiri, W. Wu, and S. I. Yaniger, Mol. Cryst. Liq. Cryst., 121, 173 (1985). [4] T. Ito, H. Shirakawa, S. Ikeda, J. Polym. Sci. Polym. Chem. 12, 11 (1974). [5] C. K. Chiang, C. R. Fincher, Y. W. Park, A. J. Heeger, H. Shirakawa, E. J. Louis, S. C. Gau, and A. G. MacDiarmid, Phys. Rev. Lett. 39, 1098 (1977). [6] L. W. Shacklette, R. R. Chance, D. M. Ivory, D. G. Miller, R.H. Baughman, Synth. Met. 1, 307 (1979). [7] M. G. Kanatzdis, Chem. & Eng. News, 3 (Dec.), 36 (1990). [8] G. E. Wnek, J. C. W. Chen, F. E. Karasz, and C. Peter-Lillya, Polymer. 20, 1441, 1979. [9] K. K. Kanazawa, A. F. Diaz, W. D. Gill, P. M. Grant, G. M. Street, G. D. Garadini, J. F. Kwak, Synth. Met. 1, 329 (1979). [10] Chem. & Eng. News, 3 (Dec., 36 (1990). [11] T. Yamamoto, K. Sanekika, A. Yamamoto, J. Polym. Lett. Ed., 18, 9 (1980). [12] A.G. MacDiarmid et al, to be published. [13] A. J. Epstein, J. Joo, C.-Y. Wu, A. Bebatar, C. F. Faisst, J. Zegarski, and A. G. MacDiarmid, Intrisically Conducting Polymers: An Emerging Technology , ed. by M. Aldissi (Kluwer Academic Publishers, Netherlands, 1993).

212 213

[14] J. Yue and A. J. Epstein, J. Chem. Soc., Chem. Commun., 2 3 , 1540 (1992). [15] B. ViUeret and M. Nechtschein, Phys. Rev. Lett., 63, 1285 (1989).

[16] L. W. Shacklette and N. F. Colaneri, Bull. Am. Phys. Soc., 3 6 ( 3 ) , 427 (1991). [17] A. J. Heeger, D. Moses, and M. Sinclair, ACS Synposium Series, 346, 372 (1987).

[18] T. Nakajima and T. Kawagoe, Synth. Met., 2 8 , 629 (1990). [19] D. De-Berry, J. Electrochem Soc., 132 1022 (1985).

[20] D. Braun and A. J. Heege, Bull. Am. Phys. Soc., 3 6 ( 3 ) , 377 (1991).

[21] H. Reiss, Bull. Am. Phys. Soc., 3 6 ( 3 ) 426 (1991). [22] Allied Signal-Americhem-Zipperline, Press Release (May, 1991). [23] C. Kittel, Introduction to Solid State Physics, 6th Ed (John Wiley &; Sons, New York, 1986). [24] Z.H. Wang, C. Li, E.M. Scherr, A.G. MacDiarmid, and A.J. Epstein. Phys. Rev. Lett. 66, 1745 (1991). [25] N.W. Ashcroft and N.D. Mermin, Solid State Physics (Saunders College Pub­ lishing, New York, 1976). [26] R.L. Carlin, Magnetochemistry (Springer-Verlag, Berlin Heidelberg, 1986). [27] J. Joo, Z. Oblakowski, G. Du, J. P. Pouget, J. M. Wiesinger, E. J. Oh, Y. Min, A. G. MacDiarmid, and A. J. Epstein, Phys. Rev. B. 4 9 , 2977 (1994).

[28] W. R. Salaneck and J. L. Bredas, Synth. Met., 6 7 , 15 (1994).

[29] M.C. Santos and J.L. Bredas, Phys. Rev. Lett., 6 2 , 2499 (1989).

[30] A. J. Epstein, J. M. Ginder, F. Zuo, R. W. Bigelow, H. S. Woo, D. B. Tanner, A. F. Richter, W. S. Huang, and A. G. MacDiarmid, Synth. Met. 18, 307 (1987). [31] J.M. Ginder, A.J. Epstein, and A.G. MacDiarmid, Solid State Commun., 72, 987 (1989). [32] S. Stafstrom and J.L. Bredas, Synth. Met., 14, 297 (1986).

[33] C.B. Duke, E.M. Conwell, and A. Paton, Chem. Phys. Lett., 8 2 , 131 (1986).

