Polyhipes MORPHOLOGY, SURFACE MODIFICATION and TRANSPORT

Home , Capillary length

POLYHIPEs MORPHOLOGY, SURFACE MODIFICATION AND TRANSPORT

PROPERTIES

BORAN ZHAO

Submitted in partial fulfillment of the requirements for the degree of

DOCTOR OF PHILOSOPHY

Dissertation Advisers: Dr. Donald L. Feke and Dr. Ica Manas-Zloczower

Department of Chemical and Biomolecular Engineering

CASE WESTERN RESERVE UNIVERSITY

January 2019

CASE WESTERN RESERVE UNIVERSITY

SCHOOL OF GRADUATE STUDIES

We hereby approve the thesis/dissertation of

Boran Zhao

candidate for the degree of Doctor of Philosophy *.

Committee Chair

Donald L. Feke

Committee Members

Ica Manas-Zloczower

Jay Adin Mann Jr.

Uziel Landau

David Schiraldi

Date of Defense

September 18th, 2018

*We also certify that written approval has been obtained

for any proprietary material contained therein.

DEDICATION

To faith in truth, who carries me in seeking answers;

To faith in consciousness, who drives me in the right directions;

To my beloved family, who are the source of my happiness;

To myself, for not yielding in pressure and not being satisfied in absorbing knowledge.

TABLE OF CONTENTS

Dedication ...... i

Table of Contents ...... ii

List of Tables ...... viii

List of Figures ...... ix

Acknowledgments...... xxi

Abstract ...... 1

Chapter 1 Introduction ...... 3

1.1 The Significance of Porous Materials ...... 3

1.2 Porous Materials Classification and Processing Methods ...... 4

1.2.1 Metal porous materials ...... 4

1.2.2 Porous ceramics ...... 5

1.2.3 Polymeric porous materials...... 5

1.3 Dissertation Outline ...... 9

Chapter 2 Emulsion Preparation and the Corresponding Effects on

Poly(HIPE) Foam Morphological and Mechanical Properties ...... 11

2.1 Synopsis ...... 11

2.2 Introduction ...... 11

2.2.1 Droplet breakup mechanism ...... 13

2.2.2 Emulsification methods ...... 17

2.2.3 General emulsion stability ...... 18

2.2.4 Causes for emulsion instability ...... 18 ii

2.2.5 High internal phase emulsions ...... 20

2.2.6 Curing of high internal phase emulsions ...... 24

2.2.7 Interfacial tension kinetics ...... 26

2.3 Experimental ...... 29

2.3.1 Materials ...... 29

2.3.2 Mixing procedures: tuning HIPE morphology ...... 29

2.3.3 Interfacial tension kinetics using Wilhelmy Plate method ...... 32

2.3.4 HIPE and polyHIPE morphology analysis ...... 33

2.3.5 Chemorheology study ...... 34

2.3.6 Mechanical properties ...... 35

2.3.7 Thermal stability analysis ...... 35

2.4 Results and Discussion ...... 35

2.4.1 The effect of propeller mixer setup on emulsion morphology

...... 35

2.4.2 The effect of syringe mixer setup on emulsion morphology ...... 41

2.4.3 Emulsion morphology under centrifugal force ...... 50

2.4.4 Interfacial tension (IFT) dynamics ...... 58

2.5 Conclusions ...... 62

Chapter 3 Morphological and Surface Effects on PolyHIPE Foam Transport

Properties: Permeability and Spontaneous Imbibition ...... 63

3.1 Synopsis ...... 63

3.2 Introduction ...... 64

iii

3.2.1 Flow in porous media...... 64

3.2.2 Substrate roughness effect ...... 67

3.2.3 Roughness effect on micro channels ...... 68

3.2.4 Wetting ...... 69

3.2.5 Fluid transport in polyHIPE foams ...... 73

3.3 Experimental ...... 75

3.3.1 Materials ...... 75

3.3.2 Emulsion preparation ...... 76

3.3.3 Foam morphology ...... 76

3.3.4 Relative foam density ...... 79

3.3.5 Foam mechanical properties ...... 79

3.3.6 Permeability measurement ...... 80

3.3.7 Spontaneous imbibition measurement ...... 82

3.3.8 Surfactant coating ...... 83

3.4 Results and Discussion ...... 84

3.4.1 Variation of foam morphology ...... 84

3.4.2 Foams mechanical properties ...... 85

3.4.3 Darcy’s flow through polyHIPE foams ...... 88

3.4.4 Spontaneous imbibition in polyHIPEs ...... 90

3.4.5 Silicon oil imbibition ...... 92

3.4.6 Partial wetting-water imbibition in surfactant coated foams ...... 97

3.5 Conclusions ...... 107

Chapter 4 Surface Modification of Polyhipe Foams ...... 108

4.1 Synopsis ...... 108

4.2 Introduction ...... 108

4.3 Experimental ...... 110

4.3.1 Foam preparation ...... 110

4.3.2 Cellulose nanocrystals (CNC) preparation ...... 110

4.3.3 CNC dispersion preparation ...... 111

4.3.4 Graphene oxide (GO) preparation ...... 111

4.3.5 Coating process ...... 112

4.3.5.1PolyHIPE foam coated with CNC ...... 112

4.3.5.2PolyHIPE foam coated with GO and achieving

electrical conductivity ...... 112

4.4 Results and Discussion ...... 113

4.4.1 CNC-coated polyHIPE nanocomposite ...... 113

4.4.2 Conductive polyHIPE via GO coating ...... 120

4.5 Conclusions ...... 133

Chapter 5 PolyHIPE Foams as Oil Sorbent: Volume Expansion, W/O

Emulsion Selectivity and Volumetric Capacity ...... 135

5.1 Synopsis ...... 135

5.2 Introduction ...... 136

5.3 Experimental ...... 139

5.3.1 Materials ...... 139

5.3.2 Foam preparation ...... 139

5.3.3 Solid polymer preparation...... 140

5.3.4 Swelling test and crosslinking density ...... 141

5.3.5 Foam volume reduction-compact ratio ...... 141

5.3.6 Morphology and image analysis ...... 142

5.3.7 Foam mechanical performance ...... 142

5.3.8 Water wettability ...... 143

5.3.9 Oil absorption and reusability ...... 143

5.3.10 Water/oil emulsion selectivity ...... 144

5.4 Theory ...... 144

5.4.1 Phase separation in free radical crosslinking

copolymerization...... 144

5.4.2 Swelling of crosslinked foam after curing ...... 149

5.4.3 Foam swelling ratio and polymer swelling ratio ...... 150

5.5 Results and Discussion ...... 152

5.5.1 CVES fabrication...... 152

5.5.2 Crosslinking density, swelling behavior and foam strut

micromorphology...... 154

5.5.3 Foam mechanical behavior and volume shrinkage

mechanism...... 157

5.5.4 Wettability: contact angle, contact angle hysteresis and

selectivity of oil over water...... 163

5.5.5 Oil absorption and recovery...... 166

5.5.6 Water/oil separation...... 170

5.6 Conclusions ...... 173

Chapter 6 Future Work ...... 174

6.1 Controlled particle distribution via directional foam drying ...... 174

6.1.1 Proposed future works ...... 175

6.2 HIPE Curing Mechanism...... 176

6.2.1 Proposed future work ...... 178

6.3 CNC Crosslinking with Polyethylenimine ...... 178

6.3.1 Proposed future work ...... 178

6.4 Hybrid CNC-rGO Coating for Controlling Conductive Foam

Sensitivity ...... 179

6.4.1 Proposed future work ...... 179

References ...... 180

vii

LIST OF TABLES

Table 2.1. Morphological and physical properties of the polyHIPEs...... 48

Table 2.2. Centrifugal parameters ...... 51

Table 2.3. Oil phase and aqueous phase composition ...... 59

Table 3.1. Surface tension and surface energy of common liquids and solid

surfaces ...... 66

Table 3.2: Processing parameters, morphological properties and mechanical

properties for polyyHIPE foams...... 85

Table 3.3: Foam permeability, characteristic hydrodynamic diameter from the

permeability tests, and window size by SEM analysis...... 89

Table 3.4: Parameters used in the scaling analysis ...... 91

Table 3.5. Estimation of AOT coating configuration ...... 105

Table 3.6. Mateirals properties of foam and AOT ...... 105

Table 5-1 Foams physical properties ...... 163

Table 5-2. Foam surface energy ...... 164

Table 5-3. Materials solubility parameter ...... 168

viii

LIST OF FIGURES

Figure 1-1. Schematics of polyHIPEs preparation process...... 7

Figure 1-2. A collection of polyHIPEs materials with various shapes. The solid

pieces are made from the same monomers as used in polyHIPEs

foams...... 8

Figure 1-3. Micromorphology features of HIPEs and polyHIPEs...... 9

Figure 2-1. Effect of viscosity ratio in simple shear field...... 14

Figure 2-2. Comparison of simple shear flow and elongation shear flow...... 14

Figure 2-3. Typical observations of deformation and burst of drops in plane

hyperbolic flow. Part (a): Water drop (p=2x10-4). After burst the

drop formed unstable pointed ends from which small droplets were

ejected. Part (b): Castor oil (p=1.0). After burst the drop became

extended into a long cylindrical thread which eventually

disintegrated...... 16

Figure 2-4. Illustration of drops in shear flow showing the change in

deformation D, angle of the main axis with increasing shear rate up

to breakup. (a) p=2x10-4, small satellite droplets ejected from the

main droplet. (b) p=1.0 neck formed and three satellite droplets. (c)

drops were drawn out into long cylindrical threads. (d) no breakup

was observed...... 17

Figure 2-5. Droplets in rhomboidal dodecahedron shape[15]...... 21

Figure 2-6. Droplets in tetrakaidehedron shape...... 21

Figure 2-7 Mixing setup (a): emulsion prepared by a propeller mixer; (b)

pristine emulsion pushed through a syringe...... 30

Figure 2-8. Centrifuge setup and schematic illustration of the dimensions ...... 31

Figure 2-9 Wilhelmy plate setup ...... 31

Figure 2-10. Schematic of Wilhelmy plate method...... 32

Figure 2-11. A typical optical microscopy image of HIPE (a) and the same

image after ImageJ processing (b). The software will identify the

droplet automatically after image processing...... 34

Figure 2-12. Mixing time effect in propeller mixing setup on emulsion

morphology. (a) R19M04, (b) R19M10, (C) R19M30...... 37

Figure 2-13. Propeller mixer mixing time effect on HIPE droplet size

distribution...... 37

Figure 2-14. Propeller mixing time effect on HIPE average droplet size and

PDI...... 38

Figure 2-15. Storage modulus VS mixing time...... 39

Figure 2-16. Storage modulus plateau vs mixing time...... 40

Figure 2-17. Linear relationship between emulsion storage modulus and inverse

of droplet size...... 40

Figure 2-18.Optical microscopy images of HIPE with 12wt% surfactant for (a)

N = 0, (b) N = 1, (c) N = 5, and (d) N = 9. The scale bar depicts

100µm...... 42

Figure 2-19. The variation of droplet size and polydispersity for HIPE with

different number of shearing cycles (N)...... 42

Figure 2-20. Storage modulus of emulsions prepared without initiator and

subjected to different number of shearing cycles (N): (a) 25°C and

(b) 70°C...... 43

Figure 2-21. Dynamic modulus versus time for samples processed with

different numbers of shearing cycles and cured at 70°C...... 44

Figure 2-22. SEM images of cured polyHIPE samples prepared from HIPEs

sheared with a different number of shearing cycles: (a) N = 0; (b) N

= 1; (c) N =3; (d) N =5; (e) N =7; and (f) N = 9. The scale bar depicts

10 µm...... 46

Figure 2-23. BET surface area and modulus for polyHIPE samples prepared

from HIPES subjected to different numbers of shearing cycles P0-

P13...... 46

Figure 2-24. Stress-strain curves for polyHIPe samples under compression.

Samples were prepared from HIPEs prepared with different

numbers of shearing cycles...... 47

Figure 2-25 Derivative TGA curves for polyHIPE samples prepared from

HIPEs subjected to different numbers of shearing cycles...... 48

Figure 2-26. Emulsion morphology of R03, R04 and R05...... 52

Figure 2-27. R03 emulsion sedimentation result...... 52

Figure 2-28. R04 emulsion sedimentation results...... 53

Figure 2-29. R05 gravity and centrifugal sedimentation result...... 53

Figure 2-30. R03 emulsion centrifuged 1 min and cured...... 54

Figure 2-31. Centrifugation time effect on final emulsion water/oil ratio...... 55

xii

Figure 2-32. Schematic of centrifugation process and foam sample sectioned

for mechanical test...... 56

Figure 2-33. R09 emulsion centrifuged for 1 min and the mechanical properties

and the foam density change in the centrifugation direction...... 57

Figure 2-34. R9 foam with 1 min centrifuge time from top to bottom of the tube

change in window size...... 57

Figure 2-35. Emulsifier concentration effect on interfacial tension dynamics. DI

water as aqueous phase...... 59

Figure 2-36. Salt effect on interfacial tension dynamics: emulsifier

concentration 0.001wt%...... 60

Figure 2-37 Critical emulsifier concentration on interfacial tension...... 61

Figure 3-1. A small droplet in equilibrium over a horizontal surface: (a) and (b)

correspond to partial wetting, the wetting is stronger in (b). (c)

corresponds to complete wetting[73]...... 71

Figure 3-2. Various types of triple phase line structure [73]...... 71

Figure 3-3. Illustration for foam void size correction...... 78

Figure 3-4. Illustration for foam window size correction...... 79

xiii

Figure 3-5. Schematic illustration of permeability measurement ...... 81

Figure 3-6. Schematic illustration of spontaneous imbibition measurement...... 83

Figure 3-7. Emulsion droplet morphology for various mixing time: (a)

R19M05, (b) R19M10, (c) R19M30 scale bar is 100µm...... 84

Figure 3-8. PolyHIPE void and window morphology...... 85

Figure 3-9. R19M05-R19M30 stress-strain curve...... 86

Figure 3-10. Foam yielding point defined as deflection point in the stress strain

curve...... 87

Figure 3-11. Wet/dry foam modulus...... 87

Figure 3-12. Wet and dry foam yield strength...... 88

Figure 3-13. A typical curve from permeability measurement. Foam sample was

R19M10...... 89

Figure 3-14. Wetting of polymer substrate by silicone oil...... 92

Figure 3-15. Spontaneous imbibition test of R40M05 using three silicone oils

with 10 cst, 100 cst and 1000 cst ...... 93

Figure 3-16. Effective radius from spontaneous imbibition process...... 95

xiv

Figure 3-17: Foam saturation for foams with various porosity and mixing time.

Figure 3-18. Illustration of air pockets formation during imbibition process...... 96

Figure 3-19. Schematic demonstration of the coating procedure ...... 97

Figure 3-20. Coating concentration in dried foam versus initial surfactant in

soaking solutions...... 98

Figure 3-21. Water spontaneous imbibition inside R19M10 foams with various

emulsifier content...... 99

Figure 3-22. Illustration for uneven penetrating front in the foam ...... 99

Figure 3-23. PGS coating concentration effect on water imbibition rate...... 100

Figure 3-24. AOT coating concentration in dry foam against coating solution

concentration...... 102

Figure 3-25. Water imbibition curve on foams coated with AOT...... 103

Figure 3-26. Effect of AOT coating concentration on foams water imbibition

rate Washburn coefficient...... 104

Figure 3-27. Evidence of coated emulsifiers blocking the windows...... 104

Figure 3-28. Close packing of AOT spheres in a 2D projection...... 106 xv

Figure 3-29. molecular structure and proposed configuration of AOT

molecules[111]...... 106

Figure 4-1. Hierarchical structure of wood biomass and the characteristics of

cellulose microfibrils[116]...... 114

Figure 4-2. CNC water dispersion. 3wt%...... 115

Figure 4-3. Freeze dried neat CNC...... 115

Figure 4-4. CNC coating layer on foam struts...... 116

Figure 4-5. CNC coating concentration in dry foams vs concentration in

solution...... 117

Figure 4-6. Foams stress-strain curve as increasing CNC coating concentration.

117

Figure 4-7. Foam compression modulus VS CNC coating content...... 118

Figure 4-8. Foams 1st and 2nd cycle compression modulus after 75% strain...... 118

Figure 4-9. 3wt% CNC coated foams 5 consecutive compression cycles at 10%

and 20% strain endpoint...... 119

Figure 4-10. 5 consecutive compression cycles with 10% and 20% strain

endpoint respectively, 3wt% coated foam...... 119 xvi

Figure 4-11. Water uptake rate Washburn coefficient vs CNC content...... 120

Figure 4-12. GO aqueous dispersion. 1.0, 2.0, 4.0 and 8.0mg/ml ...... 121

Figure 4-13. Foams soaking in GO solution...... 121

Figure 4-14. rGO coated foams. 1.0, 2.0, 4.0, 8.0 mg/ml GO solution...... 122

Figure 4-15. Neat foam strut surface...... 123

Figure 4-16. rGO coating layer. 4.4mg/ml sample...... 124

Figure 4-17. Foams mechanical compression setup coupled with electrical

resistance measurement...... 125

Figure 4-18. Schematic illustration of foam electrical behavior under uniaxial

compression strain...... 125

Figure 4-19. Foams resistance change on the first strain cycle to 20% endpoint.

126

Figure 4-20. Resistance-strain behavior in the following cycles after first strain

at 20%...... 127

Figure 4-21. Resistance-strain behavior when foams were further strained to

50%, first cycle...... 128

Figure 4-22. Resistance-strain behavior of following cycles of 50% strain...... 130 xvii

Figure 4-23. Resistance-strain behavior for 75% strain...... 131

Figure 4-24. Summary of foams’ strain-history influence on resistance-strain

behavior when first strained at various endpoints...... 132

Figure 4-25. Electrical resistance behavior of CNC-GO hybrid foams with

compression strain ...... 133

Figure 5-1 Composition dependence of free energy: (left) unstable and (right)

stable. Local stability is determined by the sign of the second

derivative of free energy with respect to composition[146]...... 146

Figure 5-2 schematic illustration of the relationship between foam and strut

swelling ratio...... 151

Figure 5-3 (a) Schematic illustration of polyHIPE foam preparation; (b) Foam

R19T60 in compact state(left), volume expansion in methanol due

to recovery of collapsed cellular structure(middle) and in heptane

due to further swelling of the strut(right); (c) Swelling ratio and

crosslinking density with toluene content in the oil phase...... 153

Figure 5-4. Inert diluent effect on polyHIPE foams curing. from left to right:

R19T0, R19T20, R19T40, R19T60, R19T80...... 154

Figure 5-5: Foam swelling ratio and solid polymer swelling ratio from only

curing the oil phase...... 156 xviii

Figure 5-6 Void/window morphology (scale bar is 50 µm) and struts

morphology(scale bar 1 µm for a and 2 µm for b-d ) for foams

prepared with dilution concentration of: 0% (A,a), 20% (B,b), 40%

(C,c) and 60% (D,d) ...... 157

Figure 5-7. Foams loading/unloading compression stress-strain curve in the dry

state (a) and in the swollen state (b). The inserts are compression

modulus for 5 consecutive cycles ...... 158

Figure 5-8. Dry foam modulus VS foam relative density behavior without inert

diluent...... 160

Figure 5-9. Hysteresis loss ratio for dry and swollen foams in the loading and

unloading compression cycle...... 161

Figure 5-10 Volume reduction of foam by increasing diluent content. Left:

Dried samples; right: R19T60 compact and expansion in heptane

(dyed in red)...... 162

Figure 5-11. Wetting of polymer solid by PDMS oil ...... 164

Figure 5-12 Water contact angles for R19 foam samples...... 166

Figure 5-13. Gravimetric capacity (a) and volumetric capacity (b) of various

solvent by foams with increasing inert dilution content...... 167

xix

Figure 5-14. Sorption capacity of foams over 10 cycles of saturation with

heptane followed by squeezing, normalized by dry foam weight...... 169

Figure 5-15 Emulsion before and after foam absorption from left to right: raw

emulsion, and emulsion after foam adsorption T0, T20, T40 and

T60...... 171

Figure 5-16. Raw emulsion droplet size distribution. the insert is from 10um to

17um and above...... 171

Figure 5-17. Emulsion droplet size distribution after foam absorption...... 172

Figure 5-18. Emulsion droplets distribution above 10 µm after foam absorption.

172

Figure 6-1. Emulsifier concentration & composition effect on HIPE curing

process...... 177

ACKNOWLEDGMENTS

I would like to acknowledge my advisor Dr. Donald L. Feke for bringing me to CWRU and providing me a friendly, open-minded atmosphere in which I was always encouraged to tackle the challenges in the research. Also, Prof. Feke’s transport course was one of the best learning experiences I ever had.

I am very grateful for my co-advisor Dr. Ica Manas-Zloczower for the same reasons as for

Prof. Feke. She is more than a mentor who guides me through the challenges in research, but one for life who teaches me to follow my instinct and to always push my limit. I am in great debt to Dr. Feke and Dr. Manas.

