ABSTRACT
Title of dissertation: PROCESSING AND STRUCTURAL
CHARACTERIZATION TOWARD
ALL-CELLULOSE NANOCOMPOSITES
Doug Henderson, Doctor of Philosophy, 2021
Dissertation directed by: Professor Robert Briber
Materials Science and Engineering
Cellulose is the most abundant biopolymer on the planet and is used in a
variety of industry sectors including paper, coatings, medicine, and food. A deep
understanding of cellulose is important for its development as an alternative polymer
to those based on petroleum.
This work focuses on two cellulose systems. The first of these, cellulose
nanofibers, are the basic structural elements of naturally-occurring cellulosic materials;
they exhibit excellent mechanical characteristics due to high crystallinity and a dense
network of hydrogen bonding. These fibers can be separated from bulk cellulose
via a TEMPO oxidation reaction followed by mechanical homogenization into a
suspension in water. In this work, the production of these fibers is investigated by
monitoring the change in structure of cellulose as a function of TEMPO reaction
time and mechanical homogenization using small angle neutron scattering, atomic
force microscopy, and optical microscopy.
The second cellulose system is a molecular solution of cellulose formed using a
binary solvent mixture consisting of ionic liquid and an aprotic solvent. Cellulose is
difficult dissolve due to a dense hydrogen bonding network, and ionic liquids have
been shown to be an effective alternative to more hazardous and energy-intensive
dissolution methods for cellulose currently used in industry. The phase behavior
of these solutions has been investigated using small angle neutron scattering as a
function of temperature. The process of regenerating cellulose from these solutions
is also investigated, as dense gels of cellulose and ionic liquid were produced with a
unique multiscale ordered structure.
The ultimate goal of this work is to combine cellulose nanofibers and molecular
cellulose solutions in order to create all-cellulose nanocomposite films. These films
are characterized using tensile testing, atomic force microscopy, and water uptake
measurements in order to understand the interaction between cellulose nanofibers
and molecular cellulose solutions, water resistance and tunability of mechanical
properties.
PROCESSING AND STRUCTURAL CHARACTERIZATION
TOWARD ALL-CELLULOSE NANOCOMPOSITES
by
Douglas Arthur Henderson
Dissertation submitted to the Faculty of the Graduate School of the
University of Maryland, College Park in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
2021
Advisory Committee:
Professor Robert Briber: Advisor, Chair
Professor Teng Li: Dean?s Representative
Professor Howard Wang
Professor Liangbing Hu
Professor Isabel K. Lloyd
? Copyright by
Doug Henderson
2021
Dedication
This work is dedicated to my fiance, Mackenzie. Thank you for being
my rock, and for being supportive throughout my graduate school
career. I could not have done this without you.
ii
Acknowledgments
I would first like to acknowledge the support of my advisor Dr. Robert Briber.
The last few years in particular have required lots of one on one meetings to complete
my research and finalize my thesis, and I will always appreciate his mentorship and
time dedicated to my graduate tenure. Dr. Briber has a clear love for teaching and
educating, and it showed in my time working closely with him as much as it has in
his many roles serving the Clark School of Engineering.
I would also like to recognize the time, dedication, and patience from Dr.
Howard Wang during his advising of my work. Dr. Wang has exceptional integrity
and a vast fundamental understanding of materials science. He taught me to
appreciate and believe in collaborative and incremental scientific progress, what it
takes to be a good researcher, and how to hold myself to a higher standard. I will
always be thankful for his guidance and his effort throughout my time working with
him.
I would like to thank Dr. Xin Zhang, for his many years of being my mentor,
travel companion, and lab-mate. I learned most of my technical and hands-on
laboratory skills from watching and learning from Dr. Zhang over the years. We
attended many conferences, workshops, and other laboratories together and he was
always a joy to travel with. He is exceptionally smart and kind, and I will always be
appreciative of his time and his sage advice during my graduate tenure.
I would like to thank Dr. Yimin Mao for his support during my work. He is
a talented instrument scientist and a wonderfully kind and outgoing person, and
iii
offered constant advice on anything from interpreting complex scattering data, to
advice on writing, presenting, and a career. I would also like to thank Dr. Wonseok
Hwang for his expertise in both small angle X-ray scattering and rheology. He put in
lots of time to help me collect critical data to finish a particularly difficult project.
I would like to recognize the advising of Dr. Liangbing Hu during the early
portion of my graduate work, who initially sparked my interest in cellulose research,
nanofibers, and sustainable materials through the novel and engaging project ideas
in his group. I also would like to acknowledge the assistance I received from other
colleagues at the NIST Center for Neutron Research and for UMD affiliated labs
like the Advanced Imaging Laboratory (AIMLab), X-ray Crystallography Center,
and in particular Dr. Karen Gaskell at the Surface Analysis Center who assisted me
countless times with atomic force microscopy. I?d also like to thank Dr. Isabel Lloyd
and Dr. Peter Kofinas for their time and advice during large group meetings.
Finally, I?d like to thank my parents, Walter and Carla; my stepmom, Sandy;
my brother, Brian; my friends; my fiance, Mackenzie, and her parents Mark and
Jackie. All have been incredibly supportive throughout my graduate tenure, and
helped me get to where I am.
iv
Table of Contents
Dedication ii
Acknowledgements iii
List of Tables viii
List of Figures ix
List of Abbreviations and Symbols xii
1 Introduction 1
1.1 Cellulose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 Polymer structure . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1.2 Cellulose hierarchical structure . . . . . . . . . . . . . . . . . 3
1.1.3 Cellulose Crystal Structures . . . . . . . . . . . . . . . . . . . 5
1.1.3.1 Cellulose I . . . . . . . . . . . . . . . . . . . . . . . . 5
1.1.3.2 Cellulose II . . . . . . . . . . . . . . . . . . . . . . . 7
1.1.3.3 Cellulose III . . . . . . . . . . . . . . . . . . . . . . . 8
1.1.3.4 Cellulose IV . . . . . . . . . . . . . . . . . . . . . . . 10
1.1.3.5 Amorphous . . . . . . . . . . . . . . . . . . . . . . . 10
1.1.4 Hydrogen Bonding . . . . . . . . . . . . . . . . . . . . . . . . 10
1.2 Instrumentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.2.1 Optical Microscopy . . . . . . . . . . . . . . . . . . . . . . . . 11
1.2.2 Small angle X-ray and neutron scattering . . . . . . . . . . . . 13
1.2.2.1 Basics of elastic scattering . . . . . . . . . . . . . . . 13
1.2.2.2 Experimental setup . . . . . . . . . . . . . . . . . . . 15
1.2.2.3 Scattering contrast . . . . . . . . . . . . . . . . . . . 17
1.2.2.4 Data analysis and interpretation . . . . . . . . . . . 20
1.2.2.5 Facilities . . . . . . . . . . . . . . . . . . . . . . . . . 23
1.2.3 Fourier transform infrared spectroscopy . . . . . . . . . . . . . 24
1.2.4 Rheology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
1.2.5 Atomic force microscopy . . . . . . . . . . . . . . . . . . . . . 29
1.2.6 Hazemeter- UV/Vis spectrometer . . . . . . . . . . . . . . . . 32
v
1.2.7 Tensile testing . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
1.2.8 Scanning electron microscopy . . . . . . . . . . . . . . . . . . 36
2 Time-dependence of the TEMPO-oxidation of cellulose 40
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
2.2 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
2.2.1 Obtaining nanoscale fibrillated of cellulose . . . . . . . . . . . 41
2.2.2 TEMPO-mediated oxidation process . . . . . . . . . . . . . . 45
2.2.3 Properties of TEMPO oxidized cellulose nanofibers . . . . . . 48
2.3 Experimental methods . . . . . . . . . . . . . . . . . . . . . . . . . . 52
2.3.1 Oxidation process . . . . . . . . . . . . . . . . . . . . . . . . . 52
2.3.2 Optical microscopy . . . . . . . . . . . . . . . . . . . . . . . . 54
2.3.3 Small-angle neutron scattering . . . . . . . . . . . . . . . . . . 55
2.3.4 Mechanical homogenization . . . . . . . . . . . . . . . . . . . 55
2.3.5 Atomic force microscopy . . . . . . . . . . . . . . . . . . . . . 56
2.4 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . 56
2.4.1 Optical microscopy . . . . . . . . . . . . . . . . . . . . . . . . 56
2.4.2 Small angle neutron scattering . . . . . . . . . . . . . . . . . . 60
2.4.3 Mechanical Homogenization- AFM . . . . . . . . . . . . . . . 66
2.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
3 Molecular cellulose solutions 71
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
3.2 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
3.2.1 Dissolving cellulose . . . . . . . . . . . . . . . . . . . . . . . . 72
3.2.2 Cellulose dissolution in ionic liquids . . . . . . . . . . . . . . . 77
3.3 Molecular cellulose solutions with aprotic solvents . . . . . . . . . . . 80
3.3.1 Experimental methods . . . . . . . . . . . . . . . . . . . . . . 82
3.3.1.1 Small angle neutron scattering . . . . . . . . . . . . 83
3.3.2 Results and discussion . . . . . . . . . . . . . . . . . . . . . . 83
3.3.2.1 Phase diagrams . . . . . . . . . . . . . . . . . . . . . 83
3.3.2.2 Small angle neutron scattering results . . . . . . . . 88
3.3.3 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
3.4 Concentrated gels of cellulose and ionic liquid . . . . . . . . . . . . . 94
3.4.1 Experimental methods . . . . . . . . . . . . . . . . . . . . . . 96
3.4.1.1 Cellulose film casting . . . . . . . . . . . . . . . . . . 96
3.4.1.2 Ionic liquid removal . . . . . . . . . . . . . . . . . . 97
3.4.1.3 Small and wide angle X-ray scattering . . . . . . . . 98
3.4.1.4 Fourier transform infrared spectroscopy . . . . . . . 99
3.4.1.5 Rheology . . . . . . . . . . . . . . . . . . . . . . . . 99
3.4.2 Results and discussion . . . . . . . . . . . . . . . . . . . . . . 99
3.4.2.1 Fourier transform infrared spectroscopy . . . . . . . 99
3.4.2.2 Small angle X-ray scattering . . . . . . . . . . . . . . 102
3.4.2.3 Wide angle X-ray scattering . . . . . . . . . . . . . . 104
vi
3.4.2.4 Multiscale structural ordering of EMIMAc and cellu-
lose gels . . . . . . . . . . . . . . . . . . . . . . . . . 113
3.4.2.5 Rheology . . . . . . . . . . . . . . . . . . . . . . . . 117
3.4.2.6 Small angle X-ray scattering as a function of temper-
ature . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
3.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
4 All-cellulose composites 124
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
4.2 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
4.2.1 Pre-treatments for CNF . . . . . . . . . . . . . . . . . . . . . 125
4.2.2 Improving water-resistance of CNF-based films . . . . . . . . . 127
4.2.3 Fabricating self-standing CNF-based films and nanocomposites 129
4.2.4 Summary of all-cellulose composites . . . . . . . . . . . . . . . 137
4.2.5 All-cellulose composite- motivation and explanation . . . . . . 141
4.2.5.1 Optimizing cellulose nanofibers (CNF) for nanocom-
posite applications . . . . . . . . . . . . . . . . . . . 141
4.2.5.2 Optimizing MCS for nanocomposite applications . . 142
4.2.6 All-cellulose nanocomposite fabrication . . . . . . . . . . . . . 143
4.3 Experimental methods . . . . . . . . . . . . . . . . . . . . . . . . . . 145
4.3.1 Cellulose nanofiber preparation . . . . . . . . . . . . . . . . . 145
4.3.2 Molecular cellulose solution fabrication . . . . . . . . . . . . . 146
4.3.3 All-cellulose composite fabrication . . . . . . . . . . . . . . . . 146
4.3.4 Atomic force microscopy . . . . . . . . . . . . . . . . . . . . . 147
4.3.5 Transmission and Haze . . . . . . . . . . . . . . . . . . . . . . 147
4.3.6 Tensile testing . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
4.3.7 Scanning electron microscopy . . . . . . . . . . . . . . . . . . 148
4.3.8 Water uptake . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
4.4 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . 149
4.4.1 Atomic force microscopy . . . . . . . . . . . . . . . . . . . . . 149
4.4.2 Hazemeter measurements . . . . . . . . . . . . . . . . . . . . . 150
4.4.3 Tensile testing . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
4.4.4 Scanning electron microscopy . . . . . . . . . . . . . . . . . . 157
4.4.5 Water uptake . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
4.4.6 Composite reinforcement . . . . . . . . . . . . . . . . . . . . . 160
4.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
5 Summary and outlook 164
5.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
5.2 Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
5.2.1 The effect of water on gelation and viscoelasticity of MCS . . 169
5.2.2 Effect of coagulation process on molecular cellulose structure . 170
5.2.3 All-cellulose nanocomposites with aligned cellulose nanocrystals171
Bibliography 173
vii
List of Tables
2.1 SANS of TEMPO oxidized cellulose fibers fit parameters . . . . . . . 64
3.1 Wide angle X-ray scattering (WAXS) fitting results for 7-peak model
of cellulose/EMIMAc gel . . . . . . . . . . . . . . . . . . . . . . . . . 112
4.1 Mechanical properties displayed by cellulose samples from different
solvents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
viii
List of Figures
1.1 Cellulose repeat unit . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Cellulose synthesis and rosette structure . . . . . . . . . . . . . . . . 3
1.3 Hierarchical cellulose structure in wood . . . . . . . . . . . . . . . . . 4
1.4 Nanofiber bundles and layers of the cell wall . . . . . . . . . . . . . . 5
1.5 Cellulose polymer offset between I? and I? . . . . . . . . . . . . . . . 7
1.6 Lattice structure of native cellulose phase I? and I? . . . . . . . . . . 8
1.7 Lattice structure of native cellulose II and III . . . . . . . . . . . . . 9
1.8 Inter- and intrachain hydrogen bonding network in cellulose . . . . . 11
1.9 Optical microscope experimental setup . . . . . . . . . . . . . . . . . 12
1.10 Momentum transfer occuring in a scattering event . . . . . . . . . . . 14
1.11 Experimental setup for small angle neutron scattering . . . . . . . . . 17
1.12 Neutron and X-ray scattering cross sections of different elements and
isotopes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.13 Examples of different types of contrast . . . . . . . . . . . . . . . . . 20
1.14 Exponential dependence of scattering intensity with respect to Q . . . 22
1.15 Scattering curve and Guinier regions associated with cylindrical scat-
tering objects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
1.16 Schematic showing experimental setup of fourier transform infrared
spectroscopy (FTIR) device . . . . . . . . . . . . . . . . . . . . . . . 25
1.17 Examples of different geometries used in rheology experiments . . . . 27
1.18 Plot of the applied amplitude of the rheometer over time . . . . . . . 29
1.19 Plot describing interaction forces between an AFM tip and sample . . 30
1.20 Experimental setup for an atomic force microscopy (AFM) measurement 32
1.21 Hazemeter setup for transmittance and haze measurements . . . . . . 33
1.22 Example of a relationship between stress and strain in response to
applied tensile load . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
1.23 Tension clamp attachment on dynamic mechanical analyzer . . . . . . 36
1.24 SEM schematic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
2.1 AFM images of cellulose nanocrystal particles dried onto mica surface 43
2.2 AFM images of microfibrillated cellulose . . . . . . . . . . . . . . . . 44
2.3 Schematic of TEMPO/NaBr/NaClO oxidation process . . . . . . . . 46
2.4 Cellulose nanofiber width plotted with carboxylate content . . . . . . 47
ix
2.5 AFM image of separated, TEMPO oxidized, and mechanically homog-
enized cellulose nanofibers . . . . . . . . . . . . . . . . . . . . . . . . 48
2.6 Tensile strength and elastic modulus comparison of TEMPO oxidized
cellulose nanofibers, paper, and other laminate polymer films . . . . . 49
2.7 Understanding mechanical behavior of cellulose nanopaper films . . . 50
2.8 Transparency and haze of CNF transparent paper . . . . . . . . . . . 52
2.9 Photos of cellulose pulp and the (2,2,6,6-Tetramethylpiperidin-1-
yl)oxyl (TEMPO) oxidation reaction . . . . . . . . . . . . . . . . . . 54
2.10 Optical microscopy of cellulose aliquots throughout TEMPO oxidation 57
2.11 Diameter distributions of fiber aliquots . . . . . . . . . . . . . . . . . 58
2.12 Fiber diameter plotted against reaction time . . . . . . . . . . . . . . 59
2.13 Reduced SANS data of each aliquot sample. . . . . . . . . . . . . . . 60
2.14 Plot showing the different components of the correlation length fitting
function used for fitting the obtained SANS data . . . . . . . . . . . 62
2.15 SANS spectra of cellulose fiber aliquots . . . . . . . . . . . . . . . . . 63
2.16 Correlation length plotted against oxidation time and macrofiber
diameter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
2.17 AFM images of nanofibers at different levels of mechanical homoge-
nization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
2.18 Histograms of length and diameter from AFM images of CNF . . . . 67
2.19 Length and diameter plotted against number of homogenizations . . . 67
2.20 Schematic illustrating the hierarchical structure present within large-
scale cellulose fibers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
3.1 Scanning electron microscopy (SEM) of fibrillation in Lyocell fiber . . 75
3.2 Structure of 1-ethyl-3-methylimidazolium acetate (EMIMAc) molecule
and how it interacts with cellulose . . . . . . . . . . . . . . . . . . . . 78
3.3 Microcrystalline cellulose dissolving in EMIMAc . . . . . . . . . . . . 79
3.4 Small angle neutron scattering of cellulose solution with EMIMAc
and DMF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
3.5 Ternary phase diagrams of molecular cellulose solutions with DMSO . 84
3.6 Ternary phase diagram showing relationship between cellulose, EMI-
MAc, and D2O . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
3.7 Small angle neutron scattering (SANS) of aggregated and dissolved
cellulose solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
3.8 SANS of decreasing EMIMAc to cellulose ratios . . . . . . . . . . . . 89
3.9 SANS of molecular cellulose solution (MCS) samples at room temper-
ature, heating, and cooling . . . . . . . . . . . . . . . . . . . . . . . . 91
3.10 SANS of MCS sample undergoing two heat cycles . . . . . . . . . . . 92
3.11 FTIR results of cellulose and EMIMAc gels . . . . . . . . . . . . . . . 100
3.12 Estimate of cellulose volume fraction with FTIR data . . . . . . . . . 100
3.13 Small angle X-ray scattering (SAXS) spectra of EMIMAc/cellulose
gels during washing . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
3.14 SAXS periodic distance, Qm plotted against cellulose volume fraction 103
3.15 WAXS results of EMIMAc/cellulose gels during washing . . . . . . . 105
x
3.16 Plots showing efficacy of WAXS 7-peak model . . . . . . . . . . . . . 108
3.17 Crystallinity fraction plotted against cellulose volume fraction . . . . 109
3.18 FWHM of the 3 ordering peaks plotted against cellulose volume fraction.110
3.19 Schematic of multiscale ordering . . . . . . . . . . . . . . . . . . . . . 114
3.20 Rheology of EMIMAc/cellulose gels at different ratios and temperatures116
3.21 Tan(?) plotted against temperature . . . . . . . . . . . . . . . . . . . 118
3.22 SAXS and WAXS results of gel under a heating cycle . . . . . . . . . 121
4.1 Drying CNF to produce films . . . . . . . . . . . . . . . . . . . . . . 131
4.2 Mechanical testing of CNF and HEC nanocomposites . . . . . . . . . 133
4.3 Mechanical results from two examples of composite systems featuring
CNF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
4.4 Mechanical properties of CNF and PAM nanocomposites . . . . . . . 135
4.5 Mechanical properties of CNF and polyvinyl alcohol composites . . . 137
4.6 Fabrication of all-cellulose composites . . . . . . . . . . . . . . . . . . 138
4.7 Tensile testing of CNC composites . . . . . . . . . . . . . . . . . . . . 139
4.8 Tensile testing of CNF and alkali-urea produced cellulose . . . . . . . 140
4.9 All-cellulose nanocomposite schematic and AFM of thin film . . . . . 149
4.10 Baseline and reflective transmission measurements of each all-cellulose
nanocomposite film. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
4.11 Haze percent as a function of wavelength for each composite film . . . 151
4.12 Haze plotted as a function of MC content . . . . . . . . . . . . . . . . 152
4.13 Stress-strain curves illustrating the different mechanical behaviors of
all-cellulose nanocomposites with varying molecular cellulose (MC)
content. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
4.14 Mechanical properties of all-cellulose nanocomposite films . . . . . . . 155
4.15 SEM images of cellulose film fracture surfaces . . . . . . . . . . . . . 158
4.16 Water uptake of all-cellulose nanocomposite films . . . . . . . . . . . 159
xi
List of Abbreviations and Symbols
?COOH protonated carboxylate group.
?COONa sodium carboxylate group.
?ONO2 nitrate.
D2O deuterium oxide.
EG54DMAm46 poly[(ethylene glycol methyl ether methacrylate)-co-N,N-dimethylacrylamide].
H2SO4 sulfuric acid.
LiOH lithium hydroxide.
NaOH sodium hydroxide.
Tg glass transition temperature.
AFM atomic force microscopy.
AGU anhydroglucose unit.
AMIMCl 1-allyl-3-methylimidazolium chloride.
ATR attenuated total reflectance.
AUC alkali-urea dissolved cellulose.
CESA cellulose synthase proteins.
CHRNS center for high resolution neutron scattering.
CNC cellulose nanocrystals.
CNF cellulose nanofibers.
DMA dynamic mechanical analyzer.
DMAc dimethylacetamide.
DMAm poly(N,N-dimethylacrylamide).
DMF dimethylformamide.
DMSO dimethylsulfoxide.
DP degree of polymerization.
EMIMAc 1-ethyl-3-methylimidazolium acetate.
xii
EMIMCl 1-ethyl-3-methylimidazolium chloride.
FTIR fourier transform infrared spectroscopy.
FWHM full-width at half maximum.
HEC hydroxyethyl cellulose.
IL ionic liquid.
MC molecular cellulose.
MCS molecular cellulose solution.
MFC microfibrillated cellulose.
NIST National Institute for Standards and Technology.
NMMO N-methylmorpholine N-oxide.
NMP N-methyl-2-pyrrolidone.
NMR nuclear magnetic resonance.
OLED organic light emitting diode.
PAA poly(acrylic acid).
PAM poly(acrylamide).
PET polyethylene terephthalate.
PVA polyvinyl acetate.
SANS small angle neutron scattering.
SAXS small angle X-ray scattering.
SEM scanning electron microscopy.
SLD scattering length density.
TEMPO (2,2,6,6-Tetramethylpiperidin-1-yl)oxyl.
TOCN TEMPO-oxidized cellulose nanofibers.
WAXS wide angle X-ray scattering.
XRD X-ray diffraction.
xiii
Chapter 1: Introduction
1.1 Cellulose
Cellulose is the most abundant biopolymer in the world and has served as an
important material throughout human history. Cellulose first found use by humanity
in the form of wood as an energy source, structural material, and source of clothing.
More recently it has found usage as a polymeric material through regeneration and
chemical modification, with applications as films, coatings, membranes, pharma-
ceuticals, textiles, and as a food additive. Currently, industry is heavily reliant on
polymer materials produced from petroleum, which can have a negative impact on
the environment due to extraction methods, subsequent refining and processing, and
eventual disposal of the polymer. Understanding the properties, processing, and iden-
tifying additional applications of sustainable materials like cellulose is essential for
providing cost-effective, environmentally friendly alternatives to petroleum-derived
polymers in use today.
1
1.1.1 Polymer structure
The molecular structure of cellulose consists of two mirrored monomers called
anhydroglucose unit (AGU), the repeat unit of cellulose. Each AGU has the chemical
formula of C6H10O5. The dimer unit containing both AGU is called cellobiose, as
shown in Figure 1.1, in which each bonded AGU is rotated 180? [1]. The two mirrored
AGU are bonded via a ?-1,4-glycosidic bond. Each AGU contains three hydroxyl
groups, two secondary hydroxyl groups on C2 and C3, and a primary hydroxyl group
on C6. The degree of polymerization (DP) is defined by the number of AGU in a
cellulose chain and can vary depending on the cellulose source. Wood will typically
have a DP of between 9000-10000, but cellulose after pulping treatments have a
DP of between 300 and 1700. Regenerated cellulose will have an even lower DP of
250-500 [2, 3].
Figure 1.1: Schematic showing the monomer and dimer structure of cellulose.
The dimer, cellobiose, is composed of two mirrored-facing monomers, referred to as
AGU [1]. ? 2015
Cellulose polymer chains are synthesized within the plant cell by cellulose
synthase proteins (CESA) [4, 5]. CESA proteins are arranged into parallel arrays
2
called rosettes, forming a terminal complex with six-fold symmetry. The rosette
contains 6 CESA protein monomers. [6] The CESA subunits are arranged into a
hexagonal rosette shape resulting in roughly 36 total cellulose chains forming a single
nanofiber in which chains are held together by hydrogen bonding and van der Waals
forces [7]. A schematic representation of the CESA protein complex arranged in the
rosette shape synthesizing cellulose is displayed in Figure 1.2 [6, 8].
Figure 1.2: Schematic illustrating the synthesizing of cellulose in the terminal
complex. CESA are arranged into hexagonal rosette units and subunits to synthesize
cellulose chains that associate to form an elementary cellulose nanofiber (labelled
microfibril in image) [8]. ?2006
1.1.2 Cellulose hierarchical structure
The cellulose chains make up the structural component of the plant cell wall.
Hierarchical buildup and layering of cellulose nanofibers, interspersed with lignin and
hemicellulose, complex polysaccharides that adds structural robustness to the stiff
cellulose network within the cell wall. These complexes form layers in the plant cell
wall then ultimately form the structure found in wood as shown in Figure 1.3 [9].
3
Figure 1.3: Overall schematic of the different structures and materials present
in wood and plant cell walls. Cellulose is hierarchically structured to make up the
structure of wood, with lignin and hemicellulose interspersed throughout to reinforce
the cellulose network [9]. ? 2008
The individual elementary cellulose nanofiber is synthesized in parallel with
other nanofibers that ultimately entangle, and form bundles held together through
hydrogen and van der Waals bonding. A schematic of the nanoscale organization
of cellulose nanofibers is shown in Figure 1.4a. An individual nanofiber is made
up of crystalline regions with periodic amorphous regions. Within these bundles
of nanofibers, the chains in the crystalline regions are aligned while entanglement
occurs in the amorphous regions between crystallites. Individual nanofibers typically
have diameters around 2-4 nm while the length can be on the order of microns.
Bundles of nanofibers can range between 10-25 nm in diameter [10,11].
These bundles are then grouped together and aligned within various layers of
the cell wall, ultimately forming a cellulose macrofiber which are typically 20-40 ?m
in diameter. The layering of different arrangements of cellulose within the cell wall is
shown in Figure 1.4b [11]. The arrangement of cellulose nanofibers within each layer
of the cell wall can vary based on a number of variables all depending on the type of
4
plant or role of the cell within the plant. The average nanofiber length, number of
nanofibers per bundle, overall cellulose composition, angle of nanofiber alignment,
number of layers, etc. can vary depending on the function of the plant cell [6, 12, 13].
a b
Figure 1.4: a) Schematic showing the structure and arrangement of cellulose
nanofibers within bundles. b) Arrangement of nanofiber bundles within layers of the
plant cell wall to form a cellulose macrofiber [11]. ? 2007
1.1.3 Cellulose Crystal Structures
1.1.3.1 Cellulose I
Cellulose I is the only naturally occurring crystalline form of cellulose, mani-
festing in the crystalline regions of cellulose nanofibers. When cellulose is synthesized
from the terminal CESA complex, the polymer chains are extruded in parallel to
each other, which align the hydroxyl groups on the C6 of the AGU in the same
5
direction. This Cellulose I structure also has two different sub-allomorphic structures,
referred to as cellulose I? and I?. Depending on the cellulose source, different ratios
of allomorphs I? and I? will be present. I? and I? can be found together within
a given cellulose sample, and even along the same nanofiber. Certain algae and
bacterial cellulose are shown to be more rich in I?, while plant cellulose such as
wood, cotton, and ramie fibers contain mostly I? [14, 15].
I? is a triclinic unit cell containing one cellulose chain. The lattice parameters
of I? are: a = 6.7 A?, b = 6.0 A?, c = 10.4 A?, ? = 118.1?, ? = 114.8, ? = 80.4? [15].
I? is a monoclinic unit cell containing two cellulose chains. The lattice parameters of
I? are: a = 7.8 A?, b = 8.4 A?, c = 10.4 A?, ? = ? = 90?, ? = 96.5? [15]. One notable
difference between I? and I? is in the offset of the aligned polymer chains, and is
illustrated in Figure 1.5 when observing the (011) phase of the crystal structures. In
the I? phase the chains are offset about 1 the length of a cellulose repeat unit in
4
the direction of hydrogen bonding, while in I? the offset is still 25% of the cellulose
chain length but alternates between chains. This causes every other chain to be lined
up in a staggered pattern [10,16].
A more detailed view of the cellulose I lattice structures is shown in Figure 1.6,
which is viewed along the direction of the cellulose chain, or the ??c direction. Figure
1.6a shows the lattice structure of phase I?. The (110) plane is the hydrogen bonding
plane, while the (100) and (010) planes contain van der Waals and hydrophobic
??
interactions. The ??a and b directions illustrate the distance between sheets of
cellulose chains [10,15]. Figure 1.6b shows the lattice structure of phase I?. Hydrogen
bonding between cellulose chains occurs within the (010) plane, while van der Waals
6
Figure 1.5: Schematic showing how the difference in the cellulose polymer offset
between I? (left) and I? (right) effects the resulting plane directions and lattice
structures. The view for both phases is in the (011) phase [16]. ? 2005
??
and hydrophobic interactions occur within the (100), (110), and (110) planes. The b
direction is the direction of hydrogen bonding between chains, and the ??a direction
is the direction of cellulose sheet stacking [10,15].
1.1.3.2 Cellulose II
In native cellulose the chains are synthesized in a parallel conformation, resulting
in the two possible cellulose I crystal structures. Cellulose can be processed, dissolved,
and regenerated into a different crystal structure [17], in which the chains are arranged
in an antiparallel conformation. This means that the C6 hydroxyl groups are in
opposite locations between adjacent cellulose chains. This structure, though not
naturally occurring, is more thermodynamically stable, so cellulose dissolution and
reconstitution or strong alkali treatment can result in the cellulose II structure [18].
Part of this stability is due to the antiparallel conformation, which enables additional
7
a
b
Figure 1.6: Lattice structure of native cellulose phase a) I? and b) I?. The view in
both schematics is perpendicular to the cellulose chain direction,
c [15]. ? 2003
hydrogen bonding in the (110) plane [19, 20]. The cellulose II unit cell is monoclinic
containing two cellulose chains per cell, with lattice parameters: a = 8.0 A?, b = 9.0
A?, c = 10.4 A?, ? = ? = 90?, and ? = 117.1? [10,21]. The overall lattice structure of
cellulose II is shown in Figure 1.7a. This regenerated form of cellulose is found in
fibers such as rayon, lyocell, or viscose [22].
1.1.3.3 Cellulose III
The cellulose III crystal structure can be obtained from either native cellulose
I or cellulose II by treating with an ammonia or amine solution. This structure
8
is softer and more plastic than cellulose I or II and has also found applications in
the textile industry [23,24]. Cellulose III has similar intersheet hydrogen bonding
interactions to cellulose II, however its chains are still aligned in parallel as found in
native cellulose I. Cellulose III has a monoclinic unit cell containing one cellulose
chain, with lattice parameters: a = 4.5 A?, b = 7.9 A?, c = 10.4 A?, ? = ? = 90?, and
? = 105.1? [10,23,24]. A schematic detailing the lattice structure of cellulose III is
shown in Figure 1.7b.
a
b
Figure 1.7: Lattice structures of cellulose a) II and b) III [22]. ? 2011
9
1.1.3.4 Cellulose IV
Cellulose IV is a crystal structure of cellulose that can only be obtained once
native cellulose I has been converted to the cellulose II or III structure [10,25] through
treatment with glycerol [19]. Cellulose IV has a tetragonal unit cell containing two
cellulose polymers per unit cell, with lattice parameters: a =8.0 A?, b = 8.1 A?, c =
10.4 A?, and all angles ?, ?, and ? = 90?.
1.1.3.5 Amorphous
When cellulose is highly disordered and does not exhibit any defined crystal
structures, it is considered amorphous. Amorphous cellulose exists in native cellulose
between the crystallites containing the cellulose I structure. Amorphous cellulose
also tends to be less dense, less stiff, and more reactive than crystalline cellulose [26].
