ABSTRACT TRIPATHI, ANURODH. Bioinspired Light-Weight Materials

ABSTRACT

TRIPATHI, ANURODH. Bioinspired Light-Weight Materials Using Biopolymers: From Synthesis to Application (Under the direction of Dr. Saad A. Khan and Dr. Orlando J. Rojas).

Increasing environmental pollution together with an irreversible state of climate change necessitates the development of efficient light-weight materials from renewable resources. This dissertation explores the interaction of individual units in a biomass, providing inspiration to produce light-weight materials such as aerogels and fibers with a morphology that mimics the cellular structure of plant cells and open porous structure of a cancellous bone.

The first part of the dissertation (Chapter 2), explores the interaction of individual units: cellulose, lignin, and hemicellulose in a biomass. The two biomasses chosen for this study are waste fibers from coconut fruit (coir fibers) and palm fruit (empty fruit bunches (EFB)). To study the structural and chemical arrangement of the individual units, the fibers are subjected to mild treatments steps to isolate the fibers into micro-nano size fibrils. The restructuring of cellulose, lignin, and hemicellulose is tracked at each step of deconstruction using microscopy, chemical and thermal analyses. During the process, we also synthesize a novel lignin-rich micro nano-fibrillated cellulose (MNFC) that holds potential for use in further applications.

The second part of the dissertation (Chapter 3,4, and 5) explores the synthesis and application of aerogels using the plant-based commercial biopolymer, cellulose esters. The aerogel synthesis procedure involves chemical cross-linking of cellulose ester to form a gel, followed by critical point drying (CPD) or freeze-drying (FD). While starting from the same concentration, the

CPD produces the aerogels with high density (60-70 mg/cm3), whereas, the incorporation of solvent exchange step before FD produces aerogels with ultra-low density (4 mg/cm3), that is well below the calculated density (20 mg/cm3). The observation led to the development of a solubility parameter based approach for solvent exchange to predict and control the swelling of a cross- linked gel, thus providing control over the density and mechanical properties of the corresponding aerogel. Hence, from a given concentration, aerogels are produced in a range of mass densities

(25-113 mg/cm3), stiffness (14-340 kPa), toughness (4-102 kPa) and compressive strength (22-

372 kPa). Additionally, ice-templating introduces anisotropic pore alignment in aerogels.

The aerogel is comprehensively tested for performance in oil-spill remediation and is found to surpass its cellulosic counterparts in maintaining their mechanical integrity for fast oil cleanup and efficient oil retention from aqueous media under marine conditions. Additionally, the Zisman and Fowkes’ theoretical frameworks are used to identify the selectiveness of the aerogel and establish a criterion for separation of various non-polar fluids from water media, thereby providing a general platform for predicting the sorption ability of the aerogels for various chemicals.

The last section of this dissertation (Chapter 6) is devoted to the development of mesoporous fibers via wet-spinning that exhibit morphology similar to that of a cancellous bone.

The fiber morphology is predicted using design parameters based on the combination of (1)

Relative Energy Difference (RED) of Hansen solubility and (2) a kinetic parameter ‘T’ that incorporates mass transfer considerations. Predictions derived from (1) and (2) are experimentally validated by varying three variables relevant to fiber production by wet-spinning, namely, choice of polymer, solvent, and non-solvent, and analyzing the resultant fiber morphology via electron microscopy imaging. The practicality of the design parameters is further put to test by using data from the published literature on wet-spinning, providing a basis for future developments in this important area that currently explores nanomaterials derived from renewable resources.

by Anurodh Tripathi

A dissertation submitted to the Graduate Faculty of North Carolina State University in partial fulfillment of the requirements for the degree of Doctor of Philosophy

Chemical Engineering & Forest Biomaterials

Raleigh, North Carolina

2018

APPROVED BY:

______Dr. Saad A. Khan Dr. Orlando J. Rojas Committee Co-Chair Committee Co-Chair

______Dr. Gregory N. Parsons Dr. Richard J. Spontak

DEDICATION

In the memory of my parents Late Shri Awadhesh Kumar Tripathi and Late Smt. Archana Tripathi for their sacrifices and endless support

BIOGRAPHY

Anurodh Tripathi was born in a small town, Kannauj in Uttar Pradesh, India. He is the elder son to his parents Late Mr. Awadhesh K. Tripathi and Late Mrs. Archana Tripathi and has one younger sister Ms. Nirotma Tripathi, who is a medical doctor in India.

Anurodh completed his high school studies in Kanpur, UP and went to Indian Institute of

Technology, Kanpur (IIT-K) to study chemical engineering. During his undergrad, he explored the world of nanotechnology while working in the research lab of Dr. Sri Sivakumar for 2 years. To support his family financially, after his graduation in 2011, he joined an oil refinery (Reliance

Industries limited) in Jamnagar, Gujarat, India, where he served as a process engineer for 2 years.

However, the love for research pulled him back to academia, when in 2013, he decided to move to the United States to pursue Ph.D. in Chemical and Biomolecular Engineering (CBE) and Forest

Biomaterials at NC State University to work with Dr. Saad A. Khan and Dr. Orlando J. Rojas.

Throughout his time as a graduate student, he was involved in several extracurricular activities. He was Treasurer in the Graduate student Association of the CBE department, where he started a new student seminar series. He was part of Indian student association at NC State,

MAITRI that worked towards easing the settlement of new Indian students at NC State. He also served as Recruiting Captain for CBE department with a goal to encourage potential new graduate students to join the department.

During Ph.D., he also met his wife, Devangana Sharma and they have been married for over a year now. Outside of research, he enjoys spending his time outdoors hiking, bag packing and climbing. He enjoys reading books on the variety of topics that include fantasy, history, and environment. He found a recent interest in photography and likes to photograph Nature.

iii

ACKNOWLEDGMENTS

First, I would like to thank my advisors Dr. Saad A. Khan and Dr. Orlando J. Rojas.

Without, their constant mentorship and guidance, this Ph.D. dissertation would not have been possible. Dr. Khan was extremely helpful during my admission at NC State by providing me with the required contacts and the information needed to decide to join NC State. Since then, not only has he guided me towards my Ph.D. dissertation, he has also been there to celebrate my personal landmarks. Dr. Rojas, even though he moved to Finland just after I joined his research group, he has always been there when I needed him the most. I admire his work ethic and try to emulate his positive, hardworking attitude. I would also like to thank my committee members- Dr. Greg

Parsons for the insightful feedbacks during sponsor meetings and in my research articles, Dr. Rich

Spontak for sharing the knowledge of polymer science that inspired me to work on phase separation. A special acknowledgment to Dr. Genzer for the guidance on solubility parameter and providing me with career counseling.

I would like to thank all the active and former members of both Khan and Rojas research group at NC State and Aalto University, for their invaluable support over the years. I enjoyed working with everyone and learned a lot in the process. In addition, I extend my special thanks to staff members in Chemical and Biomolecular Engineering (CBE) and Forest Biomaterials department who assisted me with administrative issues and were always helpful in solving my travel reimbursement queries (there were many!). I would also thank staff scientists at the

Analytical Instrumentation Facility (AIF) at NC State to support me in the characterization of samples.

During my Ph.D., I was fortunate enough to work with many smart and kind-hearted people. I learned a lot while working with them. In the CBE department, I extend my thanks to the

number of people- Dr. Mandy Burns and Dr. Prajesh Adhikari for introducing me to nanodiamonds, Dr. Marius Rutkevičius, who worked with me on porous fibers, Dr. Tahira Pirzada,

Barbara Vasconcelos de Farias who helped me in crop protection experiments and Emily Facchine, with whom, I am currently exploring rheology of cellulose nanofibers. I thank Dr. Charlie

Opperman and Reny from Plant pathology department to educate me in the field of Crop protection while calmly answering my persistent questions. In the Department of Forest Biomaterials, I thank-

Dr. Hasan Jameel and Dr. Ana Ferrer for teaching me the pulping process. Dr. Abdelrahman

Abdelgawad for introducing me to Curdlan and guiding me initially on the cellulose acetate project. In Aalto University, Finland, I would like to acknowledge- Dr. Blaise Tardy and Konrad

Klockers for introducing me to the immense possibilities of cellulose nanocrystals. Dr. Mariko

Ago for training me on cellulose chemistry. Dr. Maryam Borghei and Katariina Solin, who worked tirelessly with me to make cellulose films. Liang Liu for elucidating the vast potential of chitin nanofibers and Dr. Guillermo Reyes for teaching me the correct procedure of fiber tensile testing.

In addition, I would like to thank Dr. Orlando Rojas for providing me the opportunity to visit

Finland and become a part of his wonderful research group. The dark days in Finland were made more memorable with the company of some amazing people- Devangana, Selma, Blaise, Sokha,

Bruno, Tai, Meri, Wenchao, Ling, Janika, Joice, Keisuke, Luiz, Guillermo, Tero, and Alexey.

The 5 years of Ph.D. would have been dulling without devoted friends like Dr. Rohan Patil,

Dr. Prajesh Adhikari, Dr. Ishan Joshipura and future Dr. Ankit Chandra. We shared a lot of memories and life-changing experiences at Firth of Tay Way that I will always cherish. In addition,

I will thank Dr. Dishit Parekh, Bharadwaja STP, Vasudev Haribal, Dr. Shivaramu Keelara, Shilpa

Keelara, Sangchul Roh, Amit Mishra and Dr. David Chang for providing memorable moments in form of numerous movies, dinners, night outs, camping, hiking, and climbing.

This Ph.D. dissertation is possible only because of the belief of my mom and dad, Late Mr.

Awadhesh K. Tripathi and Late Mrs. Archana Tripathi. Without their vision, endless sacrifices and constant hard-work, this achievement would have been impossible. My sister Dr. Nirotma Tripathi has always been there to encourage me when I doubted myself. I am truly thankful to my wife

Devangana Tripathi (Sharma) for being patient during last year of my Ph.D. She was a constant support during stressful days. I also thank my father and mother in law- Mr. Sanjay Sharma and

Mrs. Chhaya Sharma, and brother-in-law Akshay Sharma for providing a family away from home.

TABLE OF CONTENTS

LIST OF TABLES…………………………………………………………………………….....x LIST OF FIGURES…………………………………………………………………………….xii Chapter 1: Introduction and Background ...... 1 1.1 Introduction and Dissertation Scope ...... 1 1.2 Inspiration from Nature...... 3 1.3 Current Challenge ...... 5 1.4 Dissertation Scope: ...... 6 1.5 Background ...... 8 1.5.1 Biopolymers ...... 8 1.5.2 Lignocellulose Biomass ...... 8 1.5.3 Cellulose Esters ...... 10 1.6 Synthesis of Light-weight materials: ...... 13 1.6.1 Aerogels: Definition and Synthesis ...... 14 1.6.2 Polymer miscibility and phase separation to produce porous fibers...... 17 1.7 References ...... 30 Chapter 2: Isolating Lignin Rich Micro-Nano Fibrillar Cellulose (MNFC) from Residual Biomass ...... 42 2.1 Abstract ...... 42 2.2 Introduction ...... 42 2.3 Experimental Section ...... 44 2.3.1 Materials ...... 44 2.3.2 Autohydrolysis pretreatment ...... 45 2.3.3 Microemulsion pretreatment ...... 45 2.3.4 Fiber processing and defibrillation ...... 46 2.3.5 Thermal and chemical analyses ...... 46 2.3.6 Structural and molecular analyses ...... 47 2.4 Results and Discussion ...... 49 2.4.1 Isolation of lignin-rich micro-nano fibrillar cellulose (MNFC) ...... 49 2.4.2 Investigating chemical structure ...... 51 2.4.3 Structural characterization and lignin redistribution...... 53 2.4.4 Thermal analysis ...... 61 2.5 Conclusions ...... 67 2.6 References ...... 68 Chapter 3: Aerogel Synthesis from Non-aqueous Polymers ...... 72 3.1 Abstract: ...... 72 3.2 Introduction: ...... 72 3.2.1 In context to the dissertation ...... 72 3.2.2 Why Aerogels?...... 73 3.3 Experimental Section ...... 74 3.3.1 Materials ...... 74 3.3.2 Gelation of Cellulose esters ...... 75

vii

3.3.3 Solvent Exchange...... 75 3.3.4 Aerogel synthesis ...... 76 3.3.5 Aerogel Characterization ...... 77 3.4 Results and Discussion: ...... 80 3.4.1 Aerogel synthesis via Freeze Drying (FD) ...... 80 3.4.2 Aerogel synthesis via Critical Point Drying (CPD) ...... 88 3.5 Conclusions ...... 92 3.6 References: ...... 94 Chapter 4: Versatile Recipe for Aerogel Synthesis with Tunable Mechanical Properties .. 97 4.1 Abstract ...... 97 4.2 Introduction ...... 97 4.2.1 In context to the dissertation ...... 97 4.2.2 Need and Challenges for aerogels with tailorable mechanical properties ...... 98 4.2.3 Hypothesis to overcome the design challenge ...... 99 4.3 Experimental Section ...... 100 4.3.1 Materials ...... 100 4.3.2 Organogel synthesis ...... 101 4.3.3 Solvent exchange and swelling ...... 101 4.3.4 Aerogel synthesis ...... 102 4.3.5 Characterization ...... 102 4.4 Results and Discussion ...... 104 4.4.1 Gel swelling: ...... 104 4.4.2 Aerogel Synthesis and properties...... 109 4.4.3 Introducing Anisotropy in Aerogels ...... 116 4.4.4 Arresting the gels in a non-equilibrium state ...... 118 4.5 Conclusions ...... 121 4.6 References: ...... 122 Chapter 5: Aerogel Application in Oil-Spill Cleanup ...... 126 5.1 Abstract ...... 126 5.2 Introduction ...... 126 5.2.1 In context to the dissertation ...... 126 5.2.2 Current technologies for oil-spill remediation ...... 127 5.2.3 Aerogels for oil-spill cleanup ...... 127 5.3 Experimental Section ...... 129 5.3.1 Materials ...... 129 5.3.2 Aerogel modification ...... 129 5.3.3 Aerogel Performance tests ...... 130 5.3.4 Wicking and contact angle measurements ...... 131 5.3.5 Spin coating ...... 131 5.4 Results and Discussion: ...... 132 5.4.1 Aerogel Modification: ...... 132 5.4.2 Oil Separation from water-oil mixture:...... 135 5.4.3 Surface energy of solid surfaces ...... 138 5.4.4 Sorption & surface energy analysis for aerogels: ...... 144 5.5 Conclusions ...... 148

viii

5.6 References ...... 149 Chapter 6: Experimental and predictive description of the morphology and porosity of wet-spun fibers ...... 153 6.1 Abstract: ...... 153 6.2 Introduction ...... 153 6.2.1 In context to the dissertation ...... 153 6.2.2 Wet-spinning and porous fibers ...... 154 6.3 Developing design parameters: ...... 155 6.3.1 Hansen Solubility Parameters (HSPs): ...... 156 6.3.2 Kinetics of solvent-non-solvent inter-diffusion during phase separation: ...... 158 6.4 Experimental Section: ...... 165 6.4.1 Materials ...... 165 6.4.2 Experimental set-up ...... 165 6.4.3 Ternary Phase Diagram...... 165 6.4.4 Fiber Characterization:...... 166 6.5 Experimental Results and Discussion in relation to developed parameters: ...... 167 6.5.1 Choice of non-solvent ...... 168 6.5.2 Choice of the solvent ...... 171 6.5.3 Choice of the polymer ...... 174 6.5.4 Comparison with the published literature: ...... 176 6.6 Conclusion: ...... 178 6.7 References ...... 180 Chapter 7: Summary and Future Directions ...... 184 7.1 Chapter 1 ...... 184 7.2 Chapter 2 ...... 184 7.2.1 Future Directions: ...... 185 7.3 Chapter 3 ...... 186 7.3.1 Future directions ...... 186 7.4 Chapter 4 ...... 188 7.4.1 Future Directions ...... 189 7.5 Chapter 5 ...... 189 7.6 Chapter 6 ...... 190 7.6.1 Future Directions ...... 191 7.7 References ...... 192 APPENDICES ...... 193

LIST OF TABLES

Table 2.1: Chemical composition of coir and EFB fibers (P) and after sequential treatment via autohydrolysis (H) and microemulsion impregnation (M)...... 51

Table 2.2: XRD analysis parameters for full width half maximum (FWHM) for coir and EFB fibers at various stages of processing ...... 61

Table 2.3: The ratio of the fiber components as calculated from values from Table 1 along with thermal analysis data obtained from Figure 2.8...... 62

Table 3.1: Density values and aerogel pore volume ...... 83

Table 3.2: The reference values of the liquid CO2 used to calculate the Hansen solubility parameters from the Equations 3.4-3.6...... 91

Table 4.1: Hansen solubility parameters (HSP) corresponding to the polymer (cellulose acetate, CA) and the solvents (mixtures of water and CA).29 AVF is the Acetone Volume Fraction in water...... 105

Table 4.2: Mechanical properties of the CA aerogels ...... 116

Table 4.3: Properties of the aerogels obtained after solvent exchange for given time periods ...... 120

Table 5.1: Contact angle data & surface tension values for different probe liquids43 ...... 147

Table 5.2: Surface tension values as calculated from Fowkes’ model of surface energy ...... 148

Table 6.1: Hansen solubility parameters of selected Polymers and solvents in MPa1/2. 31,32...... 157

Table 6.2: Diffusivity values of solvent in non-solvent (DS-NS) and non-solvent in 34 solvent (DNS-S) as calculated from Wilke-Chang correlation...... 164

Table 6.3: List of all the systems investigated along with the major properties identified for the resulting fibers. CA= Cellulose Acetate, PS= Polystyrene. The MV and MC indicate macrovoids and microcellular structure respectively. Note that microcellular structure is also referred to as spongy structure in this study...... 168

Table 6.4: Our design parameters (RED and T) applied to selected studies of phase separated membranes and fibers...... 178

Table A2.1: Transmission peaks observed for coir and EFB samples along with the identified functional groups, chemical compounds and fiber component

consisting of the identified functional groups. (C = Cellulose, H = Hemicelluloses, L = Lignin)...... 195

Table A3.1: Comparison with other studies on aerogels for oil sorption reported in literature (ρ= Bulk density, P= Porosity,W=Water, NP-S= Non-polar solvent, R= Reusability (# of times sorbed and desorbed) ...... 200

Table A3.2: Material properties of the aerogels synthesized in current study and prior studies ...... 203

Table A4.1: The Hansen solubility parameters of cellulose acetate and the various solvents in which it is soluble or insoluble, as indicated. The values are taken from Hansen, C. M. Hansen Solubility Parameter- User Handbook. PhD Proposal 1, (2007) ...... 211

Table A5.1: Properties of the solvents used for sorption testing ...... 213

Table A6.1: Hansen solubility parameters for various polymers and solvents studied elsewhere.[1,2] ...... 218

LIST OF FIGURES

Figure 1.1: a) Ashby plot of specific values (normalized with density) of strength and stiffness (Young’s Modulus) for both natural and synthetic materials, b) the fracture toughness vs Young’s modulus plot of bone and nacre in comparison with a successful synthetic ceramic material (alumina/PMMA). Dashed arrows represent the change in values of individual components (closed circles) to homogeneous mixtures (open circles). The solid arrows indicate the change in toughness during crack growth (solid circles) after crack initiation (open circles). The figure is adapted from Wegst et al.10 ...... 3

Figure 1.2: Bamboo wood is composed of a hierarchal arrangement of cellulose microfibril into an aligned fibril bundle embedded in lignin and hemicellulose matrix that shapes into a honeycomb-like cellular structure. Figure adapted from Wegst et al.10 ...... 4

Figure 1.3: The scope of this Ph.D. dissertation is visually represented through the schematic of wood deconstruction. Chapter 2 investigates the structural and chemical arrangement of cellulose, lignin and hemicellulose in the biomass during the process of isolating micro/nano fibrils, whereas Chapter 3-6 explore the synthesis, properties, and application of light- weight materials (aerogels and fibers) from cellulose esters...... 7

Figure 1.4: Wood deconstruction scheme of a softwood tree illustrating the individual components (cellulose, lignin, and hemicellulose) that are assembled together to form the macro-structure of wood. Adapted from the Sorieul et al.29 ...... 9

Figure 1.5: Chemical formulae for cellulose esters such as Cellulose diacetate (CDA or CA), Cellulose Acetate Propionate (CAP) and Cellulose Acetate Butyrate (CAB). The structure of cellulose is represented by R=H...... 10

Figure 1.6: a) Silica aerogel used in Stardust mission to capture high-speed cosmic particles (Image taken in Smithsonian National Air and Space Museum, Washington DC), b) pressure vs temperature curve demonstrating freeze drying (FD) and critical point drying (CPD)...... 16

Figure 1.7: Illustration demonstrates the relaxation of tightly coiled polymer chains when mixed with an appropriate solvent. The yellow beads represent the monomer of a polymer and blue beads represent solvent molecules. Adapted from Dr. Ian Genzer’s lecture notes for introduction to polymer class at NC State University...... 18

Figure 1.8: Schematic demonstrating the concept of cohesive energy density used to form a relation for solubility parameters...... 20

xii

Figure 1.9: Visual representation of the Hansen solubility sphere along with the relative position of solvents on the three components axes. The values of the solvents on the axes are not drawn to scale...... 23

Figure 1.10: The graphical representation of extracting the temperature-composition phase diagram of a polymer-solvent system from the Gibbs free energy equation. The curve shows the upper critical solution temperature (UCST) at the intersection of binodal and spinodal curves...... 25

Figure 1.11: A representative ternary phase diagram for non-solvent induced phase separation.153 The three main regions: stable, metastable and unstable separated by binodal and spinodal curves are shown on the graph. The distance between the polymer-solvent axis and binodal curve is denoted as the Binodal Gap (BG). Three possible compositions paths for the polymer- solvent-non-solvent are represented on the graph based on the ratio of solvent outflow and non-solvent inflow during the phase separation. The major types of phase separated morphologies are also shown...... 27

Figure 2.1: The synthesis procedure for the micro-nano fibrillar cellulose from coir fiber (MNFC-coir) and the sample nomenclature is shown. The as received coir fiber (P-coir) was subjected to autohydrolysis and fiber refining (H- coir) followed by microemulsion pretreatment (M-coir) and microfluidization to produce MNFC-coir. The same procedure was carried on EFB fibers...... 50

Figure 2.2: FTIR spectra for coir and EFB at various stages of fiber processing highlighting major peaks...... 52

Figure 2.3: SEM images of coir (left column) and EFB (right column) fibers at each processing stage: untreated fibers (pristine, P) (a,b,c,d); fibers after hydrothermal H (e,f) and after microemulsion M treatments (g,h). Insets- (e,f,g,h) show high magnification images of individual fibers to indicate surface pits and to visualize the microfibril angle (denoted in the insets of e, f, g). Solid circles show the pits and vessel elements. Dotted circles highlight the small fibrils still attached to the fibers...... 56

Figure 2.4: SEM images of micro-nano fibrillar cellulose (MNFC) obtained after microfluidization of coir (top row) and EFB (bottom row). Two structural populations are presented in dotted circles. A higher magnification image of the nanofibers is shown on the left (a1,b1) and for microfiber is shown on the right (a2, b2). The solid circle in a) shows lignin deposition in form of nanoparticles...... 57

Figure 2.5: AFM images of MNFC- from coir and EFB fiber suspensions after centrifugation, demonstrating two structural population. The size scale is 3 µm. Many small particles can be observed in the images that are less than 1 µm in diameter, which are most likely coalesced lignin particles...... 58

xiii

Figure 2.6: Confocal microscopy image of coir and EFB fibers at each stage of fiber processing. The red circles indicate lignin droplets. The yellow circle indicates the vessel element...... 59

Figure 2.7: XRD diffractograms for Coir and EFB fibers at different treatment stages of processing...... 60

Figure 2.8: a,b) TGA thermographs for the coir and EFB fibers after each processing step, c,d) time derivative of TGA curves. Inset shows the bar graph for Full Width at Half Maximum (FWHM) for coir and EFB fibers...... 62

Figure 2.9: Schematic rendition of the changes illustrating coir fiber defibrillation at the given processing step, authohydrolysis, microemulsion treatment and microfluidization. The proposed mechanism is supported by confocal and SEM images. In the illustration, brown color represents lignin, blue represents hemicellulose and green represents cellulose. The fine green lines represent cellulose nano-fibrils that combine to give a micro-fibril. The micro-fibrils combine to give a macro-fibril or cellulose fiber bundle. The case of EFB fibers is expected to be similar...... 66

Figure 3.1: a) Leica CPD 300 used for critical point drying (CPD), b) Steps involved in the CPD procedure of an organogel ...... 77

Figure 3.2: Illustration of the three-step sequence of aerogel formation from a non- aqueous polymer via freeze-drying (FD)...... 80

Figure 3.3: a) Proposed cross-linking reaction between CDA chains and cross-linker, PMDA and the image showing the gelation of CDA solution into acetone- based organogel after the gel was set for 24 h, b) FTIR spectra of CDA flakes and aerogel...... 82

Figure 3.4: a) Comparison of measured and calculated density of organogels and hydrogels assuming that all solvent was replaced with air while keeping the size of the gel. The data for 2% hydrogel density is not reported because the high swelling of the respective organogel during solvent exchange made the hydrogel fragile for handling during measurements, b) image of the 2% aerogel (density 4.3 mg/ml) on a dandelion leaf, c) Compressive stress- strain profile for 4% aerogels with the maximum compressive stress of 350 kPa and maximum strain of 92%. It is to be noted that the desired shape of the 2% aerogel was difficult to synthesize due to breakage of the gel during the sequential solvent exchange process. Hence, compression testing of 2% aerogel was not performed. Along the same lines, due to the relatively high concentration of polymer in 6 and 8 % CDA gels, the final aerogel product for these gels have unavoidable artifacts during solvent exchange and freezing step. The stress-strain curve for the 6% and 8% aerogels are not reported here. However, we found that both the 6% and 8% aerogels consistently yielded below 30% strain (Figure A3.3). Inset SEM image

xiv

shows the radial cross-section of 4% CDA aerogel, and d) water uptake and kerosene oil uptake (g/g of aerogel) as exhibited by the different aerogels...... 86

Figure 3.5: SEM images of 4% a) cellulose acetate propionate (CAP) and, b) cellulose acetate butyrate (CAB) aerogel. Inset shows camera images of the respective aerogels. c) Water and oil uptake by the three types of aerogels. d) Liquid volume uptake normalized with the available pore volume per gram of the aerogel for all three aerogels...... 88

Figure 3.6: a) image of the precursor gel (left) and aerogel (right), b) density of aerogel as obtained from Critical Point Drying (CPD) compared to the aerogel obtained from Freeze Drying (FD), c) density of aerogel obtained from CPD after different cycles of solvent exchanges with liquid CO2. The red data points is the organogel density, d) RED value of cellulose acetate w.r.t liquid CO2 as a function of pressure...... 90

Figure 3.7: a) SEM image of the cross-section of 4% CDA aerogel synthesized via CPD (scale bar is 0.5 μm), c) N2 adsorption and desorption isotherm obtained for 4% CDA aerogel prepared from critical point drying. Inset shows the pore size distribution derived from desorption isotherm by Barrett-Joyner- Halenda (BJH) method...... 92

Figure 4.1: a) Relative energy difference (RED) between CA and the solvent blend of acetone and water. b) Turbidity measurement of the 4 and 8 wt% CA solutions in acetone/water solvent. The turbidity (NTU) value for 8 wt% CA in 0.75 AVF was out of the instrument limit, i.e., 10,000NTU. Images show solution of CA in solvent blends with of different AVFs, represented in the image. Note a drastic increase in turbidly in water-rich solvents, at AVFs of <90%...... 106

Figure 4.2: a) Swelling behavior of cross-linked organogels as a function of the acetone volume fraction (AVF) in the solvent. b) Equilibrium volumetric swelling ratio after exchange with water. c) Schematic illustration of the swelling of cross-linked CA gels (color code: black chains-CA polymer; red dash- cross-links; blue dots-water molecules; grey- acetone and blue- water). Hydrolysis of ester bonds is expected when the organogel comes in contact with water, resulting in irreversible swelling of the gel, termed here “hydrogel” (from the swelled organogel to the hydrogel, as indicated by the size of the image)...... 108

Figure 4.3: a) Illustration of unidirectional freezing of hydrogel. b) Image showing the unidirectional freezing arrangement. Inset shows the CA aerogel, c) SEM image of out-of-plane view (cross-section perpendicular to the freezing direction). Inset shows the magnified SEM image (scale: 100 μm) and, d) SEM image of in-plane view (cross-section in direction to the freezing direction)...... 112

Figure 4.4: a) Density and pore volume (%) of CA aerogels synthesized from hydrogels that were obtained after solvent exchange with different acetone volume fractions (AVF). b) Compressive stress-strain curve for the CA aerogels. Illustration in the inset shows the compression direction (out-of-plane compression). c) SEM images of the pore wall of the aerogels. The wall thickness of aerogels is corresponding to 1.3±0.3 µm (1A), 0.8±0.1 µm (0.9A), 0.8±0.1 µm (0.75A) and 40±12 µm (0.5A)...... 113

Figure 4.5: Compressive stress vs strain curves for aerogel 0.9A when compressed in- plane and out-of-plane. Inset shows out-of-plane SEM micrographs of the aerogel 0.9A when, uncompressed (top left), compressed out-of-plane (bottom left) and compressed in-plane (top right). Photos of aerogel samples are also shown...... 117

Figure 4.6: a) Kinetics of volumetric swelling ratio for organogel immersed in 0.9 AVF. Inset has swelling ratio normalized with maximum swelling observed at 48 h. b) Swelling ratio measured after immersing in water for 72 h. c) Density and pore volume of the aerogels synthesized after varying the solvent exchange time. d) Compressive stress vs strain curves for the aerogels obtained after varying solvent exchange time...... 119

Figure 5.1: Set-up used for modification of cellulose diacetate aerogel ...... 130

Figure 5.2: Images showing the water contact angle, a) water wicks through the surface of unmodified CDA aerogel, high water contact angle at b1) top surface, b2) transverse cross-section of modified aerogel, c) XPS survey scan for unmodified and modified aerogel (top & radial cross-section) ...... 133

Figure 5.3: a) The images show unmodified aerogel (left) and modified aerogel (right) on water surface. Within 2 hours, the unmodified aerogel sinks completely below the water surface, contrary to the modified aerogel which floats on the water surface even after 12 hours, b) water uptake rate by unmodified and modified aerogel for 48 hours and, c) water uptake by the modified aerogel just after silanation (t=0) and the same aerogel sample after 4 months. Inset images demonstrates high water contact angle on the modified aerogel surface even after 4 months...... 134

xvi

oil/water stirred system to model oil spill in a turbulent sea (motor oil: 7g, DI water: 100g, aerogel: 0.35g). The upper photographs shows a side view of the beaker used for the study prior to addition of the modified aerogel (“initial system”; upper left photograph), and an overhead view of the instant that the aerogel was added to the beaker (t = 0 sec; upper right photograph) ...... 137

Figure 5.5: Uptake of various liquids by modified aerogels. a) Setup of wicking experiments showing partially immersed cuboidal aerogel into the liquid of interest. The bottom face of the liquid is immersed 1.5 mm below the liquid surface. b) Liquid uptake rate for three oils...... 138

Figure 5.6: a) Traditional Zisman plot of cosine of contact angle vs surface tension of the probe liquid for low density polyethylene film.40 The intersection of the straight line with the X-axis where contact angle is zero is identified as the surface energy of the solid surface. b) The theoretical plot of rate of liquid uptake of 4 probe liquids with following relation of surface tension: 훾퐿4 < 훾퐿3 < 훾퐿2 < 훾퐿1. The modified Zisman theory used in this study relies on this plot to identify a range of surface tension of the solid surface, aerogel in this case...... 142

Figure 5.7: a) Liquid uptake as measured from tensiometer. The value in parentheses is the surface tension of corresponding liquids in air (mN/m), b) Graphical representation of experimental set up of tensiometer explaining the effect of buoyancy and surface tension on the weight recorded by instrument that results in negative value of liquid uptake. Image in inset demonstrates the set up for wicking, c) graph of mass of liquid sorbed vs the surface tension of the corresponding liquid shows a linear relationship, and d) volume of liquid sorbed relative to the available pore volume by the modified aerogel...... 145

Figure 6.1: a) Laboratory scale wet-spinning set up for the fiber synthesis, b) cross- section schematic of the polymer solution as soon as it comes in contact with the non-solvent. A fixed radius ‘R’ is assumed for the fiber and the skin thickness ‘W’ is varying as W(t) at t→ 0+. 퐶푁푆 is defined as the mass of non-solvent per unit volume of non-solvent. 퐶푁푆= 퐶푁푆 at r⟶∞ and 퐶푁푆 = 퐶푅(푡) at r=R. The assumed linear concentration profile of 퐶푁푆 within the skin layer is illustrated at small time scale, t⟶0+. The solvent concentration is 퐶푆= 퐶(푡) at r=R, and 퐶푆= 퐶0 at r= R-W...... 160

Figure 6.2: 10 w/v% CA in DMAC precipitated in water (top row) and methanol (bottom row). a1,a2) Fiber cross-section (scale: 50µm), b1,b2) fiber surface (scale: 100 µm), c1,c2) magnified fiber surface (scale: 10 µm). Inset shows further magnification of the fiber surface (scale: 1µm), d1,d2) magnified edge of fiber cross-section (scale: 5 µm), e1,e2) magnified fiber core (scale: 5 µm)...... 169

xvii

Figure 6.3: a) Ternary phase diagram for CA/DMAC/Non-solvent system showing the cloud point curve. Dashed arrows after the data points predict shape of the curve at higher CA concentration, b) illustration of fiber development in different non-solvents...... 171

Figure 6.4: 10 w/v% CA in DMSO precipitated in water (top row) and methanol (bottom row). a1, a2) Fiber cross-section (scale: 50µm), b1,b2) fiber surface (scale: 100 µm), c1,c2) magnified fiber surface (scale: 10 µm). Inset shows further magnification of the fiber surface (scale: 1µm), d1,d2) magnified edge of fiber cross-section (scale: 5 µm), e1,e2) magnified fiber core (scale: 5 µm)...... 172

Figure 6.5: A ternary phase diagram of PS/DMAC/Non-Solvent demonstrating the cloud point curves for non-solvents, water and methanol. The binodal gap (BG) with respect to each non-solvent is also shown on the diagram...... 174

Figure 6.6: 30 w/v% PS in DMAC precipitated in water (top row) and methanol (bottom row). a1,a2) Fiber cross-section (scale: 50µm), b1,b2) fiber surface (scale: 100 µm), c1,c2) magnified fiber surface (scale: 10 µm). Inset shows further magnification of the fiber surface (scale: 1µm), d1,d2) magnified edge of fiber cross-section (scale: 5 µm), e1,e2) magnified fiber core (scale: 5 µm)...... 176

Figure 7.1: a) Camera image of organogel of cellulose diacetate in acetone (left) and the corresponding aerogel (right) formed after solvent exchange with t- butanol followed by freeze drying, b) SEM image of the aerogel...... 188

Figure A2.1: FTIR spectra of pure cellulose with MNFC-Coir and MNFC-EFB ...... 194

Figure A2.2: The Coir and EFB fibers showing the vessel element...... 196

Figure A2.3: Cross-section of the untreated or pristine coir and EFB fibers showing the vessel element in solid circles ...... 196

Figure A3.1: N2 adsorption and desorption isotherm obtained for 4% CDA aerogel. Inset shows the pore size distribution derived from desorption curve of the isotherm by the Barrett-Joyner-Halenda (BJH) method...... 197

Figure A3.2: SEM image of the juction point at one pore in the aerogel. The wall thickness is ~1 μm...... 197

Figure A3.3: The compressive stress-strain curve for 4. 6, and 8% aerogels. The curves for 6 and 8% were not reproducible due to unavoidable artifacts generated during solvent exchange and freezing step. However, they consistently yielded below 30%...... 198

Figure A3.4: Liquid uptake w.r.t the available pore volume per gram of aerogel for 2, 4, 6 and 8 % aerogels. A ratio greater than 1 for water uptake by 4% and above xviii

aerogel indicates a combination of absorption in the pores and adsorption on the polymer walls may be due to hydrogen bonding of water with polymer chains. Hence the liquid uptake is referred to as sorption. The low value of sorption for 2% aerogel relative to the available pore volume is may be due to inability of 2% aerogel to retain such high volume of water. Relative volumetric uptake for kerosene oil is less than or equal to 1, which further implies entrapment of air bubbles during sorption...... 199

Figure A4.1: Frequency sweep demonstrating elastic modulus (G’) of the hydrogel obtained after solvent exchange for 4, 12 and 24 h. The organogels prepared for this measurement were 2.5cm in diameter. They were immersed in 0.9 AVF for 4, 12 and 24 h followed by water immersion for 72 h. The frequency sweep was run on a 2.5 cm diameter serrated steel flat plate geometry...... 206

Figure A4.2: Cryo-SEM image of the hydrogel obtained after solvent exchange with AVF =0.9, that shows (a,b) walls of CA in honey-comb pattern , and (c) the magnification of the area where sample destruction was prevented by rapid freezing. It exhibits a phase separated CA structure...... 207

Figure A4.3: SEM micrographs of the out-of-plane axis of aerogels when the acetone volume fraction in solvent exchange is varied. Inset shows the camera image of the bulk aerogels ...... 208

Figure A4.4: A cyclic compression of 0.9A aerogel when subjected to increasing strain of 25, 50 and 80%. The aerogel was kept under compression for 10 s before the release was initiated, as indicated by dwell time. The gap between the curves at maximum strain shows that the aerogel is slow to recover...... 209

Figure A4.5: SEM micrographs of the out-of-plane axis of aerogels when solvent exchange time is varied. Inset shows camera image of the bulk aerogels...... 210

Figure A5.1: a) The experimental set-up to test the modified aerogel performance for crude oil-spill in sea water, b) water and oil uptake from the mixture of crude oil and salt water...... 212

Figure A6.1: Normalized solvent concentration at the precipitation interface as a function of time. The graph is plotted from eq. 20...... 216

Figure A6.2: N2 adsorption-desorption curve for 10 w/v% CDA with DMAC as the solvent and non-solvent as a1) water, a2) methanol. The corresponding pore size distribution curve for non-solvent b1) water, and b2) methanol. The BET surface are lies in between 22- 25 m2/g and the pore size distribution is broad with the peak close to mesoporous limit of 50 nm...... 217

xix

Chapter 1: Introduction and Background

1.1 Introduction and Dissertation Scope

During the preliminary stages of human civilization, the technological advancement was supported by natural materials such as bone, wood, plants, and shells. As humanity became technologically more advanced, these materials were replaced by synthetic compounds, mainly oil-based materials that offered improved performance. However, the oil-based approach of material development is not sustainable as not only we are depleting non-renewable resources, but also polluting our environment.1,2 With the world heading towards an irreversible state of climate- change, we need to replace or aid our advanced technology towards more sustainable and efficient materials that have minimal carbon foot-print to reserve the limited non-renewable resources.

At present, there is a pressing need for novel light-weight materials that can aid or enhance technologies in a variety of fields, such as transportation, buildings materials, specialty garments and energy storage and conversion. To address this challenge, the future materials should offer unprecedented combinations of stiffness, strength, and toughness at low density and would need to be formed into bulk complex shapes and manufactured at high volume and low cost.3,4 A classical material design problem is the balance of two key structural properties- strength and toughness at a low bulk density of the material. Strong materials tend to be brittle, whereas tough materials are frequently weak.5 This is where Nature provides the source of inspiration for fresh ideas. Natural materials often consist of stiff and soft components in hierarchal porous architectures such as in the case of bone6, silk7, nacre8 or wood9. In many of these materials, a controlled transfer of load to the soft component act as a toughening mechanism.

Figure 1.1a shows the Ashby plot of specific properties (normalized by density) of strength vs stiffness for both natural and synthetic materials, the plot is adapted from Wegst et al.10 Many

engineered materials such as ceramic and metallic alloys have higher strength and stiffness than compared to natural materials, with an exception of silk fibers. The silk fibers have been reported to achieve high stiffness (10 GPa) and extraordinary toughness of 1000 MJ/m3 (discussed later)11, approaching that of Kevlar and carbon composite fibers. An initial look at the Ashby plot may suggest that natural materials are at the lower end of the performance spectrum, but unlike most synthetic materials, natural material use mechanically weak and inexpensive starting materials like proteins, polysaccharides, calcites etc. to form composites at ambient temperatures. The synergistic properties of such starting materials are greater than the sum of their parts when arranged in hierarchal interlocked and porous structure as also demonstrated in Figure 1.1b. Here, the dashed lines indicate the increase in properties of the natural composite material when the individual components (solid circles) are mixed, which represent an extraordinary increase in fracture toughness (open circle) that does not follow rule of mixing. Additionally, the toughness is much higher for crack growth (closed circle above solid arrows) than for crack initiation (open circle). As in the case of synthetic ceramic material (alumina/PMMA), while mimicking the nacre- like structure, a similar increase in toughness is achieved.12

1.2 Inspiration from Nature

Nature is filled with the examples of light-weight materials with high strength and toughness like spider silk, seashells, bone, enamel, wood and many others.13 These natural materials display high strength to weight ratio as a result of careful structuring of molecular entities into porous supramolecular structures, achieved through millions of years of evolution.10 Some well-known examples of evolution-driven strategies is a highly porous ordered architecture observed in cancellous bone, bamboo wood, and silk that effectively combines light-weight and stiffness. However, the stiff material tends to be brittle. To maintain strength, some natural

materials such as spider silk11 or wood9 incorporate nanofibers towards an intricate architectural design.

It is through optimized hierarchal architecture at each structure level, that bamboo wood, palm or coconut tree stems achieve unusually high efficiency in terms of mechanical performance per unit weight.14,15 Figure 1.2 illustrates this hierarchal architecture for bamboo wood10, which is composed of molecular cellulose assembling to form a cellulose microfibril. These aligned microfibrils are bundled together and embedded in a lignin-hemicellulose matrix shaped into hollow hexagonal cells of varying wall thickness. This intricate arrangement of nanomaterials

(cellulose) with natural chemicals (lignin and hemicellulose) generates a macroscopic honeycomb tubular structure that provides increased strength and toughness at reduced density.15 The bamboo, palm, and coconut have a more complex structure than wood, where a radial density gradient of microfibril in the honeycomb cell structure increases material flexural rigidity.14

Taking inspiration from such natural materials regarding their hierarchal architecture, we can save millions of years while developing strong and stiff yet tough materials, by not starting

from scratch. In fact, we can use some of these renewable natural materials like cellulose microfibrils, lignin or hemicellulose at each stage of their structural deconstruction and engineer them to mimic other biological materials, which may pave the way for a more sustainable future.

1.3 Current Challenge

The biggest challenge lies in not only developing a strong, stiff and tough material but also their scale-up from the lab scale in a cost-efficient manner.3 Observing our natural surroundings provides a mean of inspiration, however, mimicking the complex hierarchal structure of the natural materials is not trivial. Equally important is the development of bioinspired bio-based materials, i.e., the development of materials using benign polysaccharides, peptides, clays and natural ceramics, which is again not an easy task because they must compete with the performance of synthetic oil-based polymers.

