<<

UNDERSTANDING OIL RESISTANCE OF RUBBER:

CN GROUP INTERACTIONS AT INTERFACES

A Dissertation

Presented to

The Graduate Faculty of The University of Akron

In Partial Fulfillment

of the Requirements for the Degree

Doctor of Philosophy

Veronique Lachat

December, 2008

UNDERSTANDING OIL RESISTANCE OF :

CN GROUP INTERACTIONS AT INTERFACES

Veronique Lachat

Dissertation

Approved: Accepted:

______Advisor Department Chair Dr. Ali Dhinojwala Dr. Ali Dhinojwala

______Co-Advisor Dean of the College Dr. Mohsen S. Yeganeh Dr. Stephen Z. D. Cheng

______Committee Member Dean of the Graduate School Dr. Gary R. Hamed Dr. George R. Newkome

______Committee Member Date Dr. Gustavo A. Carri

______Committee Member Dr. Roderic P. Quirk

______Committee Member Dr. Rex D. Ramsier

ii

ABSTRACT

Nitrile rubber (NBR) is copolymer of and . It is resistant to swelling by oils. 1 Swelling and thermal degradation decrease as acrylonitrile content increases. Infrared-visible Sum Frequency Spectroscopy (SFS) is used in the present study to probe the molecular origin of oil resistance. SFS is a surface specific spectroscopic technique and was used to probe an NBR/liquid interface. Oil resistance of the NBR is reflected in changes in the SFS spectra of NBR at the interface. As reference materials, two additional polyacrylonitrile (PAN) and (PBD) were included and analyzed prior to analyzing the nitrile rubber.

SFS analysis of the film revealed a shift of the CN stretching mode at the sapphire/PAN interface compared to that of PAN bulk when the PAN film was annealed above its glass transition temperature. This demonstrates how environment affects nitrile dipole-dipole interaction. The influence of liquid environment on the PAN surface was directly assessed by comparing the SFS spectra of the PAN/air to that of PAN/heptane and

PAN/water interfaces. This showed a minor effect of the solvent on the nitrile CN stretching mode of PAN.

iii The second section was devoted to the analysis of PBD. A positive shift of the methylene stretching mode of PBD is detected in the SFS spectrum. This indicates that the polar sapphire surface influences the specific vibrational mode of PBD.

Finally, NBR rubbers with 40% or 20% acrylonitrile content (ACN) were characterized.

In the SFS spectra of NBR (40% and 20 %ACN)/air surfaces, a new vibrational band was observed at 2050 cm -1, in addition to the CN stretching mode of nitrile rubber at 2233 cm -

1. This band was not detected in bulk NBR (40% and 20% ACN). Additives in bulk NBR were determined using High Performance Liquid Chromatography (HPLC). Results suggested the presence of molecules. However, purified NBR also showed the

2050 cm -1 band. There are a limited number of assignments that can be present in this range. This includes nitrile groups interacting with salts. Probable assignments are proposed after considering the chemical compounds present in NBR .

The final part of this dissertation is concerned with the interaction between an NBR film and two solvents: heptane and toluene. No change at the sapphire/NBR interface upon heptane exposure indicates the stability of the NBR film in contact with heptane. This is in accord with the oil resistance of NBR to hydrocarbon solvents. Solubilization of the

NBR rubber thin film after toluene exposure is revealed by changes in the SFS spectrum at the sapphire/NBR interface. This shows that SFS is capable of accurately detecting molecular changes at /liquid interfaces.

iv SFS results are consistent with the resistance of NBR to aliphatic solvent and instablity in aromatic solvents. Quantitative interpretation of the oil resistance will require an assignment of the 2050 cm -1 absorption band.

v

ACKNOWLEDGMENTS

I would like to thank Dr. Dhinojwala for placing into my hands the most challenging opportunity I had the luck to live as a PhD student. Accomplishing the experimental part of this work at ExxonMobil Research and Engineering center represented for me an extremely valuable exposure to corporate research and environment. The key difference between industrial and academic research is the time constraint, that is why I would like to warmfully acknowledge Dr. Mohsen S. Yeganeh for managing the presence of a full- time student in his lab. His scientific input, professional experience and interest definitely guided my research. Working in the SFS lab was very enlightening from a technological point of view and I would like to express my gratitude to Shawn M. Dougal for his kindness and patience all along my stay. I would like also to thank Bernard Silbernagel and Paul Stevens without whom this collaboration between the University of Akron and

ExxonMobil would not have been possible.

I would like to thanks Robert Seiple and Critt Ohlemacher for their professionalism and help. Finally I would like thank my committee members, Dr. Gary R. Hamed, Dr.

Gustavo A. Carri, Dr. Roderic P. Quirk and Dr. Rex D. Ramsier.

vi I would like to specially thank Valerie Hill and Vicki England Patton for their endless resources and support. Both of them contributed in countless ways to the elaboration of this thesis. I also cannot forget Mayela Ramirez, Betul Buehler, Emilie Gautriaud, Gocke

Ugur, Vasav Sahni and Sunny Sethi for their help in editing this thesis. Finally, I would like to thank our group of students who made of the laboratory a very enjoyable research environment.

vii

TABLE OF CONTENTS

Page

LIST OF TABLES ...... xii

LIST OF FIGURES ...... xiv

CHAPTER

I. INTRODUCTION ...... 1

II. BACKGROUND...... 5

2.1 Nitrile Rubber ...... 5

2.1.1 Polyacrylonitrile...... 5

2.1.2 Polybutadiene...... 8

2.1.3 Nitrile Rubber ...... 9

2.2 Oil Resistance of Nitrile Rubber...... 11

2.2.1 Theoretical Approach...... 11

2.2.2 Key Experimental Parameter: Acrylonitrile Content...... 15

2.2.3 Industrial Approach ...... 16

2.3 Advances in Probing Polymer/Liquid Interactions...... 17

2.3.1 Experimental Parameter...... 18

2.3.2 Linear Partition Model...... 20

2.3.3 In-situ Swelling Measurements of Thin Polymer Films...... 21

2.3.4 Polymer Membrane Design ...... 24

viii 2.4 Summary...... 26

III. PROBING MOLECULAR INTERACTIONS USING SPECTROSCOPY ...... 28

3.1 Linear Spectroscopy...... 30

3.1.1 Infrared Spectroscopy...... 32

3.1.2 Raman Spectroscopy...... 34

3.1.3 Non-Linear Effects...... 39

3.1.3.1 Hot Band ...... 40

3.1.3.2 Fermi Resonance...... 42

3.2 Non-Linear Spectroscopy ...... 42

3.3 Spectroscopy for Probing Interactions...... 52

3.3.1 Model Compound ...... 52

3.3.2 Polyacrylonitrile Model Compound ...... 57

3.3.3 Frequency Shift at Surfaces ...... 61

3.4 Messages of This Dissertation ...... 62

IV. EXPERIMENTAL...... 64

4.1 Polymer Thin Film Preparation ...... 64

4.2 NBR Purification ...... 66

4.3 High Pressure Liquid Chromatography Characterization...... 66

4.4 IR-Visible Sum Frequency Generation Spectroscopy Measurements (SFS) ...67

4.4.1 ExxonMobil Narrow-line SFS Spectrometer...... 67

4.4.2 The University of Akron Broadline SFS Spectrometer ...... 70

ix 4.4.3 SFS in Total Internal Reflection (TIR) Geometry...... 72

4.5 Differential Scanning Calorimetry (DSC) Measurements...... 76

4.6 FTIR Measurements...... 76

4.7 Raman Spectroscopy Measurements ...... 76

4.8 Nuclear magnetic Resonance Measurements...... 77

4.9 Size Exclusion Chromatographic Measurements ...... 77

4.10 Thin Film Thickness Measurements...... 77

V. RESULTS AND DISCUSSION ...... 79

5.1 Spectroscopic Investigation of Nitrile Groups Interactions in PAN...... 80

5.1.1 SFS Spectra and Peaks Assignments ...... 80

5.1.2 PAN under Solvent Exposure ...... 90

5.1.3 Summary on PAN Investigation ...... 94

5.2 Analysis of PBD ...... 94

5.2.1 Bulk and SFS analysis of PBD ...... 94

5.2.2 Summary on PBD Investigation ...... 99

5.3 Analysis of NBR...... 99

5.3.1 Bulk Analysis of NBR ...... 99

5.3.2 SFS Investigation of NBR in the CN Region ...... 103

5.3.3 Summary of the Analysis of Dried NBR at Interface ...... 111

5.4 Resistance of NBR to Solvents...... 113

5.4.1 Resistance of NBR Surface to Heptane ...... 113

x 5.4.2 Resistance of NBR Surface to Toluene ...... 117

5.4.3 Resistance of NBR Surface to Water...... 121

5.4.4 Assigning the 2050 cm -1 Vibrational Band...... 123

5.4.5 Conclusions on Solvent Resistance of NBR...... 126

VI. SUMMARY AND CONCLUSIONS...... 128

REFERENCES ...... 132

APPENDICES ...... 138

APPENDIX A. PAN AND NBR THERMAL TRANSITIONS ...... 139

APPENDIX B.THIN FILMS ELLIPSOMETRY THICKNESS MEASUREMENTS...... 141

APPENDIX C. HPLC RUBBER ADDITIVES ANALYSIS...... 144

APPENDIX D. NMR SPECTRA OF NBR...... 147

APPENDIX E. SEC OF NBR...... 149

xi

LIST OF TABLES

Table Page

4.1 Solvents used during the HPLC separation ...... 67

4.2 Solvent profile used to characterize the methanol extract in Section 5.3.2 ...... 67

5.1 Positions of the CN stretching vibration for PAN from Figure 5.2 obtained by fitting using equation (3.22) ...... 84

5.2 Positions of the CN stretching vibration for PAN from Figure 5.3 obtained by fitting using equation (3.22)...... 87

5.3 PBD vibrational bands observed in Figure 5.10 compared to PBD vibrational bands observed by Binder...... 98

5.4 Observed frequencies (in cm -1) and assignments for Furukawa alternating copolymers and our NBR. (CN) and (BD) refer to the acrylonitrile and butadiene units respectively ...... 102

5.5 Positions of the CN stretching vibration for NBR (40% ACN) from Figure 5.13 obtained by fitting using Equation (3.22)...... 104

5.6 Positions of the CN stretching vibration for NBR (40% ACN) and purified NBR (40% ACN) at the NBR (40% ACN)/air interface from Figure 5.14 by fitting using Equation (3.22)...... 108

5.7 Positions of the CN stretching vibration for NBR (40% ACN) in the different regions probed obtained from Figures 5.13 and 5.15 by fitting using Equation (3.22)...... 111

5.8 Positions of the CN stretching vibration for NBR (40% ACN) in the different regions probed from Figures 5.16 and 5.17 obtained by fitting using Equation (3.22)...... 116

5.9 Positions of the CN stretching vibration for NBR (40% ACN) in the different regions probed from Figure 5.18 obtained by fitting using Equation (3.22)...... 120

xii 5.10 Vibrational modes absorbing in the (2000-2100 cm -1) region 64 ...... 124

5.11 Typical recipe for NBR emulsion polymerization 5 ...... 125

xiii

LIST OF FIGURES

Figure Page

2.1 PAN chemical structure ...... 6

2.2 PBD chemical structure ...... 8

2.3 NBR chemical structure...... 9

2.4 Schematic representation of the dilution process occurring during rubber welling...... 12

3.1 Ball and spring model for a diatomic molecule...... 30

3.2 Schematic representation of the location of the atoms in the nitrile and CO 2 moieties as the molecules vibrate with their corresponding dipole moment ...... 34

3.3 Correlation between the vibrational frequency shift and the change in magnitude of the force constant ...... 39

3.4 Energy scheme and corresponding vibrational spectrum for fundamental and hot transitions ...... 41

3.5 A sketch illustrating the production of SHG through crystalline quartz with vibrational excitation at 694 nm and the corresponding harmonic at 347 nm ...... 43

3.6 A sketch illustrating the SFG process through a non-linear material ...... 45

3.7 Definition of Euler angle relating the attached molecular axes frame (abc )to the laboratory axis system (xyz ) ...... 47

3.8 Schematic of an SFS experiment at interface in external reflection. The two polarization cases S and P have been illustrated for the incident infrared beam. The directions of the laboratory-fixed coordinate axes are indicated...... 49

xiv

3.9 Acetonitrile chemical structure ...... 52

3.10 PAN chemical structure ...... 57

3.11 Structure of singly-conjugated, pyrolyzed polyacrylonitrile ...... 59

3.12 Structure of conjugated, pyrolyzed polyacrylonitrile ...... 59

4.1 Schematic diagram of the ExxonMobil narrowline SFS spectrometer...... 69

4.2 Typical normalized SFS spectrum...... 70

4.3 Schematic diagram of the University of Akron broadline SFS spectrometer...... 71

4.4 Schematic diagram of the TIR geometry used in SFS experiments ...... 73

4.5 Angular dependence of SFS signal at the sapphire/NBR interface under heptane exposure in SSP polarization for a film of a thickness of 300 nm. The solid line corresponds to fit modeled by program written by Li 91 ...... 74

4.6 Angular dependence of SFS signal at the sapphire/NBR interface in SSP polymerization for a film of a thickness of 300 nm. Solid line corresponds to fit modeled by program written by Li 91 ...... 75

5.1 FTIR spectrum of bulk PAN in the CN stretching region ...... 81

5.2 (A) FTIR spectrum of PAN bulk, (B) SFS spectrum of PAN in PPP polarization at the sapphire/PAN interface after PAN annealing at 90 °C for 8 h followed by a second heating at 120 °C for 4 h and (C) SFS spectrum of PAN in PPP polarization of PAN at the sapphire/PAN interface after annealing at 90 °C for 8 h only, in the CN stretching region ...... 83

5.3 (A) FTIR spectrum of PAN bulk, (B) SFS spectrum of PAN in PPP polarization at the sapphire/PAN interface after PAN annealing at 90 °C for 8 h followed by a second heating at 120 °C for 4 h and (C) SFS spectrum of PAN in PPP polarization at the sapphire/PAN interface after PAN annealing at 90 °C for 8 h followed by an additional heating at 120 °C for 4 h on sapphire (temperature-treated at 750 °C for 36 h), in the CN stretching region ...... 86

5.4 (A) SFS spectrum of PAN in PPP polarization at the sapphire/PAN interface after PAN annealing at 90 °C for 8 h and 120 °C for 4 h with spin-coating on sapphire (temperature-treated at 750 °C for 36 h)

xv and (B) SFS spectra of PAN in PPP polarization after PAN annealing at 90 °C for 8 h and 120 °C for 4 h at sapphire/PAN in the CH stretching region...... 88

5.5 SFS spectrum of PAN in PPP polarization at the PAN/air interface, in the CN stretching region...... 90

5.6 SFS spectra of PAN in PPP polarization at (A) PAN/water interface sapphire in the CN stretching region (2000-2100 cm-1) and (B) PAN/water interface in the CN stretching region (2100-2300 cm -1). SFS spectra were obtained with the University of Akron broadline SFS spectrometer...... 91

5.7 SFS spectra of PAN in PPP polarization (A) at the PAN/air interface in the CN stretching region (2000-2100 cm -1) and (B) at the PAN/air interface in the CN stretching region (2100-2300 cm -1). SFS spectra were obtained with the University of Akron broadline SFS spectrometer...... 92

5.8 SFS spectra of PAN in PPP polarization (A) at the PAN/heptane interface and (B) PAN/water interface in the CN stretching region...... 93

5.9 Raman spectrum of PBD and corresponding fit ...... 96

5.10 (A) Raman spectrum of bulk PBD, (B) SFS spectrum of PBD in PPP polarization at the sapphire/PBD interface and (C) SFS spectrum of PBD in PPP polarization at the PBD/air interface in the CH stretching region. The data points number was reduced for clear labeling of the spectra...... 97

5.11 FTIR spectra of bulk (A) NBR (20% ACN), (B) NBR (30% ACN), and (C) NBR (40% ACN) in the CN region...... 100

5.12 FTIR spectra of bulk (A) NBR (20% ACN), (B) NBR (30% ACN), and (C) NBR (40% ACN) in the CH region...... 101

5.13 (A) FTIR spectrum of NBR (40% ACN), (B) SFS spectrum of NBR (40% ACN) in PPP polarization at the sapphire/NBR (40% ACN) interface and, (C) SFS spectrum in PPP polarization at the NBR (40% ACN)/air interface, in the CN stretching region. The data points number was reduced for clear labeling of the spectra ...... 104

5.14 SFS spectra of NBR (40% ACN) in PPP polarization (A) at the NBR (40% ACN)/air interface and (B) at the purified NBR (40% ACN)/Air, in the CN stretching region...... 108

5.15 (A) FTIR spectrum of NBR (20% ACN), (B) SFS spectrum of NBR (20% ACN) in PPP polarization at the sapphire/NBR (20% ACN)

xvi interface and (C) SFS spectrum of NBR (20% ACN) in PPP polarization at the NBR (20% ACN)/air interface in the CN stretching region. The data points number was reduced for clear labeling of the spectra...... 110

5.16 SFS spectra of NBR (40% ACN) in PPP polarization, (A) at the NBR (40% ACN)/air interface and (B) at the NBR (40% ACN)/heptane interface in the CN stretching region...... 113

5.17 SFS spectra of NBR (40% ACN) in PPP polarization (A) at the sapphire/NBR (40% ACN) interface before heptane exposure and (B) after 1 h of heptane exposure in the CN stretching region ...... 113

5.18 SFS spectra of NBR (40% ACN) in PPP polarization (A) at the NBR (40% ACN)/air interface, (B) at the sapphire/NBR (40% ACN) interface before toluene exposure, (C) at the sapphire/NBR (40% ACN) interface after 30 s of toluene exposure and N 2 drying, and (D) at the sapphire/NBR (40% ACN) after 4 additional min of toluene exposure and N 2 drying, in the CN stretching region...... 119

5.19 SFS spectra of NBR (40% ACN) in PPP polarization at the sapphire/NBR (40% ACN) interface (A) before water exposure and (B) after water exposure, in the CH stretching region...... 122

5.20 SFS spectra of NBR (40% ACN) in PPP polarization (A) at the NBR (40% ACN)/air interface and (B) at the NBR (40% ACN)/water Interface, in CH stretching region...... 122

xvii

CHAPTER I

INTRODUCTION

The first reference to nitrile rubber can be found in a French patent in 1931. Nitrile rubber (NBR) was commercialized four years later in Germany for its outstanding oil and fuel resistant properties. The NBR market just began growing in the United States when

WWII broke out in December 1941. Nitrile production became a full-fledged war effort, along with other synthetic , due to attacks by U-boats in the Atlantic stopping shipments of to the US. Many of the U.S. production plants that exist today were built because of the war effort 2. Sixty years later, NBR production in the U.S. alone represents 130 million pounds 3 annually and is often referred to as the workhorse of automobile rubber products. NBR is the oldest and most widely used O-ring material in fuel applications 4 and it is also used in grease sealings, creamery equipment 5, and textile applications.6 The reason why NBR is used in these applications is its resistance to swelling when in contact with these liquid environments. The polar CN group is the key to nitrile rubber’s ability to resist oil penetration.7

1 Early on, industry developed ASTM testing procedures to characterize the oil resistance of nitrile to “help” predict the lifetime of a rubber piece in a “standard” liquid environment. 2 However, these procedures up to the present day have not been able to predict the behavior of nitrile in “complex” or non-standard formulated oil. As the number of chemical compounds increases in the oil, the tests imagined by rubber designers fail to answer the important question: how will the rubber interact with the oil?

Despite the recent advances in modeling associative solvents 8, there still no microscopic picture of the rubber/oil interactions.

The approaches developed up to today have focused on indirect information like the effect of the oil on the rubber’s physical and mechanical properties. Furthermore, understanding the nature of these interactions between the rubber and the oil must involve direct information regarding the impact of the oil on the rubber molecules properties. For example, observing the first spectroscopic changes of rubber molecules at the rubber/oil interface could significantly help in shaping a molecular mechanism of the oil resistance phenomenon. Such an approach requires a tool capable of probing the rubber/oil interface and represents a technical challenge as the presence of liquid prohibits the use of vacuum techniques.

The technique of Sum Frequency Spectroscopy (SFS) is able to probe the polymer/liquid interface and retrieve molecular information about the microscopic environment at the interface. Such an interfacial description represents an important step in understanding

2 and controlling phenomena starting at the interface. This technique was used for investigations in fields including sensor designing, food processing, medical industries 9-

11 and in biotechnology, more precisely with , and biosensor disease diagnosis.12-

14

Insight on polymer/liquid interactions would not only shred light on a problem related to the rubber industry problem but would also contribute to other sciences that pursue similar goals such as membranes science,15 coatings fabrication,16 and polymer implant design.17

The approach proposed within the framework of this dissertation is to probe the surface of nitrile rubber using SFS and collect molecular information, such as solvent-induced shift on the NBR absorption bands to understand the impact of the liquid environment on the molecular properties of rubber. Therefore, the old industrial phenomenon of NBR’s oil resistance will be reinvestigated.

This dissertation is articulated in 6 chapters. The second chapter describes the experimental tools available to characterize the oil resistant properties of the focusing on two main areas. The first area is the industrial approach, which essentially focuses on the mechanical properties of the swollen rubber. The second area describes the advances in fields where the polymer/liquid interaction characterization is the prime concern. Chapter 3 sets the foundation of non-linear spectroscopy methods and illustrates

3 the spectroscopic manifestations of the interaction between the nitrile groups and the environment. Chapter 4 describes the details of the Sum Frequency Spectroscopy technique and how it will be used in this research. Chapter 5 presents the experimental results and Chapter 6 emphasizes the main conclusions of this investigation.

4

CHAPTER II

BACKGROUND

2.1 Nitrile Rubber

Nitrile rubber (NBR), a copolymer of acrylonitrile and butadiene monomer, will have properties intermediate between those of PAN and PBD. This is why a brief background about these two homopolymers that make up NBR is an appropriate starting point for this section.

2.1.1 Polyacrylonitrile.

Polyacrylonitrile (PAN) is a commercially important polymer 18 and widely used as a fiber. PAN fibers are used in protective clothing,19 biomedical materials 20 and nanosensors.21 PAN is produced by a reaction of acrylonitrile monomer with a free radical or an anionic initiator. The polymerization can be carried out in bulk, emulsion, suspension, slurry or solution.21 The growing acrylonitrile polymer chain occurs in a head-to-tail fashion, attaching to every alternate atom a compact and highly polar nitrile unit.

5

Figure 2.1. PAN Chemical structure.

This polar nitrile side group is a very small moiety and yet exerts a major influence on the electrical, mechanical, thermal and barrier properties of PAN. Calculation of the strength of the nitrile-nitrile dipole interaction in low molecular weight compound demonstrated that electrostatic forces are large and control molecular arrangements.22 If parallel intramolecular nitrile groups induce repulsion, anti-parallel intermolecular dipole moment causes attraction between polymer chains. Also, the intramolecular dipole interactions are responsible for both stiffness and high extension properties of PAN molecules. The backbone of PAN molecule is stiff and rotation around successive bonds is possible only in a cooperative manner. Pitsyn and Sharonov 23 elaborated a model where the rotation of every other bond correlated with its predecessor is required before any concerted motion of nitrile side groups could occur. PAN morphology and crystallinity does not resemble that of any other polymer.21 PAN has a low degree of crystallinity (paracrystallinity) and a high melting point.24 Analytical techniques such as

X-ray diffraction fail to provide information about PAN crystallinity. However, electron diffraction, showed two-dimensional order.25 The scientific community tends to agree on a heterogeneous structure for PAN, constituted of three phases: amorphous, quasi- crystalline and crystalline phases.21 Furthermore, the existence of two thermal transitions

6 around 100 °C and 150 °C is also recognized.18, 21 the first transition at 80-100 °C is attributed to the glass transition and the second transition around 140 °C is attributed to the loosening of dipole-dipole interactions.18, 26, 27 Okajima et al.28 attributed both transitions to two amorphous phases having a different degree of order; the highest order phase contributing to the highest transition relaxation. However, other authors such as

Minami 29 envisioned the lowest temperature transition originating from molecular motion in the paracrystalline phase. Some authors reported three thermal transitions. 30

The nitrile-nitrile interactions in PAN are strong to the extent that only a limited number of solvents can dissolve the polymer chains. At a molecular level, the solvent has to form a hydrogen bond with a strength higher or at least comparable to those forces holding polyacrylonitrile chains together.