1.1.4 Hydrogen Bonding
The hierarchical cellulose structure, from nanofiber bundles in the cell wall
to chains forming into crystal structures, is held together by a complex network of
hydrogen bonding [12]. Cellulose chains exhibit three different hydrogen bonds, the
first being interchain, or between chains, and the second being intrachain, or within a
single chain. In native cellulose the intrachain hydrogen bonds occur between O3H ?
O5 and O2H ? O6, and the interchain bond occurs between O6H ? O3 [10,15,27]. The
hydrogen bonding network is illustrated in Figure 1.8. Interchain hydrogen bonds
establish bonds between cellulose chains and adds to the strength and resilience of
10
the cellulose. It also makes cellulose difficult to dissolve [28]. Intrachain hydrogen
bonds help maintain stiffness within individual cellulose chains.
Figure 1.8: Inter- and intrachain hydrogen bonding network present between
cellulose chains [28]. ? 2009
1.2 Instrumentation
1.2.1 Optical Microscopy
Optical microscopy is a technique useful for observing and quantifying larger
scale objects in the micron to millimeter size range. Light can be employed and
manipulated in an optical microscopy to enable different techniques that have different
applications in characterization. A schematic example of an optical microscope is
shown in Figure 1.9. For bright field images, the sample is placed onto a stage in
the center of the setup, directly below the objective lens. Light is applied from
underneath the sample stage, illuminating the sample for imaging from a camera set
above the lens setup.
One additional method of imaging with an optical microscopy employs cross
polarizers, the setup for which is also shown in Figure 1.9. Light provided from a
11
source under the sample passes through a polarizer, after which the light vibrates
along only one plane. This light will pass through the sample, interacting with
birefringent material which splits the polarized light causing it to vibrate along
multiple planes. The light then passes through an additional polarizer oriented at
90 ? to the first, called an analyzer. This eliminates any light oriented along the
plane that was allowed to pass through the first polarizer, which eliminates all light
that did not interact with birefringent material, generating contrast in particular
materials and parts of a sample. Typical birefringent materials include liquid crystals,
polymers under deformation, certain crystalline materials, and anisotropically ordered
materials.
Figure 1.9: Experimental setup for an optical microscope with cross polarization
capabilities [29]. ? 2015
Cellulose fibers exhibit birefringence due to the higher index of refraction along
the cellulose chains versus perpendicular to them. Therefore, alignment of nanofiber
12
bundles within different layers of the cell wall appear as birefringent, becoming
more visible. As a result, when using cross-polarized optical microscopy to observe
TEMPO oxidized cellulose, the fibers can be seen with greater contrast, making their
changes in morphology, particularly upon swelling and disintegration, more visible.
Optical microscopy in this work was conducted using an Olympus Accura Zoom
XB70 microscope.
1.2.2 Small angle X-ray and neutron scattering
Small angle scattering is a technique that can investigate the structure of
materials by probing variations in density or electron density at the nanometer scale.
This is a useful technique for studying cellulose nanomaterials and its phase behavior
in solutions. X-rays and neutrons are used for small angle scattering. Some of
the basic theory for SAXS and SANS scattering will be described, along with the
advantages of each technique. Brief explanations will also be given for methods of
data reduction and analysis and understanding scattering data in the context of this
work.
1.2.2.1 Basics of elastic scattering
In a scattering experiment, the probe (i.e. neutrons, x-rays, or photons) will
interact with the molecules in a sample, resulting in a change in the momentum
vector, k, or the energy of the probe. X-rays will interact with electrons while
neutrons will interact with atomic nuclei in the sample, which leads to differences
13
in contrast. This event is illustrated by the schematic in Figure 1.10 [30], and the
change in momentum results in a value called the scattering vector as given by the
following equation: ??? ?? 4?k~s ? k~l? = Q~ = sin(?) (1.1)
?
Where ? is the wavelength of the scattering probe. If this change in momentum
results in no change in energy, it is referred to as elastic scattering, which is the type
of scattering employed in this work. The scattering vector, Q, can yield important
structural information for a sample based on Bragg?s Law, given in Equation 1.2, for
calculating the interplanar spacing, or d-spacing, in a lattice [30].
n? = 2dsin(?) (1.2)
Relating these two equations yields a relation between the scattering vector, Q, and
a given distance in a sample [31]:
2?
d = (1.3)
Q
Figure 1.10: Momentum transfer that occurs during a scattering event, relating to
the scattering vector, Q [30].
14
1.2.2.2 Experimental setup
For small angle scattering experiments, X-ray and neutrons are first generated
from a variety of sources. For many laboratory X-ray sources, such as the CuKa
source from the Xenocs Xeuss instrument employed in this work, are produced from
running a high voltage between two electrodes. The elements used in the source
electrodes produce x-rays with different ranges of energies, which can affect the range
of scattering angles available for measurement. An additional type of X-ray source
is called a synchrotron, which is a circular particle accelerator used to accelerate
photons to relativistic speeds and high energies. For experiments that require high
flux, for example measuring materials that do not scatter strongly, a synchrotron
source is useful as a single measurement only takes seconds, versus minutes or hours
at a smaller-scale laboratory source. Neutron sources are usually either produced
from continuously operating fission reactors, or from spallation sources. Continuous
reactors typically generate neutrons based on the fission reaction of U-235, which
yields 2-3 neutrons at energies of about 2 MeV. Generated neutrons are directed
into beam guides where they are prepared for measurement. In spallation sources,
high energy hydrogen ions are produced by a linear accelerator, deposited into a
synchrotron and accelerated to incredibly high speeds. These ions then collide with a
high atomic number target, producing 10-30 neutrons per ion collision. This design
allows for a high flux of neutrons that can be generated in pulses, which allows for
time-resolved measurements [30].
There are four basic steps to a small angle scattering experiment, shown
15
in Figure 1.11: monochromation, collimation, scattering, and detection. During
monochromation, the beam is narrowed only to allow neutrons of a certain range of
energies. This is typically done with a velocity selector and ensures the wavelength of
neutrons or x-rays in the beam are consistent. Collimation removes neutrons or x-rays
not parallel with the beam. This can be accomplished by placing an aperture after
monochromation and a second aperture before the sample, as is shown in Figure 1.11.
Collimation ensures the measurement of the scattering angle is accurate, and the
distance between apertures can be adjusted to determine the level of accuracy desired.
B. More precise monochromation and collimation will improve the wavelength and Q
resolution, but will decrease the flux of the beam. Neutron beams will typically have
lower flux than X-ray beams, so depending on the sample and the facility broadening
the beam resolution is sometimes needed to complete a measurement in a timely
manner. Scattering is performed as the neutron beam interacts with the sample.
Scattered neutrons are typically detected by a 2-dimensional area detector, which is
placed behind the sample as shown in Figure 1.11. Area detectors are 2D arrays of
individual detectors each representing a pixel on a digital image of the area detector.
Neutron or X-ray counts are measured over the course of a measurement, and a 2D
image of the detector is constructed according to the relative intensities for each
pixel. An example of this image is shown in Figure 1.11 [30].
The area detector can be placed at different distances from the sample, resulting
in a different measurable range of scattering angle, Q. This raw 2D scattering pattern
of the scattered beam can be averaged with respect to Q, resulting in the variation
in scattering intensity, I(Q). Measurements will typically be conducted at a variety
16
Figure 1.11: Experimental setup for a small angle neutron scattering measurement.
An example raw 2D scattering pattern is shown on the area detector [30].
of sample-to-detector distances in order to measure different size scales and a full
range of scattering angle. At short distances a high scattering angle (usually up
to ?0.5 A??1) and therefore shorter length scales, on the order of nanometers, are
measured. At longer distances a low scattering angle (down to ?0.001 A??1) and
longer length scales, on the order of hundreds of nanometers, are measured. The
pattern can be circularly averaged to create a plot of the scattering intensity, I(Q)
as a function of Q. Through data analysis, the relationship between I(Q) and Q can
determine important properties or structure of a sample [30].
1.2.2.3 Scattering contrast
The likelihood of interaction between the scattering probe and the component
can be described by the scattering cross section, which is a general approximation
17
of how large an atom appears to the incident beam. Differences in scattering cross
sections for elements as a function of atomic number using both neutrons and X-rays
is shown in Figure 1.12 [30].
Figure 1.12: Differences in scattering cross sectionof different elements and isotopes
between small angle x-ray and neutron scattering [30].
The contrast factor, ??2, is the difference between scattering length densities
within a given sample. The scattering length varies based on the scattering probe
(neutrons or X-rays) and is dependent on the elements present in a sample. For
example, X-rays and electrons interact with the electron cloud of an atom, while
neutrons interact with the nuclei of an atom. The equation for calculating contrast
factor, ??2, based on the scattering lengths is Equation 1.4, where b is scattering
18
length, v is the molecular volume, ? is scattering length density (SLD), and A and B
represent different components of the sample, for example, particles and solvent [30]:
( )2
bA bB
??2 = (?A ? ?B)2 = ? (1.4)
vA vB
In X-ray scattering, the size of an electron cloud generally increases with the
number of electrons, or the atomic number of an atom. Thus, metals and other
heavy atoms will scatter x-rays more strongly than lighter atoms like hydrogen,
oxygen, carbon, nitrogen, etc. Alternatively, neutrons scatter based on the isotopic
nature of an element, illustrated in the difference between scattering cross section
of hydrogen and deuterium with neutrons shown in Figure 1.12. This difference in
contrast can be useful for measurements with SANS, and allow the scattering from
different structures within the system to be highlighted using a technique referred to
as contrast matching.
An example of contrast matching is shown in Figure 1.13. Having components
in a system with a greater difference in SLDs creates greater contrast, meaning the
scattering associated with that system becomes more intense. In a multi-component
system, it can be difficult to decouple the scattering associated with each component,
but contrast matching can emphasize a single component. The large difference in
SLD of Hydrogen and Deuterium is crucial for this technique. The SLD of many
materials, such as polymers or solvents, will fall between that of Hydrogen and
Deuterium, which means a precise mixture of hydrogenated or deuterated can exactly
match the SLD of a variety of samples.
19
Figure 1.13: Examples of different types of contrast: ?finite contrast? shows a
two-component system with differing SLDs; ?zero contrast? shows a system where
each component has the same SLD; multiple contrasts shows a 3 component system,
each with different SLDs; ?contrast matching? shows a 3 component system where
two of the SLDs are equal, emphasizing the third component [30].
1.2.2.4 Data analysis and interpretation
While interpreting scattering data can vary based on the experiment, a general
case of the equation for scattering intensity is helpful for a basic understanding of
scattering data. Equation 1.5 models the overall macroscopic scattering cross section,
or absolute scattering intensity, for a system of globular inhomogeneities or objects
in a matrix or solvent.
I(Q) = ???2P (Q)S(Q) (1.5)
20
Where ? is the volume fraction of the scattered object or particle, ?? is the
contrast factor, P(Q) is the form factor, and S(Q) is the structure factor [30,31]. The
form factor is representative of intra-particle interactions and takes into account the
spatial extent and shape of the scattering object or particle [32]. Generally, the form
factor is given by the Fourier transform of the density distribution of a scattering
object. The structure factor represents the inter-particle scattering and accounts for
interactions between particles in a sample.
Small angle scattering data can be interpreted in many ways, but a common
place to start is to look at exponential dependence of intensity on the scattering
angle, Q, in the scattering data, as shown in Figure 1.14 [30]. Identifying regions of
exponential Q dependence on intensity can discern the shape or dimensionality of a
particle, or the behavior of a polymer chain. Additionally, the particle size can be
approximated from the corresponding Q range of the slope.
Identifying a plateau in scattering data can reveal a radius of gyration, Rg,
through the Guinier approximation. In a given range of data qRg < 1, the Guinier
approximation is valid, and can be estimated by the following equation [30]:
( )
Q2R2g
I(Q) ? exp ? (1.6)
3
An example scattering function of a cylindrical model is plotted in Figure 1.15.
In the plateau region at low Q, the size scale corresponds to the large R2g of the overall
particle. At intermediate Q, the size scale corresponds to the 1-dimensional aspect
of the cylinder, the radius. Beyond this region at high Q, the decay corresponds
21
Figure 1.14: Examples of how the expo nential behavior of the scattering intensity
can identify the dimensionality of a particle, or the behavior of a polymer chain [30].
to the surface fractal typical of Porod scattering, as shown previously in Figure
1.14 [30]. Obtaining data with a plateau requires a sufficiently low concentration,
as with increasing concentration, inter-particle interactions can begin to dominate
the scattering at larger length scales, resulting in the low Q plateau no longer being
visible.
One other common feature in scattering data is a peak. A peak, or a local
maximum if it is broad, results from constructive interference from inter-particle
scattering and is manifested in the structure factor. A broad peak can be a sign of
a standardized interparticle distance in a colloidal suspension, or at much smaller
length scales a sharp peak can indicate the presence of precise crystalline ordering.
Measuring crystalline peaks can be accomplished using WAXS, which is employed in
this work. WAXS is the same technique as X-ray diffraction (XRD), measuring the
22
Figure 1.15: Example scattering curve of a cylinder, showing the two guinier
regions associated with the Rg2 of the overall cylindrical particle, and Rg1 associated
with the particle radius [30].
same general Q range, but is experimentally conducted like small angle scattering
at very small sample-to-detector distances. Based on Bragg?s law as previously
discussed, the peaks in scattering intensity will correspond to the spacing of a crystal
lattice based on the relation of d = 2?/Q.
1.2.2.5 Facilities
SANS in this work was conducted at the Center for High Resolution Neutron
Research at the National Institute for Standards and Technology on the NG-B 30
meter neutron beamline [33]. SAXS and WAXS were conducted at the University of
23
Maryland X-ray Crystallography Center, using the Xenocs Xeuss instrument.
1.2.3 Fourier transform infrared spectroscopy
Infrared radiation spectroscopy involves the passing of infrared radiation
through a sample. A portion of the radiation is absorbed by the sample depending on
the molecular content and interaction within the sample. This can be used to identify
materials, used as quality control of a sample, and used to identify the relative
amounts of components in a mixture. FTIR allows for a fast scan of a sample at all
infrared frequencies at the same time. To accomplish this, an interferometer is used to
split the IR beam in two, modifying the optical path of one beam, then recombining
them to form an optical interference pattern called an interferogram. The resulting
beam exiting the interferometer is infrared radiation that is oscillating in intensity.
After the infrared radiation beam interacts with the sample, an interferogram is
obtained that is converted to a single line absorbance spectrum through Fourier
deconvolution. A typical experimental setup for FTIR is shown in Figure 1.16 [34].
As infrared radiation passes through a sample, certain wavelengths are absorbed
by the sample depending on its chemical structure. Absorption only occurs when
the vibrations of a molecule caused by the incident beam result in a change in a
dipole moment, or a change in vibrational energy level. Each quality and type of
vibration for a given molecule corresponds to the energy and frequency associated
with a given wavenumber of the incident infrared radiation beam. As particular
wavenumbers of infrared radiation are absorbed by the sample, the resulting beam
24
Figure 1.16: Schematic showing the experimental setup of an FTIR device, including
the infrared radiation source, interferometer setup with the beam splitter, fixed mirror,
and movable mirror, the sample, and the infrared radiation detector [34]. ? 1991
hitting the detector will include a lower intensity of those wavenumbers, resulting in
a change in measured intensity (presented as absorbance or transmission). An FTIR
spectrum is compiled by plotting the measured intensity for each wavenumber in the
infrared radiation spectrum with wavenumbers from about 4000 to 400 cm-1 [35].
Calculating the transmission or absorbance from the measured intensity can be done
using equations 1.7 and 1.8 respectively, where %T is percent transmission, A is
absorbance, Is is the incident beam intensity, and I is the resulting beam intensity
after passing through the sample.
( )
%T = 100? Is (1.7)
( )I
Is
A = log (1.8)
I
FTIR is a useful technique for understanding polymers. It can be used to
25
identify functional groups, measure changes in composition or relative concentration
of groups or components within the sample or used to measure structural properties
like crystallinity, level of oxidation, or particle contamination. For this work, FTIR
was employed to track the removal of EMIMAc from a concentrated cellulose and ionic
liquid gel by identifying the changes in absorbance of a particular peak associated
with EMIMAc at a wavenumber of 1640 cm-1.
A typical FTIR measurement can be done in transmission mode, in which the
infrared radiation beam passes through a sample. However, for some samples, such as
the cellulose and EMIMAc gels measured in this work, the thickness is large enough
that the beam cannot pass through the sample effectively. In this case the attenuated
total reflectance (ATR) tool is used to conduct a measurement in reflective mode.
In this case the sample is pressed against a flat crystal surface, usually diamond,
through which the infrared radiation beam is internally reflected. This is a surface
technique, so it only measures a limited region of the sample, but it allows for the
measurement of thicker samples non-destructively.
1.2.4 Rheology
Rheology is a technique that can measure the mechanical behavior of softer
materials, such as fluids or soft gels. For thick viscous fluids and gels, often a sample
is measured using two parallel plates. A rotational force is applied to the upper
plate resulting in a mechanical response, which can yield information about the
mechanical properties of a sample. For less viscous samples, concentric cylinders
26
or conical geometries are employed, but for fluids and soft gel-like solids, aligned
parallel plates can be used, as shown in Figure 1.17.
Figure 1.17: Example geometries and experimental setups used in rheology experi-
ments [36].
With parallel plates, as was used in this work, the upper plate is rotated,
causing a shear stress in the sample (see Equation 1.9).
? = ?G (1.9)
Where ? is the shear stress, ? is the shear strain, and G is the shear modulus of the
sample.
In this work, rheology was used to measure the point at which a gel transforms
into a viscous fluid. To accomplish this, the gel is first measured using a simple
strain ramp test, in order to identify the limits of the linear elastic region, or the
area of strain before the gel begins to yield or plastically deform. A value of strain
is taken within that region and an oscillatory test of that strain is conducted as a
27
function of temperature. This involves inducing an oscillatory strain on the sample
that is high enough to measure its mechanical response, but low enough that the gel
does not yield.
The induced strain is given by a sinusoidal equation as a function of the
frequency, ?, and time, t. The response of the sample is measured through the
change in the phase lag of the frequency curve. This response is quantified by the
storage modulus, G?, which measures the elastic component, and the loss modulus,
G?, which measures viscous component of the response. The strain, storage modulus,
and loss modulus are given by Equations 1.10-1.12, and in accordance with Figure
1.18. ( ?)
?(t) = ??0sin ?t+ (1.10)
2
? ?0
G = cos(?) (1.11)
?0
?? ?0
G = sin(?) (1.12)
?0
As the sinusoidal mechanical input is applied to the sample, the response is measured
through the time delay in the sample response time through the phase angle, ?,
which measures the time between the sine curves of the input and sample response,
as illustrated in Figure 1.18. The ratio of the loss modulus to the storage modulus
results in a value called the phase lag, or tan(?), which is a measure of the internal
friction within the sample.
??
G
tan(?) = ? (1.13)G
If tan(?) > 1 the viscosity is dominating the mechanical response indicating the
28
behavior of a viscous fluid, while if tan(?) < 1 the modulus is dominating the response
indicating the behavior of a solid gel. The point at which tan(?) = 1 is referred to as
the gel point and is the point at which a gel transitions between an elastic solid or a
viscous fluid. In this work, rheology is used to measure the existence of a gel point
in a soft gel of cellulose and ionic liquid, and the temperature at which it occurs.
Figure 1.18: Plot of the applied amplitude of the rheometer over time showing
the sinusoidal input during a frequency sweep and the phase lag of the material
response [36].
1.2.5 Atomic force microscopy
AFM is a technique that is ideal for measuring surface topography of a sample,
as it tends to have high contrast in the Z direction from the surface of the sample.
AFM relies on atomic level surface forces exerted on a tip immediately close to or in
contact with the surface of a sample. Typically, there are two modes for standard
AFM measurement, contact mode and tapping mode, which are employed depending
on the material being imaged and the desired resolution. The tip of an AFM is
supported on the end of a small cantilever. Subtle interatomic forces exerted on the
29
tip by the sample surface result in slight deflections in the cantilever as the tip is
moved across the surface, allowing the topography to be mapped for an area of the
sample anywhere from 100 nm to 10?s of microns in width. Contact mode can be
damaging, particularly for soft samples or substrates since the lateral motion across
the sample can create large forces, so this is typically reserved for very hard samples.
An example of the interatomic forces that govern the deflection of the cantilever
can be described in the plot shown in Figure 1.19. The repulsive region indicates
that the tip is close enough to the sample that it is being pushed away by interatomic
forces, while the attractive regime causes the cantilever to bend toward the sample.
Contact mode is conducted region A in Figure 1.19 in the repulsive regime, in which
the tip directly exerts forces onto the sample surface during scanning. The distance
between the tip and sample is kept constant by a feedback circuit that adjusts
positioning to account for changes in sample height during scanning [37].
Figure 1.19: The atomic force plotted against separation distance that governs the
cantilever response in an AFM [37].
Non-contact mode, or tapping mode, is conducted in region C in the plot,
in which minor deflections of the cantilever result from attractive forces exerted
30
by the sample surface. In tapping mode, the cantilever is intentionally vibrated
at or near its resonant frequency. A feedback controller is employed to maintain
a constant oscillatory amplitude, and the height of the sample under the tip is
indirectly measured from the feedback inputs required to maintain the oscillations.
Since tapping mode is non-destructive and more suitable for soft materials such as
polymers, it is the method employed for this work.
An example of an experimental AFM setup for tapping mode imaging is shown
in Figure 1.20. A laser is aimed at the top of the cantilever and reflected into a
photodiode, which detects any changes in the height of the cantilever. The function
generator drives the cantilever with a fixed frequency and an amplitude that is
maintained at a constant setpoint, as determined by the lock-in amplifier. The inputs
that are used to maintain the oscillatory amplitude are representative of the height
of the sample at a given lateral position across the sample surface. As scans are run
a two-dimensional topographical map of the sample surface is created to produce an
image [38].
In this work, AFM is used to characterize the dimensions of individual CNF
produced from the TEMPO oxidation and mechanical homogenization process,
and to image thin films of all-cellulose composite mixtures. Samples are prepared
by thoroughly diluting a solution or suspension of material, then applying to an
incredibly smooth substrate, such as a freshly cleaved mica disc or a silicon wafer
that has been cleaned using a UV/ozone process. The solvent is removed by drying
at ambient conditions, spinning, or in a vacuum oven. Any further solvent, in this
case ionic liquid, is removed by thoroughly washing [39].
31
Figure 1.20: Experimental setup for an AFM measurement [38]. ? 2012
1.2.6 Hazemeter- UV/Vis spectrometer
Measurements of light transmission and haze are commonly measured optical
properties of materials. Transparency is a crucial measurement for polymer samples
due to the ubiquity of transparent polymers commonly found in everyday life.
Haziness measures the ratio of light scattered by the sample versus the light that
is transmitted through the sample. Haze can be a beneficial property for materials
such as coverings for solar cell, where increased haze can maximize light scattering
across the surface, or a disadvantage when desiring full optical transparency, like in
flexible electronics or packaging.
A UV/Vis spectrometer that is fit with a 150 mm integrating sphere can
provide accurate transmittance measurements of high scattering samples, as it is
able to measure with less loss of light using internal reflection within the sphere.
The experimental setup for a transmittance measurement using an integrated sphere
32
is shown in Figure 1.21a. Measurement light is passed through a sample that is
placed immediately at the aperture opening of the integrated sphere, so both the
transmitted and the scattered light that pass through the sample are able to be
measured.
a
b
Figure 1.21: Experimental setup using an integrated sphere to measure a) standard
transmittance and b) diffuse transmittance for measuring the percent haze in a
sample [40].
For haze measurements, the standard surface located at the back of the sphere
is replaced with a light trap that captures all simply transmitted light, so the only
light measured by the integrating sphere is light that has scattered from the sample,
referred to as diffuse transmittance. A schematic detailing the setup of a diffuse
transmittance measurement is shown in Figure 1.21b. To calculate the haze of a
33
sample, you divide the diffuse transmittance value by the standard transmittance as
shown in equation 1.14, essentially calculating the amount of light that is scattered
out of all the light that is transmitted through the sample without being absorbed
or reflected [40].
Tdiffuse
Haze% = (1.14)
T
1.2.7 Tensile testing
One of the most fundamental relationships in materials science is between stress
and strain. Strain, , is the deformation response of a material under a given load,
given by Equation 1.15, where L is the length of the tested material. Stress, ?, is
measure by dividing the force of a load by the cross-sectional area of a material, given
by Equation 1.16, where F is the applied tensile load and A is the cross-sectional
area. Young?s modulus, E, is a measure of stiffness of the material, and is measured
by the change in stress and strain in the linear-elastic deformation phase, given by
Equation 1.17.
?L
 = (1.15)
L
F
? = (1.16)
A
d?
E = (1.17)
d
This elastic relationship can be observed in the example stress-strain plot in Figure
1.22.
The initial region of the curve is defined by the Young?s modulus, E, being the
34
Figure 1.22: Example stress vs. strain curve and a schematic of material response
to an applied tensile load [41].
slope of the linear elastic region. This deformation being elastic indicates that the
material will return to its initial shape after the applied load. The point at which
the material begins plastically deforming is referred to as the yield point, where ?y
is the yield stress and y is the yield strain. This yield point typically occurs shortly
after the linear elastic curve and can be calculated by using a 0.2% strain offset and
extrapolating the slope to the point at which it intersects the stress-strain curve.
After the yield point, the material begins to plastically deform, or deform in a
permanent and unrecoverable manner. Plastic deformation continues until the sample
fails, which is indicated by the ultimate stress and strain, ?u and u respectively. In
this work, tensile testing is conducted using a dynamic mechanical analyzer (DMA),
specifically the TA Q800 with the tensile film testing attachment. The experimental
setup is shown in Figure 1.23. The top clamp is fixed to the instrument, while
the bottom clamp is left floating. During tensile testing, the applied force pulls
downward on the floating bottom clamp, applying the load to the film. Sample
preparation of cellulose films for tensile testing is described in Chapter 4 of this
document.
35
Figure 1.23: Experimental setup of a dynamic mechanical analyzer with attached
film tension clamp [41].
1.2.8 Scanning electron microscopy
SEM is a technique commonly used in the field of materials science to identify
nanoscale structures in a sample with high-resolution and high magnification imaging
[42]. In particular, SEM has been a powerful tool for the field of polymer science and
engineering, both in establishing a link between synthesis or processing and polymer
microstructure, and in understanding the relationship between mechanical properties
and fracture or deformation microstructure [43?46]. For this work, the Hitachi
SU-70 provided by the University of Maryland Advanced Imaging and Microscopy
Laboratory (AIMLab) was used to image samples.
Figure 1.24 shows a schematic detailing the major components of a typical
36
SEM configuration. At the top of the column in the image is the electron gun. This
both produces electrons to form the beam, and accelerates them to an energy level
typically between 0.1 and 30 keV [42,43]. In the Hitachi SU-70 SEM in the UMD
AIMLab, a field emission gun is used. The diameter of the electron beam is focused,
aligned, and condensed, at multiple stages using electromagnetically controlled lenses
and apertures, which allows the beam to focus on very small spots on the sample.
The column and sample chamber are typically held under very high vacuum so air
does not scatter electrons from the beam. The electron beam passes through the
sample, and images are produced from electron detectors based on different types
of sample-electron interactions [43,47]. For this work, images shown were gathered
using transmitted electrons only.
Figure 1.24: Schematic of a JSM-5410 scanning electron microscope [43]. ? 2007
Polymers, like cellulose, present some difficulties when being imaged in SEM,
37
so there are techniques and preparations that must be followed to obtain the highest
quality images. First, contrast in SEM is often governed by atomic number (i.e.
number of electrons) of elements in a sample. Most polymers consist of low atomic
number materials- in the case of cellulose, carbon, hydrogen, and oxygen- so it can be
difficult to obtain sufficient contrast for identifying features at the nanoscale [45?47].
Additionally, cellulose is a non-conductive material, which results in a charging
effect due to electron irradiation from the beam [48]. There are two solutions that
can ameliorate the problems listed above. First is depositing a thin coating of a
conductive material such as carbon, gold, silver, or tungsten onto the surface [46].
For this work, a carbon coating was used.
The second solution is to measure samples at a low acceleration voltage,
typically between 0.1-0.7 keV. For this work, samples were imaged using 0.7 keV.
At higher voltages, if the beam remains on the cellulose sample for more than a few
seconds, deterioration in the form of curling, bubbling, or contrast darkening begins
to occur, so a good technique is to focus on an area away from the intended image
sight, then once the beam is aligned and focused properly, move to the feature and
record an image quickly, before beam damage begins. Using a lower voltage means
the contrast will be less sharp, so a balance is required between image quality versus
risk of beam damage.
This work used SEM to examine the fracture surface of cellulose films after
tensile testing. Samples were prepared for measurement using an SEM sample holder
with a loading surface oriented parallel to the beam. Fractured film samples were
adhered to the loading surface using carbon tape, with the fracture surface pointing
38
upward. These samples were then carbon coated, and loaded into the SEM for
imaging. The fracture surfaces were examined together with the tensile testing
results in order to identify the mechanism for mechanical failure in each film [47].
39
Chapter 2: Time-dependence of the TEMPO-oxidation of cellulose
2.1 Introduction
CNF are the most abundant natural nanomaterial in the world. They are
produced by a TEMPO oxidation pretreatment followed by mechanical homogeniza-
tion. Understanding how the structure of cellulose changes during this pretreatment
can provide insight into how CNF is produced. This chapter initially provides a
background and history of cellulose nanomaterials, including mechanical and chemical
treatments and the different morphologies and fiber dimensions that result from each
method of production. The research described in this chapter discusses the changes
in structure of cellulose during the TEMPO oxidation process. The time dependence
is studied by removing aliquots of fibers at various times during the reaction. The
large-scale cellulose macrofibers experience swelling during the reaction, which is
characterized by optical microscopy and fit to a log-normal distribution. SANS was
also used to characterize the change in structure within the cellulose macrofibers at
the nanoscale. The data was fit to a correlation length function model, indicating a
separation between CNF at the nanoscale.
40
2.2 Background
2.2.1 Obtaining nanoscale fibrillated of cellulose
Nanofiber material with diameters at the nanoscale possess a high overall
surface area to volume ratio, which can result in remarkably different mechanical,
optical, electrical, or thermal properties than would be observed in the original
bulk material [49]. Within the field of nanotechnology, research on nanofibers has
been growing in recent years with applications in catalysis, electro-optical films,
fiber-reinforced nanocomposites, microelectronics, gas-barrier films, cosmetics, flame-
resistant materials, and more [50].
The most elementary motif within the hierarchical structure of cellulose are
semi-crystalline nanofibers roughly 3-5 nm in diameter. These nanofibers are arranged
in bundles that are roughly 12-16 nm in diameter. These bundles are then embedded
within the overall matrix of the cell wall. To maximize the mechanical properties of
cellulose based materials, a significant amount of research has gone into breaking
down cellulose to its most elementary state, that of nanoscale fibers with high surface
area to volume ratios. Before cellulose pretreatments were employed, the methods
used to break down cellulose were solely mechanical, for example, high pressure
homogenization, grinding, cryo-crushing, or ultrasonic and enzymatic methods [51].
However, because of the hierarchically organized layers of cellulose nanofibers held
together with hydrogen and van der Waals bonding, these methods result in fiber
diameters between 15-100 nm. This indicates that while mechanical methods can
41
break down the hierarchical structure of cellulose, it cannot break apart the bundles
of nanofibers to the most elementary nanofiber structure [51, 52]. These purely
mechanical processes for obtaining nanofibers are energy intensive and inefficient due
to the tight network of hydrogen bonding between nanofibers within these bundles.
Therefore, additional processing is required to break cellulose down to elementary
nanofibers and obtain maximized surface area to volume ratio [52].
Chemistry-assisted surface modification of cellulose has been employed to work
in tandem with mechanical methods, improve fibrillation efficiency, and fully separate
cellulose nanofibers in the process. Typically charged functional groups are introduced
on the surface of individual nanofibers to create a coulombic repulsion between them,
encouraging full separation of the elementary nanofibers in the resulting suspensions.