Even though, many studies have probed the nano/microstructure of a wide variety of natural materials ranging from wood, bone, and teeth, to silk, fish scales, bird beaks, and shells13, very few have comprehensively characterized the most critical mechanical properties of these materials, such as strength and toughness and even fewer studies have identified the relevant nano/microscale mechanisms responsible for such properties. More importantly, there are even fewer examples of practical bioinspired materials and we are still a long way from replacing all synthetic polymers from practical applications. However, many industries are pledging towards sustainable growth. Recently, Ford announced the use of biopolymers in its automobile body parts leading to a total weight reduction by 10% without detrimental effect on its material properties.16

Additionally, the development of nacre8 and silk inspired17 materials towards practical use instills hope for highly damage-tolerant bioinspired bio-based structures.

1.4 Dissertation Scope

The research area of bioinspired and bio-based light-weight materials is an extremely broad topic. This dissertation focuses on 3 key aspects of this research area. First, is to understand the structural arrangement and chemical association of lignin, cellulose, and hemicellulose (Chapter

2) by delaminating the residual fibers from coconut and palm tree, and analyzing their morphological and thermo-chemical properties. Second, we mimic the cellular structure observed in such fibers and in other natural materials to develop novel synthesis routes for tough and strong aerogels from biopolymers (Chapter 3). We further explore the idea of tailoring the mechanical properties of such aerogels using the concepts of solubility parameters in Chapter 4. Additionally, in this chapter, we also perform ice templating to mimic close pore honeycomb-like structure that induces anisotropy in aerogels. In Chapter 5, we investigate one practical application for these bioinspired bio-based aerogels for oil-spill remediation, where the mechanical and oil-uptake performance of aerogels is tested in simulated marine conditions. Moreover, a theoretical framework is developed using modified Zisman theory and Fowkes’ theory to establish a criterion for separation of various non-polar fluids from water media. And lastly, the concepts of polymer thermodynamics and mass transport phenomenon is applied to biopolymers in the setting of non- solvent induced phase separation to develop mesoporous fibers that mimic the porous structures of cancellous bone (Chapter 6). The biopolymers studied in Chapter 3 to 6 are cellulose esters.

The flow of this dissertation can be visually illustrated on the delamination schematic of wood as explained in Figure 1.3.

Figure 1.3: The scope of this Ph.D. dissertation is visually represented through the schematic of wood deconstruction. Chapter 2 investigates the structural and chemical arrangement of cellulose, lignin and hemicellulose in the biomass during the process of isolating micro/nano fibrils, whereas Chapter 3-6 explore the synthesis, properties, and application of light-weight materials (aerogels and fibers) from cellulose esters.

1.5 Background

1.5.1 Biopolymers

The polymers produced by living organisms are called biopolymers. Just like synthetic polymers, the biopolymers consist of long chains made of covalently bonded repeating units. These repeating units can be nucleotides such as RNA or DNA producing polynucleotides, the repeating units can be amino acids or proteins called polypeptides, or the units can be monosaccharides generating polysaccharides.18 Apart from these classifications, there are other types of biopolymers such as rubber, melanin, suberin, and lignin. However, certain biopolymers like lignin do not have a well-defined structure, which makes them difficult to study, but a large scientific and industrial effort is underway for the valorization of lignin since 20-30% of the plant biomass is lignin that has low industrial value.19,20

1.5.2 Lignocellulose Biomass

Lignocellulosic biomass refers to plant dry matter produced by agricultural crops such as corn, grass, and sugarcane as well as woody forest-based resources such as hardwood and softwood trees. In addition, the lignocellulosic biomass can come from plant and fruit residue such as coconut shell or fibers and palm fruit residue known as empty fruit bunches.21–23 The terrestrial growth of lignocellulosic biomass is of the order magnitude hundreds of billions of tons per year and only a few percent of this biomass is utilized by humans.24 Lignocellulose consists of the heterogeneous distribution of three major components: cellulose, hemicellulose, and lignin.25

Figure 1.4 shows a simplified illustration of the deconstruction of a softwood tree. The macroscopic elements of the tree are made of wood or fiber cells that are closely packed by middle lamellae, which consist of 50-70% of lignin by weight.25 Thus, lignin acts as an adhesive between fibers. The fibers cell wall is composed of the primary cell wall (P) and the three layers of

secondary cell walls (S1, S2 and S3). The cell walls (P, S1, S2, and S3) differ in microfibril organization and the ratio of cellulose, lignin, and hemicellulose, which gives the different mechanical properties to the wood fiber.26 The crystalline cellulose microfibril bundles are held together with amorphous lignin with hemicellulose that acts as the binder. The primary components: cellulose, lignin, and hemicellulose are synthesized by the plant to produce fibers, vessel elements and tracheids that are responsible for fluid transport within the plant.27 It is this intricate arrangement of the primary components into a porous structure that makes wood one of the lightest yet toughest natural material.28 In Chapter 2, we explore this intricate interaction between cellulose, lignin, and hemicellulose by defibrillating the residual biomass obtained in form of coir fibers and empty fruit bunches (EFB). In the process, we also introduce a novel class of micro-nano materials that hold potential for the next class of light-weight composites from lignocellulose.

1.5.3 Cellulose Esters

After the chemical treatment and purification of lignocellulose, a high purity cellulose is obtained, that can be processed into useful forms such as textiles, filters, films and many others.30,31

The chemical structure of cellulose is shown in Figure 1.5. The cellulose has abundant functional groups with C-2, C-3 and C-6 carbon positions of the anhydroglucopyranose unit occupied by hydroxyl groups.32 The hydroxyl groups of cellulose can not only be substituted with various molecular species like acetyl, propyl, butyl and many others, but also these hydroxyl groups can be partially or fully substituted to produce different degrees of substitution (DS). The difference between cellulose acetate (CA or CDA), cellulose acetate propionate (CAP) and cellulose acetate butyrate (CAB) is the length of the substituted carbon chain that induces difference in hydrophilicity and solubility.33 The hydroxyl groups on cellulose chains govern many physical and chemical properties of this polymer and the value of DS provides useful insight into these properties.31,34,35

Cellulose and its derivatives exhibit intramolecular and intermolecular hydrogen bonding which gives rigidity to the molecule and affects their solubility and molecular configuration.35 The intramolecular hydrogen bonds are disrupted at a low degree of substitution (DS= 0.8-1) which makes cellulose acetate soluble in water. Increasing the DS (incorporation of hydrophobic acetate groups) decreases water solubility. However, at higher DS, cellulose acetate is soluble in organic solvents like acetone, tetrahydrofuran, dioxane.31,36 It has been reported that basic solvents such as acetone interact primarily with the hydroxyl group of cellulose acetate while acidic solvents like formic acid solvate acetyl groups.35,37 In Chapter 4, we explore the idea of solubility parameter

(discussed later) to tune the solubility of cellulose acetate in various solvent blends. a) Production and Uses:

Purified cellulose, free from lignin and hemicellulose, is mixed with anhydrous acetic acid and acetic anhydride (reactive agent) and a catalyst (sulfuric acid).38 The reaction mixture is aged for 20 h during which the acetylated cellulose chains dissolve into the reaction media to give a clear solution. The cellulose triacetate is precipitated with water as flakes. The cellulose triacetate is purified by dissolving in acetone and precipitating in water. Thereafter, the cellulose triacetate is hydrolyzed to give cellulose acetates with the desired degree of substitution.

Cellulose acetate is an attractive choice for a number of applications such as textiles manufacture, specialty paper, cigarette filters, and airplane coatings due to its toughness, flexibility, deep gloss and low toxicity34. The transparency and smoothness of cellulose acetate plastics have resulted in its uses in umbrellas, toothbrushes, spectacle frames and screwdriver handle39. Additionally, CAP and CAB are also used in inks and coatings.

While cellulose is abundant in nature, synthesis of cellulose and its derivatives is quite expensive40 and it cannot compete with petroleum-based products on the grounds of cost. Hence,

the current research on cellulose and its derivatives is focused on specialty products with high net value. Cellulose-based gels, aerogels, films, and fibers are studied for biosensors, drug delivery, bio-separations, optical devices, and environmental remediation.41–44 In this dissertation, our focus is on the aerogels and fibers of cellulose esters for environmental remediation. b) Biodegradability of cellulose esters:

The idea of using cellulose esters for environmental remediation raises a question if they are biodegradable or their use may create secondary contamination. This question is a currently under investigation by research groups and industries alike. While there are only a few degradation studies done of CAB and CAP, the degradation studies of CA reveal that the rate of degradation of CA in the natural environment depends on the degree of substitution (DS).45–48 It was found that the rate of biodegradation was reduced, but not inhibited by higher DS. The CA have been shown to undergo degradation in both aerobic46 and anaerobic conditions.49,50 Gu et al.51, demonstrated that both aerobic and anaerobic microorganisms are known to produce complete hydrolases, including esterases, which are necessary to degrade polysaccharides. In biologically active soil,

CA fibers are completely degraded in 4-9 months.52,53

The patent literature reveals that commercial products from cellulose acetate are modified to enhance biodegradation.54 The modifications include the use of cellulose chain-splitting enzymes, nitrogenous organic compounds or hydrolyzing the fiber surface to enhance degradation rate.55 Cannon et al.56, in their patent, disclosed the use of lower DS cellulose acetate for fiber spinning by using a blend of acetone and water that results in fibers with a high rate of biodegradation.

Recently, there has been significant research into the enzymes that degrade CA.

Cellobiohydrolases (a type of cellulase enzyme) can attack amorphous and crystalline cellulose

regions but are sensitive to acetyl groups.57 It has been shown that after initial deacetylation the backbone of cellulose acetate is biodegraded by cellulase enzyme.58,59 But, the presence of an esterase with cellobiohydrolases is reported to facilitate the cleaving of CA chains.60,61

Beside biodegradation, photodegradation is also an important route for CA degradation. It has been found that photodegradation is optimal with 280 nm or shorter wavelength UV-

62 irradiation. Additionally, the presence of photoactive compounds like TiO2 or benzophenone enhances the photodegradation of CA.63,64

While these studies reveal that CA-based materials have the potential for environmental degradation, further studies are needed in different environmental conditions and with different cellulose esters to confirm the biodegradation behavior of cellulose esters.

1.6 Synthesis of Light-weight materials

The desired porous structure of a light-weight material can be realized by either of the two classical approached in material science: “top-down” or “bottom-up” approach. The “top-down” approach involves carefully removing the material from the bulk to induce porosity, whereas, the

“bottom-up approach requires assembly of the molecular species to generate a porous structure.

The examples of “top-down” approaches can include aerogel formation65 and lithography.66 On the contrary, the “bottom-up” approach includes foam formation67, layer by layer printing (3d printing)68 and phase separation (precipitation or coagulation).69 The light-weight material synthesis can also involve a combination of “top-down” and “bottom-up” approaches, such as defibrillation of biomass to get nano-micro entities that could be assembled to get a novel light- weight material (Chapter 2). In addition, the dissertation explores one “top-down” approach in form of aerogels (Chapter 3-5) and one “bottom-up” approach in form of pore structure formation in fibers via phase separation (Chapter 6).

1.6.1 Aerogels: Definition and Synthesis

Aerogels, as the name suggests, are gels in which the solvent is replaced with air without collapsing the microstructure. Kistler, in 1931 prepared the first aerogel from silica gel by exchanging water to ethanol and heating the alcogel up to the critical point of ethanol to eliminate capillary forces70. In the late 1970s, the time consuming, tedious process of solvent exchange was eliminated by use of organic solvents during gel synthesis. Combined with the development of

71 72 supercritical CO2 for fluid drying made the process relatively fast and easy . Brock et al. , in

2004 reported a novel sol-gel process for aerogel synthesis which opened a new door of single metal aerogel synthesis. Aerogels gained massive media attention in 2006 when NASA’s Stardust mission returned successfully after capturing high-speed cosmic particles using silica aerogels

(Figure 1.6a)73. Many novel aerogels have been made after that including the carbon nanotube, graphene, cellulose, and nanocellulose74–76.

A crucial step in aerogel synthesis is the removal of the supporting fluid within the organogel without disrupting the network structure of the solid phase.77 Various techniques are employed to produce the final porous structure, including supercritical drying, ambient drying, and freeze-drying (lyophilization). Traditionally, materials produced by supercritical drying are called aerogels, whereas, materials produced by ambient drying and freeze-drying are frequently called xerogels and cryogels, respectively. However, materials with pore volume above 90%, synthesized by any of these processes are often termed as “aerogels”.77 For simplicity, we use the term

“aerogel” for all materials prepared in this dissertation.

The transition from a gel to aerogel requires solvent removal without causing structural collapse. The damage to microstructure can be avoided by minimizing the capillary force that occurs during solvent removal. The surface tension of the solvent exerts capillary pressure which is given by Equation 1.178 14

2훾푐표푠휙 푃 = ---Equation 1.1 푐푎푝푖푙푙푎푟푦 푅 where, γ = Surface tension of the liquid,

R = Radius of the capillary, and cos휙 = Angle liquid forms with the capillary surface.

The smaller the radius ‘R’, the larger the applied pressure. The liquid inside the very thin capillaries (on the order of microns) of the aerogel exerts inward force while evaporating, causing the structural collapse. To prevent this, surface tension must be eliminated which is generally achieved by two processes: freeze-drying (FD) and critical point drying (CPD). As shown in the pressure vs temperature curve in Figure 1.6b, a direct transition of liquid to gaseous state, which can cause structural collapse, can be avoided by either freezing the liquid and subjecting it to sublimation called FD or taking the solvent to a supercritical state and venting it out called CPD.

The critical point conditions for most of the solvents involve extreme temperature and pressure but with the development of supercritical CO2, the critical point drying can be done at relatively mild conditions of temperature (310C) and pressure (72 bar) by replacing the transition liquid with

71 liquid CO2. There have been reports of using ambient drying for “aerogel” synthesis, where a low surface energy solvent is introduced to reduce capillary forces during solvent evaporation.79,80

However, such “aerogels” have comparably high density and low pore volume.

The aerogels demonstrate a unique mechanical and physical properties, such as high pore volume, low density, high surface area, high strength to weight ratio, and impressive thermal & acoustic insulation.81–84 Carbon aerogels synthesized using carbon nanotube and graphene holds the world record for the lightest material with a density of 0.16mg/cm3 85. These properties make aerogels an attractive choice for applications such as insulation, transportation, catalysis, sorption agent, drug delivery, tissue engineering and even in space exploration73,86–89. Although research on aerogels is very old, yet not much work has been done on cellulose ester aerogels. Rigacci et al. 90,91, reported cellulose acetate aerogel synthesized via supercritical drying which had a high bulk density of 250-850 mg/cm3 and low porosity of 40-80%. A similar study by Tan et al. 92, reported a cellulose acetate aerogel via urethane linkage of cellulose acetate and cellulose butyrate followed by supercritical drying. This opens the immense possibility to study cellulose ester

aerogels, where we can synergistically combine unique properties of aerogels with inherent properties of cellulose esters to derive a novel material. The Chapters 3 and 4 of this dissertation are devoted to the investigation of extremely light (< 30 mg/cm3) and highly porous (>98%) cellulose ester aerogels with competitive mechanical properties. In Chapter 5, we explore the application of such materials for oil-spill cleanup.

1.6.2 Polymer miscibility and phase separation to produce porous fibers

The phase separation phenomenon is identified as the conversion of a single phase system into a multiphase system. Even though the concept of phase separation can be observed in simplest of forms such as separation of oil and water into two immiscible liquid phases, yet the concept of phase separation is a broad idea encompassing different scientific fields such as colloidal science93, biology94,95 and material science96,97. In polymer science, the phase separation technique is usually used to make porous solid materials.98–103 The concept of phase separation can be classified as the

“bottom-up” approach for producing light-weight porous materials as the dissolved polymer chains aggregate during phase separation to reveal a variety of porous architectures. Before understanding phase separation in the polymer solution, we must briefly introduce the thermodynamics and governing equations of the polymer miscibility. a) Polymer Miscibility

A polymer chain consists of the repeating units of a molecule. The random polymer coils are intertwined together which can be simply correlated at the macroscopic scale with the cooked spaghetti.104,105 In their solution state, the polymer chains are relaxed and are free to move around as illustrated in Figure 1.7 and the extent of polymer chain relaxation depends on the compatibility of a polymer with a particular solvent, where the polymer chains arrange to achieve the lowest energy state. In other words, the Gibbs free energy of polymer-solvent mixing is minimized: ΔG

= ΔH -TΔS ≤0. Here ΔH and ΔS are the enthalpy and entropy of mixing respectively, T is the absolute temperature in Kelvin.105

Paul Flory and Maurice Huggins through their separate works in 1941 came up with a neat

Equation (1.2) to define this ΔG for polymer solutions, which is called Flory-Huggins theory106–

109:

∆퐺푚푖푥 휙퐴 휙퐵 = 휙퐴휙퐵휒퐴퐵 + 푙푛휙퐴 + 푙푛휙퐵 ---Equation 1.2 푅푇 푁퐴 푁퐵

Where, 휙, 휌, 푁 are the volume fraction, density and the number of monomers of the solvent (A) and polymer (B), with the chi parameter ′휒′ to account for the energy of mixing of polymer and solvent molecules. The Equation 1.2 of the free energy of mixing constitutes 3 terms on the right- hand side, the 1st term represents a non-combinatorial contribution of the enthalpy of mixing, whereas the 2nd and 3rd terms represent the combinatorial entropy contribution which arises due to

an increase in the number of distinguishable states in the mixture relative to pure components. For polymer solutions, the entropy term for the polymer in Equation 1.2 (3rd term) is negligible due to the high molecular weight of the polymers. Therefore, the contribution to the ΔGmix is only from the enthalpy term (1st term) and the entropy term of the solvent (2nd term). For polymer blends, even the 2nd term is negligible. This necessitates the calculation of the enthalpy term to identify the polymer miscibility in a particular solvent or a polymer. To that end, the 휒 parameter in the enthalpy term is represented as a function of temperature (T), Equation 1.3110:

퐵 퐶 휒(푇) = 퐴 + + ---Equation 1.3 푇 푇2

In most cases, C=0 and a linearity is observed for 휒 with 1/T. However, for some polymer solutions, a non-linearity could be observed resulting in more complex phase diagrams.107,110 If

휒<0, polymer-solvent interaction is exothermic, otherwise it is endothermic. Although most polymer-solvent pairs exhibit endothermic mixing, there are some that exhibit exothermic mixing and it is not possible to correctly estimate 휒 when it is negative.111 While 휒 is generally found to be independent of ‘N’, linear or non-linear dependence on 휙 have been reported in certain instances, particularly for poor solvents.112,113 The 휒 parameter can be obtained either theoretically114 and experimentally from binodal and spinodal curves115, or it can also be predicted from the solubility parameters (explained later).

The Flory Huggins theory for polymer miscibility is the oldest theory that is still actively used but it has certain limitations such as it is based on the lattice model that uses various approximations in the counting process. The theory does not account for different shapes and sizes of the monomers and solvent molecules. It also ignores the free volume of polymer and assumes perfectly incompressible systems. The most important limitation is that it only applies to non-polar

molecules whereas most of the biopolymers exhibit polar and hydrogen bonding interactions. The reader is referred to Introduction of Polymer by Young and Lovell for further reading.105 b) Solubility Parameters:

Solubility parameters serve as an important tool for correlating and predicting cohesive and adhesive properties of the material from a knowledge of only the properties of components. In particular, for polymers, solubility parameters can be used to find the compatible solvent or polymer for blends.116,117

(i) Hildebrand Solubility Parameters:

Unlike 휒 values, that are pairwise-specific, solubility parameters (δ) are pure component values. The solubility parameters provide a measure of energy needed to break the interaction of a species to permit the interaction with a different species as illustrated in Figure 1.8. The physiochemical properties of a system can be identified by its cohesive energy density (E/V), where ‘E’ is the energy of vaporization and ‘V’ is the molar volume of the material. The square root of the cohesive energy density of the material is defined as Hildebrand solubility parameter

(δ)118–120.

Figure 1.8: Schematic demonstrating the concept of cohesive energy density used to form a relation for solubility parameters.

Patterson and co-workers predicted the 휒 parameter using the corresponding states theory

(CST), as well as using Hildebrand solubility parameters in non-polar systems.121,122 The equation to calculate 휒 from Hildebrand solubility parameter for a non-polar polymer-solvent system is defined by Equation 1.4119,120

푉 휒 = 0.34 + 푟 (훿 − 훿 )2 ---Equation 1.4 푅푇 퐴 퐵 where, Vr is the molar volume, R is gas constant and T is the absolute temperature in K. The unit of 휒 is MPa1/2. The constant of 0.34 is generally accepted as a necessary constant for polymer systems as a correction to Flory-Huggins combinatorial entropy.123 Some studies have also suggested the use of constant 0.34 to align the values obtained from CST and Hildebrand solubility parameter. The CST predicts 휒 about 0.3 units larger than the Hildebrand solubility parameter.124

The values of δ can be found experimentally or can be estimated via the use of group

125,126 contribution method. Generally, if δA = δB (or 휒<0.5), the polymer mixes well in the solvent.127 Even though, the Hildebrand solubility parameter is still used to predict polymer solvency, the parameter was developed only for non-polar and non-associating system and the parameter does not account for polar or hydrogen bonding interactions as exhibited by most of the biopolymers.119,120 The Hildebrand solubility parameter, however, has been extended to include all kind of systems. Van Arkel128, Small,129 and Prausnitz, and co-workers123,130 dissociated the solubility parameter into polar and non-polar components. Although, this neglects the induction interactions (dipole-dipole). Charles Hansen in 1967 proposed a practical extension of the

Hildebrand solubility parameter, that includes polar, dispersive and hydrogen bonding interaction components, which are popularly termed as Hansen solubility parameters.131

(ii) Hansen Solubility Parameters (HSPs)

Hansen further improved upon Hildebrand solubility theory by dissociating the cohesive energy density into three components: the dispersion (δd), the polar (δp) and the hydrogen bonding

131 (δh) that arise, respectively, from van der Waals, dipole and hydrogen bonding interactions.

These three components, the Hansen solubility parameters (HSP), can be factored in or calculated as Ra, Equation 1.5:

2 2 2 2 2 2 Ra = 4(( d1 −  d 2 )+ ( p1 −  p2 )+ ( p1 −  p2 )) ---Equation 1.5 where, Ra is the difference between the HSP of a solvent (1) and a polymer (2). The constant 4 is from an empirical correlation. The solubility is maintained if Ra < R0, defined as the “interaction radius” of the polymer, which is measured experimentally. A Relative Energy Difference (RED) is defined as Ra/R0. A value of RED < 1 implies good solubility for the polymer in the given solvent. The value of interaction radius R0 for a polymer is usually identified experimentally. The solubility of the polymer is tested with various solvents of known solubility parameters (δd, δp, δH) and the R0 is identified based on the polymer solubility in the corresponding solvent. Similar to

Hildebrand solubility parameter, the HSPs can be found experimentally or can be calculated by group contribution method.126,132–134

The HSPs can be visually represented on a plot with a set of three mutually perpendicular axes represented by dispersion, polar, and hydrogen bonding parameters respectively as illustrated in Figure 1.9. Within this “Hansen space”, the interaction radius of a polymer (R0) can be represented in form of a sphere radius. The center of the sphere is the HSP values (δd, δp, δH) of the polymer of interest. As represented on Figure 1.9, the solvent 1 lies outside the interaction radius of the polymer (RED>1), the solvent 1 should not be compatible with the polymer. On the

contrary, solvent 2, lies within the interaction radius of the polymer (RED<1), therefore, solvent

2 is expected to fully dissolve the polymer.

The application of HSP approach is a relatively simple that accounts for polar and hydrogen bonding interactions, which are very common in the polymers of interest. In addition, the HSP approach affords prediction capability in multicomponent systems131. However, there are still exceptions for the polymer-solvent systems that do not follow the HSP rule, which may be due to some limitation of this approach, such as the shape and size of the molecules is not accounted, that plays a significant role in the rate of solvency.131 Even then, it is a practical model and with a large amount of database available for the solubility parameters, it is easy to apply.135

It is noted that recently Hansen solubility parameters have been expanded to investigate particle dispersions such as carbon nanotubes136, cellulose nanocrystals137, clays,138 and others.

However, in these cases modification is needed for the model to be successfully implemented it these and related systems. Therefore, we have limited our study to fully solubilized polymers,

which are appropriate in the framework of the conventional HSP model. In Chapter 4, we use

HSPs to tune the swelling behavior of cellulose acetate gels and consequently modify the mechanical properties of corresponding aerogels. In Chapter 6, we again use HSP, but to predict the non-solvency of a polymer. Coupled with the mass transfer model, we predict the fiber structure formed during the phase separation process of wet-spinning. c) Phase separation

For the two-component polymer-solvent system, the equilibrium condition is obtained by

휕∆퐺 equating the chemical potential (Δ휇 = ) of the polymer to zero as demonstrated in Figure 휕∅퐵

1.10.139 This gives the binodal curve separating the single phase and phase separated system. By equating the derivative of chemical potential with respect to the polymer volume fraction, to zero generates the spinodal curve separating the phase separated region into metastable and unstable regions. Finally, the critical point at the intersection of binodal and spinodal curve is obtained by the double derivative of chemical potential with respect to polymer volume fraction.110 Figure

1.10 demonstrates upper critical solution temperature (UCST), which means that polymer solution exists in single phase above UCST and phase separates below that temperature. Alternatively, some polymeric systems exhibit lower critical solution temperature (LCST) that implies that the polymer-solvent exists in single phase below LCST. This phase diagram alludes to a very important phenomenon in phase separation, thermal induced phase separation (TIPS) that is used to create porous materials.

Another way to induce polymer phase separation involves the addition of vapors (called vapor induced phase separation, VIPS) or liquid bath (non-solvent induced phase separation,

NIPS).98 In both these cases, the third component behaves as a non-solvent for the polymer and the porous solid structure is developed during the process of phase separation. In all the cases of phase separation (TIPS, VIPS or NIPS), the solvent or non-solvent can be removed from the porous solid structure by evaporation, freeze-drying or supercritical drying.101,140

(i) Non-solvent induced phase separation (NIPS):

Since the invention of an asymmetric membrane by Loeb and Sourirajan141, a lot of research has been conducted on NIPS or immersion precipitation that lead to many commercially available filter membranes. The NIPS or immersion precipitation or bath coagulation is more complex 3-component phase transition phenomenon than TIPS. In its simplest form, a film of the concentrated polymer solution is immersed in a non-solvent bath leading to precipitation or phase separation of the polymer. The principle structure formation occurs during the immersion step which is accompanied by the multicomponent mass transfer of solvent-nonsolvent leading to concentration fluctuations in the system and hence, leading to various phase transitions. Naturally, the mathematical models need to address both the thermodynamic and kinetic aspect of the system.

The pioneering work by Pouchly et al.142, to develop the thermodynamic model for the ternary system from Flory-Huggins equations was further refined by Altena and Smolders143 and Yilmaz and McHugh144, who developed the complete set of equations for the binodal and spinodal curves, as well as tie line slopes and critical points. Cohen et al.145, were the first ones to calculate the diffusion path using the ternary phase diagram, combining the thermodynamic and kinetic aspect of ternary phase separation. The model was further refined by Reuvers146,147, Tsay148,

Radovanovic,149,150 and Cheng151 These models were based on the diffusion and continuity equations combined with the thermodynamics of irreversible processes.

Conventionally, ternary phase diagrams like the one shown in Figure 1.11, are used to explain the thermodynamic aspect of non-solvent induced phase separation. A ternary phase diagram of polymer-solvent-non-solvent shows three regions: stable, unstable and metastable, which are divided by binodal and spinodal curves.152,153 In the stable region, all three components

(polymer, solvent, and non-solvent) co-exist. Contrarily, the region covered by the spinodal curve is unstable and the region between the binodal and spinodal curve is metastable. Understanding 26

the phase transitions of a system allows predicting the polymer precipitation mechanism, where liquid-liquid demixing plays a vital role in non-solvent induced phase separation and the demixing starts as soon as the system composition crosses binodal curve. It is well known that the polymer phase separates via nucleation and growth in the metastable region, whereas spinodal decomposition occurs in the unstable region. Nucleation and growth in the polymer poor phase in the upper metastable region (high polymer concentration) results in a porous morphology. On the contrary, nucleation and growth in polymer rich phase in the lower metastable region (low polymer concentration) gives polymer powders or mechanically weak structures. The spinodal decomposition typically produces open-cell or interconnected network structure.

Figure 1.11: A representative ternary phase diagram for non-solvent induced phase separation.153 The three main regions: stable, metastable and unstable separated by binodal and spinodal curves are shown on the graph. The distance between the polymer-solvent axis and binodal curve is denoted as the Binodal Gap (BG). Three possible compositions paths for the polymer-solvent-non-solvent are represented on the graph based on the ratio of solvent outflow and non-solvent inflow during the phase separation. The major types of phase separated morphologies are also shown.

The relative position of the binodal curve on the ternary phase diagram is a function of interaction among polymer, solvent, and non-solvent. The distance between the binodal curve and polymer-solvent axis (Figure 1.11) is defined as Binodal Gap (BG), which can be directly correlated to the tendency of liquid-liquid de-mixing. Smolders et al.154,155, introduced the concept of delayed and instantaneous de-mixing to explain morphology in the phase separated membranes.

In delayed de-mixing, the polymer composition just beneath the membrane surface remains in the stable region for a long time (few seconds to minutes), implying that no de-mixing occurs immediately after immersion. After a long time, the polymer composition beneath the surface crosses the binodal curve, as shown by the composition path 1 on Figure 1.11. On the contrary, during instantaneous de-mixing, the polymer composition may cross the binodal curve instantly

(less than a second), implying that liquid-liquid de-mixing will start immediately after immersion, as shown by the composition path 2 on Figure 1.11. During instantaneous de-mixing, the polymer composition may directly go into the unstable region (path 3) if it crosses the critical point. It is evident from the ternary phase diagram that a system with large BG may exhibit delayed de-mixing and a system with small BG may undergo instantaneous de-mixing. The composition path taken by the system also depends on the diffusion kinetics of the solvent and non-solvent that is qualitatively related to the ratio of solvent outflow to non-solvent inflow, defined as k. A system with k>1 may undergo delayed de-mixing, whereas a system with k<1 may undergo instantaneous de-mixing, resulting in two fundamentally different morphologies.153 A delayed de-mixing produces membranes with sponge-like morphologies, whereas, instantaneous de-mixing produces macrovoids.

The ternary phase diagram is usually used to explain phase-separated systems. As discussed, such diagram can be developed theoretically or experimentally, but it is a tedious

process and even then, it gives a qualitative understanding of the bulk morphology.156–158

Furthermore, the complexity involved in the theoretical development of a model in non-solvent induced phase separation process entails development of a simpler predictive model that is quick to implement. In Chapter 6, we focus on developing two parameters, one based on Hansen solubility parameters (RED) and the other based on the time constant for solvent diffusion (T), that can rationalize the choice of solvent and non-solvent for a polymer of interest. Additionally, these parameters can allow to predict and investigate the fiber structure.

1.7 References

[1] ITOPF. Oil tanker spill statistics 2011. 2015.

[2] Burton NHK, Musgrove a. J, Rehfisch MM, Clark N a. Birds of the Severn Estuary and Bristol Channel: Their current status and key environmental issues. Mar Pollut Bull 2010;61:115–23. doi:10.1016/j.marpolbul.2009.12.018.

[3] Wegst UGK, Ashby MF. The mechanical efficiency of natural materials. Philos Mag 2004;84:2167–81. doi:10.1080/14786430410001680935.

[4] Gibson LJ, Ashby MF. Cellular Solids: Structure and Properties. Cell. Solids Struct. Prop. 2nd ed., Cambridge University Press, Cambridge, UK; 1997.

[5] Ritchie RO. The conflicts between strength and toughness. Nat Mater 2011;10:817–22. doi:10.1038/nmat3115.

[6] Meyers MA, McKittrick J, Chen P-Y. Structural biological materials: critical mechanics- materials connections. Science 2013;339:773–9. doi:10.1126/science.1220854.

[7] Liu X, Zhang K-Q. Silk Fiber — Molecular Formation Mechanism, Structure- Property Relationship and Advanced Applications. InTech Open, 2014, p. 69–102. doi:dx.doi.org/10.5772/57611.

[8] Walther A, Bjurhager I, Malho JM, Pere J, Ruokolainen J, Berglund LA, et al. Large-area, lightweight and thick biomimetic composites with superior material properties via fast, economic, and green pathways. Nano Lett 2010;10:2742–8. doi:10.1021/nl1003224.

[9] Berglund LA, Burgert I. Bioinspired Wood Nanotechnology for Functional Materials. Adv Mater 2018;30:1–15. doi:10.1002/adma.201704285.

[10] Wegst UG., Bai H, Saiz E, Tomsia AP, Ritchie RO. Bioinspired structural materials. Nat Mater 2014;14:23–6. doi:10.1038/NMAT4089.

[11] Keten S, Xu Z, Ihle B, Buehler MJ. Nanoconfinement controls stiffness, strength and mechanical toughness of Β-sheet crystals in silk. Nat Mater 2010;9:359–67. doi:10.1038/nmat2704.

[12] Bouville F, Maire E, Meille S, Van de Moortele B, Stevenson AJ, Deville S. Strong, tough and stiff bioinspired ceramics from brittle constitutents. Nat Mater 2014;13:5018–514.

[13] Fratzl P, Weinkamer R. Nature’s hierarchical materials. Prog Mater Sci 2007;52:1263–334. doi:10.1016/j.pmatsci.2007.06.001.

[14] Wegst UGK. Bending efficiency through property gradients in bamboo, palm, and wood- based composites. J Mech Behav Biomed Mater 2011;4:744–55. doi:10.1016/j.jmbbm.2011.02.013.

[15] Wegst UGK, Ashby MF. The structural efficiency of orthotropic stalks, stems and tubes. J Mater Sci 2007;42:9005–14. doi:10.1007/s10853-007-1936-8.

[16] Polkowski R, Kiziltas A, Ueki M. Mechanical Properties of Biorenewable Blends of Polyamide 10,10 and Polyamide 6,10. SAE Tech Pap 2017;01:5. doi:https://doi.org/10.4271/2017-01-0490.

[17] Mittal N, Ansari F, Gowda.V K, Brouzet C, Chen P, Larsson PT, et al. Multiscale Control of Nanocellulose Assembly: Transferring Remarkable Nanoscale Fibril Mechanics to Macroscale Fibers. ACS Nano 2018:acsnano.8b01084. doi:10.1021/acsnano.8b01084.

[18] Mohanty AK, Misra M, Drzal LT. Natural Fibers, Biopolymer, and Biocomposites. CRC Press, Inc.; 2005.

[19] Ragnar M, Henriksson G, Lindström ME, Wimby M, Blechschmidt J, Heinemann S. “Pulp and Paper”: Ullmann’s Encyclopedia of Industrial Chemistry. Weinheim: Wiley-VCH; 2014.

[20] Vanholme R, Demedts B, Morreel K, Ralph J, Boerjan W. Lignin Biosynthesis and Structure. Plant Physiol 2010;153:895–905. doi:10.1104/pp.110.155119.

[21] Dimmel DR, Bozell JJ, Power AJ. Pulping catalysts from lignin: Industrial uses of agricultural materials. Situat Outlook Rep 1994:27–33.

[22] Ferrer A, Vega A, Ligero P, Rodríguez A. Pulping of empty fruit bunches (EFB) from the palm oil industry by formic acid. BioResources 2011;6:4282–301.

[23] Tripathi A, Ferrer A, Khan SA, Rojas OJ. Morphological and Thermochemical Changes upon Autohydrolysis and Microemulsion Treatments of Coir and Empty Fruit Bunch Residual Biomass to Isolate Lignin-Rich Micro- and Nanofibrillar Cellulose. ACS Sustain Chem Eng 2017. doi:10.1021/acssuschemeng.6b02838.

[24] Lucia LA. Lignocellulosic biomass: a potential feedstock to replace petroleum. BioResources 2008;3:981–2.

[25] Sjostrom E. Wood Chemistry: Fundamentals and Applications 2nd ed. Academic Press, New York; 1993.

[26] Hatfield RD, Ralph J, Grabber JH. Cell Wall Structural Foundations: Molecular Basis for Improving Forage Digestibilities. Crop Sci 1999;39:27–37.

[27] Gibson LJ. Inside Plants: An Engineer’s View of the Arnold Arboretum n.d.

[28] Song J, Chen C, Zhu S, Zhu M, Dai J, Ray U, et al. Processing bulk natural wood into a high-performance structural material. Nature 2018;554:224–8. doi:10.1038/nature25476.

[29] Sorieul M, Dickson A, Hill JS, Pearson H. Plant Fibre: Molecular Structure and Biomechanical Properties, of a Complex Living Material, Influencing its Deconstruction

towards a Biobased Composite. Materials (Basel) 2016;9:618.

[30] Wang S, Lu A, Zhang L. Recent advances in regenerated cellulose materials. Prog Polym Sci 2016;53:169–206. doi:10.1016/j.progpolymsci.2015.07.003.

[31] Fischer S, Thümmler K, Volkert B, Hettrich K, Schmidt I, Fischer K. Properties and Applications of Cellulose Acetate. Macromol Symp 2008;262:89–96. doi:10.1002/masy.200850210.

[32] Hendriks ATWM. Pretreatments to enhance the digestibility of lignocellulosic biomass. Bioresour Technol 2008;100:10–20.

[33] Mukherjee T, Kao N. PLA Based Biopolymer Reinforced with Natural Fibre: A Review. J Polym Environ 2011;19:714–25. doi:10.1007/s10924-011-0320-6.

[34] Zugenmaier P. Characteristics of cellulose acetates— Characterization and physical properties of cellulose acetates. Macromol Symp 2004;208:81–166. doi:10.1002/masy.200450407.

[35] Kamide K, Saito M. Effect of total degree of substitution on molecular parameters of cellulose acetate 1984;20:903–14.

[36] Bochek AM, Kalyuzhnaya LM. Interaction of water with cellulose and cellulose acetates as influenced by the hydrogen bond system and hydrophilic-hydrophobic balance of the macromolecules. Russ J Appl Chem 2002;75:989–93. doi:10.1023/A:1020309418203.

[37] Pintarić B, Rogošić M, Jasna Mencer H. Dilute solution properties of cellulose diacetate in mixed solvents. J Mol Liq 2000;85:331–50. doi:10.1016/S0167-7322(00)89017-0.

[38] Malm CJ, Tanghe LJ, Laird BC. Preparation of Cellulose Acetate - Action of Sulfuric Acid. Ind Eng Chem 1946;38:77–82. doi:10.1021/ie50433a033.

[39] Rustemeyer P. History of CA and evolution of the markets. Macromol Symp 2004;208:1– 6. doi:10.1002/masy.200450401.

[40] Schurz J. Trends in Polymer Science: A bright future for cellulose. Prog Polym Sci 1999;24:481–3.

[41] Qiu X, Hu S. “Smart” Materials Based on Cellulose: A Review of the Preparations, Properties, and Applications. Materials (Basel) 2013;6:738–81. doi:10.3390/ma6030738.

[42] Konwarh R, Karak N, Misra M. Electrospun cellulose acetate nanofibers: the present status and gamut of biotechnological applications. Biotechnol Adv 2013;31:421–37. doi:10.1016/j.biotechadv.2013.01.002.

[43] Chang C, Zhang L. Cellulose-based hydrogels: Present status and application prospects. Carbohydr Polym 2011;84:40–53. doi:10.1016/j.carbpol.2010.12.023.

[44] Kettunen M, Silvennoinen RJ, Houbenov N, Nykänen A, Ruokolainen J, Sainio J, et al. Photoswitchable superabsorbency based on nanocellulose aerogels. Adv Funct Mater 2011;21:510–7. doi:10.1002/adfm.201001431.

[45] Komarek RJ, Gardner RM, Buchanan CM, Gedon S. Biodegradation of radiolabeled cellulose acetate and cellulose propionate. Jounral Appl Polym Sci 1993;50:1739–46.

[46] Buchanan CM, Gardner RM, Komarek RJ. Aerobic biodegradation of cellulose acetate. Jounral Appl Polym Sci 1993;47:1709–19.

[47] Zee M va. der, Stoutjesdijk J., Feijen J. Relevance of aquatic biodegradation tests for predicting degradation of polymeric materials during biological solid waste treatment. Chemosphere1 1998;36:461–73.

[48] Puls J, Wilson S a., Hölter D. Degradation of Cellulose Acetate-Based Materials: A Review. J Polym Environ 2011;19:152–65. doi:10.1007/s10924-010-0258-0.

[49] Rivard CJ, Adney WS, Himmel ME, Mitchell DJ, Vinzant TB, Grohmann K, et al. Effects of natural polymer acetylation on the anaerobic bioconversion to methane and carbon dioxide. Appl Biochem Biotechnol 1992;34:725–36.

[50] Gartiser S, Wallrabenstein M, Stiene G. Assessment of several test methods for the determination of the anaerobic biodegradability of polymers. J Environ Polym Degrad 1998;6:159–73.

[51] Gu J-D, Eberiel DT, McCarthy SP, Gross RA. Cellulose acetate biodegradability upon exposure to simulated aerobic composting and anaerobic bioreactor environments. J Environ Polym Degrad 1993;1:143–53.

[52] Northrop DM, Rowe WF. Effect of the Soil Environment on the Biodeterioration of the Man-Made Textiles. Plenum Press, New York; 1987. doi:doi.org/10.1007/978-1-4613- 0949-9_3.

[53] Gu J-D, Yang S, Welton R, Eberiel DT, McCarthy SP, Gross RA. Effect of environmental parameters on the degradability of polymer films in laboratory-scale composting reactors. J Environ Polym Degrad 1994;2:129–35.

[54] Willmund R. Filter Tow, method for the production thereof, as well as tobacco smoke filter element and method for its production. 5,427,852, 1995.

[55] Teufel E, Willmund R. Filter tow and method for its manufacture as well as tobacco smoke filter element and method for its manufacture. US5427852A, 1995.

[56] Cannon JN. Process for making cellulose acetate fibers. US5512230A, 1996.

[57] Knowles JK., Lentovaara P, Murray M, Sinnott ML. Stereochemical course of the action of the cellobioside hydrolases I and II of Trichoderma reesei. J Chem Soc Chem Commun 1988;0:1401–2.

[58] Altaner C, Saake B, Tenkanen M, Eyzaguirre J, Faulds CB, Biely P, et al. Regioselective deacetylation of cellulose acetates by acetyl xylan esterases of different CE-families. J Biotechnol 2003;105:95–104.

[59] Altaner C, Saake B, Puls J. Specificity of an Aspergillus Niger Esterase Deacetylating Cellulose Acetate. Cellulose 2003;10:85–95.

[60] Sakai K, Yamauchi T, Nakasu F, Ohe T. Biodegradation of Cellulose Acetate by Neisseria sicca. Biosci Biotechnol Biochem 1996;60:1617–22.

[61] Moriyoshi K, Yamanaka H, Ohmoto T, Ohe T, Sakai K. Mode of Action on Deacetylation of Acetylated Methyl Glycoside by Cellulose Acetate Esterase from Neisseria sicca SB. Biosci Biotechnol Biochem 2005;69:1292–9.

[62] Hon N-S. Photodegradation of cellulose acetate fibers. J Polym Sci Polym Chem 1977;15:725–44.