I would like to acknowledge my colleagues, peer graduate students and friends, Andrew

Wang, Tianqi Liu, Ammar Patel, Xuehui Gong, Kristen Rohm, Dr. Ronald Zeszut, Dr.

Xinyu Liu, Dr. Kai Ke, Dr. Gabriel Gedler, Dr. Vahid Karimkhani, Dr. Liang Yue, and soon to be doctor Vahab Solouki Bonab. They made my time at CWRU joyful. I will cherish our friendship forever.

I would like to thank Procter & Gamble for the support of this work. Specifically, I would like to thank Wade hubbert, Steve Merrigan, Maxwell Wingert for the insightful discussions about the project.

I would like to thank the Chinese Scholar Council for the financial support for my Ph.D.

Thanks to my love Xin Zeng and my lovely daughter Evelyn, your companion and confidence in me helped me through the difficult times. I cannot put in words or sentences of how much I love you.

xxi

Last not but least I would like to thank my parents. Thanks for bringing me to this world.

Your love is the best thing in the world.

xxii

PolyHIPEs Morphology, Surface Modification and Transport Properties

BORAN ZHAO

ABSTRACT

Polymerized High Internal Phase Emulsions (polyHIPEs) are highly porous polymeric foams with a variety of potential applications such as in filtration, as sorbent materials, or for tissue scaffolding. Since these foams are prepared through emulsion templating, important aspects in polyHIPE research and development include (but are not limited to) the control of the porous structure, understanding the influence of the structure on fluid transport properties and the exploration of new applications for polyHIPEs by tuning surface properties.

In this work, three emulsification procedures were adopted for the purpose of controlling the structure of the HIPE template. These procedures include two methods of different combinations of simple shear flow and extensional shear flow as can be found in a propeller-mixer or syringe setups, as well as the application of a centrifugal field. The influence of the mixing configurations on the HIPE was investigated in terms of droplet size/distribution, flow properties, curing process, and structural and mechanical properties of the polyHIPE foams.

Furthermore, two important transport phenomena important to many polyHIPE applications, namely Darcy’s flow and capillary flow, were studied using polyHIPE foams

with controlled morphology and surface properties. By conducting a permeability study and a spontaneous imbibition dynamic study, the interconnecting void (window) size was identified as the dominant structural parameter for both flows. In the spontaneous imbibition study, the window size distribution was found to have a direct impact on fluid saturation. Besides the morphology effect, foam surface wettability was also controlled by coating with surfactant-dioctyl sulfosuccinate (AOT). A critical surfactant coating concentration was observed above which a significant improvement in the water wettability and absorption rate were achieved. Such a critical concentration corresponds to a monolayer surface coverage of the foam interior surface by the surfactant.

In addition to improving surface wettability, the electrical conductivity and mechanical compression modulus of the foam could also be controlled by using a coating protocol with various functional nanomaterials, namely cellulose nanocrystals and graphene oxide. Such a coating process might be a promising method for polyHIPE surface functionalization.

Finally, polyHIPE absorption capacity was greatly improved by creating a material that exhibits a transformation from a compact volume state to an expanded volume state during the absorption process. This compact-expansion transition was realized by introducing an inert diluent in the HIPE continuous phase, which affects the foam crosslinking density thus controlling poly(HIPE) swelling behavior. Such foams show improved volumetric absorbing capacity, rendering them to become a promising sorbent for solvent-spill cleanup applications.

CHAPTER 1

INTRODUCTION

1.1 The Significance of Porous Materials

Porous media are a fundamental part of the material world. Their existence is universal.

Soils are porous such that air, water and nutrition can infiltrate and be absorbed by the plants or animals. Oil reservoirs are made of porous rocks that contain recoverable gas and oil. Bones are porous so that new blood cells produced (by marrow) can be transferred out of the bone. The same substance can form a solid or a porous material. For example, a block of solid rock and a pile of sand are both made of basically the same material-silica, yet they function quite differently. The rock blocks the gas and liquid from penetrating through while sand allows both through. Ancient architecture was often built with heavy solid rocks which could bear a limited overall loading, provided only poor heat insulation and were fairly unstable during earthquakes. Nowadays, porous composite ceramics or metal materials allow engineers to design structures with good energy usage efficiency and improved mechanical stability. Another example involves the use of polymers.

Polyurethanes are used to manufacture high performance O-ring sealing and water repellant coatings while the same material can be used as porous foams or cushions found in cars. Polyvinylidene fluoride is used for wire insulation, chemical resistant pipes as well as in porous membranes for filtration. From these examples, it can be found that by converting a solid substance to a porous one, materials expand their application from structural purpose to more versatile functions. Their function ranges from isolating to 3

exchanging “chemicals”, or it can behave to enhance or retard the transfer of “heat” from one side to the other. In other words, mass transport, momentum transport and energy transport processes like filtration, absorption, damping or thermal insulation processes can be engineered by making a material porous. Thus, the focus of porous materials research, development and engineering is always understanding the relationships between structural properties like mechanical performance and functional properties like permeability, thermal and electrical conductivity.

1.2 Porous Materials Classification and Processing Methods

Porous media can be categorized into two types: natural and artificial. Natural porous media can be found universally, such as above mentioned soils, oil reservoir and bones, as well as lava, pumice, foods, etc. Artificial porous media can be further sub-classified into metal, ceramic, and polymeric types. Each type of artificial porous materials involves a unique fabrication process and has its own pro and cons.

1.2.1 Metal porous materials

In order to lower energy consumption, porous metals are often applied as structural materials in automobile, aviation and construction industries due to the specific strength, stiffness, ductility, and light weight [1, 2]. Moreover, since the metals are good thermal and electrical conductors and have good catalytic properties, porous metals have been applied in heat exchangers, electrodes, filters, and catalysts[1].

Powder metallurgy is a common method to prepare porous metals through metal powder mixing, molding, and sintering. Metal porous materials can be also prepared by methods like fiber sintering, melt-casting, and metal-deposition, vapor deposition and electrodeposition etc.

1.2.2 Porous ceramics

Ceramic materials have great hardness, large specific surface area, low thermal conductivity, and resistance to high-temperature corrosion. Thus they are often applied in filtration and separation, combustion and fire retardant applications, and heat insulation.

Porous ceramics are usually processed by partial sintering, sacrificial fugitives, replica templates, and direct foaming [3, 4].

1.2.3 Polymeric porous materials

Polymeric porous materials have unique properties such as a relatively light weight, a wide range of mechanical properties, and are easy processable. A foam in this context refers to a solid-gas structure. Polymer foams can be classified into various groups.

Morphologically, polymer foams can be classified as open- and closed-cell types. Open- cell foams have interconnected voids while closed-cell foams have voids that are separate from one another. Conveniently based on apparent density, polymer foams can be categorized into high- (>0.4 g/cm3), medium- (0.1-0.4 g/cm3), and low-(<0.1 g/cm3) density types. Mechanically, polymer foams can also be classified as rigid, semi-rigid, and flexible types. Rigid foams consist of materials that have a glass transition temperature (Tg) 5

or crystalline temperature (Tc) that is higher than room temperature (RT) such that they are in a glassy and rigid form at RT. Conversely, flexible polymer foams are made of polymers that have Tg or Tc lower than RT. For example, rigid foams are normally made of polystyrene (PS), polycarbonate (PC), epoxy resin (ER), and phenol formaldehyde resin

(PF). Flexible foams are often made of elastic polyurethane (PU), polyvinyl chloride

(PVC), acrylate-methacrylate copolymers.

The most popular method of processing polymer porous material is foaming by blowing chemical/physical agents first developed in the 1930s. Chemical blowing agents are the ones that produce gas via chemical reaction, for example, baking powder and isocyanates.

Physical blowing agents are those that can be dissolved in polymer melts and then gasify under foaming conditions. One of the most popular physical blowing agents is supercritical

CO2 due to the following aspects: low cost, non-toxic and environmentally benign. The polymer foaming process is often coupled with the extrusion-injection molding operation which has been extensively developed in the polymer processing industry. This enables the polymer foams to be mass produced and leads to its wide application. Other than foaming, porous polymer materials can be processed by methods like phase inversion [5] and various types of templating processes [6-8]. Phase inversion is the main principle behind fabricating porous membranes. Polymer membranes are widely applied in filtration, separation, and biomedical applications. Among various templating methods, polyHIPEs prepared via high internal phase emulsion templating has achieved great attention recently

[9, 10]. It normally involves polymerizing (crosslinking) the continuous phase of the emulsion and subsequent removal of the dispersed phase. Comparing to the foams prepared

with blowing agents, emulsion templating needs less extreme processing parameters like high temperature and high pressure. Also, as the emulsion template has controllable flow properties and fast curing rate, the preparation can be processed continuously.

Furthermore, with recycling the dispersed aqueous phase, it can be both environmentally friendly and cost competitive with conventional foaming techniques.

Figure 1-1. Schematics of polyHIPEs preparation process.

PolyHIPEs preparation process normally involves three procedures as illustrated in Figure

1-1: 1) an emulsification step where dispersed phase gets mixed into the continuous phase and a stable emulsion forms; 2) a curing step where the monomers in the continuous phase get polymerized (crosslinked) and a porous cellular structure forms; 3) post-treatment step, for example, removal of the dispersed phase, washing off the residuals, and foam surface modification; usually in this step the foams can gain certain features for a specific application. The first step could be viewed as creating the template-emulsion. The features of the template, for example, stability, morphology (droplet size), phase volume fraction and the formulation can directly affect the final porous foam properties. The second step, curing procedure, serves the purpose of transforming the template (a liquid-liquid

emulsion) to a solid-liquid porous material. Emulsion stability during this step could also affect the final porous foam properties. The last step, post-treatment, enables the porous media new properties by either physical or chemical treatment. This post treatment can readily alter the foam surface properties (for example, wettability) and broaden the application scope of the foams.

The flexibility of HIPE rheological properties enables polyHIPEs to be either molded or machined into various geometries as demonstrated in Figure 1-2. Figure 1-3 shows the morphological features of the HIPEs and polyHIPEs under microscope.

Figure 1-2. A collection of polyHIPEs materials with various shapes. The solid pieces are

made from the same monomers as used in polyHIPEs foams.

Figure 1-3. Micromorphology features of HIPEs and polyHIPEs.

1.3 Dissertation Outline

In the second chapter, the influence of mixing methods during the emulsification step on the curing process, foam morphology, and mechanical properties will be discussed. This study serves as the foundation throughout the work that follows. The third chapter discusses the influence of the foam morphology on two important fluid transport processes, namely Darcy’s flow as in filtration process and capillary flow as in spontaneous imbibition. Also discussed in this chapter is that the wettability of the foam will change when applying a surfactant to the foam surface. In the fourth chapter, it is shown how a physical coating process in the post-treatment step can affect the foam’s mechanical and electrical properties and related potential applications like strain sensor. In the fifth chapter, we circle back to the first step, emulsification procedure, and demonstrate that an addition

of inert diluent in the oil phase can enable significant improvement in the solvent absorbent capacity of the foams.

CHAPTER 2

EMULSION PREPARATION AND THE CORRESPONDING EFFECTS ON

POLY(HIPE) FOAM MORPHOLOGICAL AND MECHANICAL

PROPERTIES

2.1 Synopsis

High internal phase emulsion morphology was studied using three emulsification approaches: propeller mixer, shearing within a syringe-needle and centrifugation. The effects of emulsion morphology on its flow properties and the curing process were investigated by chemorheology. The dynamic interfacial tension was studied with a simplified oil/aqueous phase system by the Wilhelmy plate method. The influence of oil phase emulsifier concentration and aqueous phase salt concentration was studied. The emulsion morphology effect on the cured foam mechanical property was studied through uniaxial compression measurements. Thermal gravimetric analysis indicates that the chemical composition of the cured foams (struts) was altered by emulsion droplet size.

2.2 Introduction

High internal phase emulsions (HIPEs) serve as template for the fabrication of polyHIPE foams. Besides serving as a template, HIPEs themselves have seen various applications in food preparation and cosmetics industry. Many foods are actually HIPEs or in the preparation process become HIPEs. For sample, the mayonnaise is made by blending oil

to egg yolks. The lecithin and protein from the yolk is the emulsifier that stabilizes this oil in water HIPE.

The high internal phase emulsion in this work refers to water-in-oil emulsions, so the oil phase is the continuous phase while the aqueous phase is the internal phase or dispersed phase. The continuous oil phase will transform into a crosslinked network that makes up the foam struts. The dispersed phase will remain as droplets during the polymerization and will eventually form the voids after evaporation. As the density of the polymer is larger than that of its precursor monomer, contraction of the oil films within the HIPE will happen during the polymerization. This contraction is believed responsible for the formation of interconnecting throats (windows) between neighboring voids. It is necessary here to define some nomenclatures. “Void” refers to the space that was occupied by the dispersed phase droplets. “Window” refers to the opening that connects neighboring voids. Later on in chapter 5, “pores” will refer to the meso-voids that exist in the struts themselves; these features are created by phase separation during free radical crosslinking copolymerization in the oil phase.

Briefly introduced above was the polyHIPE foam formation mechanism, and so it is not hard to conclude that any changes to the HIPE morphology will be reflected on the polyHIPE foams. Thus engineering the emulsification process will enable controlling polyHIPE morphology. Key emulsification parameters consist of the continuous-phase-to- internal/dispersed-phase volume ratio, the composition of each phase (monomers, crosslinkers, emulsifiers, and salt), and the mixing method for droplet breakup.

2.2.1 Droplet breakup mechanism

The mechanism of droplet breakup was first studied by G.I. Taylor[11] under two flow conditions, namely a pure elongation flow (four-roll-mill device) and a combination of elongation and rotation flow (parallel plates). These two flow fields were simple yet very common in practical mixing device. It was found that the elongation flow was more effective in droplet breakup since the elongation stress and the deformation are in the same direction while in the parallel plates apparatus there was a 45° angle between them, thus resulting in additional rotational movement instead of just droplet breaking. Besides the effect of flow field, also studied in the same work was the viscosity ratio of two phases

(dispersed phase viscosity divided by continuous phase viscosity). For very small viscosity ratios, the droplet remains coherent in spite of the fact that it gets very long and narrow for both flows. As the viscosity ratio increases, it gets easier for droplet breakup and again the four-roll setup was found to break droplets more effectively. When the droplet viscosity is several times that of the continuous phase, the four-roll setup was found to be able to break up the droplets however, the parallel plates were no longer capable to do so. It is very interesting that the manner of droplet breakup was found to be first an elongation to a threadlike form then breakup into much smaller sizes of 1/100th of the original size. When a droplet was deformed from a spherical to an ellipsoidal shape, the deformation was defined as:

퐿−퐵 D = (2-1) 퐿+퐵

where L is the long axis and B is the short axis. Taylor showed that in order for a droplet to “burst”, the system needs to exceed a critical deformation defined as:

푟̇푟 휇 푓(푝) E = 푑 푐 (2-2) 훾 where 푟̇ is the shear rate, 푟푑 is the droplet radius, 휇푐 is the continuous phase viscosity, 훾 is the interfacial tension, 푓(푝) is a function of the viscosity ratio 푝 (dispersed phase divided by continuous phase). Mason[12] later demonstrated that for viscosity ratio between 0.1 to

10, D equals the critical deformation E in the range of 0.5 to 0.6.

Figure 2-1. Effect of viscosity ratio in simple shear field.

Figure 2-2. Comparison of simple shear flow and elongation shear flow. 14

The emulsion droplet breakup mechanism was later further studied by Grace[13]. A plot of the critical deformation (capillary number) versus viscosity ratio was generated by Grace as shown in Figure 2-1. It is of great practical value because the continuous phase viscosity spans from 45 to 2800 poise and the interfacial tension ranges from a fraction of 1 mJ/m2 to 25 mJ/m2. This general correlation allows prediction of the critical shear rate 푟̇ required to break a drop of radius 푟푑 at a continuous phase viscosity 휇푐 and an interfacial tension 훾 for any viscosity ratio from 10-6 to 3.5 in simple shear field (like a Couette device). The minimum occurs in the range of 0.1-1.0 disperse-to–continuous-phase ratio, whereas at either lower viscosity ratios or higher viscosity ratios the critical shear required to break up the droplet increases rapidly. From Figure 2-2, comparison was made between a simple shear flow and hyperbolic flow (4-roller) for critical shear (for droplet breakup) versus viscosity ratio. It shows the critical shear for achieving minimum value is about 0.6 for simple shear flow while for hyperbolic flow this value is reduced to 0.2. Also seen in Figure

2-2 is that critical shear for hyperbolic flow was even smaller compared with simple shear when moving in both directions away from the minimum, which indicates that indeed the hyperbolic flow is more effective for droplet breakup.

Figure 2-3. Typical observations of deformation and burst of drops in plane hyperbolic flow. Part (a): Water drop (p=2x10-4). After burst the drop formed unstable pointed ends from which small droplets were ejected. Part (b): Castor oil (p=1.0). After burst the drop

became extended into a long cylindrical thread which eventually disintegrated.

Continuous phase is silicone oil (52.6 poise) in both cases.

In Mason’s work[12], droplet breakups were categorized into several classes for both simple shear flow and hyperbolic flow. For hyperbolic flows generated by four roller apparatus, two main breakup scenarios were observed as shown in Figure 2-3. At low viscosity ratios, the ends of the particle form sharp points from which fragments of satellite droplets were released; it was found when p>0.2, instead of forming pointed ends, the drop was deformed into a thread which increased in length until it reached a diameter on the order of 10~50 μm, when it broke into smaller droplets. Similar observations were also made from the simple shear flow situation as shown in Figure 2-4.

Figure 2-4. Illustration of drops in shear flow showing the change in deformation D,

angle of the main axis with increasing shear rate up to breakup. (a) p=2x10-4, small satellite droplets ejected from the main droplet. (b) p=1.0 neck formed and three satellite

droplets. (c) drops were drawn out into long cylindrical threads. (d) no breakup was

observed.

2.2.2 Emulsification methods

Various devices can be used for emulsification purpose from simple pipe flow (low efficiency); static mixers and general stirrers (medium efficiency) to colloid mills, high- pressure homogenizers and ultra-sonication process. The emulsification process can be continuous (pipe flow/static mixers) or batch wise (stirrers). The regime of the flow

(laminar/turbulent) also affects the emulsification outcome. A higher viscosity of the continuous phase can lead to smaller droplets in the turbulent viscous regime and this effect will be more significant in the laminar viscous flow regime. However, it was found that

the viscosity of the continuous phase doesn’t affect the droplet size in the turbulent inertia regime[14].

2.2.3 General emulsion stability

Normally an emulsion is not a thermodynamically stable mixture since it has a much larger interfacial area than would a separated layered mixture with the same composition.

Kinetically an emulsion can be stable for longer period of time with the help of surfactants

(referred to as emulsifiers in emulsion industry). A surfactant molecule usually has multifunctional groups that enable it to lower the mixture energy by specifically absorbing at the two-phase interface. Hundreds of emulsifiers have been synthesized to accommodate different emulsion systems. The HLB (Hydrophile-Lipophile Balance) value which typically ranges from 1 to 15 has been used to evaluate the polar to nonpolar properties of the emulsifier. Generally, the higher the HLB value is, the stronger is the polarity of the surfactant. Emulsifiers at the higher end of the HLB spectrum are suitable for preparation of oil in water emulsions and vice versa. Because emulsifiers reduce the interfacial tension, a lower capillary numberresults, and thus less viscous stress (or energy) is required for breaking up droplets. Another important role for the emulsifier is the fact that they prevent the droplets from coalescence by generating an interfacial tension gradient[14].

2.2.4 Causes for emulsion instability

An emulsion could break down via several processes. Emulsion creaming and sedimentation describes the movement of droplet under the gravitational or centrifugal 18

forces. A concentration gradient (or phase volume fraction) builds up in the system with the larger droplets moving faster to the top (if their density is lower than that of the medium like oil in water) or to the bottom (in case of water in oil emulsion). The limiting case will be a close-packed array of droplets with the remainder of the volume occupied by the continuous phase. Flocculation refers to aggregation of the droplets (without any change in primary droplet size) into larger groups. It originates from the van der Waals attraction that is universal with all dispersed systems when there is not sufficient repulsion to keep droplets apart. Ostwald ripening results from the finite solubility of the liquid phases. Even though liquids are labeled as immiscible, there will still be mutual solubility. For example, the oil phase monomer EHA has water solubility about 0.01wt% and EGDMA has water solubility of 0.5wt%. Due to the polydisperse nature of emulsions, the smaller droplets will have larger solubility when compared with the larger ones (because of curvature effects).

With time, the smaller droplets disappear and their molecules diffuse through the bulk continuous phase and deposit into the larger droplets. And the droplet size distribution shifts to larger values. Coalescence refers to the process of thinning and disruption of the liquid film between the droplets with the result of fusion of multiple droplet into larger ones. The limiting case will be the complete separation of the emulsion into two distinct liquid phases. Phase inversion refers to the process of exchange between the disperse phase and the medium (need to elaborate).

