effect of molecular structure on the viscoelastic

15 downloads 0 Views 3MB Size Report
Dec 9, 2010 - The molecular structure of the polymer influenced the viscoelastic properties of the ...... of the extensive intra- and intermolecular hydrogen bonding in ...... (0.016 mol) of HBr solution was added at room temperature under ...
EFFECT OF MOLECULAR STRUCTURE ON THE VISCOELASTIC PROPERTIES OF CELLULOSE ACETATE IN A TERNARY SYSTEM

by

CHIA-WEN CARMEN HSIEH B.Sc., The University of British Columbia, 2007

A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF

MASTER OF SCIENCE in The Faculty of Graduate Studies (Forestry)

THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) December 2010 © Chia-wen Carmen Hsieh

Abstract

A series of ternary systems composed of cellulose acetate (CA), N,N-dimethylacetamide (DMA), and water were prepared by varying the mixing temperature, order of component addition, and polymer substitution pattern with increasing water content. The viscoelastic properties of the resulting ternary systems were measured using steady state and dynamic rheology. The CA/DMA/H2O mixture formed physical gels at 17.5 and 19 wt% nonsolvent concentrations after heating to 50 and 70/90°C respectively. Gel formation was characterized by the loss of a Newtonian plateau in the steady state as well as the transition of the elastic (G) modulus becoming greater than the viscous (G) modulus in the dynamic state. The molecular structure of the polymer influenced the viscoelastic properties of the resulting gel. Commercially available CA was found to be partially acetylated at the C2, C3, and C6 positions and contained a total degree of substitution (DS) of 2.47. As CA cluster size in solution decreases with increasing temperature, viscosity measurements showed higher viscosity for samples heated at 50°C, where the loss of the linear stress-strain relationship occurred at 17.5 wt% water. In the dynamic state, higher heating temperature produced higher elastic moduli with a longer linear viscoelastic region (LVR), indicative of a stable system. Changing the sequence of polymer addition by adding CA to DMA/H2O solution resulted in lower overall viscoelastic moduli as compared to adding water to a CA/DMA solution. CA that was regioselectively synthesized to a DS of 2.4 showed different viscoelastic behaviour than the commercial CA. This polymer was completely acetylated at C2 and C3 and partially acetylated at position 6. The system underwent phase separation induced gelation at much lower nonsolvent content. Stress sweep experiments confirmed a shorter LVR and higher G than commercial CA. Increasing the DS of the regioselective polymer to 2.8 led to a longer LVR and higher G than all other polymers at the same nonsolvent content. The enhanced steady shear viscosity and dynamic viscoelastic properties were a result of the intensification of hydrogen bonding and hydrophobic interactions between the polymer, solvent, and nonsolvent.

ii

Table of contents

Abstract .......................................................................................................................................... ii Table of contents .......................................................................................................................... iii List of tables................................................................................................................................... v List of figures ................................................................................................................................ vi List of schemes............................................................................................................................... x List of symbols and abbreviations .............................................................................................. xi Acknowledgements .................................................................................................................... xiv Dedication .................................................................................................................................... xv 1

Introduction ........................................................................................................................... 1 1.1

Cellulose ........................................................................................................................... 1

1.2

Cellulose modification ...................................................................................................... 9

1.3

Cellulose derivatives ....................................................................................................... 11

1.4

Cellulose acetate and its properties................................................................................. 13

1.4.1

Synthesis of cellulose acetates ................................................................................ 15

1.4.2

Dissolution of cellulose acetates ............................................................................. 20

1.4.2.1 Cellulose acetate properties in solution .................................................................. 21

2

3

1.5

Cellulose and cellulose derivative gels ........................................................................... 26

1.6

Rheological characterization of gels ............................................................................... 31

1.7

Goal of the project .......................................................................................................... 37

1.8

Thesis outline .................................................................................................................. 39

Experimental materials and methods ................................................................................ 40 2.1

Materials ......................................................................................................................... 40

2.2

Instrumentation ............................................................................................................... 40

2.3

Data analysis ................................................................................................................... 43

2.4

Synthesis of regioselective cellulose acetate .................................................................. 44

2.5

Characterization of cellulose acetate samples ................................................................ 50

2.6

Ternary system preparation ............................................................................................ 52

Results and discussion ......................................................................................................... 54 iii

3.1

General ............................................................................................................................ 54

3.2

Dissolution method 1: effect of varying mixing temperature on ternary system formation......................................................................................................................... 54

3.3

Dissolution method 2: effect of varying mixing temperature on ternary system formation......................................................................................................................... 75

3.4

Ternary system formation by varying polymer regioselectivity..................................... 87

3.4.1

Synthesis of regioselective cellulose acetate at the C2 and C3 positions ............... 87

3.4.2

Role of regiochemistry in ternary system formation .............................................. 97

3.4.2.1 Commercial CA vs. regio2.4 CA ............................................................................ 97 3.4.2.2 Effect of increasing acetyl substitution ................................................................. 105 4

5

Conclusions......................................................................................................................... 112 4.1

Effect of ternary system mixing temperature ............................................................... 113

4.2

Effect of ternary system dissolution method ................................................................ 114

4.3

Effect of varying degrees of acetylation at the C6 position.......................................... 115

Future work........................................................................................................................ 117 5.1

Recommendations for future work ............................................................................... 117

Bibliography .............................................................................................................................. 119 Appendices ................................................................................................................................. 131 A: Selected spectra .............................................................................................................. 132 B: Figures and tables ........................................................................................................... 149 C: Calculations .................................................................................................................... 152

iv

List of tables

Table 1.1. Unit cell dimensions of various cellulose polymorphs (Krässig 1993; Claffey & Blackwell 1976; Sugiyama et al. 1991) ................................................................................. 6 Table 1.2. Examples of cellulose esters (bold) and ethers of commercial importance (Balser et al. 2004; Thielking & Marc Schmidt 2006) ............................................................................. 12 Table 1.3. Effect of DS on solubility with CA having different substitution patterns (Deus et al. 1991) .................................................................................................................................... 21 Table 3.1. Hansen solubility parameters for CA, DMA, and water at 25°C (Brandrup et al. 2003) ............................................................................................................................................. 62 Table 3.2. Zero shear viscosity values for samples with 20 wt% nonsolvent prepared by dissolution method 1, dissolution method 2 and dissolution method 2 reheated to 90°C ... 82

v

List of figures

Figure 1.1. Cellulose molecular structure ....................................................................................... 2 Figure 1.2. Hydrogen bonding system of cellulose I, where unfilled arrows are pointing to intramolecular hydrogen bonds and filled arrows point to intermolecular hydrogen bonds (Klemm et al. 2005). Copyright Wiley-VCH Verlag GmbH & Co. KGaA. Reproduced with permission. .................................................................................................................... 3 Figure 1.3. Representation of the different orientations of the C6-O6 bonds: gauche-trans (gt), gauche-gauche (gg), and trans-gauche (tg) conformations ................................................... 4 Figure 1.4. Cellulose I unit cell model according to Meyer and Misch (Meyer & Misch 1937). Copyright Wiley-VCH Verlag GmbH & Co. KGaA. Reproduced with permission. ........... 5 Figure 1.5. Hydrogen bonding in cellulose II. Only atoms involved in hydrogen bonding (represented by dotted lines) are labeled. Top left: intermolecular hydrogen bonds are O6D---O2 in sheets containing only center molecules. Top right: intermolecular hydrogen bonds are O2-D---O6 in sheets containing only origin molecules. Bottom: sheets containing origin and center molecules contain O6-D---O6 and O2-D---O2 intermolecular hydrogen bonds. In the former case O5 and O3 can also act as acceptors. Intramolecular hydrogen bonds are O3-D---O5 in each molecule with a minor component involving O6 as acceptor (Langan et al. 1999). Reproduced with permission from the American Chemical Society ................................................................................................................................... 7 Figure 1.6. Transformation of cellulose into its various polymorphs (Klemm et al. 2005) ........... 8 Figure 1.7. Cellulose regeneration processes involving a cellulose derivative (Viscose technology) and direct dissolution (Lyocell technology) (Klemm et al. 2005) .................. 10 Figure 1.8. Structural representation of cellulose acetate ............................................................. 13 Figure 1.9. Dependence of glass transition temperature, melting temperature, and decomposition temperature on the cellulose acetate degree of substitution (Kamide & Saito 1985) ......... 19 Figure 1.10. Interactions between cellulose and the DMA/LiCl solvent system. Interactions at the C3 hydroxyl position are not pictured for simplicity .......................................................... 22 Figure 1.11. Relationship between radius of gyration Rg and hydrodynamic radius Rh in solution. Rg is the same for both particles (Schulz et al. 2000). Copyright Wiley-VCH Verlag GmbH & Co. KGaA. Reproduced with permission. ....................................................................... 25 Figure 1.12. Schematic phase diagram showing different phase separation mechanisms (Tsunashima et al. 2002) ..................................................................................................... 28 Figure 1.13. Examples of phase separation achieved via (left) nucleation and growth characterized by droplets and (b) spinodal decomposition characterized by bicontinuous morphology (Butler & Heppenstall-Butler 2003). Reprinted with permission from Elsevier ............................................................................................................................................. 29 Figure 1.14. Schematic representation of cluster growth during cross-linking leading to gel formation. The variable p, known as the cross-linking index, measures the extent of the cross-linking reaction measured as the ratio of the number of bonds formed to the total vi

number of possible bonds. R is the radius of the molecular clusters formed (Winter & Mours 1997). Reproduced with kind permission from Springer Science+Business Media 32 Figure 1.15. Viscosity as a function of shear rate ......................................................................... 34 Figure 1.16. Expected dynamic mechanical spectra of the elastic (G) and viscous (G) moduli of the shear storage modulus for an (left) entanglement network system (pseudo-gel) and (right) a covalently cross-linked network (true gel) (Ross-Murphy 1987). Reprinted with permission from Elsevier ..................................................................................................... 36 Figure 2.1. Proposed synthesis of regioselectively substituted cellulose acetate at the C2 and C3 positions............................................................................................................................... 44 Figure 3.1. Effect of mixing temperature on steady shear viscosity for ternary systems with 10 wt% CA and 12.5-17.5 wt% water. Samples were prepared using dissolution method 1addition of water to a CA/DMA solution ............................................................................ 55 Figure 3.2. Evolution from clear colorless solution to cloudy gel network for a 10 wt% CA ternary system with (left to right) 12.5, 15, 17.5, and 20 wt% wate. Samples were prepared using dissolution method 1 and a mixing temperature of 90°C. ......................................... 56 Figure 3.3. Effect of mixing temperature on the viscosity enhancement of 10 wt% CA ternary systems with 12.5-20 wt% water (values extrapolated to zero-shear viscosity for samples exhibiting a Newtonian plateau and at 0.05 sec-1 shear rate for samples exhibiting shear thinning behaviour at low shear rate). Data represent the average of 3 replicates .............. 57 Figure 3.4. Elastic (G) and viscous (G) moduli for 10 wt% CA solution at 15, 20, and 25 wt% water mixed at 90°C. With increasing water content a transition can be visually observed from a clear solution, to a cloudy system, and ultimately to a gel. ..................................... 59 Figure 3.5. Effect of increasing water content from 10 to 27.5 wt% in a 10 wt% CA ternary system on G at an angular frequency of 1 rad/s at various temperatures. .......................... 60 Figure 3.6. Effect of solvent solubility parameters on the elastic modulus G for gels heated at various temperatures at 10 wt% CA at a fixed frequency of 1 rad/s ................................... 63 Figure 3.7. Stress sweep experiment for samples with 10 wt% CA and 15-20 wt% nonsolvent at different mixing temperatures. Experiments were performed at a fixed frequency of 1 Hz 64 Figure 3.8. Frequency dependence of the elastic and viscous moduli for 10 wt% CA/DMA/H2O mixtures, mixed at 50°C. G is represented by solid filled shapes while G is unfilled...... 66 Figure 3.9. Elastic modulus (obtained at a frequency of 1 rad/s) of 10 wt% CA/DMA/H2O gels as a function of water content .................................................................................................. 67 Figure 3.10. Elastic modulus versus % strain for a 10 wt% CA ternary system mixed at 50°C. The limit of linearity (percentage strain at the intersection between the tangent of the drop in G and the linear viscoelastic region) shifts to lower strain as nonsolvent concentration increases from 10 to 22 wt% ............................................................................................... 68 Figure 3.11. Onset point of nonlinearity of the elastic modulus for samples with 10 wt% CA mixed at 50, 70, and 90°C and 17.5 to 21 wt% water content ............................................ 69 Figure 3.12. LSM images of 10 wt% CA gels containing 20 wt% water heated at (a) 50°C, (b) 70°C, and (c) 90°C. Gels were tagged with calcofluor white (0.01 wt% of CA). Each image dimension is 127m x 127 m. ........................................................................................... 72 vii

Figure 3.13. Temperature sweep experiments conducted at 1°C/minute for 10 wt% CA and 20 wt% nonsolvent mixed at 50, 70, and 90°C. G is represented by solid filled shapes while G is unfilled ....................................................................................................................... 74 Figure 3.14. Temperature effect on steady shear viscosity for a 10 wt% CA ternary systems at 12.5-17.5 wt% water, dissolution method 2- addition of polymer to a water/DMA solution ............................................................................................................................................. 76 Figure 3.15. Visual observation of gel formation at (a) 50°C, (b) 70°C, and (c) 90°C with 10 wt% CA and increasing nonsolvent content from 12.5 to 22.5 wt% with 2.5 wt% increments .. 77 Figure 3.16. Elastic (G, filled shapes) and viscous (G, unfilled shapes) moduli of 10 wt% CA gels as a function of frequency mixed at 50°C. ................................................................... 79 Figure 3.17. Effect of increasing nonsolvent content in 10 wt% CA ternary system on the elastic modulus at 1 rad/s with varying temperature by dissolution method 2 ............................... 80 Figure 3.18. Temperature effect on stress sweep for 10 wt% CA ternary systems prepared with dissolution method 2............................................................................................................ 81 Figure 3.19. Stress sweep experiment for 10 wt% CA gels prepared at 50, 70 and 90oC as well as the corresponding gels reheated to 90oC at 20 wt% nonsolvent ......................................... 84 Figure 3.20. G values (obtained at 1 rad/s) for 10 wt% CA ternary systems heated at (top) 50°C, (bottom left) 70°C, and (bottom right) 90°C ....................................................................... 85 Figure 3.21. Stacked 1H NMR plots of (1A) 6-O-(4-methoxytriphenylmethyl)-cellulose, 6TC; (2A) 2,3-di-O-acetyl-6-O-(4-methoxytrityl)-cellulose, 2,3Ac6TC; and (3A) 2,3-di-Oacetylcellulose. The peak at 2.5 is the DMSO-d6 solvent. .................................................. 89 Figure 3.22. 13C NMR spectrum of 2,3-di-O-acetylcellulose (3A). The insert from 169 to 172 ppm shows the carbonyl C=O peaks arising from the acetyl groups. C3*: acetyl carbonyl group at C3 with a neighboring C6 hydroxyl ...................................................................... 90 Figure 3.23. Stacked 1H NMR plots of (1B) 6-O-triphenylmethylcellulose, 6TC II; (2B) 2,3-diO-acetyl-6-O-tritylcellulose, 2,3Ac6TC II; and (3B) 2,3-di-O-acetylcellulose II .............. 92 Figure 3.24. Stacked FTIR plots of (1) commercial cellulose acetate, (1B) 6-O-triphenylmethylcellulose (6TC II), and (2B) 2,3-di-O-acetyl-6-O-trityl-cellulose (2,3Ac6TC II) .............. 93 Figure 3.25. 13C NMR spectrum of propanoated commercial cellulose acetate (4) in CDCl3 at 75 MHz at 25°C. The acetyl and propanoyl C=O triplet signals are shown in the inset.......... 94 Figure 3.26. 13C NMR spectra of propanoated 2,3-di-O-acetylcellulose 4A (top) and 4B (bottom) ............................................................................................................................................. 96 Figure 3.27. Stress sweep experiment for 10 wt% regio2.4 CA samples with 15, 20, and 25 wt% water prepared at 50°C (frequency = 1 Hz). G is represented by solid filled shapes while G is unfilled ....................................................................................................................... 98 Figure 3.28. Elastic modulus as a function of % strain for 10 wt% regio2.4 CA gels prepared at 50°C (unfilled shapes) and 90°C (filled shapes) (frequency = 1 Hz) .................................. 99 Figure 3.29. Frequency sweep experiment for regio2.4 CA samples at 10 wt% CA prepared at 50°C. Elastic moduli are represented by filled shapes while viscous moduli are shown as unfilled shapes. .................................................................................................................. 100 viii

Figure 3.30. Stress sweep experiment for 10 wt% regio2.4 CA samples prepared at 50, 70 and 90°C and 15 wt% nonsolvent content (frequency = 1 Hz). G is represented by solid filled shapes while G is unfilled ................................................................................................ 101 Figure 3.31. Elastic modulus under applied stress for 10 wt% regio2.4 CA (filled shapes) and commercial CA (unfilled shapes) prepared by heating to 90°C ........................................ 103 Figure 3.32. Viscosity for 10 wt% regio2.4 CA and commercial CA gels at 20 wt% nonsolvent prepared at 50 and 90°C. Regio2.4 gel viscosity values are represented by solid shapes while commercial CA are the unfilled shapes ................................................................... 104 Figure 3.33. Elastic (G, filled shapes) and viscous (G, unfilled shapes) moduli for 10 wt% regio2.4 and commercial CA at 15 wt% (left) and 20 wt% (right) water contents (samples were prepared at 90°C) ...................................................................................................... 105 Figure 3.34. Frequency sweep experiment for 10 wt% regio2.8 CA gels prepared at 70°C with 10-17.5 wt% nonsolvent. G is represented by solid filled shapes while G is unfilled ... 106 Figure 3.35. Stress sweep experiment for 10 wt% cellulose acetate samples (commercial CA, regio2.4 CA, and regio2.8 CA) at 15 wt% nonsolvent prepared at 70°C. G is represented by solid filled shapes while G is unfilled ......................................................................... 108 Figure 3.36. Stress sweep experiment for 10 wt% cellulose acetate samples at approximately the same G (3700 Pa) after heating to 70°C. Regio2.4 CA contained 20 wt% nonsolvent, regio2.8 contained 15 wt% nonsolvent, and commercial CA contained 20 wt% nonsolvent ........................................................................................................................................... 109 Figure 3.37. Stress sweep experiment for 10 wt% cellulose acetate samples (commercial, regio2.4, and regio2.8 CA) at 20 wt% nonsolvent prepared at 70°C ................................ 110

ix

List of schemes

Scheme 2.1. Synthesis of 1A, 6-O-(4-methoxytriphenylmethyl)-cellulose (6TC) ....................... 46 Scheme 2.2. Synthesis of 2A, 2,3-di-O-acetyl-6-O-(4-methoxytrityl)-cellulose (2,3Ac6TC) ..... 47 Scheme 2.3. Propanoation of CA samples to determine the individual DS at the C2, C3, and C6 positions............................................................................................................................... 50

x

List of symbols and abbreviations

°C

degree Celsius

δd

dispersive solubility parameter

δh

hydrogen bonding solubility parameter

δp

permanent dipole-dipole solubility parameter

1

proton

H

13

carbon-13

Å

ångström

AGU

anhydroglucopyranose unit

CA

cellulose acetate

CDA

cellulose diacetate

CDCl3

deuterated chloroform

CED

cohesive energy density

CHCl3

chloroform

CMC

carboxymethylcellulose

CTA

cellulose triacetate

[Cu(NH3)4]OH2

cuprammonium hydroxide

DLS

dynamic light scattering

DMA

N,N-dimethylacetamide

DMF

N,N-dimethylformamide

DMI

1,3-dimethyl-2-imidazolidinone

DMSO

dimethylsulfoxide

DP

degree of polymerization

DS

degree of substitution

DSC

differential scanning calorimetry

FTIR

Fourier-transform infrared

G

elastic (storage) modulus

G

viscous (loss) modulus

gg

gauche-gauche

C

xi

GPC

gel permeation chromatography

gt

gauche-trans

H2O

water

HPLC

high-performance liquid chromatography

IR

infrared

KBr

potassium bromide

LiCl

lithium chloride

LSCM

laser scanning confocal microscopy

LSM

laser scanning microscopy

LST

liquid-solid transition

LVR

linear viscoelastic region

Mn

number average molar mass

Mol

moles

Mw

weight average molar mass

N2O4

dinitrogen tetroxide

NaOH

sodium hydroxide

NMF

N-methyl formamide

NMMO

N-methylmorpholine-N-oxide

NMR

nuclear magnetic resonance

P2O5

phosphorous pentoxide

Pa

pascal

rad

radian

Rg

radius of gyration

Rh

hydrodynamic radius

RID

refractive index detector

s

second

SEM

scanning electron microscopy

SLS

static light scattering

TBAF

tetrabutylammonium fluoride

TBDMS

tert-butyldimethylsilyl

Td

decomposition temperature xii

TDMS

thexyldimethylsilyl

TFA

trifluoroacetic acid

TGA

thermal gravimetric analysis

Tm

melting transition temperature

tg

trans-gauche

Tg

glass transition temperature

THF

tetrahydrofuran

wt%

weight percent

ZnSe

zinc selenide

xiii

Acknowledgements

I owe thanks to many people that have accompanied me throughout this journey. First and foremost, I thank my supervisor Dr. John F. Kadla for giving me the opportunity to work in his outstanding group and the immense knowledge that he imparted on me in the completion of this thesis. I would also like to acknowledge the helpful advice and constructive criticism from my committee members, Dr. Savvas Hatzikiriakos and Dr. Shawn Mansfield, as well as my nondepartmental examiner Dr. Frank Ko. I thank the Department of Wood Science faculty, staff, and fellow students for their friendship and advice. In particular, members (and former members) of the Advanced Biomaterials Group: Reza Korehei, Ana Filipa Xavier, Ian Dallmeyer, Wei Qin, Drs. Shinsuke Ifuku, Batia Bar-Nir, Yong-sik Kim, Jennifer Braun, and Scott Wasko for their help in the preparation of this thesis and sharing their areas of expertise with me. Honourable mention goes to members of the Forestry Products Biotechnology group who opened the door to research for me in the first place. My heartfelt thanks to my colleagues in the Department of Chemistry, in particular PhD students Emmanuel Castillo and Montse Rueda, for their advice and insightful discussions on organic synthesis. I also thank Mr. Kevin Hodgson at the UBC BioImaging Facility for great help and expertise in two-photon microscopy. I would like to express my sincerest gratitude and appreciation to my family for their constant encouragement, patience, and support in both good and difficult times. Last but not least, I owe a very special thank you to my mentors. Thank you for your guidance, advice, and dedication. I consider myself very fortunate to have met exceptional people like you who never cease to push me beyond my potential. To Drs. Renata Bura and Valdeir Arantes: thank you for believing in me.

xiv

Dedication

To my parents

xv

1

1.1

Introduction

Cellulose

Cellulose is the most abundant natural polymer on earth. It is the main constituent of plant cell walls and is also produced by algae, fungi, animal (tunicate), and bacteria (Acetobacter xylinum). Industrially, its main sources come from wood, cotton fibre, and cotton linters. Prior to its discovery, it was used by humans as an energy source, building material, and for clothing (Klemm et al. 2005). It has been estimated that between 1010 and 1012 tons of cellulose are biosynthesized globally each year (Klemm et al. 2002). Within the past century, cellulose has also been used as a chemical raw material in the production of cellulose nitrate (“gun cotton” - a low-order explosive, and celluloid a flexible thermoplastic film), cellulose xanthogenate (a precursor for viscose), and wood pulp (for production of paper and cardboard). Cellulose is a linear polymer comprised of D-anhydroglucopyranose units (AGU) linked by β-(1,4)-glycosidic bonds, as shown in Figure 1.1. In 1838 Anselme Payen derived the empirical formula for cellulose to be C6H10O5 (Payen 1838). It was believed that cellulose was made up of a few small molecules of glucose or cellobiose. Almost a century later in 1920, Herman Staudinger recognized the existence of covalent bonds linking the glucose units forming long molecular chains (Staudinger 1920). The cellulose chain consists of a D-glucose unit with a C4-OH on one end (non-reducing end) and a C1-OH on the other end which is in equilibrium with the aldehyde structure (reducing end). The 1,4-linkage in cellulose is trans (diequatorial) which is what leads to its high crystallinity, good strength and mechanical properties, decreased solubility, and increased stability to hydrolysis (Odian 1991). The anhydroglucose units exist in the thermodynamically favoured 4C1 conformation, and to accommodate the preferred bond angles of the acetal oxygen bridges, every second AGU ring is rotated 180° in the plane, hence making cellobiose the basic repeat unit (Rao et al. 1998). The glucosidic bond at the 4 position was determined by Haworth through degradation reactions of cellobiose (Haworth et al. 1927).

1

OH

OH OH

HO HO

O

4

O HO

6

O 5 2

3

OH

1

OH

HO O

OH cellobiose repeating unit non-reducing end group

OH O

O HO

OH

H

OH O

n reducing end group

Figure 1.1. Cellulose molecular structure

The molecular weight of native cellulose, defined by the degree of polymerization (DP), differs widely depending on the origin and method of isolation. Native cotton has a DP of around 13,000 (Marx-Figini 2007), while native wood cellulose has a DP of around 10,000 (Wilson & Hamilton 1986). After chemical treatment such as pulping, the DP can decrease to between 500 and 2,600. In 1913, by reacting cellulose with 15% NaOH and dimethyl sulphate, a trimethyl derivative was created (Denham & Woodhouse 1913), and its further acid hydrolysis (Irvine & Hirst 1922) proved the existence of two secondary and one primary hydroxyl group in each C6H10O5 residue in the positions 2, 3, and 6. The three hydroxyl groups in cellulose also differ in their reactivities (Yin & Brown 1959; Bochek & Kalyuzhnaya 2002). The hydroxyl at C2 is slightly more reactive than the one at C6 and four times more so than that at C3 (Lenz 1960). The reactivity of the respective hydroxyl groups is also affected by sterics, with the primary hydroxyl group at position 6 being the least sterically hindered (Hebeish & Guthrie 1981). The presence of hydroxyl groups and oxygen atoms on both the pyranose ring and the glycosidic bond make cellulose form an ordered hydrogen bonding system, which leads to semi-crystalline structures (Krässig 1993). Both intra-and intermolecular hydrogen bonding are present in cellulose. Intramolecular hydrogen bonds are responsible for the stiff and rigid nature, as well as the “two-fold screw axis” of the cellulose molecule (Figure 1.2). From IR, NMR spectroscopy, and X-ray diffraction studies (Gardner & Blackwell 1974), it is known that intramolecular hydrogen bonds are formed along both sides of the native cellulose chain. One exists between the C3-hydroxyl hydrogen of one AGU and the pyranose ring oxygen (O5) of an adjacent unit (O3-H---O5, bond length 2

2.75Å), and the other between the C2-hydroxyl hydrogen and the adjacent C6-hydroxyl oxygen (O2-H---O6, bond length 2.87Å) which is in a tg (trans-gauche) orientation (Marrinan & Mann 1954).

Figure 1.2. Hydrogen bonding system of cellulose I, where unfilled arrows are pointing to intramolecular hydrogen bonds and filled arrows point to intermolecular hydrogen bonds (Klemm et al. 2005). Copyright Wiley-VCH Verlag GmbH & Co. KGaA. Reproduced with permission.

Trans-gauche (tg) refers to the orientation of the C6-O6 bond with respect to the rest of the cellulose backbone. In the tg orientation, “trans” describes the torsion angle (O5-C5-C6-O6) and “gauche” indicates the torsion angle (C4-C5-C6-O6). That is the C6-O6 bond is trans to the C5-O5 bond and guache to the C5-C4 bond as illustrated in Figure 1.3. NMR studies using monosaccharides, oligosaccharides, and cellulose show three possible conformations of the exocyclic C5-C6 bond as shown in Figure 1.3: gauche-gauche (gg), gauche-trans (gt), and tg (Horii et al. 1983). The orientation of the C6-O6 bond influences the intermolecular hydrogen bonding in cellulose which is responsible for the sheet-like nature and uniform packing of the polymer. This occurs between the OH group at the C6 and the O3 of the neighbouring chain of cellulose molecules adjacently located in the same lattice plane (Gardner & Blackwell 1974).

3

Figure 1.3. Representation of the different orientations of the C6-O6 bonds: gauche-trans (gt), gauche-gauche (gg), and trans-gauche (tg) conformations

Cellulose chains aggregate to form fibrils, long thread-like bundles of molecules stabilized by hydrogen bonding between hydroxyl groups of adjacent molecules. The bundles arrange themselves in a regular pattern to form microfibrils that exhibit a crystalline X-ray diffraction pattern. The elementary fibril of native cellulose is considered the smallest morphological unit. Recent data indicates that its diameter can vary between 3 to 35 nm in diameter depending on the cellulose source (Fink et al. 1990). The order of cellulose aggregates is highly crystalline due to extensive hydrogen bonding, although crystallinity is not uniform throughout the macromolecule. The fringed fibril model (Hearle 1963) describes the cellulose crystal structure as two-phased, with both non-crystalline (low order) and crystalline (high order) regions. This is the most generally accepted structural description for cellulose and cellulose fibres. Four major polymorphs of cellulose have been reported: celluloses I, II, III, and IV. These polymorphisms refer to the existence of more than one crystalline form, differing in 4

physical and chemical properties. Cellulose I, also known as native cellulose, is the most predominant allomorph. In 1928, Meyer and Mark (Meyer & H. Mark 1928) proposed a unit cell of the crystal lattice, assuming a monoclinic unit cell and two anti-parallel cellobiose chain segments running in opposite directions along the fibre axis. The unit cell dimensions are illustrated in Figure 1.4.