[63] Hosono K, Kanazawa A, Mori H, Endo T. Photodegradation of cellulose acetate film in the presence of benzophenone as a photosensitizer. J Appl Polym Sci 2007;105:3235–9.

[64] Egerton GS, Shah KM. The Effect of Temperature on the Photochemical Degradation of Textile Materials Part I : Degradation Sensitized by Titanium Dioxide. Text Res J 1968;38:130–5.

[65] Aegerter M, Leventis N, Koebel M. Aerogels handbook (Advances in Sol-Gel Derived Materials and Technologies). 2011.

[66] Xia Y, Whitesides G. Soft lithography. Annu Rev Mater Sci 1998;28:153–84. doi:10.1146/annurev.matsci.28.1.153.

[67] Schramm L. Emulsions, Foams, and Suspensions: Fundamentals and Applications. 1st ed. Wiley-VCH, Weinham, Germany; 2005.

[68] Stansbury JW, Idacavage MJ. 3D printing with polymers: Challenges among expanding options and opportunities. Dent Mater 2016;32:54–64. doi:10.1016/j.dental.2015.09.018.

[69] Kadla JF, Korehei R. Tuning the morphology of cellulose acetate gels by manipulating the mechanism of phase separation. Biomacromolecules 2011;12:43–9. doi:10.1021/bm100917j.

[70] Kistler SS. Coherent Expanded-Aerogels. J Phys Chem 1931;36:52–64. doi:10.1021/j150331a003.

[71] Tewari PH, Hunt AJ, Lofftus KD. Ambient-temperature supercritical drying of transparent silica aerogels. Mater Lett 1985;3:363–7. doi:10.1016/0167-577X(85)90077-1.

[72] Mohanan JL, Brock SL. A new addition to the aerogel community: unsupported CdS aerogels with tunable optical properties. J Non Cryst Solids 2004;350:1–8.

doi:10.1016/j.jnoncrysol.2004.05.020.

[73] Tsou P. Silica aerogel captures cosmic dust intact. J Non Cryst Solids 1995;186:415–27. doi:10.1016/0022-3093(95)00065-8.

[74] Shi Z, Zhang W, Zhang F, Liu X, Wang D, Jin J, et al. Ultrafast separation of emulsified oil/water mixtures by ultrathin free-standing single-walled carbon nanotube network films. Adv Mater 2013;25:2422–7. doi:10.1002/adma.201204873.

[75] Zhao Y, Hu C, Hu Y, Cheng H, Shi G, Qu L. A versatile, ultralight, nitrogen-doped graphene framework. Angew Chemie - Int Ed 2012;51:11371–5. doi:10.1002/anie.201206554.

[76] Zhang W, Zhang Y, Lu C, Deng Y. Aerogels from crosslinked cellulose nano/micro-fibrils and their fast shape recovery property in water. J Mater Chem 2012;22:11642. doi:10.1039/c2jm30688c.

[77] Pierre AC, Pajonk GM. Chemistry of aerogels and their applications. Chem Rev 2002;102:4243–65. doi:10.1021/cr0101306.

[78] Israelachvili JN. Intermolecular and Surface Forces Third Edition. Academic Press; 2010. doi:10.1016/B978-0-12-375182-9.10025-9.

[79] Prakash SS, Brinker CJ, Hurd AJ, Rao SM. Silica aerogel films prepared at ambient pressure by using surface derivatization to induce reversible drying shrinkage. Nature 1995;374:439–43. doi:10.1038/374439a0.

[80] Toivonen MS, Kaskela A, Rojas OJ, Kauppinen EI, Ikkala O. Ambient-Dried Cellulose Nanofibril Aerogel Membranes with High Tensile Strength and Their Use for Aerosol Collection and Templates for Transparent, Flexible Devices. Adv Funct Mater 2015;25:6618–26. doi:10.1002/adfm.201502566.

[81] Yang X, Cranston ED. Chemically Cross-Linked Cellulose Nanocrystal Aerogels with Shape Recovery and Superabsorbent Properties. Chem Mater 2014;35:6016–25. doi:10.1021/cm502873c.

[82] Donius AE, Liu A, Berglund LA, Wegst UGK. Superior mechanical performance of highly porous, anisotropic nanocellulose-montmorillonite aerogels prepared by freeze casting. J Mech Behav Biomed Mater 2014;37:88–99. doi:10.1016/j.jmbbm.2014.05.012.

[83] Meador MAB, Scherzer CM, Vivod SL, Quade D, Nguyen BN. Epoxy reinforced aerogels made using a streamlined process. ACS Appl Mater Interfaces 2010;2:2162–8. doi:10.1021/am100422x.

[84] Zou J, Liu J, Karakoti AS, Kumar A, Joung D, Li Q, et al. Ultralight multiwalled carbon nanotube aerogel. ACS Nano 2010;4:7293–302. doi:10.1021/nn102246a.

[85] Sun H, Xu Z, Gao C. Multifunctional, ultra-flyweight, synergistically assembled carbon

aerogels. Adv Mater 2013;25:2554–60. doi:10.1002/adma.201204576.

[86] Du A, Zhou B, Zhang Z, Shen J. A special material or a new state of matter: A review and reconsideration of the aerogel. Materials (Basel) 2013;6:941–68. doi:10.3390/ma6030941.

[87] Sachithanadam M, Joshi SC. High strain recovery with improved mechanical properties of gelatin-silica aerogel composites post-binding treatment. J Mater Sci 2014;49:163–79. doi:10.1007/s10853-013-7690-1.

[88] Meena AK, Mishra GK, Rai PK, Rajagopal C, Nagar PN. Removal of heavy metal ions from aqueous solutions using carbon aerogel as an adsorbent. J Hazard Mater 2005;122:161–70. doi:10.1016/j.jhazmat.2005.03.024.

[89] García-González CA, Alnaief M, Smirnova I. Polysaccharide-based aerogels - Promising biodegradable carriers for drug delivery systems. Carbohydr Polym 2011;86:1425–38. doi:10.1016/j.carbpol.2011.06.066.

[90] Fischer F, Rigacci A, Pirard R, Berthon-Fabry S, Achard P. Cellulose-based aerogels. Polymer (Guildf) 2006;47:7636–45. doi:10.1016/j.polymer.2006.09.004.

[91] Rigacci A. Cellulosic aerogels for energy applications. ACS- Am. Chem. Soc. 2008 Fall Natl. ACS Meet., 2008.

[92] Tan C, Fung BM, Newman JK, Vu C. Organic aerogels with very high impact strength. Adv Mater 2001;13:644–6. doi:10.1002/1521-4095(200105)13:9<644::AID- ADMA644>3.0.CO;2-#.

[93] Loudet J-C, Barois P, Poulin P. Colloidal ordering from phase separation in a liquid- crystalline continuous phase. Nature 2000;407:611–3.

[94] Alberti S. Phase separation in biology. Curr Biol 2017;27:R1097–102. doi:10.1016/j.cub.2017.08.069.

[95] Bordier C. Phase separation of integral membrane proteins in Triton X-114 solution. J Biol Chem 1981;256:1604–7.

[96] Dagotto E, Hotta T, Moreo A. Colossal magnetoresistant materials : the key role of phase separation. Book 2001;344:1–153.

[97] Moreo A, Yunoki S, Dagotto E. Solid state physics - Phase separation scenario for manganese oxides and related materials. Science (80- ) 1999;283:2034–40. doi:10.1126/science.283.5410.2034.

[98] van de Witte P, Dijkstra PJ, van den Berg JW a., Feijen J. Phase separation processes in polymer solutions in relation to membrane formation. J Memb Sci 1996;117:1–31. doi:10.1016/0376-7388(96)00088-9.

[99] Beltsios KG, Bedard MCM. Structure formation phenomena during phase inversion. II.

Semiflexible polymers and liquid crystallinity. J Macromol Sci 2000;B39:623–44. doi:Doi 10.1081/Mb-100102475.

[100] Guillen GR, Pan Y, Li M, Hoek EM V. Preparation and characterization of membranes formed by nonsolvent induced phase separation: A review. Ind Eng Chem Res 2011;50:3798–817. doi:10.1021/ie101928r.

[101] Rezabeigi E, Wood-adams PM, Drew RAL. Production of porous polylactic acid monoliths via nonsolvent induced phase separation. Polymer (Guildf) 2014:1–11. doi:10.1016/j.polymer.2014.10.063.

[102] Lloyd DR, Kinzer KE, Tseng HS. Microporous membrane formation via thermally induced phase separation. I. Solid-liquid phase separation. J Memb Sci 1990;52:239–61. doi:10.1016/S0376-7388(00)85130-3.

[103] Kabra BG, Gehrke SH, Spontak RJ. Microporous, Responsive Hydroxypropyl Cellulose Gels. 1. Synthesis and Microstructure. Macromolecules 1998;31:2166–73. doi:10.1021/ma970418q.

[104] Stevens MP. Polymer Chemistry: An introduction, 2nd ed. Oxford, NY: Oxford University Press; 1990.

[105] Young RJ, Lovell PA. Introduction to Polymers. 3rd ed. CRC Press, Inc.; 2011.

[106] Flory PJ. Themodynamics of high polymer solutions. J Chem Phys 1941;10:51–61. doi:10.1063/1.1723621.

[107] Flory PJ. Principles of Polymer Chemistry. Cornell University Press, Ithaca, NY; 1953.

[108] Huggins ML. Some properties of solutions of long-chain compounds. J Phys Chem 1942;46:151–8. doi:10.1021/j150415a018.

[109] Huggins ML. Solutions of long chain compounds. J Chem Phys 1941;9:440. doi:10.1063/1.1750930.

[110] Eitouni HB, Balsara NP. Thermodynamics of Polymer Blends, Volume 1. AIP Press; 1998.

[111] Graessley WW, Krishnamoorti R, Reichart GC, Balsara NP, Fetters LJ, Lohse DJ. Regular and Irregular Mixing in Blends of Saturated Hydrocarbon Polymers. Macromolecules 1995;28:1260–70. doi:10.1021/ma00108a065.

[112] Bates FS, Muthukumar M, Wignall GD, Fetters LJ. Thermodynamics of isotopic polymer mixtures: Significance of local structural symmetry. J Chem Phys 1988;89:535–44. doi:10.1063/1.455442.

[113] Han CC, Bauer BJ, Clark JC, Muroga Y, Matsushita Y, Okada M, et al. Temperature, composition and molecular-weight dependence of the binary interaction parameter of polystyrene/poly(vinyl methyl ether) blends. Polymer (Guildf) 1988;29:2002–14.

doi:10.1016/0032-3861(88)90174-7.

[114] Sun Z, Wang CH. Determination of Flory-Huggins interaction parameter and self-diffusion coefficients in ternary polymer solutions by quasielastic light scattering. J Chem Phys 1995;103:3762–6. doi:10.1063/1.470055.

[115] Olabisi O, Robeson LM, Shaw MT. Polymer-Polymer Miscibility. Academic Press, New York; 1979.

[116] Shaw MT. Studies of polymer–polymer solubility using a two‐dimensional solubility parameter approach. J Appl Polym Sci 1974;18:449–72. doi:10.1002/app.1974.070180212.

[117] Piccarolo S, Titomanlio G. Synergism in the swelling and solubility of poly(methyl methacrylate) in presence of ethanol/water mixtures. Die Makromol Chemie, Rapid Commun 1982;3:383–7. doi:10.1002/marc.1982.030030602.

[118] Hildebrand JH, Scott RL. Solubility of Non-Electrolytes, 3rd Ed. Reinhold, New York,1950; Dover, New York, 1964; n.d.

[119] Hildebrand JH, Scott RL. Regular Solutions. Prentice-Hall, Englewood Cliffs, NJ; 1962.

[120] Hildebrand JH, Prausnitz JM, Scott RL. Regular and Related Solutions. Van Nostrand- Reinhold, Princeton, NJ; 1970.

[121] Bhattacharyya SN, Patterson D, Somcynsky T. The principle of corresponding states and the excess functions of N-alkane mixtures. Physica 1964;30:1276–92. doi:10.1016/0031- 8914(64)90082-5.

[122] Patterson D. Role of free volume changes in polymer solution thermodynamics. J Polym Sci Part C Polym Symp 1967;16:3379–89. doi:10.1002/polc.5070160632.

[123] Blanks RF, Prausnitz JM. Thermodynamics of polymer solubility in polar and nonpolar systems. Ind Eng Chem Fundam 1964;3:1–8. doi:10.1021/i160009a001.

[124] Biroš J, Zeman L, Patterson D. Prediction of the x Parameter by the Solubility Parameter and Corresponding States Theories. Macromolecules 1971;4:30–5. doi:10.1021/ma60019a008.

[125] Fedors RF. A method for estimating both the solubility parameters and molar volumes of liquids. Polym Eng Sci 1974;14:147–54. doi:10.1002/pen.760140211.

[126] Just S, Sievert F, Thommes M, Breitkreutz J. Improved group contribution parameter set for the application of solubility parameters to melt extrusion. Eur J Pharm Biopharm 2013;85:1191–9. doi:10.1016/j.ejpb.2013.04.006.

[127] Du Y, Xue Y, Frisch HL. Solubility Parameters. vol. 15. AIP Press, New York; 1996.

[128] van Arkel AE. Mutual Solubility of Liquids. Trans Faraday Soc 1946;42B:81–4.

[129] Small PA. Some factors affecting the solubility of polymers. J Appl Chem 2007;3:71–80. doi:10.1002/jctb.5010030205.

[130] Prausnitz JM, Anderson R. Thermodynamics of solvent selectivity in extractive distillation of hydrocarbons. AIChE J 1961;7:96–101. doi:10.1002/aic.690070123.

[131] Hansen CM. Hansen Solubility Parameter- User Handbook. vol. 1. 2007. doi:10.1017/CBO9781107415324.004.

[132] Ramanaiah S, Rani PR, Reddy KS. Hansen solubility parameters for the solid surface of cellulose acetate propionate by inverse gas chromatography. J Macromol Sci Part B 2012;51:2191–200. doi:10.1080/00222348.2012.669681.

[133] Segarceanu O, Leca M. Improved method to calculate Hansen solubility parameters of a polymer. Prog Org Coatings 1997;31:307–10. doi:10.1016/S0300-9440(97)00088-X.

[134] Milliman HW, Boris D, Schiraldi DA. Experimental determination of Hansen solubility parameters for select POSS and polymer compounds as a guide to POSS-polymer interaction potentials. Macromolecules 2012;45:1931–6. doi:10.1021/ma202685j.

[135] Hansen CM, Abbot S. Hansen Solubility Parameters 2018. https://www.hansen- solubility.com/ (accessed April 7, 2018).

[136] Bergin SD, Sun Z, Rickard D, Streich P V., Hamilton JP, Coleman JN. Multicomponent solubility parameters for single-walled carbon nanotube-solvent mixtures. ACS Nano 2009;3:2340–50. doi:10.1021/nn900493u.

[137] Gårdebjer S, Andersson M, Engström J, Restorp P, Persson M, Larsson A. Using Hansen solubility parameters to predict the dispersion of nano-particles in polymeric films. Polym Chem 2016;7:1756–64. doi:10.1039/C5PY01935D.

[138] Ho DL, Glinka CJ. Effects of solvent solubility parameters on organoclay dispersions. Chem Mater 2003;15:1309–12. doi:10.1021/cm0217194.

[139] Strathmann H. Production of Microporous Media by Phase Inversion Processes. Mater Sci Synth Membr 1985;269:165–95. doi:10.1021/bk-1985-0269.ch008.

[140] Li CL, Wang DM, Deratani A, Quémener D, Bouyer D, Lai JY. Insight into the preparation of poly(vinylidene fluoride) membranes by vapor-induced phase separation. J Memb Sci 2010;361:154–66. doi:10.1016/j.memsci.2010.05.064.

[141] Loeb S, Sourirajan S. Sea water demineralization by means of an osmotic membrane. Adv Chem 1963;38:117–32. doi:10.1021/ba-1963-0038.ch009.

[142] Pouchly J, Zivny A, Solc K. Thermodynamic Equilibrium in the System Macromolecular Coil-binary Solvent. J Polym Sci Part C-Polymer Symp 1968;23:245–56.

[143] Altena FW, Smolders CA. Calculation of Liquid-Liquid Phase Separation in a Ternary

System of a Polymer in a Mixture of a Solvent and a Nonsolvent. Macromolecules 1982;15:1491–7. doi:10.1021/ma00234a008.

[144] Yilmaz L, Mchugh AJ. Analysis of nonsolvent–solvent–polymer phase diagrams and their relevance to membrane formation modeling. J Appl Polym Sci 1986;31:997–1018.

[145] Cohen C, Tanny GB, Prager S. Diffusion-controlled formation of porous structures in ternary polymer systems. J Polym Sci Polym Phys Ed 1979;17:477–89. doi:10.1002/pol.1979.180170312.

[146] Reuvers AJ, Smolders CA. Formation of membranes by means of immersion precipitation. Part I. A model to describe mass transfer during immersion precipitation. J Memb Sci 1987;34:67–86. doi:10.1016/S0376-7388(00)80021-6.

[147] Reuvers AJ, Smolders CA. Formation of membranes by means of immersion precipitation Part II. The mechanism of formation of membranes prepared from system of cellulose acetate-acetone-water 1987;34:67–86.

[148] Tsay CS, Mchugh AJ. Mass transfer modeling of asymmetric membrane formation by phase inversion. J Polym Sci Part B Polym Phys 1990;28:1327–65. doi:10.1002/polb.1990.090280810.

[149] Radovanovic P, Thiel SW, Hwang ST. Formation of asymmetric polysulfone membranes by immersion precipitation. Part II. The effects of casting solution and gelation bath compositions on membrane structure and skin formation. J Memb Sci 1992;65:231–46. doi:10.1016/0376-7388(92)87025-S.

[150] Radovanovic P, Thiel SW, Hwang ST. Formation of asymmetric polysulfone membranes by immersion precipitation. Part I. Modelling mass transport during gelation. J Memb Sci 1992;65:213–29. doi:10.1016/0376-7388(92)87024-R.

[151] Cheng L-P, Soh YS, Dwan A-H, Gryte CC. An improved model for mass transfer during the formation of polymeric membranes by the Immersion-Precipitation process. J Polym Sci Part B Polym Phys 1994;32:1413–25. doi:10.1002/polb.1994.090320813.

[152] Widjojo N, Lipscomb GG, Chung T. Evolution of polymeric hollow fibers as sustainable technologies : Past , present , and future. Prog Polym Sci 2012;37:1401–24. doi:10.1016/j.progpolymsci.2012.01.001.

[153] Matsuura T. Synthetic Membranes and Membrane Separation Processes. CRC Press, Inc.; 1994.

[154] Smolders CA, Reuvers AJ, Boom RM, Wienk IM. Microstructures in phase-inversion membranes. Part 1. Formation of macrovoids. J Memb Sci 1992;73:259–75. doi:10.1016/0376-7388(92)80134-6.

[155] van’t Hof JA, Reuvers AJ, Boom RM, Rolevink HHM, Smolders CA. Preparation of asymmetric gas separation membranes with high selectivity by a dual-bath coagulation

method. J Memb Sci 1992;70:17–30. doi:10.1016/0376-7388(92)80076-V.

[156] Hao J, Wang S. Calculation of Alcohol-Acetone-Cellulose Acetate Ternary Phase Diagram and Their Relevance to Membrane Formation. J Appl Polym Sci 2001;80:1650–7.

[157] Sun AC, Kosar W, Zhang Y, Feng X. A study of thermodynamics and kinetics pertinent to formation of PVDF membranes by phase inversion. Desalination 2013;309:156–64. doi:10.1016/j.desal.2012.10.005.

[158] Barzin J, Sadatnia B. Correlation between macrovoid formation and the ternary phase diagram for polyethersulfone membranes prepared from two nearly similar solvents. J Memb Sci 2008;325:92–7. doi:10.1016/j.memsci.2008.07.003.

Chapter 2: Isolating Lignin Rich Micro-Nano Fibrillar Cellulose (MNFC) from Residual Biomass

2.1 Abstract

We approach light-weighting by exploring nature through the study of the interaction of cellulose, lignin and hemicellulose and their structural arrangement within the biomass. Two prominent residual biomasses generated as waste product in Southeast Asia are chosen: coir fibers

(from coconut fruit) and Empty Fruit Bunches fibers (EFB, residue from palm oil). The residual biomass of coir fibers and palm tree Empty Fruit Bunches (EFB) are subjected mild treatments such as autohydrolysis and microemulsion treatments followed by microfluidization to isolate

Micro and Nano Fibrillar Cellulose (MNFC) with high residual lignin content of ~24 and ~31 wt%, respectively. This serves two purposes. First, in the process of defibrillation, the structural and chemical arrangement of cellulose, lignin, and hemicellulose is studied within coir fibers and EFB.

Second, the structural and thermo-chemical analyses of MNFC presented in this chapter expects to facilitate further interest in preparation of new light-weight materials from these novel micro- nano lignocellulose. The most significant findings include the fact that two MNFC populations, with distinctive structural differences are produced, with characteristic lateral dimensions of 20-

70 nm and 1-3 μm. The lignin distribution after possible re-condensation occurred in the form of nano-droplets.

2.2 Introduction

Lignocellulosic biomass in the form of wood, grasses, fibers and agricultural residues are efficient components in natural biocomposites, as produced by nature after centuries of iterations and evolution. Thus, man-made materials from sustainable, bio-based, biodegradable and renewable biomass is an excellent starting point in efforts to replace fossil sources for such

purpose.1–3 However, materials engineered from lignocellulosic fibers often use chemicals in their isolation from biomass that may require complex recovery processes; in addition, the extensive degree of chemical and physical deconstruction may limit a full exploitation of the properties inherent to wood fibers. There has been a large research impetus on producing fibers from natural resources using mild treatments to better preserve the original structure of the system and to reduce the environmental footprint.4,5

The natural fibers used in this study are residuals from processing the coconut husk and the empty fruit bunches of palm trees. Coir fibers (density of 0.6-0.9 g/cm3) from the coconut husks represent 5-6 million tons per year with roughly 80% contribution coming from Southeast

Asia.6 Only 10% of this potential enters the commercial trade and, traditionally, they are used to make mats, carpets, ropes, mattings, fishing nets and brushes because of their high ductility (15-

40% strain).7 Recently, few studies proposed coir fiber utilization as raw material in polymer composites and efforts are underway to explore wider markets.8–11

The other non-woody biomass studied here, empty fruit bunches (EFB) fibers (density of

~1 g/cm3), is the residual lignocellulosic material after palm oil production.4,12 Malaysia is the largest palm oil producer, contributing to 51% of the market worldwide. The global oil production reached almost 21 million tons in 2010.13 Thus, an enormous amount of residues are generated by this industry, mainly, EFB fibers, which are proposed here for the production of micro-nano fibers by methods different than those we4,14 and others5,15–18 have reported so far. One critical issue common to coir and EFB fibers is the high lignin content, as high as 45%. Therefore, any attempt in utilization should consider preserving this component as the only way to make any industrial effort truly sustainable.

Efforts to extract nanocelluloses from coir and EFB fibers have been reported.4,5,16–18

However, the studies so far have generally used chemical treatments that can be improved in terms of their severity and its consequences. Specifically, micro- and nanofibers carrying residual lignin, hemicelluloses and extractives may present an attractive raw option for higher yield, low production cost and much lower environmental impact. Moreover, the presence of lignin has been shown to enhance barrier properties and tunable surface properties of nanocellulose films and nanopapers.19,20 With the goal of retaining the native lignin in coir and EFB fibers, we report a methodology to isolate from these residual biomass, lignin-rich micro and nano fibrillar cellulose

(generally termed here as MNFC). The multi-stage procedure involves previously reported procedures for autohydrolysis21 and microemulsion treatment22, both of which are mild treatment procedures that allow us to retain the fiber composition. Our study includes morphological, structural and thermal analyses which were performed after each processing step to develop a thorough understanding of defibrillation and lignin redistribution. Based on the developed understanding and wood fiber morphology, a defibrillation mechanism is proposed.

2.3 Experimental Section

2.3.1 Materials

Coir fibers were obtained from the National Coir Research and Management Institute

(NCRMI) under the Government of Kerala, India. These were retted fibers, manually separated from husk. EFB from a Malaysian oil palm mill was supplied by Straw Pulping Engineering S.L.

(Zaragoza, Spain). The fiber samples were used as received.

The chemicals for micro-emulsion pretreatment included R-limonene sourced from Fulka,

St. Louis, USA. Sodium dodecyl sulfate (SDS), n-pentanol, sodium chloride (NaCl), sodium

hydroxide (NaOH) and urea were obtained from Sigma-Aldrich. DI water was used for all experimental purposes.

2.3.2 Autohydrolysis pretreatment

The pretreatment was carried out in a 10 L digester. 30 g of the fiber sample (coir or EFB) was loaded with distilled (DI) water at a fiber-to-water mass ratio of 1:60. The fibers were digested at 180 °C for 60 min followed by mechanical disintegration in a Sprout-Bauer. The disintegrated fibers were filtered using a cheese cloth whereupon they were filtered by centrifugation at 15000 rpm for 1 min. The pretreated samples were stored in plastic bags at 4 °C until further use. These grinded fiber samples are referred to as H-Coir or H-EFB.

2.3.3 Microemulsion pretreatment

A microemulsion was prepared with an anionic surfactant (sodium dodecyl sulfate, SDS) and R-limonene, creating a thermodynamically stable, single phase surfactant-oil-water system

(SOW).22 As reported before, the treatment was performed on the fiber samples to disrupt the hydrogen bonding between the cellulosic fibrils, thus facilitating the mechanical defibrillation.

Briefly, a 100 ml microemulsion was prepared by adding SDS (3.0 wt %) to water (67.7 wt%) and

NaCl (2.7 wt%) solution. The active agent, urea (12.5 wt%), was added followed by limonene (8.0 wt%), n-pentanol (3.1 wt%) and NaOH (3.0 wt%). The mixture was mixed under low shear

(magnetic bar) to produce a clear, isotropic liquid microemulsion. The H-Coir or H-EFB fibers

(3.0 wt% solid content) were soaked in this microemulsion for 12 h at room temperature and atmospheric pressure. The fibers were then filtered under vacuum and washed with DI water several times to remove the components of the microemulsion that may have remained after filtration. Thereafter, they were centrifuged at 15000 rpm for 1 min and stored at 4 °C until further use. The fiber samples after microemulsion pretreatment are referred to as M-Coir or M-EFB.

2.3.4 Fiber processing and defibrillation

The M-coir and M-EFB fibers were re-dispersed in water (0.15 wt% solid content) using a high shear mixer at 22,000 rpm for 15 min. MNFC-coir and EFB fibers were obtained following microfluidization (Microfluidics M-110T). The microfibers were passed 20 times through an intensifier pump that increased the pressure (2000 bar), followed by an interaction chamber which defibrillated the microfibers by shear forces and impacts against the channel walls and colliding streams. Through this process, the microfibers were further broken up into nano-sized structures forming the micro-nano lignocellulosic fibrils (MNFC). The unit was operating under a constant shear rate. The temperature was not controlled but fluidization was temporarily ceased when the temperature of the dispersion reached ~90˚C to prevent pump cavitation. Processing then recommenced when the temperature was ~45˚C. The resulting LCNF suspensions were collected and stored at 4˚C in a refrigerator. It was observed in our case that the microfluidizer got clogged after 1 pass of the coir/EFB fibers that were not subjected to microemulsion pretreatment.

2.3.5 Thermal and chemical analyses

Fiber composition

The chemical composition of the raw fibers was determined following TAPPI standards:

T-222 for lignin, T-203 for α-cellulose, T-204 for ethanol–benzene extractives and T-211 for ash.

Holocellulose content was determined by the Wise and co-workers method.23,24

Fourier Transform Infrared Spectroscopy (FTIR)

FTIR was conducted using a Perkin Elmer spectrophotometer in the Attenuated Total

Reflectance mode (FTIR-ATR). Samples were analyzed using the Pike Miracle accessory equipped with a Germanium (GE) crystal. The spectrum was collected for 12 scans and corrected for background noise. The multipoint baseline correction was realized for each spectrum and the

curves were normalized to 20% transmittance with respect to the peak at 1027 cm-1. All samples were dried in an oven at 60 °C and kept dry in a dessicator prior to analysis.

Thermogravimetric analysis (TGA)

Thermal degradation analysis was performed using a TA instruments TGA Q500. The analysis was carried out in a N2 environment at a flow rate of 60 ml/min with a constant heating rate of 10 °C/min over a temperature range of 40-700 °C. The fiber samples were dried in oven at

60 °C and stored in desiccator prior to the analysis. The analysis was done in triplicate for data reproducibility. The time-derivative of the weight loss curve was obtained and smoothed by adjacent averaging over 50 points. The differential curve was plotted against temperature.

2.3.6 Structural and molecular analyses

Scanning electron microscopy (SEM)

Imaging was performed by Field Emission Scanning Electron Microscope (FESEM), FEI

Verios 460L. The P-coir and P-EFB fibers were cryo-fractured under liquid N2. A sharp, clean blade was used to fracture the material and to obtain images of the radial cross-sections. The remaining samples (H, M and MNFC) were suspended in water at 0.01 wt% and freeze dried after vitrifying in liquid N2. All the samples were mounted on an SEM stub using a double-sided carbon tape. The prepared samples were coated with 5-nm layer of gold and platinum to capture secondary electrons from the surface and thus reduce charging. The imaging was carried out at the landing voltage of 2 kV and 13 pA current at a working distance of 5 mm. The Everhart-Thornley (ETD) detector was employed to detect secondary electrons from the samples.

Atomic force microscopy (AFM)

A silica wafer was carefully cut into 1x1 cm2 and the surface was cleaned by soaking in

0.1N NaOH for 10 s. After rinsing in DI water, the wafer was subjected to UV light for 15 min. A

0.01 wt % of the fiber suspension was sonicated at 30% amplitude. A drop of the resulting suspension was placed on the previously cleaned silica surface, which was allowed to dry in the oven at 60 0C. The fibers dried on silica surfaces were mounted on an aluminum sample holder and examined with a Dimension 3000 scanning probe microscope from Veeco Metrology Group.

Scanning was performed in the tapping mode in air using silicon cantilevers (NSCl5/AIBS) delivered by Olympus AC160TS. The drive frequency of the cantilever was about 275–325 kHz

(nominal resonance of 300 kHz). The scanned areas were imaged. No image processing except flattening was made. Images were taken with a feedback loop to keep the amplitude of oscillation constant and the response of the feedback loop was measured. The response of the feedback loop was used to determine how far the scanner was moved in Z direction in order to keep the amplitude of oscillation constant.

Laser scanning confocal fluorescence microscopy (LSCM)

A Zeiss LSM 710 laser scanning confocal microscope with Zen 2008 was used with the excitation laser at Ar 488 nm over an emission range of 490 to 560 nm and a 40x C-Apochromat objective (1.1 W NA) zoom to acquire multichannel fluorescence image. The fiber samples were suspended in water prior to the analysis.

X-ray Diffraction (XRD) analysis

XRD was conducted using Rigaku SmartLab X-ray diffractometer using the operating mode of Bragg-Brentano reflection geometry. A Cu Kα radiation source of wavelength 0.154 nm and operating at 40 kV and 44 mA was incident on the solid sample. The XRD patterns were

recorded over the angular range 2θ ranging from 5 to 50°. The fiber samples were dried in the oven at 60 °C and manually cut into small pieces using a pair of scissors prior to the analysis. Full width at half maximum (FWHM) was calculated by calculating peak width of I200 at half its maximum intensity.

2.4 Results and Discussion

We begin by describing the procedure used for the isolation of the lignin-rich micro-nano fibrillar cellulose (MNFC) from coir and empty fruit bunches (EFB) fibers. During each step of fiber processing, the fibers were characterized for chemical composition (TAPPI standard and

FTIR), morphology (AFM, SEM), lignin distribution (LSCM), crystallinity (XRD) and thermal properties (TGA). Based on the results a defibrillation model is proposed.

2.4.1 Isolation of lignin-rich micro-nano fibrillar cellulose (MNFC)

The experimental procedure for the isolation of MNFC fibers (Figure 2.1) was designed to retain the lignin present in the coir or EFB fibers (amounting 36-43%). The as-received or pristine coir and EFB fibers (P-coir and P-EFB) were subjected to autohydrolysis (hydrothermal treatment) followed by fiber disintegration and refining in the Sprout-Bauer producing H-coir and

H-EFB, respectively. A microemulsion pretreatment was performed on these fibers to delaminate and disrupt the hydrogen bonding to give M-coir and M-EFB fibers. These fibers were then microfluidized and are referred to as MNFC-coir or MNFC-EFB, respectively. The microemulsion step was necessary to prevent microfluidizer from clogging. The details of the procedure are included in the experimental section.

Figure 2.1: The synthesis procedure for the micro-nano fibrillar cellulose from coir fiber (MNFC-coir) and the sample nomenclature is shown. The as received coir fiber (P-coir) was subjected to autohydrolysis and fiber refining (H-coir) followed by microemulsion pretreatment (M-coir) and microfluidization to produce MNFC-coir. The same procedure was carried on EFB fibers.

Autohydrolysis is a mild fiber treatment process, in which fibers are heated in DI water at

180 °C to partially leach the soluble lignin and hemicelluloses from the fibers, while leaving behind high residual amounts of insoluble lignin and cellulose in the fibers.21,25 The chemical composition of the coir and EFB fibers after each processing stage is summarized in Table 2.1.

After autohydrolysis, the hemicelluloses content is reduced from 23 and 20 wt% to 7 and 4 wt% for coir and EFB fibers, respectively. Likewise, for coir and EFB fibers, a more limited reduction in lignin content, by 9 and 7 wt% respectively, was determined after autohydrolysis. The microemulsion treatment does not lead to significant changes in chemical composition of the H- coir or H-EFB fibers. A very slight decrease in the lignin content (by 3-5 %) is attributed to the alkaline medium of the microemulsion system, due to presence of NaOH.22 The composition analysis after microfluidization was not attempted since it was assumed that mechanical defibrillation did not alter substantially the chemical composition of the fibers.26 With the unique combination of these mild pretreatment processes, the MNFC from coir and EFB fibers retained the original lignin, to a large degree, with final content of 31 and 24 wt%, respectively. To the best of our knowledge, this is the first study to report on the isolation of MNFC from coir and EFB fibers while retaining the native lignin. 50

Table 2.1: Chemical composition of coir and EFB fibers (P) and after sequential treatment via autohydrolysis (H) and microemulsion impregnation (M).

Sample Extractives Lignin Cellulose Hemicellulose (wt%) (wt%) (wt%) (wt%) P-Coir / P-EFB 0.5 / 0.8 43 / 36 30 / 34 23 / 20 H-Coir / H-EFB <0.1 / <0.5 34 / 29 56 / 58 7 / 4 M-Coir / M-EFB <0.1 / <0.5 31 / 24 56 / 67 5 / 3

2.4.2 Investigating chemical structure

FTIR analysis was performed to investigate any changes in the chemical structure during fiber processing. The infrared spectra of cellulose, hemicellulose and lignin has been extensively studied in the literature.15,27,28 The three fiber components mainly comprise functional groups such as alcohols, ketones, ether, aromatics, saturated alkanes and esters. The FTIR spectra obtained from coir and EFB fiber samples at each processing stage is shown in Figure 2.2. Some of the significant peaks are highlighted in the FTIR spectra. Table A2.1 lists all the major peaks that are identified with their corresponding functional groups, chemical compounds and the given fiber component. Two main transmittance regions ranging from 3500-2900 cm-1 and 1740-700 cm-1 are seen in all the curves. A broad and prominent peak at 3329 cm-1 and a distinct peak at 2910 cm-1 are due to O-H stretch and sp3 C-H stretch respectively, demonstrating availability of large number of acid, alcohol groups and saturated alkanes, that are present in all three fiber components. One major difference observed during processing of coir and EFB is the C=O stretch at 1736 cm-1 that arises from esters present in hemicellulose in P-coir and P-EFB. This peak is shifted to ca. 1715 cm-1 after autohydrolysis and remains there following further processing. This may be due to two reasons. First, most of the hemicellulose leaches out after autohydrolysis. Second, the ester groups in the remaining hemicellulose are most likely converted to carboxylic groups during hydrolysis.

An alkoxy C-O stretch and C-O deformation at 1027 cm-1 and an acyl C-O peak at 1240 cm-1 represents C-O-C stretching vibration arising from pyranose ring17,29. The shoulder peaks at 1161 cm-1 and 1100 cm-1 are characteristic of cellulosic components and have been reported before by

Yang et al.28 These peaks are also attributed to C-O-C stretching vibrations. The C-O-C asymmetric stretching can also be attributed to ester functional group present in hemicelluloses and the absence of this peak after autohydrolysis further indicates reduction in hemicellulose content.

Figure 2.2: FTIR spectra for coir and EFB at various stages of fiber processing highlighting major peaks.

The peaks in the range of 1470-1430 cm-1 represent methoxy groups and the peaks between

1600-1460 cm-1 represent C=C stretch of aromatics. Peaks in both regions are characteristic of methoxy groups and aromatic rings present in the lignin. These peaks are absent in the FTIR of pure cellulose, as shown in Figure A2.1. The peaks in 1400-1300 cm-1 correspond to sp3 C-H bend. Even though FTIR spectra were developed after careful drying and storage of the samples, yet a small peak at 1638 cm-1 was observed and attributed to the O-H bending arising from the

adsorbed water on the surface. This demonstrates that water molecules are difficult to completely remove from cellulose due to strong molecular interactions. The peak at 898 cm-1 is attributed to

C-O-C stretch from glycosidic linkage between glucose units. The peaks from 900-700 cm-1 are due to aromatic sp2 bend arising from aromatic hydrogen present in lignin.

From the FTIR spectra analysis, three conclusions can be drafted. First, coir and EFB have very similar chemical composition. Second, the fiber processing steps utilized in this study do not alter the chemical structure of the fiber components apart from small conversion of hemicellulose esters to carboxylic groups; lastly, a distinct lignin IR fingerprint is observed in the MNFC samples for lignin chemical groups, further confirming that the native lignin is retained in the final material.

2.4.3 Structural characterization and lignin redistribution

This section investigates the change in structure of coir and EFB fibers due to the various processing steps. Figure 2.3 shows SEM images of the coir (left column) and EFB fibers (right column). The top row shows images of the untreated or pristine fibers. The middle row shows fiber images after autohydrolysis and mechanical defibrillation and the bottom one shows fiber images after microemulsion pretreatment. Both, the untreated coir and EFB fibers, display very similar morphology. The same applies, surprisingly, after processing. The cross-sections of the coir and

EFB fibers are revealed in Figures 2.3a and c, respectively. The observed macro-fibers have high aspect ratio with a length of few centimeters and widths of the order of 200 to 500 µm. Figures

2.3b and d show the surface texture of coir and EFB fibers, respectively. The cross-section and the surface images reveal that these are thick-walled fibers covered with surface pits (shown in solid circles). Both these fibers are in a tightly packed arrangement forming hollow tubules. The hydrothermal treatment followed by mechanical defibrillation (H-Coir and H-EFB) rips apart the tight arrangement of the tubular structures into individual fibers, as shown in Figure 2.3e and f.

These individual fibers retain the high aspect ratio, with few mm in length and 10-15 µm in diameter. The forced defibrillation of the fibers results in clusters of small fibrils attached to the fibers (shown in dotted circles). The inset image of Figures 2.3e and f shows higher magnification images of H-coir and H-EFB fibers, respectively. They reveal the presence of surface pits on the individual fibers. The water and nutrient are transported across the bundle of coir and EFB macro fibers via these surface pits, as indicated before in Figure 2.3b and d. The inset image further reveals a distinctive high microfibril angle for coir (indicated by lines in the image). The angle was analyzed using Image J software and was measured to be ~420 for coir fibers and ~ 210 for EFB fibers. The microemulsion treatment did not alter the morphology of the fibers (Figure 2.3g and h). The fibers retained the same dimensions. The only structural difference observed is that “clean” fibers, i.e., the small fibrils attached to the fibers, are removed, resulting in better isolation. This may be one of the reasons why microemulsion treatment was necessary before microfluidization.

After “cleaning” the fibers, the other features such as the pitted vessel elements were still observed

(inset of Figure 2.3h and Figure A2.2), which have been reported previously.30 The vessel walls were crowded with pits that were neatly arranged in elongated oval shapes. But, these pitted vessel elements are very few. Figure A2.3 reveals vessel elements in the cross-section of the pristine coir and EFB fibers.

Figures 2.4a and 2.4b show images of coir and EFB fibrils, respectively, after microfluidization and reveal a disruption of the initial fiber structure. Figures 2.4a1 and b1 exhibit polydispersity in fibril width, ranging from 20-70 nm. Figures 2.4a2 and b2 exhibit flat ribbon like fibrils with 1-3 µm width and approximately 500 nm in thickness. The images also indicate that the microfluidization applied did not completely defibrillate the sample. The two structural populations are evident from Figure 2.5 for both MNFC-coir and EFB. The small fibers are

interlinked to the large ribbon-like structures. However, this may be an artifact resulting from SEM sample preparation. To confirm, the microfluidized fiber samples were centrifuged and AFM imaging was carried out on the supernatant and the precipitated phases, as shown in Figure 2.5.

From the figure, a variety of fiber sizes can be seen in the non-centrifuged coir and EFB fiber samples. The supernatant, on the other hand, display only small fibers whereas the precipitate is filled with wide fibers along with smaller defibrillated fibrils attached to the larger fibers.

Furthermore, the AFM image show many small particles. Similar type of spherical structures was also observed in SEM images (indicated in Figure 2.4a) that are less than 1 µm in diameter. These are believed to be coalesced lignin particles. Several studies have reported on the deposition of lignin particles on the fibers after autohydrolysis.21,31 It has been recently proved that heating the lignin above its glass transition temperature cleaves the carbohydrate-lignin bonds.

This forces the hydrophobic lignin in the aqueous media to migrate from the cell walls and recondense into small spherical droplets that redeposits on the inside and outside of the fiber surface.32 There are also some reports on the formation of lignin-like compounds also referred to as pseudo-lignin. 33,34 Previous studies have shown to produce lignin droplets from carbohydrates

(called pseudo-lignin) even with the absence of native lignin in hybrid poplar, especially under high-severity pretreatment conditions.35,36

Since no severe pretreatment steps were applied, it is likely that native lignin redistributes itself inside and outside the fiber wall. To track the lignin redistribution, confocal imaging was performed on the coir and EFB fibers at each processing stage (Figure 2.6). The green auto- fluorescence is from lignin. The P-coir and P-EFB fibers (Figure 2.6a1 and a2) have homogeneous lignin distribution on the fiber walls. After autohydrolysis, the fiber defibrillates into individual fibers with majority of lignin distributed on the fiber walls, along with small particles of lignin that are observed within the fiber, as indicated by red circles in Figure 2.6. The

yellow circle indicates the vessel element as seen before in SEM image (Figure A2.2b). The confocal image after microemulsion treatment is very much like the one after autohydrolysis, except that the concentration of lignin particles inside the fiber is reduced significantly. The presence of small fibers, less than 1 µm in width, displaying green color uniformly after microfluidization indicates that lignin has remained or redeposited evenly on the defibrillated fibers.

Figure 2.6: Confocal microscopy image of coir and EFB fibers at each stage of fiber processing. The red circles indicate lignin droplets. The yellow circle indicates the vessel element.