[34] W.B. Euler, Solid State Commun., 5 7 , 857 (1986). [35] S. Stafstrom, J. L. Bredas, A. J. Epstein, H. S. Woo, D. B. Tanner, W. S. Huang, and A. G. MacDiarmid, Phys. Rev. Lett., 5 9 , 1464 (1987). 214

[36] H. Suzuki, Electronic Absorption Spectra and Geometry of Organic Molecules (Academic Press, New York, 1967). [37] L.D. Landau and E.M. Lifshitz, Statistical Physics, Vol. 1 &; 2, 3rd ed (Pergamon Press, New York, 1980). [38] G. Baym, Lectures on Quantum Mechanics (Addison-Wesley Publishing Com­ pany, California, 1990). [39] R.B. Seymour and C.E. Carraher, Jr., Polymer Chemistry: An Introduction, 3rd Ed., Marcel Dekker, Inc., New York, 1992. [40] Conducting Polymers, ed. J.R. Ebdon, (Chapman and Hall: New York, NY, 1991).

[41] T. Yasui, Bull. Chem. Soc. Japan, 1 0 , 306 (1935).

[42] N.E. Khomutov and S.V. Gorbachev, Zhur. Fiz. Khim., 2 4 , 1101 (1950).

[43] N.E. Khomtutov and S.V. Gorbachev, Soveshch. Elektrokhim., 1950, Izdat. Akad. Nauk SSSR, Moscow, 1953, pp. 579-585. [44] D.M. Mohilner, R.N. Adams, and W.J. Argersinger, Jr., J. Am. Chem. Soc., 8 4 , 3618 (1962). [45] Y. Wei, X. Tang, Y. Sun, and W. W. Focke, J. Polym. Sci., Polym. Chem. Edn 2 7 , 2385 (1989).

[46] N. Gospodinova, P. Mokreva, and L. Terlemezyan, Polymer, 3 4 , 1330 (1993). [47] N. Gospodinova, L. Terlemezyan, P. Mokreva, and K. Kossev, Polymer, 34, 2434 (1993). [48] N. Gospodinova, P. Mokreva, and L. Terlemezyan, Polymer, 34, 2438 (1993). [49] A. Kitani, M. Kaya, J. Yano, K. Yoshikawa, and K. Sasaki, Synth. Met., 18 341 (1987).

[50] A. Kitani, J. Yano, A. Kunai, and K. Sasaki, J. Electroanal. Chem., 2 2 1 , 69 (1987).

[51] A. Thyssen and J. W. Schultse, Synth. Met., 4 1 - 4 3 , 2825 (1991).

[52] K. Tzou and R. V. Gregory, Synth. Met., 4 7 , 267 (1992).

[53] V. Tsakova, A. Milchev, and J. W. Schultze, J. Electroanal. Chem., 3 4 6 , 85 (1993). [54] K. Sasaki, M. Kaya, J. Yano, A. Kitani, and A. Kunai, J. Electroanal. Chem., 215, 401 (1986).