Figure 1.4. Cellulose I unit cell model according to Meyer and Misch (Meyer & Misch 1937). Copyright Wiley-VCH Verlag GmbH & Co. KGaA. Reproduced with permission.

Numerous authors have suggested that the native cellulose unit cell may vary depending on the source. In 1974, Gardner and Blackwell (Gardner & Blackwell 1974) proposed a different unit cell structure based on more advanced X-ray techniques. Using cellulose I from algae (Valonia), they proposed a monoclinic lattice with two parallel running chains. The same year, Sarko and Muggli (Sarko & Muggli 1974) working with the same algae found that using minimal energy considerations cellulose I favored a parallel arrangement of cellulose molecules, and that cellulose II an anti-parallel orientation. A decade later, it was shown using solid state NMR (Atalla & VanderHart 1984) and electron diffraction (Sugiyama et al. 1991) that native celluloses were actually a combination of two sub-allomorphs: cellulose Iα and cellulose Iβ. Cellulose Iα is the primary polymorph in bacterial and algal celluloses, while cellulose Iβ is 5

predominant in higher plants and animal sources. Native cellulose has a parallel chain orientation (Hieta et al. 1984; Nishiyama et al. 2008), with cellulose Iα being triclinic with one cellulose chain per unit cell and cellulose Iβ being monoclinic with two parallel chains per unit cell with dimensions similar to that originally proposed by Meyer-Misch. Unit cell dimensions of the cellulose allomorphs are shown in Table 1.1.

Table 1.1. Unit cell dimensions of various cellulose polymorphs (Krässig 1993; Claffey & Blackwell 1976; Sugiyama et al. 1991) Polymorph

a-axis (Å)

b-axis (Å)

c-axis (Å)

γ-axis (Å)

Cellulose I

6.74

5.93

10.36

81

Cellulose I (algal and bacterial)

8.01

8.17

10.36

97.3

Cellulose I (higher plants and animal)

7.85

8.17

10.34

96.4

Cellulose II

9.08

7.92

10.34

117.3

Cellulose III

9.9

7.74

10.3

122

Cellulose IV

7.9

8.11

10.3

90

Cellulose II is also referred to as regenerated or mercerised cellulose depending on the mechanism of formation. Mercerised cellulose is produced by mercerisation (most commonly with sodium hydroxide) with subsequent removal of the alkali. Regenerated cellulose on the other hand is typically produced by derivatizing cellulose to enable processing, followed by deprotection or regeneration to cellulose. The most common example is viscose rayon fibers, which are produced by extrusion of cellulose xanthane in sodium hydroxide followed by regeneration in an acid coagulation bath. Because this cellulose is reprecipitated, it is also called regenerated cellulose. Transformation from cellulose I to cellulose II is irreversible, which implies that cellulose II is a more stable form; i.e. cellulose I is metastable (Sarko & Muggli 1974). Structurally, cellulose II differs significantly from that of cellulose I. Although a monoclinic unit cell, the cellulose chains are believed to be antiparallel in orientation (Sarko & Muggli 1974; Kolpak & Blackwell 1976). In addition, the conformation of the hydroxymethyl 6

group at the C6 position differs between the corner and centre chains resulting in a larger unit cell with subtle differences in hydrogen bonding as shown in Figure 1.5. The origin (corner) chains are in a gt conformation while the center chains are in a tg conformation (Langan et al. 1999). This leads to the formation of not only interchain (present in cellulose I) but also interplane hydrogen bonds. From a thermodynamic point of view, gt is the more stable conformation and once the polymer adopts this conformation it cannot revert back to cellulose I. This is the most important crystalline form from a commercial and technical point of view, as it serves as a precursor for cellulose fibres.

Figure 1.5. Hydrogen bonding in cellulose II. Only atoms involved in hydrogen bonding (represented by dotted lines) are labeled. Top left: intermolecular hydrogen bonds are O6-D---O2 in sheets containing only center molecules. Top right: intermolecular hydrogen bonds are O2-D--O6 in sheets containing only origin molecules. Bottom: sheets containing origin and center molecules contain O6-D---O6 and O2-D---O2 intermolecular hydrogen bonds. In the former case O5 and O3 can also act as acceptors. Intramolecular hydrogen bonds are O3-D---O5 in each molecule with a minor component involving O6 as acceptor (Langan et al. 1999). Reproduced with permission from the American Chemical Society 7

The two other polymorphs, cellulose III and cellulose IV are of less significance and have not been as widely studied. Cellulose III is obtained by treating either cellulose I (IIII) or cellulose II (IIIII) with liquid ammonia at low temperature (below -30°C) and high pressure and subsequent recrystallization by evaporation of the ammonia under high temperature and washing with alcohol. By using liquid ammonia instead of alkaline solution the smoothness and strength of the corresponding cellulosic fibres is improved for use in the textile industry (Yatsu et al. 1986). However, this process has not gained wide commercial interest due to high production costs. Cellulose IV is obtained by treating cellulose I, II, or III in a suitable liquid at high temperature under tension and is generally associated with the production of high performance rayon. Figure 1.6 summarizes the chemical and physical processes by which native cellulose is converted to other polymorphs.

Figure 1.6. Transformation of cellulose into its various polymorphs (Klemm et al. 2005)

In addition to dictating crystalline structure, hydrogen bonding affects other physical properties of cellulose such as solubility. Cellulose is generally not soluble in most common solvents including water because of the extensive intra- and intermolecular hydrogen bonding in the anhydroglucopyranose chains. The interchain hydrogen bonding in the crystalline regions is strong, giving the resultant cellulose fibres good strength. They also prevent cellulose from melting (i.e., non-thermoplastic), and therefore has a decomposition temperature (Td) less than its

8

glass transition temperature (Tg). In other words, due to its extensive network of hydrogen bonds, cellulose decomposes prior to melting. To facilitate industrial processing, cellulose is usually regenerated or derivatized because of the difficulty in selecting appropriate solvents for homogeneous phase reactions. The most common cellulose derivatives are cellulose esters and ethers.

1.2

Cellulose modification

As mentioned previously, cellulose contains many intra- and intermolecular hydrogen bonds which are responsible for its high degree of crystallinity, rigidity, and poor solubility in polar solvents. A semi-crystalline polymer, it decomposes prior to melting because of these hydrogen bonds. Despite these challenges, it has very consistent chemical and physical properties, which if chemically modified, can be made into very useful materials. Commercial cellulose modification proceeds in the solid or swollen state (heterogeneous reactions). The accessibility of the hydroxyl groups depends on breaking the hydrogen bonds through sequential activation steps, as well as interactions with the reaction media. The dominating method for the production of regenerated cellulose is the viscose process. This method involves pure cellulose, typically dissolving pulp being converted with CS2 into cellulose xanthogenate as a metastable intermediate. As this intermediate is soluble in aqueous sodium hydroxide, it can be formed as a viscose solution in a wet spinning process. Coagulation of this viscose dope into an acid bath cleaves the xanthate groups and a high purity cellulose is regenerated. Viscose fibers, also known as rayon, are predominantly used for film (cellophane) and textile applications. A more direct regeneration process without derivatization can be made through copper ammonia technology. Here, regenerated cellulose filaments are produced by spinning a cellulose solution from wood pulp in a mixture of copper (II) hydroxide in aqueous ammonia (cuprammonium hydroxide) [Cu(NH3)4]OH2. This process is rarely used as it poses environmental hazards. A more environmentally friendly process, the Lyocell process, consists of regenerating cellulose from a solution of N-methylmorpholine-N-oxide (NMMO) monohydrate by solution spinning (Klemm et al. 2005). Once the regenerated fibers are 9

produced, almost all of the solvent employed is recovered. Figure 1.7 shows two processing schemes for cellulose regeneration: the viscose process and Lyocell process.

Figure 1.7. Cellulose regeneration processes involving a cellulose derivative (Viscose technology) and direct dissolution (Lyocell technology) (Klemm et al. 2005)

Non-commercial methods of cellulose modification for production of more functionalized cellulose derivatives can proceed through homogeneous media. One important solvent system is N,N-dimethylacetamide (DMA) / lithium chloride (LiCl) which is commonly used in organic synthesis and for analytical purposes (Dupont 2003). Another effective solvent system is that of tetrabutylammonium fluoride trihydrate in DMSO (DMSO/TBAF) (Östlund et al. 2009). An advantage with using the DMSO/TBAF system is that it can dissolve cellulose with very high DP without pretreatment. There has been recent development in the use of ionic liquids for cellulose dissolution (Heinze & Koschella 2005). Graenacher was the first to propose N-ethylpyridinium chloride in 10

the presence of a nitrogen-containing base as a method to dissolve cellulose in 1934 (Graenacher 1934). Ionic liquids are now becoming a viable, non-derivatizing way to dissolve cellulose. 1butyl-3-methylimidazolium chloride for example (Swatloski et al. 2002) can dissolve high molecular weight cellulose without activation or pretreatment. Microwave heating can also accelerate the dissolution process, and the regeneration process consists of simply precipitating the polymer in water, ethanol or acetone (Zhu et al. 2006). The regenerated cellulose contains almost the same degree of polymerization and polydispersity as the initial material, but its morphology is significantly changed as it forms fused microfibrils (H. Zhang et al. 2005). Ionic liquids present an advantage over other conventional solvent systems as they are more environmentally friendly and can be recycled after the regeneration process.

1.3

Cellulose derivatives

One of the main reasons driving the interest in developing new cellulosics is because cellulose is an abundant renewable resource. It is also biocompatible, hydrophilic, and has the ability to form superstructures (helix formation, cholesteric mesophases, Langmuir-Blodgett layers, etc) (Heinze & Liebert 2001). Because of their versatility cellulose derivatives are used in a wide variety of applications in the food, pharmaceutical, and cosmetic industries. Cellulose acetates, for example, have been used as membranes for reverse osmosis leading to the commercial installation of desalinization plants around the world (Committee on Polymer Science and Engineering et al. 1994). Both cellulose ethers and esters are of commercial significance. Table 1.2 shows the most common cellulose esters and ethers along with yearly production numbers and uses. Common cellulose ethers include methyl cellulose, hydroxyethyl cellulose, and carboxymethyl cellulose (CMC), the latter being the most widely produced. CMC is used as a food thickener and additive, as well as in detergents, paints, and toothpaste due to its high viscosity, nontoxicity, and biocompatibility (Klemm et al. 2005). The most important industrial cellulose esters include cellulose acetate, cellulose acetate propionate, and cellulose acetate butyrate. These are commonly used in coatings and other thermoplastic applications, but have found their way to 11

controlled-release systems in the pharmaceutical industry and optical media in the development of liquid crystalline displays.

Table 1.2. Examples of cellulose esters (bold) and ethers of commercial importance (Balser et al. 2004; Thielking & Schmidt 2006) Product

Worldwide

Functional

Application

production (tons/year) group Cellulose acetate

900,000

-COCH3

Textiles, cigarette filter tow, coatings

Cellulose nitrate

150,000

-NO2

Lacquers, low-order explosives

Cellulose xanthate

4,000,000

-C(S)SNa

Textiles (precursor to Rayon)

Carboxymethyl cellulose

230,000

-CH2COONa

Food thickener and additive

Methyl cellulose

120,000

-CH3

Thickening agent, drug delivery

Ethyl cellulose

4,000

-CH2CH3

Food additive (emulsifier)

Hydroxyethyl cellulose

60,000

-CH2CH2OH

Thickening agent

In the manufacture of cellulose esters, the reaction of cellulose typically involves the use of an organic or inorganic acid and a strong acid catalyst (Balser et al. 2004). On the other hand, the production of cellulose ethers involves reactions under alkaline conditions, with the final product usually being water soluble (Klemm et al. 2005). In both cellulose esters and ethers, the chemical substitution reactions occur at the C2, C3, and C6 hydroxyl groups. Substitution at these groups can be random or selective depending on the synthetic pathway. Esterification and etherification studies have shown that the C6 hydroxyl group is esterified ten times faster than the C2 or C3 groups, whereas etherification proceeds twice as fast at the C2 and C6 group as compared to the C3 hydroxyl group. The primary alcoholic group at C6 is distinguished from the

12

other two secondary alcohol groups in that it has an axis of free rotation around the C5-C6 bond (see Figure 1.2), which is, however, somewhat restricted by hydrogen bonding (Hebeish & Guthrie 1981). This leads to the derivatized cellulose having varying degrees of substitution, which influences its behaviour in solution and end-use. The varying extent of derivatization is due to the fact that the three hydroxyl groups have different acidity and are not equally accessible to reagents, due to their location in different ordered regions within the polymer (Tasker et al. 1994).

1.4

Cellulose acetate and its properties

Cellulose acetate is a general term that refers to cellulose that has been substituted by acetyl groups. The most common commercial cellulose acetate is so-called 2.5 cellulose acetate (secondary CA), in which an average of 2.5 hydroxyl groups of the AGU are acetylated. The term degree of substitution (DS) is used to describe the extent of substitution (to a maximum of 3) of any functional group to the hydroxyl groups of the AGU (see Figure 1.8). Cellulose diacetate (CDA) refers to acetates containing an acetyl DS between 2 to 2.5, whereas cellulose triacetate (CTA) can be used for cellulose acetates with a DS greater than 2.7.

Figure 1.8. Structural representation of cellulose acetate

Cellulose acetate is one of the most widely utilized and studied cellulose derivatives. CA was first proposed as a suitable material for photographic film in 1901, having previously been 13

made out of celluloid (cellulose nitrate). Throughout the 20th century, cellulose diacetate has also found its way into textiles, where it could be spun into a silk-like yarn known as “Celanese.” Soon after the 1970s, however, man-made fibres such as nylon and polyester started to replace CA silk on the textile market and presently there is declining demand for the material. By contrast the market for cellulose acetate filter tow continues to grow because of (i) the specific retention behaviour of CA, which gives cigarettes an agreeable “taste signature,” (ii) a high speed technology for rod making exists, and (iii) there are a large variety of possible filter characteristics. Cellulose tri- and diacetates mixed with plasticisers produce transparent thermoplastics with excellent clarity and good mechanical properties. They are non-yellowing, resistant to embrittlement, and provide good electrical insulation. Through injection moulding or extrusion, plasticised CA’s can be transformed into products like tool handles, combs, toys, and spectacle frames. CA’s have also been used as coating for airplanes and other surfaces (Rustemeyer 2004). There is also a growing demand for CA films as one of the layered components in liquid crystal displays (LCD) (Sata et al. 2004). CA films have good clarity and ease of processing. They are inert towards polarized light, and thus serve the dual purpose of protecting the polarizing plate and provide optical compensation with controlled birefringence to enhance the viewing angle of the flat layer screens. Cellulose acetate has recently been the polymer of choice for reverse osmosis membrane preparation (Tohru Shibata 2004). This application is based on the balance of hydrophilicity (hydroxyl groups) and hydrophobicity (acetyl groups) of the polymer for membrane flux and rejection, with the degree of substitution having a profound effect on selectivity. The basic process for membrane formation is through gelation of a polymer solution by phase separation and subsequent coagulation of the primary particles by solvent evaporation or contact with a nonsolvent (coagulant). The molecular structure of CA can affect chain stiffness, macromolecular shape, and interactions leading to aggregation and association which in turn influence flow properties. These are important factors to take into account for the intended applications of cellulose acetate materials. The extent and regiochemistry of acetylation can have a dramatic effect on the CA properties.

14

1.4.1 Synthesis of cellulose acetates

Cellulose acetate (CA) was first synthesized by Schützenberger in 1865 (Schützenberger 1865) by heating cellulose in a sealed glass tube with acetic anhydride. Presently, the most common industrial production method is the “acetic acid process” (Steinmeier 2004). In this process cellulose is activated in an acetic acid solution with sulphuric acid as a catalyst. Acetic anhydride is then added to form cellulose triacetate followed by subsequent hydrolysis where several acetyl groups are removed to achieve an average degree of substitution of 2.5. In the first step of this process a cellulose sulphate ester intermediate is formed which subsequently undergoes transesterification in the presence of acetic anhydride. The presence of residual hydroxyl groups means that the polymer retains a degree of moisture sensitivity, which strongly influences its application properties (Law et al. 2004). Another industrial process that was used in the early twentieth century was the methylene chloride process, also known as the Dormagen process developed by Bayer. It consists of cellulose activation with perchloric and sulfuric acid catalyst followed by triacetate hydrolysis with methylene chloride instead of acetic acid (Gedon & Fengi 2000). This “fibrous acetylation process” has also been reported using other inert diluents such as toluene, benzene, or hexane which maintain the fibrous structure of cellulose throughout the reaction. Again perchloric acid is the catalyst of choice as it does not react with cellulose to form acid esters and virtually complete acetylation (DS 3.0) can be achieved upon addition of acetic anhydride (Gedon & Fengi 2000). Zinc chloride has also been used as a catalyst in the production of cellulose acetate; however, large quantities are required, and thus this method has not been economical for commercialization. Cellulose can be acetylated at the C2, C3, and/or C6 positions (Figure 1.8). The degree of acetylation can be controlled by temperature, catalyst concentration, water content, and solvent. Complete cellulose dissolution is ideal before acetylation as the hydroxyl groups are more accessible in the homogeneous state (Steinmeier 2004). Several solvent systems are known to dissolve cellulose. However, most of these solvents, although effective at dissolution, have not achieved industrial prominence due to high costs. Thus, they have mostly been used in

15

laboratory-scale synthesis, where low concentrations of cellulose are used and solvent recovery is not critical. The cellulose activation step prior to acetylation in the acetic acid process is crucial to the formation of cellulose triacetate. Since most of the acetylation occurs in the heterogeneous state, treating cellulose with water or other swelling agents such as sodium hydroxide or liquid ammonia can lead to an increase in accessibility of the hydroxyl groups for reaction with acetic anhydride. An acid catalyst is also required for complete acetylation. High catalyst concentrations make cellulose more soluble and increase the acetylation rate. However, it also promotes faster degradation, through acidic hydrolysis of the glycosidic bonds. Acetylation rates can also be controlled with temperature, where high temperature improves swelling and increases acetylation rates, but also increases the degradation processes (Steinmeier 2004). During esterification, the hydroxyl group is substituted by an acetyl group. The esterification process is in equilibrium with the reverse reaction, saponification; the equilibrium is strongly dictated by the presence of water. The acetylation reaction comes to an end when the mixture becomes a viscous liquid. Typically, viscosity measurements are done to check the degree of polymerization and reaction completeness. Stopping the reaction consists of the addition of water or dilute acetic acid to destroy the excess anhydride present. Subsequent hydrolysis takes place with the remaining acid catalyst and water. The acid catalyst first protonates the acetyl carbonyl group, making the carbonyl carbon more electrophilic. The reaction proceeds through nucleophilic attack by water with subsequent deprotonation and formation of acetic acid. Higher water content in the hydrolysis step leads to an increase in primary hydroxyl groups, indicating that the C6 acetyl is hydrolyzed faster than the C2 and C3 in the presence of acid (Malm et al. 1950); the enhanced hydrolysis is due to steric hindrance. By contrast base-catalyzed hydrolysis preferentially deacetylates the C2 and C3 positions. It is typically performed in a ternary system consisting of DMSO, water, and a suitable aliphatic amine like dimethylamine, n-hexylamine, or hexamethylenediamine (Deus et al. 1991; Wagenknecht 1996). Under these conditions it is the pKa of the conjugate acid that influences the order of hydrolysis, i.e. pKa of C2 < C3 C2 > C3 during acetylation with pyridine in a DMA/LiCl or DMI/LiCl solvent system (Takaragi et al. 1999). One approach leading to partially substituted cellulose acetate with controlled stereochemistry is through protecting group chemistry. Using a non-derivatizing homogeneous solvent system such as DMA/LiCl, insertion of a trimethylphenyl (or trityl) group preferentially reacts at the C6 for steric reasons (Camacho Gómez et al. 1996), and cellulose can be further acetylated at the C2 and C3 positions. Deprotection can proceed through strong acid such as hydrobromic acid in acetic acid or molten salt hydrates (Fischer et al. 2001). Silane protecting groups have also been used, in particular bulky moieties such as tert-butyldimethylsilyl17

(TBDMS) and thexyldimethylsilyl- (TDMS) functional groups (Klemm et al. 1997) to selectively protect the C6 position in the case of the former and C2 and C6 in the latter. Heterogeneous reactions are also common for cellulose functionalization. A very efficient method is the “impeller” method (Morooka et al. 1984). Trifluoroacetic anhydride is usually used as the impeller and mixed anhydrides are added to the acylation reaction. Regioselective cellulose esters can also be produced this way using the trityl protecting group (Iwata et al. 1997). Techniques such as 13C NMR spectroscopy can be used for determining the distribution of substituents in the cellulose derivatives (Miyamoto et al. 1984; Iwata et al. 1992; Tezuka & Tsuchiya 1995). Functional groups introduced within the cellulose AGU are assigned a unique chemical shift depending on which carbon the substitution takes place. For example, cellulose propanoation leads to 13C NMR spectra containing three propanoyl peaks at the C2 (128 ppm), C3 (173 ppm), and C6 (174 ppm) positions (Tezuka & Tsuchiya 1995). Proton NMR spectroscopy can also be used to characterize cellulose acetates through treatment with acetyl-d3 chloride or propionic anhydride to produce an acetate with more structural regularity (i.e. complete ester functionalization as opposed to hydroxyl groups) to determine the distribution of acetyl groups within the AGU (Goodlett et al. 1971). Another convenient way for structural elucidation of CA is infrared spectroscopy (Heinze & Liebert 2004). The signals of the carbonyl groups at 1740-1750 cm-1 and hydroxyl groups at 3300-3700 cm-1 can provide useful information on the distribution of the acetyl groups within the AGU. Two signals at 1752 and 1740 cm-1 correspond to the acyl moieties at the C2/C3 and C6 positions respectively. The signal at 3660 cm-1 corresponds to the hydroxyl group at the C6 (primary OH unit), while signals at 3520 and 3460 cm-1 have been ascribed to secondary hydroxyl groups. The absorption band at 3580 cm-1 can be attributed to hydrogen bonding of the primary hydroxyl group (Heinze et al. 2006). Infrared spectroscopy cannot distinguish the absorption bands of the C2 and C3 hydroxyl and acetyl groups, and line shape analysis or deconvolution of the spectra is often necessary to obtain quantitative information. Therefore, this technique is often complementary to NMR analysis for a more thorough cellulose acetate characterization. The thermal behaviour of cellulose acetate is also influenced by the DS, regiochemistry, and method of preparation (Iwata et al. 1997). Cellulose acetate has a high melting temperature 18

but low melting entropy, suggesting that cellulose ester melts exist as semi-flexible extended chains. The glass transition temperature Tg for cellulose triacetate (DS > 2.7) has been determined to be 172°C, the melting temperature Tm to be 307°C, and the decomposition temperature Td to be 356 °C. Cellulose acetates with different degrees of acetylation show Tg values between 190 and 220°C. X-ray analysis has shown that the melting temperature and crystallinity drops for secondary CA as compared to cellulose triacetate due to smaller and less perfect crystallites. Kamide and Saito (Kamide & Saito 1985) performed an extensive study on the thermal properties of cellulose acetates as a function of DS (0.49, 1.75, 2.46, and 2.92). The results determined by differential scanning calorimetry (DSC) are shown in Figure 1.9.

Figure 1.9. Dependence of glass transition temperature, melting temperature, and decomposition temperature on the cellulose acetate degree of substitution (Kamide & Saito 1985)

Figure 1.9 shows a linear dependence of the glass transition temperature (Tg) decreasing with an increase in the average DS. The melting temperature (Tm) is more difficult to describe. CA with 0.49 DS shows no melting transition, and with increasing DS it undergoes a Tm minimum at a DS ~ 2.5. The decomposition temperature (Td) increases slightly with DS (about 8%). The DS dependence of Td suggests that molecular interactions of the chains influence thermal degradation. 19

1.4.2 Dissolution of cellulose acetates

Cellulose acetates with varying DS are obtained by either controlled acetylation of cellulose or deacetylation of CTA. Commercially the latter process is typically used. Hydrolysis of CTA using acetic acid with catalytic sulphuric acid produces a statistical deacetylation in which all of the acetyl groups are relatively accessible. The C6 acetyl position hydrolyzes faster than that of the C2 and C3 due to steric reasons. Regioselective deacetylation with amines in DMSO and water has preferential hydrolysis of acetyl groups at the C2 and C3 positions because of their lower hydroxyl pKa (Miyamoto et al. 1985). Depending on the degree of substitution (Elöd & Schrodt 1931) and regiochemistry (Philipp et al. 1996), cellulose acetates are soluble in many organic solvents (Deus et al. 1991). As with most polymers, cellulose acetate solubility is dependent on the hydrogen bonding component (δh) of the solubility parameter index for the solvent and cellulose actetate (Hoernschemeyer 1974). Cellulose acetates with a degree of substitution (DS) between 0.5 and 1 are soluble in aqueous solutions (Bochek & Kalyuzhnaya 2002), while those of DS greater than 1 tend to be insoluble in aqueous solutions but soluble in various organic solvent systems. Table 1.3 provides a brief summary for a few commonly used CA solvents and the corresponding CA DS range (Philipp & Klemm 1995). Other organic solvents that have found use in the dissolution of cellulose acetate (DS solubility range in parenthesis) include tetrahydrofuran (2.0-3.0), dioxane (2-2.5), N,N-dimethylformamide (2-3.0), and N,N-dimethylacetamide (0.47-3.0) (J. Mark 2009). The latter solvent, also known by its abbreviation DMA, is usually used for dissolving CA with a high DS insoluble in aqueous solution.

20

Table 1.3. Effect of DS on solubility with CA having different substitution patterns (Deus et al. 1991) DS range of solubility for cellulose acetate Solvent

Statistical deacetylation at the

Regioselective deacetylation

2, 3, and 6 positions

at the 2, 3 positions

Water

0.8-1.0

Insoluble

DMF

1.8-2.7

1.3-2.8

Insoluble

Insoluble

Acetone (1% water)

2.1-2.6

2.5-2.6

Acetone/chloroform (1:1 v/v)

2.3-2.6

Insoluble

Pyridine

0.8-2.7

1.2-2.8

Pyridine/water (1:1 v/v)

0.6-2.0

1.2-1.6

Pyridine/acetone (1:1 v/v)

1.3-2.7

1.5-2.8

Ethyl acetate

1.6-2.7

2.1-2.8

Acetone ( Rh at low molar mass as expected for linear chains, while Rg < Rh as molar mass increases and particles aggregate to form architectures resembling microgels (Burchard 2003). These structures have been described as worm-like chains.

24

Figure 1.11. Relationship between radius of gyration Rg and hydrodynamic radius Rh in solution. Rg is the same for both particles (Schulz et al. 2000). Copyright Wiley-VCH Verlag GmbH & Co. KGaA. Reproduced with permission.

The nature of the CA solvent influences the interactions between substituents in the polymer. Basic solvents such as acetone and DMF mainly solvate the hydroxyl groups while acidic solvents like formamide and NMF (N-methyl formamide) solvate the acetyl groups of CDA. Basic solvents form dipole interactions with the polymer while acidic solvents autoassociate through hydrogen bonding (Pintaric et al. 2000). These hydrogen bonding interactions lead to intra- and intermolecular bridging between CDA molecules making the polymer particularly rigid. Maximum rigidity is achieved when the DS is approximately 2.5, when intersegmental hydrogen bonding is largest (Kamide & Saito 1984). Solution properties of cellulose acetate differ with the degree of substitution as well as the distribution of substituents in the AGU. Acetyl substitution in cellulose reduces its efficiency of chain packing and the amount of intermolecular H-bonding, thus increasing chain flexibility and mobility. Compared to cellulose itself, the hydroxyl groups lead to strong hydrogen bonding which lead to the formation of crystalline regions that make the polymer uniformly packed and unable to dissolve in common solvents. Acetyl groups in the C2 and C3 positions can form hydrogen bonds in DMA solution, but play no role in cluster formation (Tsunashima & Hattori 2000). Chain clustering occurs in solution due to intermolecular hydrogen bonds. Thus, hydroxyl 25

groups at the C6 position are effective in forming intermolecular associations in polar solvents (Tsunashima et al. 2001). The C6 position acetyl groups can also hydrogen bond intramolecularly with a hydroxyl in the C2 position, making the chain stiffer (Kawanishi et al. 1998). Changes in temperature also affect CA in dilute solution. In DMA, CA showed lowtemperature solubility, dissolving below -12°C and phase separating above 65°C. Three types of structures exist: one is a single CA chain, and the two others are dynamic self-assemblies which are formed temporarily by solvent-mediated hydrogen bonding. These self-assemblies arise as structures reorganize with temperature and the competition between hydrogen bonding and hydrophobic interactions of the polymer (Tsunashima et al. 2002). Cellulose acetate in high concentrations behaves differently than that in dilute solution. The polymer tends to aggregate and form clusters. Flow behaviour is affected by molecular parameters such as molar mass, molar mass distribution, substitution pattern, and substituent distribution. Structural parameters such as molecular conformation, interactions, and supramolecular structures of the polymer also play a role in its behaviour in solution (Clasen & Kulicke 2001). Both poor solvents and good solvents can increase or decrease the specific viscosity respectively in concentrated solution (Isono & Nagasawa 1980). Intermolecular association between the polymer and the solvent at high enough concentrations can lead to thermoreversible gelation as observed in the CA/benzyl alcohol system (Goebel & Berry 1977).