Crystallinity:

XRD studies were performed at each stage of processing of the coir and EFB fibers to investigate the changes in crystallinity. The comparison of XRD profiles for coir and EFB fibers are shown in Figure 2.7a and b, respectively. All the diffractograms showed two peaks at 2θ of 15.800 and

22.450. These peaks may represent triclinic unit cell of cellulose I. The three main peaks for cellulose Iα are reported at 14.260, 16.70 and 21.80 for the miller indices corresponding (100), (010) and (110) respectively.37 The cellulose Iβ exhibits similar peaks with slight shift at 14.850, 16.670 and 22.980 for the miller indices corresponding (1-10), (110) and (200) respectively. The broad peak at 15.80 in Figure 2.7 can be deconvoluted into two peaks at approximately 14.50 and 16.50.

However, due to presence of large amounts of amorphous lignin, the two distinct peaks for cellulose I are difficult to resolve. Moreover, the absence of peaks at 12.20, 19.90 and 22.10 for cellulose II indicates that the treatments applied in this study did not change the hydrogen bonding arrangement and hence the cellulose retained the triclinic structure typical of cellulose I.

Figure 2.7: XRD diffractograms for Coir and EFB fibers at different treatment stages of processing.

The full width at half maximum (FWHM) for (200) peaks are listed in Table 2.2, that give a qualitative estimation of the extent of delignification of the fiber sample. An overall broad FWHM

(30-40) indicates presence of high amount of amorphous lignin, which agrees well with results

presented before. Processed Coir (H, M and MNFC) has a higher FWHM than processed EFB due to high lignin content present in the first case.

Table 2.2: XRD analysis parameters for full width half maximum (FWHM) for coir and EFB fibers at various stages of processing

P- H- M- MNFC- P-EFB H- M- MNFC- Coir Coir Coir Coir EFB EFB EFB FWHM 6.2 3.9 3.9 4.2 5.8 3.3 2.9 2.9 (degrees)

2.4.4 Thermal analysis

Thermal stability is one of the key factors to determine a material’s potential for use as reinforcing agent in composites. Due to difference in the chemical structure among cellulose, hemicelluloses and lignin, they present different thermal transitions.28,38 Hemicelluloses comprise chain combinations with many sugar units and is the first to decompose, starting from 220 to 315 ºC.

Cellulose consists of long glucopyranose chains that decomposes in the range of 315-400 ºC.

Phenolic compounds, including lignin, present a wide range of decomposition temperatures, covering a range that goes as high as 900 ºC.

The TGA thermographs for coir and EFB fibers at each treatment stage are included in

Figure 2.8 (a and b). As observed from the curves, a large amount of residue is left (>10%) for all the samples due to a high amount of lignin present. P-coir and P-EFB fibers are highest in lignin content and they have the largest residual mass at 700 ºC, 26.4 and 24%, respectively, indicating incomplete pyrolysis of lignin. In addition, there are two common features evident from TGA curves. First, a small weight loss is observed below 100 ºC, most likely due to moisture loss and secondly, the decomposition temperature for all the samples lies between 310- 360 ºC.

Table 2.3: The ratio of the fiber components as calculated from values from Table 1 along with thermal analysis data obtained from Figure 2.8.

퐂퐞퐥퐥퐮퐥퐨퐬퐞 퐂퐞퐥퐥퐮퐥퐨퐬퐞 Maximum Decompositio FWH mass loss Residu Sample 퐇퐞퐦퐢퐜퐞퐥퐥퐮퐥퐨퐬퐞 퐋퐢퐠퐧퐢퐧 n temperature M (ºC) rate e % (ºC) (wt%/min) P-Coir 1.3 0.7 315 29 -1.39 26.4 H-Coir 8.0 1.6 355 39 -2.09 16.3 M-Coir 11.2 1.8 331 45 -1.91 19.3 MNFC- -- -- 325 51 -1.33 19.6 Coir P-EFB 1.7 0.9 311 42 -1.52 24.0 H-EFB 14.5 2.0 353 34 -2.87 10.9 M-EFB 22.3 2.8 322 44 -1.67 19.0 MNFC- -- -- 335 47 -1.46 20.2 EFB

There are three aspects to notice from the derivative curves (Figure 2.8 c and d): peak location, peak width (quantified by Full Width at Half Maximum (FWHM) in graph inset) and relative peak height. The details from Figure 2.8 are listed in Table 2.3. The peak location indicates the thermal stability of the material. A shoulder peak for P-coir and EFB fiber is observed at around 270 ºC, which is distinct for coir samples compared to those from EFB. This is explained from the cellulose-to-hemicellulose ratio calculated from chemical composition data (Table 2.1), which is 1.3 for coir and 1.7 for EFB. A low cellulose-to-hemicellulose ratio indicated high hemicellulose content which is manifested in a shoulder peak for hemicellulose. Moreover, this peak disappears after autohydrolysis as most of the hemicellulose is leached away indicated by high cellulose-to-hemicellulose ratio. The prominent peak for all samples is due to cellulose decomposition and is characterized by the respective decomposition temperature in Table 2.3. The decomposition temperature increases by ~40 ºC after autohydrolysis. A similar increase in thermal stability of cellulose microfibrils from soybean hulls, upon removal of hemicelluloses has been observed in our previous studies.39 Raveendran et al.40 and Jin et al.38, have claimed that the thermal properties of the fibers are unaffected by interaction within the fiber components and are only affected by change in fiber composition. The cellulose content increases after autohydrolysis as indicated by increase in cellulose-to-hemicellulose ratio from 1.3 and 1.7 to 8.0 and 14.5 for H- coir and H-EFB, respectively. This explains a ~40 ºC rise in decomposition temperature for H- samples. However, further processing by microemulsion treatment and microfluidization decreases the decomposition temperature by 20-30 ºC. This sudden decrease in decomposition temperature cannot be explained by change in fiber composition alone, where the cellulose-to- hemicellulose ratio increases further to 11.2 and 22.3 for M-coir and EFB, respectively. A hypothesis is given later based on defibrillation mechanism to account for such behavior.

Another important feature is the broadness of the peak, characterized by the FWHM value.

Since, lignin decomposes in a wide temperature range, a broad peak should directly correlate with the ratio of cellulose-to-lignin present. The cellulose to lignin ratio increased from 0.7 and 0.9 for

P-coir and EFB, respectively, to 1.6 and 2.0 after autohydrolysis, indicating a relative decrease in lignin content. The FWHM, decreases for H-EFB, as expected but increases for H-coir. An exception is perceived due to a narrow peak for P-coir, with a FWHM of 29 ºC, explained as follows. In plant fibers, hemicelluloses act as a binder between cellulose and lignin. For P-coir, a large amount of hemicellulose decomposes first at 270 ºC, allowing for fast decomposition of cellulose as seen from 29 ºC FWHM for P-coir. However, in P-EFB and due to its low hemicellulose content, a slight overlap in hemicellulose and cellulose decomposition results in a broad cellulose degradation peak, with a FWHM of 42 ºC. After microemulsion treatment, the cellulose-to-lignin ratio further increases to 1.8 and 2.8 for M-coir and M-EFB, respectively. The

FWHM, on the contrary increases with decreasing lignin content, which further suggests that this behavior cannot be explained by fiber composition alone.

A similar trend is observed for maximum mass loss rate, which increases after autohydrolysis due to increase in cellulose content, but the maximum mass loss rate decreases after further processing, even though the cellulose content increases. This data can be explained by taking into consideration the interactions between the fiber components along with the fiber composition. Based on the microscopy data and wood fiber morphology 30, a schematic illustration of fiber defibrillation is proposed in Figure 2.9 to explain the thermal behavior. In the illustration, brown color represents lignin, blue represents hemicellulose and green represents cellulose. The fine green features represent cellulose nano-fibrils that combine to give cellulose a micro-fibril.

The micro-fibrils bundle to give a macro-fiber or cellulose fiber bundles. The microscopic images

show that P-coir fiber has a cellular cross-section with lignin concentrated on the cell walls. A further magnified illustration of the cell wall proposes cellulose fiber bundles are connected via hemicelluloses to lignin within the spaces. Upon hydrolysis, most of the hemicelluloses are removed, leaving behind cellulose fiber bundles with detached lignin, as can be seen from the confocal image, where lignin redistributes inside and outside the fiber walls as lignin particles.

This explains the fast degradation with low FWHM and a higher degradation temperature. After microemulsion treatment, the lignin, redistributes again sticking on to the cellulose fiber bundles as is evident from the decrease in concentration of the lignin particles. We speculate that the residual lignin adhered onto the cellulose fiber bundles causes a slow degradation rate and high

FWHM. The microfluidization process defibrillates the fibers into micro and nano-fibrils with two structural populations. However, the micro-fibrils still contain lignin, as seen from the corresponding confocal images, while less hemicelluloses are present (a fact that is corroborated by chemical analyses but nearly impossible to determine by the imaging techniques used here).

The homogeneous distribution of lignin in the entire sample further reduces the degradation rate and increases the FWHM.

2.5 Conclusions

Lignocellulosic micro- and nano-fibrils (MNFC) were isolated from coir and EFB fibers by using low severity treatments, which allowed to retain most of the native lignin (31 and 24 wt%, respectively). The fiber processing steps included autohydrolysis, mechanical defibrillation, microemulsion treatment and microfluidization. The fibers were characterized during each treatment step for their chemical, structural and thermal properties. The FTIR analysis confirmed no significant changes in the chemical groups of the fiber components. The electron microscopy and AFM images indicated two structural populations in coir and EFB- MNFC with fibril diameter of 20-70 nm and 1- 3 µm. The confocal images highlighted the lignin redistribution on the fiber walls and within the hollow fiber in form of particles. As expected, the crystallinity of the fibers was not affected by the treatments applied, due to presence of substantial amounts of amorphous lignin. However, the thermal degradation behavior was largely affected by the chemical composition and the redistribution of lignin during autohydrolysis and microemulsion treatments.

These results, along with the features observed in the morphological analyses, can be combined to postulate the possible structural and component changes that occurred upon defibrillation. We expect that the detailed structural and thermo-chemical analyses presented here facilitate further interest for preparation of new materials from MNFC isolated from coir and EFB, two abundant bioresources that are most suitable for their valorization.

2.6 References

[1] Satyanarayana KG, Guimaraes JL, Wypych F. Studies on lignocellulosic fibers of Brazil. Part I: Source, production, morphology, properties and applications. Compos Part A Appl Sci Manuf 2007;38:1694–709. doi:10.1016/j.compositesa.2007.02.006.

[2] Rhim J-W, Park H-M, Ha C-S. Bio-nanocomposites for food packaging applications. Prog Polym Sci 2013;38:1629–52. doi:10.1016/j.progpolymsci.2013.05.008.

[3] Nagarajan V, Mohanty AK, Misra M. Sustainable green composites: Value addition to agricultural residues and perennial grasses. ACS Sustain Chem Eng 2013;1:325–33. doi:10.1021/sc300084z.

[4] Ferrer A, Filpponen I, Rodríguez A, Laine J, Rojas OJ. Valorization of residual Empty Palm Fruit Bunch Fibers (EPFBF) by microfluidization: Production of nanofibrillated cellulose and EPFBF nanopaper. Bioresour Technol 2012;125:249–55. doi:10.1016/j.biortech.2012.08.108.

[5] Abraham E, Deepa B, Pothen L a., Cintil J, Thomas S, John MJ, et al. Environmental friendly method for the extraction of coir fibre and isolation of nanofibre. Carbohydr Polym 2013;92:1477–83. doi:10.1016/j.carbpol.2012.10.056.

[6] Van Dam JEG. Coir processing technologies: improvement of drying, softening, bleaching and dyeing coir fibre/yarn and printing coir floor coverings. Common Fund Commod 2002:51.

[7] Célino A, Fréour S, Jacquemin F, Casari P. The hygroscopic behavior of plant fibers: a review. Front Chem 2014;1:1–12. doi:10.3389/fchem.2013.00043.

[8] Geethamma VG, Joseph R, Thomas S. Short Coir Fiber-Reinforced Natural-Rubber Composites - Effects of Fiber Length, Orientation, and Alkali Treatment. J Appl Polym Sci 1995;55:583–94. doi:DOI 10.1002/app.1995.070550405.

[9] Harish S, Michael DP, Bensely A, Lal DM, Rajadurai A. Mechanical property evaluation of natural fiber coir composite. Mater Charact 2009;60:44–9. doi:10.1016/j.matchar.2008.07.001.

[10] Haque MM, Hasan M, Islam MS, Ali ME. Physico-mechanical properties of chemically treated palm and coir fiber reinforced polypropylene composites. Bioresour Technol 2009;100:4903–6. doi:10.1016/j.biortech.2009.04.072.

[11] Ayrilmis N, Jarusombuti S, Fueangvivat V, Bauchongkol P, White RH. Coir fiber reinforced polypropylene composite panel for automotive interior applications. Fibers Polym 2011;12:919–26. doi:10.1007/s12221-011-0919-1.

[12] Ferrer A, Vega A, Ligero P, Rodríguez A. Pulping of empty fruit bunches (EFB) from the palm oil industry by formic acid. BioResources 2011;6:4282–301.

[13] FAO, Food and Agriculture Organization of the United Nations.Available from: (accessed on September 30, 2016) n.d.

[14] Ferrer A, Salas C, Rojas OJ. Dewatering of MNFC containing microfibrils and microparticles from soybean hulls: mechanical and transport properties of hybrid films. Cellulose 2015;22:3919–28. doi:10.1007/s10570-015-0768-y.

[15] Moran JI, Alvarez VA, Cyras VP, Vazquez A. Extraction of cellulose and preparation of nanocellulose from sisal fibers. Cellulose 2008;15:149–59. doi:10.1007/s10570-007-9145- 9.

[16] Nascimento DM, Almeida JS, Dias AF, Figueirêdo MCB, Morais JPS, Feitosa JP a, et al. A novel green approach for the preparation of cellulose nanowhiskers from white coir. Carbohydr Polym 2014;110:456–63. doi:10.1016/j.carbpol.2014.04.053.

[17] Deepa B, Abraham E, Cordeiro N, Mozetic M, Mathew AP, Oksman K, et al. Utilization of various lignocellulosic biomass for the production of nanocellulose: a comparative study. Cellulose 2015;22:1075–90. doi:10.1007/s10570-015-0554-x.

[18] Abraham E, Deepa B, Pothan L a., Jacob M, Thomas S, Cvelbar U, et al. Extraction of nanocellulose fibrils from lignocellulosic fibres: A novel approach. Carbohydr Polym 2011;86:1468–75. doi:10.1016/j.carbpol.2011.06.034.

[19] Rojo E, Peresin MS, Sampson WW, Hoeger IC, Vartiainen J, Laine J, et al. Comprehensive elucidation of the effect of residual lignin on the physical, barrier, mechanical and surface properties of nanocellulose film. Green 2015:1853–66. doi:10.1039/c4gc02398f.

[20] Ferrer A, Quintana E, Filpponen I, Solala I, Vidal T, Rodríguez A, et al. Effect of residual lignin and heteropolysaccharides in nanofibrillar cellulose and nanopaper from wood fibers. Cellulose 2012;19:2179–93. doi:10.1007/s10570-012-9788-z.

[21] Arayaa F, Troncosob E, Mendonçab RT, Freera J, Rencoretd J, Del Riaod Carlos J. Structural Characteristics and Distribution of Lignin in Eucalyptus Globulus Pulps Obtained by a Combined Autohydrolysis/Alkaline Extraction Process for Enzymatic Saccharification of Cellulose. J Chil Chem Soc 2015;2:2954–60.

[22] Carrillo CA, Laine J, Rojas OJ. Microemulsion Systems for Fiber Deconstruction into Cellulose Nano fi brils. Appl Mater Interfaces 2014;6:22622–7. doi:10.1021/am5067332 |.

[23] Ferrer A, Vega A, Rodríguez A, Ligero P, Jiménez L. Milox fractionation of empty fruit bunches from Elaeis guineensis. Bioresour Technol 2011;102:9755–62. doi:10.1016/j.biortech.2011.07.037.

[24] Wise L, Murphy M, Daddieco A. Chlorite Holocellulose, its fractionation and bearing on summative wood analysis and on studies on the hemicelluloses. Tech Assoc Pap 1946;29:210–8.

[25] Castro JF, Parra C, Yáñez-S M, Rojas J, Teixeira Mendoncìa R, Baeza J, et al. Optimal

pretreatment of Eucalyptus globulus by hydrothermolysis and alkaline extraction for microbial production of ethanol and xylitol. Ind Eng Chem Res 2013;52:5713–20. doi:10.1021/ie301859x.

[26] Jonoobi M, Harun J, Shakeri A, Misra M, Oksmand K. Chemical composition, crystallinity, and thermal degradation of bleached and unbleached kenaf bast (Hibiscus cannabinus) pulp and nanofibers. BioResources 2009;4:626–39. doi:10.15376/biores.4.2.626-639.

[27] Oh SY, Yoo D Il, Shin Y, Seo G. FTIR analysis of cellulose treated with sodium hydroxide and carbon dioxide. Carbohydr Res 2005;340:417–28. doi:10.1016/j.carres.2004.11.027.

[28] Yang H, Yan R, Chen H, Lee DH, Zheng C. Characteristics of hemicellulose, cellulose and lignin pyrolysis. Fuel 2007;86:1781–8. doi:10.1016/j.fuel.2006.12.013.

[29] Alemdar A, Sain M. Isolation and characterization of nanofibers from agricultural residues - Wheat straw and soy hulls. Bioresour Technol 2008;99:1664–71. doi:10.1016/j.biortech.2007.04.029.

[30] Ilvessalo-Pfäffli M-S. Fiber Atlas- Identification of Papermaking Fibers. Springer Series in Wood Science; 1995. doi:10.1007/978-3-662-07212-7.

[31] Araya F, Troncoso E, Mendonca RT, Freer J. Condensed Lignin Structures and Re- Localization Achieved at High Severities in Autohydrolysis of Eucalyptus globulus Wood and their Relationship with Cellulose Accessibility. Biotechnol Bioeng 2015;112:1783–91. doi:10.1002/bit.25604.

[32] Donohoe BS, Decker SR, Tucker MP, Himmel ME, Vinzant TB. Visualizing lignin coalescence and migration through maize cell walls following thermochemical pretreatment. Biotechnol Bioeng 2008;101:913–25. doi:10.1002/bit.21959.

[33] Sannigrahi P, Kim DH, Jung S, Ragauskas A. Pseudo-lignin and pretreatment chemistry. Energy Environ Sci 2011;4:1306–10. doi:10.1039/c0ee00378f.

[34] Kumar R, Hu F, Sannigrahi P, Jung S, Ragauskas AJ, Wyman CE. Carbohydrate derived- pseudo-lignin can retard cellulose biological conversion. Biotechnol Bioeng 2013;110:737– 53. doi:10.1002/bit.24744.

[35] Hu F, Jung S, Ragauskas A. Pseudo-lignin formation and its impact on enzymatic hydrolysis. Bioresour Technol 2012;117:7–12. doi:10.1016/j.biortech.2012.04.037.

[36] Hu F, Jung S, Ragauskas A. Impact of pseudolignin versus dilute acid-pretreated lignin on enzymatic hydrolysis of cellulose. ACS Sustain Chem Eng 2013;1:62–5. doi:10.1021/sc300032j.

[37] French AD. Idealized powder diffraction patterns for cellulose polymorphs. Cellulose 2014;21:885–96. doi:10.1007/s10570-013-0030-4.

[38] Jin W, Singh K, Zondlo J. Pyrolysis Kinetics of Physical Components of Wood and Wood-

Polymers Using Isoconversion Method. Agriculture 2013;3:12–32. doi:10.3390/agriculture3010012.

[39] Ferrer A, Salas C, Rojas OJ. Physical, thermal, chemical and rheological characterization of cellulosic microfibrils and microparticles produced from soybean hulls. Ind Crops Prod 2016;84:337–43. doi:10.1016/j.indcrop.2016.02.014.

[40] Raveendran K, Ganesh A, Khilar KC. Pyrolysis characteristics of biomass and biomass components. Fuel 1996;75:987–98. doi:10.1016/0016-2361(96)00030-0.

Chapter 3: Aerogel Synthesis from Non-aqueous Polymers

3.1 Abstract

Aerogels are one of the lightest man-made materials. We study single component aerogels from non-aqueous polymers in terms of their synthesis and structure-property relationships. The two methods to synthesize aerogels are studied: Freeze Drying (FD) or lyophilization and Critical

Point Drying (CPD) using super critical CO2. The synthesis of aerogels, from non-aqueous polymers such as cellulose esters, through FD is challenging because most organic solvents exhibit extremely low triple point that is difficult to achieve through FD. This challenge is overcome by employing a unique combination of synthesis procedures involving chemical cross-linking, solvent exchange to water and subsequent FD to produce ultra-light (4.3 mg/ml) and highly porous

(99.7 %) cellulose acetate aerogels (CA) with honeycomb morphology. This versatile synthesis approach is extended to other non-aqueous polymers such as cellulose acetate propionate (CAP) and cellulose acetate butyrate (CAB) to produce single component polymer aerogel. On the contrary, the aerogels produced via CPD exhibit high shrinkage (50-70%) and consequently higher densities (~70 mg/ml). The reason for such shrinkage is investigated through concepts of solubility parameters.

3.2 Introduction

3.2.1 In context to the dissertation

In previous chapter, we explored the interaction of cellulose, lignin, and hemicellulose in the biomass. For commercial use, the cellulose is isolated from ligno-cellulose biomass and chemically modified to enhance its properties. One such class of commercial modified cellulose is cellulose esters. In this chapter, we use cellulose esters to introduce a novel synthesis route to make the lightest man-made materials called aerogels.

3.2.2 Why Aerogels?

Kistler, in 1931, prepared the first aerogel by supercritical drying of solvent in the gel.1

Subsequent development of supercritical CO2 drying process made aerogel synthesis relatively fast and safer.2 However, it was not until NASA’s stardust mission that captured high speed cosmic particles using silica aerogels, that aerogels gained media attention in 2006.3 Many novel aerogels have been synthesized since then, including carbon nanotube4,5, graphene6,7, quantum dots8, polyimides9, cellulose10 and nanocellulose11,12. The definition of aerogels also developed over time. However, an important feature shared by all aerogels is the removal of solvent from wet gel with minimum shrinkage to give high pore volume, usually higher than 90%.13,14

Even though a plethora of aerogels have been reported in the last decade, not much work has been done on single component polymer aerogels. Polymers have been used largely as a reinforcing material to improve elastic properties of inorganic aerogels.15–17 And although single component aerogels from aqueous based polysaccharides such as alginate, pectin, starch, chitin, chitosan, agar and cellulose have been synthesized, their wet strength is questionable14,18.

Polyurethane aerogels are one of the earliest synthetic polymer aerogels reported that are formed by polycondensation of isocyanates and alcohols.19 The isocyanates are however generally toxic and very reactive, that may sometimes prove hazardous to the surroundings.20 Meador and co- workers have worked extensively with non-aqueous polyimide aerogels and have demonstrated improved mechanical properties for these aerogels.9,21 In this chapter, we demonstrate the single component aerogel synthesis from a different class of polymers namely, cellulose esters that are non-aqueous polymers with hydroxyl functionality.

Cellulose esters are polymers derived from the abundantly-available cellulose, but are easy to process than their parent cellulose due to reduced intramolecular hydrogen bonding.22 The aerogel synthesis approach is developed on cellulose acetate which is extended to cellulose acetate

propionate (CAP), and cellulose acetate butyrate (CAB) exhibiting a wide-range capability of this

approach. Surprisingly, there have been few attempts at making aerogels from cellulose esters.

Tan et al.23, prepared aerogel by urethane linkage of cellulose acetate and cellulose butyrate

followed by CPD. The study was later reproduced by Rigacchi and co-workers with cellulose

acetate aerogels.24,25 But, both these efforts produced aerogels with high density (>0.1 g/ml) and

low pore volume (<90%). The synthesis of non-aqueous polymers such as cellulose esters is

challenging via FD process and has not been attempted before this study. The extremely low

pressure and temperature at the triple point of organic solvents, used to dissolve non-aqueous

polymers, is usually difficult to achieve in the commercial freeze drying apparatus.

In this chapter we introduce a unique combination of synthesis procedures involving

chemical cross-linking, careful solvent exchange to water and subsequent freeze drying to produce

ultra-light (4.3 mg/ml) ever reported from cellulose esters. Additionally, we also explore the CPD

process to produce aerogels to have a quantitative comparison of the properties of the aerogels

produced from the two processes.

3.3 Experimental Section

3.3.1 Materials

Cellulose diacetate (acetyl: 39.8%, hydroxyl: 3.5%), cellulose acetate propionate

(propionyl: 42.5%, hydroxyl: 5%) and cellulose acetate butyrate (butyryl: 46%, hydroxyl: 4.8%) were provided by Eastman Chemical Company and used as received. Acetone (99 %), ethanol, triethyl amine (TEA) and the cross-linking agent 1,2,4,5-benzenetetracarboxylic acid (also known as pyromelletic dianhydride, PMDA) were purchased from Sigma Aldrich. Deionized (DI) water was used. Dry ice was prepared in the laboratory using a liquid CO2 cylinder with siphon tube

bought from Airgas (NC). All solvents used for wicking measurements were 99 % pure and bought from Sigma Aldrich.

3.3.2 Gelation of Cellulose esters

Cellulose diacetate (CDA) gels were synthesized using PMDA cross-linker as reported

earlier.26 Briefly, a homogeneous solution of CDA in acetone was formed by stirring it in a 100

ml Pyrex bottle for 24 h. The stoichiometric amount of PMDA required for complete cross-linking

was calculated by assuming one PMDA molecule reacts with two hydroxyl groups on different

CDA chains (Figure 3.3a). The CDA: PMDA molar ratio of 2:1 is required for complete cross-

linking, but for this study CDA: PMDA molar ratio of 8:1 was used to prevent formation of a rigid

cross-linked structure and to have free hydroxyl groups available for further modification. CDA

solutions (2, 4, 6 and 8 w/v%) with the PMDA cross-linker were stirred for approximately 5 h to

ensure complete dissolution. To this solution, 0.5 v% of catalyst triethyl amine (TEA) was added

while stirring for another 30 s. The solution was then transferred to a cylindrical mold and allowed

to set into a gel for 24 h. Thereafter, the obtained organogels, hydrogels and aerogels are referred

to by using the concentration of the initial CDA solution (2, 4, 6 and 8 %). The 4% cellulose acetate

propionate (CAP) and cellulose acetate butyrate (CAB) gels were prepared similarly.

3.3.3 Solvent Exchange

The acetone-gelled cellulose ester was subjected to sequential solvent exchange steps to

gradually replace acetone with DI water. The gel was placed in a mixed solution of acetone and

DI water; the volume of the DI water was 5 times the volume of the gel and was replaced every 12

h to allow enough time for the gels to reach equilibrium with the solution. A total of six such

exchanges were performed with the following ratio of acetone: DI water used in sequence: 90:10,

75:25, 50:50, 25:75, 10:90, 0:100. After the final exchange, the gel was kept in DI water for 24 h.

This gel in water is termed as hydrogel. This hydrogel was subjected to Freeze drying.

3.3.4 Aerogel synthesis a) Freeze Drying (FD)

The hydrogels were frozen by completely immersing the gel in a dry ice/ethanol bath for

20 min. The frozen hydrogel was then transferred to a lyophilizer (Labconco FreeZone 2.5 Freeze

Dryer) operating at -53 0C and 0.113 mbar which is lower than the triple point of water.27 The frozen hydrogel was dried for ~24 h to obtain the aerogel. b) Critical Point Drying (CPD)

The organogel was solvent exchanged with pure acetone 3 times to remove unreacted species and catalyst. Thereafter, the organogel was placed in the critical point dryer. A Leica EM

CPD300 critical point dryer was used for aerogel drying (Figure 3.1a). The gel were cut into cuboidal or cylindrical shapes of known size and placed in the chamber. The procedure for CPD is illustrated in Figure 3.1b. The liquid CO2 was filled slowly in the chamber and allowed to exchange with the acetone in the organogel for few minutes before purging out the solvent. This process is termed as one cycle of liquid CO2 exchange. Such exchanges were varied from 10 to 50 cycles for same volume of the organogel sample. The mass and volume of the resulting aerogel was measured.

Figure 3.1: a) Leica CPD 300 used for critical point drying (CPD), b) Steps involved in the CPD procedure of an organogel

3.3.5 Aerogel Characterization a) Density and pore volume

Organogel density: If there was no swelling or shrinkage during the solvent exchange and freeze drying step, the density can be calculated as shown in Equation 3.1.

M  = CDA ---Equation 3.1 organogel M (1− M ) CDA + CDA CDA  water where, ρcal = calculated density of the aerogel, MCDA = weight % of the CDA added for aerogel synthesis -3 ρCDA = density of the cellulose diacetate flakes (1.3 g cm ) -3 ρwater = density of DI water (0.998 g cm )

Hydrogel density: The hydrogel density is calculated by measuring volume of each hydrogel via volume displacement using deionized water in a flat based beaker. Assuming, there was no shrinkage during freeze drying, the hydrogel density is calculated as shown in Equation 3.2:

M CDA hydrogel = ---Equation 3.2 Vhydrogel

Aerogel density: Aerogel density (ρa) was calculated by measuring its mass and volume. The aerogel was cut into cuboidal shape using a sharp clean blade. The mass of this cut aerogel was measured by analytical balance, Fisher Scientific Accu-225D, which has least count of 0.1 mg and the volume was determined by using the dimensions (digital Vernier caliper). It is noted that the

2% aerogel was cut into small cuboidal shapes to measure its volume. Average density is reported after 5 measurements. The pore volume of the aerogels was calculated using Equation 3.3.

Pore Volume = (1-ρa/ρCE)*100% ---Equation 3.3 where ρa is the bulk density of aerogel ρCE is the bulk density of cellulose esters, 1.3 g /ml for

CDA, 1.27 g/ml for CAP and 1.2 g/ml for CAB.28 b) Surface area

The Brunauer-Emmett-Teller (BET) surface area was measured by N2 absorption and desorption isotherms using Micrometrics ASAP 2020. About 0.1-0.2 g of sample was first degassed for 3 h at 115 0C prior to the analysis. BET analysis was done for a relative pressure of

0.01- 0.3 at -196 0C.

The surface area for FD aerogel was also calculated theoretically by assuming that the aerogel is composed of cylindrical pores with diameter ranging from 50-100 μm with wall thickness of 1 μm as seen from SEM (Figure 3.4c inset), we can calculate the range of surface area as follows:

1 1 gm of 4% aerogel has volume V = 1/  = cm3 aerogel 0.0234 Total surface area A = 2rhN ---1 1 V = = r 2 hN + 2rhNw ---2 0.0234

From (1) & (2),

2 A = m2 / g = 1.6 - 3.2 m2 g-1 0.0234(r + 2w) where, r= pore size radius = 25-50 μm h = height of one cylindrical pore N = total number of pores w = wall thickness of the pore ~ 1 μm c) Scanning Electron Microscopy

Imaging was done by Field Emission Scanning Electron Microscope (FESEM), FEI Verios

460L. The aerogels were fractured under liquid N2 using a sharp clean blade to image the radial cross-section. The samples were fixed on the metal stub using a double sided carbon tape. The as- prepared SEM samples were coated with a 5 nm layer of gold and platinum to capture secondary electrons from the surface and thus reduce charging. d) Mechanical Compression testing

The freeze dried aerogels synthesized for compression testing were molded in 20ml syringes with a height: diameter ratio of 2:1. The top and bottom part of a cylindrical aerogel was made smooth by using a sharp clean blade. Compressive stress-strain curves were obtained using an Instron Series IX with compressive loads of 0.5 N, lowered at the rate of 5 mm min-1. e) Chemical analyses

Fourier Transform Infrared Spectroscopy was conducted in the Attenuated Total

Reflectance mode (FTIR-ATR) was conducted using a Perkin Elmer spectrophotometer. Samples were analyzed using the Pike Miracle accessory equipped with a GE crystal. The spectrum was collected for 256 scans and corrected for background noise. The multipoint baseline correction was realized for each spectrum.

3.4 Results and Discussion:

3.4.1 Aerogel synthesis via Freeze Drying (FD)

The aerogel formation from a non-aqueous polymer via FD process is challenging. Most of the organic solvents have triple points at an extremely low pressure and temperature, which is difficult to attain. To overcome this challenge, we developed an innovative process of aerogel formation from a non-aqueous polymer that occurs in a three-step sequence (Figure 3.2) starting with the synthesis of the organogel (gel in acetone) via chemical cross-linking, followed by solvent exchange into a hydrogel (gel in water) and then converting into the aerogel via Freeze Drying

(FD).

Figure 3.2: Illustration of the three-step sequence of aerogel formation from a non-aqueous polymer via freeze-drying (FD).

a) Confirmation of the cross-linking

Figure 3.3a shows the first step in the process in which the organogel is formed by the cross-linking reaction via ester linkages between hydroxyl groups in CDA and anhydrides of

PMDA. While the formation of strong, self-standing gels at low CDA concentration (4 w/v%)

(Figure 3.3a) indicates good cross-linking, direct evidence comes from comparison of FTIR spectra (Figure 3.3b) of the aerogel with that of CDA flakes. The main differences between the two FTIR spectra are (a) out of plane angular vibrations of aromatic sp2 C-H bends (from 690 cm-

1 to 900 cm-1) and, (b) aromatic C=C stretch (~1500 cm-1). These two peaks indicate the presence of aromatic cross-linker in the aerogel. Also, the lack of paired bands for C=O stretch (indicative

of anhydride groups between 1800 cm-1-1830 cm-1 and 1740 cm-1-1775 cm-1) suggests that all

PMDA reacted with the CDA hydroxyl groups. A slight broadening of O-H stretch at 3476 cm-1 is assigned to the increased number of O-H bonds via the carboxylic groups formed. The strong acetyl C-O and alkoxy C-O peaks at 1220 cm-1 and 1032 cm-1, respectively indicate the presence of C-O-C stretching in the glucopyranose ring. These peaks combined with prominent carbonyl peak (C=O stretch) at 1736 cm-1, confirms the presence of ester groups in CDA. The peaks at around 2950 cm-1 are due to sp3 C-H stretch and the one at 1368 cm-1 corresponds to C-H bend.

Figure 3.3: a) Proposed cross-linking reaction between CDA chains and cross-linker, PMDA and the image showing the gelation of CDA solution into acetone-based organogel after the gel was set for 24 h, b) FTIR spectra of CDA flakes and aerogel.

b) Aerogel Characterization

We next examine to what extent the organogel swells (or shrinks) as we transition it to its final aerogel form via the intermediate hydrogel state, by measuring its density at each stage. Such information is important to understand the underlying structural changes a material undergoes during aerogel formation, an area in which very little work exists. Table 3.1 shows the densities of the organogel, hydrogel and aerogel together with the aerogel pore volume which are further depicted in Figure 3.4a. The organogel and hydrogel densities were calculated by assuming that all the liquid was replaced with air without changing the volume of the gel.

Table 3.1: Density values and aerogel pore volume

2% 4% 6% 8% Organogel density (mg/ml) 20.0 40.3 60.7 81.4 Hydrogel density (mg/ml) -- 19.3 ±1.6 50.8 ± 1.2 93.3 ± 2.1 Aerogel density (mg/ml) 4.3 ± 0.7 23.4 ± 1.0 77.2 ± 1.6 110.2 ±1.2 Aerogel pore volume (%) 99.7 ± 0.1 98.2 ± 0.1 94.1 ± 0.1 91.2 ± 0.1

Several features are evident from the data. First, the bulk densities of the 2 % and 4 % aerogels are much lower than the calculated organogel density, unlike that for the 6 % and 8 % aerogels. The smaller density is likely due to extensive swelling of the gels during gradual solvent exchange with water, along with small shrinkage during freeze drying. Second, we find (Figure

3.4a) that the 4 % hydrogel does not exhibit appreciable shrinkage during freeze drying, as indicated from a very small increase in the aerogel density from hydrogel density. In contrast, the

6 % and 8% hydrogels exhibits a large shrinkage during freeze drying. Third, the 8 % CDA gels seem to shrink during solvent exchange (whereas the 2, 4 and 6 % gels swell during the solvent exchange). This observation is consistent with the strong intermolecular hydrogen bonding exhibited by cellulose diacetate, which is expected to be stronger at the high CDA concentration

that limits solvent penetration. Finally, due to its extensive swelling, the 2 % aerogel is ultra-light and highly porous (density of 4.3 mg/ml and pore volume of 99.7%, Figure 3.4b). This is one of the lightest reported cellulose based aerogels (see Table A3.1 for comparison).

The morphology of the aerogels were examined using SEM. The radial cross-section of 4

% aerogels show irregularly-shaped pores formed by film-like walls of assembled CDA (SEM image in Figure 3.4c, inset). The image reveals that the pore size is dictated by the rate of freezing.22 Water in the hydrogels nucleate forming ice crystals that squeeze out the cross-linked cellulose diacetate polymer and compress them into thin walls. The crystals then sublime during the drying process leaving behind cylindrical pores with an average diameter of 50 μm. The CDA aerogels exhibit a low BET surface area (3.4 m2/g) as compared to aerogels obtained from other cellulosic sources likely due to the large pore size of the aerogels. Figure A3.1 shows the N2 adsorption-desorption isotherm for the aerogel along with the pores size distribution in the inset.

The lack of any distinct peak below 500 Å indicates lack of micro or meso porosity. Assuming that the CDA aerogel is composed of cylindrical macropores, with a diameter range of 50-100 μm running throughout the aerogel along the radial axis, and the wall thickness of 1 μm (Figure A3.2), we calculate the surface area to be in the range of 1.6-3.2 m2/g, which agrees well with the value measured by N2 adsorption. This agreement further supports the lack of micro porosity in the aerogel.

The porous morphology, including thin-walled structures observed in the cross-linked

CDA aerogels results in highly compressible systems (see compressive stress-strain curve of 4 w/v% CDA aerogel, Figure 3.4c). Cross-linking a gel with relatively high polymer concentration gives an aerogel with a maximum compressive strength of 350 kPa, which is larger than the values reported for cellulosic aerogels (see Table A3.2). In contrast to the brittle behavior of silica

aerogels29, the 4 % CDA aerogel exhibits a high compression strain (92 %) without failure. The high compression strains are generally comparable to carbon nanofiber aerogels and are higher than those of cellulosic aerogels reported in the literature.12,30–33 Interestingly, the 4 % aerogel does not exhibit any yielding, suggesting that the relatively high concentrations of CDA used during synthesis (4 w/v%), renders strength and high compressibility to the aerogel.

Given the high pore volume and closed cell morphology, the CDA aerogels are expected to perform ideally for liquid uptake, as confirmed experimentally (Figure 3.4d) for oil and water uptake. The 2 % aerogels showed the highest water sorption of 92 and 112 g/g of aerogel respectively, although capillary forces caused their collapse during evaporation. The 4, 6 and 8 %

CDA aerogels, however, were strong enough to sustain the capillary forces. The liquid uptake recorded for CDA aerogels in this study is comparable to the liquid uptake reported for nanocellulose aerogels (see Table A3.1 and references therein).

Figure 3.4: a) Comparison of measured and calculated density of organogels and hydrogels assuming that all solvent was replaced with air while keeping the size of the gel. The data for 2% hydrogel density is not reported because the high swelling of the respective organogel during solvent exchange made the hydrogel fragile for handling during measurements, b) image of the 2% aerogel (density 4.3 mg/ml) on a dandelion leaf, c) Compressive stress-strain profile for 4% aerogels with the maximum compressive stress of 350 kPa and maximum strain of 92%. It is to be noted that the desired shape of the 2% aerogel was difficult to synthesize due to breakage of the gel during the sequential solvent exchange process. Hence, compression testing of 2% aerogel was not performed. Along the same lines, due to the relatively high concentration of polymer in 6 and 8 % CDA gels, the final aerogel product for these gels have unavoidable artifacts during solvent exchange and freezing step. The stress-strain curve for the 6% and 8% aerogels are not reported here. However, we found that both the 6% and 8% aerogels consistently yielded below 30% strain (Figure A3.3). Inset SEM image shows the radial cross-section of 4% CDA aerogel, and d) water uptake and kerosene oil uptake (g/g of aerogel) as exhibited by the different aerogels.

Figure A3.4 shows the liquid uptake (oil and water) relative to the available pore volume.

The marginally higher water uptake compared to the available pore volume in the 4, 6 and 8% aerogels suggests a significant finding in that sorption may takes place in both the pores and on the polymer. Moreover, a ratio of oil uptake less than 1, may imply entrapment of air bubbles during sorption. c) Aerogels from other non-aqueous polymers

To demonstrate versatility of our aerogel synthesis approach, cellulose acetate propionate

(CAP) and cellulose acetate butyrate (CAB) aerogels were synthesized. Both these are non- aqueous polymers where some of the hydroxyl groups are replaced by propionate and butyrate functional groups. The SEM images of 4% CAP and CAB aerogel are shown in Figure 3.5a and b respectively which exhibits similar honeycomb structure as observed before for CDA aerogel.

Yet again, we were able to prepare one of the lightest single component polymer aerogels starting from a relatively high concentration of 4 w/v%. The density and pore volume of CAP was measured to be 13.0 ± 0.9 mg/ml and 99.0 ±0.1% respectively. The density of CAB was slightly on the higher side, 27.9 ± 3.2 mg/ml but still 10 times lower than reported for earlier cellulose ester based aerogels.23,24 Compared to CDA aerogels, the 4% CAP and CAB aerogel exhibit lower water uptake (Figure 3.5c) due to increased hydrophobicity arising from longer carbon chain functional groups, namely propionate and butyrate. The oil uptake is highest for CAP due to largest available pore volume, validated from Figure 3.5d where the liquid uptake is normalized with available pore volume and for oil uptake the ratio is ~0.8 for all three types of aerogels. The similar morphological and bulk characteristics of CAP and CAB aerogels to CDA aerogels suggests that the mechanical properties of CAP and CAB aerogels will not be much different from that of CDA aerogels. Hence, they were not measured.

3.4.2 Aerogel synthesis via Critical Point Drying (CPD)

Figure 3.6a shows the camera image of organogel and the corresponding aerogel obtained from CPD. We observe that a translucent gel turns opaque after transitioning to aerogel. The reason for this transition is discussed later. The density of these aerogels is surprisingly uniform with varying CDA concentration, ranging from 65- 75 mg/ml (Figure 3.6b), which is 4 times as high when compared to the density of aerogel as obtained from FD. On the contrary, the 8% aerogel from FD has higher density compared to the aerogel obtained from CPD. The density of 6%

aerogel from both the procedure was almost similar. This indicates that the 4% gel undergoes large shrinkage than 8% gel during CPD. There could be two possible reasons for high shrinkage during

CPD. First, incomplete removal of acetone can cause structural collapse during venting of CO2 in supercritical state, and second, organogel shrinkage during solvent exchange with liquid CO2. To test first hypothesis, different samples of aerogel were made from increasing the cycles of liquid

3 CO2 exchanges. The density of the aerogel obtained from 1 cm of 4% CA gel is shown in Figure

3.6c. The red data shows the organogel density (the hypothetical density if there was no gel shrinkage), whereas the black data shows aerogel density. The uniform density across all cycles of

CO2 exchanges proves that the aerogel shrinkage is not due to incomplete removal of acetone.