One method is acid hydrolysis using sulfuric acid (H2SO4), which helps break down
amorphous regions between cellulose crystallites (see Figure 1.4a). After ultrasonic
homogenization, this process yields cellulose nanocrystals, with dimensions of about
5-10 nm in diameter and 50-200 nm long. Cellulose nanocrystals (CNC) are different
from CNF in that they do not contain the amorphous regions present between
cellulose crystallites. Additionally, CNCs have larger diameters than CNF, which
have diameters of roughly 2-4 nm. CNCs are currently being studied for applications
as nanocomposite fillers and its ability to form liquid crystal suspensions, which
can produce films with interesting optical properties [53]. While the high degree
of crystallinity and one-dimensional structure make CNCs an ideal candidate for
nanocomposite applications, fully separated cellulose nanofibers have a much higher
aspect ratio and further increase the surface area to volume ratio, having greater
42
potential for improved mechanical properties in nanocomposites.
Figure 2.1: AFM images of cellulose nanocrystal particles dried onto mica surface
[54]. ? 2008
An AFM image of CNCs is shown in Figure 2.1, showing a wide array of
diameters and relatively low aspect ratios [54]. In addition to the relatively low
aspect ratios, another disadvantage is that yields of nanocrystals produced from the
acid hydrolysis with H2SO4 are low, typically around 30-50% [55].
Using a mild enzymatic hydrolysis in combination with mechanical homog-
enization can produce higher aspect ratio fibers with diameters as low as 5 nm
and lengths as long as tens of microns, as shown in the AFM images in Figure
2.2 [56,57]. However, the images also show that while some 5 nm nanofibers exist,
the bulk of the sample is dominated by larger fiber bundles as the nanofibers have
been slightly fibrillated, but not fully separated to where individual nanofibers are
visible. These mechanically homogenized cellulose fibers will be referred to as mi-
43
crofibrillated cellulose (MFC). Applications for MFC are as thickeners, components of
filters, or as additives to paper products to improve the surface finish or mechanical
properties [50].
Figure 2.2: Height (a and c) and phase (b and d) images of microfibrillated cellulose
produced from a mild enzymatic hydrolysis and mechanical homogenization, dried
onto a mica surface. Images a and b are 1 x 1 ?m and images c and d are 5 x 5
?m [56]. ? 2007
Recently, a method of extracting individual elementary fibers has been achieved,
yielding fibers of 2-4 nm in diameter and up to a few microns long through TEMPO
mediated oxidation, followed by mechanical homogenization [58]. This homogeniza-
tion can be accomplished through microfluidization, ultrasonication, or even stirring
over a longer time period. The result is a very high yield of TEMPO oxidized
cellulose nanofibers with a narrow dispersity of diameters of 2-4 nm and lengths of
up to a few microns [58]. After homogenization, the suspensions can be purified by
centrifuging to remove larger particles, resulting in yields over 90% by weight. While
44
diameters are consistent, apart from slight variations based on the original cellulose
source, lengths can vary based on conditions such as degree of oxidation or source of
cellulose [59].
2.2.2 TEMPO-mediated oxidation process
TEMPO is a water-soluble, widely-available, and stable hydroxyl radical, and
has been used for efficient and selective conversion chemistry of hydroxyl groups to
aldehydes, ketones, and carboxyl groups under mild conditions. TEMPO has been
used in the oxidation of polysaccharides such as starch, amylodextrin and pullulan for
selective conversion of the C6 primary hydroxyl to carboxylate groups present on the
surface of nanofibers [60,61]. For this process, NaBr and TEMPO were dissolved into
polysaccharide solutions in water at pH of 10-11, and NaClO is added as the primary
oxidant to begin the reaction. When this same reaction is conducted with cellulose,
a similar result is achieved. Figure 2.3 shows the TEMPO/NaBr/NaClO-mediated
oxidation of cellulose, resulting in the selective oxidation of the C6 hydroxyl group to a
charged sodium carboxylate group. During the reaction, NaOH is continuously added
to maintain a pH of about 10-11 [62]. After oxidation, the original semicrystalline
structure of the elementary cellulose nanofibers is maintained as the cellulose is not
dissolved but separated.
The results of the TEMPO oxidation process can vary depending on the type
of cellulose used. According to work done by Okita et. al., the size of the crystals
present in elementary cellulose nanofibers (which can vary depending on the cellulose
45
Figure 2.3: Selective oxidation of C6 hydroxyl groups of cellulose to carboxylate
groups through TEMPO/NaBr/NaClO oxidation at pH 10-11 [62]. ? 2010
source) can affect the carboxylate content of the cellulose [61]. Figure 2.4 shows
the relationship between the elementary nanofiber (here labeled microfibril) width
and the total carboxylate and aldehyde content after the reaction. The calculated
carboxylate and aldehyde content of the cellulose, based on the amount of exposed
hydroxyl groups on the surface of cellulose nanofibers, decreases with increasing
crystallite size, as with smaller crystals, i.e. smaller nanofiber widths [61].
Additionally, the experimental results of maximum carboxylate and aldehyde
content show a similar relationship with crystallite size, supporting the notion that
only C6 hydroxyl groups present on the surface of the nanofibers are oxidized. The
cross-sectional structure is based on well-established structures for cellulose I and is
simplified to be represented by a rectangular cross section [63].
Once the TEMPO oxidation process is completed, the cellulose still retains its
overall fiber macrostructure, and mechanical homogenization is necessary for separa-
46
Figure 2.4: The calculated potential carboxylate and aldehyde content based on the
available C6 hydroxyl groups on the surface of cellulose nanofibers (dotted line), and
the experimentally obtained carboxylate and aldehyde content after the oxidation
process (points) plotted against cellulose microfibril (i.e. nanofiber) width [61]. ?
2010
tion of nanofibers. Different mechanical methods can be employed for this process,
including ultrasonic homogenization, a high pressure microfluidizer, or a household
blender, resulting in a transparent and viscous cellulose nanofiber dispersion in water.
The TEMPO oxidation and mechanical homogenization treatments have, for the first
time, produced wood cellulose fibers that have been successfully broken down to their
individual elementary cellulose nanofibers. These nanofibers will have diameters of
roughly 3-4 nm, and lengths of up to a few microns. Additionally, yields of over
90% were recorded using this process. Cellulose nanofibers prepared from TEMPO
oxidation and subsequent mechanical homogenization will be referred to as CNF for
the purpose of this work.
Compared to MFC and nanocrystalline cellulose, CNF shown in the AFM
image in Figure 2.5 have overall smaller and more uniform diameters. Lengths are
47
Figure 2.5: AFM image of separated, TEMPO oxidized, and mechanically homoge-
nized cellulose nanofibers.
longer than nanocrystalline cellulose, as the amorphous regions within the elementary
nanofiber are still present. In MFC, since there is no chemical pretreatment the
nanofibers are often partially fibrillated off larger scale fibers and can also still be
attached, whereas CNF particles are fully separated.
2.2.3 Properties of TEMPO oxidized cellulose nanofibers
Self-standing films can be produced from suspensions of TEMPO oxidized
and homogenized cellulose nanofibers by filtering and thoroughly drying them. The
resulting films, referred to as nanopaper in this work, have mechanical properties
much stronger than that of normal paper, and conventional laminate polymers such
as polyvinyl acetate (PVA) or cellophane. Figure 2.6 shows a comparison of the
elastic modulus and tensile strength of TOCN, PVA, cellophane, paper, and PVA
composites containing 20% CNF. The mechanical properties of CNF are shown to
48
Figure 2.6: Tensile strength and elastic modulus comparison of TEMPO-oxidized
cellulose nanofibers (TOCN), paper, and other laminate polymer films [52]. ? 2011
be far superior to the other materials. The inherent stiffness and crystallinity of the
elementary nanofibers provide a high elastic modulus. When compared to traditional
paper, the overall surface-area to volume ratio of the CNF films is dramatically
increased, allowing for greater stiffness and strength due to greater entanglement
of nanofibers and increased hydrogen bonding sites [64]. Additionally, the high
tensile strength of CNF films is due to a mechanism of breaking and reforming of
hydrogen bonds during film deformation, as proposed by Zhu et. al. [65]. Figure
2.7a shows overlapping cellulose chains held together by hydrogen bonding. During
deformation the hydrogen bonds break, but as the chains continue to move past
each other additional hydrogen bonds can re-form. This allows the films to maintain
a relatively large stiffness despite having already yielded plastically, as shown in
Figure 2.7b. This mechanism results in a higher toughness as well as a higher
stiffness and strength [65]. Conventionally as the stiffness and strength of a material
increase, it becomes more brittle and more susceptible to a sudden crack-induced
49
failure, and decreased toughness. As a result, materials that are able to demonstrate
high toughness and high stiffness and strength are often sought after in material
selection [66].
a
b
Figure 2.7: a) Proposed mechanism for breaking and reforming of hydrogen
bonds between cellulose chains during deformation. b) Mechanical testing results
of nanofibrillated cellulose films showing high toughness due to elongated plastic
deformation region. Mechanical behavior varies depending on the average diameter
of the fibers present in each film [65]. ? 2015
CNF nanopaper also has interesting optical properties. Due to the gaps between
nanofibers being smaller than the wavelength of visible light, nanopaper films are
optically transparent [64]. Figure 2.8a shows the transparency of nanopaper, as
text can clearly be read through the film. Haze refers to the scattering of visible
50
light that occurs in the film. Figure 2.8b shows an example of haze, as the initial
size beam of light can be observed on the near side of the film, but after passing
through the film scattering has caused the beam to broaden. The nanopaper film
observed in Figure 2.8b has had MFC included in order to maximize haze for solar
cell applications. The light transmission of CNF nanopaper films can be as high
as 90-93%, which is greater than conventional transparent polymer films such as
polyethylene terephthalate (PET), having a light transmission of about 92%. A
comparison of the light transmission values can be observed in Figure 2.8c, indicating
that CNF has transparency high enough to be used in transparent polymer film
applications [67,68].
One promising application for CNF nanopaper is as a transparent polymer
substrate. Transparent and flexible electronic devices have been printed onto CNF
nanopaper, such as transistors, antennae, sensors, organic light emitting diode
(OLED) and solar cells. The relatively low surface roughness makes it better suited
for device printing than conventional paper, and having a sustainable material that
is not derived from petroleum for flexible electronic devices is important given the
current global dependence on fossil fuels [68,69]. CNF films also have excellent oxygen
barrier properties, and can therefore be useful in packaging and food containment
materials [70]. CNF also is a promising candidate for composite reinforcement due
to its one-dimensional shape and high stiffness and strength.
51
a b
c
Figure 2.8: Photograph showing optical (a) transparency and (b) haze properties
or transparent paper from dried CNF [66,67]. ? 2014
2.3 Experimental methods
2.3.1 Oxidation process
In this experiment cellulose was treated with the TEMPO oxidation reaction.
Aliquots were removed from the reaction at different times for characterization.
The process was adapted from previously reported protocols developed by Hu et.
al. [71?74]. For the reaction, southern yellow pine bleached cellulose wood pulp
with a moisture content of about 85% was used. A photograph of the cellulose
52
pulp is shown in Figure 2.9. Sodium hypochlorite solution, sodium bromide, sodium
hydroxide, sodium carbonate, and sodium bicarbonate were all purchased from
Sigma-Aldrich. Initially, 35 g of moist cellulose pulp (?5 g dry) is mixed into DI
water at 1% cellulose by mass, then TEMPO and NaBr are added and dissolved at
concentrations of 0.1 and 1 mmol/g of cellulose, respectively. The reaction began
by adding 35 mL of sodium hypochlorite solution (at 10-15% available chlorine),
dropwise, into the solution. The pH of the solution was monitored with a pH meter
and was initially raised to 10.5 with a 3 M solution of sodium hydroxide (NaOH),
then maintained at 10.5 through the remainder of the reaction. The reaction lasted
for 120 min, and products were collected throughout to be used for preparing CNF
and for SANS study. A photograph showing the TEMPO oxidation reaction in
progress is shown in Figure 2.9. At the start of the reaction, the pulp is not broken
down and clumps can be observed, but the mixture becomes more homogeneous
throughout the reaction as the oxidation occurs.
To characterize the fibers as a function of reaction time, aliquots of approx-
imately 10 mL were removed from the reaction at intervals of 5, 15, 30, 60, and
120 minutes. After removal, the aliquots were immediately quenched with ten-fold
volume of deionized water applied to halt the reaction. Aliquots were washed and
filtered three times using a Buchner funnel with applied vacuum in order to fully
remove the reactive agents. The collected cellulose was weighed to estimate remaining
water content, then redispersed in water and diluted to concentrations of about 1%
by mass. Cellulose concentrations were verified using oven drying.
53
a b
Figure 2.9: a) Bleached cellulose pulp at roughly 85% moisture content. This is the
raw cellulose material used in the TEMPO oxidation reaction. a) TEMPO oxidation
reaction underway. The pulp begins to break down and disintegrate throughout the
reaction.
2.3.2 Optical microscopy
Optical microscopy was performed to examine the morphology of the samples
using an Olympus Accura Zoom XB70 microscope. Samples were prepared by placing
wet fiber aliquots of varying reaction times between a glass slide and cover slip. The
fibers seen in optical microscopy are the macroscale cellulose fibers, i.e. the cellulose
fibers that are observable in conventional paper. Diameters of each large-scale fiber
seen in optical images were measured using ImageJ, and the resulting distributions
were fit to a log-normal distribution function to estimate the center and width of the
distribution [75].
54
2.3.3 Small-angle neutron scattering
To prepare the cellulose fibers for SANS characterization, the water in each
aliquot was solvent exchanged with deuterium oxide (D2O) by centrifuging and
exchanging the supernatant for pure D2O until the water (H2O) content of the liquid
was less than 1%. The fibers were then redispersed into D2O at concentrations
of 1% by mass using magnetic stirring and sonication. SANS measurements were
conducted at the center for high resolution neutron scattering (CHRNS) on the NGB
30 m SANS instrument [33]. The samples were loaded into titanium SANS cells with
quartz windows and a 1 mm path length. Each sample was measured at 1.33, 4.00,
and 13.17 meter detector distances and a neutron wavelength of 6.0 A?. An MgF2
lens configuration was also employed at 13 meters measurement at a wavelength of
8.4 A?, this focuses the neutron beam and minimizes the beam spot size to measure a
lower Q range [76]. The overall Q range for the measurement was 0.001 to 0.5 A??1,
allowing for probing at size scales from nanometers to hundreds of nanometers. The
wavelength spread, ??/?, was 14.8%.
2.3.4 Mechanical homogenization
CNF was produced from the TEMPO-oxidized fibers by homogenization using
a NanoDeBEE microfluidizer from BEE International at a pressure of 30 kpsi. 120-
minute TEMPO oxidized cellulose is prepared at a concentration of about 0.7 vol%
(1 wt%) cellulose and then passed through the microfluidizer 1 to 4 times.
55
2.3.5 Atomic force microscopy
1 cm mica disks were freshly cleaved with a razor blade. Suspensions of each
CNF sample were diluted 10,000 times with water, then a drop was placed on
the freshly cleaved mica and allowed to dry at ambient conditions. Micrographs
were obtained using an AFM -DI Nanoscope III multimode in the University of
Maryland Surface Analysis Center. Diameter was measured using Gwyddion and
measuring height profiles across individual fibers. Lengths were measured using
ImageJ. Both lengths and diameters were organized into histograms and fit to a
log-normal distribution. Roughly 300 fibers were measured for each sample set.
2.4 Results and discussion
2.4.1 Optical microscopy
The macrofiber samples from aliquots taken throughout the TEMPO oxidation
reaction were imaged with an optical microscope as shown in Figure 2.10. Micrographs
in Figure 2.10a and 2.10b were obtained using bright field, and micrographs in Figure
2.10c and 2.10d were obtained with cross-polarizers. At lower oxidation times, the
fibers maintain their overall structure, with some swelling observed. At 120 min
oxidation time, the macrofibers have swollen in diameter, and the overall structure of
the fiber starts to break down. The fiber interface becomes less visible, and smaller
scale fibers can be observed in the cross polarized images expanding outward.
The macrofiber widths were measured using ImageJ by manually measuring
56
a b
c d
Figure 2.10: Optical microscopy micrographs taken of 5 (a,c) and 120 min (b,d)
fiber aliquot samples in bright field (a,b) and using cross polarizers (c,d). throughout
the TEMPO oxidation reaction. The left-hand column are bright field images, and
the right-hand column is cross-polarized images taken of the 5- and 120-minute
aliquot samples.
the cross section of each fiber. Results were obtained from a 1.4 mm by 1.8 mm
micrograph for each sample. A virtual grid was overlaid onto aliquot micrographs,
with each macrofiber cross section measured in each square of the grid to account for
variations in width along each macrofiber length. At least 180 width measurements
were obtained to estimate macrofiber diameter for each sample. The fiber diameter
measurements were plotted in histograms, shown in Figure 2.11, with bins set by
Equation 2.1, where ? is the standard deviation and N is the number of measurements.
1
# of bins = 3.5? ? ?N 3 (2.1)
57
Figure 2.11: Log-normal distribution function fit to the macrofiber diameter
histograms for 5- and 120-minute fiber samples. A best fit line for a log normal
distribution is plotted for the histogram data.
The plot presented shows diameters for the 5 and 120 minute oxidized samples.
The 5-minute sample histogram is centered on a diameter of about 30 ?m, while
the 120-minute sample histogram is broader and centered on a diameter of about 75
?m. The distributions were fit to a lognormal equation with the probability density
function:
1 (ln(x)??)2
PDF = ? e? 2?2 (2.2)
x? 2?
The parameters ? and ? are used to define the location and shape of the natural
logarithm of the distribution, they do not directly correspond to real diameter values.
The mean diameter of a given log-normal distribution is given by the following
equation:
1 2
Mean Diameter = ?M = e
?+ ?
2 (2.3)
58
The standard deviation of the distribution, used to describe how the distribution
deviates from the mean diameter, is given by the following equation:
?
Standard deviation = ?SD = (e2?+?
2)(e?2 ? 1) (2.4)
Figure 2.12: Mean diameter large-scale fibers of each aliquot sample from the
TEMPO reaction plotted against reaction time. Diameters are measured from optical
microscope images.
Figure 2.12 shows the mean diameter compiled from the aliquot samples of
varying reaction time, with the error bars showing the standard deviation of the
distributions. The mean diameter increases with reaction time, as was observed
in optical microscopy. Additionally the standard deviation of these distributions
increases dramatically at 120 minutes of reaction time, indicating that the macrofibers
do not swell uniformly resulting in a larger variation in macrofiber diameter. This
increase in diameter can be attributed to the effects of the TEMPO oxidation
reaction. The surface level hydroxyl groups responsible for hydrogen bonding between
59
individual cellulose nanofibers have been converted to anionically charged sodium
carboxylate groups. The anionic charge creates a repulsion that separates the
individual nanofibers within the nanofiber bundles that make up the layers of the
plant cell wall, resulting in swelling that is visible at the macroscale. The jump in
diameter between 60 and 120 minutes indicates that the diameter does not necessarily
increase linearly with time.
2.4.2 Small angle neutron scattering
SANS data of all 5 D2O exchanged aliquots taken from the TEMPO oxidation
reaction were obtained. The circular average of each corrected data file plotted
against the scattering vector, Q, are plotted in Figure 2.13.
Figure 2.13: Reduced SANS data of each aliquot sample.
The scattering data for the cellulose fiber sample was reduced using SANS
macros from the CHRNS through Igor, software [77] and the scattering data was
modeled using the fitting software SASView. The data was modeled with a three-term
60
function of the form as shown in Figure 2.14:
A C
I(Q) = + +B (2.5)
Qn 1 + (Q?)m
Where A is the Porod prefactor, n is the Porod exponent, C is the Lorentzian
prefactor, m is the Lorentzian exponent, ? is the correlation length, and B is incoherent
background. The three terms in this function represent three contributions to overall
scattering, and are shown in Figure 2.14. The first term corresponds to intensity
at low-Q, representing large aggregate structures. The second term corresponds to
mid-Q scattering, representing the presence of nanometer scale structures. The third
term is for incoherent background, which depends mostly on the hydrogen content
in the sample [78]. SANS data is reported in Figure 2.13. From 5 minutes to 120
minutes, there is a distinct evolution in the mid-Q scattering between 0.01 A??1 > Q
> 0.04 A??1, while high- and low-Q scattering remains relatively constant.
The scattering background is subtracted by fitting a constant to the high-q
region, accounting for the incoherent backgound portion of the fit function (shown in
orange) in Figure 2.14. Due to the large size of macrofibers the samples may not have
been homogeneous when loaded into SANS cells resulting in slightly different overall
ratios of cellulose to D2O in each sample. To account for this, after background
subtraction SANS curves were normalized at the high-Q region. Since intensity will
scale with overall volume fraction, this normalizes each sample based on varying
cellulose content.
After normalization, a clear trend in the evolution of the mid-q region, roughly
61
Figure 2.14: Plot showing the different components of the correlation length fitting
function used for fitting the obtained SANS data.
0.01 A??1 < Q < 0.04 A??1, can be observed in the inset shown in Figure 2.15a. The
intensity of the mid-q region increases with oxidation time, with the most significant
increase being between 60 and 120 minutes. The intensity of the high-Q region
remains the same for all samples, with the same power law dependence. The low-Q
regions show a slight decrease in intensity with increased reaction time. The fitting
parameters for each measured sample are displayed in Table 2.1.
The Lorentzian exponent, m, was fixed during the fitting, the background was
measured separately prior to fitting and subtracted off, the remaining parameters
were left unfixed. The power law exponent, n, shows a slight decrease with oxidation
time, while both the Lorentzian scale and correlation length increase.
The correlation length is related to the distance between individual nanofibers
within overall macrofibers. As reaction time increases, hydroxyl groups at the surface
of the nanofibers are oxidized and become charged, creating electrostatic repulsion
62
a
b
Figure 2.15: a) SANS spectra that have been normalized with incoherent back-
grounds subtracted. Included is an inset showing mid-q evolution with increasing
oxidation time. Curves shown shifted with fit curves of correlation length fitting. b)
The same data have been plotted with an offset, so each curve is visible, along with
lines obtained from fitting each data to the correlation length function.
between fibers. As this surface charge builds up, the distance between fibers begins
to increase. The correlation length corresponds to a mesh size for the network of
nanofibers within the macrofibers. A similar interpretation is given in Penttila et. al.,
63
Time (min) 5 min 15 min 30 min 60 min 120 min
A ( ?10?5) 6.6 11.9 13.6 31.2 55.0
n 2.25 2.12 2.12 1.93 1.87
C 0.50 0.53 0.59 0.64 1.09
Length, ? (A?) 18.83 19.67 20.94 22.88 29.68
m 3.3 3.3 3.3 3.3 3.3
Table 2.1: Fit parameters resulting from the correlation length fitting. A is the
porod scale factor, n is the porod exponent, C is the Lorentzian scale factor, ? is the
correlation length, m is the Lorentzian exponent, and ?2 is the estimate for goodness
of fit.
in which the mesh size in cellulose nanofiber networks was shown to decrease during
enzymatic hydrolysis, where the mesh size was again attributed to the correlation
length [79]. This length does not, however, indicate a specific object size or distance,
but the length is on the order of the diameter of individual nanofibers, and the
increase indicates an increasing mesh size.
The ?2 parameter is a measure of the goodness of fit, and is relatively constant
until the 120 minute sample, when it increases dramatically. This is likely due to the
variability in macrofiber size observed in optical microscopy. Just as swelling of the
macrofibers leads to more variability in size and to a broader diameter distribution,
the interfiber distances at 120 minutes are more variable, resulting in a broader
shoulder in the scattering data that does not fit the correlation length equation
as successfully. The Lorentzian slope, m, is fixed during fitting, and denotes the
decay of the of the scattering at intermediate Q. The power law scale, A, increases
slightly with oxidation time, where the power law exponent, n, decreases slightly. It
is difficult to interpret variability of these parameters as there are only a few points
64
to fit this part of the data, as the power law scattering extends to much lower Q that
is out of range of the instrument. The power law scattering is due to relatively large
scattering objects, potentially the overall structure of larger scale cellulose fibers or
nanofiber bundles. While the individual nanofibers have begun to separate due to the
anionic charge groups introduced during the TEMPO oxidation reaction, the overall
structure of the large-scale fibers is still present, and mechanical homogenization is
required to fully separate the nanofibers.
a b
Figure 2.16: a) Correlation length and Lorentzian scalar values obtained from
SANS plotted against oxidation reaction time. b) The large-scale macrofiber diameter
obtained from optical microscopy plotted against the correlation length. A linear fit
line is plotted with this data as well.
The correlation length is an indicator of the increase in interfibrillar distance
within each sample. Using optical microscopy data, it has been shown that at large
scales, this correlation length increase indicates swelling in the cellulose macrofibers.
In Figure 2.16, the correlation length from SANS has been plotted against the
macrofiber peak diameter from optical microscopy. There is a linear correlation of
the two measurements consistent with the interpretation that the separation of the
nanofibers with TEMPO oxidation leads to swelling of the macrofibers.
65
2.4.3 Mechanical Homogenization- AFM
a b
Figure 2.17: AFM images taken of diluted nanofiber samples from 1 pass (a) and
4 passes (b) through the microfluidizer. A line profile was drawn in figure b to
illustrate the uniformity of nanofiber height in the 4 pass image. Most nanofiber
heights fall between 1 and 2 nm.
CNF was prepared by homogenizing the 120-minute TEMPO-oxidized cellulose
at different levels. Images were obtained of each sample displaying isolated fibers
on a flat mica surface. AFM images of 1 and 4 pass CNF suspensions are shown
in Figure 2.17a and 2.17b, respectively. In both samples, the heights of the CNF
appear consistent, as evidenced by the line profile displayed in Figure 2.17b. Figure
2.18a shows nanofiber lengths of both 1 and 4 pass CNF, displaying considerable
variability in the fiber lengths. Figure 2.18b shows diameters of both 1 and 4 pass
CNF, with histograms appearing very similar.
Diameters of the nanofibers were estimated by measuring height using line
profiles as shown in Figure 2.17b. For each sample, more than 300 nanofiber heights
were measured to estimate the diameter, and the resulting data were plotted as
histograms shown in Figures 2.18a and 2.18b. Histograms were fit to a lognormal
66
a b
Figure 2.18: Histograms and corresponding log-normal fits for 1 and 4 pass length
(a) and diameter (b) distributions. The length distribution of the 1 pass fibers
have slightly longer lengths and a broader distribution than the 4 pass fibers. The
diameter distributions appear almost identical.
distribution function, with mean and standard deviation calculated as determined by
Equations 2.2, 2.3, 2.4 respectively, with the lognormal fitting plotted alongside the
histograms. In Figure 2.18a, the length histogram of the 1 pass sample is broader
and centered around a larger length than the 4-pass sample. In Figure 2.18b, the
diameter histogram of the 1 pass data is comparably broad to the 4-pass data, and
each histogram is centered at a similar diameter.
a b
Figure 2.19: Mean nanofiber length (a) and diameter (b) as a function of the
number of microfluidizer passes, with a line of best fit. Figure a illustrates the
overall decrease in fiber length and distribution broadness with increased mechanical
homogenization. Figure b shows how the diameter remains constant with increasing
homogenization steps.
67
The mean and standard deviation for length and diameter of each sample are
plotted in Figures 2.19a and 2.19b, respectively. The dotted line in Figure 2.19b is
a line showing the average diameter of all samples, equaling a diameter of about
2.1 nm, with a small range of error. Length values shown in Figure 2.19a decrease
with number of passes, with a dotted line of best fit shown between points. CNF
prepared with 1 pass had a mean length of 428 ? 160 nm, while 4 passes had a mean
length of 311 ? 100 nm.
With increased mechanical homogenization, the diameters do not decrease.
The nanofiber length decreases with increased homogenization, and the length
distributions also become narrower. This indicates that the forces generated from
the high-pressure homogenization, reaching pressures of 30 kpsi, are sufficient to
sever the length of the nanofibers, but not breaking the cellulose crystallite structure
that makes up the nanofiber cross section. When incorporating the CNF into all-
cellulose composites, a narrower distribution can lead to better quantification of the
mechanical reinforcement effects within the composite, therefore CNF with 4 passes
of homogenization were used to produce all cellulose composites.
2.5 Summary
The structural evolution of cellulose throughout the TEMPO oxidation process
was investigated. A typical oxidation reaction takes 120 minutes, so the cellulose
structure was investigated throughout the length of the reaction. Using optical
microscopy, swelling of large-scale cellulose macrofibers was observed, from roughly
68
36 ?m at 5 min to 89 ?m at 120 min. The internal nanostructure of the cellulose
macrofibers was probed using SANS, in which a scattering correlation was observed
corresponding to the elementary interfiber distance that increased from 19 A? at
5 min to 30 A? at 120 min. While this correlation does not directly correspond
to a physical length scale within the cellulose, the increasing correlation length
indicates a qualitative increase in interfiber distance. There is a linear correlation
between macrofiber diameters, indicating that the increase in interfiber distance at
the nanoscale is a possible mechanism for the macro-scale swelling observed using
optical microscopy, as illustrated in the schematic shown in Figure 2.20.
Figure 2.20: Schematic illustrating the hierarchical structure present within large-
scale cellulose fibers. Cellulose nanofibers are aligned into bundles, which are arranged
within the cell wall within the large-scale fibers. During the TEMPO-oxidation
reaction, repulsive charges cause separation between nanofibers, resulting in swelling
in the large-scale macrofibers.
Considerable work has gone into studying the mechanical fibrillation of cellulose
nanofibers. With the advent of the TEMPO oxidation reaction as a chemical
pretreatment, CNF can be obtained with minimal mechanical fibrillation, resulting in
fully separated elementary nanofibers. While the structure of the CNF has previously
been characterized, this work has investigated the intermediate chemical pretreatment
69
process of the cellulose before mechanical fibrillation, and how this process affects
the overall cellulose structure. While the carboxylation of cellulose has been shown
to allow for full separation of CNF after mechanical pretreatment, this work shows
that it also results in a physical change in the swelling of cellulose macrofibers due
to a change in interfiber distance at the nanoscale, providing new insight on how the
separation of CNF is understood.
70
Chapter 3: Molecular cellulose solutions
3.1 Introduction
This chapter discusses the dissolution of cellulose with a binary solvent of ionic
liquid and polar aprotic solvent, referred to here as a MCS. The background material
will discuss previously employed methods of dissolving cellulose, which can mostly be
separated into 3 categories: chemical modification or derivatization, aqueous or protic
systems, or non-aqueous and non-derivatizing direct solvent dissolution. Additional
methods of processing cellulose will also be discussed including the Lyocell method,
mercerization, and an alkali-urea system. A summary of ionic liquid processing of
cellulose will be discussed as well, to inform the work on MCS.
MCS is a fully dissolved solution of cellulose, achieved with minimal heating
and no apparent aggregation. The research conducted regarding MCS will start
with phase behavior and overall scattering behavior as a function of temperature.
Neat cellulose films can be produced from MCS by first evaporating the cosolvent,
dimethylsulfoxide (DMSO), resulting in a dense gel of cellulose and ionic liquid. The
ionic liquid can be removed through a multi-step soaking process using acetone and
ethanol. The structure of these gels was studied by small- and wide-angle x-ray
scattering and rheology, revealing a unique structure present in the films that is
71
dependent on ionic liquid composition. The evolution of this structure was studied
throughout the ionic liquid removal process as well. A model has been proposed to
describe and understand this structure.
3.2 Background
3.2.1 Dissolving cellulose
Cellulose is the most abundant naturally occurring polymer on earth and has
been a material in use throughout human history in applications such as paper
and fabrics. However, cellulose also has a significant history in the field of polymer
science. The first man-made fibers were of Chardonnet silk, introduced in the late
19th century. This silk was the first type of rayon, a manufactured textile product
derived from cellulose that has been regenerated into a fiber. This type of fiber
is referred to as semi-synthetic, as the long-chain polymer structure of cellulose
is provided by plants naturally and is only partially degraded in the fabrication
process. These rayon, or Chardonnet silk, fibers were developed as an alternative to
conventional silks, and by the early 20th century became a significant facet of the
textile market [80].
The first rayon fibers were discovered by treating cellulose with sulfuric and
nitric acid to produce a cellulose derivative called cellulose nitrate in 1846. The
introduction of the nitrate (?ONO2) groups allowed separation between adjacent
cellulose chains, allowing solubility in common solvents and reducing crystallinity.
Cellulose nitrate could then be regenerated into cellulose by extruding thin fibers,
72
allowing the solvent to evaporate, and removing the nitrate groups with a hydrolysis
and sodium hydrodisulfide treatment. This same process used to produce cellulose
nitrate was employed to fabricate the first man-made polymers, such as celluloid
which was commonly used in early film production [81].