2.2.5 High internal phase emulsions

When the droplet-phase concentration is dilute, the droplets are far apart such that the droplets behavior (movement) and the emulsion overall properties could be well approximated as a dilute dispersion (Stokes’ equation, the Einstein equation for diluted dispersion, viscosity, etc.). As the internal phase concentration (volume fraction) gets larger, the hindrance effect of droplet movement from collision/attraction of close-by droplets gets more significant. Emulsions are conveniently grouped by their internal volume fraction. Emulsions with less than 20 percent by volume of internal phase are arbitrarily designated as low internal phase ratio emulsions. Those with about 20% to about

65% of internal phase are medium internal phase ratio emulsions and those above 70% are called high internal phase ratio emulsions. Emulsions properties like rheology are greatly affected by phase volume fractions. To be specific, the packing limit of disperse phase is

74% by volume for rigid, monodispersed spheres and this limit can be passed either by increasing the polydispersity or deforming the droplets shape from sphere to polyhedral.

Assuming mono-dispersed and elastic droplets, Lissant studied the packing conformation of high internal emulsion droplets [15-17]. The droplets of the emulsions have freedom of movement until their volume fraction reaches 68%. In the region between 68% and 74%, spherical is still the droplets preferred stable shape. Between 74% and 94%, the droplets will prefer a rhomboidal dodecahedron configuration as shown in Figure 2-5. While above

94% the droplets will prefer a tetrakaidehedron configuration as shown in Figure 2-6. In the same work, Lissant pointed out that the aging/relaxation would affect emulsion flow

properties like viscosity and yield stress due to reconfiguration of the droplets near the boundaries mentioned above.

Figure 2-5. Droplets in rhomboidal dodecahedron shape[15].

Figure 2-6. Droplets in tetrakaidehedron shape.

High-internal-phase emulsions have received increased attention in the last few decades. It was found that high-internal-phase emulsion (W/O) properties were influenced by the

following parameters: electrolyte concentration in the aqueous phase, surfactant concentration, droplet size and droplet volume fraction. The presence of electrolyte (salts) in the aqueous phase was found to increase the emulsion stability significantly[18]. It was reported that HIPEs prepared with pure DI water as internal phase coarsened with time and produced a fraction of large droplets that grow at the expense of smaller droplets (Ostwald ripening). This coarsening process resulted in a decrease in emulsion yield stress and finally leads to phase separation. Emulsions thus made also couldn’t withstand a single freeze/thaw process. Addition of salts in the aqueous phase as low as 0.02 M significantly slowed down the coarsening process and decreased the droplet size and the HIPE could withstand the freeze/thaw process. The increased stability of the HIPE was attributed to the fact that the salt will increase the absorption of emulsifiers at the interface (due to decreased solubility in water) and the Ostwald ripening will be inhibited as pointed out also by others

[19-21]. Besides salt concentration in the aqueous phase, the emulsifiers in the oil phase play a significant role in determining emulsion morphology and stability. It was found that there is an optimum emulsifier concentration which is between 10-15 percent by weight.

Below this region, the stability of the emulsion was decreased due to the fact that there isn’t enough surfactant to stabilize the water/oil interface and thus the emulsion droplet size is usually large and the final droplet phase volume fraction is limited. Above this region, although the stability of the emulsion was enhanced, the quality (mechanical integrity) of the porous foam after curing was found to decrease significantly. This is due to the fact that the monomer fraction in the oil phase was decreased and phase separation could happen during polymerization in the oil phase[22, 23].

The flow properties of an emulsion are among its most important characteristics in both fundamental research and in manufacturing such as in mixing, pumping, filling or packing.

From the discussion about the emulsion droplet breakup mechanism, it is obvious that emulsion rheology properties are related to the area averaged droplet radius 푅32, interfacial tension γ, internal phase and continuous viscosity ratio, and for HIPE the internal phase volume fraction ϕ and shear rate ṙ are also very important parameters.

Princen[24] studied foam (air-liquid) and emulsion rheological properties in small-strain shear, static situation. Models that relate droplet size, interfacial tension and volume fraction to yield stress 휏0 and storage modulus G′ were proposed as follows:

훾 1 휏0 = 1.28 휙3퐹푚푎푥(휙) (0.74 < 휙 < 1) (2-3) 푅32

훾 1 G′ = 휙3퐸(휙) (0.74 < 휙 < 1) (2-4) 푅32

F(ϕ) and E(ϕ) are both strong functions of internal volume fraction ϕ. Through the study of small-strain shear modulus of a series of real, polydispersed emulsions, the model was further tested and modified and found to be given by

훾 1 G = 1.769 휙3(휙 − 0.712) (휙 > 0.712) (2-5) 푅32

훾 G = 0.509 (휙 ≈ 1) (2-6) 푅32

This model was shown to be more accurate than the previous work done by Derjaguin[25] which overestimated the constant in the following fashion:

훾푆 훾 G = 퐶′ ≥ 0.88 , 퐶′ = 4/15 (휙 ≈ 1) (2-7) 푉 푅32

In the same work, the Sauter mean radius 푅32, spherical droplets areas per unit volume of the dispersed phase followed:

3 ∑푖 푛푖푅푖 3푉 푅32 ≡ 2 = (2-8) ∑푖 푛푖푅푖 푆0

Where 푅푖 the radius of droplet and 푛푖 the number of these droplets, 푆0 the surface area of close-packed spheres and V the volume of the droplets. The real droplets in HIPE are polyhedral as discussed already, and the actual S/S0 for polyhedron droplets should be larger than the one for spherical case. However, for polyhedrons, the average faces per droplets is between 12 and 14, and it turns out that S/S0 is not very sensitive to the exact shape. The ratio is shown to be 1.098 for pentagonal dodecahedron, 1.105 for the rhombic dodecahedron and 1.099 for the tetrakaidecahedron.

2.2.6 Curing of high internal phase emulsions

HIPE can serve as a template for fabrication of porous polymer foams. When the continuous phase consists of monomers and crosslinkers, it can be transformed into a crosslinked network. After the internal phase gets properly removed, the space occupied by droplets transform into voids and the interconnecting pores will normally be generated via chemical contraction. Polymer foams prepared by HIPE template are usually called polyHIPE which stands for poly high internal phase emulsion. The polymerization process can be initiated by radicals in either aqueous or oil phases and will go on via free radical polymerization. It is obvious that HIPE droplet size and internal volume fraction dictate two important parameters of the porous foams, namely void size/distribution and porosity.

It is less intuitive to see how droplet and its volume fraction affect the interconnecting windows. However, it should also be a function of the two due to window formation mechanism from the rapture/contract of the thin films separating adjacent droplets.

Free radicals can be initiated by thermal heating at various temperatures or UV radiation.

Other polymerization mechanisms are also reported in various monomer systems

(ATRP)[26]. The polyHIPE cured from monomers and crosslinker in the oil phase can be treated as thermosets. The curing of thermosets can be studied by various techniques, for example, liquid chromatography, differential scanning calorimetry, IT spectroscopy, dielectric measurement and particularly viscoelastic behavior measurement- chemorheology. The word chemorheology arises from the two areas of interest being studied[27]. The first interest comes in the change of physical (rheological) properties during the reaction. The processability or flow (rheology) of the emulsions system is of primary importance because process changes such as heating rates, hold temperatures, or pressure are made during this early period based on material viscosity. The other area of concern is the chemistry of the process: the rate of reaction, the mechanisms, the kinetics and the end of the reaction. The interaction between these two concerns are obvious because the rate of rheological change in either the liquid or solid state and the development of the optimum glass transition temperatures cannot be separated from the chemical contribution to these effects. Not only chemorheology connects the reaction kinetics with rheology, its convenience, one step measurement provides great capability for various thermosetting polymer systems. During curing of thermoset polymers, the reactive system transforms from liquid monomers/crosslinkers to a crosslinked solid network. For free

radical crosslinking copolymerization like the case in the oil phase in this study, the continuous phase consists of Newtonian liquid monomers at the beginning. As the radicals propagates, the viscosity of the oil phase increases with the increase in molecular weight until the gelation point where the crosslinked network is formed and the viscosity at this point increases significantly.

Since initially the system is liquid-like (either Newtonian or non-Newtonian fluid), the viscous loss modulus dominates compared with elastic storage modulus. At the gelation point, the storage modulus will cross and pass the loss modulus. So this intersection can be viewed as the gelation point from rheology measurement. When measuring the viscoelastic behavior of the pure oil phase, we should see clearly the transition of viscosity during the curing process. The viscoelasticity of high internal phase emulsion is unique. The storage modulus dominates the loss modulus over the whole curing process due to the elastic contribution from the large oil/water interfacial films. So it is less obvious to define the exact point of gelation. In a previous study of curing of HIPEs, the gelation point was picked as the change in the slope of the loss modulus curve[28].

2.2.7 Interfacial tension kinetics

The interfacial tension lowering effect of the emulsifiers plays a crucial role for the HIPE droplet size reduction and increased stability. This results from the adsorption of emulsifiers at the liquid-liquid (water/oil) interfaces. This phenomenon has a remarkable impact not only on emulsion system but all systems that involve multiphase systems. The adsorption phenomena of a small molecule on a surface was first systematically studied 26

almost a century ago by the Langmuir[29-32]. In this remarkable work, the adsorption/desorption phenomena were modelled as a reaction between the available adsorbing sites and adsorbing molecules. This process is affected by adsorption/desorption rate constant, available adsorbing sites concentration (total surface area) and the gas molecule concentration. The case for emulsifiers is similar to the above in principle, the unique feature about the emulsion system is the partitioning of the surfactant between the two liquid phases, which is not happening in the gas-solid adsorption case. The origin of the partitioning comes from the limited but non-zero solubility of the emulsifiers in the aqueous phase (in the literatures focused on this area, the smaller solubility phase is called the extracting phase describing that it extracts surfactant from the other phase) [33, 34].

Furthermore, it was reported that partitioning of the emulsifiers can affect emulsion stability[35]. Thus in addition to the parameters mentioned above, information of partition properties of the emulsifier and the volume ratio between two phases are crucial for fully understanding or predicting interfacial phenomena in the emulsion system.

Adsorption dynamics can be divided into several aspects during which the emulsifier can be transferred between various regions. First of all, the incident that triggers the adsorption dynamics can be a variation in the interface area, either by generating a fresh interface or demising an interface, or a variation of the concentration in the bulk. To restore an equilibrium state, emulsifiers can transfer between the interface and the solution layer right next to it (called the sublayer in both phases), between the sublayers and the bulk phases.

The mass transfer of the former is by adsorption (viewed as a reaction) while the latter process is merely diffusion controlled. Normally the adsorption process is much faster than

diffusion, thus diffusion is usually the controlling mechanism[36]. Exceptions with adsorption the controlling step were reported also and were found due to re-orientation of the emulsifiers[37-40]. It can be found that in (semi-) infinite systems the adsorption process and transfer of emulsifier hardly affect the bulk concentration far from the interface and a monotonic variation of interfacial tension is always expected. However, in many practical situations, the semi- or infinite assumption is not real. For example, when the aqueous-to-oil phase ratio is very large (>>1) and the emulsifier concentration in the aqueous phase is zero initially, the large extracting driving force is expected to eventually deplete the emulsifiers in the oil phase. Thus, a correction for a finite system is necessary for correctly modelling or predicting the emulsifier effect on emulsion properties. An approach to this problem is given by Shimbashi[41] in the case of two equal phase volume at the partitioning equilibrium. It was found that the depletion effect is remarkable when the volume is limited and the initial concentration of one phase is far from equilibrium. A minima was first observed by Mansfield[42] and later modelled by Rubin and Radke[43].

The transport between the bulk phase and the sublayer was described according to the

Nernst film model, while the exchange between the sublayer and the interface was described by a linear or a Langmuir model. The appearance of minima in the dynamic interfacial tension was theoretically assessed and the condition for the appearance and their characteristics were shown to be dependent on the volume ratio between the two phases, on the ratio of the diffusion coefficients and on the adsorption and desorption rate constants. In particular, this latter dependence is equivalent to a dependence on the partition coefficient.

2.3 Experimental

2.3.1 Materials

Oil Phase: HIPE oil phase consists of monomers, crosslinker and emulsifiers. The monomer and crosslinker used in the experiments were 2-ethylhexyl acrylate (2-EHA) and ethylene glycol dimethacrylate (EGDMA), respectively, and were obtained from Sigma

Aldrich. The emulsifier polyglycerol succinate (PGS) and DTDMAMS were used as received. The oil phase was made up of a 3.35:1 ratio of 2-EHA: EGDMA if not specifically pointed out otherwise. A default 12% by weight emulsifier was dissolved in the oil phase prior to emulsification at 40°C. The interfacial tension of this oil phase with the aqueous phase is 0.8 mJ/m2 measured by the pendent drop method. The emulsifier concentration may vary when studying the interfacial dynamics between the two phases.

Aqueous phase: The aqueous phase consists of deionized water, 2.0 wt% sodium chloride

(NaCl) from Fisher Scientific and 0.33wt% sodium persulfate (NaPS) as the initiator

(obtained from Sigma Aldrich, 98%).

2.3.2 Mixing procedures: tuning HIPE morphology

Propeller mixer: In a typical emulsification procedure, the aqueous solution was heated to 40°C and added into the oil phase kept approximately at the same temperature to form

HIPE using a digital gear pump (Cole-Parmer Instrument Company) at 40 mL/min. The system was mixed in a HDPE cylinder of inner diameter 67.2mm thoroughly during

addition of the aqueous phase by a Fisher Scientific StedFast Stirrer Model SL 1200 mechanical mixer (propeller diameter of 52.1mm) at 300 rpm. The HIPE was then stirred further for various amounts of time. A coding system RXXMXX was adopted for such prepared emulsions. For example, R19M04 represents the emulsion made with a water to oil weight ratio of 19 and being mixed for 4 min.

Syringe-needle mixer: Emulsions prepared with the above mentioned propeller mixer with 4 min of total mixing time were cooled down to room temperature and passed N times

(N = number of passes) through a syringe with diameter 2 cm and a needle with diameter

0.838 mm and length 25.4 mm. This setup induces a combination of shear and elongation flow on the HIPE samples. Since the volumetric flow rate of pumping process was kept at

~0.5 cm3/s, the maximum strain rate during flow is calculated to be on the order of 104 s-1.

It should be noted that the real strain rate is lower than this approximation due to the wall- slip of the HIPE noticed at flow boundaries[44]. Figure 2-7 is schematic illustration of both mixing setups. HIPEs prepared by either setup were cured in 10ml glass beaker at 70C° in a convective oven overnight.

Figure 2-7 Mixing setup (a): emulsion prepared by a propeller mixer; (b) pristine

emulsion pushed through a syringe. 30

Figure 2-8. Centrifuge setup and schematic illustration of the dimensions.

Centrifuging the HIPE: HIPE was first prepared by propeller mixer method and was transferred into 50 ml centrifuge tubes. Then the HIPE was subjected to various centrifugal operations at various rotational speed and centrifugation time. After centrifugation, the free oil phase was removed from the top and the HIPEs were cured in the same tube at the same condition as previously mentioned. Figure 2-8 shows the schematic of the centrifugation setup.

Figure 2-9. Wilhelmy plate setup.

Figure 2-10. Schematic of Wilhelmy plate method.

2.3.3 Interfacial tension kinetics using Wilhelmy Plate method

Interfacial tension between emulsion aqueous phase and oil phase was measured using a

Wilhelmy plate setup (Kruss K100 tensiometer) as shown in Figure 2-9 and Figure 2-10.

Oil phases with emulsifier concentration ranging from 0.01wt% to 1.0wt% were prepared.

Various NaCl concentration in DI water was prepared as the aqueous phase. The aqueous- and oil-phase volumes were kept constant at 10 and 25ml respectively. Measurements were taken over a 2 hr period with a data sampling rate of 2Hz. After each measurement, the

Wilhelmy plate was rinsed with IPA and DI water, heating with a Bunsen burner to a cherry red color, and then cooled down to room temperature. This was repeated 3 times before subsequent measurements.

2.3.4 HIPE and polyHIPE morphology analysis

Emulsion droplet-size distributions were analyzed using an optical microscopy (Olympus

BX51). About 10µl of emulsion sample was placed on a flat glass slide and the cover slide was carefully placed on top of it. In order to achieve good image quality and prevent droplet breakup gentle pressure was put on the cover slide to ensure a monolayer of emulsion droplets. The images were then processed using ImageJ to get the size distribution. Typical optical microscopy images before and after ImageJ processing are shown in Figure 2-11.

A scanning electron microscope (SEM, JEOL JSM-6510LV) operating at 30 KV was used for studies of the morphology of the cured polyHIPE samples. Residual salt and emulsifier were removed from the polyHIPE specimens by Soxhlet extraction using first DI water and then isopropanol, each for 24 hr. Subsequently, the samples were sputter-coated with a thin layer of gold to achieve conductivity prior to SEM imaging.

Nitrogen adsorption/desorption measurements were performed at 77.3 K on a

Micromeritics TriStar II gas adsorption analyzer. PolyHIPE samples were degassed at 80

°C overnight under nitrogen. An 11-point adsorption isotherm collected over 0.01-0.30

P⁄P0 was used for the surface-area measurements and the data were analyzed via the BET method theory[45, 46].

Figure 2-11. A typical optical microscopy image of HIPE (a) and the same image after

ImageJ processing (b). The software will identify the droplet automatically after image

processing.

2.3.5 Chemorheology study

Isothermal time-sweep tests (TA Instruments ARES G2 Rheometer) were run on the polyHIPE samples. Data were obtained in the linear viscoelastic regime at a constant frequency of f = 1 Hz and strain of 0.5%. Measurements were conducted with a 1 mm gap at 70°C. The accuracy of temperature control was on the order of ±0.1°C. A serrated-plate geometry was employed for the rheological measurements to avoid wall-slip due to the solid-like behavior of the HIPE[44]. A water immersion setup was devised to study the chemorheology of samples in the wet state.

2.3.6 Mechanical properties

An Instron 1101 device was used to perform compression testing on cylindrical (diameter

= 25 mm, height = 10 mm) specimens of as-cured polyHIPE. Tests were performed at a 1 mm/min compression rate at room temperature. The samples were found to exhibit negligible bending during measurements.

2.3.7 Thermal stability analysis

The thermal stability of the polyHIPE samples was characterized by thermal gravimetric analysis (TGA, TA Instruments Q500). The temperature was ramped from 25°C to 600°C at 5°C /min. Prior to testing, residual salt and emulsifier were removed from the samples by Soxhlet washing using first DI water and then isopropanol each for 24 hr.

2.4 Results and Discussion

2.4.1 The effect of propeller mixer setup on emulsion morphology

The microstructure of emulsions can be characterized through its droplet size distribution.

Two common indices are the arithmetic-mean diameter (D10) and the area weighted Sauter- mean diameter (D32), defined as:

n d D   i i 10 n  i (2-9)

n d 3 D   i i 32 n d 2  i i (2-10)

where the summation is taken over a representative population of droplets. A polydispersity index (PDI) can be defined as the ratio of D32/D10. If PDI = 1, the system is monodispersed, with larger values of PDI indicating more polydispersity.

The tendency of liquid droplets to break under the action of hydrodynamic shearing is governed by the balance between shear stress and interfacial forces acting on the droplet.

These can be quantified through the capillary number, which gives the ratio of applied shear stress to the Laplace pressure:

휂 훾̇ 퐷 퐶 = 푑 32 (2-11) 푎 2훾

Where ηd is the dispersed phase viscosity, 훾̇ is the shear rate, and γ is the interfacial tension.

In order for droplet breakup to occur, Ca must exceed a critical value that depends on the ratio of viscosity in the dispersed phase (d) to that of the continuous phase. For the materials used in this work this viscosity ratio is about 0.006, and a value of Ca ~ 10 is expected for the high shear conditions applied to the HIPEs. The critical capillary number for breakup in shear and elongation flows at such viscosity ratio is about 3.0 and 0.4, respectively [47]. Thus, significant breakup of the water droplets is expected for both of the shearing conditions (propeller mixer or syringe setup) applied. With additional interfacial area created as a result of droplet breakup, at some point the amount of emulsifier present in the system may not be sufficient to stabilize the system, thereby 36

allowing coalescence to occur. In addition, van Aken and Zoet [48] showed that flow- induced coalescence in a HIPE can follow a fracture-type mechanism which involves the coalescence of many emulsion droplets along a slip or fracture plane. The resulting droplet size distribution within these zones can become very heterogeneous with very large droplets present.

Figure 2-12. Mixing time effect in propeller mixing setup on emulsion morphology. (a)

R19M04, (b) R19M10, (C) R19M30.