Additionally, we can confirm that 10 cycles of liquid CO2 exchange is enough to completely dry

1 cm3 of organogel.

To explore the second hypothesis of gel shrinkage during the solvent exchange with liquid

CO2, we invoke the idea of Hansen solubility parameters. The RED value much greater than 1 will confirm that liquid CO2 is not the “good solvent” for CDA, which causes gel shrinkage during solvent exchange with liquid CO2. The Hansen solubility parameters (δd, δp, δh) of the liquid CO2 are calculated using the following equations:

δ V −1.25 dref = ( ref) ---Equation 3.4 δd V 90

δ V −0.5 pref = ( ref) ---Equation 3.5 δp V

0.5 δhref −3 Vref = exp [−1.32x10 (Tref − T) − ln ( ) ] ---Equation 3.6 δh V

The reference values of liquid CO2 are listed in the Table 3.2.

Table 3.2: The reference values of the liquid CO2 used to calculate the Hansen solubility parameters from the Equations 3.4-3.6

1/2 1/2 1/2 3 δdref (MPa ) δpref (MPa ) δhref (MPa ) Vref (cm /mol) Tref (K) 15.6 5.2 5.8 39.13 298.15

The RED values of CDA with respect to liquid CO2 for different pressures was calculated using Equation 1.5 and the resulting graph is shown in Figure 3.6d, which demonstrates that RED gets close to 1 around 700 bar pressure, indicating that the liquid CO2 may become “good solvent” for CDA at such pressure. It is expected that if the critical point dryer were to run at 700 bar, we could expect CA translucent aerogel with much lower density. The operating pressure in the critical point dryer is around 56 bar. At such pressure, the RED is close to 5. Therefore, we see a large shrinkage in the gel during solvent exchange with liquid CO2 along with high aggregation inducing opaque nature to the aerogel.

Figure 3.7a shows the SEM images of the aerogel cross-section shows an open porous aggregated structure as seen in case of spinodal decomposition.34 This further agrees to the theory of gel shrinkage that causes aggregation of polymer chains during the solvent exchange with liquid

CO2. Figure 3.7b shows the N2 adsorption-desorption isotherm for these aerogels. The isotherm identifies as Type V category isotherm as classified by IUPAC convention and the hysteresis indicates mesoporosity in the aerogel.35 The pore size distribution curve in the inset of Figure 3.7b 91

further confirms the mesoporous structure of the aerogel with pore size in the range of 20-50 nm.

The BET surface area obtained from the low relative pressure range of the isotherm (0- 0.3) is 263 m2/g, which is approximately 300 times higher than the CDA aerogels obtained from freeze drying.

Figure 3.7: a) SEM image of the cross-section of 4% CDA aerogel synthesized via CPD (scale bar is 0.5

μm), c) N2 adsorption and desorption isotherm obtained for 4% CDA aerogel prepared from critical point drying. Inset shows the pore size distribution derived from desorption isotherm by Barrett-Joyner-Halenda (BJH) method.

3.5 Conclusions

In this chapter, we laid out the synthesis procedure for aerogels using freeze drying and critical point drying. While exploring a way to make light-weight single component aerogels from polymers, we developed a versatile method to synthesize aerogel from a non-aqueous polymer and demonstrated the idea with cellulose esters. A unique combination of chemical crosslinking, solvent exchange and freeze-drying process was used to produce ultralight (4.3 mg/ml) and highly porous (99.7 %) cellulose diacetate aerogels for the first time. A low density cellulose ester aerogel

(< 30 mg/ml) was easily synthesized even with a relatively high starting concentration of the

polymer (4 w/v%). The honeycomb morphology achieved during ice-templating provided the aerogels high compressive strength (up to 350 kPa) and maximum strain of about 92%.

Additionally, the critical point drying method using supercritical CO2 was also investigated using CDA organogels. Even though such aerogels have relatively high density, 4 times high compared to the aerogels obtained from freeze drying, the critical point dried aerogels demonstrated mesoporous structure with BET surface area of 263 m2/g, which is approximately

300 times higher than the freeze-dried aerogel of CDA.

3.6 References

[1] Kistler SS. Coherent Expanded-Aerogels. J Phys Chem 1931;36:52–64. doi:10.1021/j150331a003.

[2] Tewari PH, Hunt AJ, Lofftus KD. Ambient-temperature supercritical drying of transparent silica aerogels. Mater Lett 1985;3:363–7. doi:10.1016/0167-577X(85)90077-1.

[3] Tsou P. Silica aerogel captures cosmic dust intact. J Non Cryst Solids 1995;186:415–27. doi:10.1016/0022-3093(95)00065-8.

[4] Sun H, Xu Z, Gao C. Multifunctional, ultra-flyweight, synergistically assembled carbon aerogels. Adv Mater 2013;25:2554–60. doi:10.1002/adma.201204576.

[5] Shi Z, Zhang W, Zhang F, Liu X, Wang D, Jin J, et al. Ultrafast separation of emulsified oil/water mixtures by ultrathin free-standing single-walled carbon nanotube network films. Adv Mater 2013;25:2422–7. doi:10.1002/adma.201204873.

[6] Bi H, Xie X, Yin K, Zhou Y, Wan S, He L, et al. Spongy graphene as a highly efficient and recyclable sorbent for oils and organic solvents. Adv Funct Mater 2012;22:4421–5. doi:10.1002/adfm.201200888.

[7] Jung SM, Mafra DL, Lin C-T, Jung HY, Kong J. Controlled porous structures of graphene aerogels and their effect on supercapacitor performance. Nanoscale 2015;7:4386–93.

[8] Wang H, Shao Z, Bacher M, Liebner F, Rosenau T. Fluorescent cellulose aerogels containing covalently immobilized (ZnS)x(CuInS2)1-x/ZnS (core/shell) quantum dots. Cellulose 2013;20:3007–24. doi:10.1007/s10570-013-0035-z.

[9] Guo H, Meador MAB, McCorkle L, Quade DJ, Guo J, Hamilton B, et al. Polyimide aerogels cross-linked through amine functionalized polyoligomeric silsesquioxane. ACS Appl Mater Interfaces 2011;3:546–52. doi:10.1021/am101123h.

[10] Feng J, Nguyen ST, Fan Z, Duong HM. Advanced Fabrication and Oil Absorption Properties of Super-hydrophobic Recycled Cellulose Aerogels. Chem Eng J 2015;270:168– 75.

[11] Sehaqui H, Zhou Q, Berglund LA. High-porosity aerogels of high specific surface area prepared from nanofibrillated cellulose (NFC). Compos Sci Technol 2011;71:1593–9. doi:10.1016/j.compscitech.2011.07.003.

[12] Sehaqui H, Salajková M, Zhou Q, Berglund LA. Mechanical performance tailoring of tough ultra-high porosity foams prepared from cellulose I nanofiber suspensions. Soft Matter 2010;6:1824. doi:10.1039/b927505c.

[13] Pierre AC, Pajonk GM. Chemistry of aerogels and their applications. Chem Rev 2002;102:4243–65. doi:10.1021/cr0101306.

[14] Du A, Zhou B, Zhang Z, Shen J. A special material or a new state of matter: A review and reconsideration of the aerogel. Materials (Basel) 2013;6:941–68. doi:10.3390/ma6030941.

[15] Meador MAB, Scherzer CM, Vivod SL, Quade D, Nguyen BN. Epoxy reinforced aerogels made using a streamlined process. ACS Appl Mater Interfaces 2010;2:2162–8. doi:10.1021/am100422x.

[16] Leventis N, Sotiriou-Leventis C, Zhang G, Rawashdeh AMM. Nanoengineering Strong Silica Aerogels. Nano Lett 2002;2:957–60. doi:10.1021/nl025690e.

[17] Nguyen BN, Meador MAB, Tousley ME, Shonkwiler B, McCorkle L, Scheiman DA, et al. Tailoring elastic properties of silica aerogels cross-linked with polystyrene. ACS Appl Mater Interfaces 2009;1:621–30. doi:10.1021/am8001617.

[18] García-González CA, Alnaief M, Smirnova I. Polysaccharide-based aerogels - Promising biodegradable carriers for drug delivery systems. Carbohydr Polym 2011;86:1425–38. doi:10.1016/j.carbpol.2011.06.066.

[19] Biesmans G, Randall D, Francais E, Perrut M. Polyurethane-based organic aerogels’ thermal performance. J Non Cryst Solids 1998;225:36–40. doi:http://dx.doi.org/10.1016/S0022-3093(98)00103-3.

[20] Baur X, Marek W, Ammon J, Czuppon AB, Marczynski B, Raulf-Heimsoth M, et al. Respiratory and other hazards of isocyanates. Int Arch Occup Environ Health 1994;66:141– 52. doi:10.1007/BF00380772.

[21] Meador MAB, Alemán CR, Hanson K, Ramirez N, Vivod SL, Wilmoth N, et al. Polyimide aerogels with amide cross-links: A low cost alternative for mechanically strong polymer aerogels. ACS Appl Mater Interfaces 2015;7:1240–9. doi:10.1021/am507268c.

[22] Bochek AM, Kalyuzhnaya LM. Interaction of water with cellulose and cellulose acetates as influenced by the hydrogen bond system and hydrophilic-hydrophobic balance of the macromolecules. Russ J Appl Chem 2002;75:989–93. doi:10.1023/A:1020309418203.

[23] Tan C, Fung BM, Newman JK, Vu C. Organic aerogels with very high impact strength. Adv Mater 2001;13:644–6. doi:10.1002/1521-4095(200105)13:9<644::AID- ADMA644>3.0.CO;2-#.

[24] Fischer F, Rigacci A, Pirard R, Berthon-Fabry S, Achard P. Cellulose-based aerogels. Polymer (Guildf) 2006;47:7636–45. doi:10.1016/j.polymer.2006.09.004.

[25] Rigacci A. Cellulosic aerogels for energy applications. ACS- Am. Chem. Soc. 2008 Fall Natl. ACS Meet., 2008.

[26] Oliveira VA, Veloso TC, Leão VA, Dos Santos CG, Botaro VR. Hydrogels of cellulose acetate crosslinked with pyromellitic dianhydride: Part I: Synthesis and swelling kinetics. Quim Nova 2013;36:102–6. doi:10.1590/S0100-40422013000100019.

[27] Wagner W, Saul A, Prub A. International Equations for the Pressure along the Melting and along the Sublimation Curve of Ordinary Water Substance. J Phsical Chem 1994;23.

[28] Malm CJ, Genung LB, Fleckenstein J V. Densities of Cellulose Esters. Ind Eng Chem 1947;39:1499–504.

[29] Gurav JL, Jung I-K, Park H-H, Kang ES, Nadargi DY. Silica Aerogel: Synthesis and Applications. J Nanomater 2010;2010:1–11. doi:10.1155/2010/409310.

[30] Yang X, Cranston ED. Chemically Cross-Linked Cellulose Nanocrystal Aerogels with Shape Recovery and Superabsorbent Properties. Chem Mater 2014;35:6016–25. doi:10.1021/cm502873c.

[31] Xu X, Li H, Zhang Q, Hu H, Zhao Z, Li J, et al. 3D Graphene / Iron Oxide Aerogel Elastomer Deformable in a Magnetic Field 2015:3969–77.

[32] Wu Z-Y, Li C, Liang H-W, Chen J-F, Yu S-H. Ultralight, flexible, and fire-resistant carbon nanofiber aerogels from bacterial cellulose. Angew Chem Int Ed Engl 2013;52:2925–9. doi:10.1002/anie.201209676.

[33] Pääkkö M, Vapaavuori J, Silvennoinen R, Kosonen H, Ankerfors M, Lindström T, et al. Long and entangled native cellulose I nanofibers allow flexible aerogels and hierarchically porous templates for functionalities. Soft Matter 2008;4:2492. doi:10.1039/b810371b.

[34] Kabra BG, Gehrke SH, Spontak RJ. Microporous, Responsive Hydroxypropyl Cellulose Gels. 1. Synthesis and Microstructure. Macromolecules 1998;31:2166–73. doi:10.1021/ma970418q.

[35] Kruk M, Jaroniec M. Gas adsorption characterization of ordered organic-inorganic nanocomposite materials. Chem Mater 2001;13:3169–83. doi:10.1021/cm0101069.

Chapter 4: Versatile Recipe for Aerogel Synthesis with Tunable Mechanical Properties

4.1 Abstract

We introduce a generalized approach to synthesize aerogels that allows remarkable control over its mechanical properties. The Hansen solubility parameters are used to predict and regulate the swelling properties of the precursor gels and, consequently, to achieve aerogels with tailored density and mechanical properties. As a demonstration, crosslinked organogels were synthesized from cellulose esters to generate aerogels. By determination of Hansen’s Relative Energy

Difference, it was possible to overcome the limitations of current approaches that solely rely on the choice of precursor polymer concentration to achieve a set of aerogel properties. Hence, from a given concentration, aerogels were produced in a range of mass densities, from 25 to 113 mg/cm3.

Consequently, it was possible to tailor the stiffness, toughness and compressive strength of the aerogels, in the ranges between 14-340, 4-103 and 22-373 kPa, respectively. Additionally, unidirectional freeze-drying introduced pore alignment in aerogels with honeycomb morphologies and anisotropy. Interestingly, when the swelling of the polymeric gel was arrested in a non- equilibrium state, it was possible to gain additional control of the property space. The proposed method is a novel and generic solution to achieving full control of aerogel development, which up to now has been an intractable challenge.

4.2 Introduction

4.2.1 In context to the dissertation

In Chapter 3 of this dissertation, we laid down the basic framework for aerogel synthesis using two approaches: freeze drying and critical point drying. During the course of study, we realized that using the freeze-drying approach we can synthesize aerogels from non-aqueous

polymers with densities lower than the calculated density (organogel density), which motivated us to explore the idea of tailoring the density and consequently, the mechanical properties of aerogel without changing the precursor concentration. Hence, in this chapter, we modify the properties of the aerogel by tuning the swelling ratio of the precedent polymer hydrogel utilizing the concepts of Hansen Solubility Parameters.

4.2.2 Need and Challenges for aerogels with tailorable mechanical properties

Since the introduction of aerogels in 19311, they have been synthesized from a variety of sources that include inorganic materials2, synthetic polymers3, biopolymers4 and carbons5; they have been recognized as promising systems for a plethora of applications, including membrane separation6, catalysis7, thermal and acoustic insulation8, flame resistance9 and sorption.10 They are also used in fabrication of advanced materials such as supercapacitors11, cosmic dust collectors12, drug delivery devices13, photonics14, optics15 and mechanical energy absorbers.16 Most of the applications cited require aerogels with high pore volume and low density along with high mechanical performance.

Aerogels are often produced from organogel precursors made by combining inorganic silica nanoparticles with suitable organic solvents. However these aerogels contain inherently weak interlinks within their structure that lead to overall poor mechanical performance.17 Making aerogels from inorganic-organic composites has been offered as a solution for these limitations.18

An attractive approach to make aerogels is the use of nano-materials such as nanocellulose19 and carbon nanotubes,20 which are not solubilized but dispersed in appropriate media and have shown good mechanical stability, mainly due to presence of physical entanglements that provide ample contact points. However, chemical cross-linking is often needed for such nano-materials, as demonstrated by Yang et al.21, for cellulose nanocrystals, Zhang et al.22, for cellulose nano/micro

fibrils and Zou et al.23, in case of carbon nanotube aerogels. Additionally, conventional polymers such as polyurea3 and polyimide24,25 have been used to form chemically cross-linked aerogels.

These aerogels displayed impressive mechanical strength. However, despite such advancements, there is still a major need to systematically tailor the mechanical properties of polymeric aerogels.

As a result, attempts have been made to control the density and pore volume of polymeric aerogels, mainly by changing the concentration of the precursor polymer.16,26–28 However, this cannot be universally applied, and in some cases, it is not practical. For example, a low polymer concentration can result in aerogels of low density but display a limited number of chain- interactions, compromising the mechanical performance of the aerogel. Herein, we introduce a novel approach to address these issues: while keeping a given polymer concentration, the swelling state of precursor organogel is systematically controlled, which in turn defines the strength-to- weight ratio of the ensuing aerogel.

4.2.3 Hypothesis to overcome the design challenge

We hypothesize that solvent exchange for the aerogels produced during freeze-drying can provide novel routes to create aerogels with a better control over their mechanical properties. The solvent exchange leads to either gel swelling or shrinking. Systematically choosing the exchange solvent and conditions can control the organogel volume change. By controlling organogel swelling and shrinkage, we can tailor the density of the resulting aerogel and consequently, its mechanical properties.

The main premise of this study is to understand the involved phenomenon that can facilitate the prediction and control of swelling/shrinking behavior of the precursor organogel. Thus, we performed a systematic analysis based on the Hansen solubility parameter, relevant to swelling and shrinkage of organogels during solvent exchange. The Hansen solubility parameter approach

is a relatively simple approach that accounts for polar and hydrogen bonding interactions, which are very common in the polymers of interest. In addition, the Hansen solubility parameter approach affords prediction capability in multicomponent systems29.

As a proof-of-concept, we explore this method using cellulose acetate. Cellulose acetate is a biopolymer, derived from cellulose, that exhibits polar and hydrogen bonding interactions. It also displays good wet strength, which is a great limitation for traditional cellulosic materials.30

The very limited work available on cellulose acetate aerogels reports low pore volume (41 %) and high density (0.85 g/cm3) materials after synthesis via supercritical drying.27,31 In our previous

Chapter, we synthesized ultra-light aerogels from cellulose acetate of density as low as 0.004 g/cm3. Herein, we build on that study to introduce a simple, systematic and general approach based on the selection of solvency (as defined by the solubility parameter) to tune swelling state of the resulting organogel thereby providing a means to control morphology and properties of the as- produced organic aerogel product.

Therefore, through this study, we uniquely demonstrate that combining rational solvent exchange with freeze-drying can enable synthesis of aerogels with more favorable properties than those made by more traditional routes.

4.3 Experimental Section

4.3.1 Materials

Cellulose acetate (CA) flakes provided by Eastman Chemical Co. with degree of substitution of 2.45 and acetyl content of 39.7% and were used as received. Reagent grade acetone

(99.5%), triethyl amine (TEA) and the cross-linking agent 1,2,4,5-benzenetetracarboxylic acid

(Pyromelletic Dianhydride, PMDA) were purchased from Sigma Aldrich. Deionized (DI) water with pH 6.74 was used. Liquid N2 cylinder was bought from Airgas (NC).

100

4.3.2 Organogel synthesis

Cellulose acetate (CA) gels were synthesized from a homogeneous solution of 4 wt% CA in acetone was formed by stirring in a 100 ml Pyrex bottle for 24 hours. Based on the assumption that one PMDA molecule reacts with two hydroxyl groups on different CA chains, a CA: PMDA molar ratio of 2:1 is expected for complete cross-linking. In the present study a CA: PMDA molar ratio of 8:1 was used. The molecular weight of one unit of CA with degree of substitution 2.45 was calculated as 264.6 g/mol. To obtain the gel (organogel) in acetone, the CA solution was stirred with PMDA cross-linker for approximately 5 h to ensure complete dissolution. The catalyst

(triethyl amine, TEA), 0.05 vol%, was added to the previous solution while stirring for another 30 s. A 10 ml of the solution was then transferred to a cylindrical mold and allowed to set into an elastic gel for 24 h.

4.3.3 Solvent exchange and swelling

The organogel was cut into cuboidal shape of 1 x 0.8 x 0.5 cm. In order to investigate the effect of solvency on gel swelling, acetone volume fractions (AVF) of 1, 0.9, 0.75, 0.5, 0.25 and

0 were used as media for immersion (48 h). The effect of solvent exchange time on gel swelling behavior was accessed for organogels with 0.9 AVF; the organogels were sampled at 4, 12, 24 and

48 h intervals.

The volume of the gels was measured via volume displacement method, where the gel was carefully placed in a glass cylinder filled with known solvent and the change in height of the solvent was measured via a Vernier caliper. The volumetric swelling/shrinking is reported by normalizing the volume of the organogel at time ‘t’ with respect to the initial volume of the organogel (Equation 4.1).

푉 −푉 Volumetric Swelling Ratio = 푡 0 ---Equation 4.1 푉0

101

where, Vt is volume of the gel at time ‘t’ and V0 is the initial volume.

A total of five experiments were carried out for each solvent concentration and each time interval. The average values of swelling along with the standard deviation are reported. The resulting organogels were immersed for 72 h in DI water, which was replaced every 24 h.

Thereafter, the obtained gels in water are referred to as hydrogels. The gels are abbreviated as fA, where, ‘f’ is the AVF of the solvent in which the organogel was immersed. For example, 0.9A indicates a gel that resulted from immersion of the organogel in a solvent with AVF = 0.9.

4.3.4 Aerogel synthesis

The respective hydrogels were frozen directionally to ensure pore alignment. This was achieved using a computer-controlled temperature base plate (Linkam LTS350) using cooling with

0 liquid N2 and electrically heated coils to maintain the temperature at -80 C. The frozen hydrogel were then transferred to a lyophilizer (Labconco FreeZone 2.5 Freeze Dryer) operating at -53 0C and 0.113 mbar, which is below the triple point of water.32 The frozen hydrogels were dried for

~24 h to give the CA aerogel. The label given to the aerogels use an abbreviation according to the immersion solvent (solvent exchange) for the corresponding organogel. For example, an aerogel prepared after immersing organogel in 90 vol.% acetone (AVF = 0.9) is referred to as “0.9A”.

4.3.5 Characterization a) Density and Pore Volume

Aerogel density (ρa) was calculated by measuring its mass and volume. The mass was measured by an analytical balance, Fisher Scientific Accu-225D, which has least count of 0.1 mg.

The volume was determined by measuring the dimensions using a digital Vernier caliper. Average density is reported after 5 measurements for 3 different aerogels. The % pore volume of the aerogels was calculated using Equation 4.2:

102

 Pore _Volume = (1− a ) *100% ---Equation 4.2 CDA

33 where, ρa is bulk density of aerogel ρCA is bulk density of CA flakes (1.3 g/cc). b) Turbidity measurements

Turbidity was measured by using a Thermo Scientific turbidity meter (OrionTM AQ4500).

The instrument was calibrated using 5 USEPA approved primary calibration standards. The samples were measured in transmittance mode. The values are reported in turbidity units (NTU). c) Scanning Electron Microscopy (SEM)

The imaging was performed with a Field Emission Scanning Electron Microscope

(FESEM), FEI Verios 460L. The aerogels were fractured under liquid N2 using a sharp clean blade to image the in-pane and out-of-plane cross-sections. The samples were fixed on the metal stub using a double-sided carbon tape. The as prepared SEM samples were coated with a 5-nm layer of gold and platinum to capture secondary electrons from the surface and to reducing charging. d) Mechanical Compression testing

The compressive stress-strain profiles were obtained via Instron Series IX using a compressive load of 0.5 N that was lowered at the rate of 5mm/min. The aerogels were compressed in-plane (parallel to freezing direction) and out-of-plane (perpendicular to freezing direction) directions. The compression modulus was obtained as the slope of initial linear region (at 1% strain). The energy of absorption (related to toughness) was calculated as the area under the curve, from 0 to 70% strain. The compressive strength was reported as the stress obtained at 70% strain and the densification strain was found as the x-intercept of the tangent from the densification region. Prior to compression testing, the aerogels were equilibrated for at least 48 h in a room under controlled relative humidity of 50%.

103

4.4 Results and Discussion

For this study, organogels are formed using cellulose acetate (CA) in acetone, followed by solvent exchange in a mixture of acetone and water and freeze-drying to achieve the final aerogel product. To understand the effect of the mixture of acetone-water solvent on the organogel during solvent exchange, we first describe how the Hansen solubility parameters affect the swelling of cellulose acetate. Following that analysis, we discuss the effect of swelling/shrinkage on the morphological and mechanical properties of the resulting aerogels, including an analysis of the formation of freeze-dried aerogels via unidirectional freezing to isolate the effect of freezing on the mechanical properties of the aerogel. Finally, we elaborate on the role of solvent exchange on aerogel properties by evaluating the effect of arresting the organogel swelling in a non-equilibrium state during solvent exchange.

4.4.1 Gel swelling

The Hansen solubility parameters provide a quantifiable guide to determine solubility of one material into another. In this study, we use the Hansen solubility method to understand solubility of cellulose acetate (CA), with acetyl content of 39.7%, in acetone/water solvent mixtures. A general concept of Hansen solubility parameter is described in the Chapter 1 of this dissertation. Briefly, the Hansen solubility parameter accounts for dispersion (δd), the polar (δp) and hydrogen bonding (δh) that arise, respectively, from van der Waals, dipole and hydrogen bonding interactions. These three components, the Hansen solubility parameters (HSP), can be factored in or calculated as Ra, Equation 4.3:

2 2 2 2 2 2 Ra = 4(( d1 −  d 2 )+ ( p1 −  p2 )+ ( p1 −  p2 )) …Equation 4.3 where, Ra is the difference between the HSP of a solvent (1) and a polymer (2). The constant 4 is from an empirical correlation. The solubility of the polymer in a solvent maintained if Ra < R0,

104

29 where R0 is defined as the “interaction radius” of the polymer, which is measured experimentally.

A Relative Energy Difference (RED) is defined as Ra/R0. A value of RED < 1 implies good solubility for the polymer in the given solvent.

For cellulose acetate in acetone/water mixtures, the values of the respective HSPs and R0 of the polymer is displayed in Table 4.1. The values for CA, water and acetone are determined experimentally elsewhere29.

Table 4.1: Hansen solubility parameters (HSP) corresponding to the polymer (cellulose acetate, CA) and the solvents (mixtures of water and CA).29 AVF is the Acetone Volume Fraction in water.

δ δ δ R d p h 0 (MPa)1/2 (MPa)1/2 (MPa)1/2 (MPa)1/2 CA 18.6 12.7 11.0 7.6 1.0 AVF (Acetone) 15.5 10.4 7.0 -- 0.9 AVF 15.5 10.9 10.2 -- 0.75 AVF 15.5 11.8 15.8 -- 0.5 AVF 15.6 13.2 24.7 -- 0.25 AVF 15.6 14.6 33.5 -- 0 AVF (Water) 15.6 16.0 42.3 --

The HSPs of the solvent blend consisting of small molecules, such as acetone and water, can be calculated by simple rule of mixing. Figure 4.1a shows the RED as a function of acetone volume fraction (AVF) in the solvent blend. It is evident from the graph that the blend with 0.9

AVF is theoretically the best solvent for CA. This claim is corroborated experimentally from turbidity analysis of the CA solutions (4 and 8 wt%) in solvent blends of AVF 1, 0.9 and 0.75. As can be seen from Figure 4.1b, the CA solution in 0.9 AVF exhibits the lowest turbidity. The corresponding images of the solutions are also shown in Figure 4.1b.

105

This understanding of the behavior of CA polymer in the solvent blend of acetone/water can be used to tune the swelling behavior of CA gels. The analysis indicates that CA gel must exhibit maximum swelling in the best solvent (0.9 AVF) due to relaxation of the polymer chains.

As can be seen in Table A4.1, the polar and hydrogen bonding component is critical in defining the interaction between the solvent and CA. In addition to a minimum value of the polar component

(>5 MPa1/2), the hydrogen bonding component of the solvent must be similar to that of CA.

Therefore, on addition of 10 % water, the hydrogen bonding component of the blend becomes similar to that of CA, allowing the 10% blend (AVF = 0.9) to become ¨better¨ solvent for CA.

The CA gel or organogel was formed by dissolving CA in anhydrous acetone along with the pyromellitic dianhydride cross-linker. The cross-linking reaction was catalyzed by triethylamine. Note that the swelling of gels also depends on the cross-linking density. Here, the

106

cross-linking density was kept constant to isolate the effect of solvent composition (see experimental section for details). The CA gels were cut in cuboidal shapes and immersed in the mixture of acetone and water with known AVF. The volumetric swelling ratio was measured, and result are given in Figure 4.2a. The maximum swelling predicted for CA organogel from HSP is at 0.9 AVF. However, the swelling exhibits a maximum at AVF between 0.75 and 0.9. The observed range of AVF instead of a fixed value, as predicted from HSP, is most likely due to the chemical cross-linking of CA, which may cause a slight shift in the “good solvent” concentration for CA organogels. As expected from the Hansen solubility parameters, solvent blends below 0.5

AVF are poor solvents for CA and they do not show any appreciable swelling.

The obtained organogels were then immersed in DI water for 72 h and the DI water was replaced two times in between to remove excess acetone. The as obtained gels are termed hydrogels. Interestingly, even after immersing the swelled organogels in the poor solvent (water), they did not shrink back to the original state; in fact, they equilibrated at a given final swelling ratio, Figure 4.2b. The final swelling ratio ranges from 2.7 to 1.4 for AVF of 0.75 to 1. The organogel in 0.5 AVF did not show any further shrinkage after immersing in water. The fact that the swelled organogel did not revert to its original volume or shrunk further when immersed in a poor solvent can be explained by the elastic behavior of the cross-linked networks and presence of hydroxyl groups on cellulose backbone.

We hypothesize that the elastic network behaves as illustrated schematically in Figure

4.2c. The chains drawn in solid lines denote the CA polymer while the lighter, red lines denote cross-links. The (blue) dots represent water molecules and the grey and blue represent acetone and water respectively. Immersing organogel in a good solvent blend (0.9 AVF, middle image in

Figure 4.2c) causes swelling and allows water molecules to seep into the network along with

107

acetone. These water molecules may hydrolyze some of the ester bonds, thus reducing the elasticity of the cross-linked network34. Upon immersing swelled organogels in a poor solvent

(AVF=0, right image in Figure 4.2c) for a certain period, the acetone molecules diffuse out while the water molecules remain trapped inside, due to hydrogen bonding interactions with the hydroxyl groups of CA polymer chains.

Figure 4.2: a) Swelling behavior of cross-linked organogels as a function of the acetone volume fraction (AVF) in the solvent. b) Equilibrium volumetric swelling ratio after exchange with water. c) Schematic illustration of the swelling of cross-linked CA gels (color code: black chains-CA polymer; red dash- cross- links; blue dots-water molecules; grey- acetone and blue- water). Hydrolysis of ester bonds is expected when the organogel comes in contact with water, resulting in irreversible swelling of the gel, termed here “hydrogel” (from the swelled organogel to the hydrogel, as indicated by the size of the image). 108

The hydrolysis of ester bonds should cause weakening of the elastic network that can be qualitatively correlated with lowering of elastic modulus (G’) of the resulting hydrogel. To test this hypothesis, three hydrogel samples were prepared by immersing the organogel for 4, 12 and

24 h in solvent of AVF=0.9 followed by immersion in water for 72 h and rheological testing. The rheology tests involved performing a frequency sweep (1-100 rad/s) on the resulting hydrogels using a serrated flat plate geometry. This test provides values of G’ which is the measure of elastic modulus of the hydrogel. If the hydrolysis is ester bonds is occurring, then it is predicted that organogel immersed for longer period (24 h) will have the lowest elastic modulus (G’). The result shown in Figure A4.1, indicates a three-fold decrease in G’ upon increasing the immersion time for the gel, from 4 to 24 h, thereby lending credence to our premise.

It should be noted that even though cellulose acetate was chosen as the model polymer, the illustrated concept of controlling polymer swelling via Hanson Solubility Parameter analysis can be easily extended to other polymers. Upon identification of the interaction radius R0 of the polymer, suitable solvents or solvent mixtures can be chosen based on their solubility parameter

(δd, δp, δH), that either swell or shrink the gel relative to their position on the Hansen sphere. For example, Diehn et al.35, used HSP approach to predict a priori, the self-assembly of 1,3,2,4- dibenzylidene sorbitol in presence of various solvents with different HSPs and their distance from

R0.

4.4.2 Aerogel Synthesis and properties

For our studies, we use nomenclature to identify the aerogels produced in different solvent exchange conditions. For example, an aerogel obtained from the gel that was solvent-exchanged in AVF = 0.9 is referred to as “0.9A”. In general, it can be expected that the mechanical properties of such aerogels depend on the pore structure that includes pore volume and alignment as well as

109

inherent wall thickness. It has been shown that that freezing conditions strongly affects the pore structure of the resulting aerogel.36,37 Therefore, to isolate the effect of freezing conditions on the mechanical properties of the aerogels, unidirectional freezing was performed to produce uniform pore alignment and structure. The procedure and setup is shown in Figure 4.3a,b. The hydrogel was placed on a copper plate that was kept at a constant temperature of -80 0C. The other faces of the hydrogel were kept open to atmosphere at 25 0C. The frozen hydrogel was freeze dried at -56

0C and 0.113 mbar, to give the aerogel as shown in the inset of Figure 4.3b. The out-of-plane

(cross-section perpendicular to the freezing direction) and in-plane (cross-section along the freezing direction) view of SEM micrographs of CA aerogels are presented in Figure 4.3c,d, respectively. The out-of-plane view of CA aerogel (Figure 4.3c) exhibits a honey-comb structure with cell sizes ranging from 50 to 100 μm. The closed pore structure resembles a morphology that is reminiscent of sublimated ice crystals. Since the growth rate of ice crystals is highly anisotropic in one direction, the CA polymer is forced to align along the solidification front. The CA polymer becomes concentrated and squeezed on to the crystal boundaries giving the shown highly ordered honeycomb structure. The inset in Figure 4.3c shows the magnified SEM image of the out-of- plane view. The in-plane view, Figure 4.3d, exhibits a directionality in the pores. Most of the CA polymer was aligned in the direction of ice crystal growth and no honeycomb structure was observed in this view. Similar pore alignment with honeycomb structure has been observed for directional freezing of cellulose nanowhisker foams38 and a composite foam based on nanocellulose and graphene oxide39. However, the pore structure is not limited to columnar geometries. Chau et al.40, for example, demonstrated fibrillar and lamellar morphology, in addition of honeycomb-like columnar systems, during directional freezing of hydrazone crosslinked

CNC/POEGMA aerogels.

110

It should be noted that the immersion of CA gel into water (i.e. a poor solvent according to

HSP) may cause phase separation, leading to hierarchical porosity with pore sizes of < 1μm.

However, results show only large pores, 50-100 μm in size, formed between the non-porous walls.

We believe that phase separated structures of CA, if any, were eliminated during the freezing step, possibly by being compressed into thin walls of the polymer during slow freezing of water. To confirm this idea, we performed cryo-SEM of a hydrogel prepared from solvent exchange in AVF

= 0.9. During sample preparation for cryo-SEM, the hydrogel was frozen rapidly in liquid nitrogen to prevent sample destruction by volume expansion of ice. This was followed by sublimation of ice, coating with gold and imaging of the sample. The obtained SEM image shown in Figure A4.2 a,b demonstrate CA arrangement in honey-comb patterns, as observed before. Magnification of a small section of Figure A4.2b, where sample destruction was prevented by rapid freezing, indicates a phase-separated CA polymer structure (Figure A4.2c), as predicted. The final properties of the dried aerogels, however, are mainly affected by their bulk properties, which are governed by the macroporous structure obtained upon unidirectional freezing of the hydrogel.

111

The CA aerogels produced by solvent exchange and freeze drying were analyzed for their structural and mechanical properties. As seen from the Figure 4.4a, aerogels 0.9A and 0.75A have minimum density and highest pore volume. The 4 wt% CA hydrogel undergoes minimal shrinkage during freeze drying. Hence, the difference in densities and porosities is attributed to the swelling behavior of CA organogels when immersed in solvents with various AVF. The organogels immersed in 0.9 and 0.75 AVF exhibited maximum swelling, resulting in aerogels with lowest densities and highest pore volume. The organogel immersed in 0.5 AVF (poor solvent for CA) did not exhibit swelling, which resulted in aggregation of the polymer chains even before freezing,

112

thus producing an aerogel with highest density and lowest pore volume. Figure A4.3 shows SEM images of the aerogels 1A-0.5A. The aerogels 1A, 0.9A and 0.75A exhibit a honeycomb pattern that spans the entire aerogel. The aerogel 0.5A, on the other hand, shows pores that do not span the entire aerogel.

The mechanical properties of the aerogel were analyzed by measuring the stress as a function of compression strain. Figure 4.4b shows the out-of-plane compressive stress-strain curve for the CA aerogels, with the inset demonstrating the experimental schematic. The curves resemble compression curves of elastomeric foams, as introduced by Gibson and Ashby.41 The

113

elastomeric foams are characterized by a linear elastic region followed by a plateau region and a final densification region. The linear elastic region is controlled by cell wall bending. A small linear elastic region is observed (less than 3 % strain) for CA aerogels, which is due to low relative density of the aerogels (< 0.09). The aerogels 1A, 0.9A and 0.75A have extremely low relative density (0.03 - 0.02), which indicates the presence of thin cell walls and edges, as seen in Figure

4.4c. The cell wall thickness of aerogels 1A, 0.9A and 0.75A corresponds to 1.3±0.3, 0.8±0.1 and

0.8±0.1 μm, respectively. Furthermore, a honeycomb pattern of the thin cell walls (Figure A4.3) induces a large pore volume and provides an added strength to the aerogel.

The linear elastic region is followed by an extended plateau region (for aerogels 1A, 0.9A and 0.75A, up to 70% strain), which is associated with collapse of the cells walls by elastic buckling. The arrangement of thin cell walls in a honeycomb pattern gives an excellent load bearing capacity to the aerogel structure and allows aerogels 1A-0.75A to be compressed to large strains of more than 80%. A cyclic compression and release for 0.9A aerogels upon increasing strain is shown in Figure A4.4. This indicates that the 0.9A aerogel maintains its mechanical integrity until 50% strain but a permanent deformation is achieved at 80% strain. Furthermore, the same 0.9A aerogel, after being subjected to 80% compression cycles, demonstrated an excellent elasticity, when compressed repeatedly, out-of-plane. This is most likely because the structure reached an equilibrium state after initial collapse and permanent bending of some cell walls. The aerogel 0.5A, in contrast, has a high relative density (0.09), corresponding to the observed thick cell walls and edges (40±12 μm). In addition, the honeycomb structure of aerogel 0.5A did not span the entire system (Figure A4.3), which generates structural defects exhibited in the form of kinks on the stress-strain curve. When the cells are completely collapsed, further strain causes the cell walls to touch each other resulting in a sudden rise of stress. This region of densification is

114

very distinct for aerogel 1A-0.75A due to a large pore volume. In contrast, aerogel 0.5A exhibits a short plateau region that transitions quickly to the densification region, as explained by its low pore volume.

Generally, the mechanical properties of an aerogel such as compression modulus, absorption energy, compressive strength and densification strain is highly dependent on its relative density.41 Usually, aerogel density is tailored by varying the polymer content. However, reducing the polymer content compromises mechanical performance of the aerogels. Significantly, in this study, the density was tuned by controlling the swelling behavior of the precursor CA organogel.

Using the controlled swelling approach developed here, we find that the density and desired mechanical properties of the aerogels be tailored, based on the basic understanding of swelling phenomenon. Moreover, we find that the mechanical integrity of the aerogels can be favorably maintained. As can be seen from Table 4.2, the resulting aerogels show a wide range of values for aerogel stiffness (compression modulus from 14 to 340 kPa), toughness (absorption energy from

4 to 103 kPa) and strength (compressive strength from 22 to 373 kPa). A wide range of densification strain (35 to 87 %) was also observed. The densification strain gives an indication to the compressibility of an aerogel that arises from the large pore volume and elastic pore walls arranged in the observed honeycomb pattern. The values imply the flexibility of the proposed solvent exchange approach used to synthesize aerogels with a range of comprehensive mechanical properties without undergoing any chemical or physical modification.

115

Table 4.2: Mechanical properties of the CA aerogels

1A 0.9A 0.75A 0.5A 3 Density, ρa (g/cm ) 0.039 0.025 0.026 0.113 Pore Volume (%) 97.0 98.1 98.0 91.3 Relative density (ρa/ρCA) 0.030 0.019 0.020 0.087 Compression Modulus (kPa) 110 14 32 340 Absorption Energy (kPa) 18 4 6 103 Compressive strength (kPa) 71 22 24 373 Densification strain (%) 78 86 87 35

4.4.3 Introducing Anisotropy in Aerogels

The pore alignment of the aerogels due to unidirectional freezing endows an anisotropic behavior. Figure 4.5 compares the compressive stress vs strain curve of aerogel 0.9A in the two main planes, namely out-of-plane and in-plane. The in-plane compression, which is in direction of the pore axis, shows a very distinct elastic regime followed by a plateau and densification regime.

The aerogel exhibits a yielding behavior (at around 15% strain), which is a characteristic of plastic foams.41 A similar behavior was observed by Donius et al.42 for anisotropic nanocellulose- montmorillonite aerogels. When compressed along the axis, the pore walls elastically buckle until

15% strain and collapse when strained further resulting in a yielding and a constant stress of around

20 kPa. The out-of-plane view of SEM micrographs of the uncompressed and compressed (out-of- plane, in-plane) 0.9A aerogel sample is shown in inset of Figure 4.5. The samples were compressed up to 85% strain. The out-of-plane compressed image mainly shows pore walls bending at the hinges. There are a few pore walls that are cracked under the strain. The in-plane compression, in contrast, shows a complete structural collapse with loss of pore morphology. As can be seen from the inset images, out-of-plane compression allows aerogels to bounce back whereas in-plane compression causes a structural failure, when compressed up to a strain of 85%.

116

The elastic behavior of aerogels upon out-of-plane compression can also be seen in Figure A4.4.

The compression modulus for in-plane compression is 112 kPa compared to 14 kPa for out-of- plane compression. The energy of absorption is 14 kPa while compressive strength is 35 kPa for in-plane compression, compared to 4 kPa and 22 kPa when the same aerogel is compressed out- of-plane. The densification strain is 74% for in-plane compression and 86% for out-of-plane compression. The data indicate highly anisotropic behavior of these aerogels, which can behave as an elastic or plastic foam based on the direction of property measurement.

Figure 4.5: Compressive stress vs strain curves for aerogel 0.9A when compressed in-plane and out-of- plane. Inset shows out-of-plane SEM micrographs of the aerogel 0.9A when, uncompressed (top left), compressed out-of-plane (bottom left) and compressed in-plane (top right). Photos of aerogel samples are also shown.

117

4.4.4 Arresting the gels in a non-equilibrium state

Building on the understanding that the swelling characteristics of a gel influences the morphology and mechanical properties of the resulting aerogel, we hypothesized that the swelling kinetics of a gel influences its properties. We therefore analyze the swelling kinetics of the gel and how the swelling relates to the properties of the corresponding aerogels. Figure 4.6a shows the swelling kinetics of the organogel for a period of 48 h when immersed in 0.9 AVF. The organogel reached a maximum swelling ratio of about 6. The values in the inset of Figure 4.6a are normalized with the swelling ratio at 48 h, which indicates that the swelling reached an equilibrium value at 24 h. This further implies that any appreciable change in the aerogel properties is not expected after 24 h of swelling. Shrinking was observed when the swelled gels were immersed in pure water, which equilibrated at the volumetric swelling ratio of around 1.5-2.3

(Figure 4.6b). The density and pore volume of the aerogels synthesized from these hydrogels is demonstrated in Figure 4.6c. The density decreases as the solvent exchange time for the organogel increases and, consequently, the pore volume of the aerogel increases. This is explained by the swelling of the organogel, which increases as a function of time.