An additional method of preparing rayon cellulose fibers was developed called
the cuprammonium process, which became preferred to the Chardonnet method due
to its lower cost. This process involved dissolving cellulose in an ammonium-based
copper(II) hydroxide solution. The solution is extruded through thin spinnerets or
nozzles, and into a dilute sulfuric acid solution in which the cellulose is regenerated,
then washed with acetic acid.
A third type of rayon fibers are known as viscose fibers. This process employs
a derivative of cellulose called cellulose xanthate, produced from a reaction between
carbon disulfide and alkaline cellulose. The cellulose is then regenerated in a dilute
sodium hydroxide solution. This process has been optimized and industrialized to be
cheaper and more efficient than cuprammonium when manufactured at a large scale,
and currently the viscose method accounts for most of the total rayon production
in the world. Due to the many mechanical steps employed throughout production,
significant versatility in properties of the final product can be achieved, however this
means the process must be very carefully controlled in order to obtain reproduceable
fibers [80].
Rayon fiber production found extensive use in the textile industry and was
originally implemented at a large scale in the United States, but has been less
frequently used in recent years as a result of the hazardous nature of the process.
73
The process uses a large amount of water, and production is energy intensive. The
reactants used can all be considered toxic and can be hazardous for workers. The
waste products and emissions from this reaction are harmful to the environment, such
as sulfur, carbon disulfide and hydrogen disulfide. While the process can be described
as sustainable, since the raw cellulose material is derived from plants instead of
from petroleum like synthetic textile materials [80], the chemical modification and
processing of cellulose are still hazardous to the environment, so it is difficult to
rationalize cuprammonium or viscose rayon as a sustainable textile material. In fact,
rayon can no longer be manufactured in the US due to producers not being able to
meet air and water-quality regulations [82, 83]. Currently, viscose rayon is primarily
produced in China, where environmental regulations are laxer [84].
The hazardous nature of the cupramonnium and viscose rayon production has
resulted in a demand for a more environmentally friendly manufacturing pathway
for sustainable cellulose-derived fibers. One such method called the Lyocell process
employs the solvent N-methylmorpholine N-oxide (NMMO) and water, which can
directly dissolve cellulose without any prior treatment or derivatization of cellulose,
such as cellulose xanthate or cellulose nitrate that are employed in rayon production.
Since this is a direct-dissolution process, there are far fewer steps than a derivatized
cellulose process like the production of viscose. Cellulose pulp is dissolved, then the
solution is spun, washed, dried, and dyed. The NMMO is recyclable throughout
this process, which is not the case for all reactants in the viscose process. (Rosenau,
2001) NMMO also exhibits low toxicity, is biodegradable, has a high recovery rate at
>99%. NMMO is currently employed at an industrial scale in the US to produce
74
regenerated cellulose fibers for textile applications [85].
NMMO can disrupt the hydrogen bonding within cellulose, first dissolving the
amorphous regions and forming complexes with the hydroxyl groups, then gradually
breaking down crystalline regions [83]. Additionally, the crystal structure of cellulose
during the Lyocell process has been shown to change from cellulose I to cellulose II,
which can improve the ability of the cellulose to be dyed, an advantage in textile
processing [22].
Figure 3.1: SEM Image showing fibrillation of a Lyocell fiber [86]. ? 2003
However, NMMO is more likely to form hydrogen bonds with water instead
of cellulose. Certain hydrates of NMMO require high temperatures for cellulose
dissolution that would cause degradation, so the amount of water needs to be carefully
controlled. NMMO also requires an antioxidant stabilizer to prevent explosion during
transport and storage. Additionally, Lyocell fibers tend to fibrillate depending on
how carefully the spinning process is controlled, which can negatively affect the
75
resulting mechanical properties and structural integrity. This fibrillation is shown in
Figure 3.1 [83,85,87,88].
The transition of the crystal structure from cellulose I to cellulose II is desirable
in many cellulose applications in textiles for aesthetic reasons. One common method
of converting cellulose I to cellulose II is a process known as mercerization, in
which cellulose is subjected to strong alkali conditions. Mercerized fibers have a
noticeably improved luster and are more water absorbent so they can absorb more
dye than unmercerized fibers. However, mercerization does not fully dissolve cellulose,
it merely results in the transition to cellulose II. Chains within nanofibers change
conformation from parallel to antiparallel packing, but the overall nanofiber structure
remains intact [18].
A similar solvent system has been shown to rapidly and directly dissolve
cellulose by using an alkali (including both NaOH and lithium hydroxide (LiOH))
and urea-based solvent. Typically processes for dissolving cellulose require long time
scales, however this solvent will dissolve cellulose within 2 minutes. Interestingly,
this solvent will only dissolve cellulose at low temperatures. The mechanism for this
solvent system relies on the formation of clusters of NaOH, urea, and water. The
NaOH and urea form ?hydrate? clusters with water molecules. As the temperature
decreases, the formation of solvent clusters increases. The low-temperature rapid
dissolution of cellulose in this system is an entropy driven process, in which cellulose
chain rearrangement is driven by entropy competition between conformational entropy
of cellulose and translational entropy of alkali and urea hydrates. The alkali clusters
break apart cellulose chain packing by disrupting intra and intermolecular hydrogen
76
bonds. The urea hydrates play a role by forming an inclusion complex around the
cellulose chains and alkali-hydrates, this keeps the chain complexes separated and
prevents cellulose aggregation, allowing for full dissolution [89,90].
3.2.2 Cellulose dissolution in ionic liquids
The excessive use of volatile solvents is damaging to the environment through
large volume production, disposal, or ozone depletion potential, and can present
health hazards to workers frequently exposed to evaporating solvents. As a result,
considerable effort has gone into defining parameters for green solvents, such as
production from renewable sources, improved biodegradability, or low volatility.
Recently, ionic liquids (ILs) have shown promise as a green solvent, in particular
due to their low volatility, in many applications in chemical synthesis, catalysis and
solvation [91]. IL also have good thermal and chemical stability, can be recyclable in
certain applications, and have a low melting temperature (below 100 ?C), making
them an ideal candidate for a green chemical solvent [92].
There are particular ILs that are ideal for dissolving cellulose, depending
on the anion and cation chosen. It is widely accepted that anions that are good
hydrogen bond acceptors, such as acetate, formate, or chloride, are effective at
dissolving cellulose [93]. It is estimated that the higher the hydrogen bond basicity
and dipolarity, the greater the ability to dissolve cellulose [94]. The role played by
the anion during dissolution of cellulose is penetrating the overall crystalline and
hierarchical cellulose structure and disrupting interchain hydrogen bonding [95].
77
Figure 3.2: EMIMAc interacting with cellulose repeat unit. Acetate anion, circled
in red, bonding with the hydrogen atom of the hydroxyl group. Imidazolium cation,
circled in blue, bonding with the oxygen atom on the hydroxyl group [96]. ? 2015
The role of the cation in the IL is not as clear and has been the subject of much
debate in the literature. ILs that are capable of dissolving cellulose have been shown
to contain one of the following cations when paired with an anion with a strong
hydrogen bond basicity: imidazolium, pyridinium, ammonium, and phosphonium.
Of these, imidazolium and pyridinium have been most effective. In understanding
the role of the cation, an ionic liquid 1-ethyl-3-methylimidazolium chloride (EMIMCl)
was studied, having an imidazolium cation and a chloride anion. It was shown by
nuclear magnetic resonance (NMR) measurements that no specific interaction existed
between the cation and the cellulose, and the dissolution appeared to be purely the
result of the anion interactions. However, when the ionic liquid EMIMAc, having an
acetate instead of chloride anion, was studied, NMR data revealed an interaction
between the oxygen of the cellulose hydroxyl groups and the acidic protons in the
78
imidazolium cation. This could be due to a weaker interaction between the anion and
cation within EMIMAc, allowing the cation to interact with the cellulose as well [97].
Figure 3.2 shows an example of the proposed interaction between the anion and
cation of EMIMAc and a cellobiose molecule, resulting in cellulose dissolution [96].
Zavrel et. al. reports the dissolution of cellulose in EMIMAc within 20 minutes
as shown using in situ optical microscopy (shown in Figure 3.3). The experiment
had 1% by weight of Avicel microcrystalline cellulose powder dissolved in EMIMAc
at about 30 ?C while images were taken. By roughly 20 minutes the large powder
particles can no longer be observed, and the cellulose appears to be mostly dissolved
without added agitation [98].
Figure 3.3: In situ optical microscopy images of microcrystalline cellulose powder
dissolving in EMIMAc [98]. ? 2009
Once the cellulose is dissolved, regenerating cellulose from solution is a crucial
step in evaluating the efficacy of ionic liquids as a solvent. To regenerate cellulose
from ionic liquid solutions, a nonsolvent, or coagulant such as water, acetone, ethanol,
methanol, acetonitrile, etc. is used. The resulting morphology and microstructure of
the cellulose will vary depending on the regeneration solvent used and the method
of regeneration. For example, rapid and immediate mixing of the solution with
79
a nonsolvent causes cellulose to flocculate, resulting in large powder-like particles.
Extrusion into a nonsolvent bath can result in thin fibers or rods, casting onto a
flat surface and washing with nonsolvent will result in a film, and aerogels can be
prepared by washing and drying under supercritical conditions [99].
After dissolving and reconstituting, the resulting cellulose tends to have a
lower degree of crystallinity than the starting cellulose material, and the regenerated
cellulose also changes from a cellulose I structure to cellulose II, similar to other
cellulose processing such as mercerization.
3.3 Molecular cellulose solutions with aprotic solvents
While ILs have been shown to be effective and green solvents for dissolving
cellulose, they still face a number of obstacles. ILs are highly viscous, which
results in thick solutions that can require excessive mechanical agitation and higher
temperatures to effectively break down cellulose, particularly when dissolving at
higher concentrations. ILs are also expensive, and while they have been shown
to be partially recyclable, more work is needed to limit the amount of IL used in
cellulose dissolution [100]. Additionally, while EMIMAc can effectively break down
cellulose, aggregation is still present in solution. Light scattering studies indicate
that aggregates with a radius of gyration up to 150 nm are still present in solutions of
cellulose in EMIMAc, a size much larger than a single cellulose macromolecule [101].
In work done by Rein et al., a binary solvent mixture using IL and various
polar aprotic cosolvents was studied, claiming to create a true molecular solution
80
of cellulose [100]. This system, particularly with dimethylformamide (DMF) and
DMSO as cosolvents, was shown to effectively dissolve cellulose, with no remaining
aggregates. Figure 3.4 shows SANS data obtained by Rein et al. of a 4 wt%
cellulose solution with a 9:1 DMF to EMIMAc ratio solvent. In this data the lack
of excess scattering at low Q indicates no significant aggregation of cellulose in the
solution [100].
Adding a cosolvent to cellulose and ionic liquid solutions can reduce the viscosity,
speed up the dissolution process, and more effectively dissolve cellulose. This work
details the investigation of the phase behavior of this cellulose solution, including
its dependence on composition and temperature, the structure within the solutions
investigated by SANS, and the effect that water has on the solution. Understanding
the compositional limits of this ternary solution system can expand the knowledge
of the structure and thermodynamics of true molecular level solutions.
Figure 3.4: SANS data for 4 wt% solution of cellulose in a 9:1 DMF to EMIMAc
binary solvent mixture. Circles- scattering from the solution; triangles- scattering
just from the 9:1 DMF to EMIMAc binary solvent mixture. [100]? 2014
81
Understanding the behavior of this and similar solutions is crucial for its
potential use for cellulose dissolution and regeneration. In this work, cellulose
solutions in EMIMAc and DMSO or DMF have been studied using SANS to observe
the presence or absence of aggregates in solution in order to understand chain
conformation and phase diagram behavior.
3.3.1 Experimental methods
Avicel PH101 cellulose powder is purified cellulose with a degree of polymer-
ization of the average number of AGU, DP ? 250, and was purchased from Sigma
Aldrich. EMIMAc (95% purity) was also purchased from Sigma Aldrich and was
stored in a dry glove box to minimize moisture content along with the Avicel cellulose.
DMSO was stored containing molecular sieves to remove water, then was filtered
before use. Deuterated DMSO (dDMSO) was stored in sealed glass vials in a dry
glovebox before use and was purchased from Cambridge Isotopes. Solutions were
prepared over a wide range of cellulose, dDMSO, and EMIMAc compositions by
weighing out a mass of cellulose, then stirring with a specified mass of dDMSO and
sealing in a vial. To avoid exposure to moisture, as all components are hygroscopic,
EMIMAc is added with a syringe through a seal in the top of the vial. The mixture
undergoes mechanical agitation and heating to 90 ?C until the cellulose dissolves and
the solution becomes transparent. Samples that dissolved would become clear within
a few minutes. Some solutions were prepared to intentionally include water (D2O for
SANS measurements), in order to investigate the effect of moisture on the solvation
82
capabilities of the system. In this case, set amounts of D2O were added to dDMSO,
and this mixture was added to the cellulose before being mixed with EMIMAc.
3.3.1.1 Small angle neutron scattering
SANS measurements were carried out at the National Institute for Standards
and Technology (NIST) Center for Neutron Research (NCNR) at the NIST, Gaithers-
burg, Maryland. Typical SANS spectra were obtained with a neutron wavelength, ?
? 6 A? , and a wavelength spread of ??/? ? 0.12. The instrument configurations of
13, 4, and 1 m sample-to-detector distance were used to collect scattering data span-
ning a Q range between 0.003 A? ?1 to 0.5 A? ?1 [33]. Data reduction was conducted
using Igor Pro 6 and the NCNR Reduction Macros [77].
3.3.2 Results and discussion
3.3.2.1 Phase diagrams
Solutions were prepared at a wide range of concentrations to determine the
phase behavior and compositional limits of this system. Based on the behavior of
the scattering in the mid- to low-Q regime, particularly between 0.01 A??1 < Q < 0.1
A??1, a determination was made based as to whether the sample was fully dissolved.
For dissolved samples, the scattering in the given Q range was mostly flat, with
the intensity only slightly larger than the incoherent background and is typically
between 0.3 cm?1 < I(Q) < 0.8 cm?1. Undissolved or aggregated samples displayed
significantly higher intensities within the given Q range, and showed a prominent
83
shoulder indicating monostructure formation and/or aggregates. SANS data from
the samples tended to exhibit one of three behaviors, which were used to construct a
phase diagram: (1) the solution is fully dissolved at room temperature (roughly 20
?C), labeled in green on the phase diagram; (2) the solution is aggregated at room
temperature but dissolved upon heating to 100 ?C, then upon cooling a full or partial
reaggregation is observed, labelled in orange; (3) the solution is not dissolved at
room temperature and remains undissolved upon heating, labelled in red. Solutions
containing added D2O are labelled with a solid circle symbol, while solutions with
no added D2O are labelled with hollow circles.
a b
Figure 3.5: Ternary phase diagram showing the relative compositions of EMI-
MAc, cellulose, and DMSO. a) Shows the lower right corner of the overall diagram,
containing the DMSO-rich region within 50% cellulose and EMIMAc. b) Shows a
smaller section for ease of visualization within 20% cellulose and EMIMAc in the
solvent-rich region. Red samples do not dissolve at all. Orange samples dissolve at
high temperature and reaggregate upon cooling. Green samples remain dissolved
upon cooling.
Two ternary phase diagrams were constructed to visualize the phase behavior
of the system with and without the effect of water. Since there are 4 components in
84
the system (Cellulose, EMIMAc, dDMSO, and D2O) the relative compositions of 3
components will be displayed on each diagram. The first diagram for example, shown
in figure 3.5a, shows the relative compositions of cellulose, EMIMAc, and DMSO.
In this case the position on the diagram is calculated based on the relative molar
amounts of these three components, and the composition of water is not included.
For example, if a solution contained equal parts of all four components, cellulose,
EMIMAc, D2O, and dDMSO (i.e., 25% of each), this diagram would display the
solution compositions with D2O removed, so each of cellulose, EMIMAc, and dDMSO
would be displayed at 33%. Essentially, the phase diagrams shown in Figures 3.5
and 3.6 is a single face on a pyramid shaped quaternary phase diagram including all
four components.
Since EMIMAc is the primary cellulose solvent, the most significant comparison
will be between cellulose and EMIMAc compositions. While the lowest EMIMAc to
cellulose ratio achieved for dissolution in the literature is roughly 3:1, [99,102] in this
work a ratio as low as 1.66:1 was achieved. Therefore using DMSO as a cosolvent can
dramatically reduce the amount of EMIMAc necessary for dissolving cellulose. The
DMSO concentration in the solutions is important for determining its flexibility as a
cosolvent. DMSO shows that even at high rates of dilution the solution will remain
dissolved at the same ratios of cellulose to EMIMAc, whereas for other investigated
cosolvents, such as DMF, dilution is only possible up to a point at which the solvent
is about 5 mol% in DMF [103]. At higher cellulose concentrations, however a higher
ratio is typically required. For example at about 15 mol% cellulose, a ratio of about
2:1 is not sufficient to dissolve cellulose, so cellulose concentration plays a role as
85
well.
The same set of samples were plotted on a phase diagram displaying the relative
compositions of cellulose, EMIMAc, and D2O, shown in Figure 3.6. The hollow
circle symbols, indicating samples containing no added water, are displayed purely
along the cellulose-EMIMAc axis, which helps to visualize the ratio between these
two components. In this diagram EMIMAc to cellulose ratio is still a significant
determining factor, despite a significant D2O composition.
Figure 3.6: Ternary phase diagram featuring the relative compositions of cellulose,
EMIMAc, and D2O.
Water was initially anticipated to play a key role in determining whether
cellulose would be soluble in the blend of EMIMAc and DMSO. In work done by
Kuzmina et. al. the effect of water on cellulose solubility in ionic liquids (with no
86
cosolvent) was investigated. As previously mentioned, aggregates of cellulose are
still present when dissolved in only ionic liquid. With the introduction of water,
upon increasing the water content the size of these aggregates was shown to increase,
though it was not known if this was due to aggregate swelling in water or the efficacy
of ionic liquid as a solvent diminishing due to the presence of water [101]. However,
the data shown in the phase diagram in figure 3.6 indicates that introducing water
into the solution does not necessarily diminish, as the EMIMAc to cellulose ratio
necessarily for dissolution is maintained with increasing water content.
This does not necessarily mean that water has no effect on the components
of this solvent system. The dissolution process is pathway dependent, if water is
introduced with cellulose instead of with DMSO, the cellulose will not dissolve
properly, if at all. The cellulose can become kinetically trapped in a low energy
aggregated state due to the presence of water, making it difficult to dissolve and
requiring excessive ionic liquid, heating, and mixing. Previous work has also shown
that using water as a cosolvent in a cellulose and ionic liquid solution without
DMSO results in a very limited solubility window, so water can inhibit the efficacy
of EMIMAc as a cellulose solvent [104]. However, because water is added at the
same time as DMSO, and DMSO has also been shown to interact with water at
the molecular level [105], it is reasonable to assume that this lessens the interaction
between water and EMIMAc, which can allow cellulose to be dissolved despite
relatively high water content. Additionally, samples containing water exhibit a much
higher viscosity than those with a similar cellulose concentration, ultimately leading
to a gel-like behavior. It is likely that water can cause bridging or local aggregation of
87
cellulose molecular chains, which can explain the increase in viscosity in the solutions.
Despite the fact that the inclusion of water does not appear to inhibit EMIMAc as a
cellulose solvent, this gelation and overall higher viscosity can make these solutions
more difficult to dissolve, requiring more stirring and heating.
3.3.2.2 Small angle neutron scattering results
The SANS spectra of each sample were used as a form of cloud point measure-
ment to evaluate the state of solvation. Spectra of samples that were fully dissolved
displayed curves with flat scattering in the mid- to low-Q regions of low relative
intensities between 0.3 cm?1 < I(Q) < 0.8 cm?1, some with small upturns at very low
Q likely due to dust or contamination. Spectra featuring aggregated or undissolved
samples feature curves with significantly higher intensities and a shoulder in the
mid-Q region.
a b
Figure 3.7: SANS spectra with variable temperature of two MCS solution samples.
a) Solution with an EMIMAc to cellulose ratio of roughly 1.67:1. This sample is
aggregated at room temperature but dissolves at high temperature. b) Solution well
within the solubility region having an EMIMAc to cellulose ratio of roughly 3:1 and
being fully dissolved at room temperature.
88
Figure 3.7a shows a sample that is aggregated at room temperature with an
EMIMAc to cellulose ratio of about 1.7:1. At room temperature, there is a prominent
shoulder observed in the mid-Q region with relatively high intensity, indicating
aggregation in the sample. Upon heating the intensity starts decreasing beginning
between 40 and 60 ?C, then becomes flat scattering between 80 and 100 ?C, indicating
that the cellulose aggregates have fully dissolved. When the sample is cooled back
down to 20 ?C, the intensity increases, almost returning to its initial high intensity
after one hour of equilibration time. Figure 3.7b shows a fully dissolved sample at
room temperature, which upon heating exhibits near identical scattering behavior,
showing minimal scattering at low-Q. The scattering behavior of the sample shown
in Figure 3.7b is remarkably similar to the flat scattering observed in the 100 ?C
scattering shown in Figure 3.7a, indicating that both are representations of solution
scattering for MCS.
Figure 3.8: SANS spectra of a sample series plotted with decreasing ratios of
EMIMAc to cellulose. The relative amounts of D2O are also variable but are not
plotted in that order.
89
One series of samples was run varying the ratio of EMIMAc:Cellulose:D2O, and
is shown in Figure 3.8. The scattering curves are shown for varying EMIMAc:Cellulose
ratio, starting at 2.03:1, and a slight change in intensity can be observed. Between
2.03 and 1.75, the samples become aggregated, displaying high low-Q intensity and
a similar nanostructure present in the mid-Q region observed previously in Figure
3.8. The D2O content, however, does not appear to affect the solubility of the
solutions, because water content decreases as the samples begin to show aggregation.
Figure 3.9 shows the aggregated samples, having EMIMAc to cellulose ratios of
2.03 (red), 1.68 (blue), and 1.66:1 (green), from Figure 3.8 heated and then cooled
back to room temperature. For each of the two most aggregated samples (blue and
green), the aggregation fully dissolves upon heating to 100 ?C, then reforms at a
slightly lower intensity upon cooling to room temperature. The observation that the
aggregation reforms upon cooling indicates a lack of solubility for the samples at
about a 1.7:1 EMIMAc to cellulose ratio. Since they can dissolve at high temperature
but reaggregate upon cooling, this indicates a region of partial solubility that is
limited by temperature.
The sample with EMIMAc to cellulose ratio of 2.03:1 shows a slight decrease
in intensity upon heating, then after cooling the intensity does not increase again,
indicating full dissolution. Additionally, in the two aggregated samples, the water
content is 50% higher in the sample containing 1.66:1 EMIMAc to cellulose than in
the 1.68:1 sample, but the scattering behavior is roughly the same at all temperatures.
Therefore, this significant change in water content does not appear to have an impact
on whether aggregation does or does not reform upon cooling after being fully
90
dissolved.
Figure 3.9: SANS spectra of 3 samples, two of which (blue and green) have
aggregated at room temperature, one (red) is fully dissolved. Plot also shows spectra
from heating each sample to 100 ?C, and cooling to room temperature.
In Figures 3.8 and 3.9, a slight decrease in intensity is observed after heat
cycling the aggregated samples with an EMIMAc to cellulose ratio of 1.68 and 1.66:1,
which could either indicate an irreversible change in the sample after heat cycling
or could indicate that the samples require longer than one hour to fully equilibrate
and recover the original aggregated structure. Figure 3.10a shows a heat cycle of
a sample labelled orange on the previously shown phase diagrams, indicating it is
dissolved at high temperatures and reaggregates upon cooling to room temperature.
As before, we also observe decreases in intensity at 50 and 80 ?C, with flat scattering
being reached between 80 and 100 ?C. Figures 3.10a and 3.10b show two heat cycles
of an aggregated sample. After the first heat cycle, a similar drop in intensity to
what was observed in figures 3.8 and 3.9 is again observed upon cooling.
To determine the reversibility of this transition, this sample was heat-cycled one
91
additional time, the scattering results of which are shown in figure 3.10b. Again, the
scattering intensity of the shoulder associated with the aggregation begins reducing in
intensity as temperature increases. When the sample is cycled again the intensity at
the mid-Q remains constant upon cooling. The initial drop in intensity from the first
heat cycle is likely observed because the sample has not had adequate time to fully
equilibrate, which explains why an additional drop in intensity is not observed after
the second heat cycle. Therefore, after identical heat-cycle protocols, the scattering
behavior of the sample remains the same. This indicates that the dissolution process
for these aggregated samples is reversible.
a b
Figure 3.10: SANS spectra of a sample on the a) first heat cycle (20 ?C to 120 ?C),
and b) second heat cycle (20 ?C to 95 ?C).
3.3.3 Conclusions
Solutions of cellulose fabricated from DMSO, D2O, and EMIMAc were investi-
gated. Using temperature-controlled SANS measurements, the phase behavior of the
solutions was investigated, with the lowest achieved EMIMAc to cellulose ratio for
92
dissolution being 1.66:1. This is the lowest observed EMIMAc to cellulose ratio ob-
served thus far, as the lowest recorded ratio in the literature is roughly 3:1. This low
EMIMAc ratio is attributed to precise humidity control during sample preparation,
and refined protocol for mixing and dissolving the solution, as introducing moisture
at the incorrect stage in the dissolution process can hinder cellulose solubility.
The effect of water, which has been previously shown to limit the solubility
of cellulose in EMIMAc substantially, was also investigated. It was shown that
introducing water into the solution along with DMSO allows the cellulose to be
dissolved at ratios as low as 1.66:1 without inhibiting the potency of EMIMAc as
a solvent. Solutions containing water tended to form a more gel-like structure and
have much higher viscosity, which made the solutions more difficult to dissolve, but
ultimately did not change the minimum required EMIMAc to cellulose ratio for
dissolution. This indicates that so long as water is introduced into the solution along
with DMSO and the humidity and moisture content of the other solution components
is carefully controlled, water will not affect the solubility of cellulose in EMIMAc.
This could be a result of the interaction between water and DMSO, since if water
were to be introduced into the solution individually, it would cause the cellulose to
precipitate and prevent a solution from forming. However, if the water is introduced
with DMSO, a molecular solution can still be obtained.
The ability to dissolve cellulose with as low an EMIMAc ratio as 1.66:1 sig-
nificantly reduces the EMIMAc needed for dissolution. As EMIMAc is a relatively
expensive component, this work makes it more economically viable as a cellulose
solvent. Additionally, it has been shown that using the EMIMAc/DMSO binary
93
solvent system results in a true molecular solution of cellulose, as shown by SANS,
which dissolves relatively quickly and with minimal heating required. The remainder
of this chapter will discuss the process for casting and reclaiming cellulose films from
solution, as well as the structural evolution of the cellulose throughout this process.
3.4 Concentrated gels of cellulose and ionic liquid
To obtain cellulose films from MCS, DMSO is evaporated from the ternary
solution a vacuum oven, leaving a transparent, concentrated gel of cellulose and
ionic liquid. In order to fully understand the structure of cellulose films cast from
MCS, this intermediate gel structure needs to be studied. Because the DMSO in the
MCS allows for such low EMIMAc to cellulose ratios, the resulting gels can contain
much higher cellulose concentrations than previously reported in the literature.
For example, in this work MCS is fabricated with 3.4% cellulose, 6.0% EMIMAc,
and 90.6% DMSO by molar percentage, and because of the relatively low cellulose
concentration it is not difficult to mix and will dissolve within a few minutes of
heating and stirring. The solution is cast into a dish and DMSO is evaporated in
a vacuum oven. The resulting gel will be roughly 38 mol% cellulose (30 vol%) in
EMIMAc, where previous efforts to dissolve Avicel cellulose in EMIMAc have only
reached about 16 mol% (reported value is about 15 wt% cellulose in EMIMAc) [106].
The gels produced in this manner are flexible and self-standing, and stiff enough
to be easily removed from the glass casting dish. This intermediate structure of
cellulose gels produced from MCS after removal of DMSO has yet to be studied.
94
Understanding the gel structure and the structural evolution of the gel as EMIMAc
is removed to produce a cellulose film can reveal important information about the
resulting final cellulose film structure. IL is removed from these gels by washing with
acetone followed by washing with ethanol to obtain the cellulose film. The gels were
studied using SAXS, WAXS and rheology to understand the changes in structure
with increasing cellulose composition, and the structure as a function of temperature
and composition.
Previous work done by Rui Lu in our research group explored spin-coating
thin films of MCS made with EMIMAc and DMSO, resulting in very thin and
amorphous cellulose films. These thin films showed excellent stability in hot water,
acid, and mild base solutions, as well as most organic solvents, making it a promising
application for protective coatings [26]. It has been widely reported in literature that
after regenerating cellulose from an ionic liquid solution using water as a nonsolvent,
a conversion to a cellulose II crystal structure occurs [3, 102, 107, 108], however
with the DMSO-EMIMAc binary solvent used in MCS, cellulose can be regenerated
without being crystallized. This potential for controlling stability, water resistance,
and variability in mechanical properties makes cellulose cast from MCS especially
interesting. In this chapter, the structural evolution of cellulose throughout the
casting process will be explored.
95
3.4.1 Experimental methods
3.4.1.1 Cellulose film casting
MCS was found to readily wet the surface of glass, but casting in a glass dish
produced uneven films, as it would dry to the edges of the dish and the resulting film
would have a greater thickness toward the edges of the dish. MCS would not wet a
Teflon surface, so a dish was designed in which the substrate was glass, but Teflon
sides were cut and fixed to the glass surface with silicone. This resulted in more
evenly dried films that did not cling to the edges of the casting dish. The casting
dish is weighed before adding MCS to monitor the progress of evaporation.
DMSO was evaporated by adding about 10 g of MCS (containing 3.4% cellulose,
6.0% EMIMAc, and 90.6% DMSO by molar percentage) to the prepared casting
dish, then placing it in a 50 ?C vacuum oven with a low vacuum of about 51 kPa.
This temperature and pressure combination allows for evaporation of DMSO without
allowing it to boil, which would create inhomogeneities in the resulting gels. The
evaporation proceeds for about 24 hours, at which point the casting dish is removed
from the oven and weighed and the mass of cellulose and EMIMAc that should
be remaining in the gel is calculated once the DMSO has fully evaporated. When
the total mass of DMSO has evaporated, the evaporation process is complete. The
resulting gel is allowed to cool to room temperature, then is peeled carefully from the
substrate and stored in a desiccator to prevent the absorption of water into the gel.
96
3.4.1.2 Ionic liquid removal
Since EMIMAc is nonvolatile, it must be removed by washing with a solvent
that will dissolve EMIMAc and remove it from the gel. Acetone is shown to be a
weak solvent for EMIMAc, capable of dissolving less than 1% of EMIMAc by weight.
As a result, acetone will remove EMIMAc from the gel slowly, preventing wrinkling
in the resulting cellulose film and maintaining a more homogeneous thickness and
structure throughout the gel. After the initial soaking step, acetone is no longer an
effective solvent for EMIMAc and it will not continue to remove it from the gel, at
which point a stronger EMIMAc solvent, ethanol, is used for washing. The remaining
gel is washed with ethanol until all EMIMAc is removed from the gel, producing a
transparent cellulose film.
To study the structural evolution of the film throughout the washing process,
a gel film of cellulose and EMIMAc is washed with the appropriate nonsolvent for a
set period of time, then dried in the vacuum oven for 2 hours to remove any residual
nonsolvent (acetone or ethanol). The dry gel is weighed to determine the mass of
EMIMAc that was removed during each soaking event. After weighing, a small piece
of the gel is cut, removed, and placed in a desiccator for later characterization.
Initial washing is done with acetone, starting with a soak of roughly 30 minutes
and increasing time to remove additional EMIMAc is removed. Acetone is considered
an ineffective solvent for EMIMAc when after 2 days no further decrease in mass is
observed, indicating that EMIMAc is no longer being removed. At this point, the
ethanol wash begins, beginning with a soaking time of 30 minutes and increasing
97
with each soaking event. The solvent removal process is complete when the film is
soaked for 2 days in ethanol and no decrease in mass is observed.
3.4.1.3 Small and wide angle X-ray scattering
SAXS and WAXS measurements on each gel were carried out using the Xenocs
Xeuss instrument in the University of Maryland X-ray Crystallography Center
featuring a CuK? X-ray source, producing X-rays with wavelength 1.542 A?. Two
configurations were measured, one covering the small-angle regime at about a 2500
cm sample-to-detector distance with a Q range of about 0.005 A??1 < Q < 0.2 A??1,
and one covering the wide-angle x-ray scattering regime. WAXS was conducted
at a close distance of 133 cm with a 2? of about 2 ? < 2? < 42 ?. Gel samples
were measured under vacuum, close to the center of each sample for 90 minutes of
collection time. SAXS data were normalized during the data reduction based on the
transmission relative to an empty beam, and the thickness of each sample measured
with a micrometer.