1.5

Cellulose and cellulose derivative gels

Gels are defined as systems which exhibit an elastic rheological response. That is, a gel should not flow (or relax) under the action of a mechanical stress imposed for an infinite period of time (Raghavan & Cipriano 2006). This implies the existence of a yield stress below which there is no steady state flow (Ferry 1980). At high (nonlinear) stress or strain, however, the system may flow. A polymer gel is a cross-linked solution linked through either chemical or physical bonds. The former are known as chemical gels while the latter are physical gels. Cellulose and cellulose derivatives form physical gels consisting of three-dimensional polymeric 26

networks in which the polysaccharide aggregates through interactions such as hydrogen bonding and hydrophobic association. Phase separation is a means to physical gel formation in polymeric systems (Khalil 1973). The polymer forms a cross-linked network through physical interactions such as ionic bonding, hydrogen bonding, dipole-dipole interactions, Coulombic forces, or chain entanglements of polymer-polymer interactions (Ross-Murphy 1992); there are no covalent bonds breaking or forming. For example, a phase separated liquid crystalline cellulose solution in ammonia/ammonium thiocyanate solvent can form thermoreversible gels at temperatures below 30°C (Frey et al. 1996). Although the gels do not possess any crystalline or covalently bonded cross-links, the rheological properties of the system at gel point had similar values as those with crystalline or covalent cross-links. Hydrogen bonding in the liquid crystalline regions formed the junction zones leading to gel formation. Phase separation has long been used as a technique for manufacturing permeable polymer membranes, and cellulose acetate provides a good template for sophisticated network formation. Scanning electron microscopy (SEM) imaging shows porous networks of void spaces during casting of membranes made using CA/acetone and CA/THF in a water coagulation bath (Reuvers et al. 1986). Laser scanning confocal microscopy (LSCM) is another technique that has been used as a tool to determine gel morphology and mechanism of gelation (Butler & Heppenstall-Butler 2003; Hans Tromp et al. 2001). Cellulose acetate/DMA/alcohol systems showed various degrees of pore size distribution depending on the alcohol alkyl chain length and number of hydroxyl (hydrogen bonding) groups of the alcohol nonsolvent (Kadla & Korehei 2010). Microscopy images give evidence of microstructure formation, and provide insight on the mechanism of gelation. Phase separation induced gelation via ternary system has been applied to systems consisting of cellulose acetate-acetone-nonsolvent in the production of wet phase inversion membranes (Hao & Shichang Wang 2001). The physical properties of the nonsolvent affected the morphology of the membrane. Water was a much stronger coagulant than methanol, ethanol, or isopropanol, and as a consequence showed large macrovoids compared to those quenched with alcohols which formed a honeycomb-like membrane structure. Calculations based on the Flory-Huggins equation demonstrated different phase diagrams for the different nonsolvents used, indicating membrane formation via different mechanisms. 27

The liquid-liquid phase separation process consists of a solution lowering its free enthalpy by separating into two liquid equilibrium phases. There are two ways in which this demixing occurs; one is by nucleation and growth of the second phase, and the other is by instantaneous spinodal decomposition. The spinodal gap is surrounded by the area where nucleation and growth phase separation takes place, and since the latter phenomenon is a relatively fast process, spinodal demixing is not very probable in polymer solutions (Wijmans et al. 1983).

Figure 1.12. Schematic phase diagram showing different phase separation mechanisms (Tsunashima et al. 2002)

Figure 1.12 shows different regions of phase separation. The two main regions are the homogeneous region, in which different components coexist in one phase, and the demixed region, in which phase separation occurs. The line separating these two regions is known as the binodal line. The phase separated region is further sub-divided into two sections separated by the spinodal line. Both the spinodal and binodal lines are convex downward, and the solution is stable below the binodal line. The nucleation and growth process occurs in the metastable region while spinodal decomposition takes place to the unstable region. Initial morphologies for gels formed via nucleation and growth result in a random polydisperse array of droplets with sharp 28

interfaces. Spinodal decomposition, on the other hand, shows either a droplet or bicontinuous morphology as large concentration fluctuations lead to the spontaneous formation of two separate regions as shown in Figure 1.13. The cluster size for the phase-separated nucleation and growth process is larger than that for spinodal decomposition. Phase separation is dependent on temperature, which dictates the structure and chain organizations of the polymer in solution (Tsunashima et al. 2002).

Figure 1.13. Examples of phase separation achieved via (left) nucleation and growth characterized by droplets and (b) spinodal decomposition characterized by bicontinuous morphology (Butler & Heppenstall-Butler 2003). Reprinted with permission from Elsevier

Cellulose aggregates form in dilute solution, and increasing temperature increases interand intrachain hydrophobic interactions leading to aggregate formation. In this way, cellulose hydrogels have been developed using microcrystalline cellulose in aqueous sodium hydroxide solution (Roy et al. 2003). With increasing polymer concentration, increasing temperature led to higher elastic modulus and faster gelation. Hydrogels can also be prepared by mixing cellulose with NaOH/urea aqueous solution. Gelation occurred depending on the polymer molecular weight, concentration, temperature, and time. The uniqueness of this system is that gelation is irreversible as evidenced by changing viscous (G) and elastic (G) modulus during heatingcooling-heating processes (Cai & Zhang 2006). Methylcellulose has been shown to form thermoreversible gels in aqueous solution upon heating. Thermal gelation of methylcellulose (DS of 1.2 to 2.2) take place because of hydrophobic associations involving the hydrophobic methyl substituents that become 29

pronounced at elevated temperatures (Hirrien et al. 1998). In addition to the hydrophobic interactions there is also intermolecular hydrogen bonding between the C6 hydroxyl groups of the methylcellulose chains (Sekiguchi et al. 2003). At high concentrations of polymer and high temperature, these interactions lead to an increase in viscosity and clear gel formation. Cellulose acetate gels also exhibit thermal reversible properties, which depend on variables such as polymer concentration, degree of acetylation, and type of solvent. One such example is the CA/benzyl alcohol system (Ryskina & Aver'yanova 1975). Mixtures of CA and benzyl alcohol were heated for approximately two hours are 100°C and allowed to cool to 25°C leading to gelation after an interval of thirty six hours. It was proposed that strong intermolecular associations in the system initiated gel formation. CA gels have also been made with anhydrous liquid ammonia (Taft & Stareck 1931) and were reported to be both heat-irreversible and heatreversible depending on the temperature during sample preparation. Unlike other cellulose derivatives, most CA/solvent systems gel after the polymer solution is heated to a specific temperature and subsequently cooled. During this process, intra- and intermolecular interactions rearrange in solution to form gels. Nonsolvents like water or alcohols added to a CA/DMA mixture will form an entangled polymeric network with enhanced rheological properties (Appaw et al. 2007). Gel formation was influenced by polymer and nonsolvent concentrations as well as the interactions between each of the ternary system components. This system exhibited a liquid-solid transition and phase separation with increasing nonsolvent concentration prior to gel formation. Although acetyl hydrophobic interactions played a role in the gelation process, it appears that gel properties are also influenced by competitive hydrogen bonding between system components which explained the different polymer network structures observed by rheology (Kadla & Korehei 2010). Khalil (1973) proposed that the dielectric constants of the nonsolvent dictated the formation of aggregates in the system leading to network formation. Reuvers et al. (1986) created gels via nonsolvent addition to CA with acetone, dioxane, and tetrahydrofuran as solvents. The extent of aggregation depended heavily on the solvent used. With water as the nonsolvent, polymer aggregation leading to phase separation occurred in the acetone system at nonsolvent concentrations as low as 2%, whereas with THF aggregation only occurred at concentrations above 40%. A high concentration of CA was required for gel

30

formation, where the polymer crystalline regions were one of the main reasons behind the physical cross-link formation in the demixing that led to gelation (Altena et al. 1986) . Polymer viscosity plays a role in the rate of gel formation. Solvent composition influences the morphology of the resulting membrane formed after nonsolvent addition to a concentrated polymer solution (Hoernschemeyer 1974). Polymers in good solvents have extended chains, while in poor solvents the chains are tightly coiled to reduce thermodynamically unfavourable contacts with the solvent. Specific viscosity gives information about chain configuration in solution and the number of chain entanglements increases with decreasing solvent power (Hoernschemeyer 1974).

1.6

Rheological characterization of gels

Rheology is the study of material deformation under the application of various forces in order to derive a predictive model for the flow behaviour. In practice, rheology is concerned with extending the disciplines of elasticity and (Newtonian) fluid mechanics to materials whose mechanical behaviour cannot be described with classical theories. The classical definition of a fluid is provided by Newton’s law of viscosity, where stress is proportional to the rate of strain. An elastic solid is defined by Hooke’s law, where stress is proportional to strain. Fluids are characterized by the fact that they have no resistance to shear, while a solid returns to its rest shape after an applied stress. Polymeric solutions are termed viscoelastic, as they possess both viscous and elastic properties. Correlation of molecular parameters with rheological material functions allow structureproperty relationships to be established (Clasen & Kulicke 2001). Using rheological shear and oscillation and extensional measurements, it is possible to quantitatively determine the viscoelastic properties of a system. The universality of rheological behaviour has been extensively explored during the past decades and has been used to study the liquid-solid transition (LST) of polymers (Winter & Mours 1997). Studies of the LST are technically important because they occur in many common fabrication processes such as injection molding and processing of cross-linking polymers. 31

Material properties close to the gel point combine liquid and solid characteristics, giving the potential for novel properties. Prior to gelation, the system consists of a distribution of finite clusters termed “sol” as it is soluble in good solvents. Beyond the gel point, it is called a “gel”. A gel is an extensive network of molecular or particular clusters unable to dissolve in a solvent.

Figure 1.14. Schematic representation of cluster growth during cross-linking leading to gel formation. The variable p, known as the cross-linking index, measures the extent of the crosslinking reaction measured as the ratio of the number of bonds formed to the total number of possible bonds. R is the radius of the molecular clusters formed (Winter & Mours 1997). Reproduced with kind permission from Springer Science+Business Media

The gelation process varies from material to material. Transition from sol to gel can be measured by the cross-linking index p, which is an indication of connectivity between different components of the polymer network. Figure 1.14 illustrates cluster growth during the LST in a typical polymeric system. At low values of p, the polymer is dispersed in solution. As p increases, the connectivity increases and molecular clusters (radius R) grow in size until the network spans the entire sample. For physical gels, the presence of non-covalent cross-links change the characteristics of the network as the physical bonds fluctuate with time and temperature. The gelation process for physical gels is thus reversible and possesses a potential advantage in that junctions can be formed and broken by altering the surrounding environment (temperature, pH, pressure, etc.) (Butler & Heppenstall-Butler 2003; Eissa & Khan 2005). Therefore, these gels tend to have a shorter lifespan as they possess a more fluid-like behaviour when used in long term applications. 32

The LST strongly affects molecular mobility which can be monitored by changes in the rheological behaviour. Mechanical experiments (shear flow, extensional flow, and oscillation) investigate the state of stress of a polymer solution depending on the type of deformation used. The shear experiment can be described via the stress tensor shown by Equation 1.2 (Clasen & Kulicke 2001):

 1.2

The first term describes the ambient temperature, the second the solution viscosity, and the third term represents the influence of the polymer on the overall state of stress. Steady shear measurements are performed by subjecting a sample to a constant shear rate

generating shear

stress τ21. The ratio of shear stress to shear rate describes the viscosity  of the polymer in solution (Equation 1.3):



1.3

Materials in which  is independent of shear rate are termed Newtonian. Most polymer solutions, however, deviate from Newtonian behaviour and are thus called non-Newtonian fluids. The flow behaviour can be shear thinning (pseudoplastic) or shear thickening (dilatant). Shear thinning consists of a reduction in viscosity with increased stress. With the removal of stress, the system usually undergoes a gradual recovery in structure also known as thixotropy (Ferry 1980). Shear thickening arises due to an increase in viscosity with increased stress, where deforming the material leads to rearrangement of its microstructure so the material resists flow with increasing shear rate. In this case, viscosity recovery upon stress removal is called anti-thixotropy. Figure 33

1.15 shows the relationship between viscosity and shear rate for different types of rheological behaviour.

Figure 1.15. Viscosity as a function of shear rate

Solutions of cellulose derivatives typically exhibit zero shear viscosity (0), characterized by a Newtonian region at low shear rates and shear thinning at high shear rates (Wang & Fried 1992). Cellulose acetate is no exception and exhibits a shear rate-dependent viscosity above a critical shear rate in solution (Zugenmaier 2004). Shear thinning may be caused by disentanglements of coils for high molecular weights or a breaking up of associations and increased orientation of molecular segments in flow direction (Zugenmaier 2004). Zero-shear viscosity rises with concentration as well as molar mass and causes an increase in intermolecular interactions. This can be explained by the presence of chain entanglements in a quasi-network structure of high molecular weight polymers at high concentrations (Wang & Fried 1992). Structural properties of polymeric solutions can also be explored by applying a small deformation oscillatory shear. If the strain amplitude is small enough, the fluid microstructure will not be significantly deformed and the resulting shear stress can give information about the rates of spontaneous rearrangements or relaxations present in the system. Dynamic rheology experiments consist of applying a sinusoidally varying strain (Equation 1.4) in the linear viscoelastic region (LVR). Here,  is the oscillation frequency and γ0 is the strain amplitude. The 34

LVR is defined as the region corresponding to the stress varying linearly with strain for the analyzed sample.

1.4

The stress sinusoidally varies with time and can be represented by Equation 1.5:

1.5

Here, G() is the storage modulus in phase with the strain, and G() is the loss modulus in phase with the rate of strain . The storage modulus represents the storage of elastic energy while the loss modulus represents the viscous dissipation of energy. Materials that exhibit both viscous and elastic properties are known as viscoelastic materials. Figure 1.16 shows a typical spectra of the elastic (G) and viscous (G) moduli of entanglement networks. For the fluid state G will be greater than G. At low frequencies polymer networks behave as high viscosity liquids. On a logarithmic scale, G has a slope close to 2, while G has a slope close to 1. As the frequency increases there is a cross-over in G and G. The moduli are relatively frequency independent, indicating that the structure of the material is approaching its gelation point. For a completely elastic gel, both moduli become frequency independent with G > G. For chemical gels formed by covalent cross-linking, both viscous and elastic moduli are strongly frequency independent and exhibit a slope of approximately zero.

35

Figure 1.16. Expected dynamic mechanical spectra of the elastic (G) and viscous (G) moduli of the shear storage modulus for an (left) entanglement network system (pseudo-gel) and (right) a covalently cross-linked network (true gel) (Ross-Murphy 1987). Reprinted with permission from Elsevier

Winter and Chambon (1986) have proposed a way to determine the gelation point by using rheological methods. In a frequency sweep experiment, the cross-over point from solution to gel occurs when both log G and log G show power law behaviour with the same positive exponent n. That is, when G = K1n and G = K2n, where K1 and K2 are constants and are not necessarily equal (Ross-Murphy 1995). Rheology has been used to characterize cellulose acetate gels through their sol-gel transition. This technique has been applied to various cellulosic systems such as microcrystalline cellulose/sodium carboxymethyl cellulose (Rudraraju & Wyandt 2005) and methylcellulose (Li 2002). Previous work (Appaw et al. 2010; Kadla & Korehei 2010) has shown that rheology can also be a complimentary method in determining the fractal dimension (Eissa & Khan 2005) of a system to probe microstructure and physical gel formation.

36

1.7

Goal of the project

Materials made from renewable resources have gained popularity in recent years due to the rapid depletion and increasing energy demand for fossil fuels. Cellulose, perhaps the most abundant natural yet underutilized biopolymer, can be chemically or mechanically modified to complement or replace certain materials previously dominated by the synthetic polymer market. One of the challenges in working with cellulose, however, is its inability to dissolve in common solvents due to its extensive hydrogen bonding network which make it difficult and expensive to process using the methods designed for synthetic polymers such as extrusion and molding. Cellulose cannot undergo thermal processing because, unlike most polymers, it decomposes to its natural elements before it can melt. A solution to this problem is through cellulose derivatization. By introducing acetyl groups to cellulose, the polymer does not possess as strong of a hydrogen bonding network as cellulose and is able to dissolve in common organic solvents such as acetone, chloroform, and THF. As a result, novel applications for cellulose acetates have entered the market which include, but are not limited to, films, filters, coatings, and membranes. The creation of CA ternary systems is of current interest as a means towards asymmetric membrane formation for separation and purification purposes (Tohru Shibata 2004). Acetate membranes have been used, for example, in the desalination industry for the purification of drinking water. CA is the ideal polymeric template as it is biocompatible and contains both hydrophilic (OH groups) and hydrophobic (acetyl groups) components giving a good balance between flux and rejection. The most common method for the production of acetate membranes is by the phase-inversion method where the polymer solution undergoes gelation by phase separation followed by subsequent coagulation by either solvent evaporation or contact with a nonsolvent. In the latter case a ternary system forms in the coagulation bath. The formation of a CA ternary system with specific viscoelastic and microstructural properties is dependent on many factors. The structure and concentration of the solvent can influence pore size formation (Reuvers et al. 1986). Nonsolvents with hydrophilic and hydrophobic components are able to manipulate the rheological properties of the resulting gels (Kadla & Korehei 2010). The molecular structure of the polymer, thus, would also be a factor that may change the viscoelastic and microstructural properties of the physical gels formed.

37

Structures of CA associates change or reorganize with temperature (Tsunashima et al. 2002), so we hypothesize that the cellulose acetate ternary system will exhibit different rheological and microstructural properties when subjected to different thermal treatment. Higher temperature makes CA in the ternary system more soluble and homogeneous, increasing its interaction with the solvent and nonsolvent components as the system cools and gels. CA solution studies show that phase separation can be enhanced by temperature changes (Suzuki et al. 1981) where intermolecular associations lead to physical cross-linking and the formation of three dimensional networks. Both intermolecular hydrogen bonding and hydrophobic acetyl interactions between the polymer and the solvent affect its behaviour in solution, which in turn can ultimately influence gel formation. Tsunashima et al. (2002) have investigated the effect of temperature in the formation of dynamic CA self-assemblies in DMA solution. Cellulose acetate chain conformations change due to competition between hydroxyl group hydrogen bonding and hydrophobic acetyl group interactions within the polymer. As the C6 hydroxyl group is responsible for intermolecular hydrogen bonding within the cellulose chains, it can cause the polymer to form clusters in polar solvents. We hypothesize that modifying the regiochemistry of cellulose acetate to exhibit only intermolecular hydrogen bonding, i.e. CA with only C6-OH, will change the viscoelastic properties of the ternary system as the chain becomes less rigid due to the lack of intramolecular hydrogen bonding interactions within the polymer chain. The physical properties of cellulose acetate were studied using GPC-SEC for molecular weight determination, DSC for melt transitions, and TGA for decomposition temperature. The distribution of acetyl groups within the CA polymer chain was studied by 1H and 13C NMR spectroscopy. The effect of mixing temperature, the order of component addition, and the role of polymer acetyl substitution in the formation of the CA ternary system was studied by means of steady state viscosity and dynamic rheology. Microstructural characterization of the gels was done using fluorescence laser scanning microscopy.

38

1.8

Thesis outline

Previous studies have shown that increasing the intermolecular interactions in a ternary system can lead to faster aggregation, phase separation, and gel formation. In this thesis, three different studies were carried out to investigate the effect of molecular structure (regiochemistry of the CA polymer) and processing parameters (temperature, order of addition) on a cellulose acetate ternary system. In the first part of this thesis, a series of cellulose acetate ternary systems using commercially available CA were prepared by varying the mixing temperature from 50 to 90°C. In the next section, the effect of predissolving or adding polymer to the co-solvent system and its relationship to mixing temperature was investigated. Finally, the role of hydrogen bonding and hydrophobic interactions was investigated using regioselectively modified CA polymers. In all of these studies, the effect on intermolecular interactions within the ternary system was evaluated using rheological and microscopic methods.

39

2

2.1

Experimental materials and methods

Materials

Cellulose acetate (39.7 wt% acetyl content, average Mn ca. 50,000) was purchased from Sigma-Aldrich and used as received. Reagent grade methanol, hydrochloric acid, N,Ndimethylacetamide (DMA), and chloroform were purchased from Fisher Scientific and used as received. Acetic anhydride (Fisher Scientific) was dried over P2O5 and distilled prior to use. Lithium chloride and potassium hydroxide were also purchased from Fisher. Anhydrous pyridine, anhydrous N,N-dimethylacetamide, trityl chloride, 4-methoxytriphenylmethyl chloride, 4-dimethylaminopyridine, imidazole, and propionic anhydride were purchased from SigmaAldrich and used as received. Hydrobromic acid (33 wt. % in glacial in acetic acid) was purchased from Acros Organics.

2.2

Instrumentation

Fourier-Transform Infrared (FTIR) spectroscopy spectra were recorded on a Perkin Elmer Spectrum One FTIR spectrometer. For chloroform soluble polymers thin films were deposited on ZnSe plates using 2-3 drops of 5% w/w chloroform solutions. For chloroform insoluble materials potassium bromide (KBr) pellets were made using approximately 5 mg of cellulosic material mixed with 15 mg of KBr (1:3 w/w polymer:KBr). The KBr powder was dried at 40°C under vacuum (50 mTorr) prior to use. The wavenumber range was 4000-500 cm-1. The data was collected after 32 scans at a resolution of 2 cm-1. 1

H and 13C nuclear magnetic resonance (NMR) spectra were recorded using a 300 MHz

Bruker Avance Ultrashield NMR spectrometer (300.13 and 75.03 MHz respectively) at concentrations of approximately 60 mg/ml and referenced to CDCl3 (7.26 ppm). The 1H NMR spectra were recorded at 25°C (unless otherwise noted), with a 90º pulse width and a 1.3 s 40

acquisition time. A 7 s relaxation delay (d1) was used to ensure complete relaxation of the aldehyde protons. A total of sixteen scans were recorded. 13C NMR spectra were also recorded at 25°C (unless otherwise noted) with a d1 of 3 sec (based on T1 of C=O groups as determined by Tsunashima and Hattori (2000) for quantitative analysis and a total of thirty thousand scans. Gel permeation chromatography (GPC) was performed using an Agilent 1100 Series GPC Analysis System (Agilent Technologies, Palo Alto, USA) to determine the relative average molecular mass and molecular mass distribution. Samples passed through Styragel HR-4 and HR-1 columns (Waters Corp., Milford, USA) at 35°C with THF as the eluting solvent at a flow rate of 1 ml/min. Signals were detected using a refractive index detector (RID). The GPC system was calibrated with polystyrene standards (Polysciences, Warrington, PA) with molecular weights ranging between 9 and 300K (see C5 in Appendix for calibration curve). The injection volume was 100 L, and polymers were dissolved in HPLC-grade THF at concentrations of 1 mg/ml. Elemental analysis was done on a Carlo Erba Elemental Analyzer EA 1108. The instrument determined total carbon, hydrogen, and nitrogen contents. Oxygen content was determined by difference. Approximately 5 mg of sample was dried under vacuum for 30 minutes and sealed with argon gas before analysis. Thermal gravimetric analysis (TGA) measurements were carried out using a TA Instruments TGA Q500. This technique was used to determine the decomposition temperature (Td) of the polymers. The procedure consisted of weighing approximately 4 mg of sample in a calibrated platinum pan and measuring the weight loss subject to a temperature ramp from 25600°C under nitrogen. The decomposition temperature was then taken as the peak maxima of the derivative of the weight loss curve. Differential scanning calorimetry (DSC) measurements were carried out on a TA Instruments DSC Q1000 calibrated with indium. All experiments were run with approximately 3 mg of sample in sealed aluminum standard pans at a heating rate of 20°C/min from 30° to 300°C. The optimal heating run temperature was determined from TGA measurements; by knowing the decomposition temperature, DSC experiments are performed to below that value. The glass transition temperature (Tg) using DSC was then measured as the midpoint temperature of the step change in heat capacity on the heat flow curve of the second heating run. In some cases, a melting transition (Tm) was observed as the peak maximum on the heat flow curve. 41

Rheological experiments were conducted with an AR 2000 Rheometer (TA Instruments, New Castle, DE) at 25°C. Cone (60mm 2° angle) geometry was used for samples below the gelation point while parallel plate (25mm) geometry was used for the gels. Samples of 3-4 ml were loaded onto the rheometer at the centre of the bottom peltier plate. For solutions, the cone geometry was lowered to the adjusted zero gap and excess solution around the geometry was removed. For gels, the glass vial in which they were made was broken at the neck to maintain the solid intact and prevent disruption of the gel structure. A sharp blade was used to slowly cut the gel evenly between 1-2 mm thick before mounting on the rheometer. In the case of biphasic gel systems, rheological measurements were performed on the lower gel-like layer, where the upper liquid layer was removed by pipette before cutting and loading the gel into the rheometer. A variable geometry gap of 1000-2000 m with an applied force of less than 0.2 N was applied for the gel samples with the parallel plate geometry. Steady shear measurements were performed by subjecting a sample to steady shear at a rate of 0.01-500 sec-1. The viscosity () was then measured as a function of steady shear rate. Prior to frequency sweep experiments, dynamic stress sweep experiments (0.01 to 1000 Pa) were performed to determine the linear viscoelastic (LVR) region while maintaining a constant frequency of 1 Hz. Stress sweep experiments were also used to determine the critical stress or onset of nonlinearity of the gels. This is the point where the samples no longer exhibit a linear stress-strain relationship, but rather deviate from linear viscoelastic behaviour and start to deform or rupture. The elastic (G) and viscous (G) moduli were determined by frequency sweep experiments over the range of 0.1 to 100 rad/s at a constant oscillation stress of 0.1 or 1 Pa depending on the LVR. Laser scanning microscopy (LSM) was used to visualize network formation in the gels. Calcofluor white (Sigma-Aldrich), a cellulose selective fluorescent dye was used to tag the cellulose acetate to image the evolution of network formation. The dye (0.01 wt %) was added to the CA system at room temperature, mixed, and heated to the required temperature as per rheological analysis. In a typical experiment a 15 g ternary system with 10 wt% CA will contain 0.15 mg of fluorescent dye. Right after heating, a few drops (~ 0.2 ml) of the hot solution was placed onto a single concavity microscope slide (76 x 26 mm, 3.12 mm thick, 0.8 mm deep – Electron Microscopy Sciences, USA), covered with a square 22 x 22 mm glass cover slip (No. 1, 0.13-0.17 mm thick, Fisher, USA), and sealed with clear nail polish. The sample was then

42

conditioned for one week in a dark environment at ambient temperature to avoid exposure to UV radiation during which enough time elapsed for the solution to solidify to a gel. Two-photon microscopy was used to achieve optical sectioning of the gels. The Mai Tai DeepSee (Spectra-Physics) pulsed Ti: sapphire IR laser connected to an Olympus FluoView FV 1000 upright microscope was used to image the gels. Measurements were performed with an Olympus multiphoton XLPLN 25x 1.05 NA water immersion objective lens. The IR laser excitation source was set to 720 nm with one channel emission bypass filter at 420-460 nm. The interface between the sample and the cover slip was set to 0 m and images were taken at a depth of 30 m.

2.3

Data analysis

All rheological analyses were performed within an error of less than 10%. The error was determined by running replicates per sample (minimum 4 replicates per sample). The multiple samples were averaged and the difference between the highest and lowest values was less than 10%. This was a consistent case for all of the samples studied. Therefore, rheological analyses were run in triplicates and the results were within the 10% error range. The raw data presented in this thesis is from one of the replicate runs. In the LSM experiments, image analysis was done using FluoView Ver.2.1c Viewer (Olympus Corp., Tokyo) software. Fractal dimensions were determined using the fractal box count method from five layers through the sample thickness using Image J software (NIH, USA). The spatial image resolution was 1024 x 1024 pixels. Average fractal dimension and standard deviation were then obtained.

43

2.4

Synthesis of regioselective cellulose acetate

The proposed synthesis of partially substituted cellulose acetate with controlled stereochemistry involves protecting group chemistry. Figure 2.1 shows the proposed synthetic pathway starting from commercially available CA (for characterization see Figures A1 to A6 in Appendix) based on the procedure by Tsunashima and Hattori (2000).

Figure 2.1. Proposed synthesis of regioselectively substituted cellulose acetate at the C2 and C3 positions

Cellulose was obtained through saponification of cellulose acetate with potassium hydroxide in methanol. 60 g of commercial cellulose acetate was stirred in 1.4 L methanol in an ice bath and an excess of 84 g of potassium hydroxide was added all at once. After stirring for half an hour, the ice bath was removed and the reaction was left to stir at room temperature for 24 hours. The reaction was then neutralized with approximately 70 ml concentrated hydrochloric 44

acid solution until the pH was 6-7, as indicated by pH paper. The reaction product was then vacuum filtered, washed with 1:1 methanol: water, washed with water followed by diethyl ether, and dried in vacuo overnight and stored in a desiccator until further use. Complete saponification was monitored by FTIR with KBr pellets. Infrared spectral assignments were based on those published by Garside and Wyeth (2003) and were in good agreement with the obtained results. The band at 1645 cm-1 pertains to residual water molecules adsorbed to the surface of the cellulose polymer; this was determined to be H-O-H bending vibrations of (Skorayakov & Komar 1994). FTIR (Figure A7 in Appendix): OH 3445, CH 2893, adsorbed water 1645, δCH 1423, 1377, 1318, δCH2 1265, δC-OH 1229, CC 1164, C-OH 1063, 1019.