Furthermore, PAN has the capability of interacting with specific surrounding molecules via two sites. The lone pair orbital of the atom is readily inclined to hydrogen bonding with molecules such as water, and Bronsted and Lewis acids.31 Also, the π - orbital electrons of the triple bond can interact with transition metal ions. These two interactions give two distinct infrared spectroscopic manifestations, which will be discussed later in Chapter 3.

7 2.1.2 Polybutadiene

Polybutadiene (PBD) is made by the polymerization of butadiene monomer. The reactivity of butadiene monomer is characteristic of the chemistry of conjuguated unsaturated double bonds. The polymer microstructure is characterized by the disposition of the double bonds present in the polymer chains. Butadiene polymerizes by addition of a monomer butadiene unit on an allyl radical resonance structure. The three possible structures are vinyl-1,2, cis -1,4 and trans -1,4

Figure 2.2. PBD chemical structure.

The 1,4 addition process leads to the formation of either cis-1,4 (Z) or trans-1,4 (E) groups. In its final form, the polymer consists of one of these forms or is a mixture of all of them depending on the polymer recipe and polymerization conditions.32

PBD is a non-polar rubber whose chains are attracted to one another via dispersion forces. It is essentially a rubbery material whose susceptibility to oxidation is high because of the presence of double bonds.

8 2.1.3 Nitrile Rubber

When butadiene is copolymerized with acrylonitrile, a completely new set of macroscopic properties appears for the poly(acrylonitrile-co -butadiene) or nitrile rubber

(NBR) as it is commonly referred to in the industry. From a synthesis point of view, acrylonitrile and butadiene are polymerized using standard free radical emulsion polymerization techniques. A typical recipe for NBR includes the monomers of acrylonitrile and butadiene, the emulsifier or soap, stearic acid, t-dodecylmercaptan and chloride, sodium pyrophosphate, ferrous peroxide and of course, water. 1 The

5 copolymerization reactivity ratios are rB= 0.4, rAN =0.04 (at 50 °C) . The product of rB.r AN is close to 0 and this indicates an alternate copolymerization (a-b-a-b-a-b).

-a-

-b-

Figure 2.3. NBR chemical structure.

9 Emulsion polymerization is done using a batch process where 40% acrylonitrile in the original reaction charge corresponds to the 40% ACN in the final polymer composition.

This allows formation of an azeotrope copolymer. The major consequence is that the copolymer has a constant composition over the entire polymerization process. If the composition of monomer is either below or above 35-40% ACN, the acrylonitrile copolymer composition will, respectively, decrease or increase with increasing conversion. Composition drift could be responsible for butadiene-rich chains which could turn into a specific point of entry for the oil.

Many macroscopic properties of NBR directly depend on this acrylonitrile content. For example, the tensile strength, abrasion resistance, hardness, and heat resistance increases as the acrylonitrile content (ACN) increases.2 Certainly, the most famous feature of nitrile rubber that increases with the ACN content is its remarkable resistance to oil.

Many polar rubbers such as polysulfide, the first solvent resistant polymer ever synthesized, also exhibit oil resistance.33 The superior solvent resistance of polysulfide rubber is explained by the presence of sulfur (highly polar atoms) in the polymer backbone. The role and importance of the polarity of sulfur is demonstrated by the fact that the solvent resistance increases with the content of sulfur atoms. Other important polar rubbers are the acrylic elastomers (ASTM designation ACM) and epichorohydrine

(EPI), a polar polymer containing a chorine atom on every other .

10 2.2 Oil Resistance of Nitrile Rubber.

The following sections describe the various approaches attempted to characterize polymer-solvent interactions and model elastomer swells in liquid environments.

2.2.1 Theoretical Approach

From a thermodynamic point of view, the Gibbs free energy balance in a swollen elastomer is controlled by enthalpy, entropy, and elastic retraction energy. Gee and

Treloar 34-42 thoroughly investigated the interaction between rubber and liquid. Later on,

Treloar43 used this information to shape an entire chapter in his book dedicated to the swelling phenomenon of rubber. It is a well known fact that aromatic oils cause rubber swelling. This is why the rubber/ system was chosen to represent the main issues related to the swelling of rubber in a solvent.

Rubber swelling is entropically driven and is not a consequence of an attractive force

∆ > between the rubber molecule and the solvent, sinceH m 0 . This behavior differs from

∆ < that of water-swelling of a substance that typically showsH m 0 , characteristic of an enthalpic-driven swelling. The swelling of the rubber molecule is therefore a reflection of the tendency of the rubber molecule to mix with oil by thermal motion in the same exact manner as a gas or common liquids. Interestingly though, the swelling of rubber by chloroform is enthalpically driven.

11 Rubber swelling corresponds to a equilibrium between a pure phase (liquid) and a mixed phase (swollen rubber). A dilution of the pure liquid into the rubber and a stretching of the chains of the rubber network must be accounted for to understand the thermodynamic equilibrium of rubber in a swelling agent.

Two phases: Solvent in Swollen Rubber

One phase: Pure Solvent (Benzene) ~ ~ ~ ~ ~ ~~ ~ ~

DILUTION PHENOMENON

Figure 2.4.Schematic representation of the dilution process occurring during rubber swelling.

Mathematically this means:

∆ = ∆ + ∆ G G1 Gel (2.1)

∆ G1 stands for the change in Gibbs free energy of the system due to the transfer of 1 mol of liquid from the liquid phase to the mixed phase, also called the Gibbs free energy of

∆ dilution. Gel corresponds to the change in Free Enthalpy (per mole of liquid absorbed) resulting from the elastic expansion of the network.

12 The statistical treatment of swelling determines ∆G and ∆S by counting the number of the configurations available versus the composition of the system and capture the behavior of the mixture during the dilution process. Many textbooks explain the calculation, the final result of which is expressed as:

∆ = [( − )+ + χ 2 ] G1 RT ln 1 v2 v2 v2 (2.1)

where v2 is the volume fraction of polymer in the mixture, and χ is the Flory-Huggins energy parameter.44 Equation (2.1) is known as the Flory-Huggins equation. Equation

(2.1) only describes the dilution of the liquid into the rubber.

The stretching affects the Free Enthalpy variation of the system via the dependence of the

work,W , on the number of liquid molecules, n1 , swelling the rubber network. Since

n1 influences the stretching ratio corresponding to the rubber isotropic expansion such as

1 1 3ρRT  − 2  v = and that = 1+ nV , it is easy to show that if W = λ 3 −1 and 2 λ3 1 1   v2 2M c  

1 ∆S = − (l 2 + l 2 + l 2 − 3) (2.2) 2Nk 1 2 3

∂W ρV 1 then, ∆G = = 1 v 3 (2.3) el ∂ 2 n1 M c which gives:

 ρV 1  ∆ = ()− + + χ 2 + 1 3 G RT ln 1 v2 v2 v2 v2  (2.4)  M c  with the corresponding swelling equilibrium:

13 ρV 1 ()− + + χ 2 + 1 3 = ln 1 v2 v2 v2 v2 0 (2.5) M c

If instead of Equation (2.2) the following expression is used:

∆S = − 1 Nk {l 2 + l 2 + l 2 − 3 − ln ( 2 2lll 2 )} (2.6) el 2 1 2 3 1 2 3 then Equation (2.4) becomes:

ρV  1 v  ∆ = ()− + + χ 2 + 1 3 − 2 G ln 1 v2 v2 v2 v2  (2.7) M c  2  which is also known as Flory-Rehner equation.

Predictive models for the rubber swelling emerged with the concept of the solubility parameter (SP). The SP model is based on several important assumptions. First, moderate swelling-power liquids must be used so that ∆S is not modified by the crosslinking density. This was experimentally established as similar ∆S were obtained for solvents of low swelling power. Furthermore, ∆S is the same in all solvents, which makes the interaction between rubber and liquid entirely governed by ∆H . Essentially, then, the model predicts the solubility of an elastomer in liquid environment via the determination of ∆H and its connection to the physical properties of the rubber and the liquid.

Hildebrand and Scott defined the heat of vaporation as the cohesive energy density (c.e.d) and it’s square root as the SP,

1  ∆H − RT  2 δ =   (2.8)  V 

14 Equation (2.8) stipulates that the heat of vaporization, ∆H − RT , is the energy needed to maintain the liquid state. This equation is valid only for vapor following the ideal gas law. For non-polar materials then, the maximum swelling is predicted when the rubber and liquid have a solubility parameter close in magnitude which defines the “force- matching” concept underlying SP ideas.

In order to predict the swelling of polar liquids, Hansen 45 modified Hildebrand’s theory by considering the driving forces controlling interactions namely, dispersion forces, hydrogen bonds, and dipole moment interactions and further decomposed the SP into three contributions:

δ 2 = δ 2 + δ 2 + δ 2 D P H (2.8)

2.2.2 Key Experimental Parameter: the Acrylonitrile Content

Salomon 7 initiated the characterization of the solvent resistance of polar rubbers such as

NBR by immersing elastomeric pieces in various solvents until the swelling equilibrium was reached. The swelling equilibria of a series of copolymers from acrylonitrile with three different comonomers (butadiene, isoprene, and dimethylbutadiene) was investigated. Historically, the extent of the swelling was expressed using “weight fraction” which represents the increase in weight after rubber plates were immersed in various liquids at 25 °C for a specific amount of time (see Figure 1 of reference 7). The three copolymers were immersed separately in pentane, hexane, heptane, cyclohexane and benzene. The major conclusion from his work was that for the three copolymers the

15 extent of the swelling decreased with the amount of acrylonitrile; the least swelling corresponded to copolymers containing 30 to 40 % ACN. This study pointed to the polar unit of NBR as a major parameter governing the oil resistance of NBR.

2.2.3 Industrial Approach

The definition the oil resistance of a rubber is, as a matter of fact, related to the “ability of the vulcanized product to retain its original physical properties, such as modulus, tensile

strength, abrasion resistance, dimensions”.32 A 20% change in volume due to oil absorption can cause a reduction up to 60% in properties such as hardness, stiffness and strength.46

The industrial approach uses mainly the extent of swelling at thermodynamic equilibrium and the effects of swelling on stress-strain isotherms to predict the oil resistance performance of rubber pieces. The swelling measurements were seen as accurate to predict the serviceability of commercial rubbers. Rubber designers evaluated then the effect of liquids on vulcanizates by studying at the changes in thickness as explained in

ASTM D471.

Tests for tensile strength, percent elongation at break, hardness, and percent volume swelling are performed after a specified immersion time period at a specified temperature in the characterization of oil-resistant rubbers as described in ASTM D5964. In the case

16 of engine oil formulations, the lack of compatibility with certain elastomers used for seals in automotive engines is tested using ASTM D7216.

Only a few academic studies have devoted attention to the mechanical properties of swelled rubber. Magryta et al.47 et al. showed that there are observable changes in the loss angle values of the swelled samples compared with those of non-swelled, making mechanical characterization of swollen rubber a sensitive characterization tool.

Furthermore, the mechanical properties such as tensile strength, elongation at break, moduli, and Shore A hardness of swollen CR and NBR vulcanizates are lower than the non-swollen vulcanizates.

The effect of liquids on the dynamic properties of carbon black filled natural rubber has been investigated by Busfield at al.48 They used a free oscillating technique to assess the dynamic storage and loss moduli of carbon black filled natural rubber. The researchers actually probed how the fillers resist liquid exposure. They evidenced a significant decrease in the storage and loss moduli of swollen filled rubber compared to the one of non-swollen filled rubber. They attributed this to a combined effect of swelling of the rubber matrix and a reduction of the volume fraction of the fillers.

2.3 Advances in Probing Polymer/Liquid Interactions

Alternative methods to predict the swelling behavior of elastomers in solvent involves activity coefficients and polarity indices and are developed in the next section.

17 2.3.1 Experimental Parameters

Predicting the oil resistance phenomenon comes down to predict the swelling power of a liquid. The solubility parameter model, presented earlier, has important limitations when the study of a polar liquid is concerned. In some case, experimental anomalies were observed for polymers such as nitrile rubber that appeared to have two solubility parameters. 8 Furthermore, the SP concept is unable to correlate accurately the interaction between two liquids characterized by a negative excess heat of mixing like in the case of gasoline- mixture. Also, the SP model relies on the hypothesis that ∆S does not depend on the swelling and that is not always true. Furthermore, it is reasonable to expect that the energy needed to vaporize a liquid to be much larger than that needed to mix a liquid.

Finally, even the modified Hansen SP, when valid, requires treatment of a large number of data in the case of a multi-component solution. The amount of data needed to define the system comprises 3 constants for the rubber and 3 additional constants for each of the solvent parts of the mixture.

Pre-determining the rubber swelling behavior in polar and non-polar solvent requires correlating swelling measurement data to rubber and solvent parameters or to an interaction parameter between rubber and solvent. Starmer 49, 50 reduced the information contained in a swelling measurement of a rubber/polymer pair to one single value A. The swelling behavior is determined by plotting the swelling extent (in % volumic ratio)

18 versus the %ACN content in the copolymer, like Salomon 7 did. The curve was fitted to a

Gaussian function that passes through a maximum noted A, A being the ACN content at the maximum swelling. For example, NBR/isooctane, the swelling curve has maximum swelling with a fictitious copolymer having a negative ACN. In the case of

NBR/aromatics, A 20-25 %ACN and finally in the case of NBR/methanol, A 40% ACN.

Starmer investigated the correlation between A and parameters describing the properties of the rubber and the solvent. There are five potential parameters available to describe the polarity of a liquid, the dielectric constant ε , the miscibility parameter M, the molar

49-52 transition energy E T(30), and the polarity index. The dielectric constant is, like the solubility parameter, entirely related to the properties of the material. The miscibility parameter (M) separates the solvents into 31 classes and predicts the solubility of pairs according to the miscibility difference number. The molar transition energy, E T(30), is an empirical parameter measuring the polarity of a solvent based on a solvent-dependent process called solvatochromism. The effect of the solvent on the position and absorption intensity of UV-Vis absorption band of a dye dispersed in the medium under analysis is monitored. ET(30) is defined as the molar transition energy for the solvatochromic band of the dye in the solvent. The last parameter is the polarity parameter measured from liquid chromatography. These three last parameters are based on the interaction between the liquid and one or other materials. Starmer examined the correlation between, respectively, the dielectric constant, the miscibility parameter, E T (30) and the polarity indices with the A value. The correlation between the dielectric constant and A, if it exists, is weak. Acceptable correlation was observed for the miscibility parameter and

ET(30) if the case of was left out. The highest correlation was observed for the

19 polarity indices. Finally Starmer, like Gee 37 did using the solubility parameter, offers an experimental mean of deducing the polarity indices of a polymer from that of the liquid which swells that polymer more than any other. Using such a parameter to predict the oil resistance of nitrile rubber is not successful yet as the reverse approach is used in the sense that it is the swelling of the rubber that determines the polarity index of the polymer. Note that, according to Starmer approach, the location of the maximum of the swelling curve, only characterizes the swelling, which leaves out another parameter, the extent of the swelling.

2.3.2 Linear Partition Model

The linear partition model was developed to predict the swelling behavior in a mixed associative solvent.8 In the case of an elastomer swelling in an ideal mixture, the observed swelling is directly determined by the solvent composition and the swelling behavior of the rubber in the pure solvent part of the mixture. An example of a nearly ideal solution is a gasoline mixture formed by oil and methyl tertiary -butyl

(MTBE).8 MTBE is a Lewis base that will interact with acidic protons. The other constituent of this specific gasoline mixture does not contain any acidic hydrogen making this oil a “non-associative” oil. If MTBE does not interact with any other chemical in the solution, it does present affinity toward the polar moiety of poly(vinylidene -co - hexafluoropropylene) (FKM-66). FKM-66, specifically, contains atoms and, consequently, acidic hydrogens. The component that will essentially be absorbed by the polymer is MTBE rather than the rest of the oil.

20

Now, the swelling behavior of an elastomer by a non-ideal solution is accounted for by multiplying the partition coefficient by the volume-fraction based activities for each solvent. Methanol is an example of a non-ideal solution as the molecule forms tetramers at room temperature, thought to be more aggressive toward FKM-66 than its non- associated form. The partition coefficient model assumes that the tendency of a chemical to swell a rubber is characterized by its activity in solution such that the partition coefficient must be multiplied by the activity of solvent in solution.

= 0 vm,e ∑ai,svi,s (2.9) i

ν 0 with, i,e the partition coefficient and ai,s the activity of the solvent. Despite the fact that prior knowledge of the chemical composition of the elastomer is not needed, the model requires thermodynamic data for each component.

2.3.3 In-situ Swelling Measurements of Thin Polymer Films

The thin film approach involves thickness uptake measurements of polymeric thin film using, for example, X-ray reflectivity. Singh and Mukherjee 53 looked at the dynamic behavior of a spin-coated polyacrylamide thin film on a silicon substrate. The film was exposed to water vapor at a saturated vapor pressure. The mass uptake was measured by gravimetry (microbalance) measurement and the thickness uptake by X-ray reflectivity.

Working with a thin film supported onto a substrate is delicate due to a possible interaction between the substrate and the polymer. The outcome of such a system is

21 unpredictable. In the work by Singh, it was proven that the diffusion coefficient of water is not dependent on the thickness of the film which means that the substrate does not modify the diffusion process. Also, their data showed that there was a time window during which the entire film of polyacrylamide was exposed to water before the film begins to swell. According to the present data, the time window was roughly on the order of 20 minutes. That time window is a useful time frame that can be used to probe the interactions between polymer and solvent before the macroscopic changes, such as swelling, occur and complicate the system.

The second example concerns the in-situ measurement of thickness uptake of polymeric thin film using ellipsometry. Overall, the investigators were interested in determining the role of the interface in the anomalous swelling of various polymers in supercritical CO 2

(Sc-CO 2). The conclusion on the influence of the interface in the swelling behavior of these polymers was based on the direct comparison between the known swelling behavior of the bulk polymer to that of the thin film swelling behavior. Li et al. 54 investigated the swelling of poly( ) (PEO) and polystyrene (PS). Because of a low cohesive energy density, CO 2 was thought to be in excess at the polymer/CO 2 interface causing an anomalous swelling of the polymer . Li and coworkers carried out in-situ swelling measurements of a polymer using an ellipsometer apparatus. Each pressure point of CO 2 was correlated to a measurement of thickness. By plotting the % of swelling versus the pressure of CO 2 or what they called swelling isotherm for different polymers, they finally showed that the interface alone cannot account for the excess of CO 2 adsorption at the polymer interface. The investigation of the role of the interface in the polymer/solvent 22 interaction is therefore possible when in-situ swelling measurement of thin polymeric films is undertaken.

The last example is related to the use of a Quartz micro-balance coupled with a heat sensor to in-situ record the mass uptake of a thin polyether urethane elastomer film

(roughly 1 m thick) exposed to ethanol vapor with heat changes. Smith and Shirazi 55 et al. simultaneously monitored the heat flow and mass/unit area change versus the composition of the gas and thus were able to record a direct measurement of the thermodynamic quantities that characterize the interaction between polymer and solvent vapor. The reliability of these measurements over those obtained from inverse gas chromatography resides in knowing when the equilibration between organic gas and the polymer occurs so that reliable enthalpic changes for gas-surface interactions are determined. Furthermore, the device design would be especially appropriate for analyzing surface specific phenomenon such as catalyst poisoning.

In-situ swelling measurements of thin polymeric films start with a time window during which the system under analysis evolves toward its equilibrium. This time frame is experimentally observable for the first time is a unique opportunity to probe the influence of the interface in polymer/liquid interactions.

23 2.3.4 Polymer Membrane Design

Membrane science is interested in developing methods to control and quantify the flux of specific molecules through polymer matrices. The interaction between polymer and solvent is the basis of membrane science where the appropriate chemical function of a polymer (forming the membrane) is able to selectively separate out one component in a mixture containing many. Such a phenomenon is called selection by a permselective membrane. Permselective membrane separation is an attractive alternative to distillation when the temperature-based difference between the components to separate is small. For example, the distillation of benzene/hexane is difficult because their boiling points are only separated by 0.6 °C. So instead of distilling the benzene/hexane mixture, it is preferable to force the mixture through a polymeric membrane and have the polymer act as a third phase to separate the two constituents. The goal will be to find a polymer with a great affinity toward one component so that it has a high selectivity. For instance, the removal of an organic solvent from aqueous solution will require an oranophilic membrane while a hydrophilic membrane will be selected for a dehydration of an organic solvent.

For example, Yamaguchi et al. 56 filled the pores of a HDPE film with poly(methyl methacrylate) and due to the presence of this polymer in the membrane, benzene over cyclohexane was able to permeate through it. Koops et al.57 studied permeation of solute in ethanol and n-hexane through cellulose acetate and concluded that solute-membrane- solvent interactions are important for mass transport.

24 Membrane/solvent interactions were revealed by directly comparing the performance of two membranes: poly(ethylene oxide)-b-poly(dimethyl siloxane)-b-poly(ethylene oxide)

(PEO-PDMS-PEO) membrane and hydrophobic poly(dimethyl siloxane) (PDMS) membrane in contact with a common solvent such as ethanol or isopropanol.58 A ternary interaction was determined by keeping two parameters constant and changing the third one. So to analyze polymer-solvent-solute interactions, they needed three sets of data to permute the variable. First, they kept the solute and membrane as a constant and changed the solvent. Second, they kept the membrane and the solvent as a constant and changed the solute and so on. By comparing the performance of not one, but two membranes with the same solvent, the conclusion was that membrane-solute interactions are possible. For example, they compared the flux of ethanol, isopropanol, methyl ethyl , toluene and hexane on first PEO-PDMS-PEO/PAN and second PAN/PDMS. PAN was used as a support. The authors intended to compare the properties of hydrophobic PDMS and hydrophilic PEO-PDMS-PEO. They kept the membrane and the solute as a constant and changed the solvent as they analyzed the PAN/PDMS membrane and compared the transport of oil/hexane to that of oil/toluene. Then, they correlated the variables together.

They connected the permeability to the nature of the membrane. They effectively noticed that for the hydrophilic membrane, PEO-PDMS-PEO, the flux of the polar solvents

(ethanol, MEK) was higher while in the case of the hydrophobic PDMS membrane the flux of non-polar solvents (hexane, toluene) was higher. Stamatialis et al.58 revealed that the flux of oil/toluene mixtures depends on the concentration of toluene in the oil

(triglyceride or ). Oil (triglycerides or ester) slows the diffusion of the mixture. The higher coupling of oil/toluene than that of the oil/hexane may be due to the higher

25 viscosity of oil/toluene. Higher coupling implies higher solvent dragging. Another example is the ion pairing of tetraoctylammonium bromide, TOABr, in toluene solution.58 The nature of the composition is pivotal for the transport of (TOABr)/toluene through polyimide. It means that ion pairing of TOABr in solvent (toluene) leads to the observation of an anomalous behavior.

A recent study 59 of polyelectrolyte ultra filtration membrane based on poly(acrylonitrile- b-2-dimethylamino ethyl methacrylate) (PAN-DMAEMA) copolymer revealed an enriched surface of PDMAEMA. The spontaneous migration of membrane casting solution bulk to membrane surface is due to the incompatibility between hydrophobic

PAN and hydrophilic PDMAEMA. Furthermore, immersion of the membrane in basic

(pH = 8) solution alters the surface composition.