During temperature ramp measurements SAXS and WAXS were measured
simultaneously, so WAXS data was obtained using the 100k Dectris Pilatus detector.
This detector is smaller and more narrow than the 300k detector, so the detector
image obtained is more 1-dimensional resulting in less precise measurements.
98
3.4.1.4 Fourier transform infrared spectroscopy
FTIR was conducted on each gel sample using the ATR attachment because the
gel samples were too thick for transmission measurement and the resulting spectra
was oversaturated. Each sample was measured in two different locations on each side.
The resulting spectra for each sample were averaged and were normalized based on
the absorption values at wavenumbers 3800 nm?1, which have absorptions of roughly
0.
3.4.1.5 Rheology
Rheology was conducted on gel samples. Initially, a strain sweep was run at
a frequency of 0.1 Hz in order to determine a strain value within the linear elastic
region of the gel. A strain of 0.15% was found to be within the elastic region but
still high enough to elicit a response. A temperature sweep was run at a strain of
0.15% and a frequency of 0.1 Hz from room temperature to 120 ?C.
3.4.2 Results and discussion
3.4.2.1 Fourier transform infrared spectroscopy
The composition of cellulose in the film throughout the ionic liquid removal
process is calculated by monitoring the mass of ionic liquid removed throughout
the washing process. This method is an indirect measurement of the ionic liquid
remaining in the gel, so an alternative method was be employed to confirm these
99
Figure 3.11: Plot of FTIR results showing the decreasing absorbance spectra of
the EMIMAc associated peak.
measurements. FTIR-ATR was conducted on cellulose and EMIMAc samples in
order to qualitatively monitor the EMIMAc content when washing [109]. At roughly
1570 nm there is a peak that indicates the presence of EMIMAc, as shown in Figure
3.11. This peak was fit to a gaussian curve, to quantify the relative cellulose content
in the films.
Figure 3.12: Area of EMIMAc peak at 1570 nm?1 plotted against cellulose volume
fraction. A line of best fit was drawn showing good linearity with the data.
Figure 3.2 shows the peak area plotted against the estimated cellulose volume
100
fraction. The data was fit to a linear trendline with R2 = 0.973, showing good
agreement. If the trendline is extrapolated to a peak area of 0, the estimated volume
fraction will be roughly 84%. This could be due to a number of reasons: 1) the
detection limit for EMIMAc with FTIR is limited, and the peak at 1570 nm will not
appear above background at low enough concentrations; 2) if EMIMAc is removed
more readily from the surface than the center of the gel, and FTIR-ATR measures a
limited volume close to the sample surface, there could be a concentration gradient
resulting in the measured composition being different than the composition at the
center; 3) mass measurements estimating how much EMIMAc is removed at each
soaking procedure are not accurate. One source of inaccuracies in mass measurement
can come from trace amounts of water that may be absorbed into the gel upon
brief exposure to ambient conditions. While the water absorbed is likely negligible
for a single mass measurement, compounded over multiple rounds this could lead
to inaccurate mass measurements. An additional source of error is assuming the
acetone and ethanol is fully removed from the gel between measurements. Due to
the thickness of the films, and how long it takes for the soaking solvents to penetrate
the gels, it can take many hours to fully remove trace amounts of solvent, and it is
difficult to discern when the gels are fully dried. This means some mass measurements
may include trace amounts of solvent remaining in the film, causing an inflated mass
measurement, an underestimation in the amount of EMIMAc removed during that
round of soaking, and thus an underestimation in the overall cellulose composition.
These potential sources of error- peak detection limit, composition gradient, and
mass inaccuracies- all would indicate an underestimation in the cellulose composition,
101
so the real cellulose composition is likely higher than presented in the following
section. For the following analysis, while the overall cellulose volume percent may
not be accurate, the trend of increase cellulose composition is still valid, and still
provides useful information for the purposes of this work.
3.4.2.2 Small angle X-ray scattering
a
b
Figure 3.13: SAXS spectra of each EMIMAc/cellulose gel sample washed with a)
acetone, and b) followed by ethanol, shown in the small angle region. A distinct
structural correlation peak can be observed in each sample.
Gel samples of cellulose and EMIMAc were measured using SAXS. Figure 3.13a
102
shows the change in the gel scattering during acetone washing to remove EMIMAc.
Scattering curves were normalized based on thickness and transmission, resulting in
curves that overlap with similar behavior at high-Q. There is a clear ordering peak
in the 30% gel sample, located at Qm, visible before any ionic liquid removal has
occurred, at roughly Qm=0.020 A?
?1, corresponding to about d=31 nm. As ionic
liquid is removed through acetone washing, Qm shifts to a higher scattering angle
and begins to diminish in intensity and prominence. By the end of acetone soaking,
with roughly 54% cellulose, the peak has almost become a shoulder, and has shifted
to about Qm=0.029 A?
?1, corresponding to about d=22 nm. Figure 3.13b shows
the scattering behavior throughout ethanol soaking. At roughly 63% cellulose the
intensity increases dramatically, while the peak position remains relatively constant.
Overall, throughout ethanol soaking there is a slight change in the peak position
and a trend toward increasing intensity, however the increase is not consistent after
normalization, indicating heterogeneities in the samples.
Figure 3.14: Periodic distance, determined by the location of each SAXS correlation
peak, plotted against cellulose volume fraction.
103
A periodic distance, Qm, associated with this ordering peak can be calculated
based on the following relation:
2?
d = (3.1)
Q
Qm has been plotted against cellulose volume fraction, VF shown in Figure
3.14. The relationship between this Qm and VF can yield information about the
structure of the scattering objects causing the structural ordering. A curve of the
following equation has been fit to the data:
1
d ? ?V 2F (3.2)
This relationship with -1 power law indicates that Qm is changing in 2 di-2
mensions, implying that the scattering objects are 1-dimensional or rod-like in
nature.
3.4.2.3 Wide angle X-ray scattering
WAXS was also conducted of these samples. Figure 3.15a shows the scattering
data throughout acetone soaking. There is a large peak located at roughly 8?, and as
cellulose composition increases the peak intensity increases as well and narrows a bit.
There are also notable shoulders present at about 2? = 19 and 27?, and as cellulose
content increases these shoulders become sharper and more prominent. Additionally,
the large amorphous halo, centered around 2? = 22?, shifts slightly to the left as
ionic liquid is removed. This is likely because the amorphous halo for cellulose is
104
a
b
Figure 3.15: WAXS data for the gels throughout EMIMAc removal process, shown
through a) acetone and b) ethanol washing.
centered at roughly 21?, whereas for EMIMAc it is at roughly 24?, so with less
EMIMAc present the peak shifts slightly to the left. Figure 3.15b shows the WAXS
data throughout the ethanol soaking process. After a very brief period of ethanol
soaking the strong peak at 2? = 8? collapses, along with the shoulders at 19 and
27?. At the end of the ethanol soaking process, all three of these peaks have minimal
intensity remaining, indicating a collapse in this short-range ordered structure due
to ethanol soaking. The resulting WAXS curves underwent an additional two-step
105
normalization process. The first step involved matching the relative intensity to the
corresponding SAXS data for the same sample. Additionally, there is a significant
amount of scattering that carries over from the small angle regime and overlaps with
the lower angle WAXS data. A power law curve was fit to the exponential decay
following the peak in the small angle regime, such that the peak at 8? was cleanly
presented. The scale, exponent, and flat background of this curve was estimated
until the overlapping intensity was no longer visible in the WAXS data after being
subtracted. After subtraction, the WAXS data were normalized by scaling according
to the high angle scattering (2? > 36?) as this area of the spectra has the same slope
across all samples.
A FWHM/2
I(2?) = (3.3)
? (2? ? x )20 + (FWHM/2)2
The WAXS data for each sample was fit to a model composed of 7 Lorentzian
peaks, each described by Equation 3.3, where I is the peak intensity, A is the area
under the curve, x0 is the center of the peak, and full-width at half maximum
(FWHM) is the full width at half maximum of the peak. The Multipeak Fitting
package in Igor was used to accomplish the fitting for this data set, and for each
peak, the area, FWHM, and peak center dictate the peak shape and size Resulting
fitting parameters for all peaks at each cellulose volume fraction are shown in
Table X. Each peak represented a different aspect of the cellulose/EMIMAc gel,
and was treated differently during the fitting process. Parameters in bold were
fixed during fitting. Peaks 0, 3, and 5 (highlighted in green) are included in the
106
cellulose and ionic liquid ordered structure, and the areas of these peaks were set
in fixed proportions of 1:0.22:0.13, of the first, second and third ordering peaks
respectively since they exhibited the same behavior during both acetone and ethanol
soaking. Values in red are small peaks believed to be inherent to cellulose that do
not change throughout fitting. The peak at about 36? is attributed to the length of
the cellulose anhydroglucose unit, and is present in other reported data on cellulose
allomorphs [110]. It is not clear what the peak at 9.5? corresponds to, though it
seems to maintain a constant width and amplitude throughout the EMIMAc removal
process so the area of this peak was fixed throughout fitting. However, the FWHM
of the peak changes slightly, particularly after ethanol soaking begins and peak 0
has shrunk considerably. Values in blue are attributed to either a background (peak
2) or amorphous scattering halos (peak 4). Amorphous cellulose exhibits a broad
halo centered at about 2? = 21? [111] and EMIMAc exhibits a broad halo centered
at about 23? as shown by in-house WAXS measurements. Peak 4 shifts from 22.2?
at the start of the washing process, 30% cellulose, down to around 21.0? at 50%
cellulose, then a slight increase to 21.4? by the end of the washing process, 90%
cellulose. This shift supports the claim that peak 4 is attributed to the amorphous
components in cellulose, since as EMIMAc is removed from the gel, the peak center
shifts closer to that of amorphous cellulose at 21?.
Figures 3.16a and 3.16b show two examples of the 7 peak fitting process for
both the 30 and 45% cellulose samples respectively. Each figure is composed of 3
stacked plots. The bottom plot shows the models for each individual peak. The
middle plot shows the total model, meaning all peaks summed together, overlaid onto
107
a b
Figure 3.16: Examples of fitting of WAXS data using 7 peaks for a) 30% cellulose
and b) 45% cellulose. Each figure consists of 3 plots- bottom) plot showing each
individual peak included in the WAXS modeling; middle) the WAXS results plotted
with the model, a sum of all the WAXS peaks; top) the residual after subtraction of
the data from the applied model.
the experimental WAXS results. The top plot shows the residuals, or the difference
between the experimental data and the multi-peak model. The overall trend of the
ordering peaks (peaks 0, 3, and 5) being narrowed and increased in intensity can be
observed between 30 and 45% cellulose content. Additionally peaks 1 and 6 remain
narrow, low intensity, and relatively constant, while the background and amorphous
peaks (peaks 2 and 4) shift slightly in position and remain very broad with a large
peak area.
The crystallinity index of the cellulose and EMIMAc gel over the course of
the washing process was calculated assuming peaks 0, 3, and 5 were associated with
the ordered structure and is shown in Figure 3.17. Initially, in the gel that has
just finished DMSO evaporation with no soaking, having a cellulose composition
of roughly 30% by volume, the crystallinity index is roughly 0.11. Upon partial
108
Figure 3.17: Crystallinity fraction plotted against cellulose volume fraction, the
plot is divided between acetone and ethanol soaking.
removal of EMIMAc from the gels using acetone the crystallinity increases to roughly
0.18, after which acetone is no longer an effective solvent for EMIMAc removal, the
cellulose composition of the gel at that point is roughly 54%. Soaking continued
using ethanol, which very abruptly results in a collapse in ordering fraction, down to
about 0.02.
The cellulose volume fraction after extraction of the EMIMAc by ethanol is
approximately 90%. The composition of the remaining 10% of the film is not clear,
it could be EMIMAc that is trapped due to the large thickness of the film. The
film was prepared with increased thickness in order to improve the signal in SAXS
and WAXS measurements, which makes EMIMAc removal a longer process. The
FTIR results indicate the EMIMAc content is lower than the volume percent would
suggest but the ATR measurement only probes just under the surface of the film. If
the EMIMAc concentration is greater in the center of the film than at the surface,
109
this would explain why the plot in figure 3.12 appears to overestimate the EMIMAc
content. The cellulose content may also be higher than 90%, due to potential error
accrued during the repeated washing, drying, and weighing steps.
Figure 3.18: FWHM of peaks associated with ordering (peaks 0, 3, and 5) plotted
against cellulose volume fraction.
The width of the peaks taken at half the maximum amplitude of each peak,
or FWHM was a fitting parameter during peak fitting. Figure 3.18 shows the
FWHM throughout the soaking process to remove EMIMAc. Between about 30
and 60% cellulose, the FWHM decreases significantly, then begins to plateau at
higher cellulose compositions. The initial decrease in FWHM occurs mostly during
the acetone soaking period, then when ethanol soaking begins, the FWHM stays
relatively constant. The narrowing of the peaks (decrease in FWHM) indicates the
ordering is improving in the lattice directions associated with the peaks. It is observed
that acetone soaking results in an improvement in ordering for all three ordering
peaks and the relative crystallinity is increasing, where during ethanol soaking when
110
the peak areas are decreasing dramatically and crystallinity drops significantly, the
FWHM remains relatively constant. FWHM has an inverse relationship with the size
of crystallite in an ordered sample, while the peak area indicates the volume percent
of ordering present in the sample. Since FWHM remains constant this indicates that
the quality of the ordering is not decreasing, but the dramatic reduction in area
indicates the decrease in volume fraction of ordering within the film. Upon briefly
washing with ethanol, the ordering is eliminated almost entirely.
111
112
Peak 0 Peak 1 Peak 2 Peak 3 Peak 4 Peak 5 Peak 6
Volume
fraction 2? (?) Area FWHM 2? (?) Area FWHM 2? (?) Area FWHM 2? (?) Area FWHM 2? (?) Area FWHM 2? (?) Area FWHM 2? (?) Area FWHM
0.30 7.6 19.0 0.23 9.5 1.8 0.17 16.9 109.1 1.33 18.4 4.2 0.24 22.2 96.9 0.58 27.9 2.4 0.24 36.1 0.80 0.13
0.30 7.6 18.2 0.23 9.5 1.8 0.18 16.9 103.0 1.33 18.5 4.0 0.23 22.2 96.7 0.60 27.8 2.3 0.23 36.1 0.80 0.13
0.31 7.6 18.2 0.23 9.5 1.8 0.17 16.9 103.6 1.36 18.4 4.1 0.23 22.2 94.4 0.59 27.9 2.3 0.23 36.1 0.80 0.13
0.33 7.6 16.4 0.21 9.5 1.8 0.18 16.9 81.4 1.38 18.5 3.6 0.21 22.0 108.1 0.67 27.9 2.1 0.19 36.1 0.80 0.13
0.35 7.6 17.9 0.21 9.5 1.8 0.17 16.9 82.9 1.39 18.5 4.0 0.20 21.9 107.9 0.65 27.9 2.3 0.19 36.1 0.80 0.13
0.38 7.7 19.4 0.20 9.5 1.8 0.15 16.9 81.2 1.37 18.5 4.3 0.20 21.9 105.8 0.62 27.9 2.5 0.19 36.1 0.80 0.13
0.40 7.7 21.2 0.20 9.5 1.8 0.16 16.9 78.3 1.39 18.5 4.7 0.19 21.8 106.1 0.61 27.9 2.7 0.20 36.1 0.80 0.13
0.45 7.8 22.4 0.18 9.5 1.8 0.13 16.9 73.5 1.39 18.5 5.0 0.17 21.6 108.0 0.60 27.8 2.8 0.19 36.1 0.80 0.13
0.48 7.8 23.6 0.19 9.5 1.8 0.15 16.9 60.6 1.45 18.6 5.2 0.18 21.5 107.7 0.60 28.0 3.0 0.20 36.1 0.80 0.13
0.50 7.7 23.1 0.18 9.5 1.8 0.20 16.9 48.4 2.33 18.7 5.1 0.18 21.0 130.6 0.65 27.8 2.9 0.18 36.1 0.80 0.13
0.53 7.8 26.1 0.19 9.5 1.8 0.19 16.9 37.4 1.63 18.7 5.8 0.18 21.1 120.1 0.60 27.9 3.3 0.19 36.1 0.80 0.13
0.54 7.8 25.6 0.19 9.5 1.8 0.20 16.9 44.3 1.98 18.8 5.7 0.17 21.0 119.6 0.59 27.9 3.2 0.19 36.1 0.80 0.13
0.56 7.8 24.1 0.19 9.5 1.8 0.21 16.9 57.3 3.21 18.8 5.4 0.17 21.0 117.7 0.58 27.9 3.1 0.18 36.1 0.80 0.13
0.63 7.6 1.6 0.24 9.5 1.8 0.23 16.9 15.7 0.51 18.7 1.0 0.17 21.2 86.6 0.50 28.3 2.0 0.20 36.1 0.80 0.13
0.70 7.7 1.4 0.29 9.5 1.8 0.23 16.9 21.2 0.51 18.6 1.0 0.17 21.3 79.6 0.47 28.5 1.5 0.17 36.1 0.80 0.13
0.75 7.8 1.0 0.28 9.5 1.8 0.26 16.9 23.1 0.51 18.6 1.0 0.17 21.2 74.8 0.44 28.7 1.4 0.18 36.1 0.80 0.13
0.78 7.8 2.3 0.31 9.5 1.8 0.23 16.9 20.0 0.51 18.6 1.0 0.17 21.2 85.0 0.48 28.7 1.7 0.19 36.1 0.80 0.13
0.90 7.8 0.0 0.00 9.5 1.8 1.45 16.9 30.6 0.48 18.8 1.0 0.17 21.4 72.0 0.41 28.7 1.2 0.16 36.1 0.80 0.13
0.90 7.4 0.0 0.00 9.5 1.8 0.52 16.9 30.8 0.46 18.8 1.0 0.17 21.4 73.1 0.41 28.7 1.2 0.16 36.1 0.80 0.13
Table 3.1: Fitting parameters from WAXS modeling, including the peak angle 2?, area, and FWHM. Parameters in bold were
fixed during fitting to provide a good fit. Peaks 0, 3, and 5 are included in the cellulose and ionic liquid ordered structure, areas
of these peaks were fixed at ratios of 1:0.22:0.13 for peaks 0, 3, and 5 respectively. Peaks 1 and 6 are small peaks believed to be
inherent to the materials that do not change throughout fitting. Peak 2 is a broad background that varied by sample. Peak 4 is
attributed to amorphous scattering from both cellulose and EMIMAc.
3.4.2.4 Multiscale structural ordering of EMIMAc and cellulose gels
Figure 3.19 shows a schematic of the proposed model that describes the behavior
of the cellulose and EMIMAc gels observed using X-ray scattering. From WAXS, we
observe a significant local ordering peak corresponding to a d-spacing of about 1.1
nm, and from SAXS, we observe Qm corresponding to a periodic distance of about
30 nm, indicating a multi-scale ordering structure. The schematic shows the local
ordering of cellulose/EMIMAc regions, potentially bundles of aligned cellulose chains
separated by EMIMAc at a fixed spacing of 1.1 nm, shown inside the dashed ovals.
The locally ordered regions give rise to nematic long-range ordering corresponding
to the distance between these ordered regions, which is occupied by an amorphous
gel-like region of cellulose and EMIMAc and separating the bundles by a distance
starting at 30 nm and decreasing to 19 nm after soaking with acetone.
During acetone soaking, ordering within the gels is improved based on the
decrease in FWHM and increase in crystallinity index from about 0.11 to 0.17. Qm
decreases from about 30 nm to 22 nm. Based on these results, acetone is selectively
removing EMIMAc from the amorphous regions of the gel but cannot penetrate the
locally ordered bundles. As the total volume of the amorphous regions decreases,
the distance between ordered regions decreases as the overall volume of the film
decreases. Due to the -1 exponential relationship between VF and Qm, the ordered2
structures of cellulose and EMIMAc bundles are estimated to be one-dimensional in
nature. Additionally, the local ordering of the gel improves as the overall volume
fraction of primarily amorphous material decreases. This is further supported by
113
the fact that acetone is a weak solvent for EMIMAc, and is unable to penetrate the
cellulose/EMIMAc ordered structure within the bundled regions. When acetone is
no longer an effective solvent for EMIMAc, at roughly 54% cellulose, this implies
that the EMIMAc present in the amorphous regions has been mostly removed, and
the remaining EMIMAc is largely present in the ordered regions.
Figure 3.19: Schematic illustrating the multiscale ordering system observed in
SAXS and WAXS, and the relationship between the small- and large-scale ordering
distances.
During ethanol soaking, the previously inaccessible EMIMAc in the ordered
regions can be removed. Despite the EMIMAc removal process with ethanol being
gradual over multiple soakings, the ordering peaks are collapsed almost immediately
at 65% cellulose. This implies that despite EMIMAc still remaining within the dense
regions of the gel, the introduction of ethanol disrupts the ordering causing the peaks
to collapse. During this process, there is a slight change in Qm, from 22 nm to 19
nm, due to small changes in volume within the gel. This could ether correspond
114
to small amounts of EMIMAc remaining in the amorphous regions being removed
by ethanol and causing the ordered regions to move closer together, or to a volume
change exclusively within the ordered regions that results in a decrease in long-range
ordering as a result. Additionally, despite the fact that the ordering within the dense
regions is lost, a difference in density between this and the amorphous regions is still
present, and is manifested in the SAXS data which still shows a significant correlation
peak. However, the SAXS data also shows a significant increase in intensity upon
the introduction of ethanol. This implies an increase in overall scattering contrast,
which could be the result of void formation from EMIMAc removal. At the time
ethanol soaking begins, the overall cellulose composition has increased, resulting
in a much stiffer gel. This high stiffness could prevent the gel from relaxing after
removing EMIMAc, resulting in voids instead of a change in volume. This could also
explain why the change in Qm during ethanol soaking is so minimal, since a change
in volume would result in a change in spacing as well.
This schematic explains the relative behaviors of the amorphous and ordered
regions present in the gels based on X-ray scattering data, however more information
is needed to fully understand the system. The amorphous and ordered regions likely
have different compositions of cellulose and ionic liquid. These values can be esti-
mated based on relative changes in EMIMAc content during soaking, but additional
experiments can help more directly measure the compositions. Additionally, further
characterization is required to determine whether void formation is the cause of the
increase in scattering intensity during ethanol soaking. Finally, elemental analysis
would help in more accurately determining the cellulose composition of each gel,
115
a
b
Figure 3.20: Rheology data showing the results of a) strain ramp testing of each
gel sample showing storage and loss moduli plotted against strain, and b) rheology
results showing storage and loss moduli plotted as a function of temperature under
alternating strain of 0.15%.
since it is likely there is some error associated with this measurement.
The amorphous nature of the resulting cellulose films is also worth noting.
Cellulose dissolution and regeneration of cellulose has been extensively reported in
literature. While the conventionally used nonsolvent is water, this results in a rather
significant conversion to cellulose II with relatively high crystallinity fractions of up
to 70% [3, 102, 107, 108]. The structures obtained in this study were almost fully
amorphous, illustrating the unique impact the DMSO cosolvent system can have on
116
the resulting cellulose structure.
3.4.2.5 Rheology
The effect of temperature on the behavior of the concentrated gels is important
for understanding the solubility of cellulose within this system and whether this
ordered structure can be controlled. Rheological measurements were conducted on
gels with 30% cellulose and 19% cellulose by volume. Figure 3.20a shows the storage
and loss moduli of these two gels as a function of shear strain. In the 30% cellulose
gel, there appears to be two transitions in modulus occuring at approximately 0.004%
and 0.06% strain, while in the 19% gel there is only one, with the transition occurring
at approximately 0.05% strain. Modulus data for both gels showed relatively linear
behavior at strains of roughly 0.002%, so this strain value was chosen for a temperature
ramp measurement, the results of which are shown in Figure 3.20b. The 30% gel
showed little to no change in overall modulus behavior upon heating, but the 19%
gel showed a dramatic decrease in modulus between 80 and 100 ?C.
Figure 3.21 shows a plot of tan(?), which was obtained from the relationship
shown in equation 3.4.
G??
tan(?) = (3.4)
G?
A transition of tan(?) in which it increases to over a value of 1 indicates the
viscous behavior is dominating the rheological response of the sample, indicating a
transition from solid to fluid. The 19% gel clearly shows a sharp increase in this
117
value beginning at about 100 ?C, increasing to about 10, showing a clear transition
from solid to fluid and an absence of a significant elastic response of the gel. The
30% gel however, showed a slight increase at 100 ?C, but not increasing beyond 0.4,
and during the rest of heating the value stayed relatively constant, indicating this
sample remained as a gel throughout heating and the elastic response dominates the
rheological behavior of the gel.
Figure 3.21: Plot of tan(?) versus temperature of each of the cellulose and EMIMAc
gels, showing the transition from gel to viscous fluid for the lower concentration gel.
The transition from gel to fluid is representative of the EMIMAc/cellulose gel
dissolving and forming a cellulose solution. This experiment indicates that while the
MCS is fully dissolved with an EMIMAc to cellulose ratio of roughly 1.75:1, after
evaporating DMSO the resulting gel does not redissolve upon heating. This indicates
the role that DMSO plays in allowing cellulose to more easily dissolve. Additionally,
when the EMIMAc:cellulose ratio is raised to about 2.7:1, well over the solubility
limit, the gel can be redissolved upon heating.
118
3.4.2.6 Small angle X-ray scattering as a function of temperature
Figure 3.22 shows the effect of temperature on the multiscale ordering observed
in previous measurements, with Figure 3.22a showing SAXS data and Figure 3.22b
showing WAXS data. In the SAXS data, there is a significant drop in intensity
upon heating to 60 ?C, by 80 ?C the ordering peak begins to disappear, and by 100
?C there is no no significant structure present in the sample. The peak does not
appear to shift to a higher or lower Q during the heating process, indicating that
the ordering distance is preserved as the structure in the gel dissolves.
The WAXS data appears different from previously reported data, shown in
Figure 3.15, because it was measured at the same time as the SAXS data and had
to be obtained with the 100k Dectris Pilatus detector. This results in a less precise
estimate of 2?, so it is difficult to compare directly with previously shown WAXS
results. However, there is a significant change in the WAXS pattern upon heating
from about 23 < 2? < 33, that likely corresponds with peak 5 from the previous
WAXS analysis shown in Table 3.1. Peak 5 was one of the three peaks associated
with local ordering in the cellulose/EMIMAc gels. At 100 ?C the remaining scattering
is dominated by a broad peak at about 21?, which likely corresponds to peak 4
from previous analysis and is a combination of EMIMAc scattering and the cellulose
amorphous halo.
With an excess of EMIMAc present with the gel, ordering is eliminated in
both SAXS and WAXS measurements. Between 80 and 100 ?C, the ordering is fully
eliminated and the gel melts into a viscous fluid, as observed in rheology. After
119
allowing the gel to cool for 7 days, the ordering reappears in both the SAXS and
WAXS data, showing that the melting process is reversible. It is notable that in
SAXS data, the ordering peak is gradually eliminated with increasing temperature,
then upon cooling gradually returns to a comparable intensity. In the WAXS data,
the ordering is eliminated almost completely at 60 ?C, at which point the SAXS
peak was still prominent. Upon cooling, the WAXS ordering at about 27? appears
suddenly, and only after 7 days, at which point the long-range ordering in SAXS
has completely reformed. This indicates that the local ordering may be forming as
a result of longer range ordering. One possible mechanism is that the long range
ordering is a result of a phase separation that occurs upon cooling that creates density
and concentration fluctuations within the gel. The denser regions of cellulose and
EMIMAc in the gel exist at a particular composition that results in a cohabitated
locally ordered structure.
3.5 Conclusions
The structure of cellulose and EMIMAc gels was studied throughout the
EMIMAc removal process using SAXS and WAXS. A multiscale ordering structure
was observed in which 1.1 nm small-scale local ordering containing both cellulose
and ionic liquid is present in ordered regions of the gels. This local ordering forms
due to longer range ordering corresponding to the periodic distance, Qm, between
the locally ordered regions. The long range ordering was shown to change with the
EMIMAc washing process, shifting from 30 to 19 nm. This decrease was related to
120
a
b
Figure 3.22: a) SAXS and b) WAXS with temperature dependence of a gel with
added EMIMAc results in elimination of multiscale ordering structure.
the change in cellulose volume fraction by a power law of -1 , indicating a change in
2
distance between 1-dimensional or rod-like structures. While this local ordering is
ultimately eliminated once ethanol is employed to remove the remaining EMIMAc
121
from the gels, the long-range ordering peak is still present in the small-angle regime
which could play a significant role in defining the microstructure of the resulting
cellulose films.
Additionally, the amorphous nature of these cellulose films makes them unique
amongst other cellulose structures in the literature that have been regenerated from
ionic liquids, as this typically results in a crystalline cellulose II film. This can result
in different mechanical properties, as well as different interactions within a composite
system. In more crystalline cellulose it is more favorable for cellulose chains to
assemble into ordered structures, therefore bonding with like cellulose components.
Amorphous cellulose will not readily form ordered structures, and therefore has a
greater probability to interact with other cellulose structures within a composite.
The multiscale ordering present in these gels is stable upon heating at low
enough ratios of EMIMAc to cellulose. At higher concentrations of EMIMAc, the
ordering structure can be eliminated upon heating, and will ultimately reform when
given sufficient time to cool and equilibrate. This indicates the importance of DMSO
within MCS, as with DMSO present in solution, the cellulose will fully dissolve
at lower ratios of ionic liquid than with no DMSO. Additionally, this shows the
dense cellulose and ionic liquid gels solid nature is likely derived from the multiscale
ordered structure shown in SAXS and WAXS data, because both are eliminating
upon heating and reform upon cooling. The short range ordering is eliminated before
the long range ordering during heating, and reforms after the long range ordering
during cooling. This indicates that the short range ordering forms as a result of the
long range ordering, possibly due to density and concentration fluctuations resulting
122
from a phase separation that occurs during cooling.
While the implication of this multiscale ordering structure on the resulting
cellulose film properties are not known, having the ability to control the presence of
this ordering could be significant in controlling the behavior of these regenerated
cellulose films. In the following chapter, mechanical and water uptake properties of
amorphous cellulose films cast from MCS are investigated, along with their role in
all-cellulose nanocomposite films.
123
Chapter 4: All-cellulose composites
4.1 Introduction
Cellulose nanomaterials, namely CNF and CNC, have garnered significant
attention in polymer nanocomposite research as a way of improving the mechanical
properties of matrix polymers through reinforcement. However, various treatment
is often necessary to enhance compatibility of CNF in nanocomposite systems and
improve the properties of CNF-based films. Common pretreatments of CNF will be
discussed as they pertain to nanocomposite fabrication, such as modifying surface
groups, solvent exchange, and reducing water uptake of CNF. Production methods
of CNF-based films and nanocomposites will be discussed. A summary of work
conducted on CNF nanocomposites will be presented, as well as a discussion on the
range of mechanical properties that result from these nanocomposites.
A brief review of the work done on all-cellulose composites will be presented,
as well as a summary of this thesis work regarding nanocomposites formed by
combining MCS and CNF. Motivations for this project will be presented along
with preparation, resulting properties, and a proposed reinforcement mechanism.
The resulting MCS/CNF nanocomposite films underwent tensile testing along with
testing their water uptake and optical properties. The results of this testing and the
124
interactions between the cellulose components will be discussed.
4.2 Background
4.2.1 Pre-treatments for CNF
In fabricating films from CNF and incorporating them into nanocomposites,
significant work has gone into various treatments to make it a more versatile material.
This section will discuss the modification of CNF surface groups to promote interfacial
compatibility in nanocomposites and improve mechanical properties, redistribution
of CNF into organic solvents to expand potential nanocomposite opportunities, and
reduction of CNF water uptake to combat the hydrophilic nature of cellulose and its
tendency to weaken upon exposure to moisture.
One of the steps in producing CNF involves converting one of three surface
hydroxyl groups to a sodium carboxylate group with a negative charge. While this is
useful in processing as it allows CNF to be fully separated, this charged carboxylate
group cannot form hydrogen bonds [112]. In work done by Fujisawa et. al. they
successfully replaced the sodium carboxylate group (?COONa) with free protonated
carboxylate group (?COOH) by treating the initial CNF with acid. Each suspension
of CNF, both ?COOH and ?COONa, were dried to form self-standing transparent
and flexible films, with varying mechanical properties. While the Young?s modulus
of the ?COOH CNF films was increased, the overall strain to failure decreased,
resulting in stiffer but less ductile films [113].