Synthesis of regioselective CA with DS ~ 2. Cellulose was dissolved in a DMA/LiCl solution by first subjecting the cellulose through a solvent exchange with water, methanol, and then DMA by centrifugation (3000 rpm, 1690 g). Specifically, 2 g of cellulose was weighed in a 50 ml polypropylene Falcon tube and filled to the mark with solvent. After manually shaking for one minute, the mixture was centrifuged. The supernatant was then decanted and the cellulose was not allowed to dry. Cellulose was treated first with water, washed twice with methanol and twice with DMA before transferring the polymer to the reaction flask. In a tared 250 ml round bottom flask, 2 grams (0.0123 mol) of cellulose that was washed with DMA was added and mixed with more DMA for a total of 50g solvent and left to stir overnight under argon atmosphere at room temperature. To start the dissolution, the flask was heated at 120°C for two hours using a thermostat oil bath. The slurry was then allowed to cool to 100°C (approximately 15 min) at which point 3 grams anhydrous LiCl (dried at 50°, 0.05 Torr) was added to the reaction. The mixture was left to stir vigorously for 10 minutes at 100°C and then slowly cooled to room temperature by turning off the thermostat while keeping the flask under an oil bath. After approximately 12 hours the cellulose was completely dissolved, turning into a clear colorless solution, in which then 4.5 ml (4.5 mol per mol AGU) of anhydrous pyridine was added. Three molar equivalents of 4-methoxytriphenylmethyl chloride (11.44 g, 0.0370 mol) in 20 ml DMA were added to the mixture at room temperature and then heated to 70°C for 16 hours. Once the reaction was complete, the mixture was cooled to room temperature and precipitated in 2L methanol, filtered, washed twice with 250 ml methanol, and dried in 45

vacuo at 40°C at 0.1 Torr to yield a white powder. The product (1A) was soluble in DMSO and insoluble in chloroform and THF. Scheme 2.1 summarizes the tritylation step to produce product 1A. The yield was 89-91%. Elemental analysis: Calculated C 71.9, H 6.0%; Found C 69.4, H 6.1%. Thermal properties: Td 312°C 1

HNMR (300 MHz, 10% in DMSO-d6, 333K, Figure A8 in Appendix):  = 7.40 (H11),

7.29, 7.27 (H10), 7.24, 7.22 (H9), 7.13, 7.10 (H9’), 6.87, 6.84, 6.81 (H10’), 4.35 (H3), 4.15 (H2), 3.83 (H1), 3.75 (OCH3), 3.67 (H6), 3.39 (H4), 3.11 (H5). 13

C NMR (75 MHz, 10% in DMSO-d6, 333K, Figure A9 in Appendix): δ = 158.70

(C11’), 144.85 (C8), 135.81 (C8’), 130.57 (C9), 128.83 (C10), 128.15 (C11), 127.11 (C9’), 113.75 (C10’), 101.76 (C1), 86.18 (C7), 77.49 (C4), 74.78 (C2), 74.55 (C3), 74.17 (C5), 62.63 (C6), 55.52 (OCH3).

Scheme 2.1. Synthesis of 1A, 6-O-(4-methoxytriphenylmethyl)-cellulose (6TC)

One gram (0.0023 mol) of 6TC (1A) was dissolved in 10 ml anhydrous DMA with 0.05 g (5 wt %, 0.0004 mol) 4-dimethylaminopyridine (DMAP) catalyst under an argon atmosphere. Freshly distilled acetic anhydride (11.75 ml, 0.115 mol- 25 equivalents per mol of OH) was added and the solution was heated to 50°C and continually stirred for 24 hours. The reaction mixture was cooled and poured into 1.2 L methanol to obtain 2,3-Ac6TC (2A) as a white precipitate. The precipitate was purified by washing repeatedly with methanol (2 x 250 ml) and filtering, and finally dried under vacuum at 40°C. The product (2A) was soluble in chloroform,

46

THF, and DMA and insoluble in DMSO. Scheme 2.2 summarizes the acetylation step for the synthesis of 2A. The yield was 72-75%. Elemental analysis: Calculated C 69.5, H 5.8%; Found C 57.4, H 5.8%. Thermal properties: Td 369°C FTIR (Figure A10 in AppendixAppendices): CH arom 2938, C=O 1758, CC arom 1607, 1510, δCH 1446, 1365, δC-OH 1299, 1237, 1217, CC pyranose ring 1178, C-OH 1053, δCH arom 834, 758, 703. 1

HNMR (300 MHz, 10% in CDCl3, 298K, Figure A11 in Appendix): δ = 7.39 (H11),

7.30 (H10,H9), 7.28 (H9’), 6.86 (H10’), 5.09 (H3), 4.81 (H2), 4.42 (H1), 4.09 (H6), 3.82 (OCH3), 3.81 (H4), 3.57 (H5), 2.14, 2.02, 1.96, 1.76, 1.65 (-CH3 acetyl). 13

C NMR (75 MHz, 10% in CDCl3, 298K, Figure A12 in Appendix): δ = 170.20, 169.71,

169.28 (C=O), 158.69 (C11’), 147.11 (C8), 139.24 (C8’), 130.97 (C9), 129.65 (C10), 129.22 (C11), 127.16 (C9’), 113.21 (C10’), 100.52 (C1), 81.71 (C7), 76.10 (C4), 72.83 (C2), 71.53 (C3), 71.83 (C5), 62.00 (C1), 55.25 (OCH3), 20.79, 20.56, 20.47 (CH3 acetyl).

Scheme 2.2. Synthesis of 2A, 2,3-di-O-acetyl-6-O-(4-methoxytrityl)-cellulose (2,3Ac6TC)

For the detritylation step, 0.5 grams (0.96 mmol) 2,3Ac6TC (2A) was dissolved in 25 ml chloroform and 1 ml of hydrobromic acid solution (4.1 mmol) was added drop wise under vigorous stirring at room temperature. After 15 minutes the solution was poured into 1L of 70% aqueous methanol solution to obtain a white precipitate. The polymer was washed twice with 200 ml methanol and dried in vacuo to give 2,3-di-O-acetylcellulose (3A) with a yield of 0.16 g, 67% yield. The product was only soluble in DMA and DMSO. 47

Elemental analysis: Calculated C 48.8, H 5.7%; Found C 49.8, H 5.6%. Thermal properties: Tg 204°C, Tm 252°C, Td 367°C 1

H NMR (300 MHz, 10% in DMSO-d6, 313 K, Figure A13 in Appendix): δ = 5.07 (H3),

4.56 (H2), 4.28 (H1), 4.03 (H6), 3.71 (H4), 3.27 (H5), 2.07, 1.95, 1.89 (-CH3 acetyl). 13

C NMR (75 MHz, 10% in DMSO-d6, 313 K, Figure A14 in Appendix): δ = 170.66,

169.73, 169.40 (C=O), 99.75 (C1), 76.43 (C4), 75.29 (C2), 72.66 (C3), 71.93 (C5), 62.65 (C1), 20.98, 20.54 (CH3 acetyl). Synthesis of regioselective CA with DS > 2. The synthetic pathway was similar to that proposed in Figure 2.1 with the exception of the saponification step. Cellulose acetate (5 grams, 0.019 mol) was dissolved in 100 ml DMA at 80°C for half an hour and left to stir overnight under argon atmosphere at room temperature. After complete dissolution, as indicated by a clear colorless solution, 7 ml of pyridine followed by 15.5 g (0.056 mol) trityl chloride in 30 ml DMA were added to the reaction mixture and heated to 70°C for 48 hours (Camacho Gómez et al. 1996). Once the reaction was complete, the mixture was precipitated in 2L methanol, filtered, washed twice with 500 ml methanol, and dried in vacuo at 40°C at 0.1 Torr to yield a white powder. The product, 1B, was soluble in DMSO, chloroform, and THF. The yield was 5.19 g, 88%. Elemental analysis: Calculated C 56.9, H 5.7%; Found C 55.1, H 5.7%. Thermal properties: Td 363°C FTIR (Figure A15 in Appendix): OH 3488, CH arom 3024, 2943, 2883, C=O 1752, CC arom

1635, 1492, δCH 1449, 1369, 1319, δC-OH 1233, CC pyranose ring 1160, C-OH 1051, δCH arom 755,

707, 667. 1

H NMR (300 MHz, 10% in CDCl3, 298 K, Figure A16 in Appendix): δ = 7.41 (H-

arom), 5.09 (H3), 4.82 (H2), 4.45 (H1), 3.72 (H6), 3.57 (H4), 3.50 (H5), 2.14, 2.03, 1.97, 1.68 (CH3 acetyl). 13

C NMR (75 MHz, 10% in CDCl3, 298 K, Figure A17 in Appendix): δ = 170.23,

169.75, 169.30 (C=O acetyl), 142.86 (C8), , 128.18 (C9), 127.92 (C10), 127.25 (C11), 100.42 (C1), 86.67 (C7), 76.12 (C4), 72.81 (C-5), 72.29 (C3), 71.90 (C2), 62.18 (C1), 20.77, 20.57, 20.47 (-CH3 acetyl).

48

1B (5 g, 0.016 mol) was dissolved in 40 ml anhydrous DMA with 19 g imidazole (0.28 mol) under argon atmosphere at room temperature. Freshly distilled acetic anhydride (50 ml, 0.53 mol) was added and the solution was heated to 90°C and continually stirred for 48 hours. The reaction mixture was cooled and poured into 2L methanol to obtain 2B as a white precipitate. The precipitate was purified by washing repeatedly with methanol (3 x 250 ml) and filtering and finally dried under vacuum at 40°C. The acetylation product was soluble in chloroform, THF, and DMA. The yield was 4.42 g, 85%. Elemental analysis: Calculated C 56.9, H 5.6%; Found 56.1, H 5.7%. Thermal properties: Td 363°C FTIR (Figure A18 in Appendix): CH arom 2944, C=O 1755, CC arom 1433, δCH 1369, δC-OH 1235, CC pyranose ring 1161, C-OH 1052, δCH arom 736, 707. 1

HNMR (300 MHz, 10% in CDCl3, 298 K, Figure A19 in Appendix): δ = 7.41 (H-arom),

5.09 (H3), 4.81 (H2), 4.42 (H1), 4.10 (H6), 3.73 (H4), 3.57 (H5), 2.14, 2.02, 1.96, 1.90, 1.72 (CH3 acetyl). 13

C NMR (75 MHz, 10% in CDCl3, 298 K, Figure A20 in Appendix): δ = 170.21,

169.73, 169.28 (C=O), 142.86 (C8), 128.69 (C9), 128.21 (C10), 127.84 (C11), 100.49 (C1), 86.63 (C7), 76.05 (C4), 74.90 (C5), 72.80 (C3), 71.84 (C2), 62.04 (C1), 20.78, 20.46 (CH3 acetyl).

The final detritylation step consisted of the same procedure as that for 2,3-di-Oacetylcellulose. Specifically, 2 g (0.0072 mol) of 2B was dissolved in 40 ml CHCl3 and 4 ml (0.016 mol) of HBr solution was added at room temperature under vigorous stirring for 15 minutes. The mixture was then poured into 1 L of 70% aqueous methanol solution and a white precipitate was obtained. The polymer was washed twice with 100 ml methanol and dried in vacuo to give 3B as a white powder with a yield of 78%. The product was soluble only in DMA and DMSO. Elemental analysis: Calculated C 49.8, H 5.5%; Found C 49.6, H 5.7%. Thermal properties: Tg 201°C, Tm 284°C, Td 366°C 1

H NMR (300 MHz, 10% in DMSO-d6, 313 K, Figure A21 in Appendix): δ = 5.07 (H3),

4.56 (H2), 4.28 (H1), 4.03 (H6), 3.71 (H4), 3.27 (H5), 2.07, 1.95, 1.89 (-CH3 acetyl).

49

13

C NMR (75 MHz, 10% in DMSO-d6, 313 K, Figure A22 in AppendixAppendices): δ =

170.66, 169.73, 169.39 (C=O), 99.76 (C1), 76.51 (C4), 75.26 (C2), 72.68 (C3), 71.84 (C5), 62.66 (C1), 20.97, 20.60, 20.53 (CH3 acetyl).

2.5

Characterization of cellulose acetate samples

To estimate the degree of acetylation at each of the C2, C3, and C6 positions in the AGU of the cellulose acetate samples, the hydroxyl groups in the CA were fully substituted by propanoyl groups (-COCH2CH3) via the propanoation process (Tsunashima & Hattori 2000) as shown in Scheme 2.3. Here, 1.0 g of commercial CA (0.0038 mol), 2,3-di-O-acetylcellulose 3A (0.0038 mol), and 3B (0.0036 mol) was dissolved in 10 ml anhydrous pyridine and 15 ml (0.12 mol) propionic anhydride with 0.05 g (5 wt%, 0.0041 mol) DMAP catalyst. The solution was heated to 90°C and stirred for one hour under argon atmosphere and, after cooling, poured into 1 L methanol. The precipitate, cellulose acetate propanoate, was obtained as a white powder after repeated methanol washes (2 x 250 ml) and drying under vacuum at 40°C overnight. The products 4, 4A, and 4B were all soluble in chloroform. 1H and 13C NMR assignments were in good agreement with those published by Iwata et al. (Iwata et al. 1992).

Scheme 2.3. Propanoation of CA samples to determine the individual DS at the C2, C3, and C6 positions

Propanoated commercial CA 4. GPC (Figure A32 in Appendix): Mn 4.78 x 104, Mw 1.56 x 105, PDI 3.26. 50

Elemental analysis: Calculated C 50.9, H 5.8%; Found C 50.9, H 5.8%. Thermal properties: Tg 175°C, Tm 285°C, Td 368°C FTIR (Figure A23 in Appendix): CH arom 2946, C=O 1750, δCH 1429, 1370, 1322, δC-OH 1234, CC pyranose ring 1166, C-O-C glycosidic 1124, C-OH 1051, C-O-C 902. 1

H NMR (300 MHz, 10% in CDCl3, 298 K, Figure A24 in Appendix): δ = 5.05(H3),

4.77 (H2), 4.39 (H1), 4.03 (H6), 3.69 (H4), 3.52 (H5), 2.11 (-CH2 propanoyl), 1.99, 1.92 (-CH3 acetyl), 1.16, 1.06, 1.04 (-CH3 propanoyl). 1 13

C NMR (75 MHz, 10% in CDCl3, 298 K, Figure A25 in Appendix): δ = 173.66 (C6

C=O propanoyl), 173.19 (C3 C=O propanoyl), 172.77 (C2 C=O propanoyl), 170.21 (C6 C=O acetyl), 169.74 (C3 C=O acetyl), 169.29 (C2 C=O acetyl), 100.54 (C1), 76.06 (C4), 72.83 (C5), 72.51 (C3), 71.82 (C2), 62.01 (C6), 27.37, 27.17 (-CH2 propanoyl), 20.78, 20.55, 20.46 (-CH3 acetyl), 9.08, 8.97, 8.94 (-CH3 propanoyl). Propanoated 2,3-di-O-acetylcellulose 4A. GPC (Figure A32 in Appendix): Mn 5.05 x 104, Mw 5.06 x 104, PDI 1.00. Elemental analysis: Calculated C 51.7, H 6.0%; Found C 50.1, H 5.7%. Thermal properties: Td 371°C FTIR (Figure A26 in Appendix): CH arom 2946, C=O 1752, δCH 1431, 1370, 1320, δC-OH 1233, CC pyranose ring 1164, C-OH 1051, C-O-C 902. 1

H NMR (300 MHz, 10% in CDCl3, 298 K, Figure A27 in Appendix): δ = 5.08 (H3),

4.81 (H2), 4.44 (H1), 4.09 (H6), 3.72 (H4), 3.56 (H5), 2.43, 2.40 (-CH2 propanoyl), 2.14, 2.02, 1.96, 1.95 (-CH3 acetyl), 1.22, 1.20, 1.17 (-CH3 propanoyl). 13

C NMR (75 MHz, 10% in CDCl3, 298 K, Figure A28 in Appendix): δ = 173.63 (C6

C=O propanoyl), 170.18 (C6 C=O acetyl), 169.72 (C3 C=O acetyl), 169.27 (C2 C=O acetyl), 100.55 (C1), 76.10 (C4), 72.91 (C5), 72.53 (C3), 71.82 (C2), 62.01 (C6), 27.37 (-CH2 propanoyl ), 20.78, 20.56, 20.47 (-CH3 acetyl), 9.08 (-CH3 propanoyl). Propanoated 2,3-di-O-acetylcellulose II 4B. GPC (Figure A32 in Appendix): Mn 5.38 x 104, Mw 1.29 x 105, PDI 2.40. Elemental analysis: Calculated C 49.8, H 5.6%; Found C 50.3, H 5.7%. Thermal properties: Tg 201°C, Tm 284°C, Td 366°C 51

FTIR (Figure A29 in Appendix): CH arom 2946, C=O 1749, δCH 1434, 1370, 1320, δC-OH 1234, CC pyranose ring 1164, C-O-C glycosidic 1123, C-OH 1050, C-O-C 903. 1

H NMR (300 MHz, 10% in CDCl3, 298 K, Figure A30 in Appendix): δ = 5.05 (H3),

4.77 (H2), 4.39 (H1), 4.05 (H6), 3.69 (H4), 3.53 (H5), 2.39, 2.36 (-CH2 propanoyl), 2.10, 2.06, 1.98, 1.92 (-CH3 acetyl), 1.18, 1.16, 1.13, 1.06 (-CH3 propanoyl). 13

C NMR (75 MHz, 10% in CDCl3, 298 K, Figure A31 in Appendix): δ = 173.62 (C6

C=O propanoyl), 170.17 (C6 C=O acetyl), 169.69 (C3 C=O acetyl), 169.25 (C2 C=O acetyl), 100.46 (C1), 76.04 (C4), 72.79 (C5), 72.48 (C3), 71.80 (C2), 61.99 (C6), 27.34 (-CH2 propanoyl), 20.75, 20.52, 20.43 (-CH3 acetyl), 9.05 (-CH3 propanoyl).

2.6

Ternary system preparation

The mixed solvent system was prepared on a weight basis two different ways. Dissolution method 1 involved first dissolving a constant weight of CA in DMA and further adding the appropriate mass of water. Dissolution method 2 involved first mixing different ratios of DMA/water and further adding a constant weight of CA. The final CA concentration of each sample was 10% wt.

Dissolution method 1. A bulk solution (150 ml) of cellulose acetate (14.3 wt% in DMA) was first prepared in a 250 ml Erlenmeyer flask. The mixture was heated to 100°C in an oven for about 20 minutes and mechanically mixed to produce a homogeneous solution. For microscopy measurement purposes, calcofluor white (2.5 mg, 0.01%wt of CA) was added to the flask at this moment. The mixed solvent systems were then prepared by weighing the appropriate mass and proportion of water/DMA to the CA/DMA solution in a 20 ml glass scintillation vial. The final polymer concentration was 10 wt% and the total weight of each ternary system was 15 g. Five CA concentrations were prepared: 12.5, 15, 17.5, 20, and 22.5 wt%. The samples were mechanically mixed and conditioned for 24 hours at room temperature after adding the nonsolvent. Three sets of samples were prepared and heated to 50, 70, and 90°C for 10 minutes

52

in an oven and mixed again to ensure complete miscibility. Once cooled, they were purged with nitrogen and left at room temperature for 1 week prior to analysis.

Dissolution method 2. Three sets of five concentrations of nonsolvent, 12.5, 15, 17.5, 20, and 22.5 wt% were prepared. In 20 ml glass scintillation vials, five samples with 1.875, 2.25, 2.625, 3, and 3.75 g water were weighed and filled with DMA to 13.5 g. Once the samples equilibrated to room temperature, 1.5 g of CA was added to each vial to make the final mass 15 g. At this point, 0.15 mg (0.01 wt% of CA) calcofluor white was added to the samples that were being utilized for microscopy. The vials were mixed and conditioned for 24 hours at room temperature and heated to 50, 70, and 90°C for 10 minutes in an oven and mixed mechanically to ensure complete miscibility. The samples were purged with nitrogen and stored at room temperature for a week prior to analysis.

53

3

3.1

Results and discussion

General

A series of cellulose acetate ternary systems (CA/DMA/water) were prepared by varying the nonsolvent concentration, mixing temperature, and order of addition of chemical components. As the nonsolvent concentration was varied the system underwent a sol-gel transition. Changing the temperature at which the components were mixed revealed the presence of aggregates in solution, while varying the order of component addition led to phase separation; based on changes in the solvent quality causing different rates of solubility. Comparisons between commercially available and regioselectively synthesized CA in the ternary system resulted in different viscoelastic properties due to differences in intra- and intermolecular hydrogen bonding and hydrophobic interactions present between the polymer and the solvent.

3.2

Dissolution method 1: effect of varying mixing temperature on ternary system formation

Aggregation-induced phase separation via nonsolvent addition has been utilized as a method to create uniform pore structures for membrane formation (Smolders et al. 1992). In this study, steady shear rheology was used to examine the solution behaviour of the CA ternary system. The CA/DMA/water ternary systems contained different water concentrations and prepared at different mixing temperatures, 50, 70 and 90°C. Figure 3.1 shows the effect of increasing mixing temperature on the steady state viscosity as water content increases from 12.5 to 17.5 wt%. The ternary system exhibits Newtonian behaviour at low shear stress followed by shear thinning as shear stress increases - trends which are consistent with previously reported results (Appaw et al. 2007; Kadla & Korehei 2010). Addition of nonsolvent to a CA/DMA solution led to an enhancement in viscosity.

54

104 50°C 70°C 90°C

103 viscosity (Pa.s)

17.5 wt%

X100

102 15 wt%

X10

101 12.5 wt%

100 10-2

10-1

100

101

102

103

shear rate (1/s)

Figure 3.1. Effect of mixing temperature on steady shear viscosity for ternary systems with 10 wt% CA and 12.5-17.5 wt% water. Samples were prepared using dissolution method 1- addition of water to a CA/DMA solution

At low nonsolvent concentration (12.5 and 15 wt%), the different mixing temperatures give similar viscosity curves. However, at higher water content, 17.5 wt% a difference in viscosity profiles is apparent between the mixing temperatures, the sample mixed at 50°C shows the disappearance of the zero-shear viscosity plateau, giving evidence of network formation. The increase in viscosity with increasing nonsolvent content is an indication of increased polymerpolymer interactions in the ternary system. Samples mixed at 70 and 90°C still show polymer solution-like behaviour and lack structure development at 17.5 wt% nonsolvent and do not possess the stronger polymer-polymer interactions evidenced with 50°C mixing. The disappearance of the Newtonian plateau at the lower mixing temperature marks the start of the 55

non-Newtonian or power-law region. Here, shear thinning is likely caused by disentanglement of the polymer coils in solution or increased orientation of the polymer coils in the direction of flow (Clasen & Kulicke 2001). At this point, the system starts to become cloudy most likely due to aggregate formation. A visual observation of the transition from clear colorless solution to cloudy gel with the increase in water content is shown in Figure 3.2.

Figure 3.2. Evolution from clear colorless solution to cloudy gel network for a 10 wt% CA ternary system with (left to right) 12.5, 15, 17.5, and 20 wt% wate. Samples were prepared using dissolution method 1 and a mixing temperature of 90°C.

Increasing viscosity with nonsolvent content is an indication of increasing molecular entanglement in the polymer solution (Hoernschemeyer 1974). These entanglements, or polymer-polymer interactions, act as temporary cross-links or junction zones, decreasing the relative motion of molecules and their ability to flow, characteristic of physical gel formation (Winter 1999; Ross-Murphy 1995). The loss of the Newtonian plateau at low shear rate as nonsolvent concentration increases suggest the development of microstructure along with the intensification of intermolecular interactions within the sample. In the CA/DMA/water ternary system, as the solutions cool to room temperature after mixing at higher temperatures, the polymer-solvent-nonsolvent molecules rearrange and interact depending not only on component concentrations, but also pretreatment temperature. As the concentration of nonsolvent increases the constituent chemicals arrange/interact in such a way that the system forms a gel network cross-linked by hydrogen bonding, i.e. a thermoreversible physical gel leading to an increase in viscosity. Figure 3.3 illustrates the progressive increase in shear viscosity (data obtained by extrapolating the viscosity curves to zero shear for samples 56

exhibiting a Newtonian plateau and values at 0.05 sec-1 shear rate for samples exhibiting shear thinning at low shear rate) with increasing nonsolvent content for the different mixing temperatures, 50, 70, and 90°C. Heating at lower temperature produced ternary systems with the highest viscosity values at the liquid to solid transition between 17.5 and 19 wt% nonsolvent; once a solid gel formed at 20 wt% water, the viscosity values were within 20% of each other regardless of the different mixing temperatures used.

104

viscosity (Pa.s)

103

102 50°C 70°C 90°C

101

100 10

12

14

16

18

20

22

water content (wt%)

Figure 3.3. Effect of mixing temperature on the viscosity enhancement of 10 wt% CA ternary systems with 12.5-20 wt% water (values extrapolated to zero-shear viscosity for samples exhibiting a Newtonian plateau and at 0.05 sec-1 shear rate for samples exhibiting shear thinning behaviour at low shear rate). Data represent the average of 3 replicates

With the addition of water, the solvent quality decreases because the polymer becomes less soluble in DMA. Weaker solvents lead to stronger polymer-polymer interactions which 57

affect cluster formation. The binary solvent system can be classified as a good or poor solvent depending on how well the polymer dissolves. Polymer chains tend to be extended in good solvents, while the chains are coiled tightly in poor solvents to reduce thermodynamically unfavourable contacts with the solvent (Hoernschemeyer 1974). DMA/H2O combinations with low water contents (10-12.5 wt%) solubilize the polymer to a greater extent than the systems with greater than 17.5 wt% water, as indicated by the lower solution viscosity and onset of phase separation (Figure 3.3). In concentrated solutions, viscosity increases with increasing water concentration more rapidly in poor solvent systems than in in good solvent systems. As solvent power or range of solvency decreases, the number of chain entanglements increases, as the solvent-nonsolvent system is unable to solubilize the polymer (Hoernschemeyer 1974). Thus, samples mixed at 50°C contain more chain entanglements as the polymer is present in a poorer solvent system than those mixed at 70 and 90°C. A sol-gel transition was observed when the ternary system elastic (G) and viscous (G) moduli were plotted as a function of frequency at three different water contents for 10 wt% CA solution heated at 90°C (Figure 3.4). At 15 wt% water, G is greater than G, indicating a solution. Both G and G exhibit strong frequency dependence, G having a slope of 1.3 and G 0.96. Sample deformation at this point is very slow so that the majority of the energy is dissipated by viscous flow. The polymer chains have enough time to avoid external deformation by relaxing to an energetically more favourable state. The relaxation process occurs by slippage of the entanglement points of intertwined polymer chains (Kulicke et al. 1996). Increasing the water content to 20 wt% decreases the slope of G to 0.40 and G to 0.60 and the system is now approaching the sol-gel transition. The Winter and Chambon criterion for gel formation is that both the viscous and elastic moduli have the same slope. Both moduli are increasing and become relatively frequency independent compared to the sample with 15 wt% nonsolvent. By 20 wt% nonsolvent, the G value starts higher than G at small angular frequency and a G and G crossover can be observed at an angular frequency of 3.40 rad/s. With increasing water content, more entanglements are present in the system, leading to slower relaxation times for the polymer chains. The entanglement points now act more like a fixed network. Consequently, the ability of the polymer network to store the imposed energy increases, and therefore starts to behave more like an elastic solid. As the ternary system reaches 25 wt% water, solid gel formation takes place

58

and both G and G are frequency independent with slopes  0, and G > G, characteristic of a 3dimensional elastic gel.

105 104

G' and G" (Pa)

103

102 101 G' (25 wt%)

100

G" (25 wt%) G' (20 wt%)

10-1

G" (20 wt%) G" (15 wt%)

10-2

G' (15 wt%)

10-3 10-2

10-1

101

102

103

angular frequency (rad/s)

Figure 3.4. Elastic (G) and viscous (G) moduli for 10 wt% CA solution at 15, 20, and 25 wt% water mixed at 90°C. With increasing water content a transition can be visually observed from a clear solution, to a cloudy system, and ultimately to a gel.

With increasing nonsolvent content, the system transitions from a clear colorless fluid to a solid gel. The process can be monitored with G at a fixed angular frequency (1 rad/s, which is considered the low frequency limit (Sonntag & Russel 1987)) as shown in Figure 3.5. As water concentration increases, the system goes through three distinct regions. At low water content, G is small and increases very slowly, characteristic of a solution. The system then transitions to an 59

intermediate phase where G increases sharply and undergoes a liquid to solid transition, and finally reaches a plateau where a solid gel is formed.

105 104

G' (Pa)

103 102 101

50°C 70°C 90°C

100 10-1 9

14

19

24

29

water content (wt%)

Figure 3.5. Effect of increasing water content from 10 to 27.5 wt% in a 10 wt% CA ternary system on G at an angular frequency of 1 rad/s at various temperatures.