118

The compressive stress vs strain curves generated from out-of-plane compression of the aerogels are shown in Figure 4.6d. Similar curves were observed in Figure 4.4b, with short linear elastic region and a long plateau region followed by a densification region. It is noteworthy that the stress-strain curves for the aerogels obtained after 24 h of solvent exchange did not vary significantly. This is most likely due to the extremely low relative density (~0.02) and high pore volume (98%) observed for aerogels obtained after 24 h of swelling. The swelling reaches an equilibrium within 24 h and hence the corresponding aerogels from 24 and 48 h of swelling have

119

the highest pore volume and lowest relative density, achievable by the solvent exchange approach for this polymer concentration. The SEM micrographs of the out-of-plane view for the aerogels obtained after varying solvent exchange time is shown in Figure A4.5. All the samples show a honeycomb pattern.

The resulting mechanical properties of the aerogels produced under different swelling conditions is listed in Table 4.3. It is evident from the values in the table that the properties of aerogels can be conveniently tailored by varying the time for solvent exchange. The compression modulus ranges from 13-160 kPa, absorption energy ranges from 4-35 kPa, and the compressive strength ranges from 22-151 kPa. A wide densification strain is also observed ranging from 65 to

88%. This wide latitude of values opens the opportunity to achieve a predetermined aerogel mechanical property (stiffness, toughness and strength) by the proposed solvent exchange approach.

Table 4.3: Properties of the aerogels obtained after solvent exchange for given time periods

4h 12h 24h 48h 3 Density, ρa(g/cm ) 0.049 0.038 0.030 0.025 Pore Volume (%) 96.2 97.1 97.7 98.1 Relative density (ρa/ρCA) 0.038 0.029 0.023 0.019 Compression Modulus (kPa) 160 54 13 14 Absorption Energy (kPa) 35 10 7 4 Compressive strength (kPa) 151 40 26 22 Densification strain (%) 65 82 88 86

We suggest that the approach described in this study can be extended to other polymers or even inorganic aerogels, provided a suitable solvent is identified based on the Hansen solubility parameters. While this study is focused on freeze dried aerogels, the concept can be expected to 120

be equally applicable for supercritical dried aerogels, upon identifying the interaction of the polymer or the inorganic compound with supercritical CO2 based on the HSP of supercritical fluids,43 as shown in Chapter 3.

4.5 Conclusions

An innovative approach to synthesize ultralight, anisotropic aerogels with tailored mechanical properties is demonstrated. The use of the solubility parameter theory to control swelling behavior of the gels allowed a remarkable control over the final mechanical performance of the aerogels. The solvent exchange approach allowed us to synthesize aerogels with densities as low as 0.025 g/cm3 with no need for optimizing the precursor polymer concentration. The synthesized aerogels exhibited a wide range of stiffness (14-340 kPa), toughness (4-103 kPa), strength (22-373 kPa) and compressibility (35-87 %). Additionally, a unidirectional and controlled freezing approach introduced anisotropy in the aerogels inducing both elastic and plastic deformations, depending on the loading direction. The morphological and mechanical properties of the aerogels can be further tuned by arresting the gels in a non-equilibrium state during solvent exchange. By using this approach, aerogels can be achieved with tunable stiffness (13 to 160 kPa), toughness (4 to 35 kPa), strength (22 to 151 kPa) and compressibility (65 to 88 %).With impressive and tunable mechanical properties along with anisotropy, these aerogels can potentially be used as shock absorbers, in thermal and acoustic insulator, among many others.

121

4.6 References

[1] Kistler SS. Coherent Expanded-Aerogels. J Phys Chem 1931;36:52–64. doi:10.1021/j150331a003.

[2] Gurav JL, Jung I-K, Park H-H, Kang ES, Nadargi DY. Silica Aerogel: Synthesis and Applications. J Nanomater 2010;2010:1–11. doi:10.1155/2010/409310.

[3] Lee JK, Gould GL, Rhine W. Polyurea based aerogel for a high performance thermal insulation material. J Sol-Gel Sci Technol 2009;49:209–20. doi:DOI 10.1007/s10971-008- 1861-6.

[4] Nguyen ST, Feng J, Le NT, Le ATT, Hoang N, Tan VBC, et al. Cellulose aerogel from paper waste for crude oil spill cleaning. Ind Eng Chem Res 2013;52:18386–91. doi:10.1021/ie4032567.

[5] Sun H, Xu Z, Gao C. Multifunctional, ultra-flyweight, synergistically assembled carbon aerogels. Adv Mater 2013;25:2554–60. doi:10.1002/adma.201204576.

[6] Chaudhary JP, Vadodariya N, Nataraj SK, Meena R. Chitosan-Based Aerogel Membrane for Robust Oil-in-Water Emulsion Separation. ACS Appl Mater Interfaces 2015;7:24957– 62. doi:10.1021/acsami.5b08705.

[7] Guilminot E, Fischer F, Chatenet M, Rigacci A, Berthon-Fabry S, Achard P, et al. Use of cellulose-based carbon aerogels as catalyst support for PEM fuel cell electrodes: Electrochemical characterization. J Power Sources 2007;166:104–11. doi:10.1016/j.jpowsour.2006.12.084.

[8] Oh KW, Kim DK, Kim SH. Ultra-porous flexible PET/Aerogel blanket for sound absorption and thermal insulation. Fibers Polym 2009;10:731–7. doi:10.1007/s12221-010-0731-3.

[9] Li J, Li J, Meng H, Xie S, Zhang B, Li L, et al. Ultra-light, compressible and fire-resistant graphene aerogel as a highly efficient and recyclable absorbent for organic liquids. J Mater Chem A 2014;2:2934. doi:10.1039/c3ta14725h.

[10] Wang S, Peng X, Zhong L, Tan J, Jing S, Cao X, et al. An ultralight, elastic, cost-effective, and highly recyclable superabsorbent from microfibrillated cellulose fibers for oil spillage cleanup. J Mater Chem A 2015;3:8772–81. doi:10.1039/C4TA07057G.

[11] Jung SM, Mafra DL, Lin C-T, Jung HY, Kong J. Controlled porous structures of graphene aerogels and their effect on supercapacitor performance. Nanoscale 2015;7:4386–93.

[12] Tsou P. Silica aerogel captures cosmic dust intact. J Non Cryst Solids 1995;186:415–27. doi:10.1016/0022-3093(95)00065-8.

[13] García-González CA, Alnaief M, Smirnova I. Polysaccharide-based aerogels - Promising biodegradable carriers for drug delivery systems. Carbohydr Polym 2011;86:1425–38. doi:10.1016/j.carbpol.2011.06.066.

122

[14] Tong L, Lou J, Gattass RR, He S, Chen X, Liu L, et al. Assembly of silica nanowires on silica aerogels for microphotonic devices. Nano Lett 2005;5:259–62. doi:10.1021/nl0481977.

[15] Toivonen MS, Kaskela A, Rojas OJ, Kauppinen EI, Ikkala O. Ambient-Dried Cellulose Nanofibril Aerogel Membranes with High Tensile Strength and Their Use for Aerosol Collection and Templates for Transparent, Flexible Devices. Adv Funct Mater 2015;25:6618–26. doi:10.1002/adfm.201502566.

[16] Sehaqui H, Salajková M, Zhou Q, Berglund LA. Mechanical performance tailoring of tough ultra-high porosity foams prepared from cellulose I nanofiber suspensions. Soft Matter 2010;6:1824. doi:10.1039/b927505c.

[17] Leventis N, Sotiriou-Leventis C, Zhang G, Rawashdeh AMM. Nanoengineering Strong Silica Aerogels. Nano Lett 2002;2:957–60. doi:10.1021/nl025690e.

[18] Ma Q, Liu Y, Dong Z, Wang J, Hou X. Hydrophobic and nanoporous chitosan-silica composite aerogels for oil absorption. J Appl Polym Sci 2015;132:41770. doi:10.1002/app.41770.

[19] Korhonen JT, Kettunen M, Ras RH a, Ikkala O. Hydrophobic nanocellulose aerogels as floating, sustainable, reusable, and recyclable oil absorbents. ACS Appl Mater Interfaces 2011;3:1813–6. doi:10.1021/am200475b.

[20] Yang C, Chen C, Pan Y, Li S, Wang F, Li J, et al. Flexible highly specific capacitance aerogel electrodes based on cellulose nanofibers, carbon nanotubes and polyaniline. Electrochim Acta 2015;182:264–71. doi:10.1016/j.electacta.2015.09.096.

[21] Yang X, Cranston ED. Chemically Cross-Linked Cellulose Nanocrystal Aerogels with Shape Recovery and Superabsorbent Properties. Chem Mater 2014;35:6016–25. doi:10.1021/cm502873c.

[22] Zhang W, Zhang Y, Lu C, Deng Y. Aerogels from crosslinked cellulose nano/micro-fibrils and their fast shape recovery property in water. J Mater Chem 2012;22:11642. doi:10.1039/c2jm30688c.

[23] Zou J, Liu J, Karakoti AS, Kumar A, Joung D, Li Q, et al. Ultralight multiwalled carbon nanotube aerogel. ACS Nano 2010;4:7293–302. doi:10.1021/nn102246a.

[24] Guo H, Meador MAB, McCorkle L, Quade DJ, Guo J, Hamilton B, et al. Polyimide aerogels cross-linked through amine functionalized polyoligomeric silsesquioxane. ACS Appl Mater Interfaces 2011;3:546–52. doi:10.1021/am101123h.

[25] Meador MAB, Alemán CR, Hanson K, Ramirez N, Vivod SL, Wilmoth N, et al. Polyimide aerogels with amide cross-links: A low cost alternative for mechanically strong polymer aerogels. ACS Appl Mater Interfaces 2015;7:1240–9. doi:10.1021/am507268c.

[26] Sachithanadam M, Joshi SC. High strain recovery with improved mechanical properties of

123

gelatin-silica aerogel composites post-binding treatment. J Mater Sci 2014;49:163–79. doi:10.1007/s10853-013-7690-1.

[27] Tan C, Fung BM, Newman JK, Vu C. Organic aerogels with very high impact strength. Adv Mater 2001;13:644–6. doi:10.1002/1521-4095(200105)13:9<644::AID- ADMA644>3.0.CO;2-#.

[28] Jiang F, Hsieh Y-L. Amphiphilic superabsorbent cellulose nanofibril aerogels. J Mater Chem A 2014;2:6337–42. doi:10.1039/c4ta00743c.

[29] Hansen CM. Hansen Solubility Parameter- User Handbook. vol. 1. 2007. doi:10.1017/CBO9781107415324.004.

[30] Zugenmaier P. Characteristics of cellulose acetates— Characterization and physical properties of cellulose acetates. Macromol Symp 2004;208:81–166. doi:10.1002/masy.200450407.

[31] Fischer F, Rigacci A, Pirard R, Berthon-Fabry S, Achard P. Cellulose-based aerogels. Polymer (Guildf) 2006;47:7636–45. doi:10.1016/j.polymer.2006.09.004.

[32] Wagner W, Saul A, Prub A. International Equations for the Pressure along the Melting and along the Sublimation Curve of Ordinary Water Substance. J Phsical Chem 1994;23.

[33] Malm CJ, Genung LB, Fleckenstein J V. Densities of Cellulose Esters. Ind Eng Chem 1947;39:1499–504.

[34] Mabey W, Mill T. Critical Review of Hydrolysis of Organic Compounds in Water Under Environmental Conditions. J Phsical Chem 1978;7:383–415.

[35] Diehn KK, Oh H, Hashemipour R, Weiss RG, Raghavan SR. Insights into organogelation and its kinetics from Hansen solubility parameters. Toward a priori predictions of molecular gelation. Soft Matter 2014;10:2632. doi:10.1039/c3sm52297k.

[36] Zeng X, Ye L, Yu S, Sun R, Xu J, Wong CP. Facile Preparation of Superelastic and Ultralow Dielectric Boron Nitride Nanosheet Aerogels via Freeze-Casting Process. Chem Mater 2015;27:5849–55. doi:10.1021/acs.chemmater.5b00505.

[37] Jiang F, Hsieh Y-L. Super water absorbing and shape memory nanocellulose aerogels from TEMPO-oxidized cellulose nanofibrils via cyclic freezing–thawing. J Mater Chem A 2014;2:350–9. doi:10.1039/c3ta13629a.

[38] Dash R, Li Y, Ragauskas AJ. Cellulose nanowhisker foams by freeze casting. Carbohydr Polym 2012;88:789–92. doi:10.1016/j.carbpol.2011.12.035.

[39] Wicklein B, Kocjan A, Salazar-Alvarez G, Carosio F, Camino G, Antonietti M, et al. Thermally insulating and fire-retardant lightweight anisotropic foams based on nanocellulose and graphene oxide. Nat Nanotechnol 2015;10:277–83. doi:10.1038/nnano.2014.248.

124

[40] Chau M, De France KJ, Kopera B, Machado VR, Rosenfeldt S, Reyes L, et al. Composite Hydrogels with Tunable Anisotropic Morphologies and Mechanical Properties. Chem Mater 2016;28:3406–15. doi:10.1021/acs.chemmater.6b00792.

[41] Gibson LJ, Ashby MF. Cellular Solids: Structure and Properties. Cell. Solids Struct. Prop. 2nd ed., Cambridge University Press, Cambridge, UK; 1997.

[42] Donius AE, Liu A, Berglund LA, Wegst UGK. Superior mechanical performance of highly porous, anisotropic nanocellulose-montmorillonite aerogels prepared by freeze casting. J Mech Behav Biomed Mater 2014;37:88–99. doi:10.1016/j.jmbbm.2014.05.012.

[43] Williams LL, Rubin JB, Edwards HW. Calculation of Hansen Solubility Parameter Values for a Range of Pressure and Temperature Conditions, Including the Supercritical Fluid Region. Ind Eng Chem Res 2004;43:4967–72. doi:10.1021/ie0497543.

125

Chapter 5: Aerogel Application in Oil-Spill Cleanup

5.1 Abstract

The 4% CDA aerogel prepared in previous chapters was rendered hydrophobic and oleophilic via chemical vapor deposition with an organo-silane. The modified CDA aerogel surpass their counterparts in maintaining their mechanical integrity for fast oil cleanup and efficient oil retention from aqueous media under marine conditions. These aerogels are identified as reusable and durable for an extended period. The Zisman and Fowkes theoretical frameworks are used to identify the selectiveness of the aerogel and establish a criterion for separation of various non-polar fluids from water media thereby providing a general platform for predicting the sorption ability of the aerogels for various chemicals.

5.2 Introduction

5.2.1 In context to the dissertation

In Chapter 3 we introduced a synthesis approach for aerogels from non-aqueous polymers, mainly cellulose esters. In Chapter 4, we build on the approach to perform a systematic investigation of a novel method to tune the density and consequently the mechanical properties of cellulose diacetate (CDA or CA) aerogels, while starting from same precursor concentration. In this chapter, we explore one practical application of such CA aerogels. The 4 w/v% CA aerogels with density of 23.4 mg/ml (0.023g/cm3) were modified using chemical vapor deposition of a silane to render them hydrophobic and oleophilic. The performance of such modified aerogel for oil-spill remediation was tested in the simulated oceanic environment. The Zisman and Fowkes’ theoretical frameworks for calculating solid surface energy are used to identify the selectiveness of the aerogel and establish a criterion for separation of various non-polar fluids from water media.

126

5.2.2 Current technologies for oil-spill remediation

Uncontrolled oil or chemical spillage into fresh water triggers a cascade of events including toxic loading and contamination of food resources, all of which are highly detrimental to the ecosystem.1,2 Prevalent technologies proposed to deal with oil spills include oil skimming, in-situ burning, mechanical containment and utilization of dispersants, solidifiers and degrading microorganisms.3–8 However, these methods are either inefficient or environmentally unfriendly.

The need for efficient oil spill cleaning technologies was imminent in the 2010 Deepwater Horizon spill, where oil gushed out in the Gulf of Mexico for 87 consecutive days, pouring out over 200 million gallons of crude oil.9 In this type of scenario, the use of sorbents is attractive as it is easy to deploy and do not generate byproducts. There are three categories of sorbent materials: organic

(milkweed, wood chips, rice straw), inorganic (organo-clay, perlite and sand, zeolites) and synthetic (non-woven polypropylene mats).10–12 However, they suffer from low sorption capability

(typically less than 10 g per gram of substrate), are difficult to recover (they are not buoyant), and are not biodegradable.

5.2.3 Aerogels for oil-spill cleanup

In recent years, aerogels have gained attention as sorbent materials because of their light- weight and high pore volume. However, conventional silica aerogels are brittle, making it difficult for removal from oil-water mixtures, and thus are prone to secondary contamination.13 Silica- chitosan composite aerogels display suitable mechanical properties but they have low porosity

(<96%) and high density (>58 mg/cc), which results in low sorption.14 Carbon nanotube and graphene-based aerogels demonstrated a very high sorption of non-polar solvent (500-600 g/g).15–

17 These aerogels are, however, not biodegradable and their synthesis involves loss of expensive material under the high temperature treatment used. The advent of nanocellulose signified new

127

biodegradable aerogels that are relatively easy to synthesize and have better mechanical properties than those based on conventional silica. Sehaqui et. al,18,19 synthesized nanofibrillated cellulose aerogels with high surface area and porosity. Compared to silica aerogels, they displayed better strength and compressibility. Unfortunately, these nanocellulose aerogels were not tested for sorption properties. Ikkala and co-workers synthesized hydrophobic and oleophilic nanocellulose aerogels suitable for oil separation.20,21 The aerogels were rendered hydrophobic via chemical vapor deposition of chlorosilanes and atomic layer deposition of TiO2. In a similar study, the sorption capacity of nanocellulose aerogels possessing different pore structure and size was quantified.22 These aerogels were reported to have oil sorption capacity of 30g/g. Recently, the highest sorption ability was reported for aerogels comprising cellulose nanofibrils (defibrillated from rice straw), and found to be 200-300 g/g for non –polar solvents.23,24 However, they suffered from weak mechanical integrity and had a maximum compression strength of only 2 kPa.

Durability and mechanical strength are essential properties for a sorbent material, in addition to buoyancy, and rapid and large sorption capacity, as these materials may be exposed to wind, waves and ocean currents. While the abundance of hydroxyl functionalities in cellulosic aerogels facilitates modification to induce hydrophobicity and increase sorption ability, some form of cross-linking is required to increase their mechanical strength and uphold their mechanical integrity in agitated water. Cellulose nanocrystal aerogels were prepared via hydrazine-aldehyde cross-linking25; likewise, epoxide cross-linking was used to synthesize aerogels based on cellulose nano/micro fibrils.26 These studies demonstrated an impressive shape recovery under water.

However, the cross-linking was carried out in the solid phase, which is generally non-uniform and thus produce uneven mechanical strength in the aerogel. The limited performance under extreme

128

conditions in agitated systems makes the reported cellulosic aerogels unlikely candidates for oil spill management in oceanic environment.

Results of these studies taken together reveal that nanocellulosic aerogels while offering high oil sorption capacities lacked mechanical integrity23,24, earlier reported cellulose acetate aerogels have poor properties and no reported sorption results27,28, and no theoretical framework exists to explain and generalize the selectivity of the aerogels. In this chapter, we modify cellulose acetate aerogels using chemical vapor deposition to render them hydrophobic and oleophilic.

Thereafter, we explore the efficacy of modified freeze dried cellulose acetate aerogels in simulated oil-spill scenario. We also provide a theoretical framework based on the surface energy theories to determine selectivity of these and possibly other aerogels towards various chemicals with different surface energies.

5.3 Experimental Section

5.3.1 Materials

Cellulose diacetate flakes provided by Eastman Chemical Co. with degree of substitution of 2.45 were used as received. The trichloro-octyl silane was purchased from Sigma Aldrich.

Deionized (DI) water was used. All solvents used for wicking measurements were 99 % pure and bought from Sigma Aldrich. Spent kerosene oil was purchased obtained from vacuum pump cleaning. Used motor oil was gathered from the oil change of the car. The crude oil was purchased from the Texas Raw Crude Oil.

5.3.2 Aerogel modification

The CDA aerogels were subjected to chemical vapor deposition (CVD) with trichloro octyl silane (TCOS) to convert them hydrophobic and oleophilic.41 A bottle-in-a-bottle set up was used

129

for this purpose (Figure 5.1) where TCOS was kept in the smaller container and the aerogel was kept on a wire mesh atop. The system was kept in an oven at 80 0C for 2 h.

Aerogel

TCOS

Figure 5.1: Set-up used for modification of cellulose diacetate aerogel

5.3.3 Aerogel Performance tests a) Sorption test

The aerogel was immersed in a liquid and allowed to saturate. After immersion, the surface of the saturated aerogel was blotted with a paper wipe to remove surface liquid and weighed.

Liquid uptake was calculated using Equation 5.1:

W −W Liquid uptake = ab a ---Equation 5.1 Wa

Where, Wab and Wa are the weights of the saturated and dry aerogels respectively. Note, for sorption of organic solvents, the blotting step was avoided to prevent wicking of solvent to the paper wipe.

Instead, the saturated aerogels were weighed immediately. b) Reusability tests

A representative solvent of a low surface energy oil, n-hexane, was used to test the reusability of modified aerogels. Loaded aerogels were mechanically compressed between paper wipes to remove the solvent and weighed again to ensure at least 70 % of the solvent was squeezed

130

out. The process of sorption and compression was repeated until the aerogel lost its mechanical integrity, which was identified when the aerogel was not able to recoil back as observed from the unaided eye. The average data with standard deviation is reported after repeated tests with three different samples.

5.3.4 Wicking and contact angle measurements

A dynamic contact angle analyzer (CAHN DCA-312) was used for the wicking measurements. The aerogels were cut into cuboid shape and lowered into the liquid (see Figure

5a). Extreme care was taken to cut the aerogels in one direction to ensure that pore alignment was the same in all the measured samples. The aerogels were immersed 1.5mm into the surface of the given liquid to ensure complete immersion of aerogel into the solvent during the course of experiment. A goniometer, SEO Phoenix was used for contact angle measurements. The images of the sessile drop reported for aerogel were taken for qualitative assessment of the extent of hydrophobic modification of the aerogels. Each value of contact angle for the flat surface (average of 5 measurements) was recorded after drop deposition for 30 s. Since the measured contact angle of liquid on the modified aerogels using a camera image is not a correct representation of the aerogel’s actual surface properties due to difficulty in obtaining a perfectly flat surface for the inventive aerogels, the change in surface energy of an aerogel due to surface modification was analyzed. The Fowkes’ model32 was used to calculate the polar and dispersive components of the solid surface energy using the contact angle data.

5.3.5 Spin coating

Silica wafers (1cm x 1cm) were used as supports that were coated with 1 wt% of CDA solution in acetone, as reported earlier.42 The silica wafers were cleaned by washing with Piranha solution for 5 min. The silica wafers were then immersed in DI water for 5 min and thoroughly

131

washed with DI water for 3 times followed by drying with compressed N2 gas. The as-cleaned silica wafers were spun coated with 1wt% CDA solution at 4000 rpm for 40 s at an acceleration of 2400 s-1.

5.4 Results and Discussion

5.4.1 Aerogel Modification

The presence of free hydroxyl groups and carbon backbone renders CDA aerogels generally amphiphilic. To increase the hydrophobicity and oleophilicity of CDA aerogels,

Chemical Vapor Deposition (CVD) with trichloro-octylsilane (TCOS) was used to cap free hydroxyl functionalities with the hydrophobic chains. As shown in Figure 5.2a, a water droplet completely wicks through the surface of the unmodified 4 % CDA aerogel, whereas a modified

4% CDA aerogel shows a water contact angle exceeding 120° on both the surfaces of the aerogel

(Figure 5.2 b1). Moreover, the fact that the transverse cross-section of modified aerogel exhibits a high, water contact angle (Figure 5.2 b2), indicates that TCOS diffuses below the surface of the aerogel, rendering the bulk aerogel hydrophobic. In general, all observed contact angles for the modified aerogels were greater than 120°. Note however, the typical large variation in contact angle values due to the uneven surface on the aerogel. An XPS survey scan of the top surface and transverse cross-section of modified aerogel (Figure 5.2c) further confirms the diffusion of TCOS into the bulk of aerogel. The calculated Si/O atomic% at the top surface and in the center is 19 and

2 at.%, respectively.

132

The samples of modified and unmodified 4% CDA aerogels were subjected to water sorption test for 48 h, with their behavior monitored after placing them on the surface of water.

Figure 5.3a shows the experimental condition, where modified (right) and unmodified (left) aerogels were left on the water surface. The unmodified aerogel sunk in the water within 2 h while the modified aerogel remains afloat even after 12 h. The unmodified CDA aerogel quickly sorbed water, between ~35 and 40 g/g aerogel during approximately 48 h, whereas the TCOS-modified

CDA aerogel was hydrophobic and sorbed less than 3 g/g aerogel during the same time period

(Figure 5.3b). This data demonstrates that the modified aerogel is stable and retains its

133

hydrophobicity even after constant exposure to water for 48 h. An additional durability test was performed where the modified aerogel sample was left on lab bench under ambient conditions for

4 months. Figure 5.3c compares the water uptake by modified aerogel after 4 months. A high durability and stability of modified aerogel is indicated by the high water contact angle on their surface (>120 degrees) and the low water uptake (< 3g /g of aerogel ).

134

5.4.2 Oil Separation from water-oil mixture

The TCOS-modified 4 w/v% CDA aerogel was tested in various scenarios of oil-in-water systems to evaluate their sorptive properties and mechanical integrity. The model oil used in these studies was a spent kerosene grade oil (viscosity of ~1-2 cP). A sample of 0.26 g of the modified aerogel was placed into a jar which contained 10.5 g of the oil and about 59.4 g of water. The photographs of Figure 5.4a are the snapshots of Movie A5.1 that show the process of the selective sorption over time of the oil from the oil-water mixture by the modified aerogel. The aerogel was saturated with oil within a minute, as shown in Figure 5.4a ii) to iii). The oil-loaded aerogel was hand-pressed, as shown in Figure 5.4a iv), to recover the oil, and was reused to remove the oil remaining in the oil-water mixture, as shown in Figure 5.4a v) to vi). The mass balance of the liquids from the separation shown in photographs of Figure 5.4a, is shown in Figure 5.4b. The weight of the water, oil and aerogel was measured before the experiment, and the weight of the oil pressed from the aerogel was measured after the experiment. The weight of the used aerogel and the remaining water was measured to complete the mass balance. The data shown in Figure 5.4b shows that about 83 wt% of the kerosene oil was recovered, with the unrecovered oil staying trapped in the aerogel. In addition, the water sorption by the CDA aerogel was negligible (about

0.02 wt%), which suggests that the recovered oil could be reused.

No major changes in the aerogel structure were observed after multiple cycles of sorption and mechanical compression. The reusability of modified aerogels was tested via n-hexane sorption and desorption by mechanical compression (Figure 5.4c). We found that a TCOS- modified 4 w/v% CDA aerogel can be subjected to at least 10 cycles of sorption and compression before undergoing structural failure. As shown in Figure 5.4c, the modified aerogel can sorb over

25 times its weight of n-hexane in the first cycle and can still sorb about 10 grams of n-hexane per gram of aerogel even after 9 cycles of sorption/compression. 135

The modified aerogels were further tested under shear in an oil in water medium (Movie

A5.2). The spent kerosene was dyed red for better observation. Figure 5.4d shows photographs to indicate that a clear solution is obtained within a minute, as nearly all the dyed oil is soaked up by the modified aerogel. The modified aerogel retains both the oil and its mechanical integrity even in the stirred media, indicating its utility for oil spills in an agitated media. To further simulate high viscosity oil spill in a turbulent environment (typical in marine environments), a stirred oil in water system was prepared (Movie A5.3), in which the oil used was spent motor oil (spent car oil with a viscosity of about 170 cP. As indicated by the photographs of Figure 5.4e, the modified aerogel sorbed all the high viscosity motor oil within 2 min, even in the stirred media.

As the final performance test for modified aerogel, it was tested in salt water with the same chemical composition as the sea water. The experimental set-up to test the modified aerogel performance in sea water is shown in Figure A5.1a. The three salts that constitutes a significant percentage of salt water in sea are sodium chloride (NaCl), magnesium sulfate (MgSO4) and calcium chloride (CaCl2) were mixed with water to make a total dissolved salt concentration of

3.5 wt%. The crude oil with the viscosity of 10 cP was mixed with this salt water. Figure A5.1b shows the water uptake and crude oil uptake by modified aerogels in presence of salts and pure water. The results indicate that the aerogel performance is not inhibited in presence of salts.

Therefore, we expect these modified aerogels to perform as well in sea water as in the lab conditions.

136

Figure 5.4: a) Snapshots from movie S1 demonstrating the separation of spent kerosene oil from oil/water stagnant media. The aerogel’s reusability is demonstrated in the figure, b) graph for the mass balance of the water and oil before and after separation indicating the complete removal of oil with very negligible sorption of water, c) reusability of the aerogel was tested for 10 cycles of sorption and desorption (by mechanical compression) of n-hexane, d) snapshots from movie S2 to demonstrate the separation of dyed kerosene oil from oil in water stirred emulsion system (kerosene: 6g, DI water: 100g, aerogel: 0.26g, stirring rate: 500rpm) e) snapshots from movie S3 demonstrating the separation of high viscosity spent motor oil from motor oil/water stirred system to model oil spill in a turbulent sea (motor oil: 7g, DI water: 100g, aerogel: 0.35g). The upper photographs shows a side view of the beaker used for the study prior to addition of the modified aerogel (“initial system”; upper left photograph), and an overhead view of the instant that the aerogel was added to the beaker (t = 0 sec; upper right photograph)

The rate of oil uptake was measured via wicking experiments using a tensiometer (CAHN

DCA-312). The aerogels were cut into a cuboid shape and lowered into the liquid (see Figure

5.5a). Extreme care was taken to cut the aerogels in one direction to ensure that pore alignment was same in all the measured samples. The bottom edge (surface) of an aerogel was submerged

1.5 mm below the surface of a given liquid to ensure continuous contact of aerogel with solvent

137

during the course of the experiment. Kerosene oil, crude oil and motor oil were the liquids used with viscosity 1.4, 10 and 168 cP respectively. Figure 5.5b shows the liquid uptake rate by the modified aerogel. We observe that low viscosity kerosene oil wicks the modified aerogel instantaneously, in contrast to high viscosity motor or crude oil. Motor oil with the highest viscosity has the slowest liquid uptake rate, as expected. The maximum uptake of crude oil is highest, as crude oil (0.97 g/cm3) has a higher density than kerosene oil (0.8 g/cm3). It should be noted that the wicking rate of any oil can be further tuned by changing the pore size of the aerogels or by changing the surface energy of the aerogel with other compounds such as fluorocarbons.

5.4.3 Surface energy of solid surfaces

An estimation of the surface energy of the modified aerogels is an intriguing idea as we can identify the liquids that the modified aerogel can separate from polar solvents such as water.

Unfortunately, the surface energy of a solid vapor interface (훾푆푉 표푟 훾푆) such that of an aerogel

138

cannot be measured directly. The value of surface energy is calculated indirectly from a set of liquid/solid contact angles, which implies that the calculated surface energy not only depends on the surface energy of the liquid used for contact angle measurement, but also on the surface energy theory applied for data analysis.29 The oldest equation relating surface energies of solid and liquid is Young’s equation:

훾푆푉 = 훾퐿푐표푠휃 + 훾푆퐿

To find surface energy of solid (훾푆), we need the value of 훾푆퐿. A unifying theory for

30 determination of 훾푆퐿and 훾푆 does not exist; however, pioneering work by Zisman , further extended by Fowkes31,32, Owen-Wendt33 and Van Oss34 are well accepted approaches. Such approaches rely on measurement of contact angle on the solid surface using the liquids with known surface energies. All theoretical models can be classified into either theory of surface tension components (STC) pioneered by Fowkes or equation of state (EOS) approach based on macroscopic thermodynamics proposed by Neumann.35–37 While both these approaches require measurement of contact angle on solid surface, the theory of STC is based on resolving 훾푆퐿 into contributions from polar (that includes dipole-dipole, hydrogen bonding and electrostatic interactions) and dispersive (van der Waals) intermolecular forces. The EOS approach on the other hand identifies 훾푆퐿 as the function of total surface energy of the solid and liquid phases leading to a model equation that can be solved simultaneously with Young’s equation to determine two unknowns 훾푆퐿 and 훾푆푉. Although EOS approach requires measurement to be done using only one probe liquid and in principle the surface energy should be same regardless of the liquid chosen, this approach has drawn a lot of criticism dealing with whether the experimentally determined constant β is the universal constant of the material or just a quantity obtained as a result of the iterative procedure applied. The reader is directed to previous work describing limitations of EOS

139

approach.38 Therefore, among all the models available in the literature, Zisman and Fowkes’ model based on theory of STC were considered most appropriate for our system of aerogels.

However, aerogels present an additional challenge when compared to other flat solid surfaces. Aerogel have surface undulations that makes the contact angle measurements from the camera images unreliable. In fact, the contact angle varies for the same sample while changing the location of measurement. Moreover, a porous structure of the aerogel induces an additional wetting phenomenon resulting in the changing shape and size of liquid droplet on the surface due to wicking.39 To overcome this challenge, a two-fold approach was employed. First, a modified

Zisman theory (explained later) was applied to the modified cellulose acetate aerogels to estimate the range of surface energy of the aerogels and second, the Fowkes’ model was applied on the modified cellulose acetate films to further narrow down the value of surface energy of modified aerogel arising from the chemical modification. a) Modified Zisman Theory

푐 According to Zisman theory, a critical surface tension of solid 훾푆 can be attributed to the modified aerogel. If this critical surface tension is greater than the surface tension of liquid

푐 푐 (훾푆 > 훾퐿), the liquid will complete wet the aerogel and get sorbed; if 훾푆 < 훾퐿, a partial wetting will be observed. When applied in its traditional sense, the Zisman plot looks something like

Figure 5.6a. The cosine of contact angle is plotted against the liquid surface tension and the surface tension where the line crosses zero contact angle or cosine value of 1 is identified as the surface tension of the solid. Since, the aerogel surface has undulations, the contact angle measurements are not reliable and should not be used to while applying the Zisman theory. Instead, wicking tests were performed on the modified CDA aerogels using solvents of different surface tension. The aerogels were carefully cut in the pore direction and immersed in the liquid of known surface

140

tension. The rate of weight gain of the aerogel is recorded. Based on the slope of the graph an estimation of the range of surface energy of the solid surface can be made. As shown in theoretical

Figure 5.6b, there are 4 types of curve possible in this case of wicking where the surface tension

4 3 2 1 푐 of liquid follows this relation: 훾퐿 < 훾퐿 < 훾퐿 < 훾퐿 . The idea is to find the position of 훾푆 in this relation. The two extreme cases are liquid 1, where the liquid does not wick at all, and liquid 4

푐 where the liquid wicks instantaneously. From these two curves, we can safely deduce that 훾푆 <

1 푐 4 훾퐿 and 훾푆 > 훾퐿 . The other two cases include intermediate wicking rate where liquid 3 exhibits

푐 2 slower wicking rate than liquid 2. In such scenario, 훾푆 > 훾퐿 because curve 2 has higher slope of wicking rate and due to slower wicking rate with liquid 3, the following relation can be deduced:

푐 2 2 푐 1 훾푆 ≥ 훾퐿 . Therefore, the final range of surface tension of the solid will be 훾퐿 ≤ 훾푆 < 훾퐿 . Note, that wicking also depends on the viscosity of the liquid. So, all the liquid tested here had viscosity lower than 3 cP. is The absolute value of weight gain is inconsequential because it depends on the density of the liquid.

141

Figure 5.6: a) Traditional Zisman plot of cosine of contact angle vs surface tension of the probe liquid for low density polyethylene film.40 The intersection of the straight line with the X-axis where contact angle is zero is identified as the surface energy of the solid surface. b) The theoretical plot of rate of liquid uptake 4 3 2 1 of 4 probe liquids with following relation of surface tension: 훾퐿 < 훾퐿 < 훾퐿 < 훾퐿 . The modified Zisman theory used in this study relies on this plot to identify a range of surface tension of the solid surface, aerogel in this case.

b) Fowkes’ Theory

The limitation of our modified Zisman theory to give only a range of surface energy for the chemically modified aerogels necessitated the use of Fowkes’ theory to isolate the effect of chemical modification of aerogels on its surface energy. Since, the Fowkes’ theory also relies on the measurement of contact angle on the solid surface and the value of contact angle on the aerogel surface is not reliable, the contact angle was measured on the spin coated layer of cellulose acetate

(method described later) that was chemically modified in the similar fashion as the aerogel.

Fowkes’ theory is a two-component surface energy model that describes surface energy of the solid surface as composed of dispersive component and polar component. The theory is based on three fundamental equations:

142

Young’s Equation

훾푆 = 훾퐿푐표푠휃 + 훾푆퐿 ---Equation 5.2

Dupre’s Definition of Adhesion Energy

퐼푆퐿 = 훾푆 + 훾퐿 − 훾푆퐿 --- Equation 5.3

And Fowkes’ theory that the adhesive energy can be separated into dispersive and polar component

1 1 1 1 퐷 퐷 푃 푃 퐼푆퐿 = 2[(훾퐿 )2 (훾푆 )2 + (훾퐿 )2 (훾푆 )2] --- Equation 5.4

Here, 훾푆= surface tension of solid,

훾퐿 = surface tension of liquid

훾푆퐿 = interfacial tension of solid and liquid

휃 = contact angle between solid and liquid

퐼푆퐿 = energy of adhesion per unit area between liquid and solid surface

훾퐷 and 훾푃 are dispersive and polar part of the surface tension respectively

Combining these three equation, we get primary equation for the Fowkes’ surface energy theory:

1 1 1 1 훾퐿(1+푐표푠휃) 퐷 퐷 푃 푃 = [(훾 )2 (훾 )2 + (훾 )2 (훾 )2] ---Equation 5.5 2 퐿 푆 퐿 푆

The first step in determining the solid surface energy using this model is to measure the contact angle with a liquid that has only dispersive component (diiodomethane is usually used for

퐷 퐷 the purpose, 훾퐿 = 훾퐿 = 50.8 푚푁/푚). This gives the dispersive component (훾푆 ) solid surface energy. Thereafter, the contact angle measurement can be done with water to solve for the

푃 remaining unknown, that is the polar component (훾푆 ) of the solid surface energy. The total solid

퐷 푃 surface energy is the sum of these two components: 훾푆 = 훾푆 + 훾푆 . In our case, the solid surface energy from the Fowkes’ theory should fall between the range of solid surface energy identified from the modified Zisman theory.

143

5.4.4 Sorption & surface energy analysis for aerogels

In this and the subsequent sections, we attempt to establish a simple framework to a priori predict what liquids can be sorbed by the modified aerogel using the modified Zisman theory

(explained before) and Fowkes’ theory. Our basic premise is that the mass of the liquid sorbed depends on the interfacial surface tension of the corresponding liquid with the surface of the aerogel, for the liquids of similar viscosity. a) Application of modified Zisman theory for aerogels

According to the Zisman theory, a critical surface tension γc can be attributed to the

30 modified aerogel. If this critical surface tension is greater than the surface tension of liquid (γc >

γliq), the liquid will complete wet the aerogel and get sorbed; if γliq > γc, a partial wetting will be observed. Since, the aerogel surface is not perfectly flat, contact angle measurements are not accurate representation of surface tension and not a means to test the Zisman theory. Instead, wicking tests were performed on the modified CDA aerogels using solvents of different surface tension, as tabulated in Table A5.1. A plot of the amount of liquid sorbed as a function of time for four solvents is shown in Figure 5.7a. The modified aerogel exhibits an instantaneous sorption of low surface energy liquid, n-hexane, but shows a negative value for the liquid uptake of water and diiodomethane. The explanation for the negative value recorded for mass gain is given in the

Appendix A5. Briefly, the buoyant force (Fb) on aerogels (Figure 5.7b) is not negligible and for hydrophobic aerogels, the surface tension force is acting upwards whereas the gravitational force

(Mabg) is very small resulting in a total negative force (F) recorded by the instrument. A negative mass gain implies that the surface tension of liquid is greater than the critical surface tension of aerogel surface (γc). As indicated from Figure 5.7a, γc < γdiiodomethane and instantaneous sorption of n-hexane indicates that γc> γhexane. However, 1,2-dichlorobenzene (with surface tension of

144

35.7mN/m), shows a sorption gradient which implies that initially the aerogel surface resisted penetration of 1,2-dichlorobenzene but as soon as the liquid penetrated the aerogel surface, the aerogel became saturated within a minute, due to capillary action. This means that γ1,2-dichlorobenzene

< γc. From this analysis, one can conclude that γc of the modified aerogel lies between 36 and 51 mN/m.

145

Once the liquid is sorbed, the equilibrium mass uptake of the corresponding liquid will depend on the capillary force due to microporous structure of aerogel, and the mass of liquid sorbed

(M) can be related as

Mg = 2r cos ---Equation 5.6 where ‘r’ is the radius of the pore, ‘γ’ is the surface tension of liquid with respect to air and ‘θ’ is the contact angle of the fluid with the pore wall.30 As evident from Figure 5.7c, the mass of the liquid sorbed is directly proportional to the surface energy of the corresponding liquid as long as the surface tension of the liquid is below the critical surface tension (γc) of the aerogel. The volume of liquid sorbed, however, depends on the available pore volume as shown in Figure 5.7d. A ratio greater than 1 for most of the solvents may indicate that the liquid is chemically interacting with the polymer. However, a ratio of less than 1 for completely non-polar n-hexane may suggest minimal chemical interaction. b) Calculation of surface energy using Fowkes’ model

Results from modified Zisman theory demonstrate that the value of γc of the modified aerogel is between 36 and 51 mN/m. The value can be confirmed and more precisely determined by considering that the interfacial surface energy of the modified aerogel is governed by surface groups, surface roughness and tortuosity. We use the Fowkes’ model to identify the dispersive and polar components of the surface energy of TCOS-modified CDA films, which was obtained by spin coating CDA on a silicon wafer. The spin coated CDA was subjected to chemical vapor deposition of TCOS using the same procedure as described for the aerogels. Water and diiodomethane were the two probe liquids used. Table 5.1 shows the average contact angle measured with the probe liquids along with the corresponding surface energy values.

146

Table 5.1: Contact angle data & surface tension values for different probe liquids43

Contact angle θ(degree) Polar γp Dispersive Total γt Probe liquid Untreated Modified (mN/m) γd (mN/m) (mN/m) CDA CDA Water 45 80 51 21.8 72.8 Diiodomethane 26 38 0 50.8 50.8

Using the Fowkes’ model, the surface energy of the untreated CDA surface and TCOS modified CDA surface were calculated as 64 and 44 mN/m, respectively as listed in Table 5.2.

Here, the dispersive component of the surface energy dominates: 71 and 93 % in untreated and modified CDA, respectively. This is likely because CDA is primarily composed of non-polar carbon backbone, which is responsible for induced dipole interactions (i.e., London forces). The polar component of the surface energy for untreated CDA is mainly due to surface hydroxyl groups. Upon modification of the CDA, by capping the hydroxyl group with non-polar eight carbon chain, the polar component drops from 18 to 3.3 mN/m. The reduction in γp renders the substrate oleophilic. Finally, we know from the prediction of the Zisman theory that the surface energy of the modified aerogel lies between 36 and 51 mN/m. The calculated surface energy of the modified film based on the Fowkes model is 44 mN/m. The apparent surface energy of the aerogel is likely smaller than the calculated surface energy for the modified thin film because of surface roughness of the aerogels. Therefore, since the modified aerogel successfully separates the non-polar liquid we conclude that the critical surface tension of the non-polar liquid must be less than 44 mN/m.