Limiting nanocomposites to the inclusion of water-soluble polymers is rather
125
stifling as many commonly-used synthetic polymers are hydrophobic [114,115]. There
has been considerable work done examining distributing suspensions of CNF into
organic solvents. Recently, in work done by Okita et. al, solvent exchange for
CNF suspensions in water was done using DMF, 1,3-dimethyl-2-imidazolidinone
(DMI), dimethylacetamide (DMAc), N-methyl-2-pyrrolidone (NMP), and DMSO.
This was achieved using two different methods. The first involved stirring the aqueous
CNF suspension into an equal amount of organic solvent, then heating to 50 ?C
under constant stirring until the water had mostly evaporated, then sonicated to
aid in dispersion. The second method involves exchanging the ?COONa groups for
free carboxylate, ?COOH, groups, as mentioned previously. The ?COOH CNF
suspensions were washed with aqueous 0.1 M HCl followed by acetone with filtration.
The intended organic solvent was added to the acetone-containing suspensions, then
stirred at 50 ?C to allow all acetone to evaporate, then sonicated for dispersion [115].
The results indicated that using the protonation method and exchanging ?COONa
groups for ?COOH, CNF was able to be redispersed at an individual fiber scale in all
the tested organic solvents. Without protonation, the only organic solvent that was
capable of good CNF dispersion after solvent exchange was DMSO. (Okita et al., n.d.)
The ability to redistribute CNF into organic solvents allows for nanocomposites to be
fabricated with a wider range of polymers such as polystyrene [116] or poly(vinylidene
fluoride) [117].
126
4.2.2 Improving water-resistance of CNF-based films
A considerable disadvantage of CNF is its affinity for water. Cellulose is already
hygroscopic, resulting in considerable weakening of cellulosic materials upon water
absorption. The mechanical strength of CNF films is partially based on the dense
network of interfibrillar hydrogen bonding, but as water is absorbed it interacts
primarily with the surface hydroxyl groups on cellulose nanofibers. This can disrupt
the hydrogen bonding between cellulose nanofibers, resulting in a weakening of
mechanical properties. Additionally, the charged surface ?COONa present on CNF
means that as water is absorbed, the charge groups dissociate and will result in
coulombic repulsion between the nanofibers, similar to the dissociation mechanism
that results in a dispersion of CNF via TEMPO oxidation. Improving the water
resistance of CNF is crucial to improving their viability as a structural polymer for
laminate film applications. Common methods for evaluating water resistance are
measuring water uptake and wet tensile strength or stiffness, and both will be noted
in the following review on this work.
The most intensely studied method of water-proofing CNF is through surface
modification. By modifying the surface ?COONa group, the hydrophilicity of CNF
can be controlled. One example involves using metal chloride solutions to trigger
ion-exchange and change the sodium ion on the ?COONa group to a different metal
ion, namely aluminum, iron (III), calcium, and magnesium. The results showed that
trivalent metal ions such as iron (III) and aluminum improved mechanical properties
of CNF films under exposure to humidity, maintaining stiffness on the order of ?10
127
GPa which is comparable to dry CNF. Additionally, upon soaking in water the CNF
films containing metal ions other than Na were shown to absorb significantly less
water. However, each film still absorbed >3 times its own weight in water after one
hour of soaking [118].
Mild esterification processes have been employed to modify CNF films by
selectively grafting alkyl chains to surface hydroxyl groups of cellulose. The resulting
films were shown to absorb roughly 50% less water when exposed to high humidity,
and decrease the loss in mechanical properties upon exposure to moisture as well [119].
Other methods for improving water resistance of CNF center around crosslinking
with other polymers in a nanocomposite film. One example features poly(acrylic
acid) (PAA), which is used to thermally crosslink with cellulose, forming ester bonds.
The crosslinking reduces the number of exposed hydroxyl groups which can reduce
the overall water uptake of the film. This crosslinking was also shown to enhance
the mechanical properties of the films, improving the stiffness and tensile strength as
compared to neat CNF, both in dry and in wet conditions [120]. Similar crosslinked
nanocomposites were formed with other polymers such as PVA, (Hakalahti et al.,
2015) chitosan [121], and carboxymethyl cellulose [122], resulting in improved water
resistance and wet mechanical properties.
These methods have been shown to be effective means of providing CNF-based
films with water-resistance, although it introduces further processing steps into a
material like TEMPO-oxidized cellulose nanofibers. A simpler step for improving
water-resistance is desirable.
128
4.2.3 Fabricating self-standing CNF-based films and nanocomposites
Cellulose nanomaterials, namely both CNF and CNC, have garnered significant
attention in polymer nanocomposite research as a way of improving the mechanical
properties of flexible matrix polymers through rod-like reinforcement. Typically, this
is accomplished by combining a suspension of cellulose nanomaterials with a polymer
soluble in a compatible solvent. This tends to limit applications of CNF and CNC
since they are typically suspended in aqueous media, requiring the matrix polymers
to be soluble in or compatible with water [123,124]. Polymers can be dissolved in
water, then mixed with a suspension of CNF. One requirement is that the polymers
do not contain positively charged end groups, as this will result in agglomeration
since charge groups on CNF contain negative charges. Interactions between the
polymer and CNF should be minimal to also prevent aggregation that could occur
in suspension.
One method for nanocomposite film production with CNF-based materials
includes suspension casting. The composite mixture is deposited into a dish and the
water is allowed to evaporate, which can be done at ambient conditions, an oven
with or without applied vacuum, or at specified humidity conditions. This method
causes a relatively homogeneous concentration increase as the solvent evaporates,
as illustrated in the schematic provided on the left in Figure 4.1 [114]. As fiber
suspensions are relatively low concentration (typically around 1% mass), the sheer
volume of solvent required for a film of adequate thickness is high and requires
long evaporation times. You can decrease this time by increasing the evaporation
129
temperature and/or vacuum to remove solvent more quickly. Faster removal of
solvent can create a gradient of solvent vapor in the environment as it is removed
from the dish, resulting in a more uneven rate of evaporation across the film and
uneven final film thickness. High temperatures, particularly above 60-70 ?C, were
also shown to cause light discoloration and browning in resulting films. Due to these
limitations on temperature and vacuum, well executed suspension casting can require
multiple days of evaporation [125].
Vacuum filtration is another method through which CNF-based films can be
fabricated. The composite mixture is poured into a funnel equipped with micropore
filters with an applied vacuum. While some material can pass through the filter
along with the solvent, the CNF toward the base of the funnel form a dense mesh
network that prevents more material loss. This causes a concentration gradient
moving up from the base of the funnel as more layers of CNF will stack and collapse
as the suspension is filtered. A schematic of this concentration increase during
filtration is illustrated on the right in Figure 4.1 [114]. Once complete, a dense gel
containing the nanofibers remains, which can be dried under hot pressing or surface-
clamped vacuum drying. The gel needs to be dried under constrained conditions or
wrinkling and shrinking can occur. One disadvantage is that depending on the size
of nanofibers, a significant loss of smaller fibers can occur through the membrane
filter. However, it has been shown that drying through vacuum filtration can increase
in-plane orientation of nanofibers as compared to suspension casting, which can
improve their mechanical properties.
For CNF-based nanocomposite films, both suspension casting and vacuum
130
Figure 4.1: Schematic showing the differences in the concentration process in A)
solvent casting and B) vacuum filtration [114]. ? 2017
filtration can be employed. For both methods, a material dissolved or suspended in
the same or a compatible solvent is mixed into the suspension of CNF before the
solvent is removed. In suspension casting, material is contained within the casting
dish and is dried and concentrated uniformly, however evaporation times can be
quite long.
With vacuum filtration there are potential obstacles to overcome when making
nanocomposite films. With pure CNF films, a significant amount of material is lost
and for smaller components within the composite, such as polymer chains or small
nanoparticles, an even larger amount can be lost. The CNF can form a fiber mat
that prevents the loss of excess material, but other components in a nanocomposite
may not be trapped by this network or may be lost at a higher rate than CNF. This
not only makes the fabrication method less efficient as more material is required for
production, it also means one cannot be confident in the relative amounts of the
nanocomposite components present in the final film. Additional characterization
would be necessary to determine the final composition in the nanocomposite. Because
there is a gradient in the concentration for vacuum filtration, this could also result in
131
an uneven distribution of nanocomposite components, depending on how it interacts
with the nanofiber network.
Due to the remarkably high stiffness of CNF, often the goal for CNF-based
nanocomposites is as a fiber reinforcement for a more ductile polymer matrix. One
example incorporates a cellulose derivative, hydroxyethyl cellulose (HEC), into a
CNF-based nanocomposite using vacuum filtration and a heated vacuum drying
process. The resulting composite exhibited interesting mechanical behavior, which is
shown in Figure 4.2. At relatively low percentages of HEC, the overall stiff nature
of the films was maintained but the point of fracture was suppressed, allowing for
increased elongation and tensile strength. The softer HEC helps to plasticize the
CNF and improves the overall toughness. Additionally, a secondary yield point is
present, which could indicate the point where the primary hydrogen bonding of the
CNF network begins to fail. The deformation region after the secondary yield point
is governed by the relatively low modulus of the matrix polymer.
In a related study, the effect of overall polymer stiffness and glass transition
temperature (Tg) was tested by using two different matrix polymer CNF-based
nanocomposites. The first is poly(N,N-dimethylacrylamide) (DMAm) and has a Tg
of approximately 130 ?C, making it a relatively stiff polymer. The second polymer is
poly[(ethylene glycol methyl ether methacrylate)-co-N,N-dimethylacrylamide] (EG54DMAm46)
which has a Tg of roughly 26
?C and a lower modulus than DMAm. Nanocomposite
films were fabricated at compositions between 0 and 50% matrix polymer and mechanical
properties were measured [127].
The stress-strain curves of the nanocomposite containing DMAm are shown in Figure
132
Figure 4.2: Stress-strain curves showing the mechanical behavior of CNF and HEC
nanocomposite at a range of compositions [126]. ? 2011
4.3a. As the DMAm concentration increases the stiffness immediately decreases, but the
percent elongation remains roughly constant. The overall behavior of the curves mimics
the pure CNF film in that the initial linear elastic region reaches the yield point at roughly
the same strain, followed by an extended plastic deformation region until the film fractures.
The stiffness continues to decrease upon increasing DMAm concentration and the tensile
strength and overall toughness decrease. The mechanical behavior of the nanocomposite
film containing EG54DMAm46, shown in Figure 4.3b, is different. Again, as the matrix
polymer concentration increases the stiffness begins to decrease, though the decrease is
larger than for DMAm. The major difference is in the percent elongation, as this value
increases as the matrix polymer concentration increases, from about 5% up to about 22%
strain. The composites reach a slightly higher tensile stress with more brittle behavior
which also results in a higher toughness. Additionally, the two arrows on Figure 4.3b point
to a secondary yield point observed at high matrix polymer composition [127].
In both nanocomposites the presence of the matrix polymer results in a decrease in
133
stiffness. In the case of EG54DMAm46, a more ductile polymer, the CNF network fails
(at the initial yield point around 1% strain), and the lower stiffness of the matrix polymer
allows the film to continue to deform. This results in larger elongation, tensile strength,
and toughness. However, with DMAm as the matrix polymer, the high stiffness prevents
interfibrillar motion during deformation. This results in a more brittle failure governed by
CNF. Despite using a higher stiffness matrix polymer, the mechanical performance was
ultimately less stiff than that for the more ductile polymer.
a b
Figure 4.3: Stress-strain curves showing the mechanical behavior of a) nanocom-
posites from CNF and DMAm and b) nanocomposites from CNF and EG54DMAm46
copolymer at increasing matrix polymer content [127]. ? 2016
One avenue for CNF nanocomposites of particular interest is when the matrix polymer
can interact with the network of CNF. One such nanocomposite system was investigated
by Kurihara et. al., in which CNF was combined with poly(acrylamide) (PAM). PAM
was chosen as a matrix polymer due to the presence of C=O bonds and NH2 groups,
which can both form hydrogen bonds with the cellulose hydroxyl groups. Nanocomposites
were formed by mixing aqueous solutions of PAM with aqueous suspensions of CNF and
films were fabricated by solvent casting and drying at 40 ?C for 2 days. The resulting
mechanical properties, shown in Figure 4.4, exhibited a slight increase in stiffness and
134
tensile strength between 0 and 20% PAM, followed by a decrease between 20 and 100%
PAM. The elongation at break shows an unusual trend of decreasing from 0-50% then
increasing with PAM content. Toughness decreases with PAM content. Additionally, CNF
(termed TOCN in the figures) was incorporated into the composites having both ?COONa
and ?COOH groups in order to determine its effect on the mechanical properties [14,128].
Figure 4.4: Tensile testing results of CNF/PAM nanocomposites showing the
Young?s modulus, elongation at break, tensile strength, and work of fracture. The
results of nanocomposites containing both ?COONa and ?COOH groups [14]. ?
2002
The mechanism for this variation in properties is particularly interesting, as it is
expected that PAM can form hydrogen bonds with CNF, and ?COOH groups are expected
to improve the level of hydrogen bonding within neat CNF films. It was found that the
135
carboxylate ion, either H or Na, did not have a significant effect on the resulting mechanical
properties. Kurihara et. al. claim that the mechanism of the increase in stiffness and
strength up to 20% PAM is due to the ability of CNF to form self-aligned and nematically
ordered structures in suspension as a result of electrostatic repulsion. (de Nooy et al.,
1995) The limited amount of PAM is present between these randomly distributed domains,
resulting in improved stiffness and strength. FTIR shows the formation of hydrogen bonds
between the PAM and ?COOH groups of cellulose, but this does not explain the observed
increase in stiffness and strength. (Nishiyama et al., 2000) While PAM presenting a
self-aligned network of CNF can explain this increase, it is likely that the hydrogen bonding
between cellulose hydroxyl groups and PAM plays a significant role as well. A similar
mechanical effect is observed in composites of cellulose and soy protein isolate. If the soy
protein is treated and denatured, the internal hydrogen bonding sites become available and
can interact with cellulose systems, such as a pad of bacterial cellulose. The interaction
through hydrogen bonding produces a similar increase in stiffness with limited inclusion of
soy protein [129,130]. The ability of these materials to hydrogen bond with cellulose set
them apart from other composites that were reported in the literature.
In a review written by Wang et. al. a variety of cellulose based composite systems
are discussed. One example is cellulose and polyvinyl alcohol (PVOH) composites, since
PVOH is a water soluble polymer having free hydroxyl groups. Figure 4.5 shows the tensile
strength and modulus of a collection of cellulose and PVOH composite systems in all forms,
including electrospun fiber mats, hydrogels, films, and fibers. The plot illustrates that the
form of the composite has an inherent effect on the resulting properties, and that the range
of properties of the standard composite films is roughly a tensile strength of between 5-200
MPa, and tensile modulus of between 0.2-10 GPa [131].
136
Figure 4.5: Range of mechanical properties from a variety of cellulose composite
materials, including films, fibers, hydrogels, and mats [131]. ? 2021
4.2.4 Summary of all-cellulose composites
The idea of an all-cellulose composite is driven primarily by its ability to form
hydrogen bonds between cellulose chains. Therefore, a composite of two different allomorphs
or geometries with varying properties could be fabricated but there would be no coupling
or surface modification required in order to generate an interaction between them. Due to
the one-dimensional nature of many cellulosic materials such as MFC, CNF, or CNC, this
can result in directionally reinforced nanocomposites with unique interactions between the
filler and matrix of the nanocomposite.
One method of creating all-cellulose composites is a one-step process achieved through
partial dissolution, in which a solvent is used to partially dissolve a fibrous morphology
of cellulose, then the solvent is removed, leaving a matrix of regenerated cellulose with
embedded cellulose fibers [132?137]. A schematic illustrating this method is shown in
137
Figure 4.6a. The filler can be any fibrous form of cellulose, from large scale cellulose
macrofibers to CNF or CNC. More information about the solvents that can be used to
partially or selectively dissolve the fibrous component can be found in Chapter 3 of this
thesis, which provides background on dissolving cellulose.
a
b
Figure 4.6: Schematics showing two different methods of preparing fiber-based
all-cellulose nanocomposites. a) One-step method involving partial dissolution of
cellulose fibers followed by regeneration, and b) two-step method in which cellulose
fibers are combined with a solution of cellulose which is then regenerated around
them [132].
In work done by Ma et al., for example, CNC, is immersed and heated in ionic
liquid, partially dissolving the nanocrystals, and added to a solution of alkali cellulose
and ionic liquid. The ionic liquid is removed from the mixture by washing with water
and drying, resulting in transparent all-cellulose nanocomposite films. The resulting films
varied by the initial amount of CNC that was added. The relative amounts of CNC that
were dissolved were determined by XRD, identifying relative crystallinities of cellulose I
138
and II crystal structures in the films. The films underwent tensile testing, the results of
which are presented in Figure 4.7. The stiffness increases with an increase in CNC content,
and a clear optimization can be observed at roughly 10%, in which a high tensile strength,
percent elongation, and toughness are all achieved. This tunability between plastic and
ductile versus stiff and brittle behavior illustrates the advantages of reinforced all-cellulose
composites. (Ma et al., 2011) While some tunable mechanical behavior was achieved, there
are disadvantages to this partial dissolution method. It is difficult to determine how much
of each cellulose component is present in each sample. While this can be inferred through
methods like XRD or FTIR, it is difficult to be confident in the relative compositions of
each component. Additionally, the partial dissolution of CNC could weaken the properties
of individual nanoparticles by damaging crystal structures.
Figure 4.7: Tensile testing results showing the effect of CNC composition on
mechanical behavior [134]. ? 2011
Another method for producing all-cellulose composites is a process of solution
infiltration, in which one cellulose component is fully dissolved, then regenerated in the
139
presence of an undissolved fibrous cellulose component, as shown in Figure 4.6b [132].
This can provide more control over the exact concentrations of each component, and is
potentially better for understanding the interactions between components. One example
of this method is by Yang et. al., where TEMPO oxidized CNF were combined with
regenerated cellulose from the alkali-urea solvent system described in Chapter 3. The
results showed an increase in the stiffness and strength of the regenerated cellulose films
upon the introduction of CNF, as shown in Figure 4.8. Remarkably low CNF compositions
were required to achieve improvements in mechanical properties, likely due to the high
length of individual nanofibers. These films also exhibit two different crystal structures,
as the CNF contains cellulose I, but the regenerated matrix cellulose contains cellulose
II, which is typical for cellulose regenerated from solution [138]. However, while solution
infiltration can be useful in identifying interactions within a composite, these reports have
focused primarily on the matrix-rich composition region of all-cellulose composites, in
which a small amount of a fibrous component provides composite reinforcement.
Figure 4.8: Stress-strain curves from tensile testing results of all-cellulose nanocom-
posites made from TOCN and cellulose regenerated from an alkali-urea dissolved
cellulose (AUC) [138]. ? 2015
140
4.2.5 All-cellulose composite- motivation and explanation
The goal of this work is to combine the two cellulose systems discussed in chapters
2 and 3, CNF and MCS, into a single self-reinforced all-cellulose nanocomposite. Both
cellulose systems represent basic fundamental cellulose structures, for CNF it is the most
naturally occurring elementary cellulose structure, and MCS the basic chemical species of
cellulose, polymer chains. Therefore, these two systems should create a situation where
the interactions of each form of cellulose can be understood. Due to the water-resistant
properties of cellulose films cast from MCS, and the nature of cellulose to form hydrogen
bonds with itself, it is expected that the addition of MCS to CNF films will reduce overall
water uptake. Additionally, the system will be examined to determine the mechanical
properties with a potentially wide range of tunability. By investigating a full range of
compositions, this work provides a new understanding of cellulose interaction within
all-cellulose composites.
For each cellulose system there are many variables that can alter the properties and
structures of the resulting cellulose. These have been optimized for the purpose of forming
nanocomposites as discussed in this section.
4.2.5.1 Optimizing CNF for nanocomposite applications
The first step to adapting CNF for nanocomposites with MCS is to replace the water
in the CNF suspension. If MCS is added to water the dissolved cellulose will immediately
precipitate, which would prevent adequate dispersion for nanocomposite production. The
water in the CNF suspension is exchanged for DMSO, which has been shown to allow good
dispersion of the CNF. A rotary evaporator was used to scale up the process. This will
141
allow the suspension to be mixed with the MCS without flocculation and precipitation.
After the TEMPO-oxidation process, cellulose must be mechanically homogenized to
produce CNF. Homogenization in this work is conducted using a high pressure microfluidizer.
The number of times the pulp is run through the microfluidizer has been shown to reduce
the length and variability, resulting in a more monodisperse length distribution, which is
helpful for understanding the resulting mechanical properties. The CNF after 4 passes
through the microfluidizer was shown to have an adequately uniform distribution of length
between 200-400 nm, and diameter about 2 nm as determined from AFM measurements
discussed in Chapter 2. Aspect ratios of roughly 100 were still obtained despite reducing
the length considerably, so the CNF is still capable of significant composite reinforcement.
Once the CNF is dispersed in DMSO the concentration at which the suspension will
form a gel is lowered and shortening the lengths of the CNF allows for greater concentrations
while still allowing for a fluid suspension. Gelation would be detrimental and make the
nanocomposite much more difficult to mix, but increased concentration is important for
reducing the amount of solvent that must be removed, as this is a time- and energy-
intensive process. Through optimization of the solvent exchange and evaporation process,
a suspension as high as 0.11 mol% CNF in DMSO was produced without forming a gel.
4.2.5.2 Optimizing MCS for nanocomposite applications
One factor that must be considered for MCS used to form nanocomposites is the
cellulose concentration. The concentration must be high enough to minimize the volume
of solvent that must be removed, but must remain low enough that the MCS can be mixed
and combined with CNF and allow the mixture to remain fluid so it can be cast to form
films. This concentration was found to be roughly 3 mol% cellulose in solution.
142
The other significant factor is IL concentration. In order to dissolve cellulose it
has been shown that a minimum concentration of IL is required, resulting in the ions
associating with the cellulose and occupying hydrogen bonding sites. With an excess of IL,
combining the MCS with CNF would result in a larger amount of IL interacting with CNF.
This could cause flocculation due to charge interactions between the ?COONa surface
groups and the dissociated ions from EMIMAc. This interaction could also potentially
allow the IL to partially dissolve the CNF, which could compromise mechanical properties
of the resulting nanocomposites.
To address this issue, the MCS was formed with the goal of minimizing IL content.
After mixing the Avicel and DMSO, IL is added in small amounts and stirred vigorously.
Additionally, between each addition of IL, the solution is heated in a 90 ?C oven for
5 minutes to test for dissolution. The MCS was checked for dissolution by eye, since
undissolved Avicel powder is visible. When the solution was fully dissolved, the final
cellulose, IL, and DMSO compositions were 3.2, 5.3, and 91.5% respectively by mole,
resulting in an IL to cellulose molar ratio of roughly 1.7:1. This is quite close to the lowest
soluble ratio achieved in Chapter 3 at about 1.66:1, and should result in limited interaction
between CNF and IL.
4.2.6 All-cellulose nanocomposite fabrication
Combining CNF and MCS to form composite mixtures is accomplished by slowly
adding MCS into a suspension of CNF in DMSO in a centrifuge tube and mixing with
a Vortex stirrer until the mixture flows homogeneously. Both the CNF suspension and
MCS solution are viscous, and the mixture is also quite viscous. After mixing, bubbles are
removed before evaporation by partially submerging the tube in a sonication bath. The
143
mixture will form a loose gel if it remains undisturbed for more than a few seconds, similar
to the gelation behavior of CNF suspensions but occurring much more quickly. Sonication
must be done while constantly rotating or agitating the tube so the mixture remains fluid.
Once the bubbles are removed, the mixture is immediately poured into a casting dish.
The casting dish used was the same used for creating dense cellulose and EMIMAc
gels in Chapter 3, with Teflon sides secured onto a glass base. This design allows the film
to evaporate evenly from the top-down without the film clinging to the side of the dish as
it dries. This results in relatively even film thickness from the center to the edge of the
film.
In fabricating nanocomposites, the drying method must also be considered. Filtering
the composite mixture into a gel then hot-pressing would result in slightly improved
mechanical performance and take less time to remove DMSO, but the MCS may fall
through the micropore filter. In testing, it was estimated that some of the solution
remained present in the condensed gel after filtering, but it became far too difficult to
determine how much of each type of cellulose was present in the final film. Solvent casting
was determined to be the best option for nanocomposite film fabrication. The DMSO
was removed via evaporation in a vacuum oven. The temperature must be low enough
that the cellulose and IL do not degrade over time since evaporation will take 3-4 days.
The evaporation termperature used was 50 ?C. The vacuum must be low enough that
the DMSO does not boil and the solvent is evaporated evenly, resulting in a consistent
thickness across the films. After testing a vacuum of approximately 11 inHg was used.
The final step in producing the nanocomposite films is removing IL. The same
protocol will be followed that was employed to produce dense gels of cellulose and IL
in chapter 3, which formed amorphous cellulose. The amorphous cellulose has shown
144
significant water-resistance and should enable further interaction with the CNF compared
to a more ordered crystalline structure.
4.3 Experimental methods
4.3.1 Cellulose nanofiber preparation
Cellulose pulp from southern yellow pine wood was used for this study. The moisture
content of the pulp was about 5%. Cellulose nanofiber production was carried out according
to the TEMPO oxidation procedure described in Chapter 2, followed by thorough cleaning
and resulting in a suspension of TEMPO-oxidized cellulose pulp in water at about 1%
by weight before homogenization. Mechanical homogenization was conducted using a
NanoDeBEE microfluidizer from BEE International. Homogenization was carried out
at roughly 200 kPa (30 kpsi) and the material was passed through the machine four
times. CNF suspensions were then purified by centrifugation to remove any remaining
unhomogenized cellulose material. CNF then underwent a solvent exchange to replace the
water in the suspension with DMSO. The CNF suspension in water with concentration of
about 0.6% was added to an equal volume of DMSO and stirred gently until homogeneous.
Water was removed using a rotary evaporator, in which the suspension kept in a water
bath at about 60 ?C, and vacuum was increased until just before the suspension began to
boil. Once the suspension had been evaporated by half it was removed, then the process
was repeated two more times. The resulting suspension of CNF in DMSO was stored in a
desiccated environment to prevent exposure to moisture, and concentration was confirmed
to be 0.11 mol% (0.6 wt%) CNF by evaporation.
145
4.3.2 Molecular cellulose solution fabrication
Avicel cellulose powder was combined with DMSO and briefly stirred. EMIMAc was
added in small amounts, followed by vigorous stirring, and heating at 90 ?C for about 5
minutes. EMIMAc was added, followed by 5 minutes of heating and stirring, until the
solution was fully transparent. The solution was confirmed to be dissolved if when held up
to the light no Avicel powder particles are observed. The final compositions of cellulose,
IL, and DMSO were 6.1, 10.6, and 86.3% by weight, respectively.
4.3.3 All-cellulose composite fabrication
CNF suspension in DMSO was poured into a large centrifuge tube. MCS was added
slowly into the CNF under constant stirring, then stirred vigorously for 60 seconds. Bubbles
were removed from the composite mixture by briefly sonicating while maintaining flow in
the liquid mixture, then the mixture was deposited into a casting dish with Teflon sides
and a glass base.
The composite mixtures were dried in a vacuum oven at 50 ?C and a vacuum of 37
kPa. Once the DMSO appeared to be fully evaporated, vacuum was increased to 30 inHg
and the films were left to dry for 12 hours longer to ensure full removal of DMSO.
IL was removed from the nanocomposite films by soaking in acetone, followed by
ethanol. Films were first soaked in acetone for 1 day, then dried for 2 hours in a vacuum
oven, then soaked again in a bath of fresh acetone for 1 more day. Films were then soaked
in ethanol following the same protocol, and soaking continued until IL was fully removed.
Total mass of the film was monitored before and after each stage of soaking to track the
removal of IL, and stopped when no change in mass was observed.
146
4.3.4 Atomic force microscopy
Suspensions of CNF in DMSO and MCS were both diluted to about 0.2 mol%. The
two were mixed at a 50/50 ratio and stirred thoroughly. The resulting mixture was cast
onto a freshly cleaned silicon wafer, spincast, then dried in the oven to evaporate DMSO.
The film was then soaked in acetone and ethanol in stages to fully remove IL, then dried
in the oven. AFM was conducted in tapping mode on the thin film using a Multimode
AFM from Nanoscope.
4.3.5 Transmission and Haze
Light transmission was measured using the Lamda 1050 device from Perkin Elmer
with a 150 mm integrating sphere. Baseline transmission measurements between wave-
lengths of 400 and 800 nm at a resolution of 5 nm per minute. Haze was measured by
measuring reflective transmission and dividing by the baseline transmission for each sample.
4.3.6 Tensile testing
Tensile testing measurements were conducted using a DMA Q800 from TA Instru-
ments. While a DMA is typically used for viscoelasticity measurements, in this work it was
used for standard tensile testing. Force ramp tests were run for each sample at a rate of 1
N/min, with at least 3 successful tensile measurements taken for each sample. Samples for
tensile testing were prepared by cutting a rectangular shaped piece from the film roughly
3-4 mm wide and the length of roughly 20-30 mm. To ensure successful tests, freshly
sharpened hole punches were used to cut circular portions from the sides of the sample
center, creating an approximate ?I? shape so failure occurred at the center. If the cutter
147
was not sharp enough there could be small microcracks that could form at the cutting site,
which could result in premature failure of the test. Thickness and width were measured
before and after each measurement to ensure accuracy and account for any changes in
dimension due to high strain.
4.3.7 Scanning electron microscopy
Images were taken using a Hitachi SU-70 FEG SEM from the University of Maryland
Advanced Imaging and Microscopy Lab with an acceleration voltage of 0.7 kV and a
working distance of about 15 mm. The films from select tensile testing measurements were
taken for fracture surface imaging with SEM. The films were loaded onto a vertical sample
holder, to observe the cross section of each fracture surface and coated with a thin layer of
carbon.
4.3.8 Water uptake
Water uptake was measured by cutting a circle of each film 1 cm in diameter and
weighing using a microbalance. For each measurement the film was placed in a shallow
water dish and submerged for a set amount of time, after which the film was retrieved,
briefly patted with a Kim wipe, and weighed again.
For some films, after about 10 minutes the films begin to disintegrate, and the
measurement could not be continued.
148
4.4 Results and discussion
4.4.1 Atomic force microscopy
A proposed schematic illustrating the potential morphology of the all-cellulose
nanocomposite films, featuring a network of CNF embedded in a matrix of regenerated
cellulose from MCS is shown in Figure 4.9a. Diluted CNF suspensions in DMSO and MCS
solutions were combined and spin cast in order to determine if the cellulose components were
aggregating upon mixing. Figure 4.9b shows an AFM image of the spin cast 50% CNF/50%
molecular cellulose (Optimizing MCS for nanocomposite applications) nanocomposite
thin film. The image shows the MCS regenerated as small particles of cellulose, covering
the silicon wafer. Individualized CNF is shown distributed throughout the layer of MC,
indicating that neither cellulose component aggregated upon mixing the nanocomposite.
Figure 4.9: a) Proposed schematic of all cellulose composite. b) Spin-coated thin
film of all-cellulose nanocomposite mixture.
149
4.4.2 Hazemeter measurements
Baseline and reflective transmission measurements were taken for each nanocomposite
film and are shown in Figure 4.10. Baseline transmission (transmission 1) measurements
are consistent for each film between about 85 and 90%, with no discernable trend as a
function of MC content. This agrees with transmission values from the literature for pure
CNF films [52,64,67] and pure amorphous MC cellulose films [110] both at around 90%.
Reflective transmission (transmission 2) measurements show an overall increasing trend
with MC content. Where pure CNF films have values of about 8%, some nanocomposites
and the pure MC films have reflective transmissions as high as 30 or 35%. Each reflective
transmission measurement also follows the trend of increasing as wavelength decreases,
with this behavior becoming more pronounced with increasing MC content as well.
Figure 4.10: Baseline and reflective transmission (transmission 1 and 2) measure-
ments of each all-cellulose nanocomposite film.
Haze is a property that quantifies how strongly light is scattered through a material.
This is calculated by dividing the reflected transmission by the baseline transmission. The
haze measurement for each composite sample is shown in Figure 4.11 Because the baseline
150
transmission values are relatively constant with respect to wavelength and are close to
100%, the haze values are very similar to the reflective transmission values.