All of the systems, heated at 50, 70, and 90°C show the same sigmoidal shape in G at an angular frequency of 1 rad/s as a function of increasing water content. In the sol-gel transition zone, between 17.5 and 21 wt% water there is a slightly larger difference between the three temperatures. The rapid increase in G at the LST may be due to increasing competition between CA, DMA, and water for the thermodynamically favorable formation of hydrogen bonding interactions. Commercial CA contains hydroxyl groups in positions C2, C3, and C6 with an individual DS of 0.78, 0.86, 0.83 respectively, which make hydrogen bonding possible between 60

the –OH groups of the polymer and the carbonyl group in DMA (Pintaric et al. 2000). Water as a nonsolvent may also introduce hydrogen bonding between the polymer and solvent, especially with DMA. To quantitatively assess the significance of the intermolecular interactions between solvent/nonsolvent and CA, values for the cohesive energy density (CED) can be used to predict solubility and solution behaviour. Typically, the Hildebrand solubility parameters (δ), which are the square root of the CED, are used to describe these interactions where materials with similar values of δ are likely to be miscible (Barton 1975). For mixtures, the solubility parameter is often taken as the sum of the products of the component solubility parameters and their respective volume fractions (), summarized by Equation 3.1:

3.1

From a thermodynamic point of view, the strength of the intermolecular forces in a solution must be equal to the CED (Burrell 1973). Hansen (1969) assumed that the cohesive energy arises from dispersive (δd), permanent dipole-dipole (δp), and hydrogen bonding (δh) forces described by the following relationship (Equation 3.2):

3.2

These three terms represent the different contributions to the free energy of mixing prevalent in polar systems. In this case, hydrogen bonding forces dominate between various groups in the solvent and polymer as they are much stronger than van der Waals or dipole forces. DMA, an amine, would be considered a strong hydrogen bonding solvent (Hansen 2007). Table 3.1 lists the values of these terms for the various components of the ternary system used in this study. 61

Table 3.1. Hansen solubility parameters for CA, DMA, and water at 25°C (Brandrup et al. 2003) Solubility parameters (MPa0.5) δ

δd

δp

δh

CA

25.1

18.6

12.7

11.0

DMA

22.7

16.8

11.5

10.2

H2 O

47.9

15.5

16.0

42.4

Previously Appaw et al. (2007) showed that the hydrogen bonding solubility parameter index closely correlates with changes in G, while both δd and δp remain relatively constant with respect to G. Figure 3.6 illustrates the solvent solubility parameter (δ) as a function of G for all three temperatures. G values were obtained from frequency sweeps at 1 rad/s. The addition of nonsolvent to the CA solution leads to an increase in the solvent/nonsolvent solubility parameter since δ for water is higher than that for DMA. At low values of G, there seems to be a relatively weak dependence on δ, followed by a slow increase of δ during the sol-gel transition, and finally δ dramatically increasing at high G. Ternary systems mixed at 90°C show the highest solubility parameters at a given G. As expected, these samples form the most homogeneous gels and show evidence of network formation at higher nonsolvent content because of better component solubility.

62

Solvent solubility parameter (MPa0.5)

30.5

30 29.5 29 28.5 28 27.5 27 50°C

26.5

70°C

26

90°C

25.5 10-1

100

101

102

103

104

105

G' (Pa)

Figure 3.6. Effect of solvent solubility parameters on the elastic modulus G for gels heated at various temperatures at 10 wt% CA at a fixed frequency of 1 rad/s

To monitor network formation with the increase of nonsolvent concentration in the ternary system, a stress sweep can be applied. A system that is weakly cross-linked will exhibit a stepwise decrease in the elastic modulus, while a strongly cross-linked system will show an elastic modulus that is relatively independent of applied stress before rupture (Winter 1999). The presence of entanglements, i.e. polymer-polymer interactions, within the ternary system can be relatively quantified, as a breakdown of interactive bonds in the material usually occurs with increasing stress (Macosko 1994). The stress amplitude of the elastic modulus increases with water content as shown in Figure 3.7. The materials are relatively independent of the applied stress at low oscillation stress levels, indicating that they are within the linear viscoelastic region (LVR). The stress sweep experiment can be used as an indicator of structure stability, where higher elastic modulus is associated with a well-dispersed and stable system (Kavanagh & Ross-Murphy 1998). At low

63

nonsolvent concentration before the gel point, G remains almost the same for all three temperatures. Once past the gel point at 20 wt% nonsolvent, the elastic modulus increases with increasing mixing temperature. At high stress levels, shear thinning occurs because the aggregates start breaking up. It is also worth noting that the critical stress increases with increasing oscillation stress, making these gels strong-linked (W.H. Shih et al. 1990) . The use of scaling theory to classify colloidal gels as strong- or weak-linked will be explained later. At higher stress, a breakdown of the interactive bonds occurs and a gradual decrease in G is observed.

104 103

G' (Pa)

102 101 50°C 15 wt% 50°C 20 wt% 70°C 15 wt% 70°C 20 wt% 90°C 15 wt% 90°C 20 wt%

100 10-1 10-2 10-2

10-1

100

101

102

103

104

osc. stress (Pa)

Figure 3.7. Stress sweep experiment for samples with 10 wt% CA and 15-20 wt% nonsolvent at different mixing temperatures. Experiments were performed at a fixed frequency of 1 Hz

Physical gels consist of extensive macromolecular clusters held together by interactions such as hydrogen bonding and hydrophobic interactions. As mentioned previously, DMA interacts with CA in such a way that the solvent carbonyl group is associated with the hydroxyl 64

group of the polymer. With the addition of water, a hydrogen bond donor and acceptor, new interactions form between the nonsolvent and DMA as well as nonsolvent and CA (Appaw et al. 2010). Cellulose acetate exists as molecular aggregates in dilute DMA solution (Kawanishi et al. 2000). Depending on molecular structure and temperature, cellulose acetate can adopt different conformations (Tsunashima et al. 2002) and decrease in chain stiffness (Johnston & Sourirajan 1973). CA exhibits a semiflexibile conformation depending on the solvent, existing in three types of structures in polar solvents: one as a single chain, and two others formed by hydrogen bonding between intermolecular hydroxyl groups on the CA C6 position and the solvent. Light scattering studies have shown that the hydrodynamic radius of the polymer decreases as the temperature increases from 2 to 61°C (Tsunashima et al. 2002), and one would expect that the polymer would be more soluble at lower temperatures as the polymer possesses a lower critical solution temperature in DMA. However, a low hydrodynamic radius can also mean that the polymer ceases to exist as a globular structure and collapses at high temperature (Tsunashima et al. 2002). The presence of both acetyl and hydroxyl groups within the anhydroglucopyranose unit (AGU) promotes strong inter- and intramolecular interactions in CA, particularly hydrogen bonding. These hydrogen bonds often bridge neighbouring cellulose structural units, making these molecules particularly rigid (Zugenmaier 2004). More specifically, intramolecular hydrogen bonding is present between the C3 hydroxyl group and the ring oxygen (O5) as well as the C2 hydroxyl group with the oxygen at the C6 position. Hydroxyl groups present at both C2 and C3 create a stiffer chain as the polymer has restricted rotation about the glycosidic bond (Kawanishi et al. 1998) . Although intramolecular hydrogen bonding does not participate in cluster formation, it can influence chain conformation and stiffness which affect the position in which intermolecular interactions take place. Intermolecular hydrogen bonding can be formed between the hydroxyl group on the C6 resulting in association. The associated clusters change with varying solution environment as dynamic hydrogen bonding interactions cooperate with hydrophobic interactions (Tsunashima et al. 2002) . These hydrophobic interactions arise from the presence of acetyl groups in the polymer chain as well as the solvent. Figure 3.8 shows the effect of water content on G and G in a system with 10 wt% CA concentration mixed at 50°C. As the water content increases, the samples slowly start showing 65

gel-like features; G > G and becoming weakly frequency dependent with G having a slope of 0.10 and G a slope of 0.38, a pattern similar to that shown in Figure 3.4 for 90°C mixing. It is of significance to note that at 20 wt% water the samples mixed at 50 °C are already gels, G > G, and that both viscoelastic moduli within the same frequency range possess higher values than those in the system mixed at 90°C, which undergo a G/G cross-over at this water content. Therefore, the ternary systems mixed by dissolution method 1 undergo a gel transition at lower nonsolvent content when heated at lower temperature and exhibit stronger viscoelastic properties.

105

G' and G" (Pa)

104 103 102 101

17.5 wt% 20 wt% 22.5 wt%

100 10-1 10-1

100

101

102

angular frequency (rad/s)

Figure 3.8. Frequency dependence of the elastic and viscous moduli for 10 wt% CA/DMA/H2O mixtures, mixed at 50°C. G is represented by solid filled shapes while G is unfilled

66

Figure 3.9 shows the effect of increasing water concentration at the liquid to solid transition (LST) on the elastic modulus. It can be seen that G exhibits a power law behaviour (Eissa & Khan 2005) with increasing nonsolvent content , where G ~ n and n varying between 45 and 49. These results suggest that the gels are fractal in nature and that they possess similar microstructures as the observed values for n are quite similar (Buscall et al. 1988). Larger power law exponents, as found in this system are usually an indication of weaker gels (Wu & Morbidelli 2001).

105 104

50°C 70°C

G' (Pa)

103

90°C

102 101 y = 3E-56x45 y = 3E-62x49 y = 6E-58x46

100

R² = 1.00 (50°C) R² = 0.99 (70°C) R² = 0.99 (90°C)

10-1 17

18

19

20

21

22

water content (wt%)

Figure 3.9. Elastic modulus (obtained at a frequency of 1 rad/s) of 10 wt% CA/DMA/H2O gels as a function of water content

Elastic properties of colloidal gels can be related to network structure. At the liquid to solid transition phase, the structure of the gel network can be considered as a collection of flocs, which, according to scaling theory, are fractal objects closely packed throughout the sample. 67

Fractality is a means of quantifying the structure of non-Euclidean objects, and the fractal dimension is a way of describing the complexity of a structure’s geometry in a single number. In the case of polymer networks created by aggregation-induced gelation, the fractal dimension provides information on the mechanism of aggregation and gelation (Marangoni 2002). The particles in the gel aggregate due to interparticle attractions. The nature of these flocs can be classified as weak-linked or strong-linked depending on the values of the elastic moduli G. Using scaling theory, Shih et al (1990) described strong-linked gels as an aggregated network where particle interactions between the flocs (interflocs) are stronger than those within the flocs (intraflocs), while weak-linked gels have intraflocs stronger than interflocs. In the case of stronglinked gels, failure under deformation would occur through breaking the intrafloc linkages, and would be observed as a decrease in the linear viscoelastic region as sample concentration increases under strain (Eissa & Khan 2005). Figure 3.10 shows the effect of increasing nonsolvent on the elastic modulus as a function of percentage strain.

105 104

G' (Pa)

103 102 101

10 wt%

100

15 wt% 20 wt%

10-1 10-2 10-6

22 wt%

10-5

10-4

10-3

10-2

10-1

100

101

102

% strain

Figure 3.10. Elastic modulus versus % strain for a 10 wt% CA ternary system mixed at 50°C. The limit of linearity (percentage strain at the intersection between the tangent of the drop in G and the linear viscoelastic region) shifts to lower strain as nonsolvent concentration increases from 10 to 22 wt% 68

The limit of linearity is defined as the value of the strain at which there is an appreciable deviation in the elastic modulus from the linear viscoelastic plateau. It is evident that in this case the limit of linearity decreases as nonsolvent concentration increases, making the gels stronglinked. The onset point of non-linearity in G for samples mixed at 50, 70, and 90°C is shown in Figure 3.11; the onset point being defined as the intersection between the tangent of the drop in G and the linear viscoelastic region in a G vs % strain plot. This power law behaviour is only applicable to the liquid to solid transition region; solutions containing below 17.5 wt% water and gels above 22.5 wt% water exhibit minimal changes in percent strain at the limit of linearity with changes in nonsolvent content.

103 50°C 70°C 90°C

% strain

102

101 y = 6E+20x-15 y = 2E+18x-13 y = 6E+16x-12

R² = 0.99 (50°C) R² = 0.98 (70°C) R² = 0.99 (90°C)

100

17

17.5

18

18.5

19

19.5

20

20.5

21

21.5

water content (wt%)

Figure 3.11. Onset point of nonlinearity of the elastic modulus for samples with 10 wt% CA mixed at 50, 70, and 90°C and 17.5 to 21 wt% water content

The gels mixed at three different temperatures were all classified as strong-linked gels, but the rate of change in nonlinearity differs between temperatures. The slope for gels heated at 50, 70, and 90°C were -15, -13, and -12 respectively. Thus, the system heated at 50°C showed 69

the largest change in nonlinearity as it had the largest slope in absolute values, while those formed at 70 and 90°C were essentially the same. With increasing nonsolvent content, the gels formed at 50°C will be more sensitive to strain. Despite the higher values of G as shown in Figure 3.9, gels heated at 50°C showed lower strain tolerance and ruptured at lower strain levels compared to those heated at higher temperatures. These results imply that heating at 50°C creates networks with relatively weaker interfloc links. The elastic modulus G ~ n as mentioned earlier. For strong-linked gels, the power exponent n equals (d+x)/(d-D), where x is the backbone fractal dimension of the flocs, d is the Euclidean dimension, and D is the gel fractal dimension. For colloidal gels, the backbone fractal dimension ranges from 1 to 1.3 (H. Wu & Morbidelli 2001). The value of x was arbitrarily set to the midpoint 1.15 as the range of x only differs by 0.3 and would change the values of D by around 2% (Eissa & Khan 2005). Equating experimental and theoretical power law exponents and substituting d = 3 (particles propagate in three dimensions), the fractal dimensions D for the system mixed at 50, 70, and 90°C are all ~ 2.91 (see C1 in Appendix for calculation) based on values from Figure 3.9. For strong-linked gels, the breaking of bonds occurs within a floc, so the limit of linearity 0 ~ -(1+x)/(d-D) at which the weakest bonds break and the linear elastic behaviour vanishes. Fractal dimensions obtained by this method based on the values from Figure 3.11 yield an average of D ~ 2.84 (see C2 in Appendix for calculation). Similar values of D suggest similar structures for the gels. This fractal dimension describes the order in the arrangement of mass within the region as well as the degree of occupancy of that embedding space. Similar fractal dimension values have been reported for fat crystals using rheology (Narine & Marangoni 1999; Marangoni 2002). High fractal dimensions are associated with more homogeneous special distributions of network mass. Fractal dimensions can give an indication of the mechanism of aggregation as flocs formed can be perceived as fractal objects. Networks with higher fractal dimensions usually indicate larger cluster size (Marangoni 2002). With higher nonsolvent content, the more clustercluster interactions exist per unit volume, resulting in a higher elastic constant and lower fractal dimension. A fractal dimension of D = 1.75 is indicative of fast aggregation, coincident with the result of the cluster-cluster aggregation (CLCLA) model (Shih et al. 1990). In this case, particles cluster by means of diffusion only, and a low fractal dimension is expected since these clusters have very low density. When D approaches 3 it is more indicative of slow aggregation; the result 70

coincides with the reaction-limited aggregation (RLA) model where cluster formation is partially reversible and particles rearrange in order to form multiple contacts with neighbouring particles (Yuan et al. 2008). The value of the fractal dimension close to 3 suggests RLA for the CA/DMA/H2O system, pertaining to the formation of a compact aggregate three-dimensional network. Other network systems suggesting the RLA mechanism of aggregation have been reported with D = 2.8 (Amal et al. 1990). Two-photon excitation microscopy images show a microporous network structure at the gel transition with 20 wt% water (Figure 3.12). The brighter areas are the cellulose acetate stained with calcofluor white, a cellulose selective dye. Darker segments represent non-CA domains. The transition from gels heated at 50 to 70°C change from a uniform dense network to a more non-uniform structure. Gels heated at 90°C show a very uniform homogeneous network once again. The increased network connectivity is consistent with changes in hydrophilic (hydrogen bonding) and hydrophobic (non-polar) interactions as temperature increases. The images must be analyzed with caution, however, as microscale images of the gels are not necessarily a macrostructural representation of the whole gel. Calculated fractal dimensions using Image J software were on average 1.98 for gels prepared at 50, 70, and 90°C (see C3 in Appendix for calculation). The images were first thresholded in order to see the microstructural elements as white and all of the background reduced to black. The traditional box counting method was used where the slope of the log-log plot of the number of particles vs. box sizes corresponds to the fractal dimension D. This method corresponds to d = 2 space (Marangoni 2002). The discrepancy between the fractal dimension calculated from rheology and microscopy probably arises from calculating values related to two different dimensional Euclidean spaces – d = 2 for microscopy and 3 for rheology. It is also important to keep in mind the issue of depth of field. The clusters present in the entire volume of the microscopy image are accounted for and projected onto a two-dimensional plane (Marangoni 2002). Fractal analysis of the distribution of particles in d = 3 space will yield integer and noninteger dimensions for Euclidean arrangements, D = 1 for a line, D = 2 for a plane, and D = 3 for a cube. Analysis on a two-dimensional image, d = 2, will have contributions of D = 1 (line) and D = 2 (plane) and yield a fractal dimension less than 2, as the mass fractal dimension is always smaller than the corresponding Euclidean dimension (D < d).

71

a

b

c

Figure 3.12. LSM images of 10 wt% CA gels containing 20 wt% water heated at (a) 50°C, (b) 70°C, and (c) 90°C. Gels were tagged with calcofluor white (0.01 wt% of CA). Each image dimension is 127m x 127 m.

72

Using the microscopy method, the scaling relationship between the number of discrete particles (N) and the length of the region of interest containing particles L are expressed as N = cLD where c is a constant (Rye et al. 2005). The fractal dimension in this case describes the degree of order in the packing of the microstructural elements within the microstructures. With rheology, the scaling relationship between shear modulus G and volume fraction , as mentioned previously (Shih et al. 1990), is expressed as G = 4.15/(3-D), where  is a constant. Thus, the fractal dimension describes the distribution of mass within a region of the network. Discrepancies therefore exist between fractal dimensions estimated by different methods as they have different interpretations. Studies with fat crystal networks have also reported differences between microscopy and rheology in the calculation of fractal dimensions (Marangoni 2002). Since the ternary system studied forms physical gels, network structure can be disrupted by the application of thermal energy. By applying a temperature sweep, one can observe the transition from gel at 25°C to solution as the temperature increases. The stability of the gel to temperature change can also be monitored with changes in the elastic modulus. Figure 3.13 shows the elastic and viscous modulus of 20 wt% nonsolvent gels with 10 wt% CA heated at various temperatures undergoing a temperature sweep at 1° per minute from 25°C to 80°C. A transition from a completely opaque to a clear colorless solution could be visually observed. In this oscillation experiment, both elastic and viscous moduli decrease with increasing temperature as the mobility of the polymer increases with thermal energy. The elastic modulus decreases faster than the viscous modulus. The dependence of G and G on temperature provides a good indication of the transition from solution to gel and vice versa. Three regions in the ramp experiment could be observed. The first region is characterized by a relatively low dependence on temperature for both G and G and G is much greater than G; this region is shorter for the gel mixed at 50°C than at 70 and 90°C. Beyond this region at above approximately 40°C, G and G start to decrease rapidly as more cross-links break until reaching a crossover point at about 42, 55, and 50° C for samples mixed at 50, 70, and 90° C respectively. G is now greater than G and the samples are now solutions.

73

103

G' and G" (Pa)

102

101 50°C 70°C

100

90°C

10-1 20

30

40

50

60

70

80

90

temperature (°C)

Figure 3.13. Temperature sweep experiments conducted at 1°C/minute for 10 wt% CA and 20 wt% nonsolvent mixed at 50, 70, and 90°C. G is represented by solid filled shapes while G is unfilled

The gels formed at 50°C undergo the gel-solution crossover at lower temperature than those formed at 70 and 90°C. This could be due to the interactions between the components being much weaker as less thermal energy is required to transition from an elastic solid to a solution; this system is not as homogeneous as those mixed at higher temperature. The rate of decrease in both G and G is also much faster for the gel mixed at 50°C, exhibiting a stepwise decrease in both moduli with increasing temperature indicating lower stability of the gel. Those gels mixed at 70 and 90°C were able to maintain relatively temperature independent moduli both before and after the G and G crossover point. This relationship between mixing temperature and thermal response correlates well with what was previously observed in Figure 3.11, where the strong-linked gels prepared at 70 and 90°C were less sensitive to changes in deformations (strain) with increasing water content than the gels prepared at 50°C. Thus, the ternary system prepared at 50°C has weaker interfloc links 74

despite possessing higher G at the low frequency range of 1 rad/s (Figure 3.9). This behaviour is typical of a physical weakly cross-linked gel, in which there are a minute number of cross-link connections, and a system of finite distribution of macromolecular clusters (Winter & Mours 1997).

3.3

Dissolution method 2: effect of varying mixing temperature on ternary system formation

In this section, the effect of component addition on the formation of a CA/DMA/H2O ternary system mixed at different temperatures was investigated. Adding the polymer last and then heating produced a system similar to that formed with dissolution method 1 at high temperatures, but differed when mixed at lower temperature. The system formed at 50°C resulted in two phases (Appaw et al. 2010), one liquid and one gel layer, becoming more prominent at higher water concentration. Figure 3.14 shows the change in viscosity as nonsolvent content increases when the polymer is added to a solution of DMA and water (dissolution method 2). As both the DMA solvent and water nonsolvent are polar, they are completely miscible at all temperatures and concentrations. Addition of polymer powder increases the viscosity with increasing water concentration. The viscosity values in this case, however, are less than those obtained with dissolution method 1, signifying weaker polymer-polymer interactions in the ternary system.

75

104 50°C 70°C 90°C

viscosity (Pa.s)

103 17.5 wt%

X100

102 15 wt%

X10

101 12.5 wt%

100 10-2

10-1

100

101

102

103

shear rate (1/s)

Figure 3.14. Temperature effect on steady shear viscosity for a 10 wt% CA ternary systems at 12.5-17.5 wt% water, dissolution method 2- addition of polymer to a water/DMA solution

At low nonsolvent content Newtonian behaviour was observed at low shear rates followed by shear thinning at high shear rates. The shear thinning effect is more pronounced at low shear rates for samples with higher water content. The non-Newtonian behaviour and disappearance of the Newtonian plateau implies the development of microstructure in the sample resulting from enhanced intermolecular interactions as nonsolvent content increases, more evident in those samples heated at 50 and 70°C. The effect of increasing temperature is more pronounced in dissolution method 2. Heating the gels at 50°C led to the formation of a two-phase system at higher water contents, while heating the gels at 70 and 90°C produced a homogeneous gel. This effect can be visually observed in Figure 3.15.

76

a.

b.

c. Figure 3.15. Visual observation of gel formation at (a) 50°C, (b) 70°C, and (c) 90°C with 10 wt% CA and increasing nonsolvent content from 12.5 to 22.5 wt% with 2.5 wt% increments

These phase separation effects have been previously reported by Appaw et al. (2010). The two-phase system formation is characterized by an upper clear liquid “solution-like” layer and a lower viscous “gel-like” layer. With increasing nonsolvent, the polymer is partitioned into the lower viscous layer and enhances its viscoelastic properties. The two-phase behaviour may result from competitive interactions between CA and water for DMA, as DMA is very miscible with water and has a strong affinity for CA. Water and CA, however, are not miscible in solution which is why the system undergoes phase separation with increasing water content. The demixing behaviour in the ternary cellulose acetate solution occurs via liquid-liquid phase separation (spinodal decomposition) or the formation of aggregates (nucleation and growth) (Reuvers & Smolders 1987). During the gel formation process, the system is in a metastable state on cooling where it can either (1) separate into two liquid phases through 77

nucleation and growth of one of the phases, or (2) demix due to polymer molecules aggregating and precipitate into solution (Figure 1.12). Ternary systems phase separate via different mechanisms depending on the temperature and conformational changes in the polymer (Butler & Heppenstall-Butler 2003). In the case of dissolution method 2, the ternary system forms two phases due to the slow rate of diffusion of the polymer into the DMA/H2O mixture at high nonsolvent content, where the kinetics of phase separation is likely faster than the kinetics of gelation. At higher temperatures, however, the polymer undergoes different conformations and is able to overcome the diffusion energy barrier to form new interactions in which the polymer is completely dissolved in solution. For dissolution method 1, the polymer was already in solution before the addition of nonsolvent, which led to faster gelation resulting in a homogeneous phase elastic solid at high water content. Reuvers and Smolders (1987) have also pointed out that there is a difference between the temperature in which aggregates are formed and gelation occurs and the temperature in which the aggregates dissolve and the structure becomes fluid. The sol-gel transition of the systems can be rheologically characterized. A plot of G and G as a function of frequency for samples mixed at 50°C is shown in Figure 3.16. A G-G cross-over can be observed for the sample with 20 wt% nonsolvent at an angular frequency of 18.2 rad/s. According to the Winter and Chambon (1986) criterion for gel formation, the system transitions from solution to gel at approximately 22.5 wt% water as both the log G and log G have the same slope of 0.16.

78

105

G' and G" (Pa)

104 103 102 101

17.5 wt% 20 wt%

100

22.5 wt% 25 wt%

10-1 10-1

100

101

102

angular frequency (rad/s)

Figure 3.16. Elastic (G, filled shapes) and viscous (G, unfilled shapes) moduli of 10 wt% CA gels as a function of frequency mixed at 50°C.

Comparing dissolution method 1 to dissolution method 2, adding polymer last leads to lower G and lower G for gels at the same nonsolvent concentration (see Figure 3.8). The samples start to become weakly frequency dependent at 22.5 wt% nonsolvent and exhibit elastic moduli ten times smaller than those produced using dissolution method 1. Values for G obtained at a fixed angular frequency of 1 rad/s for samples heated at 50, 70, and 90°C with increasing water content are shown in Figure 3.17.

79

105 104

G' (Pa)

103 102 50°C

101

70°C 90°C

100 10-1 11

13

15

17

19

21

23

25

27

29

water content (wt%)

Figure 3.17. Effect of increasing nonsolvent content in 10 wt% CA ternary system on the elastic modulus at 1 rad/s with varying temperature by dissolution method 2

At low water content, the system is in the solution stage with low values of G. Ternary systems heated at 50°C exhibit higher G compared to those heated at higher temperatures, just like in dissolution method 1. The transition stage between solution and gel (18-20 wt% water) shows the elastic modulus increasing tenfold when heated to 90°C compared to 50°C. Past the 19 wt% water content, gels mixed at 90°C show the highest elastic modulus. Stress sweeps indicate weaker gels compared to those obtained from dissolution method 1 as shown in Figure 3.18. At 15 wt% nonsolvent content, the system is still in the solution phase and is relatively independent of the applied stress at low oscillation stress levels. The same shear thinning behaviour can be observed at higher stress. Close to the gel point at 20 wt% water, the sample mixed at 50°C exhibited a stepwise decrease in G with increasing applied stress. This is indicative of a weakly cross-linked polymer structure in which there are minute numbers of cross-linked connections in a system with a finite distribution of macromolecular clusters 80

(Winter & Mours 1997). At this stage of network formation, the applied stress is sufficiently high enough to disrupt the cross-links, but cannot rupture the system. As a result, G temporarily drops in the viscoelastic region until reaching the critical stress at around 400 Pa. With dissolution method 1, the system heated at 50°C did not show any stepwise decrease in G; instead, it exhibited a long linear viscoelastic region followed by a sharp drop (system rupture) at high applied stress due to the formation of a strong network structure with infinite macromolecular clusters.

104 103 50°C - 15 wt% 50°C - 20 wt% 70°C - 15 wt% 70°C - 20 wt% 90°C - 15 wt% 90°C - 20 wt%

G' (Pa)

102 101 100 10-1 10-2 10-2

10-1

100

101

102

103

104

osc. stress (Pa)

Figure 3.18. Temperature effect on stress sweep for 10 wt% CA ternary systems prepared with dissolution method 2

The ternary system prepared at 70°C and 90°C show a relatively long linear viscoelastic region followed by shear thinning at high oscillation stress. Heating at 70°C with dissolution method 2 led to a critical stress of about 27 Pa, while the sample prepared with dissolution 81

method 1 had a critical stress of about 630 Pa, a more than twenty-fold increase. Heating at 90°C showed a critical stress of 736 Pa and 32 Pa for heating with dissolution method 1 and dissolution method 2 respectively. Elastic modulus values were also higher with dissolution method 1 for samples mixed at 70 and 90°C. This could be due to a more homogeneous mixture of components being formed at higher temperature, in which during the cooling process leads to stiffer gels. Cellulose acetate ternary systems form a physical gel, in which the junctions form or break when altering temperature, making the gelation process reversible. The viscosity of the system decreases as the temperature increases because the mobility of the polymer increases with higher temperature but decreases again as the system cools. Samples heated at 90°C allow for more molecular movement within the ternary system, and the components rearrange in the most thermodynamically stable conformation as the gels cool. Samples heated at 50°C, cooled, and reheated to 90°C will end up possessing almost the same viscosity as that heated initially to 90°C. Table 3.2 shows viscosity values extrapolated to zero shear for samples containing 20 wt% water mixed by two different methods. This nonsolvent concentration is close to the twophase formation point in the ternary system, as determined by visual observation, when mixed by dissolution method 2. These samples were then reheated to 90°C to determine the thermoreversibility of the gels.