147

Table 5.2: Surface tension values as calculated from Fowkes’ model of surface energy

Polar γp (mN/m) Dispersive γd (mN/m) Total γt (mN/m)

Untreated CDA 18.3 45.8 64.1 TCOS-modified CDA 3.3 40.6 43.9

5.5 Conclusions

The cellulose diacetate (CDA or CA) aerogels were modified by chemical vapor deposition to render them hydrophobic and oleophilic. The modified aerogels demonstrated fast oil uptake and good oil retention. Even in the agitated oil in water media, the modified aerogels sorbed low viscosity kerosene oil within 45 s and the uptake of high viscosity oil (170 cP) was well within 2 min. Moreover, the modified aerogels uphold their mechanical integrity in turbulent media, thus preventing any secondary contamination. These properties make them suitable for potential use to clean up oil spills in the rough marine environments.

The modified CDA aerogels were further identified for fluid selectivity based on surface energy analysis. The results from the modified Zisman theory indicated the critical surface energy for the aerogel to be in the range from 36 to 51 mN/m. The value of interfacial surface tension of spin coated CDA modified with chemical vapor deposition was calculated to be 43.9 mN/m using

Fowkes’ theory. Thus, a non-polar liquid with surface tension lower than 44 mN/m could be separated from water media using these modified CDA aerogels. This analysis served two purposes: it showed the versatility of these aerogels in sorption of other chemicals, and a theoretical platform was established to a priori predict how a non-polar chemical can be separated by these aerogels based on the surface tension of the chemical.

148

5.6 References

[1] Aguilera F, Méndez J, Pásaroa E, Laffona B. Review on the effects of exposure to spilled oils on human health. J Appl Toxicol 2010;30:291–301. doi:10.1002/jat.1521.

[3] Fingas M. Oil Spill Science and Technology. Elsevier, Amsterdam 2011;a.

[4] Fingas M. “In-situ burning” in Oil Spill Science and Technology. Elsevier, Gulf publishing company, New York NY; 2011, p. 737–903.

[5] Michel J, Adams EE, Addassi Y, Copeland T, Greeley M, James B, et al. Oil Spill Dispersants. Efficacy and Effects. Washington, DC: 2005.

[6] Rosales PI, Suidan MT, Venosa AD. A laboratory screening study on the use of solidifiers as a response tool to remove crude oil slicks on seawater. Chemosphere 2010;80:389–95. doi:10.1016/j.chemosphere.2010.04.036.

[7] Correa M, Padron E, Petkoff I. The San Rafael De Laya oil spill : A Case of cleanup and remediation in Venezuela. Int. oil spill Conf., 1997, p. 932–5.

[8] Aldhous P, Hecht J. Beware long-term damage when cleaning up oil spills. New Sci 2010;10:2765. doi:10.1016/S0262-4079(10)61464-9.

[9] Graham B, Reilly WK, Beinecke F, Boesch DF, Garcia TD, Murray C a. Deep Water: The Gulf Oil Disaster and the Future of Offshore Drilling. 2011.

[10] Teas C, Kalligeros S, Zanikos F, Stournas S, Lois E, Anastopoulos G. Investigation of the effectiveness of absorbent materials in oil spills clean up. Desalination 2001;140:259–64. doi:10.1016/S0011-9164(01)00375-7.

[11] Adebajo MO, Frost RL, Kloprogge JT, Carmody O, Kokot S. Porous Materials for Oil Spill Cleanup : A Review of Synthesis. J Porous Mater 2003:159–70. doi:10.1023/A:1027484117065.

[12] Carmody O, Frost R, Xi Y, Kokot S. Adsorption of hydrocarbons on organo-clays- Implications for oil spill remediation. J Colloid Interface Sci 2007;305:17–24. doi:10.1016/j.jcis.2006.09.032.

[13] Reynolds JG, Coronado PR, Hrubesh LW. Hydrophobic aerogels for oil-spill clean up - Synthesis and characterization. J Non Cryst Solids 2001;292:127–37. doi:10.1016/S0022- 3093(01)00882-1.

[14] Ma Q, Liu Y, Dong Z, Wang J, Hou X. Hydrophobic and nanoporous chitosan-silica composite aerogels for oil absorption. J Appl Polym Sci 2015;132:41770.

149

doi:10.1002/app.41770.

[15] Zhao Y, Hu C, Hu Y, Cheng H, Shi G, Qu L. A versatile, ultralight, nitrogen-doped graphene framework. Angew Chemie - Int Ed 2012;51:11371–5. doi:10.1002/anie.201206554.

[16] Sun H, Xu Z, Gao C. Multifunctional, ultra-flyweight, synergistically assembled carbon aerogels. Adv Mater 2013;25:2554–60. doi:10.1002/adma.201204576.

[17] Mikhalchan A, Fan Z, Tran TQ, Liu P, Tan VBC, Tay TE, et al. Continuous and scalable fabrication and multifunctional properties of carbon nanotube aerogels from the floating catalyst method. Carbon N Y 2016;102:409–18. doi:10.1016/j.carbon.2016.02.057.

[18] Sehaqui H, Salajková M, Zhou Q, Berglund LA. Mechanical performance tailoring of tough ultra-high porosity foams prepared from cellulose I nanofiber suspensions. Soft Matter 2010;6:1824. doi:10.1039/b927505c.

[19] Sehaqui H, Zhou Q, Berglund LA. High-porosity aerogels of high specific surface area prepared from nanofibrillated cellulose (NFC). Compos Sci Technol 2011;71:1593–9. doi:10.1016/j.compscitech.2011.07.003.

[20] Jin H, Kettunen M, Laiho A, Pynnönen H, Paltakari J, Marmur A, et al. Superhydrophobic and superoleophobic nanocellulose aerogel membranes as bioinspired cargo carriers on water and oil. Langmuir 2011;27:1930–4. doi:10.1021/la103877r.

[21] Korhonen JT, Kettunen M, Ras RH a, Ikkala O. Hydrophobic nanocellulose aerogels as floating, sustainable, reusable, and recyclable oil absorbents. ACS Appl Mater Interfaces 2011;3:1813–6. doi:10.1021/am200475b.

[22] Cervin NT, Aulin C, Larsson PT, Wågberg L. Ultra porous nanocellulose aerogels as separation medium for mixtures of oil/water liquids. Cellulose 2012;19:401–10. doi:10.1007/s10570-011-9629-5.

[23] Jiang F, Hsieh Y-L. Super water absorbing and shape memory nanocellulose aerogels from TEMPO-oxidized cellulose nanofibrils via cyclic freezing–thawing. J Mater Chem A 2014;2:350–9. doi:10.1039/c3ta13629a.

[24] Jiang F, Hsieh Y-L. Amphiphilic superabsorbent cellulose nanofibril aerogels. J Mater Chem A 2014;2:6337–42. doi:10.1039/c4ta00743c.

[25] Yang X, Cranston ED. Chemically Cross-Linked Cellulose Nanocrystal Aerogels with Shape Recovery and Superabsorbent Properties. Chem Mater 2014;35:6016–25. doi:10.1021/cm502873c.

[26] Zhang W, Zhang Y, Lu C, Deng Y. Aerogels from crosslinked cellulose nano/micro-fibrils and their fast shape recovery property in water. J Mater Chem 2012;22:11642. doi:10.1039/c2jm30688c.

150

[27] Fischer F, Rigacci A, Pirard R, Berthon-Fabry S, Achard P. Cellulose-based aerogels. Polymer (Guildf) 2006;47:7636–45. doi:10.1016/j.polymer.2006.09.004.

[28] Rigacci A. Cellulosic aerogels for energy applications. ACS- Am. Chem. Soc. 2008 Fall Natl. ACS Meet., 2008.

[29] Good RJ. Contact angle, wetting, and adhesion: a critical review. vol. 6. 1992. doi:http://dx.doi.org/10.1163/156856192X00629.

[30] Zisman WA. Relation of the Equilibrium Contact Angle to Liquid and Solid Constitution. Contact Angle, Wettability, Adhes 1964;43:1–51. doi:doi:10.1021/ba-1964- 0043.ch001\n10.1021/ba-1964-0043.ch001.

[31] Fowkes FM. Additivity of intermolecular forces at interfaces. Determinattion of the contribution to surface and interfacial tensions of dispersion forces in various liquids. J Phys Chem 1963;67:2538–41. doi:10.1021/j100806a008.

[32] Fowkes FM. Attractive forces at interfaces. Ind Eng Chem 1964;56:40–52. doi:10.1021/ie50660a008.

[33] Owens DK, Wendt RC. Estimation of the surface free energy of polymers. Jounral Appl Polym Sci 1969;113:1741–7.

[34] van Oss CJ, Good RJ, Chaudhury MK. Additive and nonadditive surface tension components and the interpretation of contact angles. Langmuir 1988;4:884–91. doi:10.1021/la00082a018.

[35] Neumann AW, Good RJ, Hope CJ, Sejpal M. An equation-of-state approach to determine surface tension of low-energy solids from contact angle. J Colloid Interface Sci 1974;49:291–304.

[36] Spelt JK, Absolom DR, Neumann AW. Solid surface tension: The Interpretation of Contact Angles by the Equation of State Approach and the Theory of Surface Tension Components. Langmuir 1986;3:620–5.

[37] Spelt JK, Neumann AW. Solid surface tension: The equation of state approach and the theory of surface tension components. Theoretical and conceptual considerations. Langmuir 1987;3:588–91.

[38] Lee LH. Scope and limitations of the equation of state approach for interfacial tensions. Langmuir 1993;9:1898–905.

[39] Yuan Y, Lee RT. Surface science techniques. vol. 51. 2013. doi:10.1007/978-3-642-34243- 1.

[40] Rulison C. So you want to measure surface energy. Krus USA 1999;49.

[41] Cunha AG, Freire C, Silvestre A, Neto CP, Gandini A, Belgacem MN, et al. Preparation of

151

highly hydrophobic and lipophobic cellulose fibers by a straightforward gas-solid reaction. J Colloid Interface Sci 2010;344:588–95. doi:10.1016/j.jcis.2009.12.057.

[42] Mohan T, Kargl R, Tradt KE, Kulterer MR, Braćić M, Hribernik S, et al. Antifouling Coating of Cellulose Acetate Thin Films with Polysaccharide Multilayers. Carbohydr Polym 2014;116:149–58. doi:10.1016/j.carbpol.2014.04.068.

[43] Rojo E, Peresin MS, Sampson WW, Hoeger IC, Vartiainen J, Laine J, et al. Comprehensive elucidation of the effect of residual lignin on the physical, barrier, mechanical and surface properties of nanocellulose films. Green Chem 2015;17:1853–66.

152

Chapter 6: Experimental and predictive description of the morphology and porosity of wet-spun fibers

6.1 Abstract

Up to now the description of the morphology of non-solvent-induced phase separated structures is only possible by the tedious process of constructing ternary phase diagrams that ultimately gives only approximate information. We address this challenge by proposing and applying, for the first time, design principles based on the combination of (1) Relative Energy Difference (RED) of

Hansen solubility and (2) a kinetic parameter ‘T’ that incorporates mass transfer considerations.

Predictions derived from (1) and (2) are applied and experimentally validated for the synthesis of porous fibers via wet-spinning of polymer solutions. The design principles allowed the possibility to qualitatively tailor the porosity and morphology of the obtained systems, depending on the selection of the solvent and non-solvent for a polymer of interest. Three variables relevant to fibers production by wet-spinning are found sufficient in our description, namely, choice of polymer, solvent and non-solvent, which were varied, and the resultant fiber morphology confirmed via electron microscopy imaging. An agreement between the predicted and experimentally measured fiber morphology and size is determined. The practicality of the design parameters is further put to test by using data from published literature on wet-spinning, providing a basis for future developments in this important area that currently explores nanomaterials derived from renewable resources.

6.2 Introduction

6.2.1 In context to the dissertation

In chapters leading up to this chapter, we investigated the interaction and structural arrangement of biomass components (Chapter 2), we also utilized one form of chemically

153

modified biomass, cellulose esters, to synthesize light-weight and strong aerogels with honeycomb-like morphology (Chapter 3). The solubility parameters were invoked in Chapter 4 to tailor the aerogel mechanical properties and in Chapter 5 we explored one potential application of these aerogel for oil-spill remediation. In this Chapter, we again use to concepts of solubility parameters, however, to predict the non-solvency of a polymer, coupled with kinetics of mass transfer to explain the structure of porous fibers formed during wet-spinning.

6.2.2 Wet-spinning and porous fibers

Fiber synthesis through the wet-spinning processes have been a popular research topic for many years. Some of the more recent developments in the area include wet spinning of regenerated silk in ionic liquids,1 filaments spun from cellulose nanofibril,2 liquid crystalline graphene oxide fibers in alkaline solutions,3 and a plethora of other examples,2,4–11 enabling advances of textiles, electronics, foods, filtration, catalysis and medicine. Wet-spinning benefits over melt-blowing because it allows production of fibers from otherwise temperature sensitive biodegradable compounds, which can be used in medical applications.4 Additionally, wet-spinning may induce porous fiber morphology/structure that can be manipulated to fit a desired application, e.g. if a particular pore size is needed in filtration, or a non-porous textile is desired.

Even with expansion of the field, the reasons behind the development of fiber structure during wet-spinning have not been fully answered. The desired structure is typically achieved by optimizing the solvent, non-solvent, polymer concentration, and other variables via trial and error over a range of experiments.12,13 The pores in the fiber synthesis during wet-spinning can originate due to non-solvent induced phase separation. There are number of studies investigating the formation of porous membranes via non-solvent induced phase separation, that demonstrate thermodynamic interactions and diffusion kinetics, which are the two key processes that determine

154

the bulk morphology and thus properties of membranes formed via non-solvent induced phase separation.14–20 However, there are only a few studies that consider the aspects of pore formation in fibers during wet-spinning.13,21–24

As discussed in Chapter 1, section 1.6.2.3, conventionally, ternary phase diagrams like shown in Figure 1.11, are used to explain the thermodynamic aspect of non-solvent induced phase separation and the bulk morphology of the fiber during wet-spinning strongly depends on the composition path taken by the ternary system in relation to its binodal curve, which in turn depends on the following: a) initial polymer concentration, b) thermodynamic interactions among the polymer, the solvent and the non-solvent and c) the kinetics of inter-diffusion of the solvent and the non-solvent.

In this Chapter, we develop two parameters, one based on Hansen solubility parameters

(HSP) and the other based on diffusion kinetics of the solvent and non-solvent, to predict and explain the fiber morphology formed during wet-spinning. The efficacy of the design parameters is tested experimentally by varying polymer, solvent, non-solvent, and investigating the ensuing fiber morphology. The design parameters are also verified against previously reported studies.

Furthermore, due to the simplicity of the model, its applications are not only limited to fibers, but can be extended to understand and predict the formation of particles and films formed by non- solvent precipitation process. We envisage that this fundamental work will enhance the knowledge regarding the porous fiber production.

6.3 Developing design parameters

The ternary phase diagram is usually used to explain phase separated systems. Such diagram can be developed theoretically or experimentally, but it is a tedious process and even then, it gives qualitative understanding of the bulk morphology.25–27 Additionally, phase separation in

155

the ternary systems is a complicated phenomenon to approach theoretically due to thermodynamic effects, non-linear multiple diffusion, non-isothermal processes, convection transport and others.

Previously, Cohen28, Reuvers15,16, Tsay29, Radovanovic17,18 and Cheng30 have proposed several models to investigate the polymer composition path during precipitation in thin films. These models were based on the diffusion and continuity equations combined with thermodynamics of irreversible processes. In spite of these efforts, it is still difficult to correlate and predict the membrane/fiber morphology for any chosen system. Furthermore, the complexity involved in theoretical development of model in non-solvent induced phase separation process entails development of a simpler predictive model that is quick to implement. Here, the focus is on developing two parameters, one based on Hansen solubility parameters (RED) and the other based on time constant for solvent diffusion (T), that can rationalize the choice of solvent and non-solvent for a polymer of interest. Additionally, these parameters can allow to predict and investigate the fiber structure.

6.3.1 Hansen Solubility Parameters (HSPs)

The solubility parameters are usually employed to identify the solubility of a polymer in a solvent (explained in Chapter 1, section 1.6.2.2). Here, we use solubility parameters to identify the tendency of non-solvency of a polymer in a non-solvent. Hansen solubility parameters (HSPs) are used, as they are an improvement over conventionally used Hildebrand solubility parameters.31

In HSP, the cohesive energy density is dissociated into three components: the dispersion (δd), the polar (δp) and the hydrogen bonding (δh) that arise, respectively, from van der Waals, dipole and hydrogen bonding interactions.31 These HSP, can be factored in or calculated as Ra, Equation 6.1.

2 2 2 2 2 2 Ra = 4(( d1 −  d 2 )+ ( p1 −  p2 )+ ( p1 −  p2 )) ---Equation 6.1

156

where, Ra is the difference between the HSP of a solvent (1) and a polymer (2) and the constant 4 is an empirical correlation. The Ra is normalized with the “interaction radius” R of a polymer as

Ra/Rₒ and is termed as Relative Energy Difference (RED). The interaction radius Rₒ for a polymer can be identified experimentally by testing the solubility of the polymer with various solvents of known solubility parameters (δd, δp, δH). For most of the commercial polymers, the value of Rₒ can be found in literature. A value of RED < 1 implies good solubility for the polymer in the given solvent. Cellulose acetate (CA) and polystyrene (PS) were chosen polymers for this study.

Cellulose acetate is a commercial biopolymer derived from biodegradable and renewable resource cellulose. It is slightly hydrophilic due to the surface hydroxyl groups, whereas PS is a synthetic polymer that is entirely hydrophobic. The HSPs for CA, PS, solvents and their respective REDs are listed in Table 6.1.

Table 6.1: Hansen solubility parameters of selected Polymers and solvents in MPa1/2. 31,32

RED RED δd δp δH R ₒ (CA) (PS) CA 18.6 12.7 11 7.6 -- -- Polystyrene 21.3 5.8 4.3 12.7 -- -- DMAC 16.8 11.5 10.2 -- 0.5 0.9 Acetone 15.5 10.4 7.0 -- 1.0 1.0 DMSO 18.4 16.4 10.2 -- 0.5 1.0 DMF 17.4 13.7 11.3 -- 0.3 1.0 Water 15.6 16 42.3 -- 4.2 3.2 Methanol 15.1 12.2 22.2 -- 1.7 1.8

Based on HSP, we propose the following: First, a combination of solvent and non-solvent can be identified for a chosen polymer for fiber formation via wet-spinning. For instance, based on the RED of CA and PS with respect to different solvents, it is predicted that DMAc, acetone,

157

DMSO and DMF are good solvents for CA and PS whereas methanol and water behave as non- solvent for these polymers.

1 Second, RED  . This implies that higher the value of RED, smaller is the boundary BG gap in the phase diagram. Therefore, a quick and qualitative prediction is possible regarding tendency of de-mixing by easily calculating RED of the chosen polymer w.r.t. non-solvent. For instance, since, CA has higher RED with water (REDwater = 4.2) than methanol (REDmethanol = 1.7),

CA is expected to have low BG with water as non-solvent on ternary diagram, as compared to when methanol is used as non-solvent. Therefore, a higher tendency of de-mixing is predicted for

CA in water than in methanol. Same argument goes for PS based on its RED values. It is noteworthy that RED parameter defined here does not account for solvent-non-solvent interactions.

In practice, fiber formation via immersion precipitation is a dynamic process where the composition of the three phases change rapidly, not only with time but also radially, due to fast multi-component mass transfer. Thus, the kinetics of the inter-diffusion of solvent and non-solvent must also be considered.

6.3.2 Kinetics of solvent-non-solvent inter-diffusion during phase separation

To understand the model for the kinetics of solvent-non-solvent inter-diffusion, let us first briefly explain the wet-spinning process used in this study as illustrated in Figure 6.1a. The polymer solution is extruded through a syringe at the fixed rate using a syringe pump, during which, the needle is submerged in the non-solvent. Figure 6.1b illustrates the cross-section of the extruded polymer solution as soon as it meets the non-solvent. The polymer solution/non-solvent interface phase separates and solidifies forming a small skin layer of thickness ‘W’. This solidification front gradually moves inside the core in the radial direction as the fiber is drawn

158

down via gravity. However, the model is developed at the small-time limit, t = ε →0+ of the polymer extrusion.

The concentration of the non-solvent (퐶푁푆) is defined as the mass of non-solvent divided by the total volume of non-solvent. Therefore, 퐶푁푆 = 퐶̃푁푆 as r ⟶∞, which is nothing but the density of non-solvent (휌푁푆). At the interface of polymer solution/non-solvent (r=R), the concentration of

+ non-solvent, 퐶푁푆= 퐶푅(푡) as t⟶0 . The illustration assumes a linear concentration profile of non- solvent in the skin layer ‘W’, as well as at the small distance outside the skin layer due to outward diffusion of the solvent. The thickness ‘W’ is a function of time W(t) at small time interval. Certain other assumptions are also made to develop the model (see Appendix A6), as were also stated by

Patsis et al.33, for non-solvent induced phase separation in thin films. Most importantly, a “lag time” is assumed that corresponds to the interval between the initial polymer solution/non-solvent contact and the time tₒ, at which the solvent first diffuses through the skin.

159

+ the fiber and the skin thickness ‘W’ is varying as W(t) at t→ 0 . 퐶푁푆 is defined as the mass of non-solvent per unit volume of non-solvent. 퐶푁푆= 퐶̃푁푆 at r⟶∞ and 퐶푁푆 = 퐶푅(푡) at r=R. The assumed linear + concentration profile of 퐶푁푆 within the skin layer is illustrated at small time scale, t⟶0 . The solvent concentration is 퐶푆= 퐶(푡) at r=R, and 퐶푆= 퐶0 at r= R-W.

The growth-rate of the volume of the skin-layer can be expressed as follows: dV d22 dW skin =(RRWLLRW) − ( −) 2  ( − ) Equation 6.2 dt dt  dt

The change in the volume of the non-solvent (VNS) as the non-solvent diffuses (Vdiffusion) through the skin-layer can be written as,

VVVNS=− fixed diffusion Equation 6.3

The fixed part of the non-solvent volume, Vfixed is very large and approximately constant compared to the changing part, Vdiffusion.

Now, the change in the non-solvent volume with respect to time can be written as following:

160

dVNS d dVdiffusion =(VVfixed − diffusion ) = − Equation 6.4 dt dt dt

According to our definition of 퐶푁푆 at the surface interface that is equal to 퐶푅(푡), the mass rate of non-solvent diffusion through the skin layer can be written as follows:

dV dC  diffusion  V R Equation 6.5 NSdt fixed dt

dV V dC =diffusion  R Equation 6.6 dt dt

Where V = Vfixed and ρNS = ρ. Equation.6.6 contains the concentration accumulation term of the non-solvent in the skin layer. The accumulation of non-solvent in the skin layer is mainly due to the diffusion of non-solvent from the non-solvent envelope to skin layer (Figure 6.1b). As there is no radial velocity component at the outer boundary of non-solvent envelope (r→∞), the contribution of convection term is neglected. From the Fick’s equation of diffusion in radial coordinates, the following expression can be written:

CCCRRR 1    DNS      ==DNS  r     r   Equation 6.7 t r  r  r R  r  r skin   skin     skin

Note, that at the limit of infinite dilution, the DNS (diffusion coefficient of non-solvent in solvent)

23 and DS (diffusion coefficient of solvent in non-solvent) are not commutative.

At a very small time-scale, a linear concentration profile of the non-solvent reveals that radial concentration gradient can be approximated as follows:

CC (RWR−−) RR( )   CCRR rrskin skin  r == Equation 6.8 r  r R − W − R  r skin Skin

The radial concentration gradient of the non-solvent can also be further approximated as:

161

C CC− 0 R =NS = − NS Equation 6.9 rSkin R − W − R W

Note that the negative sign indicates that 퐶푅 is decreasing in the radial direction. Thus, from

Equations. 6.6-6.9, the change in non-solvent volume becomes: dV VVdC DC diffusion =R = − NS NS Equation 6.10 dt dtR R W

The skin thickness ‘W’ is time dependent and the rate of skin growth can be written as directly proportional to the steady state volume flux of non-solvent in to the system. For the sake of simplicity in the model, the proportionality constant is assumed as unity.

dW VD C 2 LRW( −=) NS NS Equation 6.11 dt RW

Assuming at t=0, W=0, the differential equation can be written in the following integral form:

WtW VD C W1−= dWNS NS dt Equation 6.12  2 00RLR2

WW23VD C = − = NS NS t Equation 6.13 2 3RLR 22

W 3 The term represents the effect of curvature on the growth of skin. If we consider that W<

W 2 VD C = NS NS t Equation 6.14 22LR2

=W = A t Equation 6.15

VD C Where A is constant which equals to NS NS . LR2

162

An approximate formula of the skin thickness, W at t →0 is obtained (Equation 6.15) considering the radial diffusion of the non-solvent and square root dependence of W on time is found, which is also consistent with the literature data for diffusion in thin films.33

Now, while writing the equation for solvent mass transfer in the skin-layer, time dependence of the solvent concentration ( C(t) ) is also treated by neglecting the convection transport as there is no external perturbation at the boundaries of skin layer to create radial velocity.

CCC 1    DS       =DS  r     r   Equation 6.16 tskin  r  r  r   R   r   r   skin skin

Similar to the non-solvent, the linear profile approximation for the solvent provides similar magnitudes of the first and second polar derivatives of solvent concentration:

  CC   CC0 − r == Equation 6.17 r  r  r W skin skin

C0 is the concentration of solvent in the interface of skin-layer and polymer solution, at r=R-W (it is considered to be fixed in the small-time scale) and C(t) is the concentration of solvent in the interface of skin and non-solvent layer (r=R). From Equations 6.15, 6.16 and 6.17, we obtain: dC D (C − C) = S 0 Equation 6.18 dt R A t

Assuming, at t=0, C=0 and integrating Equation 6.18, we obtain the following expression of C,

  1     t  2    C(t) = C0 1− exp−    Equation 6.19    s     

163

In the Equation 6.19, τs is a time scale for the solvent diffusion. All else being equal, τs→∞

represents no diffusion of the solvent to the non-solvent layer and τs→0 represents the complete diffusion of solvent to the non-solvent layer. Figure A6.1 reveals that as the time-scale increases, the solvent removal process becomes slower.

The expression of τs is written as follows:

 ~  V C NS  D   =   NS  Equation 6.20 S  4L  D 2    S 

D =  NS =T Equation 6.21 S D 2 S

The rate of solvent diffusion into non-solvent can be qualitatively compared with τs or ‘T’. A smaller value of T indicates a faster diffusion of solvent into non-solvent. This further implies a smaller fiber diameter if all other parameters are kept constant. Table 6.2 gives the values of diffusivities of the solvent in non-solvent (DS-NS) and non-solvent in solvent (DNS-S) as calculated from Wilke-Chang correlation.34 It should be noted that in the limit of infinite dilution the diffusivities are non-commutative.

Table 6.2: Diffusivity values of solvent in non-solvent (DS-NS) and non-solvent in solvent (DNS-S) as calculated from Wilke-Chang correlation.34

Non- DS- DNS- Solvent 5 5 Solvent NS(x10 S(x10 (S) (NS) cm2/s) cm2/s) DMAC Water 0.91 1.70 DMAC Methanol 1.93 2.48 DMSO Water 1.07 0.68 DMSO Methanol 2.26 0.98

164

6.4 Experimental Section

6.4.1 Materials

Cellulose acetate was provided by Eastman chemical company with DS of 2.45 and molecular weight of 30,000 g/mol. Polystyrene was purchased from Scientific Polymer Products

Inc. with molecular weight of 260,000 g/mol. All solvents were obtained from Sigma Aldrich and were used without further treatment. Reverse osmosis purified water was used in the experiments for all intended purposes.

6.4.2 Experimental set-up

A small-scale laboratory set up was built as shown in Figure 6.1a. Polymer solution was injected through a syringe at the fixed rate of 0.1ml/min. The needle diameter was 0.412 mm. The non-solvent was filled in a 1liter test tube and the syringe needle was immersed in the non-solvent during the experiment. The height of the measuring cylinder was 40 cm. After the experiment, the as spun fibers were manually collected onto a clean roller. There was no drawing step involved in the experiment.

6.4.3 Ternary Phase Diagram

A known concentration of the polymer solution in DMAC was prepared. The non-solvent water or methanol was added dropwise under constant vigorous stirring until the solution turned cloudy. The final concentration of the three components were noted at the cloud point. The polymer solutions were made starting from 1 wt% at unit interval. The CDA solutions with a maximum concentration of 12 wt% and PS solutions with a maximum concentration of 40 wt% were used.

165

6.4.4 Fiber Characterization a) Density and Porosity

The polymer fiber density was measured by their mass and volume. The mass of the fiber was measured by analytical balance, Fisher Scientific Accu-225D, which has least count of 0.01 mg and the volume was calculated by measuring the fiber diameter by digital screw gauge and fiber length by digital Vernier caliper. Average density is reported after 10 measurements. The porosity was calculated using Equation 6.22.

   Porosity = 1− f 100% ---Equation 6.22     p  where, ρf is the bulk density of the fiber and ρp is the bulk density of polymer as provided by the vendor, 1.3 g/cm3 for CA and 1.05 g/cm3 for PS. b) Scanning Electron microscopy (SEM)

Imaging was done using Field Emission Scanning Electron Microscope (FESEM), FEI,

Verios 490L. The polymer fibers were fractured under liquid N2 using a sharp clean blade to image the radial cross-section. The samples were fixed on the 45° inclined metal stub using a double- sided carbon tape. The samples were coated with 5nm layer of gold and palladium before imaging. c) N2 adsorption isotherms

The N2 adsorption and desorption were measured using Micrometrics ASAP 2020. The

Brunauer-Emmett-Teller (BET) surface area was calculated from N2 adsorption isotherm between the relative pressures of 0.1-0.3 at -196 °C. About 0.1-0.2 g of sample was degassed for 20 h at 60

°C prior to the analysis. The pore size distribution curve was derived from N2 desorption curve of the isotherm by Barrett-Joyner-Halenda (BJH) method.

166

6.5 Experimental Results and Discussion in relation to developed parameters

The two parameters defined by ‘RED’ and ‘T’ can give a fairly accurate qualitative prediction about the fiber structure. The parameter ‘RED’ allows to choose a set of solvent and non-solvent for a polymer of interest. The parameter ‘T’ along with ‘RED’ qualitatively predicts if the polymer tends to undergo delayed or instantaneous demixing, which in turn gives an idea about fiber morphology. Additionally, ‘T’ parameter can give a qualitative comparison of fiber diameters. From here on, we test if the predictions from these two parameters correlate with the experiments.

There are three essential components in fiber making via wet-spinning: i) Non-solvent, ii) solvent, and iii) polymer. For a chosen polymer, the solvent and non-solvent were identified based on HSP. The various combinations are investigated, and the resulting fiber morphology, density, pore volume and fiber diameter are listed in Table 6.3. In the following sections, the effect of each variable on fiber structure will be discussed in detail.

167

Polyme Solven Non- R T Morphology Densit Pore Diamete r t Solvent E (x10- y Volume r (µm) D 5) (g/cm3) (%) s/cm2 CA DMA Water 4.2 2.05 MV, MC 0.71±.0 45.3±3. 190±10 C 4 4 CA DMA Methan 1.7 0.67 MC 0.41±.0 68.2±6. 100±12 C ol 8 4 CA DMSO Water 4.2 0.59 MV, MC 0.88±.0 32.0±4. 135±10 6 3 CA DMSO Methan 1.7 0.19 MC 0.63±.0 51.6±6. 63±2 ol 8 4 PS DMA Water 3.2 2.05 MV, MC 0.35±.0 67.0±1. 360±2 C 2 6 PS DMA Methan 1.8 0.67 MV, MC 0.32±.0 69.7±0. 270±8 C ol 1 7

6.5.1 Choice of non-solvent

The SEM images of the CA in DMAc polymer fibers obtained by precipitating in water and methanol are shown in Figure 6.2, where both types of fibers demonstrate circular cross- section with bulk porosity. However, there are a few striking dissimilarities. First, the fiber obtained by precipitating in water exhibits a number of macro-voids that are concentrated close to the fiber surface, shown in Figure 6.2a1 ,d1. On the contrary, CA fiber obtained via precipitation in methanol do not exhibit any macro-voids and displays uniform bulk porosity (Figure 6.2a2, d2). Second, the surface of the fiber obtained from precipitating in water has non-porous skin

(Figure 6.2b1, c1 and c1 inset), when compared to the fiber formed by precipitating in methanol

(Figure 6.2b2, c2 and c2 inset) which has no skin and a uniform porosity throughout the surface.

In both cases, however, the fiber core exhibits microcellular or spongy morphology (Figure 6.2e1, e2). The difference in the fiber structure is also manifested in form of their density and pore volume 168

as can be seen in Table 6.3. The fibers formed in water have higher density (0.71 g/cm3) and lower pore volume (45.3 %) than fibers formed in methanol (ρ=0.41 g/cm3 and pore volume 68.2 %).

And lastly, a noteworthy difference in fiber diameters. Using the same needle, the fiber formed by precipitating in water has diameter of 190 µm, whereas the fiber formed by precipitating in methanol has diameter of 100 µm.

These morphological and structural properties can be qualitatively predicted by invoking the defined parameters (RED and T) together. The RED of CA with water is 4.2, and with methanol, it is 1.7, therefore, binodal gap (BG) of CA is lower for water than for methanol as non- solvent. Lower BG results in faster phase separation, which is confirmed experimentally from the cloud point curve obtained on the ternary phase diagram of CA-DMAC-Non-solvent as shown in

Figure 6.3a. Therefore, CA has higher tendency to phase separate in water than in methanol. As

169

soon as the CA was extruded in water, instantaneous de-mixing caused a skin formation, which acts as a physical barrier and slows down the inter diffusion of solvent and non-solvent. On the contrary, when the polymer is extruded in methanol, due to low tendency of CA to phase separate in methanol, formation of skin layer, if any, is delayed and the polymer fiber is stretched by gravity before solidification of the surface. This allows free inter-diffusion of solvent and non-solvent as illustrated in Figure 6.3b. At the same time, diffusion kinetics of solvent and non-solvent start shaping the fiber structure, forming micropores throughout the fiber cross-section. The kinetic

2 2 parameter (T) of DMAC is 2.05 s/cm in water and 0.67 s/cm in methanol (Table 6.3), implying that resistance to solvent (DMAC) diffusion is higher in water than in methanol. Therefore, due to instantaneous de-mixing and slow diffusion of DMAC in water, a fiber was formed with large diameter (190 µm), whereas, for non-solvent methanol, delayed de-mixing and faster diffusion of

DMAC in methanol yielded a thinner fiber (100 µm diameter). It is noted that needle diameter was constant in both the cases.

The formation of macrovoids while coagulating in water can also be explained by instantaneous demixing of CA in water. The solidification of surface due to instantaneous demixing, traps water pockets underneath the solidified fiber surface, resulting in formation of macrovoids. No such entrapment is possible in methanol as non-solvent due to delayed demixing of CA in methanol. The diffusion of non-solvent in both cases is slowest in the fiber core due to hindrance from few microns of sub surface solidified polymer structure that allows formation of microcellular structure. A similar kind of morphology has been observed for CA membranes where macrovoids are seen close to the surface in contact with non-solvent and the bottom surface remains spongy-like structure.25,35,36 However, to the best of our knowledge, the microcellular type structure reported here for CA fibers have not been observed previously for CA wet spun fibers.

170

These kinds of microcellular structures have been previously reported for membranes formed via thermal induced phase separation when the polymer composition crosses the spinodal curve.37 We do not have enough information to confirm that the CA fibers formed in our case are via spinodal decomposition. However, the obtained fibers have mesoporous structure as confirmed from N2 adsorption-desorption analysis (Figure A6.2) that shows pore size distribution in the range of 40-

2 50 nm. The BET area calculated from the slope of N2 adsorption curve is 20-25 m /g.

6.5.2 Choice of the solvent

In the second variation of experimental verification, solvent for CA was changed to DMSO and the fibers were formed by coagulation in either water or methanol. The RED of CA and PS with the two non-solvents will be same as before, i.e., REDCA-Water = 4.2, REDCA-Methanol = 1.7. The only difference was the choice of solvent: DMSO and DMAC, which changes the kinetic

171

parameter T. We observe that with DMSO as the solvent, fiber structure closely resembles the structure seen in CA/DMAC/Non-solvent case, that is a circular fiber with mesoporous bulk and presence of macrovoids while precipitating in water (Figure 6.4a1, d1, e1), whereas absence of macrovoids with uniform microcellular structure while precipitating in methanol (Figure 6.4a2, d2, e2). On similar lines, the fiber surface was non-porous in case of coagulation in water (Figure

6.4b1, c1 and c1 inset), whereas a uniform porous structure was observed on the fiber surface while coagulation in methanol (Figure 6.4b2, c2 and c2 inset).

With DMSO as solvent, the kinetic parameter (T) is different (Table 6.3, TDMSO = 0.59

2 2 s/cm , compared to TDMAC=2.05 s/cm , both in water). This implies that DMSO diffuses faster than DMAC in water, resulting in a thinner fiber with diameter (135 µm), compared to fiber obtained from CA/DMAC/water system (d=190 µm). Similarly, when methanol is non-solvent,

172

RED is 1.7 and fiber structure appears to be the same as that obtained in CA/DMAC/methanol

2 2 system. However, the kinetic parameter TDMSO = 0.19 s/cm , compared to TDMAC = 0.67 s/cm

(both in methanol), implying that diffusion of DMSO is faster than that of DMAC in methanol.

This results in thinner fiber with a diameter of 63 µm as compared to fiber obtained from

CA/DMAC/methanol system with a fiber diameter of 100 µm.

Apart from differences in solvent diffusivities, fiber diameter can also be affected by its stretching due to gravity, as our experiments were performed by extruding polymer solution into a measuring cylinder filled with the antisolvent. In case of precipitation in water, the fiber stretching was minimum due to instantaneous demixing and solidification of the fiber surface, whereas precipitation in methanol was delayed resulting in higher fiber stretching before solidification of the surface. This may also cause a thinning of fiber diameter in case of precipitation in methanol. However, the fact that a smaller diameter was observed in DMSO/water system (135 µm) compared to DMAC/water (190 µm) (RED parameter is the same in both cases) indicates that kinetic parameter is instrumental in defining the fiber diameter and a follow up study is necessary to quantify its effect to the fiber properties.

To summarize, our observations show that the type of de-mixing dictates the extent of macrovoids, that can be qualitatively predicted from RED values, and strongly depends on the system of solvent - non-solvent used. CA/water has high RED (=4.2) and exhibit macrovoids due to instantaneous de-mixing, whereas CA/methanol has low RED (=1.7) and exhibits microcellular morphology due to delayed de-mixing. The kinetic parameter helps to qualitatively predict the fiber diameter. A low value of T (=0.19 s/cm2) for DMSO/methanol produces CA fiber with smallest diameter (63 µm), whereas a high value of T (=2.05 s/cm2) for DMAC/water produces

CA fiber with largest diameter (190 µm).

173

6.5.3 Choice of the polymer

As RED varies for different polymers, to determine differences between fibers, we compared polystyrene (PS) and CA fibers. The DMAC was used as solvent, and water and methanol were kept as non-solvents. It is noted from previous studies on membrane formation that viscosity η of the polymer solution strongly affects the morphology of the phase separated structure.38 To have a uniform comparison with CA, we selected a high molecular weight PS and a solution concentration of 30 % w/v to match CA and PS solution viscosity (η30 w/v% PS ≈ η10 w/v%

CA ≈ 2.7 Pa-s).

The RED values of PS (with water and methanol are 3.2 and 1.8 respectively,) predict a smaller BG with water than methanol. This is also confirmed via ternary phase diagram of

PS/DMAC/Non-solvent (Figure 6.5). Based on these values, we predict a lower tendency of

174

precipitation of PS in water than in methanol causing instantaneous demixing in water and relatively delayed demixing in methanol. A large number of macrovoids are observed in the fiber formed via coagulation in water (Figure 6.6 a1, d1), due to instantaneous demixing and solidification of the fiber surface. On the contrary, a more microcellular or spongy structure is expected and was formed in the fiber formed by precipitation in methanol (Figure 6.6 a2, d2). In both cases the fiber core has microcellular structure (Figure 6.6 e1, e2) due to hindered diffusion of solvent-non-solvent by solidified surface as mentioned before. The kinetic parameter T for

DMAC/water is 2.05 s/cm2 and for DMAC/methanol is 0.67 s/cm2. Therefore, a faster diffusion of DMAC in water is expected than that of DMAC in methanol. Hence, the fiber diameter is expected to be smaller when the precipitation is carried out in methanol (270 µm) as compared to the fiber diameter when the precipitation takes place in water (360 µm). Comparing PS fiber surface obtained in precipitation in water (Figure 6.6 b1, c1 and c1 inset), and methanol (Figure

6.6 b2, c2 and c2 inset), we can observe a minute difference is fiber surface morphology. The small surface pores (or roughness) were observed on the surface of the fiber formed by coagulation in methanol in contrary to the smooth fiber surface observed for the fiber formed by coagulation in water. The surface pore structure observed for PS is different compared to that of CA fibers.

Moreover, a different kind of macrovoids and microcellular structure is observed in PS than that observed in CA. In case of PS, finger like macrovoids were observed extending towards the fiber core when non-solvent is water. On the contrary, tear drop macrovoids were observed in the PS fiber formed in methanol. This observation emphasizes one drawback of our model, i.e., it cannot be used to compare fiber structures obtained from different polymers. The RED is used qualitatively to compare the BG of the polymer with various non-solvents but cannot be successfully applied across different polymers. For instance, PS and CA have RED of 3.2 and 4.2

175

respectively with water but the BG of PS with water is smaller as compared to that of CA with water (Figure 6.3 a and 6.5). Nevertheless, the model can successfully compare the fiber structure obtained from one polymer when subjected to different solvent-non-solvent systems.

6.5.4 Comparison with the published literature:

We looked for the studies that reported ternary phase diagram of single polymer system with various solvent-non-solvent systems, along with the SEM images of resulting structures obtained by via wet-spinning or non-solvent induced phase separated membranes. We found a few studies in field of membrane science satisfying these criteria, where we tested our model. Table

A6.1 contains the values of solubility parameters for the components used in the following studies.

The RED and T values for each set of system is listed in Table 6.4. Yin et al.39, studied fiber production by extruding PAA/DMAC solution into various composition of water and DMAC. For

PAA, it is expected based on RED values that BG will increase with increase in concentration of

176

solvent (DMAC) in the non-solvent (water). Consequently, the roughness on the fiber surface increases with decrease in RED. Also, the T values will decrease upon addition of DMAC in non- solvent. This will result in decrease of fiber diameter as also reported in the study. Sukitopaneeni et al.24, investigated the morphology of PVDF hollow fiber membranes by wet-spinning solution of PVDF in NMP in the coagulation bath of water and various blends of water with methanol, ethanol and isopropyl alcohol (IPA). The study has a ternary phase diagram of PVDF/NMP and the non-solvents used, that satisfies the criteria of REDs inverse relationship with BG. Hence, a microcellular or spongy morphology is observed when the non-solvent used are 50% methanol/ethanol/IPA with water, because the PVDF undergoes delayed demixing in these non- solvents. It is difficult to comment on the fiber diameter because the study was for hollow fibers, but based on the T values with pure solvents, we hypothesize that PVDF fiber diameter (D) may follow the following order when spun in pure non-solvents: Din IPA>Din Water>Din Ethanol>Din Methanol.