Figure 4.11: Haze percent as a function of wavelength for each composite film
Figure 4.12 shows this haze, averaged across all wavelengths, and plotted as a function
of MC content. For pure CNF films, the haze is about 8%. With increasing MC content
the haze increases quickly before leveling out between 24 and 45% MC and approaching a
value of about 30%. While there is some variability, there is a clear trend of increasing
haze with increasing MC content that levels out at higher MC concentrations.
The haze of pure CNF is fairly low, around 8%, which is in agreement with haze
values for CNF found in the literature [68,139]. This minimal haze is derived from small
amounts of air trapped in the CNF dispersion during the drying process, producing voids
in the film that scatter light. With careful drying, or filtering and pressing, haze can be
reduced to around 4%. As the MC composition is increased, the amount of haze in the film
quickly increases, but the source of this haze is unknown. Crystallinity in pure MC films is
minimal, so crystallization is not a likely source of haze. One possibility is that the haze in
MC is also caused by voids, similar to CNF. In Chapter 3, the ethanol soaking process for
151
the cellulose and EMIMAc gels was discussed. The SAXS results showed that immediately
after ethanol soaking begins, the intensity increases significantly. This increase in scattering
could also indicate the presence of voids. These voids could be forming upon the removal of
EMIMAc during the second stage of washing using ethanol, but the surrounding cellulose
in the system is too stiff to relax, resulting in voids.
Figure 4.12: Optical haze plotted as a function of MC content. Haze was calculated
by averaging haze values across all wavelengths.
4.4.3 Tensile testing
The first method employed to understand the interaction between MC and CNF in
the nanocomposites was tensile testing. The tensile behavior of the pure CNF films is the
red curve shown in Figure 4.13. In composite samples containing low concentrations of
MC, the samples fracture more quickly but have a steeper initial slope, resulting in a stiffer
but more brittle film with a shorter plastic deformation region. As MC concentration
continues to increase, the slope decreases and elongation increases, where at roughly 70%
152
MC the film appears to have similar mechanical behavior to the neat CNF film. As MC
content further increases the initial slope continues to decrease, and elongation increases
substantially. At 100% MC the film deforms very plastically, having the lowest slope and
largest elongation of all measured samples.
Figure 4.13: Stress-strain curves of select all-cellulose nanocomposite samples
illustrating the different mechanical behaviors of all-cellulose nanocomposites with
varying MC content. A sample line of 0.2% strain offset is drawn onto the 100% MC
data to illustrate how yield strength was calculated, and a line is drawn establishing
the elongation at break at the end of the curve.
The modulus, percent elongation, yield strength, and toughness were calculated
based on the mechanical testing. Yield strength was calculated based on the 0.2% offset
from the linear elastic region, and toughness was calculated from the area under the curve.
The modulus of pure CNF is 8.5 GPa, which agrees with values reported in the literature
which range between about 6-12 GPa, however the yield strength is significantly lower that
the reported values [52, 140]. This is likely due to the CNF in this work having shorter
lengths. Longer lengths allow for greater entanglement and breaking and reforming of
153
hydrogen bonds during plastic deformation [65]. Additionally, the nanofibers in the CNF
film could be damaged from the excessive mechanical treatment applied in this work. This
could also explain the lower yield strength. This could also explain the percent elongation
being less than reported values.
The mechanical behavior of the pure MC film exhibits largely plastic deformation.
Modulus and yield strength values are much lower than those of CNF at around 2 GPa and
40 MPa respectively, while the elongation at break is high, around 8%, and the toughness
is also high. Table 4.1 shows a comparison of mechanical properties between cellulose
obtained from MCS, versus cellulose obtained from solutions in pure ionic liquids, EMIMAc
and 1-allyl-3-methylimidazolium chloride (AMIMCl). The cellulose obtained from pure
ionic liquid solutions contains a significant amount of cellulose II crystallinity [102], while
MC is almost fully amorphous. This explains why MCS has a lower stiffness and a larger
elongation, as the crystallinity makes the cellulose from EMIMAc and AMIMCl less ductile.
Young?s modulus Tensile strength Elongation at
Solvent
(GPa) (MPa) break (%)
EMIMAc 3.1-5.5 35-48 0.5-2.6
AMIMCl 5.1-7.1 78-120 2.1-5.6
MCS 2.2 54.4 7.5
Table 4.1: Mechanical properties displayed by cellulose samples from different
solvents: two ionic liquids EMIMAc and AMIMCl [102], and the MCS described in
this work.
It is shown in Figures 4.14a and 4.14b that the Young?s modulus and yield strength
follow a similar trend as both increase with MC up to about 20% MC, then begin to
decrease almost linearly with MC, until 100% MC. The Young?s modulus follows this trend,
and while the yield strength data shows considerable variability, it also appears to follow
this trend. An opposite trend can be observed in Figures 4.14c and 4.14d, as the percent
154
elongation and toughness both decrease initially with limited MC content up to about
20%, then increase almost linearly up to 100% MC. While there is some variability in the
samples as evidenced by the error bars, the overall trend for each property is clear.
a b
c d
Figure 4.14: a) Young?s modulus, b) yield strength (measured from 0.2% strain
offset), c) percent elongation at failure, and d) toughness plotted against MC content.
The mechanism driving the mechanical behavior of CNF is hydrogen bonding
between entangled nanofibers, combined with the intrinsic stiffness of the semicrystalline
fibers. Once the film begins to yield, hydrogen bonds between nanofibers break and
reform, allowing for the film to maintain stiffness throughout plastic deformation. This
extends the plastic deformation region and allows for higher toughness overall [65]. In the
nanocomposite samples, despite the low stiffness of neat MC films, there is an increase in
stiffness observed after adding a limited amount to nanofiber composites. This results in a
larger yield strength as well, but also limits the ability of the film to plastically deform, as
155
the elongation and toughness are decreased.
Similar behavior was observed in a CNF-based nanocomposite with PAM as described
in Figure 4.4, or in composites of cellulose with soy protein isolate [129, 130]. These
composite films also display an increased stiffness and strength with a limited inclusion of
the matrix polymer, in this case roughly 25% PAM. The key similarity between these two
systems is the inclusion of a matrix polymer capable of forming hydrogen bonds with CNF,
allowing for a self-reinforced nanocomposite. While this indicates the presence of favorable
hydrogen bonding interactions, the exact mechanism for how this interaction results in the
given mechanical behavior is unclear. A possible mechanism for this behavior is the MC is
acting as a reinforcement, forming hydrogen bonds with surrounding nanofibers. This can
reinforce contact points in the CNF network and act as a reinforcing agent and improves
mechanical properties. It also appears to limit the elongation at break and toughness. This
indicates while the network has been strengthened, the mechanism of CNF breaking and
reforming hydrogen bonds, thereby extending the plastic deformation region, no longer
applies.
By testing a full range of composition from 0 to 100% MC, a range of tunable
mechanical properties were achieved with the all cellulose nanocomposites, having a
modulus between 2.2-11.9 GPa and yield strength between 55-110 MPa. This range
in tunability is consistent with ranges of other polymer composites based on cellulose
nanomaterials, such as with polyvinyl alcohol or polypropylene [131], as well as ranges of
other isotropic and nonwoven ACCs [132,133].
156
4.4.4 Scanning electron microscopy
SEM images for 3 cellulose films are shown in Figure 4.15, with the cross-sections of
the films fracture surfaces after tensile testing in which the vertical direction is oriented
along the 2-dimensional plane of the film, and the horizontal direction is oriented with the
film thickness. Figures 4.15a and 4.15b show a CNF film with no MC included, Figures
4.15c and 4.15d show a nanocomposite film with 15% MC, and Figures 4.15e and 4.15f
show a 100% MC film. In Figure 4.15a, a structure oriented in the plane of the film
is visible. The higher-magnification image in Figure 4.15b shows dark areas amidst the
oriented structure, and a large gap between layers in which bridging nanofibers can be
observed. In Figure 4.15c and 4.15d, a similar structure oriented in the plane of the 15%
MC film is present, but with a lower contrast difference than in the 0% film, which could
indicate lower roughness in the fracture surface. While there are similar dark areas present
in Figure 4.15d as are visible in 4.15b, they appear smaller, and no larger gaps are visible.
Figure 4.15e shows a fracture surface with round structures with protruding edges in the
100% MC film. In the inset image in Figure 4.15f, tears within the crater structures are
visible.
SEM images shown in Figure 4.15 can provide insight into the mechanical behavior
of these films. Figure 4.15c shows crater-like structures, which is evidence of a ductile
fracture after significant plastic deformation. Figures 4.15a and 4.15c show a semi-ordered
or layered structure characteristic of CNF films [141], with the layered structure in 4.15a
appearing rougher and more protruded from the surface as compared to 4.15c based on
greater changes in contrast along the surface. The 15% MC film shown in 4.15c showed
more brittle behavior with a higher stiffness and lower elongation, which would result in
157
Figure 4.15: SEM images of 0% (a, b), 15% (c, d), and 100% (e, f) MC composite
films shown at two different magnifications, in which the vertical direction in the
image is oriented along plane of the film and the horizontal direction in the image is
oriented along film thickness.
a flatter fracture surface. The 0% MC film has a greater elongation period compared to
that of the 15% sample, as well as a larger region of plastic deformation. In Figure 4.15a,
the fracture surface of the 0% MC film is qualitatively rougher than the 15% MC film
shown in Figure 4.15a, as the film strains plastically and layers of CNF are pulled out of
the surface plane of the films cross section. Therefore, the SEM results are consistent with
the measured mechanical properties.
4.4.5 Water uptake
To understand the interaction of MC and CNF in the nanocomposites water uptake
measurements were conducted and the results are shown in Figure 4.16a. The films having
zero or low MC content absorb water rapidly, and the curves do not appear to plateau
or decrease in absorption rate before the films disintegrate. At about 8% MC, the film
disintegrates but a plateau in water absorption can be observed, indicating a change in
overall water uptake behavior. Films containing 12% MC or greater maintain their integrity
158
with exposure to water and do not disintegrate. Additionally, between 18 and 24% MC
there is a steep decline in water uptake. The overall water uptake is quantified by recording
the highest amount of water uptake for each sample and is shown in Figure 4.16b. This
plot shows that a steep drop in water uptake occurs between 18 and 24% MC. All films
containing greater than 24% MC uptake significantly less water, plateauing at about 30%
greater than the initial film mass at all MC contents greater than 45%.
a b
Figure 4.16: a) Plot showing the level of water uptake throughout the soaking
process of each nanocomposite cellulose film. Water uptake is presented as the
mass gain of the film relative to the initial mass of the film. b) Maximum water
uptake recorded for each film after extensive soaking time in water. Values were only
recorded for films that showed a plateau in water uptake, films lower than 8% MC
disintegrated.
Overall, the change in uptake behavior observed between 18 and 24% MC is consistent
with the mechanical property data, in which a maximum in stiffness and yield strength and
a minimum in elongation and toughness is observed at similar MC content. The existence
of favorable hydrogen bonding interaction on the mechanical behavior is further supported
by the water uptake measurements. The most significant drop in water uptake is observed
at a similar MC composition where the stiffness and strength reach a maximum. Neat CNF
films will rapidly absorb water due to surface charge and the dissociation of the sodium
carboxylate groups, which causes electrostatic repulsion between nanofibers and leads to
159
disintegration of the films. If MC in the nanocomposites is occupying the hydrogen bonding
sites on the nanofibers this could prevent interaction between the nanofibers and water
and result in significantly decreased water absorption. Since the mechanical and water
uptake properties see a change in behavior at the same compositions, roughly between 18
and 24% MC, it is reasonable to conclude that the available hydrogen bonding sites on
CNF that would enhance water uptake have been saturated by interaction with MC. This
could indicate that between 18 and 24% MC has saturated the CNF network, resulting in a
significant transition and reduction in uptake behavior. This explains why beyond 24% MC
the mechanical properties begin to decrease and plasticity increases, as the film is following
a rule of mixtures behavior between the more brittle reinforced CNF films containing
18-24% MC, and the more plastic MC films. This also explains why water uptake begins
to plateau and stabilize at the water uptake behavior of pure MC at compositions above
24% MC.
4.4.6 Composite reinforcement
While the ACCs reported in this work exhibit a tunable range of mechanical properties
comparable with others found in the literature, the reinforcement behavior in the fiber-
rich composition region has not been thoroughly discussed in the context of all-cellulose
composites. Other all-cellulose composites prepared via partial dissolution of fibrous
cellulose exhibit an optimizable region where mechanical properties are enhanced with
limited dissolution time, but further dissolution decreases crystallinity sufficiently enough
to reduce mechanical properties [142?147]. Due to the preparation method of partial
dissolution, it is difficult to determine the mechanism that would cause this reinforcement,
because precise compositions of each cellulose component are not known.
160
The nanocomposite films developed in this work showed an increase in stiffness and
strength, but overall became embrittled by the introduction of between 18 and 24% MC
into CNF, with stiffnesses increasing from about 8 GPa in pure CNF films to about 12
GPa. Beyond 24% MC the films become more plasticized, and a stiffness comparable to
pure CNF films maintained up to roughly 55% MC. Additionally, in films containing higher
MC content, an overall increase in toughness is observed. Water uptake measurements
indicated that introducing MC into CNF results in a dramatic reduction in water uptake,
decreasing from a water uptake ratio of 18 at 8% MC, to a ratio of 4 at 24% MC. Beyond
this composition at higher MC content, the uptake ratio decreases to about 0.3, similar to
the behavior of pure MC.
The water uptake and mechanical property behavior indicate favorable hydrogen
bonding between CNF and MC within the nanocomposites. The hydrogen bonding sites
on the CNF surface interact with MC, resulting in a maximum stiffness and strength
between 18 and 24% MC content which is larger than that of pure CNF. This change is
coupled with a significant reduction in water uptake, allowing the films to maintain their
mechanical integrity despite long exposure times in water.
The amorphous nature of MC makes it able to interact more freely with the hydrogen
bonding sites on the surface of CNF. The MC has a crystallinity index as low as 2% as
discussed in Chapter 3. High crystallinity results in ordered molecules that are highly
associated forming particular structures, but amorphous systems have no ordering, are
less densely packed, and allow for a greater likelihood of association with an adjacent
composite material, such as CNF [148]. Additionally, previous reports have demonstrated
that amorphous cellulose exhibits excellent stability in water, which explains the reduction
in water uptake in the reported all-cellulose composites [39,149,150].
161
Since the mechanical and water uptake properties undergo a significant change in
behavior at the same compositions, roughly between 18 and 24% MC, it is reasonable
to conclude that the available hydrogen bonding sites on CNF, that would otherwise
absorb water, have been saturated by MC. This explains why, at compositions greater
than 24% MC, the mechanical properties begin to decrease and ductility increases, as it
essentially follows a rule of mixtures behavior between the more brittle reinforced CNF
films (containing 18-24% MC) and the more ductile 100% MC film. This also explains
why water uptake begins to plateau and stabilize at the water uptake behavior of pure
MC at compositions above 35% MC. At 45% MC, the absorption ratio is below 1, with a
continued decrease with increasing MC composition up to 100%.
4.5 Conclusions
All-cellulose nanocomposites were fabricated from molecular cellulose solutions of
EMIMAc and DMSO, and DMSO CNF suspensions. The solvent exchange process from
water to DMSO allowed the CNF to be dispersed into the MCS without aggregation,
resulting in transparent nanocomposite films after drying and washing to remove DMSO
and EMIMAc solvents.
This work has shown that creating an all-cellulose nanocomposite film from amor-
phous MC and CNF nanofibers suspended in DMSO results in transparent polymer films
with a robust range of tunable mechanical properties. Additionally, in nanocomposites
containing >18% MC, the water uptake was reduced by more than an order of magnitude,
allowing the films to maintain mechanical integrity despite prolonged exposure to water
that results in disintegration of pure CNF films. This water resistance and versatility in
mechanical properties has improved the viability of CNF as a structural material. The
162
interaction of amorphous MC and CNF has also been investigated, revealing a favorable
hydrogen bonding interaction not observed in other all-cellulose nanocomposite systems.
163
Chapter 5: Summary and outlook
5.1 Conclusions
The objective of this dissertation was to investigate the processing and structures
of two fundamental cellulose systems- elementary cellulose nanofibers (CNF) produced
from TEMPO oxidation and mechanical homogenization, and molecular cellulose solutions
(MCS) dissolved using a binary solvent system of ionic liquid and polar aprotic cosolvent-
and combine these systems to develop a self-reinforcing all-cellulose nanocomposite through
which we could understand the interaction between the cellulose components. Cellulose
is a promising sustainable alternative to conventional petroleum-derived polymers, and
deepening the understanding of new cellulose processing and structures provides additional
tools for integrating sustainable polymers into plastics production in the future.
CNF were produced through TEMPO-oxidation and mechanical homogenization in
order to study the effect of the process conditions on the fiber structure. Optical microscopy
and SANS studies revealed how the TEMPO oxidation process affects the macro- and
micro-scale cellulose structures. Optical microscopy of fiber aliquots showed cellulose
macrofibers swelled with increased oxidation time. Fiber diameters were measured and the
resulting histograms were fit to a log normal distribution, showing that the mean diameter
increased from 36 ?m at 5 minutes to 89 ?m at 120 minutes during the reaction. SANS was
conducted on the fiber aliquots revealing an evolution in the mid-Q data. The data were
164
fit to a correlation length model, resulting in a correlation length at intermediate Q that
increased with oxidation time from 19 A? to 30 A?. The increase in correlation length was
ascribed to the increase in distance between elementary CNF within the overall cellulose
hierarchical structure. Plotting the macrofiber diameter against correlation length showed
a linear relationship, indicating that these behaviors were linked. This discovery provides
insight into the evolution of cellulose throughout this reaction, which will be an aid in
future application of this process. Additionally, understanding this mechanism can be
useful for identifying additional cellulose pretreatments for efficient nanofiber production
moving forward.
Fully oxidized cellulose samples were subject to varying degrees of mechanical
homogenization, resulting in separated CNF suspensions. The samples were diluted and
imaged using AFM. Lengths and diameters of the resulting CNF were measured and the
resulting histograms were fit to a log normal distribution. Mean diameters were shown to
be constant with increasing mechanical treatment and have a narrow distribution, with
an average nanofiber diameter of about 2.1 nm. Lengths were shown to decrease with
increased mechanical treatment, from 428 ? 160 nm with 1 pass to 311 ? 100 nm. CNF
fiber produced from 120 minutes of oxidation and with 4 passes through the microfluidizer
were used to produce all-cellulose nanocomposites in chapter 4.
MCS were produced using microcrystalline cellulose, dimethylsulfoxide (DMSO),
and EMIMAc. Compositions of all three components, as well as varying added water
content, were varied and ternary phase diagrams were assembled. Temperature variation
experiments using SANS were employed to study the phase behavior of MCS, revealing
a minimum ratio of EMIMAc to cellulose of approximately 1.75:1 that is required for
dissolution. The lowest reported ratio presented in the literature is about 3:1 EMIMAc to
165
cellulose, indicating that improved kinetics provided by DMSO in the solution allows for
greater interaction between cellulose and EMIMAc, and reduces the EMIMAc needed for
dissolution. It was also discovered that the addition of water to MCS does not affect the
solubility of cellulose, possibly due to the interactions between water and DMSO when
building the solution.
The process for casting cellulose from MCS was investigated. DMSO was evaporated
using a vacuum oven, forming dense gels of cellulose and EMIMAc at concentrations as high
as 30% by volume. Solvent washing with acetone and ethanol removed EMIMAc from the
gel, and the evolution of the gel structure was observed using FTIR, SAXS, and WAXS. In
the dense cellulose and EMIMAc gels, a unique multiscale ordering structure was discovered.
This ordering was determined to be the result of a unique joint cellulose/EMIMAc co-
ordering structure measured using WAXS, that gives rise to periodic ordering between
crystallites that was measured using SAXS. The behavior of this ordering was dependent
on the washing liquid employed to remove EMIMAc from the gel- acetone maintained the
short-range ordering as EMIMAc was removed resulting in a change in periodic distance,
and ethanol fully eliminated the ordering. Plotting periodic distance with cellulose volume
fraction resulted in a -12 power law dependence, indicating the dense ordered regions were
1-dimensional. While the crystalline peaks were eliminated and the resulting cellulose
films were largely amorphous, density fluctuations associated with the periodic distance
remained in the cellulose structure that require further characterization.
A model was proposed to summarize the multi-scale ordering structure. The local
ordering corresponds to a peak in WAXS showing the co-ordering structure between
EMIMAc and cellulose, the dominant peak is associated with a distance of 1.1 nm. This
local ordering results in dense regions within the gel, causing fluctuations associated with
166
the period distance measured in SAXS, decreasing from 30 to 19 nm as EMIMAc is removed
during washing. The stability of the ordering in the gels was also investigated by combining
SAXS/WAXS with rheology at variable temperature, revealing that with a sufficiently
large ratio of EMIMAc to cellulose, 2.7:1, the ordering would be quickly eliminated upon
heating to 100 ?C and reform after long cooling times, whereas with ratio close to the
solubility limit, 1.75:1, the gel remains solid upon heating.
These two cellulose systems were combined to fabricate an all-cellulose nanocomposite
comprised of CNF and amorphous molecular cellulose (MC) cast from MCS. A previously
reported process of exchanging water in CNF for DMSO was successfully scaled up to
a large volume and completed without further treatment to the surface groups present
on CNF. The resulting CNF in DMSO suspensions were mixed with MCS, DMSO was
evaporated in a vacuum oven, and EMIMAc was thoroughly washed to form laminate
all-cellulose nanocomposite films. The films underwent tensile testing, water uptake testing,
and their structures were investigated with transmission and haze measurement and SEM.
Tensile testing results showed self-reinforcing behavior in which the addition of small
amounts (<24%) of lower-modulus MC into CNF resulted in an increase in strength and
stiffness, but resulted in a more brittle film. With >24% MC the films exhibited tunable
mechanical behavior with decreased stiffness and strength and increased toughness and
elasticity. The fracture surfaces of nanocomposite films were imaged using SEM. Pure
CNF showed a fracture surface in which layers of CNF protruded in and out of the surface
as a result of extended plastic deformation. In the 15% MC nanocomposite film while
the same layered CNF structures are visible, the surface appears smoother, indicating
more brittle fracture. In the MC sample, round structures with edges protruding from the
surface indicate a clearly ductile fracture with no visible nanostructures in the images. The
167
introduction of MC dramatically reduced the water uptake of CNF by nearly an order of
magnitude upon as little as 15% MC present in the film. The prevention of significant water
uptake indicates that the C6 carboxylate groups, which are responsible for the detrimental
water uptake of pure CNF, are occupied or shielded by the MC through hydrogen bonding
interactions. The relative compositions between 15-24% MC show the most dramatic
reduction in water uptake, and increase in mechanical properties, implying a link between
these behaviors. This self-reinforcing behavior aids in the understanding of the interaction
between CNF and MC within these nanocomposites. Additionally, the reduction in water
uptake for CNF provides a pathway to its viability as a structural polymer.
In summary, the main contributions of this dissertation are:
1. Provided insight into the behavior of the cellulose hierarchical network during
TEMPO oxidation, and the mechanism through which CNF is produced.
2. Investigated the phase behavior of a new cellulose solution system, revealing a distinct
solubility limit for cellulose in EMIMAc, and investigating the effect of water and
temperature on the solubility.
3. Discovered a unique multiscale ordering structure in dense gels of cellulose and
EMIMAc cast from MCS and outlined a model to explain the behavior of the
ordering in the system. The effect of this system on the resulting structure of
amorphous cellulose.
4. Fabricated all-cellulose nanocomposites from MCS and CNF, which displayed self-
reinforcing behavior and dramatically reduced the water uptake of CNF. These
results indicate constructive hydrogen bonding interaction between CNF and MC
and increase the viability of CNF due to improved water resistance and tunable
168
range of mechanical properties.
5.2 Future work
This thesis probed the structure and behavior of elementary cellulose systems, but
there are several possible avenues through which this work can be continued.
5.2.1 The effect of water on gelation and viscoelasticity of MCS
In Chapter 3, it was determined that MCS could allow water to be incorporated into
the system while still dissolving cellulose, despite previous work showing that water will
inhibit the solvation capabilities of EMIMAc. While SANS was able to show that MCS
with significant water compositions were still fully dissolved, the mechanical behavior of
these systems was changed dramatically. The introduction of water resulted in significant
viscosity increase and eventual gelation of the solution. Understanding the effect that
water has on the mechanical behavior of this solution is beneficial for any future processing
or extrusion capabilities of this solution for producing cellulose.
The proposed experiment is to produce a series of MCS in which the cellulose to
EMIMAc ratio is fixed, but the water content is gradually increased. These solutions will
be studied with rheology to establish a relationship between water content and viscosity
of the liquid solution or modulus of the gelled solution. The effect of temperature on the
resulting mechanical behavior will also be tested in order to establish a defined point of
gelation based on temperature and composition.
In addition, the presence of water in MCS may effect the regeneration process,
resulting in different cellulose structures after removing the EMIMAc. This structure can
169
be characterized by SAXS and WAXS to observe any changes in the multiscale ordering
structure outlined in this work.
5.2.2 Effect of coagulation process on molecular cellulose structure
The ordering in these gels was studied and a model was presented to explain the
behavior as a function of composition and temperature. However, further characterization is
necessary for a more complete understanding of the system. More thorough characterization
of the composition of the gels through elemental analysis can provide a more accurate
measure of the cellulose and EMIMAc composition throughout soaking. Identifying the
relative distributions of cellulose and EMIMAc between the amorphous and ordered bundled
regions can provide more information for potentially mapping out the unit cell structure of
the cellulose/EMIMAc co-crystal structure.
Microscopy is needed to understand the morphology of the described multi-scale
ordering structure. Additionally, conducting WAXS measurements of the gels under
deformation can provide information about the co-crystal structure based on anisotropic
scattering that may result from the deformation. Finally, after the acetone washing stage,
reswelling the gel with EMIMAc can be tested. If the local ordering is maintained, the
interbundle spacing would likely increase again, providing additional compositions with
which the -12 scaling law can be explored.
More generally, if a disordered state can be maintained throughout the EMIMAc
removal process, the resulting cellulose film will likely be more amorphous and glassier
compared to the MC that was produced in this work, which could result in more elasticity
and an overall softer cellulose film. This could be accomplished by using different nonsol-
vents to remove EMIMAc such as water or isopropyl alcohol, or carefully manipulating the
170
washing process to achieve the desired coagulation results. Alternatively, using different
nonsolvents could also result in crystallization of the cellulose during regeneration, which
would result in different mechanical and water uptake behavior. The resulting cellulose
structures can be characterized using SAXS and WAXS.
5.2.3 All-cellulose nanocomposites with aligned cellulose nanocrystals
Establishing anisotropy or alignment of the rod-like filler in nanocomposite films can
result in exceptional mechanical properties. One avenue for alignment of rod-like particles
is through shearing, which can be accomplished through extrusion of fibers or films before
the drying process. The dramatic change in cross-sectional flow during extrusion produces
shear forces in the liquid that align rod-like structures, then the liquid can be quickly
dried or cast in a coagulant bath to arrest the aligned structure. In this work, the CNF
concentration was so low, that large volumes of the liquid composite mixture were required
for a sufficient amount of cellulose to be cast and shear-alignment through extrusion was
not a viable production method.
CNCs are another class of cellulose nanomaterial introduced in Chapter 2. While
they have aspect ratios lower than CNF, they are rodlike structures and have been studied
at length in the literature as a viable filler for nanocomposite reinforcement.
CNCs can be maintained in suspensions of much higher concentration than CNF.
CNC can also be successfully distributed into DMSO as well, making it compatible with
MCS for nanocomposite applications. Initially, alignment of CNC in DMSO, and in
combination with MCS will be tested. This alignment can be induced through increased
shear, or using a weak magnetic field [151]. The alignment process can be studied using
SAXS or SANS.
171
Nanocomposites will be produced at a range of CNC to MC ratios, and at each
composition films will be produced with differing degrees of alignment of CNCs. Tensile
testing will be conducted to understand both the effect of CNC reinforcement on MC films
and the effect of CNC alignment, and water uptake measurements will also be conducted.
Optical properties will also be investigated, as shear-alignment been shown to produce
birefringence in cast CNC films.
172
Bibliography
[1] Sarute Ummartyotin and Hathaikarn Manuspiya. A critical review on cellulose:
From fundamental to an approach on sensor technology. Renewable and
Sustainable Energy Reviews, 41:402?412, 2015.
[2] Eero Kontturi, Tekla Tammelin, and Monika O?sterberg. Cellulose?model films
and the fundamental approach. Chemical Society Reviews, 35(12):1287?1304,
2006.
[3] Carina Olsson, Alexander Idstro?m, Lars Nordstierna, and Gunnar West-
man. Influence of water on swelling and dissolution of cellulose in 1-ethyl-3-
methylimidazolium acetate. Carbohydrate Polymers, 99:438?446, 2014.
[4] Monika S. Doblin, Isaac Kurek, Deborah Jacob-Wilk, and Deborah P. Delmer.
Cellulose biosynthesis in plants: From genes to rosettes. Plant and Cell
Physiology, 43(12):1407?1420, 2002.
[5] Inder M. Saxena and R. Malcolm Brown. Cellulose biosynthesis: Current views
and evolving concepts. Annals of Botany, 96(1):9?21, 2005.
[6] Daniel J. Cosgrove. Growth of the plant cell wall. Nature Reviews Molecular
Cell Biology, 6(11):850?861, 2005.
[7] Susan K. Cousins and R. Malcolm Brown. Cellulose I microfibril assembly:
computational molecular mechanics energy analysis favours bonding by van der
Waals forces as the initial step in crystallization. Polymer, 36(20):3885?3888,
1995.
[8] Chris Somerville. Cellulose synthesis in higher plants. Annual Review of Cell
and Developmental Biology, 22:53?78, 2006.
[9] Hiroshi Nishimura, Akihiro Kamiya, Takashi Nagata, Masato Katahira, and
Takashi Watanabe. Direct evidence for ? ether linkage between lignin and
carbohydrates in wood cell walls. Scientific Reports, 8(1):1?11, 2018.
173
[10] Luc Vinet and Alexei Zhedanov. A ?missing? family of classical orthogonal
polynomials. PhD thesis, 2011.
[11] Lloyd Donaldson. Cellulose microfibril aggregates and their size variation with
cell wall type. Wood Science and Technology, 41(5):443?460, jan 2007.
[12] Lorna J. Gibson. The hierarchical structure and mechanics of plant materials.
Journal of the Royal Society Interface, 9(76):2749?2766, 2012.
[13] J. R. Barnett and Victoria A. Bonham. Cellulose microfibril angle in the cell
wall of wood fibres. Biological Reviews of the Cambridge Philosophical Society,
79(2):461?472, 2004.
[14] Yoshiharu Nishiyama, Paul Langan, and Henri Chanzy. Crystal structure and
hydrogen-bonding system in cellulose I? from synchrotron X-ray and neutron
fiber diffraction. Journal of the American Chemical Society, 124(31):9074?9082,
2002.
[15] Yoshiharu Nishiyama, Junji Sugiyama, Henri Chanzy, and Paul Langan. Crystal
Structure and Hydrogen Bonding System in Cellulose I? from Synchrotron X-
ray and Neutron Fiber Diffraction. Journal of the American Chemical Society,
125(47):14300?14306, nov 2003.
[16] Shiro Kobayashi. Challenge of synthetic cellulose. Journal of Polymer Science,
Part A: Polymer Chemistry, 43(4):693?710, feb 2005.
[17] Francis J. Kolpak, Mark Weih, and John Blackwell. Mercerization of cellulose:
1. Determination of the structure of Mercerized cotton. Technical Report 2,
1978.
[18] Elizabeth Dinand, Michel Vignon, Henri Chanzy, and Laurent Heux. Mer-
cerization of primary wall cellulose and its implication for the conversion of
cellulose I  cellulose II. Technical Report 1, 2002.
[19] L. M.J. Kroon-Batenburg, B. Bouma, and J. Kroon. Stability of cellulose
structures studied by MD simulations. Could mercerized cellulose II be parallel?
Macromolecules, 29(17):5695?5699, 1996.
[20] Takanori Kurihara and Akira Isogai. Mechanism of TEMPO-oxidized cellulose
nanofibril film reinforcement with poly(acrylamide). Cellulose, 22(4):2607?2617,
2015.
[21] P. Langan, Y. Nishiyama, and H. Chanzy. X-ray structure of mercerized
cellulose II at 1 A? resolution. Biomacromolecules, 2(2):410?416, 2001.