0 5 10 15 20 30 4 min

300 5 min

300 8 min

300 10 min

Percent by counts

300 15 min

300 30 min

0 0 5 10 15 20 Droplet size (um)

Figure 2-13. Propeller mixer mixing time effect on HIPE droplet size distribution. 37

As shown in Figure 2-12 and Figure 2-13, the propeller mixer set up generated droplets in the order of microns. The size distribution of the pristine emulsion R19M04 spans from fraction of a micron to 20 µm, the majority of the droplets fall near 5 µm, however a long tail was present. The droplets continued to break up as mixing time increases. The fraction of droplets (>5µm) decreases and a greater number of droplets (<5 µm) were generated.

For M30 emulsions, all the droplets fall below 10 µm and the peak shifts to 2 µm. From

Figure 2-14, it can be found that the average droplet size shifts from 15 µm to 5 µm, a decrease by a factor of 3 while the polydispersity decreased from 2.6 to 1.6, which indicates that the distribution gets narrower. It can also be found that after 10 min, the droplet size reduction was relatively insensitive to the mixing, which can be explained by equation 3 that as droplet size gets smaller, the ratio between shear force and the capillary pressure gets smaller and finally achieved kinetic equilibrium.

Figure 2-14. Propeller mixing time effect on HIPE average droplet size and PDI.

Figure 2-15. Storage modulus vs. mixing time.

The droplet size effect on emulsion rheology properties was studied. The small-strain

(0.5%) frequency sweep technique was adopted to study the structure of the emulsion.

Results for emulsions from R19M04 to R19M30 are shown in Figure 2-15. A plateau in the low frequency region was found for every sample which indicates a solid-like behavior due to the restrained movement of the droplets. This can be attributed to the closely packing of the droplets which limits the flow of the dispersed phase in the continuous oil phase. As the droplet size decreases (with increasing mixing time), the plateau value increases and extends in the higher frequency region. The plateau value was plotted with droplet size for

R19M04 to R19M30 in Figure 2-16. The increase in storage modulus of the emulsion indicates that the solid-like behavior is strengthened by reducing droplet size. Princen proposed a model[44, 49] that relates the HIPE droplet size with the storage modulus and found that the G' scales with the inverse of the droplet size. Results of G' versus 1/D for

the emulsions are plotted in Figure 2-17 and are found to be in good agreement with this model.

Figure 2-16. Storage modulus plateau vs mixing time.

300

Equation y = a + b*x Plot Emulsion storage modulus Weight Instrumental 250 Intercept -58.28605 Slope 1360.60427 Residual Sum of Squares 19.45135 Pearson's r 0.95876 R-Square(COD) 0.91921 200 Adj. R-Square 0.89228

150

100

Emulsion storage modulus (Pa)

0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20 Inverse of droplet size (m-1)

Figure 2-17. Linear relationship between emulsion storage modulus and inverse of

droplet size.

2.4.2 The effect of syringe mixer setup on emulsion morphology

Droplet size and distribution was also studied for emulsions processed in the syringe and needle setup. The pristine emulsion R19M04 was first prepared with the propeller mixer setup and then it was processed through the syringe and needle. The sudden diameter change from 20 mm to 0.84 mm will generate an elongation shear in addition to the simple shear flow.

The effect of shearing on HIPE morphology is illustrated in Figure 2-18 and Figure 2-19.

The optical microscopy images show that the droplet size reduces and the droplet size distribution becomes narrower for the first few shearing cycles of the HIPE (N < 3), but for additional shearing (N > 3) some coalescence of the droplets was noticeable.

Figure 2-18. Optical microscopy images of HIPE with 12wt% surfactant for (a) N = 0, (b)

N = 1, (c) N = 5, and (d) N = 9. The scale bar depicts 100µm.

The variations of droplet size and PDI for HIPE systems prepared with different N are shown in Figure 2-19. Both D32 and PDI decrease significantly as N increases from 0 to

3, which indicates that during these shearing events, the capillary number exceeded the critical value for droplet breakup. The droplet size-distribution becomes narrower upon repeated shear cycles due to the preferential breakup of larger droplets.

4.0 14 D 32 3.5 PDI 12 3.0 10 2.5

)

m 8



( 2.0

PDI

D 6 1.5

4 1.0

2 0.5

0 0.0 0 2 4 6 8 10 N

Figure 2-19. The variation of droplet size and polydispersity for HIPE with different

number of shearing cycles (N).

While droplet breakup was expected to continue for the shearing cycles beyond N =3

(because the Ca still exceeded the critical value for droplet breakup), for N > 3, only negligible shifts in D32 and PDI were seen. Thus, the break-up and coalescence processes struck a kinetic balance that results in a relatively stable D32 and PDI.

To further test the stability of the HIPE, emulsions containing no initiator (and thus not prone to polymerization) were subjected to shearing. Figure 2-20a shows the results of rheological tests performed at 25 °C. With increased number of shearing cycles and thus a reduction in droplet size, the storage modulus was found to increase. However, there was not a significant variation of storage modulus over time, which indicates that no coalescence occurs under this condition. However, when the test was carried out at 70°C

(the temperature which would be used for curing the HIPE), significant changes of the storage modulus with time was observed. Figure 2-20b shows the storage modulus (G') relative to its value at time zero (G0') for emulsions subjected to various cycles of shearing.

The strong reduction in G' over time for N > 3 can be attributed to the coalescence of the droplets that occurs at this higher temperature.

Figure 2-20. Storage modulus of emulsions prepared without initiator and subjected to

different number of shearing cycles (N): (a) 25°C and (b) 70°C.

Figure 2-21 shows the storage modulus for HIPE samples subjected to different shearing cycles and then cured at 70°C. Similar to previous work [36], each curve shows three regimes: (i) an initial induction period, where the storage modulus remains unchanged over time; (ii) the intermediate polymerization period, wherein a steep increase in the moduli is observed; and, (iii) a final curing period, in which a plateau in the modulus is reached. Key features in Figure 2-23 include that the initial modulus increased with the number of shearing cycles for 0 < N < 3, but then remained constant for larger number of cycles. This is consistent with the reduction in droplet size observed with the initial shearing cycles.

Secondly, it is noted that for samples with N > 3, the storage modulus decreases with time prior to its increase during the polymerization period. This decrease reflects the coalescence that can occur within the emulsion, as demonstrated with the results presented in Figure 2-20b. Lastly, the final plateau value of the storage modulus was found to be similar for N  3. This is consistent with the constancy of the morphological results displayed in Figure 2-19 for N  3.

Figure 2-21. Dynamic modulus versus time for samples processed with different numbers

of shearing cycles and cured at 70°C. 44

Figure 2-22 shows SEM images of polyHIPE samples cured from HIPEs subjected to different numbers of shearing cycles. Table 2.1 presents some morphological parameters and physical property data for these samples. The relative density (the ratio of the polyHIPE density to that of the polymer itself), which is a measure of the overall porosity of the polyHIPE, is fairly constant for all values of N. The BET surface area, which provides a coarse characteristic parameter inversely related to the void size within the polyHIPE, is also presented in Figure 2-23. The BET surface area increases with N for 0

 N  3 but remains relatively constant for N  5.

Figure 2-22. SEM images of cured polyHIPE samples prepared from HIPEs sheared with a different number of shearing cycles: (a) N = 0; (b) N = 1; (c) N =3; (d) N =5; (e) N =7;

and (f) N = 9. The scale bar depicts 10 µm.

Based on analysis of the SEM images, two representative void sizes can be defined. The primary voids correspond to the small droplets present within or formed upon shearing the

HIPE, while the larger voids are attributed to the coalescence which occurred during the curing process. Two opposite trends can be found here. The primary void size decreased with N for 0  N  3 but remained relatively constant for N  5. However, the size of the large droplets formed from coalescence was found to increase slightly with N, which indicates that the emulsions subjected to more shearing cycles (and thus had more time elapse prior to polymerization) were more prone to the formation of larger droplets.

20 80 18 70

/g) 16

2 60 14

12 50

10 40

8 30

6 20 4 10

BET surface area (m 2

Compression modulus (KPa) 0 0 0 1 3 5 7 9 11 13 N

Figure 2-23. BET surface area and modulus for polyHIPE samples prepared from HIPES

subjected to different numbers of shearing cycles P0-P13.

Figure 2-24 displays the results of mechanical testing of the polyHIPE samples in a wet state (i.e., prior to drying). The stress-strain curves show that some yield or plasticity is seen for polyHIPE samples prepared from HIPEs for which N > 3. The relationship between the observed Young’s modulus and the number of shearing cycles applied to the

HIPE prior to curing is summarized in Table 2.1 and Figure 2-24. The Young’s modulus was also found to increase with N for 0  N  3 but remains relatively constant for N  5.

To further explore the effect of shearing of the HIPEs on the mechanical properties of the polyHIPEs, a few additional samples were prepared in which N = 11 or 13 shearing cycles were applied. The results clearly demonstrate the existence of a plateau in the modulus.

This result corroborates the assertion that droplet breakup and coalescence achieve a kinetic balance for higher numbers of shearing cycles, and that this leads to a stable polyHIPE morphology.

Figure 2-24. Stress-strain curves for polyHIPE samples under compression. Samples

were prepared from HIPEs prepared with different numbers of shearing cycles.

N9 N7 N5 N3

Deriv.weight (%/ N1 N0

0 100 200 300 400 500 600 Temperature (oC)

Figure 2-25. Derivative TGA curves for polyHIPE samples prepared from HIPEs

subjected to different numbers of shearing cycles.

Table 2.1. Morphological and physical properties of the polyHIPEs.

N, Number Relative BET Young’s Primary cell Coalescence of shearing density Surface modulus size (µm) cell size cycles area (m2/g) (kPa) (µm)

0 0.0650 2.7 ± 0.5 7.0 ± 1.5 26.8 ± 0.4 N/A

1 0.0659 3.2 ± 0.6 11.5 ± 2.1 7.1 ± 0.7 11.8 ± 0.8

3 0.0699 6.6 ± 1.1 54.7 ± 2.4 5.5 ± 0.4 16.7 ± 0.3

5 0.0708 9.2 ± 1.5 65.2 ± 3.3 4.4 ± 0.4 17.0 ± 1.0

7 0.0710 11.2 ± 3.7 72.8 ± 5.0 3.6 ± 0.8 22.8 ± 2.1

9 0.0675 8.4 ± 1.6 68.2 ± 3.3 3.5 ± 0.4 22.5 ± 2.3

11 68.9 ± 5.4

13 66.1 ± 11.9

Some insight on the mechanical behavior of polyHIPEs can be derived from the work of

Gibson and Ashby [50, 51]. For cellular solids, the compressive modulus is related to the density of the material and the modulus of the struts comprising the structure. Although the model developed by Gibson and Ashby is based on the assumption of uniform void size distribution, this is certainly not the case with polyHIPEs. Wong et al. [42] have shown that for cellular materials containing a bimodal void size distribution there is an increase in modulus by comparison with uniform void size distribution. Also, the modulus of the struts themselves can be affected by the presence of nanoscale pores inside the walls as reported by Silverstein [43] and more recently by Maheo [44] and Ceglia [45]. Moreover, the samples in this work were tested in a wet state and the water transport within the polyHIPE can also influence the results.

The thermal stability of the polyHIPE samples was investigated using TGA and Figure

2-25 displays results for polyHIPE samples prepared from HIPEs that were sheared for different numbers of shearing cycles. For all samples, two peaks are found in the first derivatives of the TGA curves at around 300 and 400°C. The maximum rates of mass loss for neat poly(2-EHA) and poly(EGDMA) are reported in the literature as 403°C [52] and

300°C [53] respectively. Thus, the two peaks recorded for the polyHIPE samples are attributed to the degradation of the copolymer backbone consisting of 2-EHA and EGDMA units. Interestingly, the peak corresponding to the EGDMA units shifts from 320°C (N =

0) to 302°C (N = 9) and of the magnitude of the peak corresponding to the degradation of

EGDMA units increases as N increased for the parent HIPE. This shift to lower degradation temperature for EGDMA units indicates a weaker EGDMA unit structure

present in the foam. On the other hand, the relative enhancement of the peak at 400°C indicates the enrichment of 2-EHA units within the cellular structure for HIPEs subjected to a larger number of shearing cycles.

The shift in the relative magnitudes of the two derivative-TGA peaks can be attributed to the significant solubility difference of the two monomers used in the HIPE formulation.

The solubility of 2-EHA in water at 20°C is 0.01 wt% while that for EGDMA is 0.5 wt%.

With a larger number of shearing cycles applied to the HIPE, the EGDMA and 2-EHA have an extended opportunity for dissolution into the aqueous phase to occur. Although the transport of the monomers into the aqueous phase will be inhibited by the presence of the emulsifier at the interface [54], during shearing, the emulsifier film will be repeatedly ruptured which can lead to an enhanced rate of transport into the aqueous phase.

2.4.3 Emulsion morphology under centrifugal force

In general, dispersion stability is affected by gravity, van der Waals attraction, repulsion

(in charged system), Ostwald ripening, and coalescence process. Under normal circumstances, these processes would be detrimental to the emulsion stability and quality.

However, some of the processes could be applied to tune the emulsion morphology. For water in oil emulsions, the heavier water droplets are constantly subjected to gravity force and tend to sediment. Stokes’ law[55] could be applied to describe the sedimentation process. For a dilute dispersion system, the sedimentation velocity was related to particle size 푑푝, density 휌푑, liquid phase density 휌0, viscosity μ and gravity constant/centrifugal acceleration C in the following equation: 50

푑 2(휌 −휌 ) 푣 = 푝 푑 0 C (2-12) 푠 18휇

However, as the concentration of the particles increases, the interaction between particles during sedimentation process like friction, attraction/repulsion becomes more significant.

A power-law correction can be made to the sedimentation velocity as follows[56].

푛 푣푡 = 푣푠휀 , 휀 = 1 − 푣표푙(푑푟표푝푙푒푡)% (2-13)

The power law index n was reported to be related to Reynolds number and to be 4.5 for low Reynolds number flows[57]. For HIPE with aqueous phase volume fraction from 75% to 95% (which corresponds to emulsions of R3 to R19), the expected sedimentation velocity by Equation 2-14 was calculated to be 2.0x10-3 and 1.4x10-6, smaller than the value predicted by the Stokes velocity in Equation 2-13. Thus the sedimentation velocity is dramatically reduced as droplet volume fraction increases. The centrifugal setup is shown in Figure 2-8. The detailed dimensions and centrifuge parameters are summarized in Table 2.2.

Table 2.2. Centrifugal parameters

Centrifugal Rotational Centrifugal

radius/m speed/ rpm acceleration/(m/s2)

R1=0.11 2000 4864

R2=0.20 2000 8896

Such an inhibition of droplet settlement is beneficial in terms of preserving the initial emulsion morphology during the processing time scale. However, a gradient of emulsion properties might be beneficial in certain applications. For example, bones are natural porous materials and they have a gradient of cell size in the radial direction. Centrifugal field can be applied to force droplet sedimentation. Emulsions with 75%, 80% and 90% aqueous phase fraction were made with droplet size controlled in approximately the same range. The emulsions were characterized by optical microscopy as shown in Figure 2-26.

The droplets ranges from 50-500 μm.

Figure 2-26. Emulsion morphology of R03, R04 and R05.

Figure 2-27. R03 emulsion sedimentation result.

Figure 2-28. R04 emulsion sedimentation results.

Figure 2-29. R05 gravity and centrifugal sedimentation result.

The sedimentation results are shown in Figure 2-27 to Figure 2-29. Sedimentation by gravity were found for R03 and R04 samples. The sedimentation rate was estimated to be

4.6x10-7 m/s and 2.3x10-7 m/s, respectively. The sedimentation rate for R05 was found to be less than 1.0x10-7 m/s. The decrease in the sedimentation rate is attributed to the increase in emulsion dispersed phase fraction. For 80% and 90% dispersed phase fraction emulsions

(R04 & R05 samples), centrifugal force (parameters as in Table 2.2) was applied to facilitate sedimentation. After centrifugation, a clear continuous oil phase was observed at the top of the tube while the droplets got compressed at the bottom. R03 sample centrifuged

for 1 minute was cured and the foam morphology was obtained by SEM as shown in Figure

2-30. The droplet size distribution was found to be bimodal with the smaller droplets in the order of 50 μm and the bigger ones of 100~1000 μm. The very top region was found to be solid due to curing of free oil phase which indicates the sedimentation continues during curing. Voids next to the top solid region were found to be 100 μm. Large voids in the order of 1mm were found in the middle sections. Such large voids couldn’t be found in the emulsion droplets morphology and thus could have resulted from droplets coalescence due to the compression force from centrifugal field. Throughout the sample, the smaller void size remained similar to the emulsion droplet size 50 μm. Since the emulsion droplet size distribution was quite broad, Ostwald ripening likely contributed to the large voids found in the samples.

Figure 2-30. R03 emulsion centrifuged 1 min and cured.

Furthermore, emulsions were centrifuged for various durations and the the free continuous oil was extracted and weighted, thus the emulsion droplet volume fraction could be

calculated based on the initial emulsion weight and composition. The emulsion droplet phase volume fraction was plotted against the centrifugation time in Figure 2-31. The aqueous volume fraction was found to increase rapidly at the beginning (<5 min) and gradually levelled off. The equilibrium aqueous phase volume fraction was found to be independent of the initial volume fraction but dependent on the centrifugal field. Since the centrifugal force was maintained at the same level (2000rpm), the droplet-phase volume- fraction in the final emulsions was found in the same level which is about 23. The movement of droplets was limited because of the close packing of the droplets typical for high internal phase emulsions. Thus instead of droplet moving, droplets were compressed in the centrifugal force direction (toward the bottom of the tube) and the continuous oil phase was pushed in the opposite direction.

Figure 2-31. Centrifugation time effect on final emulsion water/oil ratio.

Figure 2-32. Schematic of centrifugation process and foam sample sectioned for

mechanical test.

Such a centrifugal filed was found not sufficient to “squeeze” emulsions with higher aqueous phase volume fractions since the droplets were much closer to each other in the beginning. Stronger centrifugal field was applied for R09 samples. Mechanical compression properties of the foams were measured at various sections along the tube as shown schematically in Figure 2-32. This schematic is obviously an ideal situation. Its purpose is to demonstrate the tendency of sedimentation of different droplet sizes.

However, the movement of droplets is confined and the oil phase is forced upwards in the centrifugal field. The compression modulus as well as the bulk density at various sections

from the top to the bottom is shown in Figure 2-33. The corresponding foam morphology can be found in Figure 2-34.

Figure 2-33. R09 emulsion centrifuged for 1 min and the mechanical properties and the

foam density change in the centrifugation direction.

Figure 2-34. R9 foam with 1 min centrifuge time from top to bottom of the tube change

in window size.

The modulus was found to be decreasing from top to bottom. The foam modulus is related to the properties of the strut materials and the local relative density at various positions 57

along the tube (centrifugation direction) as shown in Figure 2-33. The bulk density results clearly indicate a decrease in the foam relative density and an increase in the porosity.

Since the centrifugal force was increasing towards the bottom of the centrifuge tube, the emulsion droplets got compressed more in that region compared to the top, thus resulting in thinner struts and higher droplet phase volume fractions. From foam morphology shown in Figure 2-34, the void size was found to be the same throughout the different sections however, the window size was found to increase along the centrifugal direction. This again can indicate the increased compression state towards to bottom of the tube where droplets got squeezed more closely and the window size (thus formed by tearing in the thinnest point of two neighboring droplets) increases.

2.4.4 Interfacial tension (IFT) dynamics

To better understand the high internal phase emulsion behavior, the influence of surfactant concentration in the oil phase, electrolyte (NaCl) concentration in the aqueous phase on interfacial tension kinetics was studied. Since some formulations have multiple components in the oil phase, to simplify such a complicated system, only one monomer

EHA and one emulsifier PGS were studied as the oil phase. Normally the aqueous phase consists of NaCl and NaPS (initiator). Since the long duration of the experiments (NaPS might induce polymerization), only NaCl was picked as the electrolyte in the aqueous phase. Both phase compositions are listed in Table 2.3.

Table 2.3. Oil phase and aqueous phase composition

Bulk oil phase Bulk aqueous phase

Emulsifier(PGS) monomer Solvent NaCl wt% wt%

0.001 NA

EHA 0.01 DI 1

1 10

Figure 2-35. Emulsifier concentration effect on interfacial tension dynamics. DI water as

aqueous phase.

Figure 2-35 shows the emulsifier concentration effect on the interfacial tension dynamics with pure DI water as the aqueous phase. The water-oil phase volume ratio during the 59

experiments was fixed at 1:2 due to the limitation of the experimental setup. As can be found in this figure, the decrease in the interfacial tension for 0.001% PGS oil phase is extremely slow, while above 0.01% the decreasing of the interfacial tension happens rapidly. This indicates the existence of a critical concentration between 0.001% and 0.01% which corresponds to monolayer absorption of emulsifier molecules at the interface. Below such concentration as the case of 0.001% PGS, there aren’t sufficient surfactant molecules at the sublayer to form a monolayer at the interface, thus further adsorption requires surfactant molecule diffusion from the bulk oil phase which is a process much slower than adsorption. Once the concentration surpasses this critical one, there are enough surfactant molecules required for monolayer adsorption and the dynamics is adsorption controlled such that the interfacial tension drops rapidly.