[55] R. Willstatter and Ch. W. Moore, Chem. Ber., 4 0 , 2665 (1907). 215

[56] R. Willstatter and st. Dorogi, Chem. Ber., 42, 2147 (1909). [57] M. Breitenbach and K.-H. Heckner, Electroanal. Chem. Interf. Electrochem., 43, 267 (1973). [58] M. Breitenbach and K.-H. Heckner, Electroanal. Chem., 29, 309 (1971). [59] M. Breitenbach and K.-H. Heckner, Electroanal. Chem., 33, 45 (1971). [60] L. Dunsch, Electroanal. Chem. Interf. Electrochem., 61, 61 (1975). [61] L. Frank, Polymer, 35, 2915 (1994). [62] R.H. Baughman, J.F. Wolf, H. Eckhardt, L.W. Shackldette, Synth. Met., 121, 25 (1988). [63] A. G. Green and A. E. Woodhead, J. Am. Chem. Soc., 112, 2388 (1910). [64] J. C. Chiang and A. G. MacDiarmid, Synth. Met. 13, 193 (1986). [65] General review. [66] F. Zuo, A. Angelopoulos, A. G. MacDiarmid, and A. J. Epstein, Phys. Rev. B. 36, 3475 (1987). [67] L. Pavesi, S. Aldrovandi, and M. Corti, J. Phys.r Condens. Matter. 5 (su p p l. 34B), B25-B30 (1993). [68] P. N. Adams, A. P. Monkman, and D. C. Apperley, Synth. Met., 55, 725(1993). [69] P. C. Stein, W. L. Earl, and A. Ray, Synth. Met., 55, 702 (1993). [701 A. M. Kenwright, W. J. Feast, P. Adams, A. J. Monkman, and B. J. Say, Synth. Met., 55, 666 (1993). [71] Yamamoto and Takakazu, Chem. Lett., 7, 1211 (1993). [72] A. M. Kenwright, W. J. Feast, P. Adams, A. J. Milton, A. P. Monkman, and B. J. Say, Polymer, 33, 4292 (1992). [73] Shaoru Ni, Jinsong Tang, Fosong Wang, and Lianfang Shen, Polymer, 33, 3607 (1992). [74] Masaki Lida, Tetsuo Asaji, Daiyu Nakamura, Ryuichi Ikeda, Michiko B. Inoue, and Motomichi Inoue, Synth. Met., 47, 303, 1992. [75] S. Kaplan, E. M. Conwell, A. F. Richter, and A. G. MacDiarmid, Solid. State NMR Polym., Proc. Annu. Chem. Conf. North Am. Solid State NMR Polym., ed, L. Mathias (Plenum Press, New York, 1988). [76] Shaoru Ni, Jinsong Tang, Fosong Wang, Polym. Commun. (Berlin), 30,652 (1989). 216

[77] Shaoru Ni, Jinsong Tang, and Fosong Wang, Polym. Commun., 3 0 , 194 (1989). [78] A. F. Richter, A. Ray, K. V. Ramanathan, S. K. Manohar, G. T. Furst, S. J. Opella, A. G. MacDiarmid, and A. J. Epstein, Synth. Met., 2 9 , E243 (1989). [79] A. G. Marshall and F. R, Verdum, Fourier Transformations in NMR, Optical, and Mass Spectrometer: A user's Handbook (Elsevier, New York, 1990). [80] J. K. M. Sanders and B. K. Hunter, Modem NMR Spectroscopy: A Guide for Chemists, 2nd Ed. (Oxford University Press, Oxford, 1993).

[81] E. F. Mooney, An Introduction to 19F NMR Spectroscopy (Heyden & Sadtler, New York, 1970). [82] Compliments of Isotec, INC., USA, Residual H^O Peaks in Deterated Solvents by Proton NM R, data sheet (Tel: l-(800) 448-9760). [83] G. Boenhausen, H. Kogler, R. R. Ernst, J. Magn. Res., 58, 370 (1984).

[84] (a) W. P. Aue, E. Bartholdi, R. R. Erhst, J. Chem. Phys., 6 4 , 2229 (1976).

[85] S. H. Glarum and J. H. Marshall, J. Phys. Chem., 9 2 , 4210 (1988).

[86] W. W. Focke and G. E. Wnek, J. Electroanal. Chem., 2 5 6 , 343 (1988). [87] J. Tang, R. D. Allendoerfer, and R. A. Osteryoung, J. Phys. Chem., 96, 3531 (1992). [88] T. Ohsawa, T. Kabata, O. Kimura, M. Onoda, and Katsumi Yoshino, Jpn J. of Applied Physics, 28, 996 (1989).

[89] M. Lapkowski and E. M. Genies, J. Electroanal. Chem., 2 7 9 , 157 (1990).

[90] J. Kruszka, M. Nechtschein, and C. Santier, Rev. Sci. Instrum., 6 2 , 695 (1991). [91] J. F. Wolf, C. E. Forbes, S. Gould, and L. W. Shacklette, J. Electrochem. Soc. 1 3 6 , 2887 (1989). [92] J. E. Wertz and J. M. Bolton, Electron Spin Resonance: Elementary Theory and Practical Applications (McGraw Hill, New York, 1972). [93] J. M. Ko and I. J. Chung, Synth. Met., 68, 233 (1995).

[94] Y. Kang, M. H. Lee, and S. B. Rhee, Synth. Met., 5 2 , 319 (1992).

[95] M. A. Rodrigues and M. A. De Paoli, Synth. Met., 4 3 , 2957 (1991).