[22] A. K. Roy Choudhury. Pre-treatment and preparation of textile materials prior
to dyeing. In Handbook of Textile and Industrial Dyeing: Principles, Processes
and Types of Dyes, volume 1, pages 64?149. Elsevier Inc., oct 2011.
174
[23] Masahisa Wada, Henri Chanzy, Yoshiharu Nishiyama, and Paul Langan. Cel-
lulose III I crystal structure and hydrogen bonding by synchrotron X-ray and
neutron fiber diffraction. Macromolecules, 37(23):8548?8555, nov 2004.
[24] M. Wada, L. Heux, A. Isogai, Y. Nishiyama, H. Chanzy, and J. Sugiyama.
Improved structural data of cellulose IIII prepared in supercritical ammonia.
Macromolecules, 34(5):1237?1243, 2001.
[25] Masahida Wada, Laurent Heux, and Junji Sugiyama. Polymorphism of cellulose
I family: Reinvestigation of cellulose IVl. Biomacromolecules, 5(4):1385?1391,
2004.
[26] Rui Lu. Morphology of Cellulose and Cellulose Blend Thin Films. PhD thesis,
2017.
[27] Susan K. Cousins and R. Malcolm Brown. X-ray diffraction and ultrastructural
analyses of dye-altered celluloses support van der Waals forces as the initial
step in cellulose crystallization. Polymer, 38(4):897?902, 1997.
[28] Hiroyuki Ohno and Yukinobu Fukaya. Task specific ionic liquids for cellulose
technology. Chemistry Letters, 38(1):2?7, 2009.
[29] A. R. Barron and Changsheng Xiang. The Analysis of Liquid Crystal Phases
using Polarized Optical Microscopy, 2014.
[30] Boualem Hammouda. Probing Nanoscale Structures?the Sans Toolbox. Tech-
nical report, 2010.
[31] Kai-Chi Lai. Small Angle Neutron Scattering of Arborescent Graft Polymer
Solutions. page 87, 2004.
[32] S. Freeman. FORM FACTORS: Notes and Examples. Technical report,
University of Manchester.
[33] C. J. Glinka, J. G. Barker, B. Hammouda, S. Krueger, J. J. Moyer, and W. J.
Orts. The 30 m Small-Angle Neutron Scattering Instruments at the National
Institute of Standards and Technology. Journal of Applied Crystallography,
31(3):430?445, jun 1998.
[34] Robert J. Markovich and Charles Pidgeon. Introduction to Fourier Trans-
form Infrared Spectroscopy and Applications in the Pharmaceutical Sciences.
Pharmaceutical Research: An Official Journal of the American Association of
Pharmaceutical Scientists, 8(6):663?675, 1991.
[35] Peter J. Larkin. Infrared and Raman Spectroscopy: Principles and Spectral
Interpretation. Elsevier, Amsterdam, 2017.
[36] Rheology of Thermosets Part 2: Rheometers - Polymer Innovation Blog.
175
[37] Ronald Reifenberger. Front Matter. Number Part I, Foundations in Funda-
mentals of Atomic Force Microscopy. World Scientific, Singapore, 2015.
[38] Antonio Aliano, Giancarlo Cicero, Hossein Nili, Nicolas G. Green, Pablo
Garc??a-Sa?nchez, Antonio Ramos, Andreas Lenshof, Thomas Laurell, Aisha
Qi, Peggy Chan, Leslie Yeo, James Friend, Mikael Evander, Thomas Laurell,
Andreas Lenshof, Thomas Laurell, Jian Chen, Jean Christophe Lacroix, Pascal
Martin, Hyacinthe Randriamahazaka, W. Jon. P. Barnes, Bart W. Hoogenboom,
Kenji Fukuzawa, Hendrik Ho?lscher, Hendrik Ho?lscher, Alessia Bottos, Elena
Astanina, Luca Primo, Federico Bussolino, Xuefeng Gao, Vinh-Nguyen Phan,
Nam-Trung Nguyen, Chun Yang, Patrick Abgrall, Friedrich G. Barth, Pablo
Gurman, Yitzhak Rosen, Orlando Auciello, C. J. Ka?hler, C. Cierpka, M. Rossi,
Bharat Bhushan, Manuel L. B. Palacio, and Charles L. Dezelah. AFM, Tapping
Mode. In Bharat Bhushan, editor, Encyclopedia of Nanotechnology, pages
99?99. Springer Netherlands, Dordrecht, 2012.
[39] Rui Lu, Xin Zhang, Li Fu, Hongfei Wang, Robert M. Briber, and Howard
Wang. Amorphous cellulose thin films. Cellulose, 27(6):2959?2965, 2020.
[40] Integrating Spheres: SHIMADZU (Shimadzu Corporation).
[41] DMA Accessories ? TA Instruments.
[42] AMMRF. What is SEM?, feb 2014.
[43] Weilie Zhou, Robert Apkarian, Zhong Lin Wang, and David Joy. Fundamentals
of scanning electron microscopy (SEM). Springer New York, 2007.
[44] Anjam Khursheed. Scanning electron microscope optics and spectrometers.
Singapore: World Scientific, 2010.
[45] George H. Michler. Electron Microscopy of Polymers. Springer Berlin Heidel-
berg, Berlin, Heidelberg, 2008.
[46] Pavan M. V. Raja and Andrew R. Barron. 9.3: SEM and its Applications for
Polymer Science - Chemistry LibreTexts, 2020.
[47] Methodology and tools for failure analysis of polymer composites. In Failure
Analysis and Fractography of Polymer Composites, pages 23?106. Elsevier,
2009.
[48] Ki Hyun Kim, Zentaro Akase, Toshiaki Suzuki, and Daisuke Shindo. Charging
effects on SEM/SIM contrast of metal/insulator system in various metallic
coating conditions. Materials Transactions, 51(6):1080?1083, 2010.
[49] Younan Xia, Peidong Yang, Yugang Sun, Yiying Wu, Brian Mayers, Byron
Gates, Yadong Yin, Franklin Kim, and Haoquan Yan. One-dimensional nanos-
tructures: Synthesis, characterization, and applications. Advanced Materials,
15(5):353?389, mar 2003.
176
[50] D. R. Paul and L. M. Robeson. Polymer nanotechnology: Nanocomposites for
food packaging applications. Polymer, 49(15):3187?3204, 2008.
[51] Kentaro Abe, Shinichiro Iwamoto, and Hiroyuki Yano. Obtaining cellulose
nanofibers with a uniform width of 15 nm from wood. Biomacromolecules,
8(10):3276?3278, 2007.
[52] Akira Isogai, Tsuguyuki Saito, and Hayaka Fukuzumi. TEMPO-oxidized
cellulose nanofibers. Nanoscale, 3(1):71?85, jan 2011.
[53] R. H. Marchessault, F. F. Morehead, and N. M. Walter. Liquid crystal systems
from fibrillar polysaccharides, 1959.
[54] Samira Elazzouzi-Hafraoui, Yoshiharu Nishiyama, Jean Luc Putaux, Laurent
Heux, Fre?de?ric Dubreuil, and Cyrille Rochas. The shape and size distribution
of crystalline nanoparticles prepared by acid hydrolysis of native cellulose.
Biomacromolecules, 9(1):57?65, 2008.
[55] Xue Min Dong, Jean Franc?ois Revol, and Derek G. Gray. Effect of microcrys-
tallite preparation conditions on the formation of colloid crystals of cellulose.
Cellulose, 5(1):19?32, 1998.
[56] M. Pa?a?kko, M. Ankerfors, H. Kosonen, A. Nyka?nen, S. Ahola, M. O?sterberg,
J. Ruokolainen, J. Laine, P. T. Larsson, Olli Ikkala, and T. Lindstro?m. Enzy-
matic hydrolysis combined with mechanical shearing and high-pressure homog-
enization for nanoscale cellulose fibrils and strong gels. Biomacromolecules,
8(6):1934?1941, 2007.
[57] M. Henriksson, G. Henriksson, L. A. Berglund, and T. Lindstro?m. An environ-
mentally friendly method for enzyme-assisted preparation of microfibrillated
cellulose (MFC) nanofibers. European Polymer Journal, 43(8):3434?3441, 2007.
[58] Tsuguyuki Saito, Satoshi Kimura, Yoshiharu Nishiyama, and Akira Isogai.
Cellulose nanofibers prepared by TEMPO-mediated oxidation of native cellulose.
Biomacromolecules, 8(8):2485?2491, 2007.
[59] Karima Benhamou, Alain Dufresne, Albert Magnin, Ge?rard Mortha, and
Hamid Kaddami. Control of size and viscoelastic properties of nanofibrillated
cellulose from palm tree by varying the TEMPO-mediated oxidation time.
Carbohydrate Polymers, 99:74?83, 2014.
[60] P. L. Bragd, H. Van Bekkum, and A. C. Besemer. TEMPO-mediated oxidation
of polysaccharides: Survey of methods and applications. Topics in Catalysis,
27(1-4):49?66, 2004.
[61] Yusuke Okita, Tsuguyuki Saito, and Akira Isogai. Entire surface oxidation
of various cellulose microfibrils by TEMPO-mediated oxidation. Biomacro-
molecules, 11(6):1696?1700, 2010.
177
[62] Takuya Isogai, Tsuguyuki Saito, and Akira Isogai. TEMPO electromedi-
ated oxidation of some polysaccharides including regenerated cellulose fiber.
Biomacromolecules, 11(6):1593?1599, jul 2010.
[63] J. F. Revol. On the cross-sectional shape of cellulose crystallites in Valonia
ventricosa. Carbohydrate Polymers, 2(2):123?134, jan 1982.
[64] Hayaka Fukuzumi, Tsuguyuki Saito, Tadahisa Iwata, Yoshiaki Kumamoto, and
Akira Isogai. Transparent and high gas barrier films of cellulose nanofibers
prepared by TEMPO-mediated oxidation. Biomacromolecules, 10(1):162?165,
2009.
[65] Hongli Zhu, Shuze Zhu, Zheng Jia, Sepideh Parvinian, Yuanyuan Li, Oeyvind
Vaaland, Liangbing Hu, and Teng Li. Anomalous scaling law of strength and
toughness of cellulose nanopaper. Proceedings of the National Academy of
Sciences of the United States of America, 112(29):8971?8976, 2015.
[66] Robert O. Ritchie. The conflicts between strength and toughness. Nature
Materials, 10(11):817?822, 2011.
[67] Hongli Zhu, Zhiqiang Fang, Colin Preston, Yuanyuan Li, and Liangbing Hu.
Transparent paper: Fabrications, properties, and device applications. Energy
and Environmental Science, 7(1):269?287, 2014.
[68] Zhiqiang Fang, Hongli Zhu, Wenzhong Bao, Colin Preston, Zhen Liu, Jiaqi Dai,
Yuanyuan Li, and Liangbing Hu. Highly transparent paper with tunable haze
for green electronics. Energy and Environmental Science, 7(10):3313?3319,
2014.
[69] Zhiqiang Fang, Hongli Zhu, Yongbo Yuan, Dongheon Ha, Shuze Zhu, Colin
Preston, Qingxia Chen, Yuanyuan Li, Xiaogang Han, Seongwoo Lee, Gang
Chen, Teng Li, Jeremy Munday, Jinsong Huang, and Liangbing Hu. Novel
nanostructured paper with ultrahigh transparency and ultrahigh haze for solar
cells. Nano Letters, 14(2):765?773, 2014.
[70] Bozhen Wu, Biyao Geng, Yufei Chen, Hongzhi Liu, Guangyao Li, and Qiang
Wu. Preparation and characteristics of TEMPO-oxidized cellulose nanofibrils
from bamboo pulp and their oxygen-barrier application in PLA films. Frontiers
of Chemical Science and Engineering, 11(4):554?563, 2017.
[71] Jia Huang, Hongli Zhu, Yuchen Chen, Colin Preston, Kathleen Rohrbach,
John Cumings, and Liangbing Hu. Highly transparent and flexible nanopaper
transistors. ACS Nano, 7(3):2106?2113, 2013.
[72] Hongli Zhu, Sepideh Parvinian, Colin Preston, Oeyvind Vaaland, Zhichao Ruan,
and Liangbing Hu. Transparent nanopaper with tailored optical properties.
Nanoscale, 5(9):3787?3792, may 2013.
178
[73] Hongli Zhu, Zhengguo Xiao, Detao Liu, Yuanyuan Li, Nicholas J. Weadock,
Zhiqiang Fang, Jinsong Huang, and Liangbing Hu. Biodegradable transparent
substrates for flexible organic-light-emitting diodes. Energy and Environmental
Science, 6(7):2105?2111, jul 2013.
[74] Zhiqiang Fang, Hongli Zhu, Colin Preston, Xiaogang Han, Yuanyuan Li, Seong-
woo Lee, Xinsheng Chai, Gang Chen, and Liangbing Hu. Highly transparent
and writable wood all-cellulose hybrid nanostructured paper. Journal of
Materials Chemistry C, 1(39):6191?6197, oct 2013.
[75] Caroline A. Schneider, Wayne S. Rasband, and Kevin W. Eliceiri. NIH Image
to ImageJ: 25 years of image analysis. Nature Methods, 9(7):671?675, jul 2012.
[76] S. M. Choi, J. G. Barker, C. J. Glinka, Y. T. Cheng, and P. L. Gammel.
Focusing cold neutrons with multiple biconcave lenses for small-angle neutron
scattering. In Journal of Applied Crystallography, volume 33, pages 793?796,
2000.
[77] Steven R. Kline. Reduction and analysis of SANS and USANS data using
IGOR Pro. Journal of Applied Crystallography, 39(6):895?900, dec 2006.
[78] Boualem Hammouda, Derek L. Ho, and Steve Kline. Insight into clustering in
poly(ethylene oxide) solutions. Macromolecules, 37(18):6932?6937, 2004.
[79] Paavo A. Penttila?, Aniko? Va?rnai, Manuel Ferna?ndez, Inkeri Kontro, Ville
Liljestro?m, Peter Lindner, Matti Siika-aho, Liisa Viikari, and Ritva Serimaa.
Small-angle scattering study of structural changes in the microfibril network of
nanocellulose during enzymatic hydrolysis. Cellulose, 20(3):1031?1040, 2013.
[80] George B. Kauffman. Products of chemistry: Rayon: The first semi-synthetic
fiber product. Journal of Chemical Education, 70(11):887?893, 1993.
[81] Paul C. Painter and Michael M. Coleman. Essentials of Polymer Science and
Engineering, 2009.
[82] Rayon (Viscose) ? Materials Index ? CFDA.
[83] Thomas Rosenau, Antje Potthast, Herbert Sixta, and Paul Kosma. The chem-
istry of side reactions and byproduct formation in the system NMMO/cellulose
(Lyocell process). Progress in Polymer Science (Oxford), 26(9):1763?1837,
2001.
[84] What is Viscose Fabric: Properties, How its Made and Where ? Sewport.
[85] N. Mohd, S. F.S. Draman, M. S.N. Salleh, and N. B. Yusof. Dissolution of
cellulose in ionic liquid: A review. AIP Conference Proceedings, 1809(February),
2017.
179
[86] Kyung Yul Lim, Jeong Seong Yun, and Byoung Chul Kim. Reduction of
Fibrillation of Lyocell Fiber with Cellulose-g-Poly(vinyl alcohol) Copolymer.
Technical Report 9, 2003.
[87] Wencong Wang, Peng Zhang, Shuai Zhang, Faxue Li, Jianyong Yu, and Jinyou
Lin. Structure and properties of novel regenerated cellulose fibers prepared in
NaOH complex solution. Carbohydrate polymers, 98(1):1031?1038, 2013.
[88] K. E. Perepelkin. Lyocell fibres based on direct dissolution of cellulose in
N-methylmorpholine N-oxide: Development and prospects. In Fibre Chemistry,
volume 39, pages 163?172. March?April, sep 2007.
[89] Jie Cai and Lina Zhang. Rapid dissolution of cellulose in LiOH/urea and
NaOH/urea aqueous solutions. Macromolecular Bioscience, 5(6):539?548, jun
2005.
[90] Jie Cai, Lina Zhang, Chunyu Chang, Gongzhen Cheng, Xuming Chen, and
Benjamin Chu. Hydrogen-bond-induced inclusion complex in aqueous cellu-
lose/LiOH/urea solution at low temperature. ChemPhysChem, 8(10):1572?1579,
jul 2007.
[91] Christian Capello, Ulrich Fischer, and Konrad Hungerbu?hler. What is a green
solvent? A comprehensive framework for the environmental assessment of
solvents. Green Chemistry, 9(9):927?93, 2007.
[92] Ning Sun, He?ctor Rodr??guez, Mustafizur Rahman, and Robin D. Rogers. Where
are ionic liquid strategies most suited in the pursuit of chemicals and energy
from lignocellulosic biomass? Chemical Communications, 47(5):1405?1421,
2011.
[93] Richard P. Swatloski, Scott K. Spear, John D. Holbrey, and Robin D. Rogers.
Dissolution of cellose with ionic liquids. Journal of the American Chemical
Society, 124(18):4974?4975, 2002.
[94] P. Ma?ki-Arvela, I. Anugwom, P. Virtanen, R. Sjo?holm, and J. P. Mikkola.
Dissolution of lignocellulosic materials and its constituents using ionic liquids-A
review. Industrial Crops and Products, 32(3):175?201, 2010.
[95] Rajdeep Singh Payal, Karteek K. Bejagam, Anirban Mondal, and Sundaram
Balasubramanian. Dissolution of cellulose in room temperature ionic liquids:
Anion dependence. Journal of Physical Chemistry B, 119(4):1654?1659, 2015.
[96] R. M. Wahlstro?m and A. Suurna?kki. Enzymatic hydrolysis of lignocellulosic
polysaccharides in the presence of ionic liquids. Green Chemistry, 17(2):694?714,
2015.
[97] Hao Zhang, Jin Wu, Jun Zhang, and Jiasong He. 1-allyl-3-methylimidazolium
chloride room temperature ionic liquid: A new and powerful nonderivatizing
solvent for cellulose. Macromolecules, 38(20):8272?8277, 2005.
180
[98] Michael Zavrel, Daniela Bross, Matthias Funke, Jochen Bu?chs, and Antje C.
Spiess. High-throughput screening for ionic liquids dissolving (ligno-)cellulose.
Bioresource Technology, 100(9):2580?2587, 2009.
[99] Hui Wang, Gabriela Gurau, and Robin D. Rogers. Ionic liquid processing of
cellulose. Chemical Society Reviews, 41(4):1519?1537, 2012.
[100] Dmitry M. Rein, Rafail Khalfin, Noemi Szekely, and Yachin Cohen. True
molecular solutions of natural cellulose in the binary ionic liquid-containing
solvent mixtures. Carbohydrate Polymers, 112:125?133, 2014.
[101] Olga Kuzmina, Elena Sashina, Svetlana Troshenkowa, and Dariusz Wawro.
Dissolved state of cellulose in ionic liquids - the impact of water. Fibres and
Textiles in Eastern Europe, 80(3):32?37, 2010.
[102] Yan Cao, Huiquan Li, Yi Zhang, Jun Zhang, and Jiasong He. Structure
and properties of novel regenerated cellulose films prepared from cornhusk
cellulose in room temperature ionic liquids. Journal of Applied Polymer Science,
116(1):547?554, apr 2010.
[103] Xin Zhang, Yimin Mao, Madhusudan Tyagi, Feng Jiang, Doug Henderson,
Bo Jiang, Zhiwei Lin, Ronald L. Jones, Liangbing Hu, Robert M. Briber,
and Howard Wang. Molecular partitioning in ternary solutions of cellulose.
Carbohydrate Polymers, 220:157?162, 2019.
[104] Kim Anh Le, Cyrielle Rudaz, and Tatiana Budtova. Phase diagram, solubility
limit and hydrodynamic properties of cellulose in binary solvents with ionic
liquid. Carbohydrate Polymers, 105(1):237?243, 2014.
[105] Yi Lei, Haoran Li, and Shijun Han. An all-atom simulation study on in-
termolecular interaction of DMSO-water system. Chemical Physics Letters,
380(5-6):542?548, 2003.
[106] Hua Zhao, Gary A. Baker, Zhiyan Song, Olarongbe Olubajo, Tanisha Crittle,
and Darkeysha Peters. Designing enzyme-compatible ionic liquids that can
dissolve carbohydrates. Green Chemistry, 10(6):696?70, 2008.
[107] Gang Cheng, Patanjali Varanasi, Rohit Arora, Vitalie Stavila, Blake A. Sim-
mons, Michael S. Kent, and Seema Singh. Impact of ionic liquid pretreatment
conditions on cellulose crystalline structure using 1-ethyl-3-methylimidazolium
acetate. Journal of Physical Chemistry B, 116(33):10049?10054, 2012.
[108] Gang Cheng, Xin Zhang, Blake Simmons, and Seema Singh. Theory, practice
and prospects of X-ray and neutron scattering for lignocellulosic biomass
characterization: Towards understanding biomass pretreatment. Energy and
Environmental Science, 8(2):436?455, 2015.
181
[109] Amine Bendaoud, Rene Kehrbusch, Anton Baranov, Beno??t Duchemin, Jean Eu-
des Maigret, Xavier Falourd, Mark P. Staiger, Bernard Cathala, Denis Lour-
din, and Eric Leroy. Nanostructured cellulose-xyloglucan blends via ionic
liquid/water processing. Carbohydrate Polymers, 168:163?172, 2017.
[110] Bo Xing Zhang, Jun Ichi Azuma, and Hiroshi Uyama. Preparation and
characterization of a transparent amorphous cellulose film. RSC Advances,
5(4):2900?2907, 2015.
[111] F. Dourado, F. M. Gama, E. Chibowski, and M. Mota. Characterization of
cellulose surface free energy. Journal of Adhesion Science and Technology,
12(10):1081?1090, 1998.
[112] Daisuke Ishii, Tsuguyuki Saito, and Akira Isogai. Viscoelastic evaluation of
average length of cellulose nanofibers prepared by tempo-mediated oxidation.
Biomacromolecules, 12(3):548?550, 2011.
[113] Shuji Fujisawa, Yusuke Okita, Hayaka Fukuzumi, Tsuguyuki Saito, and Akira
Isogai. Preparation and characterization of TEMPO-oxidized cellulose nanofib-
ril films with free carboxyl groups. Carbohydrate Polymers, 84(1):579?583,
2011.
[114] A. J. Ben??tez and A. Walther. Cellulose nanofibril nanopapers and bioinspired
nanocomposites: A review to understand the mechanical property space.
Journal of Materials Chemistry A, 5(31):16003?16024, 2017.
[115] Yusuke Okita, Shuji Fujisawa, Tsuguyuki Saito, and Akira Isogai. TEMPO-
oxidized cellulose nanofibrils dispersed in organic solvents. Biomacromolecules,
12(2):518?522, 2011.
[116] Shuji Fujisawa, Tomoyasu Ikeuchi, Miyuki Takeuchi, Tsuguyuki Saito, and
Akira Isogai. Superior reinforcement effect of TEMPO-oxidized cellulose
nanofibrils in polystyrene matrix: optical, thermal, and mechanical studies.
Biomacromolecules, 13(7):2188?2194, 2012.
[117] Eftihia Barnes, Jennifer A. Jefcoat, Erik M. Alberts, Mason A. McKech-
nie, Hannah R. Peel, J. Paige Buchanan, Charles A. Weiss, Kyle L. Klaus,
L. Christopher Mimun, and Christopher M. Warner. Effect of cellulose nanofib-
rils and TEMPO-mediated oxidized cellulose nanofibrils on the physical and
mechanical properties of poly(vinylidene fluoride)/cellulose nanofibril compos-
ites. Polymers, 11(7), 2019.
[118] Michiko Shimizu, Tsuguyuki Saito, and Akira Isogai. Water-resistant and
high oxygen-barrier nanocellulose films with interfibrillar cross-linkages formed
through multivalent metal ions. Journal of Membrane Science, 500:1?7, 2016.
[119] Houssine Sehaqui, Tanja Zimmermann, and Philippe Tingaut. Hydropho-
bic cellulose nanopaper through a mild esterification procedure. Cellulose,
21(1):367?382, 2014.
182
[120] Steven Spoljaric, Arto Salminen, Nguyen Dang Luong, and Jukka Seppa?la?.
Crosslinked nanofibrillated cellulose: Poly(acrylic acid) nanocomposite films; en-
hanced mechanical performance in aqueous environments. Cellulose, 20(6):2991?
3005, dec 2013.
[121] Matti S. Toivonen, Sauli Kurki-Suonio, Felix H. Schacher, Sami Hietala,
Orlando J. Rojas, and Olli Ikkala. Water-Resistant, Transparent Hybrid
Nanopaper by Physical Cross-Linking with Chitosan. Biomacromolecules,
16(3):1062?1071, 2015.
[122] Nikolaos Pahimanolis, Arto Salminen, Paavo A. Penttila?, Juuso T. Korhonen,
Leena Sisko Johansson, Janne Ruokolainen, Ritva Serimaa, and Jukka Seppa?la?.
Nanofibrillated cellulose/carboxymethyl cellulose composite with improved wet
strength. Cellulose, 20(3):1459?1468, jun 2013.
[123] My Ahmed Said Azizi Samir, Fannie Alloin, and Alain Dufresne. Review of
recent research into cellulosic whiskers, their properties and their application
in nanocomposite field. Biomacromolecules, 6(2):612?626, 2005.
[124] Youssef Habibi, Lucian A. Lucia, and Orlando J. Rojas. Cellulose nanocrystals:
Chemistry, self-assembly, and applications. Chemical Reviews, 110(6):3479?
3500, 2010.
[125] Houssine Sehaqui, Andong Liu, Qi Zhou, and Lars A. Berglund. Fast prepa-
ration procedure for large, flat cellulose and cellulose/inorganic nanopaper
structures. Biomacromolecules, 11(9):2195?2198, 2010.
[126] Houssine Sehaqui, Qi Zhou, and Lars A. Berglund. Nanostructured biocom-
posites of high toughness - A wood cellulose nanofiber network in ductile
hydroxyethylcellulose matrix. Soft Matter, 7(16):7342?7350, 2011.
[127] Alejandro J. Ben??tez, Francisco Lossada, Baolei Zhu, Tobias Rudolph, and
Andreas Walther. Understanding Toughness in Bioinspired Cellulose Nanofib-
ril/Polymer Nanocomposites. Biomacromolecules, 17(7):2417?2426, jul 2016.
[128] Takanori Kurihara and Akira Isogai. Properties of poly(acrylamide)/TEMPO-
oxidized cellulose nanofibril composite films. Cellulose, 21(1):291?299, 2014.
[129] Hyunjin Kim, Ji Eun Song, and Hye Rim Kim. Comparative study on the
physical entrapment of soy and mushroom proteins on the durability of bacterial
cellulose bio-leather. Cellulose, 28(5):3183?3200, 2021.
[130] Ziyan Zhou, Hua Zheng, Ming Wei, Jin Huang, and Yun Chen. Structure
and mechanical properties of cellulose derivatives/soy protein isolate blends.
Journal of Applied Polymer Science, 107(5):3267?3274, 2008.
[131] Jinwu Wang, Lu Wang, Douglas J. Gardner, Stephen M. Shaler, and Zhiyong
Cai. Towards a cellulose-based society: opportunities and challenges. Cellulose,
28(8):4511?4543, 2021.
183
[132] Tim Huber, Jo?rg Mu?ssig, Owen Curnow, Shusheng Pang, Simon Bickerton,
and Mark P. Staiger. A critical review of all-cellulose composites. Journal of
Materials Science, 47(3):1171?1186, 2012.
[133] Behnaz Baghaei and Mikael Skrifvars. All-Cellulose Composites: A Review of
Recent Studies on Structure, Properties and Applications. Molecules, 25(12),
2020.
[134] Hao Ma, Bo Zhou, Hong Sheng Li, Yi Qun Li, and Shi Yi Ou. Green composite
films composed of nanocrystalline cellulose and a cellulose matrix regenerated
from functionalized ionic liquid solution. Carbohydrate Polymers, 84(1):383?389,
2011.
[135] Hannes Orelma, Antti Korpela, Vesa Kunnari, Ali Harlin, and Anna Suurna?kki.
Improving the mechanical properties of CNF films by NMMO partial dissolution
with hot calender activation. Cellulose, 24(4):1691?1704, 2017.
[136] Alireza Shakeri, Aji P. Mathew, and Kristiina Oksman. Self-reinforced nanocom-
posite by partial dissolution of cellulose microfibrils in ionic liquid. Journal of
Composite Materials, 46(11):1305?1311, jun 2012.
[137] Nattakan Soykeabkaew, Takashi Nishino, and Ton Peijs. All-cellulose compos-
ites of regenerated cellulose fibres by surface selective dissolution. Composites
Part A: Applied Science and Manufacturing, 40(4):321?328, apr 2009.
[138] Quanling Yang, Tsuguyuki Saito, Lars A. Berglund, and Akira Isogai. Cellulose
nanofibrils improve the properties of all-cellulose composites by the nano-
reinforcement mechanism and nanofibril-induced crystallization. Nanoscale,
7(42):17957?17963, 2015.
[139] Ming Chun Hsieh, Hirotaka Koga, Katsuaki Suganuma, and Masaya Nogi.
Hazy Transparent Cellulose Nanopaper. Scientific Reports, 7, 2017.
[140] Hayaka Fukuzumi, Tsuguyuki Saito, and Akira Isogai. Influence of TEMPO-
oxidized cellulose nanofibril length on film properties. Carbohydrate Polymers,
93(1):172?177, 2013.
[141] Tsuguyuki Saito, Takehiko Uematsu, Satoshi Kimura, Toshiharu Enomae, and
Akira Isogai. Self-aligned integration of native cellulose nanofibrils towards
producing diverse bulk materials. Soft Matter, 7(19):8804?8809, 2011.
[142] Jun Zhang, Jinming Zhang, Nan Luo, Xiaoyu Zhang, Lili Xu, Jin Wu, Jian
Yu, and Jiasong He. All-Cellulose Nanocomposites Reinforced with in Situ
Retained Cellulose Nanocrystals during Selective Dissolution of Cellulose in
an Ionic Liquid. ACS Sustainable Chemistry and Engineering, 4(8):4417?4423,
2016.
184
[143] Jinyang Li, Haq Nawaz, Jin Wu, Jinming Zhang, Jiqiang Wan, Qinyong Mi,
Jian Yu, and Jun Zhang. All-cellulose composites based on the self-reinforced
effect. Composites Communications, 9:42?53, sep 2018.
[144] Jan W. Dormanns, Jeremias Schuermann, Jo?rg Mu?ssig, Beno??t J.C. Duchemin,
and Mark P. Staiger. Solvent infusion processing of all-cellulose composite
laminates using an aqueous NaOH/urea solvent system. Composites Part A:
Applied Science and Manufacturing, 82:130?140, 2016.
[145] Moein Ghaderi, Mohammad Mousavi, Hossein Yousefi, and Mohsen Labbafi.
All-cellulose nanocomposite film made from bagasse cellulose nanofibers for
food packaging application. Carbohydrate Polymers, 104(1):59?65, 2014.
[146] J. A. Sirvio?, M. Visanko, and N. C. Hildebrandt. Rapid preparation of all-
cellulose composites by solvent welding based on the use of aqueous solvent.
European Polymer Journal, 97:292?298, 2017.
[147] M. Mat Salleh, K. Magniez, S. Pang, J. W. Dormanns, and M. P. Staiger.
Parametric optimization of the processing of all-cellulose composite laminae.
Advanced Manufacturing: Polymer and Composites Science, 3(2):73?79, apr
2017.
[148] Wan Hafizi Wan Ishak, Noor Afizah Rosli, and Ishak Ahmad. Influence of
amorphous cellulose on mechanical, thermal, and hydrolytic degradation of
poly(lactic acid) biocomposites. Scientific Reports, 10(1), 2020.
[149] Kazuyuki Hattori and Arisa Arai. Preparation and Hydrolysis of Water-
Stable Amorphous Cellulose. ACS Sustainable Chemistry and Engineering,
4(3):1180?1186, 2016.
[150] A. Isogai and R. H. Atalla. Amorphous celluloses stable in aqueous media:
Regeneration from SO2?amine solvent systems. Journal of Polymer Science
Part A: Polymer Chemistry, 29(1):113?119, 1991.
[151] Yimin Mao, Markus Bleuel, Yadong Lyu, Xin Zhang, Doug Henderson, Howard
Wang, and Robert M. Briber. Phase Separation and Stack Alignment in Aque-
ous Cellulose Nanocrystal Suspension under Weak Magnetic Field. Langmuir,
34(27):8042?8051, 2018.
185