Figure 2-36. Salt effect on interfacial tension dynamics: emulsifier concentration

0.001wt%.

Figure 2-36 shows the effect of salt concentration in the aqueous phase on the interfacial kinetics. Below the monolayer concentration, the diffusion process is the limiting step. 60

Decreasing salt concentration slightly reduces interfacial tension indicating an increased adsorption concentration at the interface[36]. The rate of IFT reduction was found similar.

When above the critical value, the typical time scale for the interface tension to reach minimum value is found to be less than 1200 seconds. Increasing emulsifier concentration from 0.01 to 1.0 wt% reduces the equilibrium time from 1200 to 200 seconds. Considering the large interface area generated in the emulsion, the bulk emulsifier concentration of the oil phase got reduced as mixing continues. The time required for reaching minimum IFT value increased, which makes it gradually more difficult to generate new interfaces. So again, the dynamics of the interfacial tension plays a crucial role in the emulsion droplet formation and in controlling the emulsion morphology. Figure 2-37 shows the effect of surfactant concentration on equilibrium interfacial tension. The sharp decrease in the IFT around 10-4 mol/L is due to the monolayer coverage of the two phase interface.

Figure 2-37. Critical emulsifier concentration on interfacial tension.

2.5 Conclusions

HIPE droplet size/distribution was controlled by propeller mixer, simple shear-syringe and needle. Emulsion droplet size approaches to equilibrium value before severe coalescence starts to happen due to prolonged shearing. The flow properties scales inversely with the droplet size before curing. Emulsion coalescence was found to increase at 70 °C by static shearing rheology test. The emulsion droplet size affected the cured struts properties as seen from the derivative thermal gravimetric analysis. The peaks that belong to individual polyEHA and polyEGDMA chains gradually segregate apart, due to the dissolution of

EGDMA into the aqueous phase. Centrifugal force caused the migration of continuous oil phase in the opposite direction and resulted in a gradient in the foams relative density, mechanical modulus and window size. The adsorption of emulsifiers to water/oil interface is a diffusion controlled process as demonstrated in dynamic interfacial tension measurement.

CHAPTER 3

MORPHOLOGICAL AND SURFACE EFFECTS ON POLYHIPE FOAM

TRANSPORT PROPERTIES: PERMEABILITY AND SPONTANEOUS

IMBIBITION

3.1 Synopsis

Due to window formation during emulsion curing, polyHIPE foams have an interconnected porous structure. Such morphology enables polyHIPE to potentially function as filtration media, absorption materials, and scaffolds for biomedical engineering. Looking at the polyHIPE morphology, there are three features that matter most for fluid transport, namely void size, the window size and the overall porosity. While in the previous chapter we demonstrated how to control the HIPE and polyHIPE structure (void size and its distribution), yet we haven’t quantified the effect of the structure of polyHIPE foams on their application that involves liquid transport within the foam. Before we dive into this aspect, it is worthwhile pointing out that while morphology of the porous media is a key parameter that affects the flow inside the foam, it is not the only one that matters.

Interactions of the flow fluid with the foam material, for example wetting, will also have a significant influence on applications that involve capillary suction. Isolating the influence of wetting from foam morphological properties is crucial for better understanding the transport process and for future material study. Experiments were carried out to investigate the effect of voids, window size and porosity on two type of flow regimes: the first type is

Darcy’s flow where water flows through the foam driven by pressure gradient and the 63

foams were saturated all the time; the second flow is spontaneous imbibition where liquid got sucked into the foam replacing the air by capillary pressure. While the first flow type is dominated primarily by foam morphology since it is pre-saturated and the amount of liquid-foam interaction due to surface wetting is very limited, the second flow type is dominated by both foam morphology and surface wetting.

3.2 Introduction

3.2.1 Flow in porous media

The study of flows in porous media is important for the petroleum industry, medical devices, agriculture, and commercial products like sorbents. An interesting case of the existence of capillary flow phenomena in everyday life is the wetting of cereals in a milk bowl. The freeze drying of aqueous dispersion of cereal fibers generates microscale pores into which the milk will be sucked. The length scale of channels in porous media ranges from millimeter to micrometer or even into the nanometer scale. The dynamics of flow in these channels are affected by the interplay of the fluid with the confining solid boundaries.

The flow in a capillary was studied by various scientists [58-60] nearly a century ago. One work which is mostly appreciated is “the dynamics of capillary flow” by Washburn[60]. In this work, an analytical equation was derived based on Hagen-Poiseuille’s law and arrived at the derivative form of the following equation:

푑푙 ∑ 푃 = (푟4 + 4휀푟3) (0-1) 푑푡 8푟2휂푙

Where 푙 is the length of the column of liquid in the capillary tube at the time 푡, ∑ 푃 is the total effective pressure which is acting to force the liquid through the capillary, 푟 is the radius of the capillary, 휂 is the liquid viscosity and 휀 its coefficient of slip. In the vertical capillary setup, ∑ 푃 the total driving pressure involves the hydrostatic pressure and the capillary pressure. From an energy point of view, the driving force involves the energy related to surface tension (capillary pressure) and hydrostatic pressure (if no pressure head, gravity is the acting force) and the counter forces are viscous loss and the loss due to inertia.

훾 In the case of small capillary radius (r ≪ 푡ℎ푒 푐푎푝𝑖푙푙푎푟푦 푙푒푛푔푡ℎ) and length (or height √휌푔 in the case of vertical) relatively short compared with equilibrium capillary height (h ≪

2훾 cos 휃 ), the influence of gravity and inertia can be neglected [60-63]. In this special 휌푔푟 scenario, the Washburn equation can be simplified in the form of:

훾 cos 휃 푙2 = ( ) 푟푡 (0-2) 2휂 where γ is the liquid surface tension and cosθ its contact angle with the solid wall. This equation has been proved effective in modelling the imbibition in various cases [62, 64].

Several works have talked about the limitation of this simple expression and proposed criteria of applying the simplified Lucas-Washburn (L-W) analysis in the presence of gravity and/or inertia [63, 65-67]. However as mentioned in Washburn’s work he already gave an explicit expression of the imbibition in the presence of gravity and inertia. What he didn’t talk about is the dynamics of liquid wetting/spreading on the capillary surface and how the roughness of the surface would affect the flow.

Traditional capillary tube materials are made from metal, glass and natural capillaries are usually a combination of materials such as mineral rocks and soils. Such materials have very large surface energy normally in the order of hundreds of mJ/m2, higher than that of water and most organic solvents. Table 0.1 lists the surface tension and energy values for some common solvents and materials.

Table 0.1. Surface tension and surface energy of common liquids and solid surfaces.

Surface energy Solvent Surface tension (mJ/m2) Solid surfaces (mJ/m2)

Water 72 silicon 1240[68]

PDMS oil 21 Glass 310[69]

Isopropanol 23 Polymethylmethacrylate 41

Polytetrafluoroethylene Ethanol 22 20 (Teflon)

To lower the surface free energy, liquids with smaller surface tension will spread on the material surface and the rate of spreading depends on the viscosity of the fluid and the surface morphology of the solid. More and more capillary devices nowadays are made of synthetic polymer materials like PDMS, PMMA or cellulose. Wetting of such surfaces by water is energetically unfavorable since the surface energy of these polymers are normally below that of water (72 mJ/m2). However, wetting of capillaries made from polymers can still be achieved by lowering water surface tension (by surfactant) or increasing the surface energy of the polymer (plasma or chemical modification). The surface coverage by 66

surfactant has a significant effect on the wetting dynamics. An overall static contact angle isn’t sufficient for the modelling of flow in such capillaries and errors can be significant due to contact angle hysteresis [70, 71]. Another cause of deviation from Washburn capillary behavior is the surface roughness of the capillaries. Except for the glass or silicon wafer, the surface of most of the materials are both chemically and geometrically heterogeneous[72].

3.2.2 Substrate roughness effect

For solid surfaces, the dynamics of wetting[73] (spreading of a droplet) are driven by capillarity and gravity. Theoretical calculations were carried out on smooth, plane surfaces for both capillary dominant (small drops) and gravity dominant (large drops) cases[74, 75].

푡 푅~V3⁄10(훾 )1⁄10 (0-3) 휂 for small drops and

휌푔푡 R~V3⁄8( )1⁄8 (0-4) 휂 for large drops, where V, 훾, 휂, 휌 represent drop volume, surface tension, viscosity and density, respectively. These models did not include the interaction between liquid and the substrate thus assuming a smooth surface and complete wetting scenario. Later on it was found that the good agreement of these models with experimental data resulted from a thin film of liquid that always precedes the drop spreading frontal. These calculations break for rough surfaces and there exists a dynamic contact angle that is a local variable. The first experiments that explored the effect of roughness were done by Cazabat and Stuart[76].

Glass surface with various roughness was adopted as the substrate and silicon oil was chosen as the model liquid due to the excellent wetting (or complete wetting) and various selection of viscosities. Very interesting experimental results were presented in that article.

At the very beginning of the spreading process, the behavior for rough surfaces is similar to that for smooth surfaces, which shows R scales with t1/8. Then as the drop spreads, part of the liquid (rims of the spherical cap) spreads in the rough grooves of the surface with a power law index ranging from 0.25 to 0.40, while a spherical cap remains in the center that continue to follow the 0.128 scaling power law. At the end, all silicon oil was consumed by the troughs and grooves and the spreading rate slows down. The behavior in the rims was explained as the capillary flow of liquid through the rough surface while the spherical cap serves as the reservoir. So as mentioned previously, for such capillary flow the length of the wetting front scales with t0.5. The outcome of the two processes results in the power law index ranging from 0.25 to 0.40. Thus, surface roughness seems to increase the spreading rate of the liquid drop and this effect will be more evident as the apparent contact angle becomes small.

3.2.3 Roughness effect on micro channels

Various experimental work and modelling have been carried out in capillary channels in the scale of submicron or nanometer [77-81]. Sbragaglia talked about the effect of hydrophobicity and roughness on the global mass flow rate in the microchannel. It was found that the existence of nano/micro grooves will reduce the drag and thus increase the flow rate due to the trapped air within the roughness layer[82]. In the imbibition of a glass 68

porous media consisting of nanopores in the range of 5 nm, Gruener found that the experiment result was lower than expected by the L-W analysis and they attributed this reduction to the absorbed water vapor layer and thus the reduced capillary radius[78].

Haneveld fabricated nano channels in the order of 10 nm or less by controlled etching of the silicon wafer; the reduction of Washburn coefficient was found to increase by a factor of 1.6 when the channel height decreased from 47nm to 5nm. Even the exact cause for this reduction was not clear, they attributed this effect to both the elastic deformation of the channel under the capillary pressure and the increased local viscosity due to ordering or layering of the liquid in the close proximity of the channel walls[80]. Shen studied the effect of roughness on nano-capillary rise assuming a porous layer on the capillary wall composed of cylindrical posts. Such roughness will cause increasing of the viscous force and the proposed model predicts the imbibition front scales with t1/2 but the Washburn coefficient was smaller due to the increased friction[79].

3.2.4 Wetting

In the case of partial wetting, a finite contact angle remains after the drop and substrate surface reached equilibrium as shown in Figure 0-1. Young and Laplace were the first to study the wetting phenomena and attributed the equilibrium contact angle cos 휃푒 to the balance of solid-vapor interface energy 훾푆푉 , solid-liquid interface energy 훾푠푙 and the liquid-vapor interface energy (surface tension) γ.

훾푆푉 − 훾푠푙 − γ cos 휃푒 = 0 (0-5)

This generic expression of contact angle works fine in the far field of the three phase contact line. When looking closely at the three phase contact line, the structure is much more complicated than what Young considered[73] as shown in Figure 0-2. Based on

Young’s expression, complete wetting (휃푒=0) will be achieved when 훾푆푉 − 훾푠푙 = γ. The spreading coefficient is defined as

푆 = 훾푆푂 − 훾푠푙 − γ (0-6)

Where 훾푆푂 is the solid surface energy under vacuum. It is different from 훾푆푉 which is the solid-vapor surface energy. If S is positive, a complete wetting will happen. It is known that water will completely wet native glass and metallic materials which usually have surface energy above 100 mJ/m2, but will only partially wet polymer surfaces which normally have surface energy about 50 mJ/m2. On the molecular level, complete wetting is achieved not because the high surface energy, but rather because the underlying solid usually has a polarizability much higher than that of the liquid. So the solid-liquid attraction is much stronger than the liquid-liquid attraction.

Figure 0-1. A small droplet in equilibrium over a horizontal surface: (a) and (b) correspond to partial wetting, the wetting is stronger in (b). (c) corresponds to complete

wetting[73].

Figure 0-2. Various types of triple phase line structure [73].

From the 1930s, wetting phenomena in a rough or porous substrate attracted increased attention in developing textiles. Wenzel[83] introduced a roughness factor which is defined as the ratio between the actual surface and the geometric surface.

푎푐푡푢푎푙 푠푢푟푓푎푐푒 푅 = roughness factor = (0-7) 푓 푔푒표푚푒푡푟푖푐 푠푢푟푓푎푐푒

Considering a droplet on a substrate which has a high surface energy, the drop will spread to lower the surface energy. The logic is that for the same increase of wetting area, a greater amount of the actual surface will be covered under the drop in the case of a rough surface and thus the droplet tends to wet the surface “more (rapidly)” and vice versa for the hydrophobic case. Then the apparent contact angle on the rough surface 휃푎 and the smooth contact angle 휃 were related by the roughness factor as: cos 휃푎 = 푅푓 cos 휃 (0-8)

Since the roughness factor is always greater than one, it magnifies the wetting behavior. In the case of water, the hydrophilic surface will become more hydrophilic (decreased contact angle) and the hydrophobic surface will increase its hydrophobicity (increased contact angle). An underlying assumption in Wenzel’s work that he didn’t point out is that the liquid is present everywhere inside the rough surface. In other words, there is no trapped air in the groves.

Cassie and Baxter[84] worked on wetting of porous surfaces where air is trapped in the pores. Let 푓1 be the area fraction of the liquid-solid interface and 푓2 be the liquid-air area fraction. Then the apparent contact angle will lie between the equilibrium contact angle of the liquid-solid (휃) and that of the liquid-air (π) as: cos 휃푎 = 푓1 cos 휃 − 푓2 (0-9) 72

Both works gained extreme popularity among subsequent researchers since their publication due to the simplicity and demonstrative ability. However, questions about the

Wenzel and the Cassie and Baxter’s work also came right after. Bartell pointed out that the experiments had shown that the contact angles of drops on surfaces containing roughness within the contact line were identical to those of smooth surfaces. It was pointed out by various authors [85-89] that the three phase contact line rather than the area beneath the drop defines the apparent contact angle.

Diimitrov[90] carried out molecular dynamic simulations of the capillary rise of a simple fluid and a polymer melt in nano-pores and found that the Lucas-Washburn equation still holds after a fairly short transient period of a few nanoseconds. For the polymer melt it was found that the flow exhibits a slip length comparable in the size with the nanotube radius.

In the same work the velocity field of the rising fluid close to the interface is not a simple diffusive spreading.

3.2.5 Fluid transport in polyHIPE foams

As mentioned before, polyHIPE foams not only have high porosity but also have good interconnectivity (via windows) between neighboring voids. Recent studies have demonstrated the potential of polyHIPEs as scaffolds in tissue engineering [91-93], oil spill recovery[94], and as supports or membranes in separations and absorption[95]. Typical applications of polyHIPEs involve fluid transport within the porous structure, which depends on two families of parameters. One relates to the morphological properties of polyHIPEs (i.e., porosity, void size and interconnect/window size and size distribution) 73

while the other relates to the interaction of polyHIPE material with the fluid (the wetting of the material by the fluid)[96-98]. For separations-related applications, morphological properties are usually more important than interfacial interactions and wettability because the polyHIPE is typically saturated with fluid, and the flow will be one phase. However, for applications that involve two-phase flow such as in oil recovery and absorbing where there may be multiple fluid phases, both the polyHIPE morphology and polymer-fluid interactions matter.

It has been shown that the morphological properties of the polyHIPEs can be controlled by the HIPE emulsification process [99]. For example, in a water-in-oil emulsion, the effect of water-to-oil phase ratio[16], emulsifier content[22] and shearing conditions on the final polyHIPE foam have been investigated in various studies. The polyHIPE structure can be further tuned after polymerization by methods like hyper-crosslinking [100]. Although polyHIPEs with various structures have been fabricated, only a few studies[101] investigated fluid transport in the Darcy’s flow regime and therefore understanding of the influence of polyHIPE morphological properties on fluid transport behavior in other than

Darcy’s flow (e.g., in capillary imbibition) is incomplete. Furthermore, to the best of our knowledge the effect of polyHIPE surface wettability on fluid transport behavior has not been reported. Elucidating these fundamental aspects of transport behavior within polyHIPEs form the objectives of this study.

Through controlling morphological parameters of polyHIPEs, this study provides fundamental understanding of the fluid flow behavior in polyHIPEs in Darcy’s flow and capillary imbibition in which the Washburn-Lucas analysis is typically applied to

investigate the fine structure of porous medium [62]. In addition, the interaction of the polyHIPE with infiltrating liquids was controlled through the use of emulsifiers and thus capillary imbibition under wetting or partial wetting conditions were investigated.

3.3 Experimental

3.3.1 Materials

The oil phase for the emulsions was prepared from a mixture of 2-ethylhexyl acrylate (2-

EHA), 2-ethylhexyl methacrylate (2-EHMA) and ethylene glycol dimethacrylate

(EGDMA) (which acts as the crosslinker) in a ratio of 2:2:1. The monomers were purchased from Sigma-Aldrich and used as received. A 7:1 mixture of ionic (polyglycerol succinate, PGS) and nonionic (ditallow dimethyl ammonium sulfate, DTDMAMS) surfactants was used at a total concentration of 7 wt%. Deionized water, purified to a conductivity below 15 µS/m was used for the aqueous phase. Sodium chloride (NaCl, 2 wt%) was added to the aqueous phase, as was the initiator sodium persulfate (NaPS, 0.33 wt %). For preparation of non-foamed polymers (used for testing the wetting of liquids on the polymer), the oil-phase initiator benzoyl peroxide (BPO) was used. Both types of initiator were purchased from Sigma-Aldrich and used as received. Silicone oils for the imbibition tests were kindly supplied by Dow-Corning and used as received. Surfactant

Aerosol-OT (sodium bis(2-ethylhexyl) sulfosuccinate) was used as received.

3.3.2 Emulsion preparation

The emulsion was prepared using the propeller-mixer setup described in detail in the previous chapter. In brief, emulsions were prepared using bench-top batch process in which the temperature of the system was maintained at 50 oC using a heating jacket. The aqueous phase, preheated to 50 C was added into the oil phase dropwise by hand and mixed utilizing an overhead mixer. Emulsion samples were collected into centrifuge tubes and polymerized at 85 oC for 4 hr. With the aim of varying the emulsion droplet size, the shearing time was varied from 5 min to 30 min. Since both the shearing time and aqueous- to-oil-phase ratio were varied, the samples were coded in the same format RxxMxx as in the previous chapter where R indicates the aqueous-to-oil-phase ratio and M gives the mixing time in minutes. For example, R19M10 represents the emulsion/foam having an aqueous-to-oil-phase ratio of 19 prepared by mixing for 10 min.

3.3.3 Foam morphology

Foam morphologies were observed using a JEOL JSM-6510LV scanning electron microscope (SEM). The voltage was adjusted between 15 and 30 kV to get proper images for the window sizes and the void sizes separately. The samples were previously washed with deionized water-isopropanol-deionized water by Soxhlet extraction for 24 hr each, and subsequently freeze-dried in order to avoid any deformation of the foam morphology.

The samples were cut with a razor blade and were subsequently coated by depositing a thin layer of gold using a Hummer 6.2 Anatech Ltd. sputter system in a nitrogen atmosphere.