Garcia et al. 60 observed that when polysulfone membranes were directly exposed to heptane after water cleaning, hexane flux is blocked. But if the membranes were pretreated by earlier exposures to solvent of decreasing polarity, the flux was recovered.

Surface compositional segregations and surface pretreatment can have a decisive influence on the behavior of a piece of polymer in contact with liquids.

2.4 Summary

The chief parameter, pointed out in the early 50s, makes the nitrile content, ACN, the prime parameter controlling the oil resistance of NBR. The ACN leading to the maximum

26 swelling depends on the chemical nature of the oil or fuel with which the rubber is in contact. Characterization of the mechanical properties of swollen rubbers remains the first means of assessing the NBR oil resistance performance by industrial rubber engineers despite inaccurate predictions in the case of complex oil formulations.

Recently, the linear partition model was able to predict the swelling behavior of mixed associative solvents by applying the concept of solvent activity to the partition coefficient of an elastomer-solvent pair. Such model concentrates on the solvent composition and the nature of the interactions in the solvent. Certain fields of research motivated by the common interest of understanding polymer/liquid interaction developed in-situ characterization techniques of polymeric thin films while others pointed out the importance of the interface in controlling the flux of specific molecules through polymer matrices. However, molecular information is still missing.

A new approach is needed to understand the mechanism of oil resistance. Along with an industrial need, the oil resistance mechanism is a very fundamental problem with many questions remaining unanswered.

27

CHAPTER III

PROBING MOLECULAR INTERACTIONS USING SPECTROSCOPY

In Chapter 2, the limitation of the thermodynamic approach to predict the rubber/liquid interactions was reviewed. It was also shown that the oil resistance of nitrile rubber depends on the amounts of nitrile groups present in NBR, pointing out the nitrile dipole interactions as the most probable cause of the oil resistance phenomenon. The prevailing polar nature of NBR directly correlates with the nitrile-nitrile dipole interactions in NBR bulk. The purpose of this chapter is to demonstrate that these nitrile-nitrile interactions give rise to spectroscopic manifestations that could be used to investigate the origin of the oil resistance phenomenon by connecting specific vibrational manifestations to the amount of nitrile in the rubber.

In addition, according to the framework of this project, the study of the environmental perturbation on these specific vibrations is also meaningful. This perturbation is related to the liquid environment to which the rubber is exposed. The C ≡N bond possesses 21% ionic character 61, 62 due to its large dipole moment (3.5 D).63 Thus, the dielectric constant of the surrounding medium plays an important role in the prediction of the strength of the intermolecular interactions between nitrile groups.

28 It is the objective of this chapter to connect the spectroscopic characteristics of nitrile rubber to the nitrile content and the nature of the environment. Molecular vibration, infrared and Raman spectroscopy, as well as Sum Frequency Spectroscopy (SFS) will be used in the characterization of the samples. Specific attention is given to establishing the suitability of SFS to investigate the oil resistance of nitrile rubber since it is the main technique used in this investigation. SFS is a non-linear spectroscopic technique which has two powerful features: surface selectivity and molecular orientation detection. The mathematical equations leading to this unique set of characteristics are presented and spectroscopic observations of nitrile associations in two CN-bearing compounds, acetonitrile and polyacrylonitrile, are reviewed.

The preliminary hypothesis of this work was that the participation of polar groups in dipole-dipole interactions is reflected in the parameters of the bands, which correspond to the vibrations of the polar fragments of the molecules. The main interest, then, was in how this affected the molecular bond properties. For this reason, understanding the parameters controlling the vibrational behavior of molecules is fundamental to this research. Therefore, the beginning of this chapter will be devoted to spectroscopy as a tool for chemical identification.

The basics of infrared and Raman spectroscopy needed to understand Sum Frequency

Spectroscopy are exposed in the next section; a complete theoretical background can be found elsewhere.64

29 3.1 Linear Spectroscopy

A molecule with n number of atoms, rotates, translates and vibrates with a total of 3n degrees of motional freedom. The molecular vibrational modes (3n-6) are characterized by the periodic motions involving changes in bond lengths and bond angles. For now, the case of a diatomic molecule is considered. When treated as a harmonic oscillator, the diatomic vibration behaves as an oscillating spring characterized by a force constant, k ,

that connects two heavy masses, m1 and m2 , as shown in Figure 3.1.

j l0 F k

i

x m m 1 2 Figure 3.1. Ball and spring model for a diatomic molecule

The force F applied to the spring that oscillates around its equilibrium position, l0 , at a distance , x , is given by Hooke’s law:

F = −kx (3.1) where F , k and x stand for the force acting on the spring, the force constant and the displacement from the equilibrium position, respectively. Applying Newton’s law of the equation of motion to the isolated spring system becomes:

d 2 (x) u = −k.x (3.2) dt 2

30 1 = 1 + 1 where u is the reduced mass such that , with m1 and m2 shown on Figure u m1 m2

3.1

Solving Equation (3.2) leads to the following expression:

1 k ν = (3.3) 2π u where ν is the frequency of vibration of the diatomic bond. If units of mass are used and the force constants are expressed in N/m, the frequency of vibration of a diatomic molecule is given by:

  ν −1 = 1 + 1 / cm 1303 k  (3.4)  m1 m2 

The force constant (k) corresponding to spring's stiffness in the ball and spring model

relates directly to the strength of the covalent bond linking m 1 and m 2. and. According

Equation (3.3), a C=N double bond is approximately twice as strong as a C-N single bond, and the C ≡N triple bond is similarly stronger than the double bond. The infrared stretching frequencies of these groups vary in the same order, ranging from 1100 cm -1 for

C-N, to 1660 cm -1 for C=N, to 2220 cm -1 for C ≡N. Intrinsically, the strength of a bond depends on the bonding nature of the molecular orbital. The wave function describing a bonding molecular orbital correspond to a situation where the electron are in between the atom creating a covalent bond. The bond order is equal to the number of bonding electrons minus the number of antibonding electrons, divided by 2.

Furthermore, a direct correlation exists between the force constant, the bond length and hybridization state of the atom.65 This directly means that if the electronic population of a 31 molecule changes upon solvation or interaction, the vibrational frequency is shifted.

Equation (3.3) states that, an increase in the force constant, k , (or a decrease in the bond length) contributes to an increase in the vibrational frequency, while a decrease in force constant, k , (or an increase in bond length) yields to a decrease in the vibrational frequency,ν . In others words a vibrational frequency shift can be either positive or negative.

Finally, the energy levels corresponding to the vibrations of a molecule are quantized and determined by solving the Schrodinger equation:66

 1  E = n + hν (3.5) vib  2  where n is the vibrational quantum number, ν is the classical vibrational frequency of the oscillator and h is the Planck constant.

3.1.1. Infrared Spectroscopy

The determination of this vibrational frequency is achieved when a change in the vibrational energy of the molecular oscillator is induced by an external photon having a frequency matching that of the fundamental frequency of the molecule. If the mechanism of that transfer of energy involves a change in the dipole moment in addition to matching the frequency of the normal coordinates, then absorption of light in the infrared region occurs such as:

32 ∆ = ()+ + 1 ν −  + 1  ν = ν where E  n 1h m n h m h m  2  2 

(3.6)

∆ ν E is the increase in vibrational energy of the molecule and m is the vibrational frequency of the molecule.

To illustrate infrared absorption, the polar CN moiety is considered. A strong dipole moment exists in the CN moiety due to the difference in electronegativity between the carbon and nitrogen atoms. Around the electronegative nitrogen atom, there will be a slight excess of negative charges. Opposite to this situation, the carbon atoms will have a slight excess of positive charges. Exaggeration of this situation allows picturing the positive and negative poles that constitute the carbon and nitrogen atoms, respectively, as a dipole. By definition the dipole moment (usually of magnitude of 10 -30 C.m) points toward the most negative atom as shown on Figure 3.2. As the CN bond stretches and contracts during vibration, the charge distribution changes along with the distance between the carbon and nitrogen atoms. That is, the dipole moment of the CN bond changes while the bond is vibrating. This is the specific mechanism which allows infrared absorption to occur. The CN vibration (around 2250 cm -1) is therefore said to be an infrared active vibration. Also displayed on Figure 3.2 is the case of CO 2. As the oxygen atoms move away from the carbon atom during the symmetric stretching mode at (1337 cm -1), the changes in the dipole moments continuously cancel each other out. Thus, the dipole moment does not change during vibration making the CO symmetric stretch of

33 -1 CO 2 molecule infrared inactive. The CO asymmetric stretching mode (at 2349 cm ), though, is infrared active.

C N

δ+ δ-

p

O C O

δ- δ+ δ-

p p Figure 3.2. Schematic representation of the location of the atoms in the nitrile moiety and CO 2 as the molecules vibrate with their corresponding dipole moment.

3.1.2 Raman Spectroscopy

Infrared and Raman spectroscopy differ in the mechanism of transferring energy between light and matter. For a given vibration the intensity of the Raman band will be different from the intensity of the infrared band despite the fact that the vibration mode will theoretically occur at the same wavenumber. When a molecule is placed in an electric field, electrons will move in the opposite direction of the field gradient while the protons will move in the same direction as the field gradient. The spatial separation between

34 electrons and protons is small but sufficient to create an induced dipole moment in the molecule. E is the external electric field applied to the molecule producing an induced dipole moment proportional to the polarizability of the molecule. By definition

= α   (3.7) where is the dipole moment of the molecule and α is the polarizability of the molecule.

Such that, if an oscillating field is applied where

 = πν E0 sin 2 t (3.8) the magnitude of the dipole moment will follow as

 = α πν E0 sin 2 t (3.9)

The magnitude of the dipole moment is proportional to the deformability of the electrons or the ease with which they move away from the positive nuclei, in other words to the deformability of the electron cloud. This property, conventionally called the atom’s polarizibility, is directly proportional to the volume and the shape of the molecule which may or may not change during the vibration of the molecule. It will also depend on the geometry of the molecule. In this case, the simplest way to account for it is to expand the polarizability according to the normal coordinate, Q, as the following:

∂α α = α + Q + ... (3.10) 0 ∂Q

35 In this equation Q is the normal coordinate rigidly fixed to the molecule and ∂α is the ∂Q rate of polarizability change as the molecule vibrates. Internal coordinates measure the change in shape of a molecule as compared to its equilibrium shape.

Since the normal coordinate is attached to the vibrating atoms, we can assume that:

= πν Q Q0 sin 2 vt (3.11) and by plugging this into Equation (3.10) it is easy to eventually demonstrate that

∂α α = α + Q sin 2πν t (3.12) 0 ∂Q 0 v and

∂α µ = α E sin 2πν t + Q E (sin 2πν t)(sin 2πν t) (3.13) 0 0 ∂Q 0 0 v

Finally, after reorganization:

∂α µ = α E sin 2πν t + Q E []cos 2π (ν −ν )t − cos 2π (ν +ν )t (3.14) 0 0 ∂Q 0 0 v v

This last equation stipulates that the induced dipole moment, , varies with three

ν −ν ν +ν component frequencies v , v , v , which are responsible for Rayleigh, Stokes and anti-Stokes scattering, respectively. The intensity of the Stokes and anti-Stokes lines are both directly related to the variation of the induced dipole moment or variation in polarizability. The intensity of the Rayleigh scattering is related to the polarizability of the molecule.

36 Experimentally, the sample is illuminated with light and the scattered light is analyzed.

The first term of Equation (3.14) corresponds to the major part of the scattering. The photon and molecule are colliding elastically leaving the rotational and vibrational energy of the molecule unchanged. Scattered photons match in energy and frequency with those of the incident light. In the ground vibrational state, when a molecule absorbs a photon, the energy of the molecule increases as the molecule reaches a higher level of energy for a very short period of time. As the molecule relaxes, the excess of energy just gained is

ν released in the form of a photon having an energy corresponding to h i , which is

Rayleigh scattering. If the rotational and vibrational energies of a molecule are changed during an inelastic collision with a photon then,

ν − ν = ∆ = ν h i h s Em h m (3.15)

ν ν where i is the frequency of the incident photon and s is the frequency of the scattered

ν ν photon. If the molecule gains energy then s is smaller than i and this is known as the

ν ν Stokes line of the Raman spectrum. If the molecule loses energy then s is larger than i and is called the Anti-Stokes line of the Raman spectrum. The Boltzmann distribution function implies that higher levels of energy are less populated. This means that the intensity of the Stokes line is greater than that of the anti Stokes lines at room temperature. The main point is that if the vibration causes no change in the polarizability, then the intensity of the band associated with this vibration would be null and the band would be said to be Raman inactive. The polarizability would be in the most general case a 3x3 tensor or matrix.

37 The magnitude of polarizability, expressed in volume unit, is related to the volume of the loose electron cloud. Large scatterers will be typically large polarizable molecules and delocalized systems such as rings, double and triple bonds. The polarizablity tensor is often represented using a so-called polarizability ellipsoid. The construction of the ellipsoid starts by drawing vector in specific directions having length proportional to

1 .The end of the vectors forms a 3D surface whose distance from the electric center α i

α of the molecules is proportional to where i is related to the symmetric real polarizability tensor.

Thus, the CN stretch vibrational band, in addition to being infrared active, will also be

Raman active. For the same reason the CO symmetric stretch of the CO 2 molecule is

Raman active and the asymmetric CO stretching mode is Raman inactive.

Whether linear infrared or Raman is employed to probe molecular vibration, the approach is the same, vibrational frequency unravels molecular identity and a vibrational frequency shift indicates interaction with specific molecules. A positive shift delocalizes toward a higher frequency as shown in Figure 3.3. It is also know as “hypsochromic” shift, from

Greco hupsos, “height, elevation” or “blue” shift. A negative shift corresponds to a decrease in the vibrational frequency, also called “bathochromic”, from Greco root

“depth” or “red” shift. A shift in vibrational frequency of a mode indicates a change in the characteristic of the bond. Therefore changes in the electronic population of a

38 molecule upon specific interactions such as hydrogen bonding or during solvation process are indicated by a spectroscopic vibrational shift.

Red shift Blue shift k decreases k increases 2.0

1.5 k k

1.0

Abs(a.u.) 0.5

0.0 Wavenumber (cm -1 )

Figure 3.3. Correlation between the vibrational frequency shift and the change in force constant magnitude.

3.1.3 Non-Linear Effect

Molecules will fail to respond harmonically by coupling of harmonic modes. In most cases it will lead only to the appearance of spectroscopic manifestations such as hot band and Fermi resonance.

From a general point of view, anharmonicity creates new levels of energy such as combination, difference and overtone. Overtones occur as a necessity to accommodate the dependence of the dipole moment on the intensity of the electric field. As the motion of the molecules slightly diverts from harmonicity, the dipole moment of the vibrating molecule will not only oscillate at the fundamental frequency, but at the double, triple

39 and so on of the fundamental frequency. Along with overtones, combinations and differences bands are also possible. Theoretically, peaks at the sum or difference of the fundamental frequency exist. However, they will be very weak and contribute only in a minor proportion to the spectral features of a molecule. In other cases though, the impact of anharmonicity is sufficient to be experimentally observable, through phenomena such as vibrational hot band,64 Fermi resonance,67 and the vibrational Stark effect. 68-70

Since the results of this project will be compared to those of acetonitrile, it is important to understand two of the three effects previously cited, namely the Fermi resonance and the hot bands. The following sections will be dedicated to explaining these two effects.

3.1.3.1 Hot Band

The term “hot” band was most likely created by Herzberg and Ottawa in the 1930's.71 A

ν ν hot band is a transition involving two excited states from 2←1 and 3←2 . As molecules are in thermal equilibrium, the population of each level contain a Boltzmann correction such that

−Ei n e kT i = (3.16) n −E0i 0 e kT

th where n i and n 0 are the number of molecules in the i and 0 levels. (E i-E0) is the energy difference between the levels, k is the Boltzmann constant and T is the absolute temperature (Kelvin scale). The intensity of the hot band is very low compared to the fundamental transition because it occurs from two excited levels with populations 40 decreasing as it obeys the Boltzmann distribution. Figure 3.4 (A) shows a transition from

ν = ν = an excited level i 1to i 2 . Such a transition is possible at room temperature only if the energy of the mode is lower or close to RT ≈ 209 cm −1 . If this is not the case, most

molecules will be found in the ground level . A second type of hot band can be expected at room temperature when low energy vibrational modes excited at room temperature couple with a high energy vibrational mode like it is shown in Figure 3.4 (B). A recent example of a hot band transition can be found in the literature. 72

A Energy scheme and vibration spectrum corresponding to vibration i:

Vibration i

Fundamental νi=2 Cold band

νi=1 Hot Band Frequency Ground State

νi= νCN

B Energy scheme and vibration spectrum corresponding to vibration i coupled with j

Fundamental νi=2 Cold band

νi=1 Hot Band Frequency Ground State

νi= νCN

ν 2 lower energy mode excited at room temperature

Figure 3.4. Energy scheme and corresponding vibrational spectrum for fundamental and hot transition.

41 3.1.3.2 Fermi Resonance

Fermi resonance arises when a combination band (or an overtone) resulting from anharmonicity lies sufficiently close to the fundamental band. Thus, instead of observing only one fundamental band, two bands involving the fundamental and the combination are observed. The intensification of the combination band is to such an extent that a second band is observable. In the event that the combination (or overtone band) is not sufficiently close to the fundamental band, the Fermi resonance will not exist (See Figure

64 1.14). Molecules exhibiting Fermi resonance are , CO 2, and acetonitrile.

3.2 Non-Linear Spectroscopy

The following section highlights the salient features of SFS spectroscopy; further details can be found in the work of Hirose 73-75 and Shen.76, 77 Non-linear spectroscopy involves physical processes of light mixing where the frequency of an input beam is converted into another one. Historically, the first experimental observation of light conversion was reported by Franken et al.78 using a ruby pulsed laser. A monochromatic light (694 nm) was focused onto to a piece of crystalline quartz and an output at 347 nm was detected.

The experiment is conceptually illustrated in Figure 3.5. Two incident photons were converted into one emerging photon with exactly twice the energy or half the wavelength.

42 Second H armonic Generation using Quartz:

γ 2 x γ Input laser beam Output light

E k E k

B Quartz B Non -linear material

Photon Figure 3.5. A sketch illustrating the production of SHG through crystalline quartz with vibrational excitation at 694 nm and the corresponding harmonic at 347 nm.

This doubling of the input beam frequency was called Second Harmonic Generation

(SHG). Note here, that this non-linear effect occurs in the bulk medium of crystalline quartz.

This interaction between light and matter arises when molecules are impinged by extremely intense electric fields such as those produced by pulsed laser.

Under laser exposure the polarization of a material is given by:

α β γ  =  +   + M + 0 ... (3.17)

α β γ where  is the polarization vector,  is the electric field vector, and 0 , and are the first, second and third order electric hyperpolarizability tensors of the molecule, respectively. This expression is also known as the electric dipole approximation where quadrupoles are neglected.

43

α In Equation (3.17), 0 is the linear polarization of the molecule, self-sufficient when the electric field applied is low. The incorporation of higher order terms captures the non-linearity of the induced polarization of the molecule. The first non-linear term is

( ) = accounted by the     . The anisotropy of condensed matter requires expanding

β as a 3x3x3 tensor. It is easy to show that:

() ( ) ( ) 1 2 ( )  = βE cos νt cos νt = βE 1+ cos 2νt (3.18) 0 2 0

= ν ν The electric field,  , is defined as E E0 cos t with an oscillating frequency . The first term of Equation (3.18) represents a static electric field and the second term corresponds to a dipole oscillating at double the frequency and emits light at 2 ν or SHG.

A direct correlation of SHG is the Sum Frequency Generation (SFG) where, two intense

ν ν beams of frequency 1 and 2 excite a material and induce a polarization which oscillates at the sum of the two incoming frequencies as shown in Figure 3.6.

44 E k γ1

B

E k

B γ3 =γ1+ γ2 E k Non-linear material γ2 B

Figure 3.6. A sketch illustrating the SFG process through non-linear material.

The mathematical expression for the molecular hyperpolarizability tensor, as a function of the two input frequencies when none of the two input frequencies is resonating with an electronic transition of the molecule, is: 76

 ∂α   ∂µ  lm n     Q  ∂Q   ∂Q  β = q q ()ν ,lmn ∑ . (3.19) 2 ()ν −ν + Γ q=1 2 q i q where (l,m,. n) is the Cartesian frame attached to the molecule associated with the set of

 ∂α  axes (a,b,c) , c is the axis of highest symmetry of the molecule,  lm  is the lm  ∂   Q q

 ∂µ  component of the Raman tensor and  n  is the n component of the transition dipole  ∂   Q q

ν Γ moment. q and q are, respectively, the frequency and width of the resonance, q. The microscopic description of the SFS tensor, represented by Equation (3.19), stipulates that

45 an SFS active mode is, by definition, an Raman and infrared active mode; if not, the hyperpolarizability tensor elements vanish according to the specific point group of the molecules. For example, the -C≡N moiety introduced to illustrate the Raman and infrared activity of the CN stretching is an SFS active vibrational mode. The hyperpolarizability tensor, therefore, comprises molecular information about the chemical structure of the molecule under light excitation.

A direct analogue of Equation (3.19) exists at a macroscopic level. When an oscillating field is applied to condensed matter, a dipole proportional to the applied electric field is induced such as:

χ (1) χ (2) χ (3)  =  +   + M + ... (3.20)

( ) ( ) ( ) where χ 1 , χ 2 and χ 3 are the first, second, and third order susceptibility tensors of the medium, respectively. Equation (3.20) is an analogue of Equation (3.19) that relates the polarization (dipole per unit volume) of molecules to the applied electric field. The

( ) dimension of the susceptibility tensor, χ 2 , matches that of the hyperpolarizability tensor and is a 3x3x3 tensor. A correspondence between the microscopic hypolarizability and

β the macroscopic susceptibility tensor can be obtained by projecting the lmn ,q into the laboratory axes (xyz ) from the attached molecular axes (abc ) using the Euler angles. The

Euler transformation coefficients are functions of ψ ,θ , and φ and are represented in

Figure 3.7. The original notation from Hirose 73 is modified for the Euler angle, χ , and is referred to,ψ , instead. The coordinate transformation includes three rotations, a first

46 rotation of the N axis of an angle φ around the c axis and a second of an angle ψ around the z axes. The third rotation is that of the c axis around z axis of an angle θ .

z c θ Φ N ψ b y x a

Figure 3.7. Definition of Euler angle relating the attached molecular axis frame (abc ) to the laboratory axis system (xyz ).

β The expression of lmn in the laboratory axis corresponds to

β (ν Ω) = (Ω)β (ν ) ijk 2 , ∑U ijk ,lmn lmn 2 (3.20) l,m,n=a,b,c

Ω Ω = (ψ θ φ) where is the Euler angle ,, and Uijk :lmn is the 27x27 Hirose projection

β coefficient. The contribution of all molecules, providing their ijk s is non-zero, adds up to an extent that depends on the number N and the molecule orientation. A probability distribution function ( f (Ω)) accounts for the orientation of each molecule, and is mathematically expressed as

  = ()Ωβ ()Ω Ω Aijk ,q N ∫ ∑U ijk :lmn lmn ,q  f d (3.21)  lmn =abc 

47 where Aijk ,q is the resonant component of the susceptibility

Q A χ ()ν = χ NR iφ + ijk ,q ijk 2 e ∑ (3.22) ij ijk k ν −ν + Γ q=1 ir q i q and where φ (not related to Euler angle) is the possible phase between the non-resonant

(wavelength independent) term and the resonant term(s).

By probing the susceptibility tensor, the symmetry of the local arrangement is captured because the symmetry of the molecules is embedded into the macroscopic response of the material to an electric field (also known as Neumann principle). This principle applies to many other physical properties such as elastic deformation, the piezo-electric effect, the magnetic effect, and pyroelectricity. The symmetry of a material affects the susceptibility tensor by reducing certain terms of the tensor to zero. The number of vanishing terms will strictly depend on the level of symmetry of the local molecular

( ) ( ) arrangement observed. For a centro-symmetric medium χ 2 = −χ 2 = 0.