Table 3.2. Zero shear viscosity values for samples with 20 wt% nonsolvent prepared by dissolution method 1, dissolution method 2 and dissolution method 2 reheated to 90°C

Gel preparation method Mixing temperature

Dissolution method 1

Dissolution method 2

Reheating to 90°C

(Pa.s)

(Pa.s)

(Pa.s)

50°C

429.2  10.93

66.43  21.23

350.8  10.36

70°C

495.7  32.41

95.11  28.33

309.9  16.83

90°C

692.1  14.07

419.4  28.15

298.0  9.121

82

Samples at 20 wt% water content prepared by dissolution method 2 show increasing viscosity with increasing mixing temperature. As the ternary systems are at the liquid to solid transition point, the viscosity decreases with increasing shear rate and no Newtonian plateau is observed. Instead, the samples exhibit a shear thinning profile characteristic of aggregate formation. By reheating the ternary systems to 90°C, the samples originally heated to 50°C and 70°C show an increase in viscosity, whereas reheating from a gel originally heated at 90°C decreased in viscosity. This may be due to weaker polymer-polymer interactions after reheating at high temperature when the gel was already exposed to 90°C heating in the first place. A characteristic of physical gels is that the non-covalent cross-links can reversibly form or break (Ross-Murphy 1995). In this case, thermal treatment led to the formation of new polymersolvent-nonsolvent interactions that led to a decrease in the elastic modulus. On the other hand, the two-phase systems formed at 50 and 70°C resulted in a homogeneous phase gel similar to that from dissolution method 1. Therefore, the results obtained from dissolution method 1 can be obtained with dissolution method 2 only by heating the gels to higher temperature. Figure 3.19 shows the response of the elastic moduli to stress for samples heated at different temperatures and reheated at 90°C. Samples heated at 50°C show a stepwise decrease in G at low oscillation stress, indicating weak network formation. After heating the sample to 90°C, however, the modulus increases and has a longer linear viscoelastic region and higher critical stress. Samples heated at 70°C and further reheated to 90°C also show an increase in elastic modulus and higher critical stress, leading to a more stable gel.

83

104

G' (Pa)

103

50°C 70°C 90°C reheat 50°C reheat 70°C reheat 90°C

102

101 10-2

10-1

100

101

102

103

osc. stress (Pa)

Figure 3.19. Stress sweep experiment for 10 wt% CA gels prepared at 50, 70 and 90oC as well as the corresponding gels reheated to 90oC at 20 wt% nonsolvent

The gels prepared using this CA/DMA/water ternary system have been shown to fractal in nature, therefore a power law behaviour should exist between G and nonsolvent volume fraction. Figure 3.20 shows that the power law exponents vary between the different preparation temperatures and the different methods of addition (i.e. dissolution method 1 and 2). In dissolution method 1, PS+N, the power-law exponent varied depending on the preparation temperature; about 38 for gels prepared at 50°C, 54 for gels prepared at 70°C, and 39 for the gels prepared at 90°C. Similarly, the gels prepared by dissolution method 2, SN+N, showed a similar trend wherein the power law exponents increased from 25 to 52 when mixed at 50°C and 70°C respectively, and decreased to 40 when mixed at 90°C.

84

105

50°C heating

104

y = 7E-48x38 R² = 0.97

G' (Pa)

103

y = 5E-31x25 R² = 0.99

102 PS+N

101

SN+P

100 16

21

26

water content (wt%) 105

105

70°C heating

104

90°C heating

104 y = 1E-49x40 R² = 0.96

2E-68x54

102

G' (Pa)

G'(Pa)

y= 103 R² = 0.94

y = 3E-65x52 R² = 0.97

101

103 102

y = 2E-49x39 R² = 0.96

101 PS+N

100

SN+P

10-1

PS+N

100

SN+P

10-1 16

18

20

22

water content (wt%)

24

16

18

20

22

24

water content (wt%)

Figure 3.20. G values (obtained at 1 rad/s) for 10 wt% CA ternary systems heated at (top) 50°C, (bottom left) 70°C, and (bottom right) 90°C

Previous studies with cellulose derivatives showed that phase separation in ternary systems with certain nonsolvent concentrations is dependent on the dielectric constant of the nonsolvent added (Khalil 1973). Water possesses a relatively high dielectric constant (ε) of 80.4 85

at 20°C and decreases with increasing temperature. For pure DMA, ε is about 38.9 at 25°C (Wohlfarth 2008). Therefore, adding water to DMA increases the dielectric constant of the binary mixture. Increasing the water content from 12.5 to 22.5 wt%, increases the dielectric constant from 40.2 to 42.2. Dielectric constants can dictate the type of phase separation; high dielectric constants produce flocculates, while lower dielectric constants produce gels. Lower ε is usually associated with lower polarity. At low nonsolvent content heated at low temperature, the binary mixture is expected to be more polar; with the addition of polymer, hydrophilic interactions are expected to dominate. With increasing temperature and nonsolvent content, lower dielectric constants lead to a more non-polar mixture and hydrophobic interactions are expected to dominate. Based on the rheological behaviour of the gels at the liquid to solid transition (Figure 3.5 and Figure 3.17), the formation of polymer aggregates is dependent on competitive hydrogen bonding and hydrophobic interactions between CA and the binary solvent. The molecular structure of the cellulose acetate in solution plays an important role in dictating the supramolecular properties of the ternary system. By modifying the cellulose acetate polymer to have hydroxyl groups present only at the C6 position and maintaining the same degree of subsitution, one may obtain an enhancement of the viscoelastic properties near the liquid to solid transition since the hydroxyl group at the C6 is the only position responsible for cluster formation. The polymer also becomes more flexible as there is no intramolecular hydrogen bonding interactions at the C2 and C3 positions. The insertion of more acetyl groups in the polymer chain increase the hydrophobicity of the polymer which might lead to higher elastic and viscous modulus at higher temperature.

86

3.4

Ternary system formation by varying polymer regioselectivity

3.4.1 Synthesis of regioselective cellulose acetate at the C2 and C3 positions

Synthesis of 2,3-di-O-acetylcellulose. In order to make a valid comparison between different cellulose acetates, they should have similar molecular weights and polydispersities as these factors can also influence its behavior in solution. Commercial cellulose acetate used in the previous ternary systems (Sections 3.2 and 3.3) was thus saponified to obtain cellulose with the same polymer backbone. Solvent exchange was necessary to facilitate dissolution of the polymer for subsequent modification. Cellulose is first activated in a polar medium, in this case water, so that the polymer chains swell and enhances the diffusion kinetics of the solvent to the tightly packed crystalline regions which are less accessible (Dupont 2003). In this process, many of the cellulose intra- and intermolecular hydrogen bonds are replaced by hydrogen bonds with H2O. In the next step methanol, a more non-polar solvent is used. Methanol expels the residual water from the system, as it hinders complexation and formation of cellulose aggregates which make dissolution more difficult. Finally, DMA is introduced to exchange the methanol and further impede the reformation of the cellulose intra- and intermolecular hydrogen-bonds. The resulting cellulose/DMA solution was then left to stir overnight to allow enough time for the solvent to penetrate the more crystalline regions of the polymer. Subsequent heat activation at 120°C creates enough vapour pressure to swell the fibre before salt addition leading to a clear colorless solution when dissolution is complete. The reaction was run under argon to prevent cellulose oxidative degradation at high temperature. Tritylation occurs via a SN2 mechanism where the nucleophilic oxygen from the cellulose hydroxyl group attacks the sp3 carbon center of the 4-methoxytriphenylmethyl chloride substrate and the chloride is subsequently ejected as the leaving group. In the case of cellulose, because 4methoxytriphenylmethyl chloride is a rather bulky molecule, and the various hydroxyl groups on the AGU are sterically hindered to varying degrees C3 > C2 >> C6, regioselective substitution takes place primarily at the primary C6 hydroxyl position. Elemental analysis of the resulting reaction product confirmed a trityl DS of 0.95 (69.4% C, 6.1% H, 24.5% O, see C4 in Appendix 87

for calculation). This result must be taken with precaution, however, as the calculation assumed no polymer degradation and the exact molecular weight of the product could not be determined as 1A was not soluble in THF (hence no GPC analysis). The trityl DS could not be determined by NMR techniques, as the presence of hydroxyl groups led to broad undefined peaks between 2.5 and 5 ppm in the 1H spectrum. The methoxy moiety in the 4-methoxytrityl group at 3,75 ppm also overlaps with the proton signals from the anhydroglucopyranose ring. 13C NMR analysis cannot provide quantitative information in this case as the ring carbon signals are too weak compared to the phenyl carbons. Acetylation of the remaining hydroxyl groups of the cellulose chains took place using acetic anhydride in DMA with DMAP as a catalyst (Tezuka & Tsuchiya 1995). The progress of the reaction can be monitored with 1H NMR in DMSO-d6 as shown in Figure 3.21. Introduction of the trityl group in cellulose and the production of 6-O-(4-methoxytrityl)-cellulose (1A) is confirmed by the presence of the aromatic peaks between 6.8 and 7.4 ppm and the methoxy functional group at 3.82 ppm. The following acetylation reaction leads to the formation of 2,3-diO-acetyl-6-O-(4-methoxytrityl)-cellulose (2A) and peaks at 1.89, 1.95, and 2.07 ppm which represent the acetyl methyl groups at the C2, C3, and C6 positions respectively (Goodlett et al. 1971). Removal of the trityl protection group leads to 2,3-di-O-acetyl-cellulose (3A) as confirmed by the disappearance of the aromatic and methoxy peaks.

88

1A

OCH3

H10, H9

H1 H4 H5

H11 H9’

H10’

H6 H3 H2

2A

CH3

H10, H9

H11 H9’ H10’

OCH3 H1 H3 H2 H6 H4H5

CH3

3A

H3

H2 H1

H6 H4 H5

Figure 3.21. Stacked 1H NMR plots of (1A) 6-O-(4-methoxytriphenylmethyl)-cellulose, 6TC; (2A) 2,3-di-O-acetyl-6-O-(4-methoxytrityl)-cellulose, 2,3Ac6TC; and (3A) 2,3-di-Oacetylcellulose. The peak at 2.5 is the DMSO-d6 solvent. 89

The deprotection step was performed underacidic conditions. The mechanism of trityl dissociation from the polymer involves a pre-equilibrium protonation of the trityl group followed by ionization and dissociation via ion-molecule pairs (SN1 cleavage) (López et al. 2008). The resulting trityl cation is a very stable species due to the inductive and resonance stabilizing effects of the three phenyl groups surrounding the central carbon atom, making the reverse reaction highly improbable. The reaction workup involved precipitation in a water: methanol mixture (1:4) which neutralizes the acidic solution and stabilizes the final product. Because the product was insoluble in common organic solvents such as THF and acetone, the product could only be characterized by NMR. The 1H NMR spectrum for compound 3A is shown in Figure 3.21 and the 13C NMR spectrum shown in Figure 3.22.

3A

C6

C3

C2

CH3

C3* C=O

C1

C2 C3 C4 C6 C5

Figure 3.22. 13C NMR spectrum of 2,3-di-O-acetylcellulose (3A). The insert from 169 to 172 ppm shows the carbonyl C=O peaks arising from the acetyl groups. C3*: acetyl carbonyl group at C3 with a neighboring C6 hydroxyl

Based on the calculated DS of trityl group (0.95) after the tritylation reaction, one would expect a highly regioselective polymer at the C6 position. However, based on the final product 3A after acetylation and deprotection, 13C NMR confirms this is not the case. The spectrum in 90 C5

Figure 3.22 shows four carbonyl peaks in the 169 to 172 ppm region. If substitution were highly regioselective with a trityl DS close to 1, only two carbonyl peaks would be present, one peak representing the acetyl carbonyl at the C2 and the other at the C3. The presence of four peaks indicates that there was some acetyl substitution at the C6 position. This could arise from discrepancies in the use of elemental analysis vs. NMR as well as loss of product during the synthesis of the intermediates, especially during purification where low molecular weight fractions of the polymer are washed away. Elemental analysis of compounds provides an empirical formula for the final product; the molecular formula can only be obtained by coupling with another technique to determine the exact mass of the molecule, such as mass spectrometry. In the case of cellulosic polymers, however, mass spectrometry is difficult as the polymers have low volatility and thermal instability. GPC is used only for a relative molecular weight determination. A thorough characterization of the final product is therefore needed to determine the exact extent of acetylation through NMR and elemental analysis by propanoation.

Synthesis of 2,3-di-O-acetylcellulose II. This synthesis proceeded with fewer steps as no solvent exchange was necessary to activate the starting material as the commercial cellulose acetate readily dissolved in DMA. Trityl chloride was the protecting group of choice as fewer C6 hydroxyl groups were available for substitution and the subsequent detritylation step showed a more facile trityl removal from 2,3Ac6TC II than the product originating from cellulose. Imidazole was used as the base for the acetylation reaction with acetic anhydride. Previous attempts using distilled pyridine (Tsunashima et al. 2001) have led to a dark brown product requiring many purification steps to obtain a white powder. Acetylation with imidazole prevented the formation of dark color in the mixture and the product precipitated as a white powder upon contact with methanol. The progress of the reactions was monitored by 1H NMR (Figure 3.23) as well as by FTIR (Figure 3.24). The FTIR spectrum of commercial CA (spectrum 1) clearly changes upon reaction with trityl chloride. Spectrum 2 showed the presence of aromatic C-C stretching at 1492 cm-1 and sharp C-H aromatic deformation bands between 755-660 cm-1 associated with the trityl protecting group, as well as a decrease in the hydroxyl stretching band at 3488 cm-1. Complete acetylation was confirmed by the disappearance of the hydroxyl stretching band at 3488 cm-1 spectrum 3. 91

1B CH3

H11, H10, H9

H3

H2 H1 H6 H4 H5

2B

CH3

H11, H10, H9

H3 H2

H1

H6 H4 H5

3B

CH3

H1 H3 H2 H4 H5 H6

Figure 3.23. Stacked 1H NMR plots of (1B) 6-O-triphenylmethylcellulose, 6TC II; (2B) 2,3-diO-acetyl-6-O-tritylcellulose, 2,3Ac6TC II; and (3B) 2,3-di-O-acetylcellulose II 92

C=O

CC ring

OH

1

OH A

CC arom

CC arom

CH arom 1B

CH 2B 4000.0

3600

3200

2800

2400

2000

1800 cm-1

1600

1400

1200

1000

800

600

450.0

Figure 3.24. Stacked FTIR plots of (1) commercial cellulose acetate, (1B) 6-O-triphenylmethylcellulose (6TC II), and (2B) 2,3-di-O-acetyl-6-O-trityl-cellulose (2,3Ac6TC II)

Characterization of cellulose acetates. To determine the distribution of acetyl substituents within the AGU, propanoation was used to completely substitute any free hydroxyl groups in the cellulose chain. Hydroxyl groups tend to form hydrogen bonds in polar solvents and have a broad range of chemical shifts in the 1H NMR spectrum. By introducing another ester moiety in the AGU, the polymer will have the necessary structural regularity to produce a clear spectrum (Goodlett et el. 1971). Using a 13C NMR technique, carbonyl signals of the cellulose esters formed will appear as clearly resolved triplets corresponding to the substitution position (C2, C3, and C6) on the glucose residue (Tezuka & Tsuchiya 1995). In particular, the propanoyl carbonyl signals will have different chemical shifts from that of the acetyl carbonyls. Commercial cellulose acetate was propanoated and the 13C NMR spectrum is shown in Figure 3.25.

93

169.29

Acetyl C=O C2

Propanoyl C=O C6

C6

solvent

C4 C5 C2 C3

169 ppm

170

C1

171

3.70 3.12

172

3.62

173 0.61 0.64

1.00

Acetyl C=O

Propanoyl C=O

174

Propanoyl CH2

C3 C2

Propanoyl CH3

C3

Acetyl CH3

C6

169.74

170.22

172.78

173.19

173.66

4

Figure 3.25. 13C NMR spectrum of propanoated commercial cellulose acetate (4) in CDCl3 at 75 MHz at 25°C. The acetyl and propanoyl C=O triplet signals are shown in the inset

As illustrated in the inset, there are two kinds of carbonyl triplet signals between 169 and 174 ppm assigned to the propanoyl (down field) and acetyl C=O (upfield) (Tsunashima & Hattori 2000). The three peaks within each triplet are assigned from downfield to upfield to the C6, C3, and C2 positions. The intensity of the acetyl C=O peaks represents the amount of substituted hydroxyl groups at the C2, C3, and C6 positions, while the intensity of the propanoyl C=O peaks denotes the amount of unsubstituted hydroxyl groups present in the original CA. The degree of acetylation at each individual position within the AGU can then be estimated from the integrated intensity ratio of acetyl to propanoyl peaks at each of the C2, C3, and C6 positions. In the case of commercial CA, C2, C3, and C6 have an individual degree of acetylation of 0.78, 94

0.86, and 0.83 respectively, leading to the polymer having a total DS of 2.47. It is important to note that both acetyl and hydroxyl functional groups are present at the C2, C3, and C6 positions allowing for both inter- and intramolecular hydrogen bonding. The same procedure to determinine individual DS can be applied to the synthesized cellulose acetates. Propanoation also makes the products soluble in common solvents such as chloroform, THF, and acetone. Consequently, gel permeation chromatography (GPC) molecular weight analysis using THF as the eluent was possible. GPC data determined minimal polymer degradation during the synthesis steps and the relative number average molecular weight was kept at around 50,000 for both of the final products (see Figure A32 in Appendix). Figure 3.26 shows the 13C NMR spectra of propanoated 2,3-di-O-acetylcellulose (4A) synthesized from cellulose and cellulose acetate respectively. The spectra show that both compounds are regioselectively substituted at the C2 and C3 positions due to the lack of propanoyl C=O peaks and contain some acetyl groups at the C6 position. Specifically, for 2,3-diO-acetylcellulose (top) the individual DS at C2, C3, and C6 are 1, 1, and 0.40 respectively, and for 4B (bottom) the respective individual DS are 1, 1, and 0.80. 2,3-di-O-acetylcellulose will now be referred to as regio2.4 CA and 2,3-di-O-acetylcellulose II will be regio2.8 CA. Regio2.4 CA was predicted to contain fewer acetyl groups at the C6 position taking into account the high degree of trityl regioselectivity at 0.95 from elemental analysis. One reason may be due to the uncertainty in the use of EA as a viable technique to characterize the product as mentioned in the previous section; another reason may be due to incomplete trityl substitution during the synthesis of 1A. As the starting material was relatively high molecular weight CA, the cellulose produced after saponification took longer to dissolve compared to that of lower molecular weight (Mn 30K). Higher molecular weight also leads to bigger cluster formation in DMA/LiCl solvent which leads to lower accessibility of C6 hydroxyl groups to the trityl chloride. Regio2.8 CA did not possess the same problem, however, as the DS at the C6 was ~0.80 just like the original starting material.

95

171 170 169.25

C4C5 C3 C2

C2

C6

Acetyl CH3

C6

Acetyl C=O

Propanoyl CH3

Propanoyl CH2

C6

solvent

C3

169 ppm

Propanoyl CH3

C6 Acetyl CH3

C6

Propanoyl CH2

172

169

C6

Propanoyl C=O 170

C4 C5 C2 C3

173

171

solvent

174

172

1.61 1.54

173

169.69

0.67

170.17

C1

1.00

4B 173.62

174

3.94 5.01 4.43

1.00

Acetyl C=O

Propanoyl C=O

Propanoyl C=O

C1

Acetyl C=O

Propanoyl C=O

169.27

169.72

170.18

173.63

4A

Acetyl C=O

ppm

C3 C2

Figure 3.26. 13C NMR spectra of propanoated 2,3-di-O-acetylcellulose 4A (top) and 4B (bottom)

96

3.4.2 Role of regiochemistry in ternary system formation

3.4.2.1 Commercial CA vs. regio2.4 CA

Both regio2.4 and regio2.8 CA dissolve readily in DMA. Unlike commercial CA, however, the synthesized cellulose acetates do not contain any hydroxyl groups at the C2 and C3 positions, and are thus devoid from forming intramolecular hydrogen bonds at those positions. This effect makes the polymer less rigid as only the C6 position is involved in hydrogen bonding. Although the C6 hydroxyl group cannot participate in intramolecular hydrogen bonds, it can participate in intermolecular hydrogen bonding with C6 hydroxyl group of adjacent CAs and/or polar solvents leading to cluster formation (Tsunashima et al. 2001). The amount and the sequential distribution of the unsubstituted C6 hydroxyl groups on the CA chain will strongly affect clustering in solution. Figure 3.27 shows a stress sweep experiment to determine the critical stress and linear viscoelastic region of the gels heated to 50°C by dissolution method 1. The elastic modulus of the gels as well as the critical stress increases with increasing water content. Comparing the regioselective polymer with a DS of 2.4 to that of the commercial CA with DS ~ 2.5 and similar molecular weight, it can be shown that regio2.4 CA forms weaker networks in DMA than commercial CA (see Figure 3.7). At the same water content (e.g. 20 wt%), both the commercial and regio2.4 CA exhibit an average G of approximately 1120 Pa, but the critical stress for commercial CA is 430 Pa while the regio2.4 CA yields at 12 Pa. However, the regio2.4 CA undergoes gelation at much lower nonsolvent content than the commercial CA, exhibiting a solgel transition at 12 wt% nonsolvent (regio2.4 CA) vs 18 wt% nonsolvent (commercial CA). At lower nonsolvent contents both the regio2.4 and commercial CA ternary system shows a stepwise decrease in G, characteristic of weak aggregates.

97

104 15 wt% 20 wt% 25 wt%

G' and G" (Pa)

103

102

101

100 10-2

10-1

100

101

102

103

osc. stress (Pa)

Figure 3.27. Stress sweep experiment for 10 wt% regio2.4 CA samples with 15, 20, and 25 wt% water prepared at 50°C (frequency = 1 Hz). G is represented by solid filled shapes while G is unfilled

The G and G crossover occurs at 17, 32, and 113 Pa for samples prepared at 50°C with 15, 20, and 25 wt% water respectively (Figure 3.27). As with commercial CA, the systems’ viscoelastic properties are enhanced with increasing nonsolvent content. Increasing the temperature led to a G and G crossover at 17, 67, and 125 at 70°C heating, and 86, 118, and 226 Pa at 90°C heating for samples prepared at 15, 20, and 25 wt%, respectively (see Figure B1 and Figure B2 in Appendix). The regio2.4 CA, however, exhibits a shorter linear viscoelastic region than the commercial CA. By plotting the elastic modulus vs. percent strain, one can see that the gels are strong-linked as the limit of linearity decreases with increasing nonsolvent concentration. Figure 3.28 shows the effect of nonsolvent content on G as a function of percent strain from samples prepared by heating to 50 and 90°C. The same phenomenon was observed for 70°C prepared gels (Figure B3 in Appendix). The limit of linearity, as defined in the 98

previous section, is the value of strain at which there is an appreciable deviation in the elastic modulus from the linear viscoelastic plateau. It was measured as the intersection between the tangent of the drop in G and the linear viscoelastic region in the G vs. % strain plot. The percent strain limit for gels with 15, 20, and 25 wt% nonsolvent are 19.8, 1.8, and 1.3% for 50°C heating, and 10.7, 3.2, and 2.5% for 90°C heating, respectively.

105

G' (Pa)

104

103

102 15 wt% 20 wt%

101

100 10-6

25 wt%

10-5

10-4

10-3

10-2

10-1

100

101

102

% strain

Figure 3.28. Elastic modulus as a function of % strain for 10 wt% regio2.4 CA gels prepared at 50°C (unfilled shapes) and 90°C (filled shapes) (frequency = 1 Hz)

Regardless of temperature, the limit of linearity shifts to lower strain as nonsolvent concentration increases, an indication of strong-linked gels. Regio2.4 gels follow the same pattern as that of commercial CA; that is, G increases with increasing heating temperature. At 15 wt% water, the system forms a gel, as indicated by a frequency sweep experiment where G > G (Figure 3.29). Due to the presence of hydroxyl groups at the C6 position which participate in 99

cluster formation, hydrogen bonding takes place selectively between these sections with the help of the polar DMA solvent. The increase in water content lead to enhanced hydrogen bonding and hydrophobic nonbonding interactions which explain the higher elastic modulus at lower nonsolvent content compared to the commercial CA.

104

G' and G" (Pa)

103

102 15 wt%

101

20 wt% 25 wt%

100 10-1

100

101

102

ang. frequency (rad/s)

Figure 3.29. Frequency sweep experiment for regio2.4 CA samples at 10 wt% CA prepared at 50°C. Elastic moduli are represented by filled shapes while viscous moduli are shown as unfilled shapes.

To determine the effect of temperature, samples with 15 wt% water were heated at 50, 70, and 90°C. There is a clear distinction between the three samples heated at different temperatures, as shown in Figure 3.30. As with commercial CA, the sample heated at 90°C shows a longer stress independent region compared to those heated at 50 and 70°C. The system also exhibits a higher critical stress, making it less prone to rupture. Similar behavior was observed for gels at 20 and 25 wt% nonsolvent (Figure B4 and B5 in Appendix). 100

G' and G" (Pa)

103

102

101

50°C 70°C 90°C

100 10-2

10-1

100

101

102

103

osc. stress (Pa)

Figure 3.30. Stress sweep experiment for 10 wt% regio2.4 CA samples prepared at 50, 70 and 90°C and 15 wt% nonsolvent content (frequency = 1 Hz). G is represented by solid filled shapes while G is unfilled

Acetylation increases the hydrophobic character of cellulose. Introduction of the hydrophobic acetyl group at the C3 position disrupts C3-O5’ intramolecular hydrogen bonding and increases polymer mobility (Bochek & Kalyuzhnaya 2002). Therefore, one explanation for the observed longer stress independent region and higher critical stress with increasing temperature temperature may be that, the chain conformation is dictated by hydrophobic interactions located within the macromolecular structure, exposing the acetylated regions to the polymer-solvent interface (Tsunashima et al. 2002). Specifically, the polymer rearranges itself in solution in such a way that the hydroxyl groups locate within the macromolecular structure, exposing the acetylated regions to the polymer solvent interface At lower temperatures, more C6

101

hydroxyl groups are present on the chain surface allowing for polar solvents to interact with the hydroxyl groups to mediate hydrogen bonding (Tsunashima et al. 2002). Both acetyl and hydroxyl groups promote strong inter- and intramolecular interactions in CA (Puleo et al. 1989). It is also known that changing the substituent distribution within a polymer can influence polymer rigidity and viscosity at a constant DS (Ferry 1980). In the commercial CA, intramolecular hydrogen bonds are present between the C3’ hydroxyl proton and the neighbouring O5 ring oxygen as well as between the C2 hydrogen and the neighbouring C6’ oxygen. As hydroxyl groups are only present at the C6 position of the regio2.4 CA, it is expected not to form such intramolecular hydrogen bonds, however, it has been suggested that in CA the C6 hydroxyl proton can form an intramolecular hydrogen bond with the carbonyl oxygen of the C2 acetyl group (Kamide et al. 1981). Although intramolecular associations do not participate in cluster formation, they do influence chain conformation and stiffness in solution. As a result, regio2.4 CA is a more flexible polymer than commercial CA. The presence of intermolecular hydrogen bonding and hydrophobic interactions, however, are more prevalent in regio2.4 CA as it contains more hydroxyl groups at the C6, and therefore have a greater propensity to aggregate and form macromolecular clusters. Commercial CA with 15 wt% water is still in the solution phase while regio2.4 CA forms a gel at the same nonsolvent content. However, the commercial CA possesses a longer linear viscoelastic region (i.e. drop in modulus occurs at higher stress) and is less prone to rupture, as the elastic modulus is observed to be relatively independent of applied stress, while a small stepwise decrease in G can be observed for regio2.4 CA before gel rupture (Figure 3.31). In the case of the regio2.4 CA, the stepwise decrease in G is characteristic of weak network formation (Winter & Mours 1997). The system does not rupture completely, rather the interactions between aggregates, including intermolecular hydrogen bonding, are more sensitive to deformation than in the commercial CA. As mentioned before, the regio2.4 CA gels at 20 and 25 wt% nonsolvent have lower critical stress than the corresponding materials from the commercial CA.

102

105

G' and G" (Pa)

104 103

102 101 15 wt% 20 wt%

100

25 wt%

10-1 10-2

10-1

100

101

102

103

osc. stress (Pa)

Figure 3.31. Elastic modulus under applied stress for 10 wt% regio2.4 CA (filled shapes) and commercial CA (unfilled shapes) prepared by heating to 90°C

Although the commercial CA and regio2.4 CA have the same G (~ 3000 Pa at 20 wt% water content) (Figure 3.8) they have dramatically different viscosities; commercial CA prepared at 90 and 50°C have higher viscosity than the corresponding regio2.4 CA (Figure 3.32). From dynamic light scattering studies it has been shown that hydrophobic interactions dominate at higher temperature (Tsunashima et al. 2002). At elevated temperature the hydroxyl groups in the CA chain are less prevalent around the polymer-solvent interface where interactions between the solvent and nonsolvent occur. Instead, they protect the chains from precipitation and induce dynamical clustering structures (Tsunashima et al. 2002).

103

104 50°C

viscosity (Pa.s)

103

90°C

102 101 100 10-1 10-2

10-1

100

101

102

103

shear rate (1/s)

Figure 3.32. Viscosity for 10 wt% regio2.4 CA and commercial CA gels at 20 wt% nonsolvent prepared at 50 and 90°C. Regio2.4 gel viscosity values are represented by solid shapes while commercial CA are the unfilled shapes

Frequency sweep experiments for gels formed at 90°C show a clear difference in the viscoelastic properties when made from regioselective CA as compared to commercial CA (Figure 3.33). Because of the increased intermolecular interactions in the regioselective polymer, arising from more C6 hydroxyl groups available to interact with the polar solvent and nonsolvent, the ternary system possesses higher elastic and viscous modulus at low nonsolvent content (15 wt%). The same phenomenom persists at 20 wt% nonsolvent, where the ternary system with regio2.4 CA is a solid gel and that with commercial CA is at the liquid to solid transition (Figure 3.5).