To obtain information on the void and interconnecting window diameters, three SEM micrographs with different magnifications were used for each foam sample and the image processed using ImageJ software. At least 300 voids/windows were sampled for each micrograph. The void diameter was characterized in terms of both the number-average diameter (D10) and Sauter-mean diameter (D32) according to:

∑ 푛푖푑푖 퐷10 = (0-10) ∑ 푛푖

3 ∑ 푛푖푑푖 퐷32 = 2 (0-11) ∑ 푛푖푑푖 where 푑푖 are the individual void/window diameter interpreted from the SEM micrographs and 푛푖 is the number of that specific void or window size in the micrograph. The Sauter- mean diameter gives more weight to the larger voids within the poly(HIPE) foams, and this measure is commonly used to characterize emulsions. The breadth of the distribution of void or window sizes in the polyHIPE is characterized by the polydispersity index (PDI) defined by:

푃퐷퐼 = 퐷32⁄퐷10 (0-12)

In order to interpret SEM micrographs properly, it is necessary to employ a statistical correction because the two-dimensional images of the voids being analyzed may show a cross-section at a random position relative to the center of the void. As shown in Figure

0-3, assuming a random cut with a distance 푟푖 from the center of the “sperical” void, the corresponding radius seen in the image is 푟푠. The assumption of a random cut through the radius promises the equal probability of 푟푖 between 0 and 푅0, thus the expexctation of the average diameter from a random cut is given by:

푅0 √ 2 2 ∫0 푅0 −푟푖 푑푟푖 휋 E(푟 ) = = 푅 (0-13) 푠 푅0 4 0 ∫0 푑푟푖

Thus, the most likely diameter of a void that appears in the SEM micrograph to have diameter dSEM is di = 4dSEM/.

Figure 0-3. Illustration for foam void size correction.

The correction factor for the the window size is different from that for the void size. As shown in Figure 0-4, asuming the distribution of windows is isotropic around the void surface, the projection of a circular window onto the 2-D SEM image will be a ellipse and the projected area is given by:

퐴푝푟표푗푒푐푡푖표푛 = 푐표푠휃퐴푤푖푛푑표푤 (0-14)

휋 where cosθ is defined as the ratio of minor and major axis of the ellipse and θ ∈ [0, ]. 2

The expectation of the window area is thus given by:

E(퐴푝푟표푗푒푐푡푖표푛) = E(푐표푠휃)퐴푤푖푛푑표푤 (0-15)

휋 2 ∫0 푐표푠휃푑휃 2 E(푐표푠휃) = 휋 = (0-16) 2 휋 ∫0 푑휃

The window size is computed as the radius of the circle that has equivalent area to the window seen in the image. Thus, the most likely size of the window (Dw) is related to the

apparent window size (Dapp,w) from the SEM micrograph, by DDw app, w /2. 78

Figure 0-4. Illustration for foam window size correction.

3.3.4 Relative foam density

The density of the various polyHIPE foams was measured according to a standard procedure (i.e. ISO 845), while the relative density was calculated by dividing this value by the density of the unfoamed sheets of polymer (obtained by curing the oil at 85 oC for

12 hr).

3.3.5 Foam mechanical properties

In order to determine the degree to which the polyHIPE deformed or compressed in the fluid flow experiments, the Young’s modulus of the foams was measured from a stress- strain curve in uniaxial compression test using an Instron 1101 platform operated at 1

mm/min. Foam samples were cut into 25-mm diameter, 10-mm thick discs for mechanical testing. The modulus was extracted from the linear region of the stress-strain curve.

3.3.6 Permeability measurement

The water flux passing through thin polyHIPE foams was measured under various applied pressure drops. For this purpose, polyHIPE foams were cut into thin (1-3 mm thick) discs.

Samples to be tested were then clamped between a glass filtration set, as is depicted schematically in Figure 0-5. The bottom piece of the filtration set is a fritted-glass base

(4.8 mm in thickness) and the top piece is a glass tube filled with water. The pressure drop across the foam was achieved by using a range of heights (0.1 - 1.5 m) of the water column in the tube and a funnel with a large cross-sectional area to maintain a constant water- column height. Water was replenished into the funnel at a certain time interval to maintain a uniform driving pressure. The effective cross-sectional area subjected to flow was equal to the fritted-glass base area, which was 2.2 cm2. At least three different pressure gradients were selected for each foam sample and data from three foams were averaged for each flow condition. The pressure applied to the foams was selected according to the foam compression stress-strain curve and under all the pressures the foam was compressed to no more than 5% strain. The foams were saturated with water throughout the process. Water fluxes were plotted against pressure drops across the foam and the slope of the curve was extracted for the permeability calculation.

Figure 0-5. Schematic illustration of permeability measurement

The overall permeability for the complete experimental assembly is given by equation

퐀퐊 ∆퐏 퐕̇ = 퐭 (0-17) 훍퐋퐭

where 푉̇ is the observed volumetric flow rate, 퐴 is the flow cross section area, Kt is the overall permeability, ∆푃 is the pressure drop across the foam, μ is the liquid viscosity and

퐿푡 is the overall length of the flow path through the foam and fritted-glass support.

The permeability for the fritted-glass member (Kg) was measured in a separate experiment

-13 2 without foam present, which resulted in Kg = 132.0 10 m . This value is at least one order-of-magnitude larger than the foam permeability (Kf). In an experiment involving foam, since the flow passes through the foam and fritted-glass support in series, the various permeabilities are related to the known values of foam thickness (Lf) and glass thickness

(Lg) through:

퐿 퐿 퐿 푡 = 푓 + 푔 (0-18) 퐾푡 퐾푓 퐾푔

Once Kf is determined, the flow channels within the polyHIPE can be associated with a characteristic hydrodynamic radius (r) calculated from[102] 81

∅푟2 ∅ 퐾푓 = , τ = 1 (0-19) 8휏 1−(1−∅) ⁄3 where ∅ is the porosity and 휏 is the tortuosity factor calculated from the porosity.[103]

Since the foams were fully saturated with the DI water before and during the test, the results of the permeability experiments provide information on only the foam morphological properties.

3.3.7 Spontaneous imbibition measurement

The uptake of liquid into polyHIPE foams by capillary action was monitored using a home- built setup illustrated schematically in Figure 0-6. PolyHIPE foams were cut into 25 mm diameter and 10 mm (or 5 mm for some samples) height discs. These were placed into open-ended plastic (machined from 50 ml centrifuge tubes) sleeves and then held facing downward vertically to a stand resting on a balance (Mettler Toledo). Subsequently, the bottom of the foam sample was brought into contact with the pool of test liquid. The accumulated weight of the penetrating liquid was monitored by the balance and recorded by computer. The raw liquid-uptake data was corrected for buoyancy effect from the reservoir of test liquid. The imbibition weight was then plotted against the square root of time and the slope of the linear region from the curve was extracted to calculate the characteristic capillary diameter. For the silicone oil imbibition test, the foams were washed separately with DI water and isopropyl alcohol (IPA) for 24 hr by Soxhlation to strip off any residual electrolyte and surfactants and then dried in convection oven @ 65 oC for 24 hr. For water imbibition tests, additional surfactant was added to the polyHIPE

foams by soaking them with various concentration of PGS (in IPA) solutions followed by squeezing. This process was repeated three times and subsequently the squeezed foams were dried under vacuum at room temperature. The physical properties of the silicone oils used in this study were measured and compared with the manufacturer-supplied data.

Figure 0-6. Schematic illustration of spontaneous imbibition measurement.

3.3.8 Surfactant coating

The foams were washed with DI water and IPA by Soxhlet extraction for 12 hr each. After drying in oven, the foams were soaked in surfactant aqueous solutions of various concentrations. Then the foams were squeezed under vacuum. The amount of solution left inside the foams was measured and compared with the neat dry foam. The coating concentration was calculated based on the dry foam weight before and after soaking.

3.4 Results and Discussion

3.4.1 Variation of foam morphology

In order to study the morphological contribution to transport processes of the foams, morphology was tuned by controlling the total mixing time during emulsification in the propeller mixer setup. Figure 0-7 and Figure 0-8 show typical morphology of polyHIPE foams templated from the HIPEs. As shown in Table 0.2, polyHIPE foams with average void sizes from 36.2 µm to 13.9 µm were achieved. The average window size was also found to be reduced from 9.4 µm to 4.2 µm as the shearing time for the emulsion was increased. The ratio of average void to window diameter remained very similar for all the foams at around 3.3.

Figure 0-7. Emulsion droplet morphology for various mixing time: (a) R19M05, (b)

R19M10, (c) R19M30 scale bar is 100µm.

Figure 0-8. PolyHIPE void and window morphology.

Table 0.2: Processing parameters, morphological properties and mechanical properties

for polyHIPE foams.

D32,v D32,w Sample t(min) PDI PDI E (kPa) S (kPa)

(µm) (µm)

R19M05 5 36.2 ± 2.3 1.3 9.4 ± 0.7 1.2 280 ± 11 29.4 ± 1.0

R19M10 10 22.4 ± 3.4 1.2 6.9 ± 0.5 1.2 260 ± 35 32.7 ± 0.4

R19M15 15 16.5 ± 0.9 1.1 6.4 ± 0.6 1.1 290 ± 7 34.2 ± 0.3

R19M30 30 13.9 ± 0.9 1.2 4.2 ± 0.4 1.2 290 ± 15 37.3 ± 0.2

Here: t = emulsion mixing time; = Sauter-mean diameter of voids; = Suter-mean diameter of windows; PDI = polydispersity index; E = Young’s modulus of the dry foam;

S= yield strength of the dry foam.

3.4.2 Foams mechanical properties

The compression mechanical properties of the foams were evaluated for the purpose of finding a suitable testing condition in the study of foam’s permeability under various cross-

foam pressure drop. The stress-strain curves for all foams were plotted in Figure 0-9, the slope below 10% strain was calculated as the foams compression modulus and the deflection point between 10% and 20% strain was defined as the yielding point as shown in Figure 0-10. Both dry and wet foams mechanical properties were summarized in Figure

0-11 and Figure 0-12. Wet foam properties were tested while saturated with water. It was found that the foams modulus was about 290 kPa and 350 kPa for dry and wet foams. When the foams saturated with water were compressed, the confinement of water trying to escape was believed to contribute to the increased wet modulus.

30 01 min 20 06 min 11 min Stress (KPa) Stress 16 min 10 31 min

0.0 0.1 0.2 0.3 0.4 0.5 Strain (mm/mm)

Figure 0-9. R19M05-R19M30 stress-strain curve.

100

Stress (KPa)

0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Strain (mm/mm)

Figure 0-10. Foam yielding point defined as deflection point in the stress strain curve.

Wet 500 Dry

400

300

200

Modulus (KPa) Modulus

100

0 1 min 6 min 11 min 16 min 31 min Mxing Time (min)

Figure 0-11. Wet/dry foam modulus.

40 Wet Dry 35

Strength (KPa) Strength

0 1 min 6 min 11 min 16 min 31 min Mixing Time (min)

Figure 0-12. Wet and dry foam yield strength.

3.4.3 Darcy’s flow through polyHIPE foams

A typical set of results for the water flux as a function of pressure drop across the foam is shown in Figure 0-13. The pressure drop across the foams was carefully maintained below the foams’ yielding strength thus the foams’ deformation was well controlled at a limited level. A similar linearity was found for all foams tested in this work, and thus Darcy’s law applies to the polyHIPE foams. Table 0.3 summarizes the measured foam permeability and the hydrodynamic diameter (DK) calculated from the permeability through equation

10. Also included in Table 0.3 is information on the window size within the foam as determined from the SEM micrograph analysis (D32,w). Note the excellent agreement between DK and D32,w. This consistency confirms the validity of SEM image analysis

method used in this work and demonstrates that the critical morphological parameter controlling Darcy flow in polyHIPEs is the window size.

Table 0.3: Foam permeability, characteristic hydrodynamic diameter from the

permeability tests, and window size by SEM analysis.

Permeability DK (µm) D32,w (µm) Foam (× 10−13 푚2 ) Permeability SEM

R10M05 16.9 ± 0.5 9.2 ± 0.4 9.4 ± 0.7

R19M10 7.5 ± 0.4 6.1 ± 0.2 6.9 ± 0.5

R19M15 4.8 ± 0.2 5.2 ± 0.3 6.4 ± 0.6

R19M30 3.0 ± 0.3 3.8 ± 0.2 4.2 ± 0.4

7x10-7

6x10-7

-7 /s) 5x10

4x10-7

3x10-7

Flow rate (m Flow rate 2x10-7

1x10-7

0 0 2000 4000 6000 8000 10000 Pressure drop (Pa)

Figure 0-13. A typical curve from permeability measurement. Foam sample was

R19M10.

3.4.4 Spontaneous imbibition in polyHIPEs

The imbibition of liquid into a porous medium is characterized by a penetration distance h that scales with the square root of time ~ t1/2 according to the well-known Lucas-

Washburn (L-W) law[59, 60] (although Bell and Cameron[58] studied this phenomenon earlier). In the present work, two scenarios were studied. One is the case of perfect wetting in which the liquid (silicone oil) interacts well with the struts within the polyHIPE, while the other is the partial-wetting case in which the wetting of the liquid (water) on the struts of the polyHIPE is controlled by coating with a surfactant.

A scaling analysis demonstrates the expected applicability of the Lucas-Washburn (L-W) analysis to the imbibition process. The L-W analysis has been reported to be inaccurate when either gravity[66] or fluid inertia plays an important role[63] in the imbibition process. The dimensionless groups relevant to capillary imbibition are defined in

Equations 3-20 ̶3-23. Table 0.4 shows definitions and estimates of the relevant physical parameters. The Bond number (Bo)[104] represents the relative importance of gravity to surface tension and was calculated to be on the order of 10-5 in the current experiment setup. The Weber number (We) represents the relative importance of inertia to surface tension and is calculated to be on the order of 10-7. The critical radius for inertia to be significant in capillary rise (Rc) was proposed by Hamrzoui[105] and was calculated to be

2.5 mm for the oil used, which is much larger than the typical void/window sizes in the

-3 foam. The capillary number (Ca) was calculated to be on the order of 10 which indicates that the viscous force is the major influence compared with gravity and inertia.

Table 0.4: Parameters used in the scaling analysis

Physical properties Characteristic value

Capillary diameter (D) 10 µm

Capillary length (L) 1 cm

Liquid density () 961.2 kg/m3

Liquid viscosity () 0.106 Pas

Liquid surface tension () 21 mJ/m2

Infiltration velocity (v) 1 mm/s

Acceleration of gravity (g) 9.8 m/s2

2 () air gD 5 Bo  4  10  (0-20)

2 vR 7 We  4  10  (0-21)

0.2 cos  2  2g 3  Rc  2 0.0025m g (0-22)

v 3 Ca   5  10  (0-23)

Given these results, the standard L-W analysis should be applicable to capillary imbibition within polyHIPE foams. Thus, the mass imbibed (m) is predicted to be given by:

2 2 2 2 3 2 A ( Swf  S wi ) a r eff  cos  mt 2 2 (0-24)

where  is the liquid density, A the cross sectional area,  the foam porosity, Swf the foam saturation by the infiltrating liquid in volume percent, Swi is the initial saturation,  the viscosity of the liquid,  the liquid surface tension, and  the contact angle between the liquid and solid. Also, reff is the effective average radius of the flow channels in which the imbibition occurs.

3.4.5 Silicon oil imbibition

The excellent wetting of the polymer polyHIPE foam material by the silicone oil is demonstrated by observation of the rapid spreading of a liquid drop on the flat polymer substrate (prepared by bulk polymerization of the monomer mixture) as shown in Figure

0-14. The vanishing contact angle demonstrated that silicone oil wets the polymer material perfectly. This is consistent with the fact that the silicone oil used here has a surface tension of 21 mJ/m2 (measured by the Wilhelmy plate method) while the polymer has a surface energy of 32 mJ/m2 measured by a two liquid method by Fowke’s equation and is close to reported surface energy of 30 mJ/m2[106].

Figure 0-14. Wetting of polymer substrate by silicone oil.

Typical capillary imbibition data for silicone oil into a polyHIPE foam are shown in Figure

0-15. The linear correlation between mass imbibed and the square root of time throughout the whole capillary rise process in all three silicone oils having wide range difference in viscosity indicates that the flow behavior agrees with the standard assumptions in the L-W analysis.

Figure 0-15. Spontaneous imbibition test of R40M05 using three silicone oils with 10 cst,

100 cst and 1000 cst.

In the L-W analysis, the structure of a porous media is depicted as an array of capillary tubes[107], the diameter of which determines the infiltration rates. The effective diameter for imbibition within the polyHIPE can be calculated using equation 3-14[61] from the

slope of the imbibition-rate curves. The results were found to be internally consistent between the three cases, with less than a 5% standard deviation between them.

To further test this result, polyHIPE foams with various window sizes (R10M05-M30) were tested with 100 cSt silicone oil, which was chosen to enable experiments with a duration long enough for precision, but short enough to allow for multiple trials, with individual trials typically lasting 5 to 60 min. Figure 0-16 shows the effective diameter calculated from the imbibition results, and compares those results to the effective hydrodynamic diameters determined from the permeability tests, and the mean window size from the imaging studies. Interestingly, it was found that the effective hydrodynamic diameter matches more closely the window size rather than the void size, which suggests that the region of greatest hydrodynamic resistance (the windows) within the foam governs the imbibition process.

Figure 0-16. Effective radius from spontaneous imbibition process.

The degree of saturation indicates how much of the available pore space is filled by the penetrating liquid. None of the foams tested in this study showed a 100% saturation by the silicone oil. As shown in Figure 0-17, the data suggests a correlation between saturation and the shearing time used during emulsion preparation. With an increase in the mixing time, the polydispersity of emulsion droplets decreases, as does the void and window sizes.

This leads to a more uniform penetration through the porous foam by the silicone oil.

Although the macroscopic transport during imbibition is vertically upwards, there will be transverse flow occurring due to the presence of many windows distributed around the void. This may lead to the entrapment of air (and correspondingly less than 100% saturation) as is depicted schematically in Figure 0-18. The infiltrating liquid will preferentially pass through the larger channels (i.e., those voids with larger windows). This

will eventually lead to a preferred flow path, and potentially trapping air pockets within the foam. Thus, the saturation data can be an indirect indication of the pore connectivity within the foam as experienced by the capillary driven flow.

1.00 R9(0.899) 0.95 R19(0.952) R40(0.978) 0.90

0.85

0.80

0.75

Saturation 0.70

0.65

0.60

5 10 15 20 25 30 Mixing time (min)

Figure 0-17: Foam saturation for foams with various porosity and mixing time.

Figure 0-18. Illustration of air pockets formation during imbibition process.

3.4.6 Partial wetting-water imbibition in surfactant coated foams

The partial wetting scenario was studied by DI water imbibition in foams coated with surfactant. R19M10 foam was chosen because of the convenience of processing. The schematic illustration of the coating process is shown in Figure 0-19.

Figure 0-19. Schematic demonstration of the coating procedure

Figure 0-20 shows the relationship of coating concentration in the dried foam with the surfactant concentration in the initial coating solutions. Since the amount of solution left in the foam was kept relatively the same, the dried coating concentration in the foam increases linearly with initial solution concentration.

Foam concentration (wt%) 2

0 0 1 2 3 4 5 6 7 8 Solution concentration (wt%)

Figure 0-20. Coating concentration in dried foam versus initial surfactant in soaking

solutions.

Figure 0-21 shows the typical water spontaneous imbibition into R19M10 polyHIPE foams which were treated with various amounts of surfactant. The sample with no added emulsifier did not imbibe water. For all of the other samples, it was found that the slope of the imbibition curves gradually increased with time. During these experiments, it was observed that the water tends to penetrate into the foams faster near their periphery and slower near their center, as is illustrated in Figure 0-22. The uneven imbibition front was found to continue developing until it reached the top surface of the sample. To minimize this effect, the thickness of the samples used in the imbibition tests was limited to 5 mm

(sample diameter of 25 mm).

3000 0 wt% 1.5 wt% 2500 6.3 wt% 10.0 wt% 2000 14.8 wt%

1500

1000

500

Imbibition weight (mg)

0 5 10 15 20 25 Time0.5 (s0.5)

Figure 0-21. Water spontaneous imbibition inside R19M10 foams with various emulsifier

content.

Figure 0-22. Illustration for uneven penetrating front in the foam.

Figure 0-23 shows the slope of the imbibition rate curves (known as the Washburn coefficients) as a function of the added surfactant content. The imbibition rate was found to increase with the surfactant content up to around 25 wt% but then decreases for higher concentrations. Surfactant coating on strut surface was believed to be affected by first, the ability of the coating solution to wet the foam materials (a uniform coating solution layer)

and second, the amount of coated surfactant layer (at least monolayer coating) and polarity of the surfactant (affinity with water). If the coating solution only partially wets the foam interior surfaces, then a non-uniform coating layer will form and only patches of coated areas will form. In this study, the coating solution was prepared using IPA as a solvent which wets the foam materials very well, thus a uniform coating solution layer is expected to form on the struts surfaces before drying procedures.

) 300

0.5

250

200

150

100

Washburn coefficient (mg/s 0

0 10 20 30 40 50 60 Coating content (wt%)

Figure 0-23. PGS coating concentration effect on water imbibition rate.

The results in Figure 0-23 can be attributed to the non-uniformity of the surfactant deposit within the foam. Thin polymer films under 100 nm may undergo dewetting on the substrate.[108] Given the internal surface area of the foams, and assuming a uniform layer of the surfactant on all of the internal surfaces, it is possible to compute an effective film

100

thickness as a function of overall surfactant concentration. For a sample with 1.5 wt% emulsifier, a coating of 4 nm (and thus non-uniform deposits of the surfactant) would be expected. A film with such thickness will probably undergo dewetting process[109, 110].