The main conclusion is that media such as bulk polymer that shows inversion symmetry cannot produce any second order effect. The interface, though, by definition, is a non- centrosymmetric medium and will exhibit a non-linear effect. For that reason, SFS is a surface selective tool.

In an SFS experiment, from the laboratory axis point of view, two incoming beams overlap on a sample and generate surface polarization which gives rise to SF signal in the 48 θ (ν ) θ (ν ) reflected and transmitted directions. i 1 and i 2 are the angle of incidence of the

ν ν θ (ν ) θ (ν ) beams 1 and 2 , respectively. The angles of reflection and refraction are r 1 , r 2 ,

θ (ν ) θ (ν ) t 1 , and t 2 respectively. The direction of the SFS beam propagating in medium 1

θ θ and 2 is given by r and t as shown in Figure 3.8.

The excitation source and the refractive index of the medium characterizing the material at the interface (see Figure 3.10) are related via the following equation 73

θ (ν ) = θ (ν ) θ (ν ) = θ(ν ) i 1 r 1 , i 2 2 (3.23)

The directions of the reflected and transmitted beam in media 1 and 2 are given by the following relationships

ν (ν ) θ (ν )+ν (ν ) θ (ν ) = ν θ 1n1 1 sin r 1 2 n1 1 sin r 2 SF n1 sin r , (3.24)

ν (ν ) θ (ν )+ν (ν ) θ (ν ) =ν θ 1n2 1 sin t 1 2n2 2 sin t 2 SF n2 sin t (3.25)

γSFS,r γ1,i γ1,r Ep θi (γ1) θr (γ1) γ2,i ES θi (γ2) θr (γ2) Linear medium γ 2,r (air,oil…)

n1 Non-Linear medium : interface

n2

γ1,t Linear medium: polymer bulk

z γSFS,t y γ2,t x

Out of page Figure 3.8. Schematic of an SFS experiment at an interface in external reflection. All the light beams lay in the plane of incidence (xz). The two polarization cases, S and P, have been illustrated for the incident infrared beam. The directions of the laboratory–fixed coordinate axes are indicated.

49 In the laboratory axes for the polymer interface, it is assumed that SFS active groups at the surface are isotropically distributed in φ and ψ . In other words, χ(2) must be rotationally invariant around the z axis. Only seven combinations of i,j,k out of 27 are

χ = χ χ = χ χ = χ χ non-vanishing; XXZ YYZ , XZX YZY , ZXX ZYY , ZZZ need to be determined.

These susceptibility tensor elements are experimentally determined by setting-up the polarization of the SFS experiment. Three polarizers convert, respectively, the SFS, visible and infrared beam to a single polarization state: S or P. It is a convention to report the polarization state of the three beams precisely in this order of decreasing frequency. S polarization corresponds to an electromagnetic wave having its electric field perpendicular to the plane of incidence (x-z, see Figure 3.7) and a P polarization signifies a polarized electric field parallel to the plane of incidence (y-axis). The polarization of the excitation and SF beams allows probing the component of an effective susceptibility

( ) χ 2 whose expression depends on the polarization of the beam. The correspondence eff

( ) between the χ 2 and the elements of the susceptibility tensor is as following: eff

χ (2 ) = (ν ) (ν ) (ν ) θ r χ eff , SSP LYY 3 LYY 1 L ZZ 2 sin 2 YYZ (3.26)

χ (2) = (ν ) (ν ) θ r (ν )χ eff ,SPS LYY 3 LYY 1 sin 1 LZZ 2 YZY (3.27)

The L pre-factor is the Fresnel optical coefficient associated with the incoming beams; it relates the intensity of the input field with that at the interface. The Fresnel factor is related to the SF beam intensity to the non-linear source polarization of the interface.

50 χ (2) ν Finally the relation between eff and the intensity of the beam at 3 is:

()2 2 I()ν =ν +ν ∝ χ I ()()ν I ν (3.28) 3 1 2 eff 1 1 2 2

It has been shown before that the tilt angle ( θ ) of a moiety at the interface can be experimentally determined. As this study concentrates on the detection of the nitrile moiety, a calculation of the angular dependency of the SF signal enhancement of the nitrile CN stretching mode is provided. The CN belongs to C ∞v symmetry group. As a usual convention, the (a,b,c) axes are attached to the molecule in such a way that the c axis coincides the rotation axis of the molecule.

The CN related to this study expands as a function of the tilt angle θ of the c axis Aijk ,q from the surface normal (z axis). The calculation is carried out by first identifying the

β β non-zero abc from the symmetry of the molecules. Each abc is transformed

β 76 into ijk using the Hirose Table and finally expressed as:

cos θ cos θ − cos 3 θ A = N ()β + β + ()− β − β + 2β (3.29) YYZ ,CN CN aac bbc 2 aac bbc ccc 4

cos θ − cos 3 θ A = N ()2β − β − β (3.30) YZY ,CN CN ccc aac bbc 4 where cos θ and cos3 θ represent the average of cos θ and cos 3 θ over all the orientations of the molecules at the surface.

51 3.3. Spectroscopy for Probing Interactions

Nitrile associations that present spectroscopic manifestations have been observed in two nitrile-bearing compounds: acetonitrile and polyacrylonitrile. In addition, acetonitrile and polyacrylonitrile are extremely sensitive to their environment and “micro-environment”.

Multiple examples will be shown where nitrile side groups of acetonitrile form specific interactions with certain molecules in the environment. In each case the spectroscopic proof of the interaction will be presented.

3.3.1 Acetonitrile Model Compound

In the family of nitrile bearing compounds, acetonitrile is the simplest molecule. It is constituted of a methyl group (CH 3) and a nitrile group (-C≡N) as shown in Figure 3.9.

Acetonitrile has been extensively analyzed using infrared and Raman spectroscopy because it offers the advantage of low spectral interferences in the region where the most characteristic feature, the C ≡N stretching mode, is detected.

Figure 3.9. Acetonitrile chemical structure.

From different points of views, acetonitrile is a molecule with singular physical and spectral properties. For example, the boiling point of acetonitrile is 82 °C compared to that of hydrogen-bonded methanol which is 64.7 °C. This observation led to the hypothesis that dimers and higher order aggregates exist in acetonitrile solution.

52 Temperature- dependent studies of acetonitrile mixtures could not confirm the existence of dimers. 31 Furthermore, from a spectroscopic point of view, when mixed with water, acetonitrile forms a hydrogen bond with the water molecule, which is associated with a positive shift of the nitrile stretch band. This shift is unusual compared to the carbonyl analogue which is indicated by a negative shift of the carbonyl stretch vibration.

Furthermore, acetonitrile is also well known to exhibit a Fermi resonance in the (2000-

2300 cm -1) spectral region 31, 79, 80 .

Studies have concentrated on acetonitrile behavior as a solvent such as the one carried by

Nyquist.79 Nyquist studied mixtures of acetonitrile with a series of solvents having different Guttmann acceptance numbers (AN). AN is a number measuring the acidity of the solvent. Nyquist observed frequency shifts of the nitrile stretching band of acetonitrile in various solvent mixtures. For acetonitrile, the bulk CN stretch occurs at 2253 cm -1.

He first noted a positive shift of the CN stretching band compared to that of neat acetonitrile in almost all of the mixtures. Acetonitrile was found to exhibit a positive shift upon hydrogen bonding or coordination with a Lewis acid with a magnitude correlating to the AN number of the solvent. The range of nitrile frequencies of the acetonitrile was from 2253 cm -1 to 2257 cm -1 as the solvent AN number ranged from 0 with hexane to 41 with methyl alcohol. One exception, though, was found in the mixture of acetonitrile/dimethylsulfoxide (DMSO) that showed a negative shift. For a mixture of with most solvents, Nyquist noted a negative shift. As he tested different compositions of mixtures, it was observed that in the 10% benzonitrile/hexane mixture the CN stretch was more red- shifted (2231 cm -1) than that of the 1% benzonitrile/hexane 53 mixture (2233 cm -1). Nyquist thought that the dipole interactions between benzonitrile molecules were more important in the 10% solution. Thus, it was directly responsible for the red shift of the C ≡N stretching mode. This set of experiments shows that hypsochromic and bathochromic shifts of the C ≡N stretching mode of acetonitrile are possible.

This unusual occurrence is predicted on the basis of the following reason. There are two sites, thus two orbitals, available for interaction in the C ≡N triple bond. The first one is the lone pair orbital of the nitrogen and the second one is one of the π -orbitals of the triple bond in C≡N. If interaction occurs via the lone pair orbital of the nitrogen atoms, a blue shift of the C ≡N stretch vibration will be observed. Interaction between nitrile and a molecule can occur via a dative (or coordinate covalent) bond, which is donated from the lone pair orbital to the acceptor orbital of another molecule. The lone pair electron orbital has a non-bonding character in nature since it is, in essence, neither bonding nor anti- bonding. Upon interaction or chemical reaction, the orbital changes; that is, the amount of bonding/anti-bonding character of the orbital changes. If the anti-bonding character of the lone pair orbital increases, then the bond length of CN will increase. If the anti-bonding character decreases, then the bond length of CN will decrease. Thus, the important concept is the reaction of the anti-bonding character of the lone pair orbital to an interaction. If the anti-bonding character of the lone pair orbital decreases, the CN bond length will decrease and a blue shift will be observed. The main reason for blue shift is an electronic effect which, in turn, changes the bond length.

54 If the interaction occurs via a π -orbital, then a negative shift of the nitrile CN stretch is observed. Another way to understand this is by correlating this concept to the case of olefin and acetylene molecules that coordinate with the π-orbitals leading to a lowering of the bond order of C=C and C ≡C vibrational stretching modes. It is the lowering of the bond order of the nitrile CN stretching during an interaction via one of the π-orbitals that explain a bathochromic shift of the CN stretch.

Intrinsically, the nitrile bond can interact via two different sites. Each interaction type has a specific shift direction allowing a straightforward identification. The specific character of the nitrile bond has been modeled in a recent simulation correlating the relative orientation of a water molecule interacting with acetonitrile, shown in Figure 1 by Doo-

Sik Ahn and Sungyul Lee. 81

Furthermore, complementary studies investigated how the acidity of the surrounding medium affected the nitrile C ≡N stretching band of acetonitrile. In those studies, acetonitrile behaved as a spectroscopic molecular probe where the monitoring of the C ≡N stretching band provided information about some physical properties of the environment.

Among the most successful molecular probes are and acetonitrile. 80 Pyridine was frequently used as a probe for IR detection of acid sites on and mixed oxides.80 Unfortunately, its strong basicity limits its applications because of its inability to distinguish surface acidic sites. Acetonitrile, due to its small size and moderate basicity, became an attractive molecule to probe acidic environments.80 In this case, the

55 mechanism for the detection of acidic sites is the sensitivity of the CN stretching vibration to the electron withdrawing power of the metal ion to which the nitrile is associated. The acidity strength distribution will be directly reflected in the wavenumber shift of this band.

Knoezinger et al. 80 compared the IR spectrum for the adsorption of acetonitrile on δ-

Al 2O3 with the IR spectrum of liquid acetonitrile. The experiment consisted of adsorbing acetonitrile onto alumina disks by increasing the pressure. At low pressure, the OH stretching bands of the surface hydroxyl groups did not undergo significant changes. The strongest modification came from the C ≡N spectral region, where two blue shifted bands occurred at 2300 and 2328 cm -1. The perturbation of the C ≡N region with the consistency of the OH region led to the conclusion that the adsorption of acetonitrile was made between the Al 3+ ions on the surface, leaving any hydroxyl sites out of the interactions. Interestingly, at higher pressure the 2253 cm -1 band increased and the OH region was perturbed, which means that, at this particular pressure, the coordination between acetonitrile and the alumina surface changed and involved the hydroxyl groups.

This raised a second unusual experimental observation as multiple peaks in the C ≡N region were observed despite the existence of one unique vibrational mode.

Assigning the IR spectrum of acetonitrile can be challenging, like in the case related to the red-shifted shoulder of the C ≡N fundamental band observed by Rowlen and Harris 31 .

While studying an acetonitrile/water mixture, Rowlen and Harris realized that the low

56 frequency band was actually a convolution of two bands. They named the low frequency

31 component as νII and the high frequency component as νIII (see Figure 1 from ref ).

The low frequency shoulder, νIII , was attributed to self-associated acetonitrile molecules and the high frequency band, νII , was attributed to the non-associated acetonitrile molecules. The respective locations of these two bands were explained by the fact that, in an associated state, the dipole moments of the acetonitrile aggregates were lower and a partial cancellation of the dipole charges lowered the strength of the bond which, in turn, red shifted the CN band with respect to the free nitrile band. Fini and Mirone 82 rejected this model based on their assertion that dimerization was the origin of the assignment of the low frequency shoulder. This shoulder was eventually called “secondary” structure.

They proposed that the red-shifted feature, according to ν2 , is a hot band. By definition, a hot band appears in the spectrum as the temperature increases and is unaffected by dilution. This debate is still open and reflects a delicate assignment situation.

3.3.2 Polyacrylonitrile Model Compound:

Figure 3.10. PAN chemical structure.

The chemical structure of polyacrylonitrile is presented in Figure 3.11. The interaction between the nitrile groups of acetonitrile was shown to be sensitive to the dipole

57 concentration and their degree of association, while the interaction between the nitrile groups of polyacrylonitrile mainly depends on the temperature. Hence, the spectroscopic behavior of polyacrylonitrile was investigated as a function of temperature, demonstrating spectroscopic manifestations of the nitrile groups during complexation and cyclization. The study of the mechanism of thermal degradation is motivated by the carbon-product fabrication as polyacrylonitrile is a superior precursor for carbon black.

The case of polyacrylonitrile is slightly more complex than that of acetonitrile because a chemical change can induce a shift of the spectra, which can complicate the spectra assignment process.

Andreeva and Burkova 83 investigated the physical changes occurring before the chemical degradation of polyacrylonitrile. They showed (see Figure 1 from ref 83 ) that the intermolecular interaction of the CN group is revealed when the CN stretching vibrational frequency changes in intensity and position during heating. They recorded the in situ position and intensity of the nitrile stretching vibrational mode during a continuous cyclic thermal treatment.

The C ≡N stretching vibrational mode shifted from 2242 cm -1 to 2236 cm -1 during the first phase of the heating from 0 to 90 °C. Above 100 °C, the location of the band stabilized and the intensity started to decrease. This sequence of spectral events reversed as the temperature was decreased.

58 Two chemical changes occur in PAN as it is heated up. First, a polymerization of the

C≡N side groups occurs at 200 °C, consuming the C ≡N bonds and producing conjugated

C=N double bonds and C=C double bonds. 84 IR spectroscopy shows the disappearance of the 2242 cm -1 band and the appearance of a band in the range 1400-1600 cm -1. This has been called a polymerization of C ≡N because, in the parallel direction to the backbone of the polymer, C ≡N groups react with each other leading to a conjugated polymer. Figure

3.11 shows how the C=N bonds are formed.

Figure 3.11. Structure of singly-conjugated pyrolyzed polyacrylonitrile.

At this point, the carbon polymer backbone is untouched, and CH 2 is still present. Real degradative loss starts at 265 °C and reaches 34% at 435 °C, where the nitrogen content begins to decrease. The second step involves the aromatization of the polymer chains. occurring in the temperature range from 265 °C to 423 °C. This reaction results in a double conjugated polymer structure, as shown in figure 3.12.

Figure 3.12. Structure of conjugated pyrolyzed polyacrylonitrile.

59 Pyrolyzed polyacrylonitrile (PPAN) on aluminium led to a conjugation reaction occurring at a lower temperature than that of the bulk PAN (400 °C).85 On aluminium, the disappearance of CH 2 methylene stretching mode indicated the formation of a conjugated backbone. This experimental observation coupled with the appearance of a peak at 800 cm -1 confirms the creation of a conjugated polymer chain at 300 °C. Interestingly, the conjugation reaction occurred first at 200°C even though there was still a C ≡N signal in the IR spectrum. This confirms the presence of unreacted CN. The C ≡N signal also shows features at a downward shifted peak at 2227 cm -1, indicating an interaction between aluminium atoms and the nitrile groups. The authors concluded that the reduction in the required temperature of the conjuguated backbone formation was a consequence of the nitrile/aluminium interaction. For the case of the copper/PAN, a conjugation reaction occurred. The C ≡N moiety did not react but complexed with copper atoms. Interestingly, a split in the vibrational frequency occurred, and two C ≡N were observed due to the diffusion of copper atoms into PAN at 200 °C. The authors observed a split of the C ≡N groups. The low component contribution arose from side-on coordination between the nitrile groups with copper atoms and the high component is attributed to end-on complexation between side nitrile groups and copper atoms 18 .

PAN interacted with water under specific conditions of temperature and pressure to allow melt spinning of PAN fibers at 180 °C under 40-70 atm.18 This experimental observation was confirmed by the fact that finely divided PAN powder formed a single phase with water at that pressure. As reported, the CN stretching band at 2234 cm -1 disappeared and a new band at 2050 cm -1 appeared. Such a large bathochromic shift 18 was not expected 60 and was described as a certain type of hydration. Furthermore, such an interaction was reversible; the band would shift back to 2234 cm -1 upon cooling of the PAN.

3.3.3 Frequency Shifts at Surfaces

Many recent studies were conducted to investigate the molecular structure of liquids at the surface of a polymer along with the interactions between a surface polymer molecule and a solvent. Some of these involved the detection of a vibrational shift only at the surface. Li et al. 17 investigated the surface of poly(2-methoxyethyl ) (PMEA).

Their studies focused on the interaction of water molecules with the carbonyl function of

PMEA. The carbonyl function has the ability to hydrogen bond with a hydrogen donor solvent, such as water or ethanol, resulting in a red shift of the carbonyl stretch. To distinguish the carbonyl at the surface of PMEA, they used SFS on a PMEA thin film in water. The water structure at the surface of PMEA was unclear and expected to be related to the blood biocompatibility of PMEA. The SFS spectrum clearly showed a red shifted carbonyl at the surface of PMEA while bulk characterization of the carbonyl groups using infrared absorption reflection showed only a bulk carbonyl at a vibrational frequency typical of a free carbonyl.

Loch et al. 86 investigated the interfacial structure between poly(ethylene terephtalate)

(PET) and 3-aminopropyltrimethoxysilane. The classical red shift of the carbonyl stretching was observed using SFS. They presented this as proof of hydrogen bonding of the silane group and ester carbonyl of PET.

61 The mixture acetonitrile/water was studied by Zhang and coworkers.87 The CN stretching vibration of acetonitrile at the interface (acetonitrile/water) mixture/air was monitored using SFS. They observed that below a certain bulk composition (X CH3CN =0.07), the CN stretch vibration frequency at the air/mixture interface was higher than that of the bulk by

14 cm-1. At a bulk concentration higher than 0.07 mole fraction, molecules at the surface vibrate at a frequency that was close to that of pure acetonitrile. They concluded that a nitrile at the surface experiences two different environments: one below the bulk molar fraction of 0.07 where the CN vibrational frequency is characteristic of a CN hydrogen bonded to water, and a second one above the bulk molar fraction of 0.07 ,where acetonitrile no longer hydrogen bonds with water. The transition between both was sharp.

Therefore, above a bulk concentration of 0.07 mole fraction, the acetonitrile molecules at the interface had spectral and structural properties approaching that of the pure acetonitrile. It is noteworthy to mention that acetonitrile is miscible in water and yet an abrupt transition occurred at the surface leading to a surface segregation in favor of acetonitrile molecules. This implied that the surface of acetonitrile could not be predicted based exclusively on bulk properties.

3.4 Messages of This Thesis

The goal of this research was to use SFS to analyze the surface of PAN, PBD and NBR

(40% and 20% ACN) in the nitrile stretching region of the spectrum. The surfaces of these three polymers were analyzed in the dried state and under liquid exposure. As previous research showed spectroscopic manifestations of interactions between nitrile groups and water, it was therefore the first liquid to be tested at the PAN and NBR 62 surface. Furthermore, in order to investigate the NBR oil resistance, the PAN and NBR surfaces were characterized under heptane exposure. The last solvent to be included in the study was toluene, a known aggressive swelling agent for rubber.

The objective of this approach was to develop an understanding of how rubber/oil interaction at a molecular scale affects the oil resistance of NBR. The overall solvation process of NBR would therefore be captured by microscopic interactions rather than individual physical data. It was of interest to investigate how polar association, or dipole interaction, affects the relationship between the oil resistance of a specific nitrile rubber and its molecular capability to associate. Thus, an old problem was investigated using a new tool to correlate the oil resistance of a polymer to simple properties of its molecules with direct spectroscopic evidence. To understand the nature of the vibrational perturbation of solvent on the CN stretch of the nitrile at the surface of nitrile rubber, the scope of the investigation covered the behavior of the CN stretching vibration in rubber of varying CN composition from 40% to 20% ACN in various environments, such as air, water, heptane and toluene. Since the copolymer was relatively complex, this investigation started by analyzing the homopolymers PAN and PBD. The investigation of

PAN provided an opportunity to focus on polar associations of PAN and formed the basis for the NBR analysis.

63

CHAPTER IV

EXPERIMENTAL

This section describes the sample preparation procedures and the various experimental techniques that were employed in this work.

4.1 Polymer Thin Film Preparation

All solvents used were HPLC grade. Reagent grade dimethylsulfoxide (DMSO), heptane, toluene and methyl ethyl ketone (MEK) were purchased from Aldrich Chemical

Corporation and used as received. Water was distilled and deionized using a MilliQ system (18 M ). Unstabilized THF was ordered from Aldrich (grade CHROMASOLV

Plus). All solvents were used as received; unstabilized THF was used no longer than a week after the bottle was opened. All substrates were cleaned as followed: rinsed with ethanol and heptane, blown dry in flowing nitrogen, and cleaned in an air-plasma oven before being used.

Polyacrylonitrile (PAN) (M N = 22,600 /M W=86,200g/mol) was purchased from Aldrich

Chemical Corporation, and used as received. Thin films (500 nm) were prepared by spin coating of a 4 wt % solution PAN in DMSO on sapphire prism at 2400 rpm for 15

64 seconds. The solution was left overnight for homogenization. After coating, the film was annealed at 90-120 °C for 8 to 12 hours.

Polybutadiene (PB) was ordered from Polymer Scientific (CAS unavailable) and used as received. Thin films (800 nm) were prepared by spin-coating of a 4 wt% PB in unstabilized THF solution on sapphire prism at 2400 rpm for 15 seconds. The solution was left to rest overnight. After coating the film was annealed at room temperature for 4 hours.

NBR (40% ACN) (M N = 113,000 /M W=344,000g/mol) was purchased from Polymer

Scientific (CAS # 532) and used as received. Thin films (250 nm) were prepared by spin coating a 4 wt% solution of NBR (40% ACN) in unstabilized THF on a sapphire prism at

2400 rpm for 15 seconds. The film was then annealed at room temperature in vacuum for

4 hours.

NBR (30% ACN) (M N = 125,000 /M W=453,000g/mol) was purchased from Polymer

Scientific (CAS # 529) and used as received. Thin films (250 nm) were prepared by spin coating a 4 wt% solution of NBR (40% ACN) in unstabilized THF on a sapphire prism at

2400 rpm for 15 seconds. The film was then annealed at room temperature in vacuum for

4 hours.

65 NBR (20% ACN) (M N = 128,200 /M W=460,800g/mol) was purchased from Lanxess

(CAS # 1846-F) and used as received. Thin films were prepared by spin coating a 4 wt% solution of NBR (20% ACN) in MEK on a sapphire prism at 2400 rpm for 15 seconds.

The film was the annealed at room temperature in vacuum for 4 hours.