Mazinani et al.40, studied phase separated membranes of Extrem polymer. The values for Extrem satisfy the criteria of RED being inversely proportional to BG, implying that it will undergo delayed demixing in methanol and glycerol exhibiting a spongy morphology. From the T values, we expect that if the fibers were made from this polymer, the diameter will be smallest in methanol and highest in glycerol. The solvent diffusion in glycerol will be minimal as the T value is very high. However, there may be issues of poor spinability due to high viscosity of glycerol. The authors also investigated different solvents. The DMAC, NMP and DMF yielded macrovoid morphology when polymer was phase separated into water, whereas spongy morphology was obtained when DMSO was used as a solvent. The authors attributed the behavior to high viscosity of Extrem in DMSO. Our model does not account for change in viscosity and needs further improvement.

177

Table 6.4: Our design parameters (RED and T) applied to selected studies of phase separated membranes and fibers.

Polyme Solven Non- RE T Morphology Comments Reference r t Solvent D No surface roughness, PAA DMAC water 5.6 2.05 Fiber dia= 50 μm 70% Moderate surface PAA DMAC water/30 3.8 -- roughness, Satisfies % DMAC Fiber dia= 40 μm the 50% Moderate surface condition Yin et al.39, PAA DMAC water/50 2.6 -- roughness of 1 % DMAC Fiber dia= 27 μm RED α 퐵퐺 40% High surface roughness, PAA DMAC water/60 2 -- Fiber dia= 32 μm % DMAC PVDF NMP Water 13.7 1.33 Macrovoids Satisfies 50% the water/ 50 PVDF NMP 9.6 -- Microcellular condition % of methanol 1 RED α . 50% 퐵퐺 PVDF NMP water/ 50 8.9 -- Microcellular If the fibers Sukitpaneeni % ethanol were made t et al.24, 50% in pure PVDF NMP water/ 8.3 -- Microcellular antisolvent. 50% IPA IPA may PVDF NMP Methanol 5.6 0.33 -- give fibers with largest PVDF NMP Ethanol 4.6 0.82 -- diameter PVDF NMP IPA 4.1 4.39 -- Extrem NMP Water 2.8 1.30 Macrovoids Extrem NMP Glycerol 1.7 >>1 Microcellular Satisfies the Extrem NMP Methanol 1.3 0.33 Microcellular condition Mazinani et 40 Extrem DMAC water 2.8 2.05 Macrovoids of al. , 1 Extrem DMF water 2.8 1.55 Macrovoids RED α 퐵퐺 Extrem DMSO water 2.8 0.59 Microcellular

6.6 Conclusion

A model is developed to qualitatively and comparatively predict the fiber morphology and size formed during wet-spinning. By using the two design parameters (RED and T), it is possible to identify a set of solvent and non-solvent combinations for use with the polymer of interest for

178

wet-spinning. Additionally, it is possible to qualitatively predict the ensuing fiber morphology and diameter. An experimental verification of the model is conducted using cellulose acetate and polystyrene. Based on RED, we chose DMAc & DMSO as solvents and water and methanol as non-solvents. Using the model, we predicted that low RED will result in mesporous fiber due to delayed demixing, and high RED will produce less porous fibers, with a possibility of an outer non-porous “skin” layer formation due to instantaneous demixing. The kinetic parameter T predominantly governs the fiber diameter. A small value of T implies fast solvent diffusion that leads to decrease in fiber diameter. This model also shows good agreement with published literature. This versatile model can be applied to any other polymer-solvent-non-solvent system if the Hansen solubility parameter is identified. Furthermore, due to the simplicity of the model, its applications are not limited to fibers, but can be extended to understand and predict the formation of particles and films formed by non-solvent precipitation process. We envisage that this fundamental work will enhance the knowledge regarding the porous fiber production in the industry.

179

6.7 References

[1] Phillips DM, Drummy LF, Naik RR, Long HC De, Fox DM, Trulove PC, et al. Regenerated silk fiber wet spinning from an ionic liquid solution. J Mater Chem 2005;15:4206. doi:10.1039/b510069k.

[2] Lundahl MJ, Cunha AG, Rojo E, Papageorgiou AC, Rautkari L, Arboleda JC, et al. Strength and Water Interactions of Cellulose I Filaments Wet-Spun from Cellulose Nanofibril Hydrogels. Sci Rep 2016;6:30695. doi:10.1038/srep30695.

[3] Jalili R, Aboutalebi SH, Esrafilzadeh D, Shepherd RL, Chen J, Aminorroaya-Yamini S, et al. Scalable One-Step Wet-Spinning of Graphene Fibers and Yarns from Liquid Crystalline Dispersions of Graphene Oxide: Towards Multifunctional Textiles. Adv Funct Mater 2013;23:5345–54. doi:10.1002/adfm.201300765.

[4] Puppi D, Chiellini F. Wet-spinning of biomedical polymers: from single-fibre production to additive manufacturing of three-dimensional scaffolds. Polym Int 2017. doi:10.1002/pi.5332.

[5] Yudin VE, Dobrovolskaya IP, Neelov IM, Dresvyanina EN, Popryadukhin P V., Ivan’kova EM, et al. Wet spinning of fibers made of chitosan and chitin nanofibrils. Carbohydr Polym 2014;108:176–82. doi:10.1016/j.carbpol.2014.02.090.

[6] Behabtu N, Young CC, Tsentalovich DE, Kleinerman O, Wang X, Ma AWK, et al. Strong, Light, Multifunctional Fibers of Carbon Nanotubes with Ultrahigh Conductivity. Science (80- ) 2013;339.

[7] Kong L, Ziegler GR. Rheological aspects in fabricating pullulan fibers by electro-wet- spinning. Food Hydrocoll 2014;38:220–6. doi:10.1016/j.foodhyd.2013.12.016.

[8] Wade SJ, Zuzic A, Foroughi J, Talebian S, Aghmesheh M, Moulton SE, et al. Preparation and in vitro assessment of wet-spun gemcitabine-loaded polymeric fibers: Towards localized drug delivery for the treatment of pancreatic cancer. Pancreatology 2017. doi:10.1016/j.pan.2017.06.001.

[9] Kou L, Huang T, Zheng B, Han Y, Zhao X, Gopalsamy K, et al. Coaxial wet-spun yarn supercapacitors for high-energy density and safe wearable electronics. Nat Commun 2014;5:3754. doi:10.1038/ncomms4754.

[10] Wang L-Y, Yong WF, Yu LE, Chung T-S. Design of high efficiency PVDF-PEG hollow fibers for air filtration of ultrafine particles. J Memb Sci 2017;535:342–9. doi:10.1016/j.memsci.2017.04.053.

[11] Zhao XH, Li Q, Ma XM, Xiong Z, Quan FY, Xia YZ. Alginate fibers embedded with silver nanoparticles as efficient catalysts for reduction of 4-nitrophenol. RSC Adv 2015;5:49534– 40. doi:10.1039/C5RA07821K.

[12] Nakajima T. Advanced fiber spinning technology. 2nd ed. Woodhead Publishing Limited;

180

1994.

[13] Arbab S, Noorpanah P, Mohammadi N, Soleimani M. Designing Index of Void Structure and Tensile Properties in Wet-Spun Polyacrylonitrile (PAN) Fiber. I. Effect of Dope Polymer or Nonsolvent Concentration S. J Appl Polym Sci 2008;109:3461–9.

[14] Guillen GR, Pan Y, Li M, Hoek EM V. Preparation and characterization of membranes formed by nonsolvent induced phase separation: A review. Ind Eng Chem Res 2011;50:3798–817. doi:10.1021/ie101928r.

[15] Reuvers AJ, Smolders CA. Formation of membranes by means of immersion precipitation. Part I. A model to describe mass transfer during immersion precipitation. J Memb Sci 1987;34:67–86. doi:10.1016/S0376-7388(00)80021-6.

[16] Reuvers AJ, Smolders CA. Formation of membranes by means of immersion precipitation Part II. The mechanism of formation of membranes prepared from system of cellulose acetate-acetone-water 1987;34:67–86.

[17] Radovanovic P, Thiel SW, Hwang ST. Formation of asymmetric polysulfone membranes by immersion precipitation. Part I. Modelling mass transport during gelation. J Memb Sci 1992;65:213–29. doi:10.1016/0376-7388(92)87024-R.

[18] Radovanovic P, Thiel SW, Hwang ST. Formation of asymmetric polysulfone membranes by immersion precipitation. Part II. The effects of casting solution and gelation bath compositions on membrane structure and skin formation. J Memb Sci 1992;65:231–46. doi:10.1016/0376-7388(92)87025-S.

[19] van’t Hof JA, Reuvers AJ, Boom RM, Rolevink HHM, Smolders CA. Preparation of asymmetric gas separation membranes with high selectivity by a dual-bath coagulation method. J Memb Sci 1992;70:17–30. doi:10.1016/0376-7388(92)80076-V.

[20] van de Witte P, Dijkstra PJJ, van den Berg JW a. W a, Feijen J. Phase separation processes in polymer solutions in relation to membrane formation. J Memb Sci 1996;117:1–31. doi:10.1016/0376-7388(96)00088-9.

[21] Weisser P, Barbier G, Richard C, Drean J-Y. Characterization of the coagulation process: wet-spinning tool development and void fraction evaluation. Text Res J 2015. doi:10.1177/0040517514551469.

[22] Liu C, Bai R. Preparation of chitosan/cellulose acetate blend hollow fibers for adsorptive performance. J Memb Sci 2005;267:68–77. doi:10.1016/j.memsci.2005.06.001.

[23] Deshmukh SP, Li K. Effect of ethanol composition in water coagulation bath on morphology of PVDF hollow fibre membranes. J Memb Sci 1998;150:75–85. doi:10.1016/S0376-7388(98)00196-3.

[24] Sukitpaneenit P, Chung TS. Molecular elucidation of morphology and mechanical properties of PVDF hollow fiber membranes from aspects of phase inversion, crystallization

181

and rheology. J Memb Sci 2009;340:192–205. doi:10.1016/j.memsci.2009.05.029.

[25] Hao J, Wang S. Calculation of Alcohol-Acetone-Cellulose Acetate Ternary Phase Diagram and Their Relevance to Membrane Formation. J Appl Polym Sci 2001;80:1650–7.

[26] Sun AC, Kosar W, Zhang Y, Feng X. A study of thermodynamics and kinetics pertinent to formation of PVDF membranes by phase inversion. Desalination 2013;309:156–64. doi:10.1016/j.desal.2012.10.005.

[27] Barzin J, Sadatnia B. Correlation between macrovoid formation and the ternary phase diagram for polyethersulfone membranes prepared from two nearly similar solvents. J Memb Sci 2008;325:92–7. doi:10.1016/j.memsci.2008.07.003.

[28] Cohen C, Tanny GB, Prager S. Diffusion-controlled formation of porous structures in ternary polymer systems. J Polym Sci Polym Phys Ed 1979;17:477–89. doi:10.1002/pol.1979.180170312.

[29] Tsay CS, Mchugh AJ. Mass transfer modeling of asymmetric membrane formation by phase inversion. J Polym Sci Part B Polym Phys 1990;28:1327–65. doi:10.1002/polb.1990.090280810.

[30] Cheng L-P, Soh YS, Dwan A-H, Gryte CC. An improved model for mass transfer during the formation of polymeric membranes by the Immersion-Precipitation process. J Polym Sci Part B Polym Phys 1994;32:1413–25. doi:10.1002/polb.1994.090320813.

[31] Hansen CM. Hansen Solubility Parameter- User Handbook. vol. 1. 2015. doi:10.1017/CBO9781107415324.004.

[32] Archer WL, a DCUS. Hansen Solubility Parameters for Selected Cellulose Ether 1992;8:599–616.

[33] Patsis A V., Henriques EH, Frisch HL. Interdiffusion in complex polymer systems used in the formation of microporous coatings. J Polym Sci Part B Polym Phys 1990;28:2681–9. doi:10.1002/polb.1990.090281314.

[34] Wilke CR, Chang PIN. Correlation of Diffusion Coefficients in Dilute Solutions. AICHE J 1955;1:264–70.

[35] Hao JH, Wang S. Influence of quench medium on the structure and gas permeation properties of cellulose acetate membranes. J Appl Polym Sci 1998;68:1269–76. doi:10.1002/(SICI)1097-4628(19980523)68:8<1269::AID-APP8>3.0.CO;2-B.

[36] Smolders CA, Reuvers AJ, Boom RM, Wienk IM. Microstructures in phase-inversion membranes. Part 1. Formation of macrovoids. J Memb Sci 1992;73:259–75. doi:10.1016/0376-7388(92)80134-6.

[37] Tsai F-J, Torkelson JM. Roles of Phase Seperation Mechanism and Coarsening in the Formation of Poly(methyl methacrylate) a symmetric Membranes. Macromolecules

182

1990;23:775–84.

[38] Peng N, Chung TS, Wang KY. Macrovoid evolution and critical factors to form macrovoid- free hollow fiber membranes. J Memb Sci 2008;318:363–72. doi:10.1016/j.memsci.2008.02.063.

[39] Yin C, Dong J, Li Z, Zhang Z, Zhang Q. Ternary phase diagram and fiber morphology for nonsolvent/DMAc/polyamic acid systems. Polym Bull 2015;72:1039–54. doi:10.1007/s00289-015-1324-5.

[40] Mazinani S, Darvishmanesh S, Ehsanzadeh A, Van der Bruggen B. Phase separation analysis of Extem/solvent/non-solvent systems and relation with membrane morphology. J Memb Sci 2017;526:301–14. doi:http://dx.doi.org/10.1016/j.memsci.2016.12.031.

183

Chapter 7: Summary and Future Directions

This chapter summarizes the previous chapters to provide major highlights to the reader in each chapter. In addition, some chapters also include future direction, which contains my thoughts regarding continuation of this research.

7.1 Chapter 1

In chapter 1, we laid out our motivation for the dissertation, which is exploring the structural and chemical interaction of components in the natural materials that inspired us to mimic the morphology of the natural materials towards developing new class of light-weight yet strong materials. The overreaching goal of the dissertation was to utilize one of the most abundant biopolymer available on earth, cellulose and its derivatives, towards developing functional light- weight materials that can replace or aid materials from synthetic polymers. The broad scope of the dissertation was narrowed down to exploring 3 avenues: First, investigate component interaction in lignocellulose biomass, Second, mimic cellular structure in biomass to develop strong aerogels and lastly, develop porous fibers that mimic structure of cancellous bone and explain the structure using concept of phase separation.

In this chapter, we also give a background of the lignocellulose biomass, cellulose and cellulose derivatives that are used to synthesize aerogels and porous fibers. The underlying fundamental of synthesizing aerogels and porous fibers are also explained. We hope, the reader gets the full overview of these materials and is guided to the relevant literature for further exploration.

7.2 Chapter 2

In chapter 2, we studied the chemical and structural interaction of cellulose, lignin and hemicellulose within the residual biomass of coir and empty fruit bunches (EFB fibers) while isolating lignocellulosic micro-and nano-fibrils (MNFC) from coir and EFB by using low severity 184

treatments, which allowed to retain most of the native lignin (31 and 24 wt%, respectively). The fiber processing steps included autohydrolysis, mechanical defibrillation, microemulsion treatment and microfluidization. The fibers were characterized during each treatment step for their chemical, structural and thermal properties. The FTIR analysis confirmed no significant changes in the chemical groups of the fiber components. The electron microscopy and AFM images indicated two structural populations in coir and EFB- MNFC with fibril diameter of 20-70 nm and

1- 3 µm. The confocal images highlighted the lignin redistribution on the fiber walls and within the hollow fiber in form of particles. As expected, the crystallinity of the fibers was not affected by the treatments applied, due to presence of substantial amounts of amorphous lignin. However, the thermal degradation behavior was largely affected by the chemical composition and the redistribution of lignin during autohydrolysis and microemulsion treatments. These results, along with the features observed in the morphological analyses, can be combined to postulate the possible structural and component changes that occurred upon defibrillation.

7.2.1 Future Directions

We terminated the study after exploring the interactions between lignin, cellulose, and hemicellulose, which resulted in synthesis of MNFC. We expect that the detailed structural and thermo-chemical analyses presented here facilitate further interest for preparation of new materials from MNFC isolated from coir and EFB, two abundant bioresources that are most suitable for their valorization. There are few approaches one could take. Rojo et al.1, elucidated the effect of residual lignin on the physical, mechanical and interfacial properties of ligno-cellulose nano paper.

They found significant changes in the interfacial and physical properties manifested by changes in water contact angle from 35° to 78° on increasing lignin content. On similar lines, the micro-nano paper from the MNFC may exhibit interesting physical and interfacial properties due to

185

redistribution of residual lignin in MNFC. Alternatively, the two structural populations of MNFC can be separated and investigated towards further use in composite materials as reinforcing component.

7.3 Chapter 3

In this chapter, we explored the synthesis procedure for aerogels using freeze-drying and critical point drying. While exploring a way to make light-weight single component aerogels from polymers, we developed a versatile method to synthesize aerogel from a non-aqueous polymer and demonstrated the idea with three cellulose esters, cellulose diacetate, cellulose acetate propionate and cellulose acetate butyrate. A unique combination of chemical crosslinking, solvent exchange and freeze-drying process was used to produce ultralight (4.3 mg/ml) and highly porous (99.7 %) cellulose diacetate aerogels for the first time. A low density cellulose ester aerogel (< 30 mg/ml) was easily synthesized even with a relatively high starting concentration of the polymer (4 w/v%).

The honeycomb morphology achieved during ice-templating provided the 4% CDA aerogels with high compressive strength (up to 350 kPa) and maximum strain of about 92%.

Additionally, the critical point drying method using supercritical CO2 was also investigated using CDA organogels. Even though such aerogels have relatively high density, 4 times high compared to the aerogels obtained from freeze-drying, the critical point dried aerogels demonstrated mesoporous structure with BET surface area of 263 m2/g, which is approximately

300 times higher than the freeze-dried aerogel of CDA.

7.3.1 Future directions

While ice-templating, the inherent network structure obtained through cellulose ester cross-linking is lost, since the nucleation and growth of ice crystals during freezing compress the polymer into thin walls. To preserve the cross-linked network structure one could perform freeze-

186

drying of gel with organic solvent (acetone). This is challenging because the triple point of acetone is extremely low to achieve (-94.5 °C) using most available equipment (see Chapter 3 for details).2

An alternative approach is the use of organic solvent that have higher triple point such as t-butanol

(25 °C), acetic acid (16.6 °C) or benzene (5.5 °C).2 There have been some articles investigating the effect of t-butanol in cellulose nanofiber aerogel.3,4 However, the approach still remains vastly unexplored from the perspective of aerogels from polymers. In one of our preliminary experiments, we performed solvent exchange of CA organogel (gel in acetone) with t-butanol, followed by freezing in freezer (-4 °C) and freeze-drying. The images of the organogel and the corresponding aerogel along with its SEM image is shown in Figure 7.1. A high degree of shrinkage was observed in solvent exchange step. The RED of CA with respect to t-butanol is 3.5. So, the high shrinkage was not unexpected. Moreover, the SEM image of an open porous structure comprising of small aggregated fibrils alludes to a phase separated structure obtained during the solvent exchange with t-butanol. A macroscopic pore structure was not evident with these aerogels.

Additionally, the resulting aerogel was highly brittle too. The challenge pertains to match the solubility parameter of the solvent of interest (for example t-butanol) with the desired polymer to prevent gel shrinkage during solvent exchange. The ideal approach will be to perform the cross- linking reaction using t-butanol as the solvent.

187

Figure 7.1: a) Camera image of organogel of cellulose diacetate in acetone (left) and the corresponding aerogel (right) formed after solvent exchange with t-butanol followed by freeze drying, b) SEM image of the aerogel.

7.4 Chapter 4

An innovative approach to synthesize ultralight, anisotropic aerogels with tailored mechanical properties is demonstrated in this chapter. The use of the solubility parameter theory to control swelling behavior of the gels allowed a remarkable control over the final mechanical performance of the aerogels. The solvent exchange approach allowed us to synthesize aerogels with densities as low as 0.025 g/cm3 with no need for optimizing the precursor polymer concentration. The synthesized aerogels exhibited a wide range of stiffness (14-340 kPa), toughness (4-103 kPa), strength (22-373 kPa) and compressibility (35-87 %). Additionally, a unidirectional and controlled freezing approach (ice-templating) introduced anisotropy in the aerogels inducing both elastic and plastic deformations, depending on the loading direction. The

188

morphological and mechanical properties of the aerogels can be further tuned by arresting the gels in a non-equilibrium state during solvent exchange. By using this approach, aerogels can be achieved with tunable stiffness (13 to 160 kPa), toughness (4 to 35 kPa), strength (22 to 151 kPa) and compressibility (65 to 88 %).

7.4.1 Future Directions

Except for looking into anisotropy in mechanical properties, the possibilities of unique pore directionality of these aerogels are not explored. Regarding exploring applications for these aerogels, with impressive and tunable mechanical properties along with anisotropy, the aerogels can potentially be used as shock absorbers, in thermal and acoustic insulator, among many others.

Apart from these, the pore structure can be utilized for templating as demonstrated by Shopsowitz et al.5, for CNC aerogels. In other approach, the toughness and strength of such materials can be further enhanced by filling the pores with a viscoelastic solid. Such solid can be incorporated into the aerogel pores in fluid form followed by in-situ UV cross-linking.6,7

7.5 Chapter 5

This chapter explored one potential application for the cellulose diacetate (CDA or CA) aerogels for selective oil uptake. The CA aerogels were modified by chemical vapor deposition to render them hydrophobic and oleophilic. The modified aerogels demonstrated fast oil uptake and good oil retention. Even in the agitated oil in water media, the modified aerogels sorbed low viscosity kerosene oil within 45 s and the uptake of high viscosity oil (170 cP) was well within 2 min. Moreover, the modified aerogels uphold their mechanical integrity in turbulent media, thus preventing any secondary contamination. These properties make them suitable for potential use to clean up oil spills in the rough marine environments.

189

7.6 Chapter 6

In this chapter, we moved away from aerogels and focused on porous fibers development during wet-spinning. A model was developed to predict the fiber morphology and size that used two design parameters (RED and T), that allowed us to identify a set of solvent and non-solvent to use for the polymer of interest during wet-spinning process. Additionally, it was possible to qualitatively predict the ensuing fiber morphology and diameter. An experimental verification of the model was conducted using cellulose acetate and polystyrene. Based on RED, we chose DMAc

& DMSO as solvents and water & methanol as non-solvents. Using the model, we predicted that low RED will result in mesporous fiber due to delayed demixing, and high RED will produce less porous fibers, with a possibility of an outer non-porous “skin” layer formation due to instantaneous demixing. The kinetic parameter T predominantly governs the fiber diameter. A small value of T implies fast solvent diffusion that leads to decrease in fiber diameter. This model also shows good agreement with published literature indicating that the versatile model can be applied to any other polymer-solvent-non-solvent system if the Hansen solubility parameter is identified. Furthermore,

190

due to the simplicity of the model, its applications are not limited to fibers, but can be extended to understand and predict the formation of particles and films formed by non-solvent precipitation process. We envisage that this fundamental work will enhance the knowledge regarding the porous fiber production in the industry.

7.6.1 Future Directions

This is a preliminary study with a bold claim of combining solubility parameter with the thermodynamics of the ternary system. Even though it provides a criterion for predicting the fiber morphology during wet-spinning using Hansen solubility parameters and kinetic parameter, there are still some questions that need to be answered. First, the solubility parameter approach does not account the interaction of solvent and non-solvent. So, the effect of change in the solvent-non- solvent pair on the fiber morphology is still not fully explained with this approach. Second, the two-parameter model does not account for the change in viscosity of the polymer solution or the polymer concentration as such. The future work for this study entails probing these questions.

191

7.7 References

[1] Rojo E, Peresin MS, Sampson WW, Hoeger IC, Vartiainen J, Laine J, et al. Comprehensive elucidation of the effect of residual lignin on the physical, barrier, mechanical and surface properties of nanocellulose films. Green Chem 2015;17:1853–66.

[2] Wilhoit RC, Chao J, Hall KR. Thermodynamic Properties of Key Organic Compounds in the Carbon Range C1 to C4. Part 1. Properties of Condensed Phases. J Phsical Chem 1985;14.

[3] Jiang F, Hsieh Y. Assembling and Redispersibility of Rice Straw Nanocellulose : E ff ect of tert -Butanol. ACS Appl Mater Interfaces 2014;6:20075–84. doi:10.1021/am505626a.

[4] Sehaqui H, Zhou Q, Berglund LA. High-porosity aerogels of high specific surface area prepared from nanofibrillated cellulose (NFC). Compos Sci Technol 2011;71:1593–9. doi:10.1016/j.compscitech.2011.07.003.

[5] Shopsowitz KE, Stahl A, Hamad WY, MacLachlan MJ. Hard templating of nanocrystalline titanium dioxide with chiral nematic ordering. Angew Chemie - Int Ed 2012;51:6886–90. doi:10.1002/anie.201201113.

[6] Chiou B, English RJ, Khan SA. UV Cross-Linking of Thiol-ene Polymers : A Rheological Study. ACS Symp. Ser., 1997, p. 150–66.

[7] Chiou B Sen, Raghavan SR, Khan SA. Effect of colloidal fillers on the cross-linking of a UV-curable polymer: Gel point rheology and the Winter-Chambon criterion. Macromolecules 2001;34:4526–33. doi:10.1021/ma010281a.

192

APPENDICES

193

APPENDIX A2: Supplementary Information for Chapter 2

3329 2910 1736 1638 1240 1027 Cellulose

MNFC-EFB

MNFC-Coir Transmittance(%)

4000 3500 3000 2500 2000 1500 1000 Wavenumber (cm-1)

Figure A2.1: FTIR spectra of pure cellulose with MNFC-Coir and MNFC-EFB

194

Table A2.1: Transmission peaks observed for coir and EFB samples along with the identified functional groups, chemical compounds and fiber component consisting of the identified functional groups. (C = Cellulose, H = Hemicelluloses, L = Lignin).

Wavenumber Functional groups Organic Compound(s) containing Fiber (cm-1) functional groups component 3329 O-H stretch Acid and alcohol C, H, L 2910 sp3 C-H stretch Alkyl, aliphatic, aromatic C, H, L 1736-1715 C=O stretch Ketone, ester and carboxylic H 1640 O-H bend Adsorbed water C 1600-1460 C=C stretch of aromatic Aromatic rings L

1470-1430 O-CH3 Methoxy (O-CH3) on aromatic ring L 1440-1400 O-H bending Acid C, H, L 1400-1300 Sp3 C-H bend Saturated alkanes C, H, L 1240 Acyl C-O (C-O-C stretching) Aryl-alkyl ether linkage C, L 1161, 1100 C-O-C stretching vibration Pyranose ring skeletal C, H 1027 Alkoxy C-O stretch and C-O C-OH (ethanol) C, H, L deformation 898 C-O-C stretching vibration Glycosidic linkage between glucose C, H units 700-900 Aromatic sp2 bend Aromatic hydrogen L

195

Figure A2.2: The Coir and EFB fibers showing the vessel element.

Figure A2.3: Cross-section of the untreated or pristine coir and EFB fibers showing the vessel element in solid circles

196

APPENDIX A3: Supplemental Information for Chapter 3

Figure A3.2: SEM image of the juction point at one pore in the aerogel. The wall thickness is ~1 μm.

197

198

Figure A3.4: Liquid uptake w.r.t the available pore volume per gram of aerogel for 2, 4, 6 and 8 % aerogels. A ratio greater than 1 for water uptake by 4% and above aerogel indicates a combination of absorption in the pores and adsorption on the polymer walls may be due to hydrogen bonding of water with polymer chains. Hence the liquid uptake is referred to as sorption. The low value of sorption for 2% aerogel relative to the available pore volume is may be due to inability of 2% aerogel to retain such high volume of water. Relative volumetric uptake for kerosene oil is less than or equal to 1, which further implies entrapment of air bubbles during sorption.

199

ρ Sorption S. Preparation Conc. P Material (mg g/g Drawbacks No method wt % % R Ref. /cc) . W NP-S

Freeze drying of Cellulose cellulose Low sorption, high 40- 94- 5- 1. fiber suspension NA 10-25 3 density compared 1 80 98 25 (patent) followed by to other aerogels, CVD of silane mechanical TEMPO Cellulose integrity in oxidation of nanofibri 99. practical scenario is CNF l from 0.1 - 1.7- 9- 21 questionable, 2. followed by 375 6 2 rice 0.6 8.1 99. 0 reusability is freeze drying straw 5 demonstrated by and CVD of (CNF) distilling toluene silane (not practical) Freeze Complicated drying NFC procedure for NFC 99. dispersion in production, Nanocell 4- 8- N 30- 3. water 0.5-2 3 mechanical 3 ulose 14 99. A 40 followed by integrity in 5 CVD of practical situation silane is questionable Freeze Nanocell N drying NFC ulose fro ot Not tested for hydrogels 98. NA 4. m birch 1.3 20 NA te sorption 4 followed by 6 kraft st applications CVD of pulp ed silane Freeze Nanocell drying of Reusability is ulose fro nanocellulose demonstrated by m hydrogel 20- N 5. 2 98 30 10 evaporating 5 hardwoo followed by 30 A Toluene (not d kraft TiO2 Atomic practical) pulp layer deposition Cellulose N Very slow fiber 99. Recycled ot separation of oil dispersion 0.25- 9- 4- N 45- 6. cellulose te and water, 6 freeze dried 1 42 97. A 62 fibers st reusability not and CVD 2 ed tested with silane

200

Table A3.1 (continued)

Long and Hydrazine 99 complicated Cellulose 5.6- cross-linking .6- synthesis 7. nanocrys 0.5-2 21.7 72 160 20 7 followed by 98. procedure, not tal freeze drying 6 studied for oil separation Kymene N Cellulos 99. crosslinking ot e 10- 9- Not studied for oil 8. followed by 0.6-1 98 NA te 8 nanofibri 20 99. separation freeze st l 8 drying ed Very expensive Liquid phase material, long silanation Bacterial 6.77 99 N synthesis 9. followed by NA 185 10 9 cellulose .6 A procedure, practical freeze application not drying demonstrated Bacterial cellulose Material loss from Carbon aerogel via >9 N 106- pyrolysis, High 10. nanofiber NA 4-6 5 10 freeze drying 9 A 312 temperature

followed by treatment pyrolysis Hydrotherma l treatment of N Graphen Material loss, Graphene ot e >9 NA 200- expensive material, 11. oxide, freeze .035 2.1 te 11 Framewo 9 600 high temperature dried st rk treatment followed by ed pyrolysis Lyophilizati on of Carbon nanotube and Graphen graphene Expensive raw e and 0.1 >9 N 12. oxide NA 580 10 material, practical 12 Carbon 6 9 A followed by use is questionable nanotube hydrazine vapor reduction Low sorption, TEOS weak mechanical polycondensa integrity, not tion in 96. 5.8 suitable for Chitosan presence of 7- N 10- 13. NA - 10 practical use, 13 -Silica chitosan 90. A 25 17.3 reusability followed by 9 demonstrated by lyophilizatio evaporation of n solvent, unpractical

201

Table A3.1 (continued)

Polycondens ation of N Low sorption, methyltrimet ot weak mechanical N NA 14. Silica hoxysilane NA NA 15 te integrity, not 14 A followed by st suitable for supercritical ed practical use drying Organic gel formation by N High density and isocyanate 250 ot Cellulose 82- low porosity, not 15. crosslinking 5-10 - NA NA te 15 acetate 42 tested for sorption followed by 750 st properties supercritical ed drying Sol-gel route to crosslink via ester linkage followed by solvent 99. 4.0- This Cellulose exchange 7- 10- 7- 16. 2-8 110. 5-112 wor Acetate with water to 91. 92 10 2 k give 2 hydrogel which is freeze dried and CVD of silane

202

Table A3.2: Material properties of the aerogels synthesized in current study and prior studies

Yiel Maximu Maximu Compressi Specific Highe Densi d m m stress on Modulus st Referen ty strai compressi at 60% Modulus (MPa/g/ strain ce (g/ml) n ve stress strain (kPa) ml) (%) (%) (kPa) (kPa) 0.023 This 4% aerogel 38 1.6 NA 92 3501 33.5 4 work 0.008 CNF aerogel 54.5 6.7 19 80 2 15 2 1 0.021 CNC aerogel 40 1.8 NA 80 71 10 7 7 0.005 Graphene 2.3 0.4 NA 95 101 1.5 16 8 Carbon 0.006 10 1.7 NA 90 641 6 10 nanofiber 1 The materials showed no yield stress. Elastic region was followed by densification region. The values represent compressive stress, i.e., the stress at maximum strain

203

References:

[1] Duong HM et al. A polysaccharide aerogel. WO2014/178797 2014. doi:WO2014/178797.

[2] Jiang F, Hsieh Y-L. Amphiphilic superabsorbent cellulose nanofibril aerogels. J Mater Chem A 2014;2:6337–42. doi:10.1039/c4ta00743c.

[3] Cervin NT, Aulin C, Larsson PT, Wågberg L. Ultra porous nanocellulose aerogels as separation medium for mixtures of oil/water liquids. Cellulose 2012;19:401–10. doi:10.1007/s10570-011-9629-5.

[4] Jin H, Kettunen M, Laiho A, Pynnönen H, Paltakari J, Marmur A, et al. Superhydrophobic and superoleophobic nanocellulose aerogel membranes as bioinspired cargo carriers on water and oil. Langmuir 2011;27:1930–4. doi:10.1021/la103877r.

[5] Korhonen JT, Kettunen M, Ras RH a, Ikkala O. Hydrophobic nanocellulose aerogels as floating, sustainable, reusable, and recyclable oil absorbents. ACS Appl Mater Interfaces 2011;3:1813–6. doi:10.1021/am200475b.

[6] Feng J, Nguyen ST, Fan Z, Duong HM. Advanced Fabrication and Oil Absorption Properties of Super-hydrophobic Recycled Cellulose Aerogels. Chem Eng J 2015;270:168– 75.

[7] Yang X, Cranston ED. Chemically Cross-Linked Cellulose Nanocrystal Aerogels with Shape Recovery and Superabsorbent Properties. Chem Mater 2014;35:6016–25. doi:10.1021/cm502873c.

[8] Zhang W, Zhang Y, Lu C, Deng Y. Aerogels from crosslinked cellulose nano/micro-fibrils and their fast shape recovery property in water. J Mater Chem 2012;22:11642. doi:10.1039/c2jm30688c.

[9] Sai H, Fu R, Xing L, Xiang J, Li Z, Li F, et al. Surface Modification of Bacterial Cellulose Aerogels’ Web-like Skeleton for Oil/Water Separation. ACS Appl Mater Interfaces 2015;7:7373–81. doi:10.1021/acsami.5b00846.

[10] Wu Z-Y, Li C, Liang H-W, Chen J-F, Yu S-H. Ultralight, flexible, and fire-resistant carbon nanofiber aerogels from bacterial cellulose. Angew Chem Int Ed Engl 2013;52:2925–9. doi:10.1002/anie.201209676.

[11] Zhao Y, Hu C, Hu Y, Cheng H, Shi G, Qu L. A versatile, ultralight, nitrogen-doped graphene framework. Angew Chemie - Int Ed 2012;51:11371–5. doi:10.1002/anie.201206554.

[12] Sun H, Xu Z, Gao C. Multifunctional, ultra-flyweight, synergistically assembled carbon aerogels. Adv Mater 2013;25:2554–60. doi:10.1002/adma.201204576.

[13] Ma Q, Liu Y, Dong Z, Wang J, Hou X. Hydrophobic and nanoporous chitosan-silica composite aerogels for oil absorption. J Appl Polym Sci 2015;132:41770.

204

doi:10.1002/app.41770.

[14] Venkateswara Rao A, Hegde ND, Hirashima H. Absorption and desorption of organic liquids in elastic superhydrophobic silica aerogels. J Colloid Interface Sci 2007;305:124– 32. doi:10.1016/j.jcis.2006.09.025.

[15] Fischer F, Rigacci A, Pirard R, Berthon-Fabry S, Achard P. Cellulose-based aerogels. Polymer (Guildf) 2006;47:7636–45. doi:10.1016/j.polymer.2006.09.004.

[16] Xu X, Li H, Zhang Q, Hu H, Zhao Z, Li J, et al. 3D Graphene / Iron Oxide Aerogel Elastomer Deformable in a Magnetic Field. ACS Nano 2015;9:3969–77.

205

APPENDIX A4: Supplemental Information for Chapter 4

206

207

Figure A4.3: SEM micrographs of the out-of-plane axis of aerogels when the acetone volume fraction in solvent exchange is varied. Inset shows the camera image of the bulk aerogels

208

209

Figure A4.5: SEM micrographs of the out-of-plane axis of aerogels when solvent exchange time is varied. Inset shows camera image of the bulk aerogels.

210

δd δp δH 1/2 1/2 1/2 MPa MPa MPa Cellulose Acetate 18.6 12.7 11.0 Acetone 15.5 10.4 7 Soluble Dimethyl Acetamide (DMAc) 16.8 11.5 10.2 Soluble Dimethyl Sulfoxide (DMSO) 18.4 16.4 10.2 Soluble Tetrahydrofuran (THF) 16.8 5.7 8 Soluble Dimethyl Formamide (DMF) 17.4 13.7 11.3 Soluble

Toluene 18.0 1.4 2 Insoluble Hexane 14.9 0 0 Insoluble Ethanol 15.8 8.8 19.4 Insoluble Water 15.6 16 42.3 Insoluble

211

APPENDIX A5: Supplemental Information for Chapter 5

Movie A5.1: Demonstration of separation of spent kerosene oil obtained from cleaning the pump.

Movie A5.2: Demonstration of separation of spent kerosene oil from oil in water stirred emulsion system.

Movie A5.3: Demonstration of separation of spent motor oil (obtained from the oil change) from water in a stirred system to model oil spill in a turbulent ocean

(Note: The Movies are hyperlinked to the supporting information of the corresponding research article)

Figure A5.1: a) The experimental set-up to test the modified aerogel performance for crude oil-spill in sea water, b) water and oil uptake from the mixture of crude oil and salt water.

212

Table A5.1: Properties of the solvents used for sorption testing

surface tension @ 20 °C Viscosity density Solvent (mN/m) cP g/cm3

Diethyl ether 16.7 0.22 0.708 n-hexane 18.4 0.3 0.655 1-propanol 20.9 1.95 0.802 Ethanol 22.1 1.07 0.787 Cyclohexane 24.7 0.89 0.773 Toluene 27.9 0.56 0.865 1.2-dichlorobenzene 35.7 1.32 1.301 Methyl Iodide 50.8 2.6 3.306 water 72.8 0.89 1

213

Explanation for the negative value of liquid sorbed in wicking setup

A negative value of the liquid sorbed is explained as follows, with reference to the graphical representation of the wicking setup shown in Figure 5b. As the aerogel sample is lowered towards the surface of the liquid, the instrument records mass gain (Mg) as the function of distance and time. Once a certain fraction of the aerogel sample is immersed in the solvent, a buoyant force is exerted in the upward direction which is not negligible in case of the aerogel samples.

For hydrophobic aerogel, F = Mabg – γAcosθ – Fb

For hydrophilic aerogel, F = Mabg + γAcosθ – Fb

Where, F = Mass gain recorded by the tensiometer Mab = Mass of the liquid sorbed the aerogel γ = Surface tension of the aerogel w.r.t liquid A = Surface area of the liquid in contact with the aerogel Fb = Buoyant force

The buoyant force and surface tension both are acting in upward direction for hydrophobic aerogel while there is very little solvent sorption (for water and diiodomethane) and hence, the gravitational force Mabg acting downwards is very small. Hence the tensiometer records negative mass gain.

214

APPENDIX A6: Supplementary Information for Chapter 6

Assumptions made to develop diffusion kinetic model

The following assumptions were made to develop the diffusion kinetic model:

1. Radial diffusion components of solvent and non-solvent are considered. As there is no external

influence on the non-solvent layer (at r→∞) to generate radial velocity component, the radial

convection process is neglected.

2. The diffusion and convection components of the other directions are neglected.

3. In the limit of experimental time, t → 0+ the skin-thickness (W) is small enough to assume linear

concentration profiles of solvent and non-solvent across the skin.

4. In the limit of experimental time, , the curvature effect of the skin layer in neglected. i.e.,

R is larger than W.

5. The concentration of non-solvent at the skin-solution interface is assumed to be negligible.

6. The concentration of solvent at the skin-solution interface is assumed to be constant, C0.

7. Initial skin formation results almost exclusively from diffusion of non-solvent into the system.

8. A quasi-steady diffusion state is established almost instantaneously.

9. The effect of temperature is negligible.

10. Diffusivities at infinite dilution are used.

215

Figure A6.1: Normalized solvent concentration at the precipitation interface as a function of time. The graph is plotted from eq. 20.

216

217

Table A6.1: Hansen solubility parameters for various polymers and solvents studied elsewhere.[1,2]

δd δp δH (MPa1/2) (MPa1/2) (MPa1/2) Polyamic acid (PAA) 15.4 14.8 13 Extrem 17.01 3.66 8.27 Polymers Poly(vinylidene fluoride) 17.2 12.5 9.2 (PVDF) NMP 17.9 12.3 7.2 DMAC 16.8 11.5 10.2 DMF 17.4 13.7 11.3 DMSO Solvents 18.4 16.4 10.2 Water 15.6 16 42.3 Glycerol 17.3 12.1 29.3 Methanol 15.2 12.3 22.3 Ethanol 15.8 8.8 19.4 Isopropyl alcohol (IPA) 15.8 6.1 16.4 Non- 70% water/30% DMAC 16.0 14.6 32.7 Solvents 50% water/50% DMAC 16.2 13.7 26.2 40% water/60% DMAC 16.3 13.3 23.0 50% water/ 50% Methanol 15.4 14.2 32.3 50% water/ 50% Ethanol 15.7 12.4 30.9 50% water/ 50% IPA 15.7 11.1 29.4

Note: R0 is assumed as 5.2 for PAA 13.2 for Extrem and 8.9 for PVDF. These values are taken for the solvents with highest Ra that dissolves these polymers. Since, our model gives a qualitative prediction based on the magnitude of values, actual values are not necessary.

References

[1] S. Mazinani, S. Darvishmanesh, A. Ehsanzadeh, B. Van der Bruggen, Phase separation analysis of Extem/solvent/non-solvent systems and relation with membrane morphology, J. Memb. Sci. 526 (2017) 301–314. doi:http://dx.doi.org/10.1016/j.memsci.2016.12.031.

[2] C. Yin, J. Dong, Z. Li, Z. Zhang, Q. Zhang, Ternary phase diagram and fiber morphology for nonsolvent/DMAc/polyamic acid systems, Polym. Bull. 72 (2015) 1039–1054. doi:10.1007/s00289-015-1324-5.

218