[96] H. Weiss, O. Pfefferkorn, G. Kotora, and D. Brian, J. Electrochem. Soc., 1 3 6 , 3711 (1989). [97] E. M. Genies and M. Lapkowski, J. Electroanal. Chem. Interfacial Electrochem., 236, 199 (1987). 217

[98] L. F. Malmonge and L. H. C. Mattoso, Polymer, 245 30 (1995). [99] C. DeArmitt, S. P. Armes, J. Winter, F. A. Uribe, S. Gottesfeld, and C. S. Mombourquette, Polymer, 34, 158 (1993). [100] W. E. Rudzinski, M. Walker, C. P. Horitz, and N. Y. Suhu, J. Electroanal. Chem., 335, 265 (1992). [101] G. Pistoia, G. Montesperelli, and P. Nunziante, J. Mol. Electron., 6, 889 (1990). [102] S. P. Armes, M. Aldissi, S. Agnew, and S. Gottesfeld, Mol. Cryst. Lig. Cryst., 190, 63 (1990). [103] Operator’s manual for Potentiostat/Galvanostat (HA-301), HOKTO Coorpo- ration, San Dimas, CA 91773. [104] A. J. Bard and L. R. Faulkner, Electrochemical Methods: Fundamentals and Applications (John Willey & Sons Press, 1980). [105] J. Yue and A. J. Epstein, J. Am. Chem. Soc., 112, 2800 (1990). [106] J. Yue, A. H. Wang, K. R. Cromack, A. J. Epstein and A. G. MacDiarmid, J. Am. Chem. Soc., 113, 2665 (1991). [107] J. Yue, G. Gordon and A. J. Epstein, Polymer 33, 4409 (1992). [108] (a) J. Yue, A. J. Epstein, Z. Zhong, P. K. Gallagher, A. G. MacDiamid, Synth. Met., 41, 765 (1991); (b) T.-C. Tsai, D. A. Tree, M. S. High, Ind. Eng. Chem. Res., 33, 2600 (1994). [109] (a) C. Barbero, M. C. Miras, B. Schnyder, O. Haas, R. Koetz, J. Matter. Chem., 4, 1775 (1994); (b) C. Barbero, M. C. Miras, R. Kotz, O. Haas, in Lithium Batteries (Proc. Electrochem. Soc.), 94-4, 281 (1994); (c) C. Barbero, M. C. Miras, R. Koetz, O. Haas, Synth. Met., 55, 1539 (1993). [110] G. Zhang, K. Gong, Yingyong Kexue Xuebao, 12, 92 (1994). [111] (a) M. Onoda, K. Yoshino, J. Appl. Phys., 78, 4456 (1995); (b) M. Onoda, K. Yoshino, Jpn. J. Appl. Phys., Part 2, 34 (2B), L260 (1995); (c) M. Ferreira, M. F. Rubner, Macromolecules, 28, 7107 (1995). [112] J. Yue and A, J. Epstein, J. Chem. Soc., Chem. Commun., 21, 1540 (1992) [113] T. Kawai, H. Mizobuchi, N. Yamasaki, H. Araki, K. Yoshino, Jpn. J. Appl. Phys., Part 2, 33, L357 (1994). [114] (a) T. Namiki, E. Yano, K. Watabe, Y. Igarashi, Y. Kuramitsu, T. Maruyama, K. Yano, T. Nakamura, S. Shimizu, T. Saito, Patent , Patent information: JP 07179754 A2 950718 Heisei; (b) K. Watable, Y. Yoneda, T. Maruyama, K. Yano, T. Nakamura, S. Shimizu, T. Saito, Patent, Patent information: JP 06003813 A2 940114 Heisei. 218

[115] S. Shimizu, T. Saitoh, T. Natamura, Patent , Patent information: DE 4244359 A1 930701.

[116] (a) A. G. MacDiarmid, A. J. Epstein, Synth. Met., 0 5 , 103 (1994); (b) A. G. MacDiarmid, A. J. Epstein, Trans., 2nd Congresso Brazileiro de Polimeros, Sao Paulo, Brazil, Oct. 5 - 8 , 544 (1993); (c) Y. Min, A. G. MacDiarmid, A. J. Epstein, Polymer Prepr., 231 (1993); (d) A. G. MacDiarmid, A. J. Epstein, Synth. Met., 69, 85 (1995).