4.2 NBR Purification

The following procedure describes the purification of NBR (40% ACN) analyzed in

Section 5.3.2. A solution of NBR 4 wt% in unstabilized THF was prepared. A beaker was filled with 700 mL of methanol and the NBR/unstabilized THF solution was slowly transferred drop by drop into the stirring solution of methanol using a Pastor pipette.

When the transfer was completed, the stirring was stopped and the solution was allowed to rest for 20 minutes. The precipitated NBR was taken from the methanol solution and placed in a vacuum oven overnight for drying.

4.3 High Pressure Liquid Chromatography Characterization

Solid pieces of NBR of (1/5 inches) 3 were extracted overnight into methanol solution to remove possible additives. HPLC separation was achieved using a Varian HPLC system

(USA), consisting of a Varian PROSTAR pump module, Rheodyne injector and a Varian

PROSTAR diode array detector and a C 18 column manufactured by Phenomenex. Data acquisition was accomplished using Galaxie software. The chromatographic eluent used to separate the low molecular weight compounds extracted in the methanol solution is summarized in Table 4.1 and 4.2:

66

Table 4.1. Solvent used during the HPLC separation 75% Water-25% Solvent A Acetonitrile Solvent B 100% Acetonitrile

Table 4.2. Solvent profile used to characterize the methanol extract in Section 5.3.2 Flow rate At Solvent composition (mL/min) 0 1.5 A=75% B=25% 2 11 A=75% B=25% 30 11 A=0% B=100% 38 11 A=0% B=100% 38.8 2 A=0% B=100% 43.5 2 A=0% B=100% 44 1.5 A=75% B=25% 46 1.5 A=75% B=25%

4.4 IR-Visible Sum Frequency Generation Spectroscopy (SFS) Measurement.

All the SFS spectra presented in the work were acquired using either one the two spectrometers described in the next two sections.

4.4.1 ExxonMobil Narrow-Line SFS Spectrometer

The SFS apparatus is based on a laser system that uses a visible pulse at (532 nm; 3mJ) overlapped with a tunable pulse (3-10 m;1-2 mJ); both have a duration of 7 ns and a repetition rate at 10 Hz. 67 The laser set-up is shown in Figure 4.1. The frequency of pulsed Nd:YAG laser (solid state laser) is doubled using a non-linear crystal. The output at 532 nm pumps the molecules of organic dye (LDS-698 or DCM) that absorbs energy of the source and re- radiates the energy as a laser radiation. A grating selects one wavelength (tunable laser dye) and sends it to the Raman cell or Multi-pass high pressure H 2 cell, for IR conversion using a third order stimulated Raman process. For most molecules, the Raman mode with the greater gain has a frequency resulting in a shift less than the tuning range of the dye laser except for molecules such as H 2 molecules with a vibrational mode corresponding to 4122 cm -1. The stimulated Raman emission is used to down shift the frequency of the input by an amount equal to the Raman mode of the H2 molecule; the input frequency for one process acts as the input of the next one. The IR pulses exiting the Raman cell are guided to the sample and are overlapped with the remaining fraction of the visible green.

The sapphire coated with sample is used in total internal reflection (TIR) geometry which is described in section 4.4.3. Three polarizers set the polarization to S or P orientation of the SFS, Green and infrared beams, respectively. The SFS signal intensity is amplified using a pre-amplifier and detected using a photon multiplier (model PS 325 from

Stanford Research System) after which the infrared intensity is recorded using an infrared meter (Rj-7200 Energy meter). This spectrometer has a narrow-resolution

(bandwitdth ≈ 0.2 cm -1) also labeled “narrowline.”

68

Figure 4.1. Schematic diagram of the ExxonMobil narrowline SFS spectrometer.88

Most of the SFS spectra obtained in this study were acquired by scanning the tunable laser dye in two regions (2000-2300 cm -1) and (2800-3200 cm -1) at intervals of 1 to 2 wavenumber(s). Each data point corresponds to an average of 20 to 50 SFS signal pulses.

For each spectrum, the background noise from the environment was measured while blocking the infrared beam and this value was subtracted from the SFS signal intensity.

Finally to account for the intensity and shape of the IR source, the SFS spectra were normalized to the profile of the infrared intensity. A typical SFS spectrum corresponds to the SFS intensity plotted as a function of the IR wavenumber as shown in Figure 4.2.

69 Further analysis of the SFS spectrum was achieved by fitting the experimental data using

Igor Pro software fitting procedures.

1.0

0.8 (A) 0.6

0.4

0.2 SFG Int. SFGInt. (a.u.)

2850 2900 2950 3000 3050 3100 -1 I.R. wavenumber (cm ) Figure 4.2. Typical normalized SFS spectrum.

The solid lines are fit to the data using Equation 3.22. The detected resonances correspond to the CH and CH 2 groups at the surface of polyacrylonitrile at the sapphire/PAN interface. The neon atomic absorption lines are used for dye calibration by observing the opto-galvanic effect of a neon hollow cathode lamp as a function of the dye laser frequency. A small part of the dye laser beam is incident between the cathode and the anode. When the dye laser frequency is tuned, the impedance of the discharge increases or decreases when resonance occurs. The dye laser frequency controller read out is monitored and adjusted to match the observed the peaks of the observe neon atomic lines.89

4.4.2 The University of Akron Broadline SFS Spectrometer

The laser set up is shown in Figure 4.3. A pulsed (femtosecond) Ti-Sapphire red (800 nm) laser (Tsunami, Spectra-Physics, Inc) and a continuous-wave, solid-state green (532 70 nm) laser (Millenia V, Spectra-Physics, Inc.) drive a regenerative amplifier (Spitfire,

Spectra-Physics, Inc). The spitfire produces an output of picosecond-long pulses at a repetition rate of 1 kHz. A part of the output is directed into the optical parametric amplifier (OPA)-(OPA-800, Spectra-Physics, Inc) which produces the tunable IR beam.

The IR and visible beam are taken out of the OPA through different paths to be spatially and temporally overlapped at the sample. The overlap in space is achieved using a dichroic mirror to transmit only the infrared beam. As the red beam takes a shorter path through the OPA, the overlap in time is achieved using a delay stage.

Figure 4.3. Schematic diagram of the University of Akron SFS broadline spectrometer.90

71 The sum frequency spectroscopy phenomenon explained in the second section of Chapter

3 is produced as the sample is guided through filter to reduce noise and is detected and amplified using a photomultiplier tube and photon counter (SR400, Stanford research

Systems, Inc). At the same time, the IR intensity for each SFS measurement at the sample is recorded. These outputs are recorded simultaneously, the SFS intensity is normalized to account for the fluctuation in IR intensity while scanning.

This spectrometer has a lower resolution as the IR pulse energy as a function of wavenumber is described by a Gaussian function of standard deviation 8.7 cm -1, also labeled “broadline”.

4.4.3 SFS in Total Internal Reflection (TIR) Geometry

SFS has been used in TIR geometry on all experiments present in this work and is presented in Figure 4.4. The polymer coated sapphire is mounted in a cell immobilized on a rotating stage (not shown) allowing to tilt the normal of the sapphire interface with respect to the incident visible and infrared beams. The visible and infrared beams strike the sapphire prism (interface A) and with angle φ are refracted before reaching the A,i

φ interface B with an angle B,i where the sum frequency beam is generated without attenuation by the sapphire.

72 Medium (n 4) C

Polymer (n ) 3 B

Prism Sapphire (n 2) SFG ΦB,3 ΦB,2 D ΦB,1 A

n1(air)

Φ νSFG A,i νvis νIR PMT

Figure 4.4. Schematic diagram of the TIR geometry used in SFS experiments.

φ The incident visible and IR beams enter the prism at interface A with an angle, A,i ,with

φ respect to the normal at the interface and strikes interface B at an angle, B,i ,which is the angle for the total internal reflection to take place. The refractive indices of the media are as indicated.

The intensity of the SFS signal detected at the sapphire/polymer interface depends on the incident angle of the IR and visible beams and the refractive indices of the media forming the interface and can be computerized as shown in Figures 4.5 and 4.6.

73 SFS Intensity (a.u.)

Angle of Incidence ( φ ) A,i Figure 4.5. Angular dependence of SFS signal at the sapphire/NBR interface under heptane exposure in SSP polarization for a film of a thickness of 300 nm. Solid line corresponds to fit modeled by a program written by Li. 91

The refractive indices values used to simulate the SFS enhancement from the

φ sapphire/NBR interface as a function of A,i are that of the sapphire and NBR, n=1.7 and n=1.52,95 respectively. Similarly, the NBR/oil interface, interface C in Figure 4.4, can be probed by choosing the proper angle that satisfies the total internal geometry at the polymer/heptane interface. The angle dependence of the SFS signal at the C interface is presented on Figure 4.6. The refractive indices values used to simulate Figure 4.6 are that of the NBR and heptane and are n=1.52 and n=1.38, respectively.

74 SFS Intensity (a.u.)

φ Angle of Incidence ( A , i )

Figure 4.6. Angular dependence of SFS signal at the NBR/heptane interface in SSP polymerization for a film of a thickness of 300 nm. Solid line corresponds to fit modeled by a program written by Li 91 .

These angles were first estimated by modeling the SFS intensity enhancement from the interface in question (Figure 4.6) and were further optimized experimentally by determining the angle giving the maximum SFS signal. Sapphire/PAN, sapphire/PB and

φ ≈ sapphire/NBR were selectively probed by using an angle of A,i 1°, and the PAN/air,

φ ≈ PB/air and NBR/air interfaces were investigated using A,i 40°. The PAN/water,

PAN/heptane, NBR/water and NBR/heptane interfaces were characterized at an angle of

φ ≈ A,i 10° as the refractive indices of water and heptane are, respectively n=1.33 and

1.38.

To study the polymer/liquid interface a stainless steel flow cell was used. For reproducibility purposes the cell was disassembled and cleaned prior to each experiment

75 by rinsing with ethanol and heptane. It was then placed in oven at 250 °C for 15 minutes, cooled to room temperature, and plasma treated for one minute. The coated prism was clamped onto the cell and a Teflon spacer was used to ensure adequate sealing between the prism and the cell.

4.5 Differential Scanning Calorimetry (DSC) Measurements

DSC on a TA DSC Q 1000 was performed on polymers at heating and cooling rates of 10

°C/min.

4.6 FTIR Measurements

FTIR spectra were obtained using a Digital Excalibur Spectrometer (FTS 3000). The data are acquired with a resolution of 4 cm -1. Solution-casted films of the polymer sample on KBr disks were used to obtain absorbance spectra in the transmission mode.

4.7 Raman Spectroscopy Measurements

Raman spectra were obtained using a Renishaw System 1000 Raman with a

514.5 nm Ar ion laser. A calibration using crystalline silica allows determining a resolution for the Raman experiment at approximately 6 cm -1.

76 4.8 Nuclear magnetic Resonance (NMR) Measurements

NMR spectra were obtained using a Varian 500 MHz spectrometer. Solution of NBR in deuterochloroform were prepared with tetramethylsilane as the internal reference. NMR spectra were placed in Appendix D.

4.9 Size Exclusion Chromatographic (SEC) Measurements.

Size Exclusion Chromatographic (SEC) analyses were carried out using an Multi Angle

Laser Light Scattering instrument equipped with three Waters HR Styragel columns with

THF as eluent at a flow rate of 1 mL/min at 35°C; the detector system combined a Wyatt

Dawn Eos laser light scattering (MALLS) detector and differential refractometer concentration detector (Waters 410). SEC chromatograms were placed in Appendix E.

4.10 Thin Film Thickness Measurements.

Thickness measurements were performed with a J.A. Woollam spectroscopic ellipsometer. Ellipsometer data were acquired in reflection mode on sapphire coated prism at an angle of incidence of 45 °. Ellipsometry measures the change in polarization state of light reflected from the surface of a thin polymer film.92 The experimental data are expressed as psi (ψ) and delta (∆) such that:

R ρ = p = tan()ψ ei∆ (4.1) Rs

ρ ~ ~ where is the ratio of the Fresnel coefficients, Rp and Rs , for p and s polarized light.

77 The reflected surface parameter such as thickness is related to the measured data through a system of non–linear and non-invertible sets of equations. The Cauchy layer model was used to model the influence of the polymer film thickness on the measured (ψ) and (∆) and fit experimental data. The geometry of the spin-coated polymer film on sapphire substrate was modeled with a first layer, layer “0”, of infinite thickness corresponding to the sapphire substrate and layer”1” a function based Cauchy layer related to the polymer, that follows Equation (4.2):

B C n()λ = A + n + n (4.2) n λ2 λ4

For the thickness calculation, experimental data were fitted to refractive indices of n=1.52 for PAN, PB and NBR.93

78

CHAPTER V

RESULTS AND DISCUSSION

The ACN content of NBR rubber is the main parameter controlling the oil resistance of

NBR. This chapter contains the results of bulk and interfacial spectroscopic characteristics of PAN, PBD and NBR polymers. Spectroscopic shifts are of primary importance in extracting information about the polar nitrile interactions. The spectroscopic methods described in Chapter 3 were used to interrogate the polar associations in bulk NBR and at the NBR surface.

Within the NBR polymer, interactions can occur between two nitrile groups but also between the nitrile side groups and the olefinic segments of butadiene which together create a more complex system to study. To break down this complexity, the stronger of the two, the nitrile-nitrile interactions are set as the reference point. The best reference material to carry out this study would be a polymer where interactions between nitrile groups dominate. That material would be the homopolymer PAN. The spectral vibrational characteristics of PAN will be investigated in the CN (2000-2300 cm -1) and in the CH region (2800-3200 cm -1).

79 The results and discussion section is articulated in four sections. First, PAN was analyzed, and then secondly, the bulk and surface character of homopolymer PB were investigated. The analysis of these two homopolymers created the basis of the investigation presented in the third section: bulk and surface characterization of NBR.

In the fourth and final section, the nitrile rubber surface was exposed to aliphatic, aromatic, and polar solvents. Surface changes were analyzed using the set of SFS data showing the first spectroscopic manifestations of the surface of nitrile rubber in a solvent environment. The commercially available NBR rubber samples used in this research differed in ACN content from 40% ACN to 20% ACN.

5.1 Spectroscopic Investigation of Nitrile Interactions in PAN.

An extended view of the PAN physical and spectroscopic features was given in Chapter

2. A DSC of PAN, placed in Appendix A on Figure A-1, confirms the detection of two thermal transitions below 160 °C. A first transition detected at 100 °C referred to as the glass transition and a second one at 150 °C, attributed to the nitrile loosening of the dipole-dipole interaction. 26, 27 The focus narrows down to the vibrational characteristics of the nitrile side group of PAN in the infrared region.

5.1.1 SFS spectra and peaks assignments.

First, the details of the CN vibration in the bulk PAN will be characterized and compared with the literature data on the location, width and intensity of the CN band. Since PAN has never been analyzed using Sum Frequency Spectroscopy

80 (SFS), previously established infrared and Raman assignments will be used to attribute the peaks corresponding to the interfacial nitrile moiety. The point of entry of this investigation is to acquire a classical FTIR spectrum of PAN to determine the features, mainly the location, of the CN stretching vibrational band. Due to its highly polar nature, the CN stretching mode should give rise to an intense transition and the CN stretch mode should be readily detected. Step one is to analyze bulk PAN using an FTIR spectrometer.

0.210

0.205

0.200

0.195

0.190 Abs.(a.u) 0.185

0.180

2100 2150 2200 2250 2300 -1 I.R. wavenumber (cm )

Figure 5.1. FTIR spectrum of bulk PAN in the CN stretching region.

The CN stretching vibration mode is detected at 2242 cm -1. Earlier experiments reported a CN stretching mode between 2238 cm -1 94 to 2243 cm -1,95 a range in agreement with the

FTIR result of a absorption band of bulk PAN at 2242 cm -1. This value will be used as the spectroscopic reference.

The primary interest of this research project is to characterize what happens to the polymer molecules in contact with liquid. A real life example for this research would be

81 to know what happens to polymer chains on the surface of an O-ring rubber piece immersed in a solvent.

To know the characteristics of the interfacial PAN molecules, and more precisely, the interfacial nitrile group, we performed SFS on PAN spin-coated thin films. The result is presented in Figure 5.2 (B) where we probed the interface between the PAN film and the sapphire prism. To the infrared spectrum in Figure 5.2(A), we compare the SFS spectrum of the PAN/sapphire interface, shown in Figure 5.2(B), which covers the same region from 2100 to 2300 cm -1. It is clear from that SFS spectrum that the CN stretch of PAN is not located at 2242 cm -1 as expected but at 2254 cm -1.

The CN resonance frequency of PAN at the sapphire/PAN interface is positioned at 2254 cm -1 which is 12 cm -1 blue-shifted with respect to the CN stretching mode of the bulk

PAN. The bulk spectroscopic data at 2242 cm -1 falls in the range reported by various works 18, 95 . However the difference of 12 wavenumbers between the vibrational frequency of PAN at the sapphire/PAN interface cannot be explained by experimental errors as the resolutions of the FTIR spectrometer and the SFS spectrometer used are below 4 and 1 cm -1 respectively. Therefore, the shift existing between the nitrile stretching mode of PAN bulk and sapphire/PAN molecule represents a potentially significant shift which requires investigation because for an unknown reason the nitrile groups in the bulk of PAN do not vibrate at the same frequency as those at the sapphire/PAN interface. The significance of CN frequency shift has been extensively investigated and explained as the following. The shifting of the CN stretching vibrational

82 band has been observed with acetonitrile 79, 80 and more recently PAN.18, 83 Positive and negative shifts are possible for both acetonitrile and polyacrylonitrile and it is a diagnosis of the interactions of nitrile with surrounding molecules.

0.210

0.200 (A) 0.190

0.180 Abs. (a.u) 10 8 6 4 (B) 2

20 15 10 (C) 5 SFG Signal Int.(a.u.) SFGSignal

2100 2150 2200 2250 2300 -1 I.R. wavenumber (cm )

Figure 5.2. (A) FTIR spectrum of PAN bulk, (B) SFS spectrum of PAN in PPP polarization at the sapphire/PAN interface after PAN annealing at 90 °C for 8 h followed by a second heating at 120 °C for 4 h and (C) SFS spectrum of PAN in PPP polarization at the sapphire/PAN interface after PAN annealing at 90 ºC for 8 h only, in the CN region.

The CN side groups have an intriguing capability to interact via either the lone pair orbital of the nitrogen “end-on” leading to an hypsochromic (positive or blue) shift of the

83 CN stretching vibrational frequency or to interact via a “side-on” coordination leading to a bathochromic (negative or red) shift of the CN stretching vibrational frequency.16

Table 5.1: Positions of the CN stretching vibration for PAN from Figure 5.2 obtained by fitting using Equation (3.22).

PAN/sapphire PAN/sapphire interface interface PAN (bulk) a PAN annealed at PAN annealed at 90 °C and 120 °C b 90 ° C only b

Appears in 5.2(A) 5.2 (B) 5.2 (C) Peak position 2242 2254 2236 (cm -1) a The error limit for peak location is ± 2 cm -1. b The error limit for peak location is ± 0.5 cm -1.

Based on previous studies, it is reasonable to believe that the observed frequency shift of

CN at the sapphire/PAN interface is due to an interaction between the sapphire substrate and the polymer via its CN groups. Moreover, Andreeva and Burkova 83 observed a shift of the nitrile groups stretching mode of PAN as the temperature was raised from 0 to 90

°C for thin PAN films. They eventually connected the shift to the density of the polymer and more precisely the distance and relative orientation of the dipole-dipole interaction of nitrile groups of PAN.

Therefore, the present surface study of PAN investigation started with the observation of this experimental shift. It raises two questions: first, why is there a positive 12 cm -1 shift of the CN stretch of PAN; and second, why such a shift is observed especially at the interface? What is so special about the environment that these nitrile groups experience

84 against the sapphire surface? Moreover, is there an experimental parameter controlling this shift?

The first parameter questioned was the temperature at which the polymer was annealed as mentioned before by Andreeva and Burkova.83 Within the framework of this study, the thermal history of the PAN thin film includes a heating to 90 ºC in a vacuum oven for 8 hours followed by cooling to room temperature followed by a second heating to 120 ºC for 4 hours. The resultant vibration peak position of CN stretch at the sapphire interface is

2254 cm -1 as shown on Figure 5.2 (B). To further investigate the nature of the sapphire-

PAN interaction, the experiment was repeated with a sample annealed for 8 hours at 90

ºC only.

As depicted in Figure 5.2(B) and 5.2(C) when the PAN film was annealed at 90 °C for 8 hours only, the SFG spectrum exhibits a CN resonance feature at 2236 cm -1 with no positive shift with respect to the same resonance feature in bulk PAN. This result could indicate that the interaction between nitrile groups and sapphire is activated by temperature, and more specifically, by heating the PAN polymer above its T g (100 ºC).

To explain this annealing effect, the chemical structure of the sapphire surface can be decomposed into two types of sites or environments that can interact with the polymer: first, it can react with residual hydroxyl groups or H bonding sites; and second, it could coordinate with unsaturated aluminum cation sites.96 To understand the spatial origin of the interaction between PAN and the sapphire surface, the surface OH groups were markedly reduced by heating the sapphire at 750 ºC for 36 hours. The effectiveness of the 85 heat treatment in reducing the hydroxyl number density at the sapphire surface was previously demonstrated using SFG spectroscopy.97 Following this specific sapphire heat treatment, the sapphire surface was coated with PAN using the same spin coating conditions.

The SFS spectrum in the CN region of the PAN/heated sapphire interface is displayed in

Figure 5.3(C). For the PAN/heated sapphire interface, the fitting reveals a CN resonance feature of PAN at 2237 cm -1 which is significantly lower than the CN stretching vibration for PAN at the PAN/sapphire interface at 2254 cm -1. The positions of the nitrile CN stretch of PAN presented in Figure 5.3 are gathered in Table 5.2.

0.210

0.200 (A) 0.190

Abs. (a.u) 0.18010 8 6 4 (B) 2 6

4

2 (C) SFG Signal Int.(a.u.) SFG Signal 2100 2150 2200 2250 2300 -1 I.R. wavenumber (cm ) Figure 5.3. (A) FTIR spectrum of PAN bulk, (B) SFS spectra of PAN in PPP polarization at the sapphire/PAN interface after PAN annealing at 90 °C for 8 h followed by a second heating at 120°C for 4 h and (C) SFS spectra of PAN in PPP polarization at the sapphire/PAN interface after PAN annealing at 90 ºC for 8 h followed by an additional heating at 120°C for 4 h with spin-coating on sapphire (temperature treated at 750 °C for 36 h), in the CN stretching region.

86 Table 5.2: Positions of the CN stretching vibration for PAN from Figure 5.3 obtained by fitting using Equation 3.22.

PAN/sapphire PAN/sapphire interface 90 °C interface 90 °C and PAN (bulk) a and 120 °C 120 °C non heated heated sapphire b sapphire b Appears in 5.3 (A) 5.3(B) 5.3(C)

Peak position 2242 2254 2237 (cm -1) a The error limit for peak location is ± 2 cm -1. b The error limit for peak location is ± 0.5 cm -1.

In addition to a marked change in the resonance frequency of the CN vibrational mode upon heat treatment of the substrate, a notable change in vibration mode of the CH and

CH 2 groups was also observed. Figure 5.4 (A) shows the SFS spectrum of the PAN/ heated sapphire interface in the 2800-3000 cm -1 range. Three resonant features at 2870,

2940 and 2950 cm -1 were observed for the PAN/heated sapphire interface in contrast to the SFS spectrum of the PAN/ sapphire interface, shown in Figure 5.4 (B) where no resonance structure was detectable. The resonance structures at 2870 and 2950 cm -1 were assigned to the methylene stretching modes and the 2930 cm -1 resonance was due a stretching mode of the CH group.94

To investigate the significance of these spectroscopic changes, first the data analysis was focused on the absence of a 12 cm -1 positive shift of the CN stretching vibration when the sapphire was heated.