104

103

104

101 100 regio2.4 CA 15 wt% nonsolvent commercial CA 15 wt% nonsolvent

10-1 10-2 10-3 10-1

100

101

102

ang. frequency (rad/s)

G' and G" (Pa)

G' and G" (Pa)

102 103 102 regio2.4 CA 20 wt% nonsolvent

101

commercial CA 20 wt% nonsolvent

100 10-1

100

101

102

ang. frequency (rad/s)

Figure 3.33. Elastic (G, filled shapes) and viscous (G, unfilled shapes) moduli for 10 wt% regio2.4 and commercial CA at 15 wt% (left) and 20 wt% (right) water contents (samples were prepared at 90°C)

3.4.2.2 Effect of increasing acetyl substitution

As previously mentioned in Section 3.2, hydrogen bonding is not the only factor in cluster formation (Tsunashima et al. 2002). With the presence of more acetyl groups, the polymer becomes more hydrophobic and can induce gelation at lower nonsolvent content. Hydrophobicity tends to lead to stronger intermolecular interactions in solution and to the formation of association superstructures. At the extreme, i.e. full acetyl substitution of the hydroxyl groups, these interactions are so strong that phase separation occurs and precludes gel formation. The absence of hydrogen bonding may be an indication that despite hydrophobic interactions playing a role in inducing gelation, hydrogen bonding interactions are the origin of gel formation. Previous studies have shown that a CA ternary system without a protic nonsolvent cannot undergo gelation (Kadla & Korehei 2010). The associated superstructures formed by increasing the number of acetyl groups are generally not stable to shear as the contact surface between the hydrophobic side chains is relatively small (Clasen & Kulicke 2001). Increasing the 105

number of acetyl groups from 2.4 to 2.8 in the polymer lead to three-dimensional physical crosslinking at lower nonsolvent content. To investigate the effect of increasing acetyl content and hydrophobicity on intermolecular interaction and associated cluster formation, gels were prepared using regio2.4 and regio2.8 CA. The CA/DMA/H2O ternary system creates strong-linked gels; however, heating at 50°C leads to networks with relatively weaker interfloc links as compared to 70 and 90°C, which are relatively similar (Figure 3.11). Regio2.4 and regio2.8 CA gels were prepared at 70°C. Figure 3.34 shows the effect of nonsolvent content on the viscous and elastic moduli for the regioselective CA with a DS of 2.8 (regio2.8 CA) ternary system.

104

G' and G" (Pa)

103 102 101 100

10 wt% 12.5 wt% 15 wt%

10-1

17.5 wt%

10-2 10-1

100

101

102

ang. frequency (rad/s)

Figure 3.34. Frequency sweep experiment for 10 wt% regio2.8 CA gels prepared at 70°C with 10-17.5 wt% nonsolvent. G is represented by solid filled shapes while G is unfilled

106

Increasing the nonsolvent content in the regio2.8 CA system from 10 – 17.5 wt% resulted in phase separation and gel formation at ~ 12.5 wt% nonsolvent (G > G). Further increasing the nonsolvent content to 15 wt% increased the moduli. However, the gels at 15 and 17.5 wt% nonsolvent show relatively similar G and G values. This indicates that the ternary system has reached its saturation point and gels with increasing water content will not significantly influence viscoelastic moduli. The G (1 rad/s) for the regio2.8 CA was substantially higher than that for the regio2.4 CA at the same nonsolvent content, 15 wt%, ~ 6500 Pa vs. 340 Pa, respectively. The saturation point for regio2.8 CA gels at 17.5 wt% water exhibited a G of ~ 5400 Pa at 1 rad/s, while regio2.4 CA gels do not reach saturation point until 25 wt% water, in this case with a G of ~ 1300 Pa. Increasing the acetyl DS at the C6 position can significantly influence the viscoelastic properties of the ternary system formed. With fewer hydroxyl groups in the polymer chain, there is a greater possibility for intermolecular aggregation. Hydrophobic interactions between the polymer and solvent influence network formation at high temperature and compete with hydrophilic OH groups for interaction with DMA. Although non-regioselective CA at the same DS of 2.8 would be expected to show higher viscoelastic moduli than that of lower DS (Flory et al. 1958), the influence of regioselective substitution at the C2 and C3 positions is still present as the polymer exhibits more chain flexibility without intramolecular hydrogen bonding between the C3-O5 and C2-O6. In addition to polymer concentration and solvent quality, the polymer chemical structure plays an important role in determining the rheological properties of cellulose acetate in a ternary system. The differences in the acetyl substituent distribution within the AGU result in a change of the intermolecular interactions and aggregation behavior. At a polymer concentration of 10 wt% and water content of 15 wt%, regioselective CA with a DS of 2.8 showed a higher tendency to aggregate based on higher viscoelastic moduli (Clasen & Kulicke 2001). Figure 3.35 shows the viscous and elastic moduli as a function of stress for the three CA samples, commercial, regio2.4, and regio2.8 CA prepared at 70°C and 15 wt% nonsolvent content.

107

104

G' and G" (Pa)

103 102 101 100

commercial CA regio2.4 CA

10-1 10-2 10-2

regio2.8 CA

10-1

100

101

102

103

osc. stress (Pa)

Figure 3.35. Stress sweep experiment for 10 wt% cellulose acetate samples (commercial CA, regio2.4 CA, and regio2.8 CA) at 15 wt% nonsolvent prepared at 70°C. G is represented by solid filled shapes while G is unfilled

It can be seen that at low oscillation stress the regio2.4 and regio2.8 CA have higher moduli than that of the commercial CA. At 15 wt% nonsolvent content both regioselective cellulose acetates have already passed the gelation point, while the commercial CA remains a solution. Although the commercial CA possesses a relatively longer linear viscoelastic region than that of regio2.4 CA, regio2.8 CA shows both higher G and G as well as a longer stress independent behaviour. In the case of regio2.8 CA, hydrophobic interactions dominate over hydrophilic interactions, leading to enhanced viscoelastic properties. The increase in acetyl content coupled with increased chain flexibility due to the absence of intramolecular hydrogen bonding between the C2 and C3 positions in the polymer cause further intermolecular interactions with the solvent-nonsolvent leading to stronger gels. Both regio2.4 and commercial CA need a water content of 20 wt% to reach the same G as that of regio2.8 CA at 15 wt% water 108

(Figure 3.36). In this case, the linear viscoelastic region is longest for commercial CA, followed by regio2.8 and regio2.4 CA in decreasing order.

105

G' (Pa)

104 103 102

101 100 10-2

regio2.4 CA 20 wt% nonsolvent regio2.8 CA 15wt% nonsolvent commercial CA 20 wt% nonosolvent

10-1

100

101

102

103

104

osc. stress (Pa)

Figure 3.36. Stress sweep experiment for 10 wt% cellulose acetate samples at approximately the same G (3700 Pa) after heating to 70°C. Regio2.4 CA contained 20 wt% nonsolvent, regio2.8 contained 15 wt% nonsolvent, and commercial CA contained 20 wt% nonsolvent

The shorter LVR for both 2.4 and 2.8 regioselective CA can be explained by the higher chain flexibility compared to the commercial CA. As these are strong-linked gels, gel failure occurs at the intrafloc domain as opposed to the interfloc domain. The commercial CA has more extensive intramolecular hydrogen bonding, e.g. via C2 and C3 hydroxyl groups which leads to higher chain stiffness and ultimately higher critical stress. In the case of regio2.4 and regio2.8 CA, which have less intramolecular interactions, regio2.8 CA exhibits a higher critical stress than regio2.4 CA. In this case regio2.8 CA has fewer C6 hydroxyl groups available to participate in intermolecular interactions, suggesting that hydrophobic intermolecular interactions play a more important role in dictating the critical stress. 109

Figure 3.37 shows the difference in stress behaviour between the three CA systems at the same nonsolvent content. At this nonsolvent content of 20 wt%, all three systems are gels. As expected, the moduli of the regio2.8 CA is significantly greater than that of regio2.4 and commercial CA. Under this condition the critical stress of regio2.8 CA is now comparable to that of the commercial CA. As these polymers have the same amount of hydroxyl groups at the C6, there seems to be a relationship between the amount of C6 hydroxyl and critical stress; critical stress increasing with decreasing C6 hydroxyl group. Similarly, by decreasing the amount of hydroxyl groups and thereby increasing the hydrophobicity of the polymer, i.e. comparing regio2.8 vs. commercial CA, G is increased. However, there does not appear to be a relationship to the amount of hydroxyl groups at C6 and G as regio2.4 CA and commercial CA have the same G.

105 104

G' (Pa)

103 102 commercial CA

101 100

10-1 10-2

regio2.4 CA regio2.8 CA

10-1

100

101

102

103

104

osc. stress (Pa)

Figure 3.37. Stress sweep experiment for 10 wt% cellulose acetate samples (commercial, regio2.4, and regio2.8 CA) at 20 wt% nonsolvent prepared at 70°C

110

Molecular structure and conformation of the polymer in solution is crucial for the determination of flow behavior in polymer solutions. The conformation of the polymer in solution primarily depends on the chemical structure of the polymer (Clasen & Kulicke 2001). Regioselective acetylation increases the polymer flexibility and intermolecular hydrogen bonding interactions leading to higher viscoelastic moduli. The addition of more hydrophobic acetyl groups to the polymer further increased the G and G at the same nonsolvent concentration, highlighting the importance of hydrophobic nonbonding interactions in the ternary system.

111

4

Conclusions

Cellulose derivatives such as cellulose acetates are soluble in common solvents and possess a glass transition and melting temperature lower than its decomposition temperature. They have found its way into a variety of commercial applications including coatings, filters, membranes, and films. Cellulose acetate in solution can form sophisticated network structures when immersed in a nonsolvent coagulation bath during the process of membrane formation (Reuvers et al. 1986). Depending on the nature of the solvent and nonsolvent, a sol-gel transition can be observed with increasing concentration of nonsolvent and result in phase separation induced gelation. Previous work by Appaw et al. (2007) used commercially available cellulose acetate (CA) dissolved in N,N-dimethylacetamide (DMA) to create a ternary system with water as the nonsolvent. With increasing polymer and nonsolvent concentration, enhanced steady state and dynamic rheological properties were observed. By adding a hydrogen bond donor and acceptor such as water, an increase in the hydrogen bonding intensity between ternary system components led to gelation at concentrations of approximately 20 wt% water. Kadla et al. (Kadla & Korehei 2010) used a variety of alcohols as nonsolvents and proposed that an intensification of hydrophobic and hydrophilic components in the ternary system led to enhanced viscoelastic properties. Hydrophilic and hydrophobic interactions can also be dictated by the nature of the cellulose acetate polymer as opposed to the nonsolvent. By manipulating the different conformations of the polymer in solution with varying temperature, different viscoelastic properties result during the formation of a CA ternary system. Temperature also affects the rate of solubilization of the polymer in a binary mixture, leading to gels with decreased viscoelastic properties. The molecular properties of CA also affect the hydrophobic and hydrophilic interactions with the solvent and can be manipulated to increase the viscosity and dynamic rheological behavior.

112

4.1

Effect of ternary system mixing temperature

A series of cellulose acetate ternary systems with DMA solvent and water nonsolvent were prepared by varying the mixing temperature. The cellulose acetate polymer had a molecular weight Mn of approximately 50,000 and a degree of substitution of 2.45. The polymer concentration in the ternary system was kept at 10 wt% while the nonsolvent concentration was varied to reflect the liquid to solid transition process with increasing water content. Based on steady state rheology measurements, temperature changes indicated the presence of aggregates in solution at 50°C (low temperature) but not at 70 and 90°C (high temperature). Although cellulose acetate does not completely dissolve even in dilute solutions, cluster formation is more evident at low temperature. Dynamic frequency sweep measurements showed higher viscous (G) and elastic (G) modulus when the samples were mixed at lower temperature, consistent with the fact that cellulose acetate possesses a lower critical solution temperature. Cellulose acetate dissolves molecularly at low temperature but undergoes a phase separation process at about 65°C leading to the formation of aggregates. Water, a hydrogen bonding donor, builds strong interactions with the solvent and polymer. However, as nonsolvent content increases, the solvent quality increases as the polymer is less soluble in the binary system and leads to phase separation at about 20 wt% water. Thermal treatment increases interactions between solvent and polymer leading to a more homogeneous mixture. As evidenced by stress sweep experiments, during the liquid to solid transition at 20 wt% water, interactions between ternary system components are stronger for gels heated at higher temperature (higher G) and possess a higher critical stress than those heated at lower temperature. Because the ternary system was subject to aggregated network formation with increasing water content, the gels could be viewed as a collection of flocs with the application of scaling theory. Subjecting the samples to strain indicated that the gels were strong-linked in nature, meaning that the interactions between the flocs (interflocs) were stronger than those within the flocs (intraflocs). The limit of linearity decreased as the water content increased; therefore the breaking of the aggregates occured from within the flocs. As predicted, the linear elastic region vanished at lower percentage strain with increasing water content. Plotting the limit of linearity 113

against nonsolvent content, gels heated at higher temperature showed more sensitivity towards nonsolvent addition than those heated at 50°C.

4.2

Effect of ternary system dissolution method

The previous method for forming a ternary system was by adding nonsolvent to a homogeneous solution of polymer and solvent. This procedure was employed as it is also the common method for the formation of membranes, where polymer solutions are immersed in a nonsolvent leading to porous structures. A different method of forming ternary systems, noted in this thesis as “dissolution method 2,” led to the formation of a two-phase phase separated system at high nonsolvent contents heated at low temperatures. In this case, the polymer was added to a binary mixture of varying concentrations of DMA and water. Low temperature thermal treatment led to two phase formation, while those gels formed at high temperature led to homogeneous gels similar to those made in the first part of this study (dissolution method 1). Rheological properties between both dissolutions methods created different viscoelastic properties. At low nonsolvent content, the ternary system remained a solution and possessed approximately the same viscosity values as those created with dissolution method 1. However, at nonsolvent content around 15 wt% the viscosity values were lower for dissolution method 2 for all temperatures. Once at the liquid to solid transition, samples heated at lower temperature tend to separate into two phases: a dense polymer-rich phase that settles at the bottom of the sample vial and a liquid phase consisting of mainly solvent and nonsolvent at the top. Samples heated at high temperature, however, form a homogeneous gel with similar viscoelastic properties as those made by dissolution method 1. It is predicted that the kinetics of gel formation and the kinetics of phase separation compete for network formation, and at lower temperature the activation energy for gelation is too high to form a homogeneous phase separated gel. To prove the kinetics of phase separation, gels heated at low and high temperatures were all subjected to a thermal treatment of 90°C. In this case, the gels previously heated at 50 and 70° showed a one-phase phase separated gel with higher viscous and elastic moduli. The results then 114

lead us to conclude that the gels prepared by dissolution method 2 lacked thermal reversibility and are in a metastable state until further heating lead to its most stable thermodynamic state. The difference in viscoelastic properties can stem from how soluble the ternary system components are in solution. The addition of water nonsolvent decreases the solvent power of DMA. Therefore, higher water content made the polymer less soluble decreasing the diffusivity of particles in solution. The separation of the ternary system into two phases was thus a consequence of the kinetics of phase separation being faster than the kinetics of gelation, where the polymer diffused into the continuous solid medium and water interacted with the solvent in the liquid dispersed phase.

4.3

Effect of varying degrees of acetylation at the C6 position

Cellulose is a linear polymer consisting of D-anhydroglucopyranose units (AGU) linked together by β-(1,4)-glycosidic bonds. Each AGU contains hydroxyl groups at the 2, 3, and 6 positions which can influence its behaviour in solution. Hydroxyl groups at the 2 and 3 positions are responsible for intramolecular hydrogen bonding in the cellulose polymer and contribute to chain stiffness. Hydroxyl groups at the 6 position are responsible for intermolecular hydrogen bonding and play an important role in cluster formation. By acetylating the 2 and 3 hydroxyl positions, a regioselective cellulose acetate was synthesized which contained no intramolecular hydrogen bonding at the C2 and C3 positions. A few acetyl groups were present at the C6 position, but still contained a majority of C6 hydroxyls capable of intermolecular hydrogen bonding with the solvent. The lack of intramolecular hydrogen bonding at the C2 and C3 positions led to a more flexible cellulose acetate polymer which was less restricted to rearrange itself in polar solvent (DMA). The consequence of this behavior resulted in more C6 hydroxyl groups present at the chain surface to interact with DMA and subsequently water, leading to gelation at lower nonsolvent content due to the strong interaction between polymer and solvent. Stress sweeps indicated higher elastic and viscous moduli at the same water content compared to the ternary

115

system containing commercial CA. Regioselective CA, however, possessed a lower critical stress and shorter linear viscoelastic region despite the gels being strong-linked. Scaling theory characterizes strong-linked gels as an aggregated network where particle interactions between the flocs (interflocs) are stronger than those within the flocs (intraflocs). In this case, failure under deformation occurs when intrafloc linkages break. Elastic moduli of regioselective CA show that with increasing water content, the limit of linearity shifts to lower strain. The formation of strong-linked gels can be correlated with intermolecular hydrogen bonding interactions between the polymer and DMA/H2O. The lack of intramolecular hydrogen bonding in the polymer led to a lower critical stress compared to commercial CA, which possessed hydroxyl groups at the C2, C3, and C6 positions. Acetyl groups in cellulose acetate increase the hydrophobicity of the polymer. The ternary system made with regioselective CA with a DS of 2.8 in DMA/H2O exhibited higher viscoelastic properties than that made with regioselective CA with a DS of 2.4 as well as commercial CA. Cluster formation is usually dictated by the presence of hydroxyl groups at the C6 position, but after a certain point, hydrophobic interactions take precedence. Aggregates form with increasing water content and form stronger gels (higher G and G) and exhibit a longer linear viscoelastic region for the regio2.8 ternary system despite the lack of C2 and C3 hydroxyls. Acetyl groups thus also play a role in cluster formation while the hydroxyl groups prevent the polymer from precipitating in solution. Therefore, both hydrophilic and hydrophobic interactions are important in the formation of phase separated gels.

116

5

5.1

Future work

Recommendations for future work

Phase diagrams of the mechanism of gel formation. Thermodynamic phase diagrams can give insight on the mechanism of liquid-liquid phase separation in a ternary system based on the binodal and spinodal limits. Using Flory-Huggins theory a series of equations can be obtained for a three component system. So far, there have not been any reports on a CA/DMA/nonsolvent system. The phase diagram can provide information on how the concentration-dependent interaction parameters between polymer-solvent, polymer-nonsolvent, and solvent-nonsolvent affect demixing. Temperature plays a role in the determination of the binodal and spinodal as phase separation is influenced by the heating and cooling process of the system. The theoretical work can help design ternary systems with the appropriate concentrations of each component for the formation of gels with desired microstructure and viscoelastic properties.

NMR diffusion studies. NMR has proven to be a useful tool in structure elucidation, but can also provide information on how the solvent penetrates the polymer based on T2 spin-spin relaxation rates of the hydrogen nuclei of the solvent. The diffusion curve obtained can give insight on the rotational and translational mobility of the polymer in solution, which is dependent on factors such as molecular weight, hydroxyl content, solvent quality, and concentration. Thus, the gelation process in a CA/DMA/H2O ternary system on increasing nonsolvent content can be monitored by measuring the solvent diffusion coefficients with respect to time. This technique further complements mechanistic gelation studies.

Static and dynamic light scattering studies of regioselective 2,3-di-O-acetylcellulose. Static light scattering is commonly used to obtain the average molecular weight of a macromolecule, in this case a polymer. Multi-angle static light scattering can be applied to obtain the radius of gyration (Rg) of the polymer. Coupled with dynamic light scattering, which determines the hydrodynamic radius (Rh), the structure of the polymer in either dilute or 117

concentrated solution can be obtained. The ratio Rg/Rh is a measure of the branching density, polydispersity, and inherent flexibility of the polymer. It would be interesting to see the contrast between polymers that are acetylated at all three positions compared to those acetylated selectively. The light scattering studies characterize further the behavior of cellulose acetate in solution and help explain how significant of a role chain flexibility and substitution pattern play in the formation of a ternary system leading to gelation.

Synthesis of 6-O-acetylcellulose and its behavior in a CA/DMA/H2O ternary system. The hydroxyl group at the C6 position can form intermolecular interactions in polar solvents leading to cluster formation. Substituting the C6 position with an acetyl group and leaving the C2 and C3 positions with hydroxyl groups would form a polymer with different solubility properties as the product would not be able to form intermolecular hydrogen bonds with polar solvents. Previous work on making C6 regioselective derivatives was published by Kondo (1993) on 6-Oalkylcelluloses. A similar scheme can be proposed where the cellulose is first tritylated at the C6 position followed by benzylic substitution at the C2 and C3 using benzyl chloride with a strong base such as sodium hydroxide in DMSO. The C6 position can then undergo a detritylation step in strong acid so that further acetylation selectively occurs at the primary alcohol position. The benzylic protecting groups can be removed by hydrogenation using a palladium-carbon catalyst. Despite the increased chain stiffness arising from intramolecular hydrogen bonds, it is expected that when mixed with DMA and water to create a ternary system the outcome would result in weak-linked gels. That is, failure under deformation would occur when the interfloc linkages break. The critical stress can give insight on the strength of the interactions holding the aggregates together and provide other material applications based on the different viscoelastic properties of the new gels formed.

118

Bibliography

Altena, F. et al., 1986. Thermoreversible gelation of cellulose acetate solutions studied by differential scanning calorimetry. Journal of Polymer Science Part B: Polymer Physics, 24(8), pp.1725-1734. Amal, R., Raper, J.A. & Waite, T.D., 1990. Fractal structure of hematite aggregates. Journal of Colloid and Interface Science, 140(1), pp.158-168. Appaw, C. et al., 2010. Phase separation and heat-induced gelation characteristics of cellulose acetate in a mixed solvent system. Cellulose, 17(3), pp.533-538. Appaw, C. et al., 2007. Viscoelastic behavior of cellulose acetate in a mixed solvent system. Biomacromolecules, 8(5), pp.1541-1547. Atalla, R.H. & VanderHart, D.L., 1984. Native cellulose: a composite of two distinct crystalline forms. Science, 223(4633), pp.283-285. Balser, K. et al., 2004. Cellulose esters. In Ullmann's Encyclopedia of Industrial Chemistry. Wiley-VCH. Available at: http://www.mrw.interscience.wiley.com/emrw/9783527306732/ueic/article/a05_419/curr ent/abstract [Accessed September 9, 2009]. Barton, A.F.M., 1975. Solubility parameters. Chemical Reviews, 75(6), pp.731-753. Bochek, A.M. & Kalyuzhnaya, L.M., 2002. Interaction of water with cellulose and cellulose acetates as influenced by the hydrogen bond system and hydrophilic-hydrophobic balance of the macromolecules. Russian Journal of Applied Chemistry, 75(6), pp.989-993. Brandrup, J., Immergut, E.H. & Grulke, E.A. eds., 2003. The Wiley database of polymer properties, Hoboken, NJ, USA: John Wiley & Sons, Inc. Available at: http://onlinelibrary.wiley.com/isbnHeader/0471532053 [Accessed November 18, 2010]. Burchard, W., 2003. Solubility and solution structure of cellulose derivatives. Cellulose, 10(3), pp.213-225. Burchard, W., Schmidt, M. & Stockmayer, W.H., 1980. Information on polydispersity and branching from combined quasi-elastic and intergrated scattering. Macromolecules, 13(5), pp.1265-1272. Burrell, H., 1973. Trends in solvent science and technology. In Solvents Theory and Practice. pp. 1-10. Available at: http://pubs.acs.org/doi/abs/10.1021/ba-1973-0124.ch001. Buscall, R. et al., 1988. Scaling behaviour of the rheology of aggregate networks formed from 119

colloidal particles. Journal of the Chemical Society, Faraday Transactions 1, 84(12), p.4249. Butler, M.F. & Heppenstall-Butler, M., 2003. Phase separation in gelatin/dextran and gelatin/maltodextrin mixtures. Food Hydrocolloids, 17(6), pp.815-830. Cai, J. & Zhang, L.N., 2006. Unique gelation behavior of cellulose in NaOH/urea aqueous solution. Biomacromolecules, 7(1), pp.183-189. Camacho Gómez, J.A., Erler, U.W. & Klemm, D., 1996. 4-methoxy substituted trityl groups in 6-O protection of cellulose: homogeneous synthesis, characterization, detritylation. Macromolecular Chemistry and Physics, 197(3), pp.953-964. Claffey, W. & Blackwell, J., 1976. Electron diffraction of Valonia cellulose. A quantitative interpretation. Biopolymers, 15(10), pp.1903-1915. Clasen, C. & Kulicke, W., 2001. Determination of viscoelastic and rheo-optical material functions of water-soluble cellulose derivatives. Progress in Polymer Science, 26(9), pp.1839-1919. Committee on Polymer Science and Engineering et al., 1994. Polymer science and engineering: the shifting research frontiers, National Academies Press. Denham, W.S. & Woodhouse, H., 1913. CLXXXVI.- The methylation of cellulose. Journal of the Chemical Society, Transactions, 103, pp.1735-1742. Deus, C., Friebolin, H. & Siefert, E., 1991. Partiell acetylierte cellulose - synthese und bestimmung der substituentenverteilung mit hilfe der 1H NMR-spektroskopie. Die Makromolekulare Chemie, 192(1), pp.75-83. Dupont, A., 2003. Cellulose in lithium chloride/N,N-dimethylacetamide, optimisation of a dissolution method using paper substrates and stability of the solutions. Polymer, 44(15), pp.4117-4126. Eissa, A.S. & Khan, S.A., 2005. Acid-induced gelation of enzymatically modified, preheated whey proteins. Journal of Agricultural and Food Chemistry, 53(12), pp.5010-5017. El-Kafrawy, A., 1982. Investigation of the cellulose/LiCl/dimethylacetamide and cellulose/LiCl/N-methyl-2-pyrrolidinone solutions by 13C NMR spectroscopy. Journal of Applied Polymer Science, 27(7), pp.2435-2443. Elöd, E. & Schrodt, A., 1931. Über die Umwandlung von primärem Celluloseacetat in Sekundäracetat. Angewandte Chemie, 44(48), pp.933-935. Ferry, J., 1980. Viscoelastic properties of polymers 3rd ed., New York: Wiley.

120

Fink, H., Hofmann, D. & Purz, H.J., 1990. Zur Fibrillarstruktur nativer Cellulose. Acta Polymerica, 41(2), pp.131-137. Fischer, S. et al., 2001. An alternative approach for deprotection of triphenylmethyl cellulose. Polymer Bulletin, 45(6), pp.517-522. Flory, P.J., Spurr, O.K. & Carpenter, D.K., 1958. Intrinsic viscosities of cellulose derivatives. Journal of Polymer Science, 27(115), pp.231-240. Frey, M.W., Cuculo, J.A. & Khan, S.A., 1996. Rheology and gelation of cellulose/ammonia/ammonium thiocyanate solutions. Journal of Polymer Science Part B: Polymer Physics, 34(14), pp.2375-2381. Gagnaire, D., Saint-Germain, J. & Vincendon, M., 1983. NMR evidence of hydrogen bonds in cellulose solutions. Journal of Applied Polymer Symposia, 37, p.261. Gardner, K.H. & Blackwell, J., 1974. The structure of native cellulose. Biopolymers, 13(10), pp.1975-2001. Garside, P. & Wyeth, P., 2003. Identification of cellulosic fibres by FTIR spectroscopy: thread and single fibre analysis by attenuated total reflectance. Studies in Conservation, 48(4), pp.269-275. Gedon, S. & Fengi, R., 2000. Cellulose esters, organic esters. In John Wiley & Sons, Inc., ed. Kirk-Othmer Encyclopedia of Chemical Technology. Hoboken, NJ, USA: John Wiley & Sons, Inc. Available at: http://onlinelibrary.wiley.com/doi/10.1002/0471238961.1518070107050415.a01/abstract [Accessed November 29, 2010]. Goebel, K.D. & Berry, G.C., 1977. Properties of cellulose acetate in solution. II. Light scattering and elasticity measurements on gels in benzyl alcohol. Journal of Polymer Science: Polymer Physics Edition, 15(3), pp.555-577. Goodlett, V.W., Dougherty, J.T. & Patton, H.W., 1971. Characterization of cellulose acetates by nuclear magnetic resonance. Journal of Polymer Science Part A-1: Polymer Chemistry, 9(1), pp.155-161. Graenacher, C., 1934. Cellulose solution. Hans Tromp, R. et al., 2001. Confocal scanning light microscopy (CSLM) on mixtures of gelatine and polysaccharides. Food Research International, 34(10), pp.931-938. Hansen, C.M., 1969. The universality of the solubility parameter. Product R&D, 8(1), pp.2-11. Hansen, C.M., 2007. Hansen solubility parameters: a user's handbook, CRC Press.