[117] (a) Y. Cao, P. Smith, A. J. Heeger, Synth. Met., 4 8 , 91 (1992); (b) Y. Cao, A. J. Heeger, Synth. Met., 5 2 , 193 ([1993); (c) Y. Cao, G. M. Treacy, P. Smith, A. J. Heeger, Appl. Phys. Lett., 00, 2711 (1992); (d) Y. Cao, P. Smith, Polymer, 3 4 , 3139 (1993). [118] A. D. Patil, Y. Ikenoue, N. Basescu, N. Colaneri, J. Chen, F. Wudl, A. J. Heeger, Synth. Met., 20, 151 (1987). [119] A. G. MacDiarmid and A. J. Epstein, in Front. Polym. Adv. Mater. (Proc. Int. Conf.), 2, 251-61 (Plenum Press, New York, 1994). [120] (a) H. Tsutsumi, S. Fukuzawa, M. Ishikawa, Y. Morita, J. Electrochem. Soc., 1 4 2 , L168 (1995); (b) K. Hwang, J. S. Kim, M. J. Kong, Synth. Met., 7 1 , 2201 (1995); fcl N. Oyama, T. Tatsuma, T. Sato, T. Sotomura, Nature, 3 7 3 , 598 (1995); (d) M. Morita, S. Miyazaki, M. Ishikawa, Y. Matsuda, H. Tajima, K. Adachi, F. Anan, J. Electrochem. Soc., 1 4 2 , L3 (1995). [121] W. Huang, B. D. Humphrey and A. G. MacDiarmid, J. Chem. Soc., Faraday Trans., 82, 2385 (1986). [122] T. Ohsawa, O. Kimura, M. Onoda, and K. Yoshino, Synth. Met., 47, 151 (1992).

[123] X. Wei and A. J. Epstein, Synth. Met., 7 4 , 123 (1995). [124] A. G. MacDiarmid, A. G. Chiang, J.A. F. Richter, N. D. L. Somasiri and A. J. Epstein, in Conducting Polymers, ed. D. Riedel, p. 97 (L. Alcacer: Dordrecht, Holland, 1987). [125] F. Zuo, A. Angelopoulos, A. G. MacDiarmid and A. J. Epstein, Phys. Rev. B., 3 6 , 3475 (1987). [126] F. A. Carey and R. J. Sundberg, Adavanced Organic Chemistry, Part A: Struc­ ture and Mechanisms (Plenum Press, New York and London, 1990). [127] X. Wei and A. J. Epstein, to be submitted. [128] N. B. Colthup, L. H. Daly and S. E. Wiberley, Introduction to Infrared and Raman Spectroscopy (Academic Press, Harcourt Brace Jovanovich, 1990).

[129] J. M. Ginder, A. J. Epstein and A. G. MacDiarmid, Solid State Commun., 7 2 , 987 (1989). 219

130] J. M. Ginder and A. J. Epstein, Phys. Rev. B., 4 1 , 10674 (1990). 1311 A. J. Epstein, J. M. Ginder, F. Zuo, R. W. Bigelow, H.-S. Woo, D. B. Tanner, A. F. Richter, W. -S. Huang and A. G. MacDiarmid, Synth. Met., 18, 303 (1987). 1321 Z. H. Wang, E. M. Scherr, A. G. MacDiarmid and A. J. Epstein, Phys. Rev. B., 45, 4190 (1992). 1331 N. F. Mott and E. Davis, Electronic Processes in Non-Crystalline Materials (Clarendonm, Oxford, 1979). 1341 I. B. Goldberg, H. R Gowe, P. R. Newman, A. J Heeger and A. G. MacDiarmid, J. Chem. Phys., 70, 1132 (1979). 1351 J- C. Scott, P. Pflunger, M. T. Krounbi and G. B. Street, Phys. Rev. B., 28, 2140 (1983). 1361 H. H. S. Javadi, R. Laversanne, A. J. Epstein, R. K. Kohli, E. M. Scherr and A. G. MacDiarmid, Synth. Met., 29, E439 (1989). 1371 A. Carrington and A. D. McLachlan (ed.), Introduction to Magnetic Resonance (Chapman and Hall, London, 1967). 1381 J. E. Werts and J. R. Bolton, Electron Spin Resonance: Elememtal Theory and Practical Applications (McGraw-Hill Book Company Press, New York, 1972). 139] A. J. Bard and L. R. Faulkner, Electrochemical Methods: Fundamentals and Applications (John Willey fc Sons Press, 1980). 140] L. Libert and J. L. Bredas, to be published. 141] D. Rodrigue, P. Snauwaert, X. Demaret, J. Riga and J. J. Verbist, Synth. Met., 4 1 - 4 3 , 769 (1991).