87 0.5 1.0 0.4 0.8 (A) 0.3 (B) 0.6 0.2 0.4 0.1

0.2 SFG Int. (a.u.) SFG Int. (a.u.) 0.0

2800 2850 2900 2950 3000 3050 3100 2850 2900 2950 3000 3050 3100 -1 -1 I.R. wavenumber (cm ) I.R. Wavenumber (cm )

Figure 5.4. (A) SFS spectrum of PAN in PPP polarization at the sapphire/PAN interface after PAN annealing at 90 °C for 8 h and 120 °C for 4 h with spin-coating on sapphire (temperature-treated at 750 °C for 36 h) and (B) SFS spectrum of PAN in PPP polarization after PAN annealing at 90 °C for 8 h and 120 °C for 4 h at sapphire/PAN interface, in the CH stretching region.

This demonstrates that when hydroxyl species are depleted from the sapphire surface, the resonance feature of the interfacial CN stretching vibrational band at the sapphire interface is very close to the vibrational frequency of PAN bulk. In other words, the main interaction between the sapphire surface and the polymer is due to OH-CN interaction and can be significantly reduced by decreasing the number of the surface hydroxyl groups. Heating PAN chains above 100 ºC (T g) would allow conformational rearrangement: CN side groups along the chains reorient in a specific direction to interact with the surface sapphire OH. Whether the disappearance of CH 2 peaks and CH in spectrum 5.4 (B) is due to a simple reorientation or a true chemical reaction is at this point unclear. The direct correlation between the presence of the OH species with the shift of the CN stretching vibrational features strongly suggests the existence of an interaction between the OH and the CN groups of PAN at the interface. It is interesting to note that this effect occurs at a temperature below the temperature assigned to the

88 loosening of the nitrile-nitrile dipole interaction in PAN bulk which tends to show that either the thermal transition associated to the nitrile dissociation is lower at the surface compared to that of the bulk (150 °C) or that the interactions between sapphire OH and nitrile groups favor the reorientation of interfacial PAN chains at 120 °C.

From the first section it is shown that PAN interacts via its nitrile groups with surrounding molecules at high temperature, this interaction weakens the nitrile dipole- dipole coupling and is indicated by a significant frequency shift of the CN stretching vibration. This experimental finding correlates with the results reported by Goodman and

Suwyn 98 where the authors observed interaction of PAN with water at 140 ºC under pressure. They further noted a corresponding negative shift of the Raman CN stretching vibration. The interaction between PAN fibers at high temperature correlates with the interaction we observed for a PAN film on a sapphire prism above 120 °C.

An unexpected shift of the CN stretch at the sapphire interface confirms the high temperature sensitivity of CN stretch and reveals the importance and specificity of the interface as a role in understanding macroscopic properties. Do those results hold for the nitrile on the opposite interface of the film where the nitrile side groups protrude at the air interface?

The angle of incidence of the visible and IR beam was changed by rotating the sapphire prism to probe the PAN/air interface; the spectrum is displayed in Figure 5.5. A double peak was fitted to the experimental SFS data. Up to this point only one resonance feature 89 was observed in the 2100-2300 cm -1 range but observing multiple peaks is a common experimental fact and is attributed to the interacting capability of the CN moiety with its environment.

20

15

10

5

0 SFG(a.u.) Intensity

2100 2150 2200 2250 2300 -1 I.R. Wavenumber (cm ) Figure 5.5. SFS spectra of PAN in PPP polarization at the PAN/air interface in the CN stretching region.

Fitting reveals two resonant features at 2238 cm -1 (positive) and 2242 cm -1 (negative).

Despite the fact that there is only one CN vibrational stretching mode, multiple peaks are commonly observed in the 2100-2245 cm -1 range for CN bearing compounds.18, 80 Each peak corresponds to a specific CN state. This tends to indicate that two different types of nitrile groups populate the PAN surface.

5.1.2 Polyacrylonitrile under Solvent Exposure

The PAN bulk, the PAN/air and sapphire/PAN interfaces were characterized using FTIR and SFS analysis, respectively. Unlike NBR, PAN is resistant to a wide range of solvents such as water, aliphatic and aromatics . The next step of this research was to

90 expose the PAN surface to a variety of specific solvents. Heptane and water, two solvents with two different polarities, were used. Heptane is an aliphatic hydrocarbon, non-polar and non-hydrogen-bonding solvent used to simulate the oil that comes into contact with rubber O-rings. Water is a highly polar material with hydrogen bonding properties, and from documented works, water has an important influence on polar association of nitrile bearing compounds. Water is the first liquid that was used in this study.

The interaction between PAN and water is known to be very strong at high temperatures causing the PAN to dissolve. Although water cannot dissolve PAN at room temperature, the existence of the PAN/water interaction at the polymer/water interface is an important issue that will be discussed below. The SFS spectra of the PAN/water interface are displayed in Figures 5.6 (A) and (B).

20 20

15 15

10 10 (A) (B) 5 5 SFG Intensity (a.u.)

SFG Intensity (a.u.) 0 2000 2020 2040 2060 2080 2100 2100 2150 2200 2250 2300 -1 -1 I.R. Wavenumber (cm ) I.R. Wavenumber (cm )

Figure 5.6. SFS spectra of PAN in PPP polarization at (A) PAN/water interface in the CN stretching region (2000-2100 cm -1) and (B) at PAN/water interface in the CN stretching region (2100-2300 cm -1). SFS spectra were obtained with the University of Akron Broadline SFS spectrometer.

Upon water injection, the binodal surface population of CN becomes uniform [spectrum on Figure 5.6 (B)] and only one CN vibrational band is observed at 2237 cm -1 which corresponds to a red shift of 1 and 5 cm -1, respectively, for the two original bands at 2238

91 and 2242 cm -1. By changing from two peaks to one peak, the data now shows that CN groups experience only one unique environment which could be reasonably in the outermost layer of the PAN film in contact with water molecules. Since Henrici-Olive and Olive 18 , commented on the observation made by Goodman 98 about an extraordinary

200 cm -1 negative shift in the vibrational CN stretch upon interaction between PAN and water molecules, it seemed necessary to probe that range for the PAN/water interface and that spectrum is shown on Figure 5.6 (A). It is clear that at the interface between PAN and water molecules no such feature is detected. Most likely, the nitrile-nitrile interactions are too strong at room temperature to allow any interaction with water molecules. A comparative spectrum of the PAN/air interface in the exact same range is included in Figures 5.7 (A) and (B).

80 80 70 70 60 60 50 50 (B) 40 (A) 40 30 30 20 20 10

SFG Intensity (a.u.) 2200 2220 2240 2260 2280 2300 2000 2020 2040 2060 2080 2100 (a.u.) Intensity SFG -1 -1 I.R. Wavenumber (cm ) I.R. Wavenumber (cm )

Figure 5.7. SFS spectra of PAN in PPP polarization at (A) PAN /air interface in the CN stretching region (2000-2100 cm -1) and (B) at the PAN/air interface in the CN stretching region (2100-2300 cm -1). SFS spectra were obtained with the University of Akron Broadline SFS spectrometer.

The purpose of this investigation was to study the PAN/heptane interface and this is treated in the following section. Interestingly, upon heptane injection a similar spectrum

92 is collected at the PAN/heptane interface (Figure 5.8) where only one vibrational band is detected at 2238 cm -1.

50 40 30 20 (A) 10

15 10 (B) 5

SFG(a.u.) Intensity 2100 2150 2200 2250 2300 -1 I.R. Wavenumber (cm )

Figure 5.8. SFS spectra of PAN in PPP polarization (A) at the PAN/heptane interface and (B) at the PAN/water in the CN stretching region.

The similarity of PAN/water and PAN/heptane interfaces could demonstrate that a small amount of water is present in the heptane, in a sufficient amount to alter the compositional environment of the PAN/heptane interface. If we further compare both interfaces with PAN bulk it appears that PAN surface properties are not significantly changed upon water or heptane exposure. From a bulk point of view, these results certainly correlate the solvent resistance of PAN toward solvent in general; from a surface point of view, the data directly indicate that even at the surface of PAN, the nitrile dipole interactions are stronger than nitrile-heptane or nitrile-water interactions.

93 5.1.3 Summary on PAN Investigation.

In conclusion, the CN groups of PAN are able to interact with the surface hydroxyl groups of a sapphire substrate at high temperatures producing a blue shift with respect to bulk PAN in the CN vibrational mode. The interaction between PAN and the sapphire surface can be markedly decreased when the surface hydroxyl groups of sapphire are reduced. This chemical modification of the buried interface affects the molecular arrangement at the PAN film surface. It was also demonstrated that PAN does not strongly interact with heptane or water at room temperature.

5.2. Analysis of PBD.

The analysis of PBD is as important as the analysis of PAN. If the acrylonitrile unit is the chemical unit bringing the oil resistance property to the NBR chain, the butadiene monomer is the unit sensitive to the oil. If the penetration of the oil is due to a specific interaction between the butadiene comonomer and aliphatic hydrocarbon, then there should be spectroscopic changes in the vibration mode of PBD CH stretch.

5.2.1 Bulk and Surface Analysis of PBD

It is well known that emulsion homopolymer PBD have three structures 1,4-trans , 1,4-cis and 1,2-vinyl. The relative amount of cis, trans, and vinyl varies according to the method of preparation and the temperature of polymerization.

94 Some regions of the infrared spectrum, especially the bending region, led to straightforward determination of the relative amount of cis, trans, and vinyl. Morero 99 developed a method that accurately quantifies the level and type of double bonds using specific bands in the 700-900 cm -1. For example, 724 cm -1 was attributed to 1,4-trans microstructure, 911 cm -1 was ascribed to 1,2-vinyl microstructure and 967 cm -1 arose from 1,4-cis microstructure. Comparing the relative intensity of these three peaks allows the determination of the relative content of the different types of double bonds in a specific PB polymer. PBD is, for that reason, a sophisticated case to analyze by spectroscopy. The bending region is unfortunately inaccessible to SFS experiments.

Therefore the analysis concentrated on the CH stretch region.

First, the bulk vibrational features of PBD were analyzed, and the Raman spectrum of

PBD is presented in Figure 5.9. Four resonance features contribute to the Raman spectrum of bulk PBD at 2840 cm -1, 2900 cm -1 and 3000 cm -1 and 3060 cm -1. These bands were compared to the PBD IR data obtained from Binder 100 . Binder, attributed

-1 2840, 2900, 3000, and 3060 cm respectively to the symmetric CH stretches of CH 2, asymmetric CH stretch of CH 2, CH stretch of cis -CH=CH and CH stretch of CH 2=CH vinyl side group.

95 5000

4000

3000

2000

1000

0 Raman Count. (a.u.) Count. Raman 2800 2900 3000 3100 3200 -1 Wavenumber (cm )

Figure 5.9: Raman spectrum of PBD and corresponding fit.

These assignments were close to what was observed in the SFS spectra and will therefore be the same assignment of peaks in this research, (according to Binder’s assignments).

The SFS spectrum at the sapphire/PBD interface is displayed in Figure 5.10 (B) and compared to the SFS spectrum of the PBD/air interface on trace C and to the Raman spectrum of bulk PBD on trace A.

96 (A) Abs. Abs. (a.u.)

1.2

0.8 (B) 0.4

1.6 1.2 0.8 SFG SFG (a.u.) Int. 0.4 (C)

2800 2900 3000 -1 3100 3200 Wavenumber (cm )

Figure 5.10. (A) Raman spectrum of PBD, (B) SFS spectrum of PBD in PPP polarization at the sapphire/PBD interface and (C) SFS spectrum of PBD in PPP polarization at the PBD/air interface in the CH stretching region. The number of data points was reduced for clear labeling of the spectra.

It is clear from Figure 5.10 that spectra (A) and (B) do not overlap perfectly; comparison

-1 -1 reveals a positive shift of the 2840 cm band arising from CH stretch of CH 2 to 2860 cm at the sapphire interface compared to that of the bulk PBD [Figure 5.10 (A)].The location and assignment for the methylene and C-H groups of PBD bulk and surface in the CH region are gathered in Table 5.3.

97 Table 5.3. PBD vibrational bands observed in Figure 5.10 compare to PBD vibrational bands observed by Binder 100 . Polybutadiene (PBD) bands in the C-H region Fig. Fig. Fig. 5.10 Binder 101 Assignment 5.10(A) a 5.10(B) b (C) b PB bulk sapphire/PB PB/air Bulk

3070 3077 CH (CH 2=CH) stretching 3010 3016 3007 3012 CH (cis-CH=CH) stretching 2910 2910 2907 2900 Asym methylene stretching 2843 2863 2849 2840 Sym. methylene stretching a The error limit for peak location is ± 3 cm -1. b The error limit for peak location is ± 0.5 cm -1.

This shift of the methylene stretch of PBD reveals the effect of a polar environment, the sapphire, on the vibrational properties of PBD. This dependence of the vibrational band of BD units on the polarity of the environment is an interesting correlation that is worth being checked later in the case of the nitrile rubber. The correlation will be slightly different and involves the influence of the polar acrylonitrile groups on the vibrational properties of butadiene units. This will be presented in the next section; for now, the attention remains on PBD spectral features displayed on from Figure 5.10. Apart from the positive shift of the methylene stretching mode of PBD at the sapphire interface, all the other features of PBD at the sapphire interface appear to closely match those of the bulk in term of their positions. The last spectrum, on Figure 5.10 (C), is the SFS spectrum of PBD at the interface between the PBD and air. The location of the peaks of PBD at the air surface matches those of the bulk PBD. Note that PBD bulk and PBD/air interface are both weakly polar media, which tend to demonstrate that the polarity of surrounding environment does affect the vibrational features of PBD. Finally, the peak at 3060 cm -1 present in the bulk was not found in either PBD/air or sapphire/PBD interfacial spectra.

98 5.2.2 Summary on PBD Investigation.

As a conclusion, the analysis of PBD revealed a very interesting modification when PBD was exposed to a polar environment at room temperature. This shift was a rare occurrence, at least for hydrocarbon molecules. This should be kept in mind when analyzing a copolymer that combines a polar unit, acrylonitrile, and a weakly polar unit, butadiene.

5.3 Analysis of NBR

Three commercial grades of NBR were provided with a content of acrylonitrile ranging from 40 to 20% ACN. NBR respectively containing 40%, 30% and 20% of ACN are industrially known as “high,” “medium” and “low” acrylonitrile content nitrile rubber, respectively.

5.3.1 Bulk Analysis of NBR

The bulk and surface spectroscopic features are presented in the first part of this section.

The bulk spectroscopic features of the NBR are also discussed for ACN ranging from

20% to 40% of acrylonitrile units.

The effect of the ACN on the nitrile stretching band is now investigated. Figure 5.11 shows the FTIR spectra of three copolymers containing 20%, 30% and 40% ACN in the nitrile stretching region 2220 to 2280 cm -1.

99 2237 cm -1 (A)

(B)

I.R.Abs.(a.u.) (C)

2200 2220 2240 2260 2280 -1 I.R. Wavenumber (cm ) Figure 5.11. FTIR spectra of bulk (A) NBR (20% ACN), (B) NBR (30% ACN), (C) NBR (40% ACN) in the CN stretching region.

To identify the peak positions, the FTIR spectra were fitted to a Gaussian function and the results are tabulated in Table 5.4. The results indicate that the nitrile CN stretching band is at 2237 cm -1 for the three copolymers, demonstrating that the nitrile CN stretching mode is independent of the ACN content. In other words, the vibrational frequencies of the CN triple bond are independent of the increasing polarity of the system.

In contrast with the nitrile band, the methylene resonance feature showed sensitivity to the amount of the ACN in nitrile rubber. Figure 5.11 shows the FTIR spectra of NBR containing 20%, 30%, and 40% ACN. The spectra were acquired in the region of 2800 to

3200 cm -1. It is observed that the location of the methylene band varies with acrylonitrile content. Higher ACN content causes a shift to higher frequency. This shift in frequency is related to the increase of the polarity of the medium. We should note that Furukawa et al.101 detected the symmetric stretching of methylene at 2850 cm -1 and attributed it to the 100 butadiene and acrylonitrile units of NBR. The locations and assignments of the observed peaks in Figure 5.12 are gathered in Table 5.4.

2920 cm -1 2844 cm -1 (A)

2923 cm -1 2848 cm -1 (B)

2928 cm -1 I.R. Abs.(a.u.) 2852 cm -1 (C)

2800 2900 3000 -13100 3200 I.R. Wavenumber (cm )

Figure 5.12. FTIR spectra of bulk (A) NBR (20% ACN) (B) bulk NBR (30% ACN) and (C) NBR (40% ACN) in the CH region.

As discussed in the previous sections, the methylene vibrational frequency at the

PBD surface was positively shifted with respect to the bulk. This shift was attributed to the interaction of the polymer with the polar sapphire surface. A comparative analysis can be made in the case of NBR. That is, the polarity of the bulk increases with the increasing

ACN content, causing the same positive shift as previously observed at surface of the

PBD. It was emphasized that the primary effect on PBD is a surface phenomenon because only the PBD chains at the sapphire surface exhibit the frequency shift in the

101 methylene symmetric stretch. However, in the NBR bulk, it is the content of ACN units in NBR that affects the methylene symmetric stretch.

Table 5.4. Observed frequencies (in cm -1) and assignments for Furukawa et al 101 alternating copolymers and our NBR. (ACN) and (BD) respectively designate the acrylonitrile and butadiene units. Copolyacrylonitrile (ACN)-co- butadiene (BD) C-H 40% 30% 20% Infrared Assignments 102 region ACN a ACNa ACN a 2852 2848 2844 Methylene sym stretching (CN) and (BD) 2928 2923 2920 Methylene antis. and CH stretching (CN) 2991 2990 2991 (2980,combination) BD 3009 3027 3028 3029 CH (BD) stretching

3077 3076 3074 CH (CH 2=CH) stretching C≡N 2237 2237 2237 Nitrile stretching (CN) region a The error limit for peak location is ± 2 cm -1.

On one hand, from the previous section, NBR bulk analysis showed that the nitrile stretch was independent of ACN of NBR. This is in contrast with the interfacial nitrile stretch of

PAN that was sensitive to the polarity of the environment on the vibration of the nitrile groups (to OH groups – after heating). The fact that the nitrile stretch of NBR is insensitive to the polarity of the bulk does not preclude a possible dependence of the nitrile stretch of NBR at the NBR interfaces.

The hypothesis states that the oil resistance is driven by the capability of NBR to interact with the environment. For example, NBR molecules are expected to interact with heptane molecules differently than with toluene molecules. Analyzing the nitrile stretch at the

102 sapphire and air interfaces is the next logical step of the investigation. The goal of this experiment was, first, to obtain a direct comparison between the nitrile spectroscopic feature of NBR bulk and that of the surface to reveal any spectroscopic differences related to the environmental impact on the interfacial nitrile. Second, a set of SFS spectra was gathered at the air and sapphire interface which became the reliable starting point for later analysis on polymer/liquid interface.

5.3.2 SFS Investigation of NBR in the CN Region

The attention is now concentrated on the spectroscopic features of NBR at the sapphire interface. The SFS spectra of NBR (40% ACN) was acquired at the sapphire/NBR and

NBR/air interface and compared to the FTIR bulk of NBR (40% ACN). Figure 5.13 trace

(A) shows the FTIR spectrum of NBR in the (2000-2300 cm -1 range). This spectrum was described earlier in Section 5.3.1 which showed a vibrational band detected at 2237 cm -1 and was assigned to the nitrile stretch of NBR (40% ACN).

Figure 5.13 also compares the SFS spectrum of NBR (40% ACN) at the sapphire interface on trace(B). The first feature is recognized at 2233 cm -1 and is assigned to the nitrile stretch of NBR. This feature red-shifted 4 cm -1 compared to that of the bulk. For the same reasons advanced in the section concerning PAN, a 4 cm -1 shift between the CN stretching band at the sapphire/NBR interface and that of the NBR bulk can be attributed to the instrumental differences.

103 (A)

Abs. (a.u.) Abs. 1.6

1.2

0.8 (B) 0.4

0.0 60

40 20 (C) 0

SFG Intensity (a.u.) SFG Intensity 2000 2050 2100 2150 2200 2250 2300cmÐ1

-1 I.R. wavenumber (cm ) Figure 5.13. (A) FTIR spectrum of NBR (40% ACN), (B) SFS spectrum of NBR (40% ACN) in PPP polarization at the sapphire/NBR (40% ACN) interface and (C) SFS spectrum of NBR (40% ACN) in PPP polarization at the NBR (40% ACN)/air interface. The data points number was reduced for clear labeling of the spectra.

It is clear from trace (B) that in addition to the nitrile stretch at 2233 cm -1 there are two peaks of weaker intensity detected in the lower frequency region (2000-2100 cm -1). For clarity, the spectroscopic features of these bands are summarized in Table 5.5.

Table 5.5. Positions of the CN stretching vibration for NBR (40%ACN) from Figure 5.13 obtained by fitting using Equation(3.22). Region Nitrile stretching position (cm -1) NBR(40% probed ACN) Bulk 2237 Sapphire/NBR 2039 2054 2233 NBR/air 2060 a The error limit for peak location is ± 0.5 cm -1. The SFS results concerning the sapphire/NBR interface are therefore partially understood to be one band at 2233 cm -1 and assigned to the nitrile stretch of NBR. The two peaks

104 observed in the low frequency region of the spectrum are attributed to a specific vibrational mode.

The SFS analysis of NBR (40% ACN) carried out on the NBR/air interface remains to be probed. Figure 5.13 shows the SFS spectrum, trace (C), for the NBR (40% ACN)/air interface. The first major observation in the SFS spectrum of the NBR/air interface is the absence of any feature in the (2200-2300 cm -1) region. If the nitrile stretch of NBR is detected at the sapphire interface, it is nowhere to be found at the NBR/air interface as no

SFS signal is detected between 2200 and 2300 cm -1. The second major difference between the spectral features of sapphire/NBR and NBR/air interface is an intense band around 2050 cm -1. Figure 5.13, therefore, shows that in the nitrile region, the spectral profile of NBR bulk differs from that of the sapphire/NBR interface; this also distinguishes it from that of the NBR/air interface. The two SFS spectra are, however, related as the additional unassigned vibrational bands are observed in both cases in the low frequency side of the (2000-2300 cm -1) region.

Two peaks are observable in the (2000-2100 cm -1) region at the sapphire/NBR interface compared to one at the NBR/air interface. But also the intensity of the 2050 cm -1 peak at the NBR/air interface is higher than that of features present in the (2000-2100 cm -1) region at the sapphire/NBR interface. A direct comparison of the number of peaks is not meaningful due to the difference in width between the features at both interfaces. For example, if the SFS spectrum of the NBR /air interface had only one narrow feature in

105 the (2000-2100 cm -1) region, a comparison of the peak number would be meaningful. The difference in intensity between the features at the NBR/air interface and sapphire/NBR interface is more interesting and shows that a specific molecule may be present at the

NBR/air interface. For the assignment of the 2050 cm -1 band, two possibilities have to be considered. The first hypothesis stipulates that this band is due to a molecule interacting with the nitrile CN side groups of NBR. There are a several assignments reported for the nitrile bands at 2050 cm -1; one of them was presented by Goodman and Suwyn.98 More precisely, it concerns a nitrile stretching mode of PAN interacting with water at high pressure and high temperature. Another possibility is an interaction of the nitrile moiety with a metal complex, such as Ni(O)CO(dialkylcyanamid),18 which led to an

∆ν = − -1 experimental bathochromic shift CN 200 cm . The second hypothesis states that the 2050 cm -1 is due to the presence of a molecule that has a vibrational mode of its own at this location. There are dosen of vibrational modes known to have an assignment in this region. Low molecular weight compounds such as , , ions, and interestingly, water molecules 64 have a vibrational band in the 2100 cm -1 region.