Thus there is lack of a continuous “hydrophilic” pathway for water imbibition. Increasing the surfactant concentration at low levels of surfactant should be expected to lead to better wettability. However, for 15 wt% surfactant added, the deposited layer would be 45 nm.

At these higher surfactant concentrations, deposits may block windows within the foam, thus leading to decreased imbibition rates. This is evidenced by the observation of films formation in the windows with increasing surfactant concentration. However, the lack of exact molecular weight and structure of surfactants obstruct further exploration of the wetting enhancement.

For this purpose, surfactant Aerosol-OT was applied using the same coating principle to achieve the best wetting effect. IPA was adopted as the solvent to eliminate the wetting problem even though AOT has a surface tension about 23 mJ/m2 at the critical micelle concentration 0.2 wt%. Again, the dry coating concentration was found to increase with solution concentration as in Figure 0-24.

101

5.00

3.75

2.50

1.25

Coating concentration (wt%) concentration Coating

0.00 0.0 0.5 1.0 1.5 2.0 Solution concentration (wt%)

Figure 0-24. AOT coating concentration in dry foam against coating solution

concentration.

The effect of AOT coated layer concentration on water imbibition behavior was plotted in

Figure 0-25. The imbibition rate with coating concentration relation was plotted in Figure

0-26. At very low coating concentrations below 1.2% by dry weight, the imbibition rate barely increases. There was a sharp increase in the slope of imbibition curve above 1.9 wt% coating. Further increasing coating concentration gradually decreases imbibition rate due to the blockage of the windows as evidenced in Figure 0-27.

102

2500 0.3wt% coating 0.8wt% coating 1.2wt% coating 2000 1.9wt% coating 3.8wt% coating 23.5wt% coating 1500

1000

500

Imbibition weight (mg)

0 5 10 Time (s0.5)

Figure 0-25. Water imbibition curve on foams coated with AOT.

103

) 1400

0.5 1200

1000

800

600

400

200

Washburn coefficient (mg/s 0

0.0 2.8 22.8 26.6 Coating concentration (wt%)

Figure 0-26. Effect of AOT coating concentration on foams water imbibition rate

Washburn coefficient.

Figure 0-27. Evidence of coated emulsifiers blocking the windows.

104

Table 0.5 lists estimated number of AOT molecules/nm2 based on coating concentration

and materials properties listed in

Table 0.6. Assuming the sharp increase in imbibition rate from 1.2 wt% to 1.9 wt% coating in dry foams is due to achieving a monolayer of AOT on the strut surface, and in 2D closing packing scenario as shown in Figure 0-28, the radius of AOT molecules was estimated to be between 0.2-0.25 nm which is in the proxy of the reported value of AOT configurations as in Figure 0-29.

Table 0.5. Estimation of AOT coating configuration

AOT radius assuming close Estimated number of AOT AOT coating packing in 2D situation molecules per nm2(rounded to concentration/wt% (78.5% area)/nm integer)

0.3 1 0.5

0.8 3 0.3

1.2 5 0.25

1.9 7 0.2

Table 0.6. Mateiral properties of foam and AOT.

Properties Estimated values

Wet/dry foam mass ratio 3.1-3.8

Solution layer thickness 0.8-1.0 µm

105

2 Number of molecular/nm 2-16

Foam specific surface area 3.7±0.7 m2/g

AOT molecular weight 445 g/mol

Figure 0-28. Close packing of AOT spheres in a 2D projection.

Figure 0-29. molecular structure and proposed configuration of AOT molecules[111].

106

3.5 Conclusions

Fluid transport within polyHIPE foams was studied in two flow scenarios, Darcy’s flow and spontaneous imbibition. The connecting window was found to be the dominant characteristic dimension for both flow types. With decreasing voids polydispersity, the imbibition saturation degree was found to increase due to the suppressed liquid wetting rate difference through various window openings. Besides morphological features, the wetting of the foam by the fluid is crucial for spontaneous imbibition process. Fluids with various surface tensions were choose as the probing liquid for imbibition process. For a partial wetting liquid for the foam material, for example water, the imbibition rate relies largely on the coating of the surfactant on the struts surface. A critical coating concentration was found beyond which the wetting rate (Washburn coefficient) increases significantly.

This is found to correspond to the monolayer coverage of surfactant on the strut surface.

107

CHAPTER 4

SURFACE MODIFICATION OF POLYHIPE FOAMS

4.1 Synopsis

As prepared, the polyHIPE foam surfaces are hydrophobic and non-conductive. Enhanced compression modulus and electrical conductivity were successfully obtained by a simple coating process of functional nanoparticles, namely aqueous dispersion of cellulose nanocrystals and graphene oxide nanosheets. The improved properties were attributed to the good integrity of coating layer by the particles. The mechanical modulus and conductivity can be conveniently tuned by the varying the concentration of the particles in the coating. When strained below the yield point of the foams, the mechanical improvement and electrical conductivity are fully recoverable. However, upon straining past the yield point, the mechanical properties show a gradually decrease and the electrical conductivity behavior is influenced by the deformation history of the foams.

4.2 Introduction

To broaden the applications of polyHIPE foams, new features or functions, for example, wetting, electrical conductivity, improved mechanical performance, selective absorbing ability, etc. are needed. At present, the ways to give polyHIPE foams new features follow two basic routes, formulation design, and chemical or physical post-treatment. In formulation design, by using of additives in the emulsification step followed by curing step, alterations to the chemically bonded molecular network can give the foam desired 108

new properties. For examples, by blending in acrylate-based monomers, the glass transition temperature of a styrene/divinyl benzene polyHIPE foam decreased significantly, which enables the foam improved elasticity at room temperature[112]. Post treatment becomes increasingly attractive when the formulation design fails. For example, polyHIPE prepared from water/oil emulsion normally results in a hydrophobic foam which limits the potential related to absorbing applications[113]. To overcome this wetting problem, monomers with more polarity can be added into either the continuous or dispersed phase [114, 115]. However, the emulsion stability suffers from increased coalescence due to increased partition of monomer into the other phase. Blending conductive carbon nanomaterials like carbon nanotubes or graphene nanoplates becomes an increasingly attractive approach to impart electrical conductivity to the foams. A practical challenge is the dispersion of such nanomaterials within the polymers due to the great tendency for agglomeration. Another challenge would be the increased viscosity associated with the percolation network formed by the nanomaterials within the blend. The increased continuous oil phase viscosity (and the corresponding increase in the viscosity difference between the continuous and dispersed phases) results in a decrease in the controllability of the emulsion morphology. The good news is that the properties of conductivity and wettability are more related to foam surface properties rather than the bulk properties. Thus a surface post-treatment may satisfactorily address this goal. Methods like chemically treatment (plasma radiation, grafting) or a simple coating treatment can result in drastic modification of the foam surface properties. Herein, solution coating was adopted to impart

109

improvements in the wettability, electrical conductivity, and enhanced mechanical modulus of the polyHIPE foams.

4.3 Experimental

4.3.1 Foam preparation

The foams used in this work were prepared with propeller mixer setup, coding R19M05.

After polymerization, the foams were purified under DI water and IPA using Soxhletion.

4.3.2 Cellulose nanocrystals (CNC) preparation

CNCs were prepared by a TEMPO-oxidation method[116]. Typically, 2 g of micro cellulose crystals (MCCs) was dispersed in 200 ml DI water by stir-mixing. Then 200 mg of sodium nitrate and 32 mg of TEMPO catalyst were dissolved in the same dispersion. To initiate the oxidation reaction, 6.45 ml hypochlorite was added into the dispersion dropwise. The dispersion pH was maintained between 10.0 and 10.5 by adding 0.5 M sodium hydroxide. The reaction was kept on for 4 hr under mild stir-mixing. Then 20 ml ethanol was added to terminate the reaction. The dispersion was centrifuged and re- dispersed until the pH became neutral. Finally, the dispersion was freeze dried for subsequent procedures.

110

4.3.3 CNC dispersion preparation

The freeze dried CNCs were dispersed in DI water by ultra-sonication. First, CNCs powders were dispersed into DI water by stir mixing for 10 min at 300 rpm. The dispersion has a milky white appearance. Then the dispersion was ultra-sonicated in order to exfoliate the microcrystal bundles. TEMPO oxidation changes the primary hydroxyl group of cellulose into carboxylate group which will impart a charged surface to the CNCs. The repulsion due to charged CNCs results in an improved dispersion. With sonication, the dispersion has a clear transparent appearance and is stable for days. The viscosities of

CNCs dispersions made with this procedure increases with CNCs concentration. All CNC dispersions contain 0.5 wt% AOT surfactant the same as used in chapter 3.

4.3.4 Graphene oxide (GO) preparation

GO was prepared by a modified hummer method [117]. Here, 5 g of natural based graphite powders were weighed and placed in a beaker into which 130 ml of concentrated sulfuric acid was slowly added into the graphite powders and the beaker was kept in an ice bath.

This suspension was stirred at 200 rpm for 30 min. Subsequently, 9 g of potassium permanganate was carefully added into the suspension within one hour, the rate of adding need to be slow to prevent overheating since the oxidation is extremely exothermal. Then

130 ml DI water was added to the mixture and the temperature was kept under 40 C. The temperature was raised to 95 °C and kept at 95 °C for 1 hour. The mixture transformed from brownish to a golden color. The GO suspension was dialyzed to remove the metal

111

ions and to neutralize the ph. The GO concentration was determined by freeze drying and calculated to be 18.5 mg/ml. Various GO concentration dispersions was prepared by diluting the above prime GO dispersion with DI water and ultra-sonication.

4.3.5 Coating process

4.3.5.1 PolyHIPE foam coated with CNC

CNCs dispersions were prepared with the previous mentioned method. Surfactant

AEROSOL OT was added in to make a 0.5 wt% concentration. A piece of foam was then immersed in this dispersion, allowing the CNC particles to penetrate inside the porous structure. Then the fully wetted foam was squeezed in a vacuum compression setup to control the ratio between the CNC dispersion and neat foam. Lastly, the wet foam was dried at 65 C in an oven overnight.

4.3.5.2 PolyHIPE foam coated with GO and achieving electrical conductivity

The surfactant AOT concentration was maintained at 0.5 wt% in all GO dispersions. The coating process was the same for CNC. After drying overnight in the oven, the foams were soaked in HI aqueous solution for in situ reducing GO to graphene to achieve electrical conductivity. Finally, the foams were washed with water in soxhlation for 24 hr.

112

4.4 Results and Discussion

4.4.1 CNC-coated polyHIPE nanocomposite

Cellulose is one of the basic building blocks of nature and perhaps the most abundant bio- material on earth. As a crystalline structural polysaccharide, it has a unique hierarchical structure as shown in Figure 4-1. Fibrillated cellulose fibers or micro-fibrillated cellulose

(MFC) have been used in papermaking as well as in applications like filtration and as a thickener additive. The fibrillation process inevitably consumes large amount of energy due to the use of high pressure homogenization or grinding machines. Chemistry-assisted nano-fibrillation of cellulose is an alternative way and has been frequently applied in the lab scale. Anionically charged functional groups can be introduced on the cellulose microfibril surfaces to form strong electrostatic repulsion between cellulose microfibrils in water. Successive mechanical disintegration of the modified cellulose residue using, for example, an ultrasonic homogenizer, generates cellulose nano-crystals (CNC). Most frequently applied chemical path for the preparation of CNC is the acid hydrolysis procedure, for example, using a 64% H2SO4 at 45 °C for 1-4 hr[118-120]. CNCs are regarded as promising bio-based nanofibers as they have uniform widths, high crystallinity and high aspect ratios. Other than the traditional acid hydrolysis method to prepare CNCs,

Akira, Tsuguyuki and Hayaka proposed a chemical paths way to prepared completely individualized cellulose nanofibers. It involves using 2,2,6,6-tetramethylpiperidine-1-oxyl radical (TEMPO)-mediated oxidation under moderate aqueous conditions[121, 122].

113

Figure 4-1. Hierarchical structure of wood biomass and the characteristics of cellulose

microfibrils[116].

TEMPO-oxidized CNC normally results in good dispersion in water due to the anionically charged C6 carboxylate groups oxidized from the primary C6 hydroxyls. The CNCs aqueous dispersions prepared in this study are shown in Figure 4-2. The dispersion was stable for days and no sedimentation was observed. The micro morphology of the freeze- dried CNCs is shown in Figure 4-3. The individual fiber diameter is on the order of 5 nm while the length is around 200 nm. Figure 4-4 illustrates the morphology of the foams coated with CNCs. The entangled CNC fiber morphology was observed on the coated foams. Such a morphology is a result from the drying of CNC suspensions on top of the foam strut surface and then a dense film layer formation.

114

Figure 4-2. CNC water dispersion. 3wt%.

Figure 4-3. Freeze dried neat CNC.

115

Figure 4-4. CNC coating layer on foam struts.

Figure 4-5 indicates the final coating concentration of CNC based on dry foams has a linear relationship with CNC concentration in initial coating solutions. Figure 4-6 and Figure 4-7 show the foam compression stress-strain curves up to 75% strain and the compression modulus with CNC coating concentration. The compression modulus was found greatly increased with CNC coating. This is attributed to the robust mechanical strength of the coated CNC layer. It is reported that the CNC films prepared from drying of suspensions have a tensile strength about 4.4 GPa. After being deformed at 75%, the coating layer was disrupted such that the compression modulus dropped in the second cycle as seen in Figure

4-8.

116

CNC concentration(dry foam) (%)

0.0 0.5 1.0 1.5 2.0 2.5 3.0 CNC concentration(in solution) (%)

Figure 4-5. CNC coating concentration in dry foams vs concentration in solution.

Figure 4-6. Foams stress-strain curve as increasing CNC coating concentration.

From Figure 4-6, the linear stress-strain region ends in the proxy of about 10% strain. The foams coated with 3 wt% CNC solution were tested in 5 consecutive cycles at 10% and

20% strain respectively. From Figure 4-9 and Figure 4-10 it can be seen that under 10% end strain cycles the compression modulus barely changed; however, the modulus 117

gradually decreased to 60% of original modulus when the foams were cycled under 20% strain. Such loss of mechanical improvement is due to the disrupted CNC coating layer.

The coated layer might undergo buckling deformation under small macroscopic compression cycles (below 10% strain). But the 20% strain will lead to plastic deformation which is not reversible.

1800

1600

1400

1200

1000

800

600

Compression modulus (KPa) modulus Compression 400

200

0 2 4 6 8 10 12 CNC concentration(dry based) (%)

Figure 4-7. Foam compression modulus VS CNC coating content.

Figure 4-8. Foams 1st and 2nd cycle compression modulus after 75% strain.

118

1200 10% strain 20% strain 1000

800

600

400

Compression modulus (KPa) modulus Compression 200

0 1 2 3 4 5 Compression cycles

Figure 4-9. 3wt% CNC coated foams 5 consecutive compression cycles at 10% and 20%

strain endpoint.

1.2

)

0 10% strain 20% strain 1.0

0.8

0.6

0.4

0.2

Normalized compression modulus (E/E modulus compression Normalized 0.0 1 2 3 4 5 Compression cycles

Figure 4-10. 5 consecutive compression cycles with 10% and 20% strain endpoint

respectively, 3wt% coated foam. 119

The coating of CNC layer would result in a slightly decrease in water uptake rate as shown in Figure 4-11. This is due to the blocking of the foam windows by the CNC films as observed and discussed in chapter 3.

Figure 4-11. Water uptake rate Washburn coefficient vs CNC content.

4.4.2 Conductive polyHIPE via GO coating

Figure 4-12 shows the aqueous suspension with various GO concentrations. The GO suspension shows good dispersion due to the charged surface functional groups. With the help of AOT surfactant, the foams could be wetted by the GO suspension easily. The pristine graphite particles have average size in the range of 20 m. After modification, the

GO particles were downsized to nano-scale range due to the exfoliation of the graphite sheets after ultra-sonication.

120

Figure 4-12. GO aqueous dispersion. 1.0, 2.0, 4.0 and 8.0mg/ml

Figure 4-13 shows the polyHIPE foams soaked in GO suspensions. The foams have large window size compared with the GO particle size thus enabling good penetration of GO particles in the foam cellular structure. After 10 min in the vacuum chamber, the foams sank to the bottom indicating that they were fully saturated with the GO suspensions.

Figure 4-13. Foams soaking in GO solution. From left to right: first row 1.0 mg/ml, 2.0

mg/ml, 4.4 mg/ml; second row 8.0 mg/ml and two neat dry foams.

The GO coated foams (dried) were reduced by soaking in HI solution for 24 hr and the resulting rGO (reduced Graphene Oxide) coated foams are shown in Figure 4-14. The periphery of the foams was uniformly coated with rGO as evidenced by the uniform black

121

appearance. The foams were then cut open to examine the coating quality inside the foam.

As the GO suspension concentration increased, the interior of the foams becomes darker in color due to the increased amount of coating. However, it was found that the coating was not uniform and the color gradually became lighter from the peripheral region to the center of the foam. This is due to the efflorescence effect typically happening in porous media drying processes [123-127]. During the drying of GO coating layers, the evaporation of water results in rearrangement of GO particles in the foam and caused the gradient of

GO concentration.

Figure 4-14. rGO coated foams. 1.0, 2.0, 4.0, 8.0 mg/ml GO solution.

With GO reduced to rGO, a conductive layer was formed on the strut. The surface morphology of neat polyHIPE and rGO-coated polyHIPE are shown in Figure 4-15 and

Figure 4-16. Without coating, the surface morphology has a wave-like topology. After rGO coating, the surface became smooth and a layered morphology was evidenced by the bumps which might be resulted from the overlapped graphene sheets.

122

Figure 4-15. Neat foam strut surface.

The electrical properties of the resulting rGO coated foams were studied under uniaxial compression strain. The experimental setup is shown in Figure 4-17 and Figure 4-18. The electrical resistance changes under strain are plotted in Figure 4-19 through Figure 4-25.

Figure 4-19 shows the resistance change when the foams were first compressed to 20% strain. The decrease in resistance of the unstrained foams from 1.0 mg/ml to 8.0 mg/ml

GO coated samples indicates that increasing GO coating concentration helped improve the conductive network within the foams. It was found that the resistance for each foam increased during the compression process. This might be due to the fact that the deformation of the struts resulted in the disruption of the coated GO layers. The physical connecting regions from the overlapping rGO layers were broken during the straining, causing the electrical resistance to increase. Multiple cycles were repeated at the same 20% end strain point as shown in Figure 4-20 and several observations were made. First, within

123

each straining cycle, it can be seen that 1.0 mg/ml and 2.0 mg/ml GO coated samples show decreased resistance while for the 4.4 mg/ml and 8.0 mg/ml GO coated samples the resistance continues to increase. Second, comparing different strain cycles, the resistance was found to gradually increase with the number of cycles, indicating the gradual disruption of the conductive network. A possible explanation would be related to the condition of the coated layer. When the coated rGO layer was thin as in the case for 1.0 and 2.0 mg/ml samples, subsequent cycles do not generate new broken points in the conductive network. Though subsequent cycles would bring the broken points further apart, within each cycle, compression would bring them back closer. For the 4.4 and 8.0 mg/ml GO coated samples, the coating layers were thick. Not only each cycle will break the connecting points further apart, within each cycle, the deformation will cause the cracks to continue to grow.

Figure 4-16. rGO coating layer. 4.4mg/ml sample.

124

Figure 4-17. Foams mechanical compression setup coupled with electrical resistance

measurement.

Figure 4-18. Schematic illustration of foam electrical behavior under uniaxial

compression strain.

125

12100 2.0 mg/ml 455000 1.0 mg/ml

12000 450000

) )  445000  11900

440000 11800

Resistance( 435000 Resistance(

11700 430000

425000 11600 0.00 0.05 0.10 0.15 0.20 0.00 0.05 0.10 0.15 0.20 2760 Strain (mm/mm) 2550 Strain (mm/mm) 4.4 mg/ml 8.0 mg/ml

2740 2500

2720 2450

) )  2700  2400

2680 2350

Resistance( 2660 Resistance( 2300 2640 2250 2620 2200 0.00 0.05 0.10 0.15 0.20 0.00 0.05 0.10 0.15 0.20 Strain (mm/mm) Strain (mm/mm)

Figure 4-19. Foams resistance change on the first strain cycle to 20% endpoint.