142] P. McManus, R. J. Cushman and S. C. Yang, J. Phys. Chem., 9 1 , 744 (1987). 143] J. B. Goodenough, Magnetism and the Chemical Bond (Interscience, New York, 1963). 144] F. Genoud, M. Guglielmi, M. Nechtschein, E. Genies and M. Salmon, Phys. Rev. L ett., 55, 118 (1985). 145] F. Devreux, Europhys. Lett., 1, 233 (1986). 146] S. Stafstrom, J. L. Bredas, A.J. Epstein, H.S. Woo, D. B. Tanner, W. S. Huang and A.J. MacDiarmid, Phys. Rev. Lett., 59, 1464 (1987). 147] J. Libert, J. L. Bredas and A.J. Epstein, Phys. Rev. B., 51, 5711 (1995). 148] H. H. S. Javadi, S. P. Treat, J. M. Ginder, J. F. Wolf and A. J. Epstein, J. Phys. Chem. Solids, 51, 107 (1990). 220

149] F. Wudl, R. O. Jr. Angus, F. L. Lu, P. M. Allemand, D. J. Vachon, M. Novak, Z. X. Liu and A. J. Heeger, J. Am. Chem. Soc., 109, 3677 (1987). 150] J. M. Ginder, A. F. Richter, A. G. MacDiarmid and A. J. Epstein, Solid. St. Commun., 63, 97 (1987). 151] J.S. Miller and A.J. Epstein, Angew. Chem. Int. Ed. Engl., 33, 385 (1994). 152] X.-L. Wei, Ph. D. dissertation. 153] B. H. Bransden and C. J. Joachain, Introduction to Quantum Mechanics (John Wiley & Sons, Inc., New York, 1989). 154] S.-A. Chen and G.-W. Hwang, J. Am. Chem. Soc., 116, 7939 (1991). 155] (a) Y Sun, A.G. MacDiarmid and A. J. Epstein, J. Chem. Soc., Chem. Com­ mun., 7, 529 (1990); (b) A.G. MacDiarmid, S.K. Manohar, J.G. Masters, Y. Sun, H. Weiss and A. J. Epstein, Synth. Met., 41-43, 621 (1991). 156] The Organic Chemistry Division, the Chemistry Department, Nanjing Univer­ sity, Organic Chemistry (People’s Academic Press, Beijing, 1978). 157] Personal communications with R. Kohlman, Dept, of Physics, The Ohio State University, March 8, 1996. 158] X.-L. Wei, Y.-Z. Wang, S. M. Long, C. Bobeczco and A. J. Epstein, J. Am. Chem. Soc., 118, 2545 (1996). 159] Statistics data from STN computer service. 160] J. Cassidy and W. Green, Synth. Met., 41-43, 3059 (1991). 161] X.-L. Wei and A. J. Epstein, Presentation at Great Lake and Central Regional ACS Conference, 1994. Phys. Rev. B., 36, 3475 (1987). 162] A. J. Heeger, S. Kivelson, J. R. Schrieffer and W.-P. Su, Rev. Mod. Phys., 60, 781 (1988). 163] (a) Library: [VGS]xpsid.lib, VG data system (version: 1994), VG Scientific, Ltd, UK; (b) J. Chastain (ed.), Handbook of X-ray Photoelectron Spectroscopy, Perkin-Elmer, Eden Prairie, Minnesota, 1992; (c) D. Briggs and M. P. Seah (ed.), Practical Surface Analysis: Vol. 1- Auger and X-ray Photoelectron Spectroscopy, 2nd Ed. (Wiley, New York, 1990). 164] D.D.C. Bradley. Synth. Met., 54, 401 (1993). 165] A. Noshay and L. M. Robeson, J. Appl. Polym. Chem., 20, 1885 (1976). 166] J. Yue and A. J. Epstein, Macromolecules, 24, 4441 (1991). 167] N. F. Mott and E. Davis, Electronic Processes in Non-crystalline Materials (Clarendon Press, Oxford, 1979). 221

[168] Z. H. Wang, H. H. S. Javadi, A. Ray, A. G. MacDiarmid and A. J. Epstein, Phys. Rev. B, 42, 5411 (1990). [169] R. T. Morrison and R. N. Boyd, Organic Chemistry , 2nd Ed. (Ally & Bacon, Inc., Boston, 1966).