Also, it is possible that the peak between 2030 and 2050 cm-1 could be due to specific molecules added to the NBR during normal polymerization. are systematically added by the rubber companies to prevent the rubber from degradation via oxidation. These are the first molecules on the list of polymerization ingredients to account for.

The necessity of finding an assignment for these bands provided the motivation to carry out a purification of nitrile rubber. The hypothesis chosen to be tested first is the presence 106 of additives in the rubber contributing to an intense vibrational band in the 2050 cm -1 region. The intention was to compare an SFS spectrum of both sapphire and air interface of a purified and non-purified NBR. If the SFS spectrum of the purified NBR shows, at the NBR/air interface, a disappearance of the 2050 cm -1 band and an appearance of the

2233 cm -1, a conclusive assignment of the 2050 cm -1 will be achieved. Here, SFS alone cannot provide an assignment. It is necessary to involve a complementary characterization analysis that will positively identify additives, if any. HPLC analysis of the purification extract was selected as the complementary analysis. The conclusion will be based on coupling the SFS spectrum with an identification of the additive via

UV/visible detection.

The purification of NBR employed the overnight extraction method using methanol as the solvent that removed the low molecular weight components in NBR. An HPLC analysis along with UV/visible analysis of the extract separated and detected the molecules present in the methanol solution. The UV visible spectrum of the extract revealed a close match between the molecule present in NBR and an

(Chromatogram and UV-Visible spectrum were placed in appendix C).

The SFS spectra of non-purified and purified NBR are presented in Figure 5.14.

107 0.5 0.4 0.3 0.2 0.1 (A) 0.0 1.2

SFG SFG Int.(a.u.) 0.8

0.4 (B)

0.0

2000 2050 2100 2150 2200 2250 2300 -1 I.R. Wavenumber (cm ) Figure 5.14. SFS spectra of NBR (40% ACN) in PPP polarization (A) at the NBR (40% ACN)/air interface and (B) at the purified NBR (40% ACN)/air interface, in the CN stretching region.

A direct comparison of purified NBR/air and NBR/air interfacial spectra from 2000 to

2340 cm -1 refutes the hypothesis of an amine being the contributor at the 2050 cm -1 features. Furthermore, the absence of detection of a peak at 2233 cm -1 at the purified/NBR interface and the persistence of the 2050 cm -1 confirms this fact.

Table 5.6. Positions of the CN stretching vibration for NBR (40% ACN) and purified NBR (40% ACN) at the NBR (40% ACN)/air interface.

Region probed Nitrile stretching position (cm -1) NBR (40% ACN)/air 2061 NBR (40% ACN) 2077 purified/air a The error limit for peak position is ± 0.5 cm -1.

According to the fitting results, a shift exists between these two bands located in the

(2000-2100 cm -1) region. The reason behind this is uncertain and at this stage of the study

108 the presence of the 2050 cm -1 band in the purified sample was the primary source of interest.

From this experiment, it was determined that the 2050 cm -1 band was not that of an amine. The possibility of an additive molecule, in this case an amine, contributing to the

2050 cm -1 band in the NBR spectrum, was ruled out. The purification method using methanol will typically wash out aromatics and polar molecules such as phenols and amines, but there is still a possibility for specific molecules to remain trapped in the NBR matrix after extraction.

To further characterize the nature of this band, attention was given to an NBR’s different acrylonitrile nitrile content. The next experiment was designed to identify an experimental parameter that can modify the spectroscopic features in the (2000-2100 cm -

1) wavenumber region. Figure 5.15 shows the SFS spectrum of NBR (20% ACN) at the sapphire and air interfaces, along with that of the bulk.

109 0.40 0.36 0.32 (A) 0.28 Abs.(a.u) 0.24 1.2

0.8

0.4 (B)

0.0 25 20 SFG Int. SFGInt. (a.u.) 15 10 (C) 5 0

2000 2050 2100 2150 2200 2250 2300 -1 I.R. Wavenumber (cm ) Figure 5.15. (A) FTIR spectrum of NBR (20% ACN), (B) SFS spectrum of NBR (20% ACN) in PPP polarization at the sapphire/NBR (20% ACN) interface and (C) SFS spectrum of NBR (20% ACN) in PPP polarization at the NBR (20% ACN)/air interface in the CN stretching region. The data points number was reduced for a clear labeling of the spectra.

Trace (A) corresponds to the FTIR spectrum of NBR (20% ACN). Trace (B) relates to the sapphire/NBR (20% ACN) interface and strongly resembles that of NBR (20%

ACN)/air interface except for its narrower width. As far as the NBR/air spectrum is concerned, there were no major difference between NBR (40% ACN) and NBR (20%

ACN). Trace (B) then, shows an SFS spectrum of the sapphire/NBR (20% ACN) interface where no nitrile CN stretch band was found in the range of (2200-2300 cm -1).

The major point is that the known assignment at 2233 cm -1 attributed to the nitrile stretch of NBR is absent from the SFS spectrum of NBR (20% ACN) at the sapphire interface.

110

Table 5.7. Positions of the CN stretching vibration for NBR (40% ACN) in the different regions probed from Figure 5.13 and 5.15 obtained by fitting using Equation (3.22). Region probed a Location (Width b)cm -1 NBR (20%ACN)/Air 2059(6) NBR (40%ACN)/Air 2059(6) Sapphire/NBR (20%ACN) 2057(3) Sapphire/NBR 2039/2024(2/1)- (40%ACN) 2233(5) a The error limit for peak position is ± 0.5 cm -1. b Γ The width of the vibrational band is defined as q in Equation 3.22.

The set of SFS data on NBR (20% ACN) demonstrates the effect of the ACN on the

NBR/sapphire spectrum in an unexpected manner. Instead of influencing the 2050 cm -1 vibrational bands, it affected the nitrile features in the (2200-2300 cm -1) range. The

NBR/air spectra were independent of the ACN content.

5.3.3 Summary of the Analysis of Dried NBR at Interfaces

The NBR’s fundamental vibration stretch was detected at 2233 cm -1 at the sapphire/NBR interface and was absent from the SFS spectrum of NBR at the NBR/air interface.

Furthermore, the investigation of the NBR revealed a series of bands and a band located in the (2000-2100 cm -1) range on the SFS spectra of the sapphire/NBR and NBR/air interfaces respectively.

The first step in elucidating the origin of these bands at 2050 cm -1 comprised a purification step (as mentioned earlier) to determine if the assignment of the 2050 cm -1

111 band to determine if the band could be attributed to additives (i.e. an amine) present at the rubber interface. The purification showed that the 2050 cm -1 band persisted after the purification and that furthermore the intensity between the bands in the SFS spectra of

NBR (40% ACN)/air and purified NBR (40% ACN)/air are similar. This experiment ruled out the additive as the contributor of the 2050 cm -1 band.

The following study was devoted to the influence of the ACN on the SFS spectra of two

NBRs with 40% and 20% ACN which revealed an effect shown in the SFS spectra at the sapphire/NBR interface. As the content in acrylonitrile decreases, only the vibrational feature at 2050 cm -1 is present at both interfaces.

If the hypothesis of an assignment of the band in the low frequency region is an interaction between a nitrile functional group and a specific molecule it is probably possible to affect this interaction by altering the dielectric constant of the medium and the ideal interface to carry on the experiment is the NBR/air interface where the 2050 cm -1 features are the strongest.

The following experiments were designed for a twofold reason: to observe first the spectroscopic changes in the 2233 cm -1 vibrational band of nitrile rubber and correlate the nature of the environment to the nitrile stretching mode. Second, the observation of the behavior of the 2050 cm -1 band in various environments should help in the assignment of this peak.

112 5.4 Resistance of NBR to Solvents

This section presents the SFS experiments on sapphire/NBR (40% ACN) and NBR (40%

ACN)/solvent interfaces. Experimentally, the air molecules against the NBR thin film are replaced by solvent molecules, while the spectroscopic feature of the nitrile rubber film are recorded.

5.4.1 Resistance of NBR Surface to Heptane

The features of the SFS spectrum of NBR/heptane interface are compared first to the

NBR/air interface to quantify the impact of the solvent on the NBR thin film. Figure 5.16 shows trace (A), the NBR/air interface, and trace (B), the NBR/heptane interface, for comparison purposes. The fitting for both traces is reported in Table 5.8.

60

40 20 (A) 0 20 15 10 5 (B) 0 SFG Intensity (a.u.)

2000 2050 2100 2150 2200 2250 2300 -1 I.R. Wavenumber (cm ) Figure 5.16. SFS spectra of NBR (40% ACN) in PPP polarization: (A) at the NBR (40% ACN)/air interface and (B) at the NBR (40% ACN)/heptane interface in the CN stretching region.

113 At a first glance, the two spectra at the NBR (40% ACN)/air and NBR (40%

ACN)/heptane interfaces reported in Figure 5.16 look similar. The dominant feature of these spectra is the vibrational band in the (2000-2100 cm -1) region. The intensity and the width of this feature change upon heptane injection as it slightly red-shifts (5 cm -1) and decreases in intensity. At a secondary level, it is noticeable to point out the appearance of a band at around 2233 cm -1 upon heptane exposure. A possible explanation would be that the solvent molecules replace molecules present at the interface, revealing a peak around

2233 cm -1. If this explanation is correct, the 2050 cm -1 band would disappear upon heptane injection and it is clear that this band at 2050 cm -1 still dominates the SFS spectrum at the NBR (40% ACN)/heptane interface.

If the 2050 cm -1 feature would be attributed to independent molecules at the NBR (40%

ACN)/air interface that are soluble in heptane, then these molecules would be washed out upon heptane exposure and the 2050 cm -1 vibrational band would disappear from the SFS spectrum. Furthermore, the hypothesis of the presence of molecules at the NBR (40%

ACN)/air interface interacting with the nitrile groups of NBR seems to be credited by the slight shift of the 2050 cm -1 feature upon heptane exposure. The molecular picture describing the NBR/heptane interface would involve specific molecules interacting strongly with the nitrile groups of NBR. The interaction between these molecules and the

NBR nitrile moieties is indicated by a large shift of the CN stretching mode, and is only slightly modified by the heptane injection.

114 If heptane is not a good solvent for NBR, the heptane molecules should stay at the interface rather than solubilize the thin film and reach out to the sapphire interface. This point can be investigated further by probing the spectroscopic feature of NBR at the sapphire/NBR interface to see if any observable changes are picked up as well. The sapphire/NBR SFS spectrum was collected one hour after the injection of heptane.

Figure 5.17 presents a comparison between the SFS spectra at the sapphire/NBR (40%

ACN) in the dried state before the injection of heptane and the corresponding SFS spectrum at the sapphire/NBR (40% ACN) after one hour of exposure to heptane.

1.6

1.2

0.8 (A) 0.4

0.0 1.5

1.0

SFG Intensity (a.u.) SFG Intensity 0.5 (B)

0.0

2000 2050 2100 2150 2200 2250 2300 -1 I.R. Wavenumber (cm ) Figure 5.17. SFS spectra of NBR (40% ACN) in PPP polarization at the sapphire/NBR (40% ACN) interface: (A) before heptane exposure and (B) after 1 h of heptane exposure in the CN stretching region.

Contrary to the previous case of the NBR (40% ACN)/heptane interface, the SFS spectrum of the interface sapphire/NBR in the dried state is almost identical to that of the

115 sapphire/NBR after one hour of exposure to heptane. The peak position, relative intensity between the peaks, and the width of the peaks are similar. None of the two spectral regions pointed out so far have changed. The 2233 cm -1 band assigned to the nitrile CN stretch of nitrile rubber dominates both SFS spectrum before and after the heptane injection. The weak bands in the (2000-2100 cm -1) region persist as well.

Table 5.8: Positions of the CN stretching vibration for NBR (40% ACN) in the different regions probed from Figure 5.16 and 5.17 obtained by fitting using Equation(3.22).

Region probed Location a(width) cm -1

NBR (40% ACN)/air 2059(6)

NBR (40% ACN)/heptane 2055(5) Sapphire/NBR (40% ACN) 2039(2) 2054(2) 2234(5) Sapphire/NBR (40% ACN) under heptane 2039(2) 2053(2) 2234(5) exposure a The error limit for peak position is ± 0.5 cm -1.

The first comment that deserves credit concerns the feasibility of such an experiment.

Investigation of polymer/liquid interface is certainly not new, but it is challenging. When investigating a polymer for the first time, it is best to first find a spectral region free of interference or with a limited number of bands. Therefore, the first conclusion based on the set of SFS data related to the NBR (40% ACN)/heptane interface and sapphire/NBR

(40% ACN) interface under heptane exposure is that SFS can be used to monitor the behavior of a polymeric thin film under liquid exposure.

116 The next step of the investigation is to extract scientific information from this set of four spectra. The obvious first remark interpreted from Figures 5.16 and 5.17 is that heptane molecules weakly interact with the NBR surface and do not diffuse. According to

Salomon 102 and Starmer 50 NBR swelling curves, a low interaction exists between the

NBR and heptane. This is confirmed by the stability over time of the spectra for the sapphire/NBR interface .

This study was designed to seek interaction information via spectroscopic manifestations.

To obtain information on the interaction of rubber with solvent, it is better to study a highly interactive system to capture spectroscopic manifestations such as NBR/toluene. 7,

50 In the following section, a duplicate investigation was conducted, this time on the NBR

(40% ACN)/toluene system.

5.4.2. Resistance of NBR Surface to Toluene

This experiment uses the same procedures as described earlier but now exposing an NBR film to toluene. Toluene is a good swelling agent of NBR; therefore, the characterization of the NBR (40% ACN)/toluene interface would lead to information on how NBR interacts specifically with toluene or on the impact of toluene solvent molecules of NBR up to an extent where the 2050 cm -1 band could significantly change. However, the interface NBR/toluene cannot be directly probed as the refractive index of toluene is higher than that of the NBR. In such a case, the requirement for total internal geometry cannot be satisfied and the reflected SFS signal is weaker. Instead, an alternate

117 experiment was designed where the sapphire interface would be monitored after successive exposure to toluene. Figure 5.18 presents the results of the experiment. Trace

(A) is the SFS spectrum of the NBR (40% ACN)/air interface, trace (B) is the SFS spectrum of the sapphire/NBR interface in the dried state, trace (C) represents the sapphire/NBR (40% ACN) interface after the NBR film was exposed during 30 s to toluene. And finally, trace (D) corresponds to the spectroscopic profile of the sapphire/NBR (40% ACN) after a second toluene exposure this time for 4 minutes.

The SFS spectra of the NBR (40% ACN)/air on trace (A) and sapphire/NBR (40% ACN) on trace (B) in the dried state were analyzed first. These spectra look almost identical to that obtained in the previous section. The band at 2233 cm -1 was attributed to the nitrile stretch of NBR. After the acquisition of the first SFS spectrum at the sapphire/NBR interface, the NBR spin-coated prism was exposed to toluene for 30s and nitrogen-dried for 2 minutes, then placed under moderate nitrogen flow for a second SFS analysis. The corresponding SFS spectrum is displayed on trace (C) of Figure 5.18. The 2233 cm -1 band still dominates the spectrum while a small contribution appeared in the lower frequency range of the spectrum at 2050 cm -1.

118

25 20 15 10 5 (A) 50 4 3 2 (B) 1 0 2.5 2.0 1.5 1.0 (C) 0.5 0.0 SFG Int. (a.u.) 3.0 2.0 1.0 (D) 0.0

2000 2050 2100 2150 2200 2250 2300 -1 Wavenumber (cm ) Figure 5.18. SFS spectra of NBR (40% ACN) in PPP polarization (A) at the NBR (40% ACN)/air interface, (B) at the sapphire/NBR (40% ACN) interface before toluene exposure, (C) at the sapphire/NBR (40% ACN) interface after 30 s of toluene exposure and N 2 drying and (D) at the sapphire/NBR (40% ACN) interface after 4 additional minutes of toluene exposure and N 2 drying, in the CN stretching region.

It seems that the toluene exposure produced some spectroscopic changes at the sapphire/NBR interface. To further confirm this trend, the NBR spin-coated prism was exposed to toluene for 4 minutes and nitrogen dried for 2 minutes. It was then placed under moderate nitrogen flow for a second SFS analysis. A third SFS spectrum of the sapphire/NBR (40% ACN) interface was acquired and is represented by trace (D) on

Figure 5.18. Evident changes are observed at the sapphire/NBR interface. After the second toluene exposure, a band as intense as the band assigned to the nitrile is detected

119 at 2050 cm -1. The sapphire/NBR interface closely resembles the NBR/air interface in the dried state with an intense band at 2233 cm -1. It is surprising that the 2233 cm -1 band is not changing as the 2050 cm -1 band continues to increase. Speculation was that this 2050 cm -1 band would continue to increase and that the 2233 cm -1 band would at some point decrease as the film will start to dissolve at the sapphire/NBR interface. This data set strongly suggests that toluene etched the NBR film because specific spectroscopic features initially present at the NBR/air interface were transmitted to the sapphire/NBR interface.

The appearance of the 2050 cm -1 band on the spectroscopic profile of the SFS spectra at the sapphire/NBR interface could also be due to a significant decrease of the NBR film thickness during the toluene exposure. It is clear that the band is present at the NBR(40%

ACN)/air interface, and could be detected at the sapphire/NBR(40% ACN) interface if the thickness decreases. Also, it means that the 2050 cm -1 peak cannot be due to an independent molecule that is soluble in toluene because it is still detected after two exposures to toluene.

Table 5.9. Positions of the CN stretching vibration for NBR (40% ACN) in the different regions probed from Figure 5.18 obtained by fitting using Equation (3.22).

Region probed Position a(width) cm -1 Sapphire/air 2059(7) Sapphire/NBR (40% CAN) (dried) 2233(5) Sapphire/NBR (40% ACN) (after first toluene 2055(3) 2233(4) exposure) Sapphire/NBR (40% ACN) (after second toluene 2054(5) 2233(5) exposure) a The error limit for peak position is ± 0.5 cm -1. 120

Spectroscopic analysis of the NBR/toluene system reveals different behavior at the sapphire surface than that of the NBR/heptane system. From a spectroscopic viewpoint, the changes in the SFS spectrum at the sapphire interface upon toluene exposure shows that the toluene molecules reach the sapphire interface. It is important to note that this effect is either a real perturbation of the sapphire interface caused by solvent molecules or it is the direct effect of a decrease in the NBR film thickness.

5.4.3 Resistance of NBR Surface to Water

As observed in the first part of the previous section, there are several hypothetical molecules capable of inducing a significant shift of the nitrile stretch. But out of all of them, the one that has the highest probability to cause a spectroscopic shift is water.

This final section will study on the behavior of NBR in contact with water.

Chapter 3 explained how the nitrile moiety is capable of interacting with water.

Acetonitrile and polyacrylonitrile both interact with water with a positive shift of 3 cm -1 at room temperature and a 200 cm -1 negative shift at high temperature and under pressure, respectively. Because the fact that the observation was made on PAN and not nitrile rubber and because the interaction between nitrile groups in NBR is weaker than that in

PAN, it is possible that another interaction between interfacial nitrile of NBR and water takes place at room temperature and atmospheric pressure.

121 A comparison between the NBR (40% ACN)/air interface and NBR (40% ACN)/water interface is presented in Figure 5.19. Various motivations drove this project over time, and when these data were recorded, special attention was given to the C-H region of

NBR. That is why the spectra were acquired in the (2800-3200 cm -1) region. An immediate loss of the SFS signal was recorded at the sapphire/NBR (40% ACN) interface presented in Figure 5.19. The same response is observed at the sapphire/NBR (40%

ACN) interface, as shown in Figure 5.20.

(A)

(B) SFG intensity (a.u.)

2800 2900 3000 3100 3200 -1 I.R. wavenumber (cm ) Figure 5.19. SFS spectra of NBR (40% ACN) in PPP polarization at the sapphire/NBR (40% ACN) interface (A) before water exposure and (B) after water exposure, in the CH stretching region.

(A)

(B) SFG intensity (a.u.)

2800 2900 3000 3100 3200 -1 I.R. wavenumber (cm )

Figure 5.20. SFS spectra of NBR (40% ACN) in PPP polarization (A) at the NBR (40% ACN)/ air interface and (B) NBR (40% ACN)/water interface, in the CH stretching region.

122 The loss of SFS signal led to a non-conclusive experimental result with respect to assigning the band at 2050 cm -1 to nitrile/water interaction. However, losing the SFS signal itself carries potential information. As explained in the first section of Chapter 3, this technique is sensitive to the symmetry of the molecular arrangement at the interface.

One could infer a complete disorganization of the NBR interface due to the high polarity of the water solvent inducing a loss of SFS signal.

Therefore water, a very polar medium, significantly affects the NBR surface while it does almost nothing to the PAN surface. The role of the nitrile groups could be easily investigated by probing NBR/water interface with ACN content intermediate between

100 and 40% ACN.

5.4.4 Assigning the 2050 cm -1Band.

There are around 25 vibrational modes in the (1900-2300 cm -1) region. The assignments for these vibrational bands include stretching of triple bonds or out-of-phase stretching of cumulated double bonds stretching vibrations.64 Out of these 25 vibrational modes, only eight appear in the (2000-2100 cm -1) region (see Table 5.10)

Several possibilities can explain the detection of a band at 2050 cm -1 at the NBR/air surface. A residual chemical present at the NBR surface having a vibrational frequency of its own at this location, a by-product of a chemical reaction between the nitrile groups of

123 nitrile rubber or a chemical species that interacts with the nitrile group of NBR inducing a negative spectroscopic shift of the nitrile stretch vibration.

Table 5.10. Vibrational modes absorbing in the (2000-2100 cm -1) region.64

Triple bonds and cumulated double bonds vibrations in the (2000 2100

2050 2150 cm 1

2000 2050 cm 1 Ketene

2012 2132 cm 1 Diazo compound

2080 2170 cm 1 Organic azid

2070 2200 cm 1 Cyanide Ions

2100 cm 1 Ferri cyanide

2020 2090 cm 1 ions

Metal (CO) 1900 2170 cm 1

To decrease the number of potential assignments for the 2050 cm -1 band, a typical recipe of NBR is presented in Table 5.11 21 and will serve to assess the probability of the each assignment.

124

Table 5.11 Typical recipe for NBR emulsion polymerization 5

Typical recipe for butadieneacrylonitrile copolymerization

Butadiene 75 Acrylonitrile 25

Soap 4.5 Stearic acid 0.6

tdodecyl mercaptan () 0.5 RSH

Potassium chloride 0.3 Sodium 0.35 Ferrous sulfate 0.1 Hydrogen peroxide 0.02

Water 180

A comparison between Table 5.10 and Table 5.11 reveals several common points. First, the presence of salt such as potassium chloride, sodium pyrophosphate and ferrous sulfate gives credit to a possible interaction between the nitrile groups of NBR and specific molecules. It also indicates that there is a possibility of finding cyanide ion in complex.

For example, cyanide ion sodium and potassium salts absorb at 2080-2070 cm -1 and ferri cyanide ion absorbs near 2100 cm -1.64

Also, a small amount of mercaptan(thiol) could hypothetically react with nitrile groups of nitrile rubber and form thiocyanate ions. As Table 5.10 shows, inorganic thiocyanate ions absorb strongly at 2020-2090 cm -1. 125 The second hypothesis is that the amine detected by HPLC could react and forms diazo compounds. Therefore, the next question is, whether further reactions could take place in the NBR, knowing that some amine and acidic molecules (stearic acid) are present in the formulations. In other words, what is needed to carry a diazotization? amine can be N-nitrosated and the resulting species rearrange to form diazonium ions. These ions are unstable unless stabilized by a resonance structure.103 The experimental conditions for the N-nitrosation involve as a reactant which is very unlikely to be found in

NBR.

The next hypothesis concerns the possibility of obtaining organic acid by-products.

Organic azides are typically synthesized by transforming a by reacting an

ion, N3 . These azides would have to be present in the emulsion recipe, which is unlikely, according to Table 5.11.