121

Hao, J.H. & Wang, S., 2001. Calculation of alcohol-acetone-cellulose acetate ternary phase diagram and their relevance to membrane formation. Journal of Applied Polymer Science, 80(10), pp.1650-1657. Haworth, W.N., Long, C.W. & Plant, J.H.G., 1927. CCLXXVI.- The constitution of the disaccharide. Part XVI. Cellobiose. Journal of the Chemical Society (Resumed), pp.28092814. Hearle, J.W.S., 1963. The fine structure of fibers and crystalline polymers. I. Fringed fibril structure. Journal of Applied Polymer Science, 7(4), pp.1175-1192. Hebeish, A. & Guthrie, J.T., 1981. The chemistry and technology of cellulosic copolymers, Berlin; New York: Springer-Verlag. Heinze, T. & Koschella, A., 2005. Solvents applied in the field of cellulose chemistry - a mini review. Polimeros, 15(2), pp.84-90. Heinze, T. & Liebert, T., 2004. 4.2 Chemical characteristics of cellulose acetate. Macromolecular Symposia, 208(1), pp.167-238. Heinze, T. & Liebert, T., 2001. Unconventional methods in cellulose functionalization. Progress in Polymer Science, 26, pp.1689-1762. Heinze, T., Liebert, T. & Koschella, A., 2006. Structure analysis of polysaccharide esters. In Esterification of Polysaccharides. Berlin/Heidelberg: Springer-Verlag. Available at: http://www.springerlink.com/content/pp2844u66j1l6323/ [Accessed November 30, 2010]. Hieta, K., Kuga, S. & Usuda, M., 1984. Electron staining of reducing ends evidences a parallelchain structure inValonia cellulose. Biopolymers, 23(10), pp.1807-1810. Hirrien, M. et al., 1998. Thermogelation of methylcelluloses: new evidence for understanding the gelation mechanism. Polymer, 39(25), pp.6251-6259. Hoernschemeyer, D., 1974. The influence of solvent type on the viscosity of concentrated polymer solutions. Journal of Applied Polymer Science, 18(1), pp.61-75. Horii, F., Hirai, A. & Kitamaru, R., 1983. Solid-state 13C-NMR study of conformations of oligosaccharides and cellulose. Polymer Bulletin, 10(7-8), pp.357-361. Husemann, V.E. & Siefert, E., 1969. N-äthyl-pyridinium-chlorid als lösungsmittel und reaktionsmedium für cellulose. Die Makromolekulare Chemie, 128(1), pp.288-291. Irvine, J.C. & Hirst, E.L., 1922. CLXXXIX.- The constitution of polysaccharides. Part V. The yield of glucose from cotton cellulose. Journal of the Chemical Society, Transactions, 121, pp.1585-1591. 122

Isogai, A., 1997. NMR analysis of cellulose dissolved in aqueous NaOH solutions. Cellulose, 4, pp.99-107. Isono, Y. & Nagasawa, M., 1980. Solvent effects on rheological properties of polymer solutions. Macromolecules, 13(4), pp.862-867. Iwata, T. et al., 1992. Preparation and n.m.r. assignments of cellulose mixed esters regioselectively substituted by acetyl and propanoyl groups. , 224, pp.277-283. Iwata, T. et al., 1997. DSC study on regioselectively substituted cellulose heteroesters. Journal of Applied Polymer Science, 65(8), pp.1511-1515. Johnston, H.K. & Sourirajan, S., 1973. Viscosity-temperature relationships for cellulose acetateacetone solutions. Journal of Applied Polymer Science, 17(12), pp.3717-3726. Kadla, J.F. & Korehei, R., 2010. Effect of hydrophilic and hydrophobic interactions on the rheological behavior and microstructure of a ternary cellulose acetate system. Biomacromolecules, 11(4), pp.1074-1081. Kamide, K., Okajima, K. & Saito, M., 1981. Nuclear magnetic resonance study of thermodynamic interaction between cellulose acetate and solvent. Polymer Journal, 13(2), pp.115-125. Kamide, K. & Saito, M., 1984. Effect of total degree of substitution on molecular parameters of cellulose acetate. European Polymer Journal, 20(9), pp.903-914. Kamide, K. & Saito, M., 1985. Thermal analysis of cellulose acetate solids with total degrees of substitution of 0.49, 1.75, 2.46, and 2.92. Polymer Journal, 17(8), pp.919-928. Kavanagh, G.M. & Ross-Murphy, S.B., 1998. Rheological characterisation of polymer gels. Progress in Polymer Science, 23(3), pp.533-562. Kawanishi, H., Tsunashima, Y. & Horii, F., 2000. Dynamic light scattering study of structural changes of cellulose diacetate in solution under couette flow. Macromolecules, 33(6), pp.2092-2097. Kawanishi, H. et al., 1998. Change in chain stiffness in viscometric and ultracentrifugal fields: Cellulose diacetate in N, N-dimethylacetamide dilute solution. The Journal of Chemical Physics, 108(14), pp.6014-6025. Khalil, S.A.H., 1973. Phase separation of cellulose derivatives: effects of polymer viscosity and dielectric constant of nonsolvent. Journal of Pharmaceutical Sciences, 62(11), pp.18831884. Klemm, D. et al., 1997. New approaches to advanced polymers by selective cellulose 123

functionalization. Acta Polymerica, 48(8), pp.277-297. Klemm, D. et al., 2005. Cellulose: fascinating biopolymer and sustainable raw material. Angewandte Chemie International Edition, 44(22), pp.3358-3393. Klemm, D., Schmauder, H. & Heinze, T., 2002. Cellulose. In E. Vandamme, S. De Beats, & A. Steinbuchel, eds. Polysaccharides II: polysaccharides from eukaryotes. Biopolymers. Weinheim, Germany: Wiley-VCH, pp. 290-292. Klemm, D., Schmauder, H. & Heinze, T., 2005. Cellulose. In Biopolymers Online. Weinheim, Germany: Wiley-VCH Verlag GmbH & Co. KGaA. Available at: http://onlinelibrary.wiley.com/doi/10.1002/3527600035.bpol6010/abstract [Accessed August 24, 2010]. Kolpak, F.J. & Blackwell, J., 1976. Determination of the structure of cellulose II. Macromolecules, 9(2), pp.273-278. Kondo, T., 1993. Preparation of 6-O-alkylcelluloses. Carbohydrate Research, 238, pp.231-240. Krässig, H.A., 1993. Cellulose: structure, accessibility, and reactivity, Yverdon, Switzerland: Gordon and Breach Science. Kulicke, W. et al., 1996. Characterization of aqueous carboxymethylcellulose solutions in terms of their molecular structure and its influence on rheological behaviour. Polymer, 37(13), pp.2723-2731. Langan, P., Nishiyama, Y. & Chanzy, H., 1999. A revised structure and hydrogen-bonding system in cellulose II from a neutron fiber diffraction analysis. Journal of the American Chemical Society, 121(43), pp.9940-9946. Law, P., Longdon, A. & Willins, G., 2004. 5.3 Solvent cast cellulose diacetate film. Macromolecular Symposia, 208(1), pp.293-322. Lenz, R.W., 1960. Distribution of methoxyl groups in the methylation of the monosodio derivatives of methyl α-D-glucopyranoside and cellulose. Journal of the American Chemical Society, 82(1), pp.182-186. Li, L., 2002. Thermal gelation of methylcellulose in water: scaling and thermoreversibility. Macromolecules, 35(15), pp.5990-5998. López, M.C. et al., 2008. Acid-catalysed hydrolysis of methoxy-substituted trityl trifluoroethyl ethers: a kinetic and computational investigation of leaving group effects. Journal of Physical Organic Chemistry, 21(7-8), pp.614-621. Macosko, C.W., 1994. Rheology - principles, measurements and applications, New York: John Wiley & Sons. 124

Malm, C.J., Tanghe, L.J. & Laird, B.C., 1950. Primary hydroxyl groups in hydrolyzed cellulose acetate. Journal of the American Chemical Society, 72(6), pp.2674-2678. Marangoni, A.G., 2002. The nature of fractality in fat crystal networks. Trends in Food Science & Technology, 13(2), pp.37-47. Mark, J., 2009. Polymer Data Handbook 2nd ed., Oxford University Press US. Marrinan, H.J. & Mann, J., 1954. A study by infra-red spectroscopy of hydrogen bonding in cellulose. Journal of Applied Chemistry, 4(4), pp.204-211. Marx-Figini, M., 2007. On the biosynthesis of cellulose in higher and lower plants. Journal of Polymer Science Part C: Polymer Symposia, 28(1), pp.57-67. McCormick, C.L. & Callais, P.A., 1987. Derivatization of cellulose in lithium chloride and N-Ndimethylacetamide solutions. Polymer, 28(13), pp.2317-2323. Meyer, K.H. & Mark, H., 1928. Über den Bau des krystallisierten Anteils der Cellulose. Berichte der deutschen chemischen Gesellschaft (A and B Series), 61(4), pp.593-614. Meyer, K.H. & Misch, L., 1937. Positions des atomes dans le nouveau modèle spatial de la cellulose. Helvetica Chimica Acta, 20(1), pp.232-244. Miyamoto, T. et al., 1984. 13C nuclear magnetic resonance studies of cellulose acetate. Journal of Polymer Science: Polymer Chemistry Edition, 22(10), pp.2363-2370. Miyamoto, T. et al., 1985. 13C-NMR spectral studies on the distribution of substituents in watersoluble cellulose acetate. Journal of Polymer Science: Polymer Chemistry Edition, 23(5), pp.1373-1381. Morooka, T. et al., 1984. Dielectric properties of cellulose acylates. Journal of Applied Polymer Science, 29(12), pp.3981-3990. Narine, S.S. & Marangoni, A.G., 1999. Fractal nature of fat crystal networks. Physical Review E, 59(2), p.1908. Nishiyama, Y. et al., 2008. Neutron crystallography, molecular dynamics, and quantum mechanics studies of the nature of hydrogen bonding in cellulose Iβ. Biomacromolecules, 9(11), pp.3133-3140. Odian, G.G., 1991. Principles of polymerization, Wiley. Östlund, Å. et al., 2009. Dissolution and gelation of cellulose in TBAF/DMSO solutions: the roles of fluoride ions and water. Biomacromolecules, 10(9), pp.2401-2407.

125

Payen, A., 1838. Comptes Rendus de l'Academie des Sciences, 7, p.1052. Philipp, B. & Klemm, D., 1995. Regioselektive Veresterung und Veretherung von Cellulose und Cellulosederivaten. Das Papier, 49(1), pp.3-7. Philipp, B. et al., 1996. Regioselective derivatization of cellulose - routes of synthesis, effects on properties, and areas of application. Polymer News, 21, pp.155-161. Pintaric, B., Rogosic, M. & Jasna Mencer, H., 2000. Dilute solution properties of cellulose diacetate in mixed solvents. Journal of Molecular Liquids, 85(3), pp.331-350. Potthast, A. et al., 2002. The cellulose solvent system N,N-dimethylacetamide/lithium chloride revisited: the effect of water on physicochemical properties and chemical stability. Cellulose, 9, pp.41-53. Puleo, A.C., Paul, D.R. & Kelley, S.S., 1989. The effect of degree of acetylation on gas sorption and transport behavior in cellulose acetate. Journal of Membrane Science, 47(3), pp.301332. Raghavan, S.R. & Cipriano, B.H., 2006. Gel formation: phase diagrams using tabletop rheology and calorimetry. In Molecular Gels Materials with Self-Assembled Fibrillar Networks. Dordrecht: Springer-Verlag, pp. 241-252. Rao, V.S.R. et al., 1998. Conformation of carbohydrates, Amsterdam: Harwood Academic Publishers. Reuvers, A., Altena, F. & Smolders, C., 1986. Demixing and gelation behavior of ternary cellulose acetate solutions. Journal of Polymer Science Part B: Polymer Physics, 24(4), pp.793-804. Reuvers, A. & Smolders, C., 1987. Formation of membranes by means of immersion precipitation : Part II. the mechanism of formation of membranes prepared from the system cellulose acetate-acetone-water. Journal of Membrane Science, 34(1), pp.67-86. Ross-Murphy, S.B., 1992. Physical gelation of biopolymers. In Viscoelasticity of Biomaterials. ACS Symposium Series. pp. 204-216. Ross-Murphy, S.B., 1987. Physical gelation of biopolymers. Food Hydrocolloids, 1(5-6), pp.485-495. Ross-Murphy, S.B., 1995. Rheological characterisation of gels. Journal of Texture Studies, 26(4), pp.391-400. Roy, C., Budtova, T. & Navard, P., 2003. Rheological properties and gelation of aqueous cellulose−NaOH solutions. Biomacromolecules, 4(2), pp.259-264.

126

Rudraraju, V.S. & Wyandt, C.M., 2005. Rheology of microcrystalline cellulose and sodiumcarboxymethyl cellulose hydrogels using a controlled stress rheometer: part II. International Journal of Pharmaceutics, 292(1-2), pp.63-73. Rustemeyer, P., 2004. 1. History of CA and evolution of the markets. Macromolecular Symposia, 208(1), pp.1-6. Rye, G.G., Litwinenko, J.W. & Marangoni, A.G., 2005. Fat crystal networks. In F. Shahidi, ed. Bailey's Industrial Oil and Fat Products. Hoboken, NJ, USA: John Wiley & Sons, Inc. Available at: http://onlinelibrary.wiley.com/doi/10.1002/047167849X.bio008/abstract [Accessed December 9, 2010]. Ryskina, I. & Aver'yanova, V., 1975. Morphological study of the three dimensional network of cellulose acetate-benzyl alcohol gelatins. Polymer Science U.S.S.R., 17(4), pp.1060-1064. Sarko, A. & Muggli, R., 1974. Packing analysis of carbohydrates and polysaccharides. III. Valonia cellulose and cellulose II. Macromolecules, 7(4), pp.486-494. Sata, H., Murayama, M. & Shimamoto, S., 2004. 5.4 Properties and applications of cellulose triacetate film. Macromolecular Symposia, 208(1), pp.323-334. Schult, T. et al., 2002. Characterization of cellulose by SEC-MALLS. Cellulose, 9(2), pp.149158. Schulz, L., Seger, B. & Burchard, W., 2000. Structures of cellulose in solution. Macromolecular Chemistry and Physics, 201(15), pp.2008-2022. Schützenberger, P., 1865. Action de l'acide acetique anhydre sur la cellulose, l'amidon, les sucres, la mannite et ses congeneres, les glucosides et certaines matieres colorantes. compt. rend. Acad. Sciences, 61, p.484. Sekiguchi, Y., Sawatari, C. & Kondo, T., 2003. A gelation mechanism depending on hydrogen bond formation in regioselectively substituted O-methylcelluloses. Carbohydrate Polymers, 53(2), pp.145-153. Shibata, T., 2004. 5.6 Cellulose acetate in separation technology. Macromolecular Symposia, 208(1), pp.353-370. Shih, W. et al., 1990. Scaling behavior of the elastic properties of colloidal gels. Physical Review A, 42(8), p.4772. Skorayakov, I. & Komar, V., 1994. Determination of the content of bound acetic acid in cellulose acetates by the method of IR spectroscopy. Journal of Applied Spectroscopy, 61(3), pp.558-563. Smolders, C. et al., 1992. Microstructures in phase-inversion membranes. Part 1. Formation of 127

macrovoids. Journal of Membrane Science, 73(2-3), pp.259-275. Sonntag, R.C. & Russel, W.B., 1987. Elastic properties of flocculated networks. Journal of Colloid and Interface Science, 116(2), pp.485-489. Staudinger, H., 1920. Über Polymerisation. Berichte der deutschen chemischen Gesellschaft (A and B Series), 53(6), pp.1073-1085. Steinmeier, H., 2004. 3. Acetate manufacturing, process and technology- 3.1 Chemistry of cellulose acetylation. Macromolecular Symposia, 208(1), pp.49-60. Sugiyama, J., Vuong, R. & Chanzy, H., 1991. Electron diffraction study on the two crystalline phases occurring in native cellulose from an algal cell wall. Macromolecules, 24(14), pp.4168-4175. Suzuki, H. et al., 1981. Critical solution temperatures in cellulose diacetate-acetone solutions. Netsusokutei, 8(2), pp.67-70. Swatloski, R.P. et al., 2002. Dissolution of cellose with ionic liquids. Journal of the American Chemical Society, 124(18), pp.4974-4975. Taft, R. & Stareck, J., 1931. Liquid ammonia as a lyophilic dispersion medium II. Ammono-gels of cellulose acetate. Journal of Physical Chemistry, 35(2), pp.578-587. Takaragi, A. et al., 1999. Reaction characteristics of cellulose in the LiCl/1,3-dimethyl-2imidazolidinone solvent system. Cellulose, 6(2), pp.93-102. Tanner, D.W. & Berry, G.C., 1974. Properties of cellulose acetate in solution. I. Light scattering, osmometry, and viscometry on dilute solutions. Journal of Polymer Science: Polymer Physics Edition, 12(5), pp.941-975. Tasker, S. et al., 1994. Hydroxyl accessibility in celluloses. Polymer, 35(22), pp.4717-4721. Tezuka, Y. & Tsuchiya, Y., 1995. Determination of substituent distribution in cellulose acetate by means of a 13C NMR study on its propanoated derivative. Carbohydrate Research, 273(1), pp.83-91. Thielking, H. & Schmidt, M., 2006. Cellulose ethers. In Wiley-VCH Verlag GmbH & Co. KGaA, ed. Ullmann's Encyclopedia of Industrial Chemistry. Weinheim, Germany: Wiley-VCH Verlag GmbH & Co. KGaA. Available at: http://onlinelibrary.wiley.com/doi/10.1002/14356007.a05_461.pub2/full [Accessed December 8, 2010]. Tsunashima, Y. & Hattori, K., 2000. Substituent distribution in cellulose acetates: its control and the effect on structure formation in solution. Journal of Colloid and Interface Science, 228(2), pp.279-286. 128

Tsunashima, Y. et al., 2001. Regioselectively substituted 6-O- and 2,3-di-O-acetyl-6-Otriphenylmethylcellulose: its chain dynamics and hydrophobic association in polar solvents. Biomacromolecules, 2(3), pp.991-1000. Tsunashima, Y., Kawanishi, H. & Horii, F., 2002. Reorganization of dynamic self-assemblies of cellulose diacetate in solution: dynamical critical-like fluctuations in the lower critical solution temperature system. Biomacromolecules, 3(6), pp.1276-1285. Wagenknecht, W., 1996. Regioselektive Cellulose-derivate durch Modifizierung technischer Celluloseaceate. Papier, 50, pp.712-720. Wang, C. & Fried, J.R., 1992. Viscoelastic properties of concentrated cellulose acetate solutions. Journal of Rheology, 36(5), pp.929-945. Wijmans, J., Baaij, J. & Smolders, C., 1983. The mechanism of formation of microporous or skinned membranes produced by immersion precipitation. Journal of Membrane Science, 14(3), pp.263-274. Wilson, J.D. & Hamilton, J.K., 1986. Wood cellulose as a chemical feedstock for the cellulose esters industry. Journal of Chemical Education, 63(1), p.49. Winter, H.H., 1999. Soft polymeric materials near the transition from liquid to solid state. KoreaAustralia Rheology Journal, 11(4), pp.275-278. Winter, H.H. & Chambon, F., 1986. Analysis of linear viscoelasticity of a crosslinking polymer at the gel point. Journal of Rheology, 30(2), pp.367-382. Winter, H.H. & Mours, M., 1997. Rheology of polymers near liquid-solid transitions. In Neutron Spin Echo Spectroscopy Viscoelasticity Rheology. pp. 165-234. Available at: http://dx.doi.org/10.1007/3-540-68449-2_3 [Accessed September 24, 2009]. Wohlfarth, C., 2008. Dielectric constant of N,N-dimetylacetamide. In Data extract from LandoltBörnstein IV/17: Static Dielectric Constants of Pure Liquids and Binary Liquid Mixtures. Springer-Verlag Berlin Heidelberg. Wu, H. & Morbidelli, M., 2001. A model relating structure of colloidal gels to their elastic properties. Langmuir, 17(4), pp.1030-1036. Yatsu, L., Calamari, T. & Benerito, R., 1986. Conversion of cellulose I to stable cellulose III. Textile Research Journal, 56(7), pp.419 -424. Yin, T. & Brown, R., 1959. The reactivity of cellulose: III. The nitration of cotton linters alternately wetted with water at various temperatures and dried, and the probable distribution of the nitrate groups. Canadian Journal of Chemistry, 37(2), pp.444-453.

129

Zhang, H. et al., 2005. 1-Allyl-3-methylimidazolium chloride room temperature ionic liquid: a new and powerful nonderivatizing solvent for cellulose. Macromolecules, 38(20), pp.8272-8277. Zhu, S. et al., 2006. Dissolution of cellulose with ionic liquids and its application: a mini-review. Green Chemistry, 8(4), p.325. Zugenmaier, P., 2004. 4. Characteristics of cellulose acetates - 4.1 Characterization and physical properties of cellulose acetates. Macromolecular Symposia, 208(1), pp.81-166.

130

Appendices

131

A: Selected spectra

Characterization of commercial CA starting material (reported Mn 50,000, acetyl content 34%, Sigma-Aldrich, USA):

A1. GPC spectrum of commercial CA: Mn 4.97 x 104, Mw 1.48 x 105, PDI 2.98

12000 30.6

RID signal

10000 8000 6000 4000 2000 0 24

26

28

30

32

34

36

time (minutes)

132

A2. TGA of commercial CA: Td 363°C

A3. DSC of commercial CA: Tg 203°C, Tm 234°C

133

A4. Infrared absorbance spectrum of commercial CA 0.493 0.48

0.46

0.44

0.42

0.40

0.38

0.36 A 0.34

0.32

0.30

0.28

0.26

0.24 0.224 4000.0

3600

3200

2800

2400

2000

1800 cm-1

1600

1400

1200

1000

800

600

500.0

A5. 1H NMR spectrum of commercial CA:

134

A6. 13C NMR spectrum of commercial CA:

Characterization of starting materials, intermediates, and products for the synthesis of 2,3-di-Oacetylcellulose: A7. Cellulose FTIR spectrum from saponified commercial CA: 0.78 0.7

0.6

0.5

0.4 A 0.3

0.2

0.1

-0.01 4000.0

3600

3200

2800

2400

2000

1800

1600

1400

1200

1000

800

600

450.0

cm-1

135

Synthesis of 1A, 6-O-(4-methoxytriphenylmethyl)-cellulose (6TC): A8. 1H NMR spectrum:

A9. 13C NMR spectrum:

136

Synthesis of 2A, 2,3-di-O-acetyl-6-O-(4-methoxytrityl)-cellulose (2,3Ac6TC): A10. FTIR spectrum: 0.0873 0.085 0.080 0.075 0.070 0.065 0.060 0.055 0.050 0.045 0.040 A 0.035 0.030 0.025 0.020 0.015 0.010 0.005 0.000 -0.005 -0.0094 4000.0

3600

3200

2800

2400

2000

1800

1600

1400

1200

1000

800

600

450.0

cm-1

A11. 1H NMR spectrum:

137

A12. 13C NMR spectrum:

Synthesis of 3A, 2,3-di-O-acetylcellulose: A13. 1H NMR spectrum:

138

A14. 13C NMR spectrum:

Characterization of starting materials, intermediates, and products for the synthesis of 3B, 2,3-diO-acetylcellulose II: Synthesis of 1B: A15. FTIR spectrum: 1.01 5 0.9 5 0.9 0 0.8 5 0.8 0 0.7 5 0.7 0 0.6 5 0.6 0 0.5 5 0.5 A 0 0.4 5 0.4 0 0.3 5 0.3 0 0.2 5 0.2 0 0.1 5 0.1 0 0.0 5 0.009 4000. 0

360 0

320 0

280 0

240 0

200 0

180 0 cm1

160 0

140 0

120 0

100 0

80 0

60 0

450. 0

139

A16. 1H NMR spectrum:

A17. 13C NMR spectrum:

140

Synthesis of 2B: A18. FTIR spectrum: 1.50 1.4 1.3 1.2 1.1 1.0 0.9 0.8 A

0.7 0.6 0.5 0.4 0.3 0.2 0.1

0.00 4000.0

3600

3200

2800

2400

2000

1800

1600

1400

1200

1000

800

600

450.0

cm-1

A19. 1H NMR spectrum:

141

A20. 13C NMR spectrum:

Synthesis of 3B: A21. 1H NMR spectrum:

142

A22. 13C NMR spectrum:

Characterization of cellulose acetate by propanoation: Synthesis of propanoated commercial CA 4: A23. IR absorbance spectrum: 1.000 0.95 0.90 0.85 0.80 0.75 0.70 0.65 0.60 0.55 0.50 A 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05 0.000 4000.0

3600

3200

2800

2400

2000

1800

1600

1400

1200

1000

800

600

450.0

cm-1

143

A24. 1H NMR spectrum:

A25. 13C NMR spectrum:

144

Synthesis of 4A, propanoated 2,3-di-O-acetylcellulose: A26. FTIR spectrum: 1.000 0.95 0.90 0.85 0.80 0.75 0.70 0.65 0.60 0.55 0.50 A 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05 -0.002 4000.0

3600

3200

2800

2400

2000

1800

1600

1400

1200

1000

800

600

450.0

cm-1

A27. 1H NMR spectrum:

145

A28. 13C NMR spectrum:

Synthesis of 4B, propanoated 2,3-di-O-acetylcellulose II: A29. FTIR spectrum: 1.000 0.95 0.90 0.85 0.80 0.75 0.70 0.65 0.60 0.55 0.50 A 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05 -0.002 4000.0

3600

3200

2800

2400

2000

1800

1600

1400

1200

1000

800

600

450.0

cm-1

146

A30. 1H NMR spectrum:

A31. 13C NMR spectrum:

147

A32. GPC traces for propanoated commercial, regio2.4, and regio2.8 CA

8000 7000

15.04

Refractive index signal

6000

15.75 15.50

5000 4000 Commercial CA

3000

Regio2.4 CA Regio2.8 CA

2000 1000 0 10

-1000

12

14

16

18

20

time (minutes)

148

B: Figures and tables B1: Stress sweep experiment for 10 wt% regio2.4 CA samples with 15, 20, and 25 wt% water prepared at 70°C (frequency = 1 Hz)

15 wt% 20 wt%

104 G' and G" (Pa)

25 wt%

103 102 101 100 10-2

10-1

100

101

102

103

osc. Stress (Pa) B2: Stress sweep experiment for 10 wt% regio2.4 CA samples with 15, 20, and 25 wt% water prepared at 90°C (frequency = 1 Hz) 106

15 wt% 20 wt%

105 G' and G" (Pa)

25 wt%

104 103 102 101

100 10-2

10-1

100

101

102

103

osc. Stress (Pa) 149

B3: Elastic modulus as a function of % strain for 10 wt% regio2.4 CA gels prepared at 70°C (frequency = 1 Hz) 105

G' (Pa)

104 103 102 15 wt%

101

20 wt% 25 wt%

100 10-6

10-5

10-4

10-3

10-2

10-1

100

101

102

% strain B4: Stress sweep experiment for 10 wt% regio2.4 CA samples prepared at 50, 70 and 90°C and 20 wt% nonsolvent content (Frequency = 1 Hz) 105

50°C 70°C 90°C

G' and G" (Pa)

104 103 102 101 100 10-2

10-1

100

101

102

103

osc. Stress (Pa) 150

B5: Stress sweep experiment for 10 wt% regio2.4 CA samples prepared at 50, 70 and 90°C and 25 wt% nonsolvent content. (Frequency = 1 Hz) 106

G' and G" (Pa)

105 104 103

102 50°C

101

70°C 90°C

100 10-2

10-1

100

101

102

103

osc. Stress (Pa)

151

C: Calculations

C1: Fractal dimensions obtained from rheology using power law exponents from Figure 3.9

Power law exponent n

=

= For 50° mixing, n = 45 45(3 – D) = 4.15 D = 2.91 Similarly, D for 70 and 90°C mixing are 2.92 and 2.91 respectively.

C2: Fractal dimensions based on the limit of linearity from Figure 3.11

Limit of linearity 0 = For 50°C mixing, the exponent value is -15 15 = 15(3 – D) = 2.15 D = 2.86 Similarly, D for 70 and 90°C mixing are 2.83 and 2.82 respectively

C3: Fractal dimensions obtained from microscopy

Images were first thresholded (50°C sample used in this example):



152

Using the Fractal Box Count in ImageJ: Box size 2 3 4 6 8 12 16 32 64 128

Count 128083 90283 60854 29145 16383 7396 4096 1024 256 64

C4: Trityl degree of substitution

Molecular formula for 1A components Functional group C AGU without hydroxyls 6 4-methoxytrityl 20

H 7 17

O 5 1

Theoretical mass calculation for tritylation of DS 1 at C6 position: Total mass = mass of (AGU without hydroxyls + 4-methoxytrityl * DS of tritylation + unsubstituted hydroxyl H) = (6*C + 7*H + O*5) + (20*1*C + 17*1 *H+ 1*1*O) + (2*H) = C(6 + 20) + H(7+17+2) + O(5+1) = 12.011(26) + 1.008(26) + 15.999(6) = 434.5 g/mol %C = (12.011*26)/434.5 = 71.9%, where 16.6% is from the AGU and 55.3% is from trityl 153

%H = (1.008*26)/434.5 %O = (15.999*6)/434.5

= 6.0%, where 1.6% is from the AGU and 4.4% is from trityl = 22.1% where 18.4% is from the AGU and 3.7% is from trityl

Found from EA: %C = 69.4, %H = 6.1 Assuming no degradation from the tritylation process, the mass of the polymer backbone should be the same, with the exception of the trityl substitution. Therefore, the %C contribution from the trityl group to the product should be the difference between the obtained %C and the theoretical %C of the AGU. %C from trityl group = 69.4 – 16.6 = 52.8% Theoretical contribution is 55.3%, therefore 52.8/55.3 = 0.95

C5: GPC calibration curve

Elution volume 13.8815 15.0263 15.4367 15.6023 16.5167 17.2798

Molecular weight 300000 113000 63000 50000 17500 9000

Slope -0.4631 -0.4631 -0.4631 -0.4631 -0.4631 -0.4631

Error 8.1477 -15.2912 -1.9038 3.5885 11.6462 -3.7842

Molecular weight

1000000

100000

10000

R² = 0.9909

1000 13

14

15

16

17

18

Elution volume

154