The most probable assignment for the 2050 cm -1 vibrational band is either an interaction with residual salt used during the NBR emulsion polymerization or that of a nitrile complex. It is unclear at this point why the 2050 cm -1 band is only detected at the

NBR/air interface.

5.4.5 Conclusions on Solvent Resistance of NBR.

The response of the NBR thin film surface to heptane, toluene and water exposure was analyzed using SFS. From a diffusion point of view it is clear that the NBR/heptane case 126 differs from that of the NBR/toluene. The SFS signal at the sapphire/NBR interfaces under heptane exposure was effectively stable over time while that corresponding to the sapphire/NBR under toluene exposure shows observable changes. It is not possible at this time to discriminate between an interference effect due to the sensitivity of SFS that is directly related to the film thickness or to a real spectroscopic manifestation resulting from the penetration of toluene through the NBR film.

The uncertainty about the assignment of the 2050 cm-1 does not affect the conclusion on

NBR thin film stability in various solvent because two SFS spectra of NBR interface are compared before and after solvent exposure. But it for now impedes any conclusion on the spectroscopic impact of the solvent on NBR surface as it is still unclear if the film went through a major change due the possible presence of chemical interacting at the

NBR surface.

The spectroscopic investigation of NBR surface under liquid exposure demonstrated the ability of SFS to distinguish two systems such as NBR/heptane and NBR/toluene. As a conclusion, the SFS experiment allows accurate tracking of the diffusion of the solvent molecules to the sapphire interface. Summary and further conclusions are placed in the next and final chapter of this dissertation.

127

CHAPTER VI

SUMMARY AND CONCLUSIONS

The main goal of this work was to determine, using SFS spectroscopy, a molecular mechanism that explains the resistance of nitrile rubber (NBR) to hydrocarbon liquids.

The ACN content of nitrile rubber was varied and it was hypothesized that the CN stretch of NBR would depend on the environment. The nitrile rubber/air interface was probed and compared to an NBR/liquid interface.

First, bulk and surface bulk characterization of PAN, PBD and NBR (40 % and 20%

ACN) was carried out using FTIR, Raman and SF spectroscopies. Next, the solvent resistances of PAN and NBR were determined.

The effect of the hydroxyl groups present at the sapphire/PAN interface on the CN stretching of PAN was determined on the SFS sapphire/PAN spectrum. The strong dipole-dipole interactions in PAN were disrupted at high temperature so that the nitrile groups of PAN can now interact with the environment. The effect of the hydroxyls at the sapphire interface on the nitrile stretching frequency was demonstrated.

128 Analysis of PBD revealed a very interesting response when PBD was exposed to a polar environment at room temperature. The SFS spectrum of PBD at the sapphire/PBD interface exhibits, at room temperature, a positive shift of the methylene stretch compared to that of bulk PBD. The sapphire shifts the PBD methylene stretch vibration.

The investigation of the surface of NBR with 40% and 20% ACN demonstrated a new

2050 cm -1 band at the NBR/air interface, even though this band is not present in NBR bulk. The CN stretch of NBR, at 2233 cm -1, was detected only at the sapphire/NBR (40%

ACN) interface. The band found at 2050 cm -1 did not seem to be due to additives in the rubber since SFS results of purified NBR were very close to those of non-purified NBR.

Molecules such as salts that complex with the nitrile groups of NBR to induce a shift of the nitrile stretching mode from 2230 to 2050 cm -1 is the most probable cause.

The second part of the studies involved exposing NBR to various solvents, heptane, toluene and water. The sapphire/NBR and NBR/liquid interfaces were analyzed.

Comparison between the SFS spectra of sapphire/NBR interface before and after solvent exposure was done. No changes were detected at the sapphire/NBR interface after one hour of heptane exposure.

But specific spectral feature appeared at the sapphire/NBR interface after exposure to toluene.

129 The appearance of a band at 2050 cm -1 for the sapphire/NBR interface may be due to toluene molecules as they reach the sapphire interface or to mixing of the SFS signals between the NBR/air and sapphire/NBR interfaces, as a consequence of a significant decrease in NBR film thickness.

Finally, the SFS spectra of NBR/water interface under water exposure show a loss of signal detection. This differs from the case of PAN where an SFS spectrum at PAN/water interface gave an SFS signal.

The hypothesis that the 2050 cm -1 band is due to salts remaining from emulsion polymerization could be further checked by looking for this band in other acrylonitrile polymers synthesized using similar salts.

For the sapphire/NBR interface after toluene exposure, dissolution of the film prevents a conclusion. Originally, since the amount of acrylonitrile was the primary parameter to consider, crosslinked NBR samples were not studied to avoid the complex situation, where two parameters, the crosslinked density and the NBR acrylonitrile content, would be analyzed simultaneously. However, the use of crosslinked samples may help in further characterizing the sapphire/NBR interface under toluene exposure. A monitoring of the nitrile CN stretch in the 2200-2300 cm -1 region of both the SFS spectra of the sapphire/NBR in, heptane and toluene exposure may show whether or not oil resistance phenomenon is a surface-related phenomenon.

130 It would also be important to use a series of solvents in the SFS studies. A comparative study of toluene and cyclohexane would be helpful in revealing the exact impact of the aromatic ring. Finally, the use of halogenated compounds would be very interesting since the rubber/Cl system has a negative change in enthalpy upon mixing.

131

REFERENCES

(1) Mark, H. F.; Bikales, N.; Overberger, C. G.; Menges, G.; Kroschwitz, J. I., Eds.; In Encyclopedia of Polymer Science and Engineering, 2nd ed.; Anionic Polymerization to Cationic Polymerization; Wiley: New York, 1985; Vol. 2; pp 558-562.

(2) Morton, M., Ed.; In , 2nd ed.; Van Nostrand Reinhold Company: New York, 1973; pp 302-308.

(3) Anonymous, Chem. Eng. News 2002 , 80(25) , 63.

(4) Graham, J. L.; Striebich, R. C.; Myers, K. J.; Minus, D. K.; Harrison, W. E.,III Energy Fuels 2006 , 20 , 759-765.

(5) Odian, G. G. In Principles of Polymerization,4th ed.; Wiley,: Hoboken, N.J., 2004; p 812.

(6) Stevens, M. P. In Polymer Chemistry: An Introduction; Oxford University Press: New York, 1999.

(7) Salomon, G. Rubber Chem. Technol. 1948 , 21 , 805-813.

(8) Westbrook, P. A.; French, R. N. Polym. Eng. Sci. 2007 , 47 , 1554-1568.

(9) Sandu, C.; Singh, R. K. Food Technol. 1995 , 45 , 84-91.

(10) Hubbell, J. A. Bio/Technology 1995 , 13 , 565-576.

(11) Ishihara, K.; Oshida, H.; Endo, Y.; Ueda, T.; Watanabe, A.; Nakabayashi, N. J. Biomed. Mater. Res. 1992 , 26 , 1543-1552.

(12) Martin, B. D.; Gaber, B. P.; Patterson, C. H.; Turner, D. C. Langmuir 1998 , 14 , 3971-3975.

(13) Inglis, W.; Sanders, G. H. W.; Williams, P. M.; Davies, M. C.; Roberts, C. J.; Tendler, S. J. B. Langmuir 2001 , 17 , 7402-7405.

(14) Rechnitz, G. A. Chem. Eng. News 1988 , 66(36) , 24-36.

(15) Dyer, R. B.; Shreve, A. P. 1998 , LA-UR-98-1831 , 1-16.

132 (16) Cheng, X.; Canavan, H. E.; Stein, M. J.; Hull, J. R.; Kweskin, S. J.; Wagner, M. S.; Somorjai, G. A.; Castner, D. G.; Ratner, B. D. Langmuir 2005 , 21 , 7833-7841.

(17) Li, G.; Ye, S.; Morita, S.; Nishida, T.; Osawa, M. J. Am. Chem. Soc. 2004 , 126 , 12198-12199.

(18) Henrici-Olive, G.; Olive, S. Adv. Polym. Sci. 1979 , 32 , 123-152.

(19) Schreuder-Gibson, H. L.; Gibson, P.; Senecal, K.; Sennett, M.; Walker, J.; Yeomans, W.; Ziegler, D.; Tsai, P. P. J. Adv. Mater. 2002 , 44 , 34-55.

(20) Fertala, A.; Han, W. B.; Ko, F. K. J. Biomed. Mater. Res. 2001 , 57 , 48-29.

(21) Mark, H. F.; Bikales, N.; Overberger, C. G.; Menges, G.,Eds.; In Encyclopedia Polymer Science and Engineering,2nd ed.; Wiley: New York, 1985; Vol. 1; pp 434- 450.

(22) Saum, A. M. J. Polym. Sci. 1960 , 42 , 57-66.

(23) Ptitsyn, O. B.; Sharonov, I. A. J. Tech. Phys. 1957 , 27 , 2744-2762.

(24) Krigbaum, W. R.; Tokita, N. J. Polym. Sci. 1960 , 43 , 467-488.

(25) Holland, V. F.; Mitchell, S. B.; Hunter, W. L.; Lindenmeyer, P. H. J. Polym. Sci. 1962 , 62 , 145-151.

(26) Kimmel, R. M.; Andrews, R. D. J. Appl. Phys. 1965 , 36 , 3063-3071.

(27) Andrews, R. D.; Kimmel, R. M. J. Polym. Sci., Part B: Polym. Lett. 1965 , 3 , 167- 169.

(28) Okajima, S.; Ikeda, M.; Takeuchi, A. J. Polym. Sci., Part A-1: Polym. Chem. 1968 , 6, 1925-1933.

(29) Minami, S. Appl. Polym. Symp. 1975 , 25 , 145-157.

(30) Ogura, K.; Kawamura, S.; Sobue, H. Macromolecules 1971 , 4 , 79-81.

(31) Rowlen, K. L.; Harris, J. M. Anal. Chem. 1991 , 63 , 964-969.

(32) Mark, H. F.; Bikales, N.; Overberger, C. G.; Menges, G., Eds.; In Encyclopedia Polymer Science and Engineering,2nd ed.; Wiley: New York, 1985; Vol. 3; pp 421- 425.

(33) Mark, H. F.; Bikales, N.; Overberger, C. G.; Menges, G., Eds.; In Encyclopedia of Polymer Science and Engineering,2nd ed.; Wiley: New York, 1985; Vol. 13; pp 186-191.

133 (34) Gee, G.; Treloar, L. R. G. Trans. Faraday Soc. 1942 , 38 , 147-165.

(35) Gee, G. Trans. Faraday Soc. 1942 , 38 , 276-284.

(36) Gee, G. Trans. Faraday Soc. 1942 , 38 , 418-422.

(37) Gee, G. Trans., Inst. Rubber Ind. 1943 , 18 , 266-281.

(38) Gee, G. Trans. Faraday Soc. 1944 , 40 , 463-468.

(39) Gee, G. Trans. Faraday Soc. 1944 , 40 , 468-480.

(40) Gee, G.; Orr, W. J. C. Trans. Faraday Soc. 1946 , 42 , 507-517.

(41) Gee, G. Trans. Faraday Soc. 1946 , 42 , 585-598.

(42) Gee, G. Trans. Faraday Soc. 1946 , 42B , 33-44; discussion, 44-50.

(43) Treloar, L. R. G. In The Physics of Rubber Elasticity.2nd ed.; Oxford University Press: New York, 1958; p 342.

(44) Flory, P. J.; Krigbaum, W. R. Ann. Rev .Phys. Chem. 1951 , 2, 383-402.

(45) Hansen, C. M. J. Paint Technol. 1967 , 39 , 104-117.

(46) Wilson, A.; Giffis, C. B.; Montermoso, J. C. Rubber World 1958 , 139 , 63-73.

(47) Magryta, J.; Debek, C.; Debek, D. J Appl. Polym. Sci. 2006 , 99 , 2010-2015.

(48) Busfield, J. J. C.; Deeprasertkul, C.; Thomas, A. G. Const. Models Rubber, Proc.Eur.Conf., 1st 1999 , 87-93.

(49) Starmer, P. H. J. Elastomers Plast. 1993 , 25 , 59-73.

(50) Starmer, P. H. J. Elastomers Plast. 1993 , 25 , 120-142.

(51) Godfrey, N. B. Chem. Technol. 1972 , 2, 359-363.

(52) Dack, M. R. J. Tech. Chem.(N.Y.) 1976 , 8, 95-157.

(53) Singh, A.; Mukherjee, M. Macromolecules 2003 , 36 , 8728-8731.

(54) Li, Y.; Park, E. J.; Lim, K. T.; Johnston, K. P.; Green, P. F. J. Polym. Sci., Part B: Polym. Phys. 2007 , 45 , 1313-1324.

(55) Smith, A. L.; Shirazi, H. M. J. Therm. Anal. Calorim. 2000 , 59 , 171-186.

(56) Yamaguchi, T.; Nakao, S.; Kimura, S. Macromolecules 1991 , 24 , 5522-5527.

134 (57) Koops, G. H.; Yamada, S.; Nakao, S. J. Membr. Sci. 2001 , 189 , 241-254.

(58) Stamatialis, D. F.; Stafie, N.; Buadu, K.; Hempenius, M.; Wessling, M. J. Membr. Sci. 2006 , 279 , 424-433.

(59) Su, Y.; Li, C. J. Membr. Sci. 2007 , 305 , 271-278.

(60) Garcia, A.; Alvarez, S.; Riera, F.; Alvarez, R.; Coca, J. J. Membr. Sci. 2005 , 253 , 139-147.

(61) Pauling, L. The Chemical Bond; a Brief Introduction to Modern Structural Chemistry; Cornell University Press: Ithaca, N.Y., 1967; pp 69-72.

(62) Pauling, L. In The Nature of the Chemical Bond and Structure of Molecules and Crystals: An Introduction to Modern Structural Chemistry,3rd ed.; Cornell University Press: Ithaca, N.Y., 1960; pp 97-103.

(63) Fawcett, W. R.; Liu, G.; Kessler, T. E. J. Phys. Chem. 1993 , 97 , 9293-9298.

(64) Colthup, N. B. In Introduction to Infrared and Raman Spectroscopy,2nd ed.; Colthup, N. B., Daly, L. H. and Wiberley, S. E., Eds.; Academic press: New York, 1975.

(65) Schrader, B. Infrared and Raman Spectroscopy: Methods and Applications; Weinheim: New York:VCH, 1995.

(66) Hollas, J. M. Basic Atomic and Molecular Spectroscopy; Wiley-Interscience: New York, 2002.

(67) Califano, S. Vibrational States; Wiley-Interscience Publication: New York, 1976.

(68) Andrews, S. S.; Boxer, S. G. J Phys. Chem. A 2000 , 104 , 11853-11863.

(69) Lambert, D. K. Phys. Rev. Lett. 1983 , 50 , 2106-2109.

(70) Lambert, D. K. J. Chem. Phys. 1988 , 89 , 3847-3860.

(71) Papousek, D.; Graner, G.; Burger, H., Eds.; In Vibrational-Rotational and Molecular Dynamics:Advances in Quantum Chemical And Spectrocopical Studies of Molecular Structures and Dynamics; Advanced Series in Physical Chemistry ; World Scientific: River Edge, N.J., 1997; Vol. 9.

(72) Zhang, V. L.; Arnolds, H.; King, D. A. Surf. Sci. 2005 , 587 , 102-109.

(73) Hirose, C.; Akamatsu, N.; Domen, K. Appl. Spectrosc. 1992 , 46 , 1051-1072.

(74) Hirose, C.; Akamatsu, N.; Domen, K. J. Chem. Phys. 1992 , 96 , 997-1004.

135 (75) Hirose, C.; Yamamoto, H.; Akamatsu, N.; Domen, K. J. Phys. Chem. 1993 , 97 , 10064-10069.

(76) Shen, Y. R. Principles of Non-Linear Optics; Wiley-Interscience Publication: New York, 1984.

(77) Zhuang, X.; Miranda, P. B.; Kim, D.; Shen, Y. R. Phys. Rev. B: Condens. Matter Mater. Phys. 1999 , 59 , 12632-12640.

(78) Franken, P. A.; Hill, A. E.; Peters, C. W.; Weinreich , G. Phys. Rev. Lett. 1961 , 7 , 118-119.

(79) Nyquist, R. A. Appl. Spectrosc. 1990 , 44 , 1405-1407.

(80) Knoezinger, H.; Krietenbrink, H. J. Chem. Soc., Faraday Trans.1 1975 , 71 , 2421- 2430.

(81) Ahn, D.; Lee, S. Bull. Korean Chem. Soc. 2003 , 24 , 545-546.

(82) Fini, G.; Mirone, P. Spectrochim. Acta, Part A 1976 , 32A , 439-440.

(83) Andreeva, O. A.; Burkova, L. A. J. Macromol. Sci., Phys. 2000 , B39 , 225-234.

(84) Chung, T. C.; Schlesinger, Y.; Etemad, S.; Macdiarmid, A. G.; Heeger, A. J. J. Polym. Sci., Polym. Phys. Ed. 1984 , 22 , 1239-1246.

(85) Wu, C. R.; Liedberg, B. J. Polym. Sci., Part B: Polym. Phys. 1988 , 26 , 1127-1136.

(86) Loch, C. L.; Ahn, D.; Chen, C.; Wang, J.; Chen, Z. Langmuir 2004 , 20 , 5467-5473.

(87) Zhang, D.; Gutow, J. H.; Eisenthal, K. B.; Heinz, T. F. J. Chem. Phys. 1993 , 98 , 5099-5101.

(88) Yeganeh, M. S.; Dougal, S. M.; Polizzotti, R. S.; Rabinowitz, P. Thin Solid Films 1995 , 270 , 226-229.

(89) 1993 , Lumonics Spectrum Master Dye Laser Service Manual Part No. 763AB031A- 00 .

(90) Rangwalla, H. Molecular Origins of Contact-Angle Hysteresis and Other Phenomena at Aqueous Interfaces of Side-Chain Comb Polymers, PhD Thesis, The University of Akron, Akron, OH, 2005.

(91) Li, G.; Yeganeh, M. S. Dhinojwala., A., to be published .

(92) Azzam, R. M. A. In Ellipsometry and Polarized Light; North Holland Pub. Co.: New York.

136 (93) Bloch, D. In Polymer Handbook, 4th ed.; Brandrup, J., Immergut, E. H. and Grulke, E. A., Eds.; Wiley: New York, 1999.

(94) Tadokoro, H.; Murahashi, S.; Yamadera, R.; Kamei, T. J. Polym. Sci., Part A: Gen. Pap. 1963 , 1, 3029-3042.

(95) Phadke, M. A.; Musale, D. A.; Kulkarni, S. S.; Karode, S. K. J. Polym. Sci., Part B: Polym. Phys. 2005 , 43 , 2061-2073.

(96) Healy, M. H.; Wieserman, L. F.; Arnett, E. M.; Wefers, K. Langmuir 1989 , 5, 114- 123.

(97) Yeganeh, M. S.; Dougal, S. M.; Pink, H. S. Phys. Rev. Lett. 1999 , 83 , 1179-1182.

(98) Goodman, A.; Suwyn, M. A. U.S. Patent 3,896,204, 1975.

(99) Morero, D.; Santambrogio, A.; Porri, L.; Ciampelli, F. Chim. Ind. (Milan, Italy) 1959 , 41 , 758-762.

(100) Binder, J. L. J. Polym. Sci. 1963 , 1 , 47-58.

(101) Furukawa, J.; Kobayashi, E.; Uratani, K.; Iseda, Y.; Umemura, J.; Takenaka, T. Polym. J. 1973 , 4, 358-365.

(102) Salomon, G. Rubber Chem. Technol. 1948 , 21 , 805-813.

(103) Vollhardt, K.; Peter, C. Organic Chemistry: Structure and Function, 3rd ed.; W. H. Freeman: New York, 1999.

137

APPENDICES

138

APPENDIX A

PAN AND NBR THERMAL TRANSITIONS

Differential Scanning Calorimetry (DSC) was performed on PAN at a heating rate of 10

°C/min. Two transitions are detected at 102 °C and 150 °C.

Figure A-1: DSC thermogram of PAN carried at a heating rate of 10 °C/min.

139

Figure A-2: DSC thermogram of NBR carried at a heating rate of 10 °C/min.

140

APPENDIX B

THIN FILMS ELLIPSOMETRY THICKNESS MEASUREMENTS

Fitting results of ellipsometer data performed on PAN, NBR and PBD spin-coated films using Cauchy layer model are presented in Figures B-1,B-2 and B-3.

Generated and Experimental 15 30.0

12 27.0 Ψ

8 degreesin Model Fit 24.0 Exp ∆-E 75° 5 Model Fit Ψ Exp -E 75° 21.0 indegrees 1 ∆

18.0 -3

-6 15.0 200 400 600 800 1000 Wavelength (nm) Figure B-1: Fitting of the measured (dotted lines) and computed values of Ψ and for the measurement on the PAN film spin-coated on a sapphire prism.

141 Generated and Experimental 35 32

25 26 Ψ i erees degr in 15 Model Fit 19 Exp ∆-E 75° Model Fit Exp Ψ-E 75° 5 13 indegrees ∆

-5 6

-15 0 200 400 600 800 1000 Wavelength (nm)

Figure B-2: Fitting of the measured (dotted lines) and computed values of Ψ and for the measurement on the NBR film spin-coated on a sapphire prism.

.

Generated and Experimental 15 35.1

12 28.1 Ψ

8 in degrees 21.1 5 14.0 in degrees in 1 ∆ Model Fit 7.0 -3 Exp ∆-E 75° Model Fit Exp Ψ-E 75° -6 0.0 200 400 600 800 1000 Wavelength (nm) Figure B-3: Fitting of the measured (dotted lines) and computed values of Ψ and for the measurement on the PPB film spin-coated on a sapphire prism.

142

Table B-1, shows the values of A n, B n and C n in the Cauchy layer model for each of the measurements and corresponding film thicknesses.

Table B-1: Fitted Cauchy parameters and polymer thin films thicknesses Polymer Thickness Cauchy parameters sample (nm) An Bn Cn + PAN 1.51 0.0051 0 435 − 1 + NBR 1.5 0.0087 0 226 − 1 + PB 1.53 0.0018 0 824 − 2

143

APPENDIX C

HPLC RUBBER ADDITIVES ANALYSIS

The chromatogram of methanol extraction solution is presented in Figure C-1.

Figure C-1: Chromatogram of NBR extract.

Additives identification was carried out by scanning UV-Visible detection (spectrum presented in Figure C-2). It is clear that the UV-Vis corresponding to the additive is not that of BHT (Figure C-3). Comparison with spectra of reference substances led to the hypothesis of an amine type of additive.

144

Figure C-2: Chemical structure of possible amine type molecule and corresponding UV spectrum.

145

Figure C-3: Chemical structure of Butyl Hydroxylated Toluene (BHT) molecule and corresponding UV spectrum.

146

APPENDIX D

NMR SPECTRA OF NBR

Olefinic protons of BD units appeared at δ =5.3-5.7 ppm. The relative intensity of the higher field peak (low chemical shift) increases with an increase in the BD units in the

NBR, the higher peak field is assigned to the BD-BD-BD sequence and the lower field peak is assigned to the AN-BD-AN sequence.

The methylene protons of the BD units in NBR appeared at δ =2.1-2.3.

Figure D-1. 500 MHz-NMR spectrum of NBR (20% ACN).

147

Figure D-2. 500 MHz-NMR spectrum of NBR (30% ACN).

Figure D-3. 500 MHz-NMR spectrum of NBR (40% ACN).

148

APPENDIX-E

SIZE EXCLUSION CHROMATOGRAPHIC MEASUREMENTS OF NBR

Figure E-1: Chromatogram of NBR (20% ACN)

149

Figure E-1: Chromatogram of NBR (30% ACN)

.

Figure E-1: Chromatogram of NBR (40% ACN)

150