126

508400 1.0 mg/ml-2 12700 2.0 mg/ml-2 1.0 mg/ml-3 2.0 mg/ml-3 500200 1.0 mg/ml-4 12600 2.0 mg/ml-4 1.0 mg/ml-5 2.0 mg/ml-5 492000 ) ) 12500

 

483800 12400

475600 12300

Resistance( Resistance(

467400 12200

459200 12100

12000 0.00 0.05 0.10 0.15 0.20 0.00 0.05 0.10 0.15 0.20 2880 Strain (mm/mm) 2600 Strain (mm/mm) 4.4 mg/ml-2 2860 2580 4.4 mg/ml-3 4.4 mg/ml-4 2840 2560 4.4 mg/ml-5

) )

 2820  2540

2800 2520

2780 2500

Resistance( Resistance( 8.0 mg/ml-2 2760 2480 8.0 mg/ml-3 8.0 mg/ml-4 2740 2460 8.0 mg/ml-5

2720 2440 0.00 0.05 0.10 0.15 0.20 0.00 0.05 0.10 0.15 0.20 Strain (mm/mm) Strain (mm/mm)

Figure 4-20. Resistance-strain behavior in the following cycles after first strain at 20%.

The foams were strained further to 50% endpoint. Figure 4-21 shows the resistance-strain relationship of the foams strained to 50% (but previously strained to 20%). From 0 to 20% strain, the resistance-strain behavior is alike the one in Figure 4-20. However, from 20% to 50%, the electrical resistance of the foams continued to increase. As explained previously, this is due to the further disruption of the conductive network as the samples were only pre-strained to 20%.

127

14100 1.0 mg/ml-1 2.0 mg/ml-1 560000 13800

540000 13500

) )

  520000 13200

500000 12900

Resistance( Resistance( 12600 480000

12300 460000 12000 0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5 Strain (mm/mm) 3400 Strain (mm/mm) 3400 4.4 mg/ml-1 8.0 mg/ml-1

3300 3200

) 3200 )   3000 3100

3000 2800

Resistance( Resistance( 2900 2600 2800

2700 2400 0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5 Strain (mm/mm) Strain (mm/mm)

Figure 4-21. Resistance-strain behavior when foams were further strained to 50%, first

cycle.

Again after the first strain, the samples were repeatedly strained at 50% strain for multiple cycles and the resistance-strain behavior is shown in Figure 4-22. For samples coated with

1.0 mg/ml GO, the resistance decreases with strain probably due to increasing the connection points in a sparsely coated network. When GO coating concentration increased to 2.0 mg/ml, the resistance first reached an intermediate minimum at around 15% then raised to a maximum at around 30% strain and finally dropped with strain. The first decrease in resistance from 5% to 15% strain corresponds to the linear elastic deformation of the foam. During this elastic deformation, the struts were slightly bended increasing the number of conductive connection points. The increase of resistance from 15% to 30% strain may correspond to the mechanical yielding of the foam at 15% strain. After this 128

point, the foams struts were gradually buckling and the coated layers were exfoliated and disrupted which results in the increase in electrical resistance. After 30% strain, even though the struts were continuously deformed and buckled, the more dominating effect is that at this point new connecting points were created with neighboring struts being compressed towards each other. With more GO coating, the minimum at 15% strain disappears and the resistance reaches maximum at 35% strain. The decrease of resistance after 35% strain was again a result of the dominating influence of the newly created conductive points. This effect was further proved by straining the foams to 75% strain as shown in Figure 4-23 and Figure 4-24. In summary, the foams’ electrical resistance-strain behavior relies greatly on the foams strain history, which determines the electrical connective networks configuration.

129

2.0 mg/ml-2 754000 1.0 mg/ml-2 16600 2.0 mg/ml-3 1.0 mg/ml-3 16400 2.0 mg/ml-4 728000 1.0 mg/ml-4 16200

16000

) 702000 )

  15800

676000 15600

15400 650000 15200

Resistance( Resistance(

624000 15000 14800

598000 14600

14400 0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5 3850 Strain (mm/mm) Strain (mm/mm) 3750 8.0 mg/ml-2 3800 4.4 mg/ml-2 3700 4.4 mg/ml-3 8.0 mg/ml-3 3750 4.4 mg/ml-4 3650 8.0 mg/ml-4 3700 3600

) ) 3550

 3650  3500 3600 3450 3550 3400 3500

Resistance( Resistance( 3350 3450 3300

3400 3250 3200 3350 3150 3300 0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5 Strain (mm/mm) Strain (mm/mm)

Figure 4-22. Resistance-strain behavior of following cycles of 50% strain.

130

5 8.0x10 4 1.0 mg/ml 2.1x10 2.0 mg/ml

5 7.5x10 2.0x104

5 4 ) 7.0x10 ) 1.8x10

 

6.5x105 1.7x104

4 6.0x105 1.6x10

Resistance ( Resistance ( 4 5.5x105 1.4x10

1.3x104 5.0x105

-0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Strain (mm/mm) Strain (mm/mm) 4.8x103 4.4 mg/ml 4.9x103 8.0 mg/ml

3 4.6x10 3 4.6x10

) 3 ) 4.4x10 3   4.3x10

3 4.2x10 4.0x103

3 4.0x10 3.8x103

Resistance ( Resistance (

3 3.8x10 3.5x103

3 3.6x10 3.2x103

-0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Strain (mm/mm) Strain (mm/mm)

Figure 4-23. Resistance-strain behavior for 75% strain.

131

8.0x105 20%-1.0 mg/ml 20%-2.0 mg/ml 4 50%-1.0 mg/ml 1.7x10 50%-2.0 mg/ml 7.5x105 75%-1.0 mg/ml 75%-2.0 mg/ml 4 7.0x105 1.6x10

) ) 5  6.5x10  1.5x104

6.0x105 1.4x104 5.5x105 Resistance( Resistance( 1.3x104 5.0x105

1.2x104 4.5x105

4.0x105 1.1x104 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Strain (mm/mm) Strain (mm/mm) 3800 3800

3600 3600 3400

) )

 3400  3200

3200 3000

2800 Resistance( 3000 Resistance( 20%-4.4 mg/ml 2600 20%-8.0 mg/ml 2800 50%-4.4 mg/ml 50%-8.0 mg/ml 2400 75%-4.4 mg/ml 75%-8.0 mg/ml 2600 2200 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Strain (mm/mm) Strain (mm/mm)

Figure 4-24. Summary of foams’ strain-history influence on resistance-strain behavior

when first strained at various endpoints.

132

450 GO 4500 +0.25wt% CNC 400 4000 350 3500 300 3000 250 2500 200

Resistance (Ohm) Resistance (Ohm) Resistance 2000 150 1500 100 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Strain (mm/mm) 640 Strain (mm/mm) +0.5wt% CNC +1.0wt%CNC 512 576

448 512

384 448

320 384

256

Resistance (Ohm) Resistance (Ohm) Resistance 320

192 256

128 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Strain (mm/mm) Strain (mm/mm)

Figure 4-25. Electrical resistance behavior of CNC-GO hybrid foams with compression

strain

4.5 Conclusions

The mechanical properties of polyHIPE foams can be enhanced by incorporating a CNC coating layer on the surface of the foam struts. Compression modulus was greatly increased from 200 kPa to 1.0 MPa, a five-fold improvement when coated with 12% CNC (on a dry foam basis). The improvement is gradually lost when the foam is strained above the mechanical yield point (which is about 10% strain). The electrical conductivity of the polyHIPE foams can be greatly increased by a rGO coating layer. The electrical resistance- strain behavior of the foams relies heavily on the strain history. When coated with small 133

rGO concentrations, the foams show a continuous decrease in resistance with strain. As the foams rGO coating concentration increases, the resistance was first increased before a critical strain point around 40% strain due to the growing of cracks in the conductive coating layer. After this point, the resistance drops with increasing strain due to the formation of new conductive paths at this large deformation.

134

CHAPTER 5

POLYHIPE FOAMS AS OIL SORBENT: VOLUME EXPANSION, W/O

EMULSION SELECTIVITY AND VOLUMETRIC CAPACITY

5.1 Synopsis

Oil and solvent spills result in catastrophic economic and environmental damage, threatening the eco-environment for all living species near the spill site. While many recently developed sorbents demonstrate exceptionally high absorbing capacity on a per mass basis with sorbents often capable of holding hundreds of times their own mass, the volumetric absorption capacity remains close to a one-to-one ratio, making transportation and storage of large volumes of sorbents a challenge. Herein, we fabricate a compact volume-expandable sorbent (CVES) by a high internal phase emulsion (HIPE) template method. The crosslinking density within the CVES is controlled by introducing an inert diluent in the HIPE continuous phase. The sorbents can transform from a compact state

(when dry) to an expanded intermediate state (absorbing a wetting solvent) and/or to a swollen fully expanded state (absorbing a good solvent). The CVES demonstrate excellent water/oil separation performance due to a multi-scale structure with micron scale voids and porous struts with nanoscale pores as a result of phase separation. Furthermore, the CVES has a low compression hysteresis loss and retains high oil absorption performance under cycles involving compression to remove collected oil. With the aforementioned properties,

CVES represents a new type of high performance sorbent.

135

5.2 Introduction

On January 14th, 2018, the tanker Sanchi sank in the East China Sea just 300 km east of

Shanghai, China’s largest city. The tanker leaked 136,000 metric tons of light crude oil, mostly natural gas condensate, threatening the safety of millions of people and posing an even greater threat to the local maritime eco-environment [128, 129]. This catastrophic accident, like the Deepwater Horizon oil spill in 2010, reminds us again of the urgency for high performance, low-cost, and eco-friendly oil spill containment and clean-up.

In practice there are several approaches to tackling oil spill clean-up including in-situ burning, chemical dispersants, skimming, bioremediation and sorbents [130]. When choosing the cleanup method, several factors are taken into consideration: location

(proximity to shore), nature (toxicity, viscosity and volatility), quantity, secondary environmental impacts, and cost. Using sorbents is one of the most cost effective approaches for shoreline or on-land spills [131]. Sorbents also find significant applications in general “housekeeping” functions in industrial environments, by improving safety for workers and preventing wider contamination. According to the International Tanker

Owners Pollution Federation, requirements for sorbent materials include oleophilicity and oil retention, low density (buoyancy), high saturation, strength and durability, anti- fermentation in water (stable in sea water), low cost and convenient storage and transportation. Traditional sorbent materials are usually made from natural materials, for example, perlite, kenaf, bark, sawdusk and chicken feathers [132-134]. Such materials have the advantages of low cost and abundant availability. However, natural sorbents suffer from issues like fermentation, efficiency (oil selectivity over water), capacity and 136

recyclability [135]. Synthetic polymer sorbents become increasingly attractive due to their design flexibility and better performance [136, 137]. Many efforts have been put in the directions of developing materials with high porosity/low density, good selectivity, and reusability. PDMS foams can be made using templates formed from sugar pellets which were leached out post PDMS resin curing. These exhibited excellent reusability due to the elastic nature of the material [138]. After chemical modification on a melamine sponge with silane agent, a commercially available product exhibits superhydrophobicity and a chloroform uptake capacity about 160 on a mass-per-mass basis [139]. Carbon foams, sponges [140, 141] and carbon aerogels [142] can be prepared by carbonization or sol- cryo method. Due to their exceptionally low density, uptake capacities up to 700 g/g were reported. These materials all exhibit great potentials having large gravimetric capacity for oil/chemical spill cleanup.

However, we would like to approach this need from another angle. The primary goal of current approaches is to increase porosity, thus decreasing the density of the sorbents and increasing the amount of oil that can absorbed. However, there is a practical limit to this approach. Even increasing sorbent porosity from 90% to 99.9%, a one-hundred-fold decrease in the sorbent relative density, only reduces the required sorbent volume by 9.9% when cleaning the same amount of spill. Given the low density of most organic solvents and petroleum products and the large volume of spills, applying even 99.9% porous sorbents would inevitably encounter great logistic challenges for storage and transportation of the large volumes of sorbent that may be needed. Volumetric based uptake capacity

137

(volume absorbed/volume of the sorbent) deserves attention as an important criterion in developing new sorbent materials.

A few studies have reported high performance materials that take advantage of polymers ability to expand beyond their original volume due to swelling by a solvent. The swelling ratio depends on the crosslinking density, the solvent/sorbent interaction and the electrolyte concentration (if the sorbent is charged) [143]. Polyelectrolyte gels were synthesized by

Ono[144] as super absorbents for nonpolar solvents. The lipophilic polyelectrolyte gels show a swelling ratio of 12 in hexane and 128 in dichloromethane. However, the swelling rate and mechanical properties of the gel were not discussed and it is not clear how to recover the absorbed solvent. A PDMS foam was prepared by Zhang[145] using xylene as an inert diluent, shows a volumetric uptake capacity of up to 5.5 ml/ml in gasoline. In this work, we proposed a new route for preparing oil sorbents with high volumetric capacity, fast absorption rate, good oil selectivity and excellent recovery performance. The foams were prepared using high internal phase emulsion (HIPE) as a template and introducing a non-reactive diluent (toluene) in the continuous oil phase as a way to control the foam volume expansion and to modify the surface roughness of the struts within the foam. The polyHIPE foams’ porous structure remains in a compact state (low porosity) while dry, expanding to an intermediate state (recovery of the cellular structure) in a wetting but non-swelling solvent (e.g., silicone oil or methanol), and further expands in a swelling solvent (e.g., heptane, toluene or chloroform).

At this point, it is worthwhile to clarify the various terminology used in this work in order to avoid confusion. First, the term “volume expansion” refers to the increase in foam

138

volume by any single or a combination of means. Secondly, “swelling” refers to the volume increase due to the chemical interaction between the foam struts and a solvent.

Lastly, “recovery from the compact state” means the volume increase due to the opening of the porous structure from a collapsed state.

The CVES foams developed in this study show excellent selectivity of oil over water and are capable of purifying water in oil emulsions. Moreover, the CVES foams exhibit good recyclability and low mechanical hysteresis over multiple absorbing/desorbing cycles and the overall uptake capacity remains basically unchanged.

5.3 Experimental

5.3.1 Materials

2-ethylhexyl acrylates (EHA), 2-ethylhexyl methacrylates (EHMA), ethylene glycol dimethacrylate (EGDMA) and Span 80, as well as sodium chloride, sodium persulfate and methane iodide were purchased from Sigma-Aldridge and used without further treatment.

Toluene, IPA, n-heptane, Chloroform and methanol were purchased from Fisher Scientific and used directly. Emulsifiers for high internal phase emulsion preparation were kindly provided by P&G.

5.3.2 Foam preparation

PolyHIPE foams are prepared by the propeller mixer setup described in Chapter 2. In the emulsion preparation procedure, the continuous oil phase consists of the monomers EHA 139

and EHMA with EGDMA as crosslinker in a ratio of 40/40/20. Then inert diluent toluene was added into the monomer mixture, and emulsifiers (PGS:DTDMAMS=10:1) were also dissolved into the mixture. The toluene concentration in the oil phase varies from 0% to

60% by weight and the emulsifiers concentration is maintained at 6% by weight for all samples. The aqueous phase consisting of 2 wt% NaCl and 0.3 wt% NaPS was added dropwise into the oil phase for 2.5 min to reach a 19:1 ratio, and then mixing was continued for 1 min. After emulsification, the emulsions were transferred to 50 ml centrifuge tubes and cured at 65 °C for 24 hr. After polymerization, foams were cut into cylindrical discs and washed with DI and IPA for 12 hr each in a Soxhlet chamber and then dried in an 85

°C oven overnight. Additional R19T60 samples were put back into DI water after IPA washing and then freeze dried. In this work we adopt a code system using RXXTXX in which R denotes the total weight ratio of aqueous phase to oil phase while T denotes the weight concentration of toluene in the oil phase. For example, R19T40 stands for a foam made from a 19 to 1 water to oil ratio emulsion and 40 wt% toluene in the continuous oil phase. Adding the inert diluent did not considerably affect the emulsification as the emulsions remain stable throughout the preparation process.

5.3.3 Solid polymer preparation

Solid polymers were prepared from the same composition as the foams continuous oil phase. For this purpose, 0.5 wt% benzoyl peroxide was added to the oil mixture as initiator.

The mixture was cured in glass tubes in an oil bath at 85 °C for 4 hr.

140

5.3.4 Swelling test and crosslinking density

Dry foams volume expansion behavior was studied using cyclohexane as a good solvent.

Foams were placed in a beaker containing the solvent then transferred to a vacuum chamber until the foam sank to the bottom of the beaker. Applying vacuum ensures that the foams were fully saturated with liquid with no air trapped inside. The foams were found to expand instantly (in a few seconds) after wetting by the solvent but the volume of the foam was found to barely increase afterward. The beaker was sealed by aluminum foil and left undisturbed for 2 hr before measurement. The weight and dimensions of the foam were measured before and after the expansion, and expansion ratio was calculated as the following:

퐹표푎푚 푉표푙푢푚푒 Expansion ratio = 푤푒푡 (5-1) 퐹표푎푚 푉표푙푢푚푒푑푟푦

Swelling behavior of solid polymers was also investigated using the same procedure as the foam except that the swelling ratio was measured by gravimetric method. The polymers were cut into 5 mm pellets and kept in the solvent for 48 hr to ensure good swelling. Then the pellets were filtered in a Buchner funnel for 10 s and weighed immediately. Such a fast filtration step makes sure there is limited (if any) solvent trapped in between the pellets and with little incorporated solvent loss due to evaporation.

5.3.5 Foam volume reduction-compact ratio

Foam’s compact ratio is defined as the following:

푬풎풖풍풔풊풐풏 풗풐풍풖풎풆 풂풔 풎풂풅풆 퐂퐨퐦퐩퐚퐜퐭 퐫퐚퐭퐢퐨 = (5-2) 푭풐풂풎 풗풐풍풖풎풆 (풅풓풚) 141

Certain amount of emulsion volume was measured prior to curing and the dry foam volume was measured after DI water and IPA Soxhlation and drying at a 60 °C oven overnight.

5.3.6 Morphology and image analysis

The internal structure and morphology of the foams was characterized by field emission scanning electron microscope (FEI Helios 650). The samples were sputter coated with 5 nm thick of gold before SEM observation. Various magnification micrographs (from 200

X to 20,000 X) were taken for observation of foam void, window and strut surface morphology. The 500 X micrographs were processed with ImageJ to analyze foam window size distribution since this magnification normally generated a good sample size (over 1000 windows). The area averaged diameter was adopted to represent the window size and a factor of √휋/2 was adopted for correction of the viewing angles of the window around a void surface (see Chapter 3).

5.3.7 Foam mechanical performance

The mechanical characteristics of the foams were studied under uniaxial compression. The impact of initial diluent on foams stress-strain behavior was studied in both dry and swollen state. Each foam was tested 5 consecutive cycles. Each cycle consists of loading the foam to 75% strain followed by unloading the strain at the same strain rate. Hysteresis loss was calculated as the ratio between the area inside the hysteresis loop and the area under the loading cycle in the following equation:

142

푎푟푒푎 표푓 ℎ푦푠푡푒푟푒푠푖푠 푙표푠푠 퐻푦푠푡푒푟푒푠𝑖푠 푙표푠푠 푟푎푡𝑖표 = (5-3) 푎푟푒푎 푢푛푑푒푟 푡ℎ푒 푙표푎푑푖푛푔 푐푢푟푣푒

5.3.8 Water wettability

The advancing contact angle for water was measured by slowly increasing volume of a standing droplet from 3 µl to 10 µl. The receding contact angle was measured by withdrawing water slowly from a standing 15 µl droplet. The data was collected from 3 positions from each sample and averaged over 3 samples. The contact angle measurement was performed on both foam and solid polymer samples.

5.3.9 Oil absorption and reusability

The solvent absorption capacity of foams was measured by placing the sample onto the solvent. After 10 min the foam sample was taken out and was left in air for 30 s to get rid of the surface attached liquid. The dimensions and weight of the sample were measured before and after the sorption. The gravimetric capacity and volumetric capacity were calculated as in equation 3 and 4:

푚푤−푚푑 퐶푔 = (5-4) 푚푑

푉푠 (푚푤−푚푑)휌푓 퐶푣 = = (5-5) 푉푑 휌푠푚푑 where 푚푤 and 푚푑 represent the weight of wet and dry foams, 푉푠 and 푉푑 represent the volume of the absorbed solvent and dry foams, and 휌푠 and 휌푓 represent the density of the solvent and dry foam.

143

The performance of the foams over multiple cycles was also examined. Dry foams were placed in heptane until saturated and then the foams were placed in Buchner funnel and the funnel was cover by a rubber film. The absorbed heptane was squeezed out of the foams by connecting the Buchner funnel with vacuum (25 in Hg in vacuum). The process was repeated 10 times consecutively to examine the durability and reusability of the foams. The sorption capacity was reported normalized to the maximum sorption capacity, defined as the gravimetric capacity of heptane absorption after 2 hr of soaking.

5.3.10 Water/oil emulsion selectivity

A water-in-heptane emulsion stabilized by the surfactant Span 80 is adopted as a model system for evaluating the polyHIPE foams’ selectivity of oil over water. Typically, 15 mg

(0.1wt%) of Span 80 was dissolved into the 15 g of heptane. Then 75 mg (0.5wt%) of water

(dyed green) was added dropwise into the prepared heptane solution. The mixture was ultrasonicated for 2 min under 40% power amplitude with stirring. The emulsion droplet size distribution before and after separation was obtained from Laser Particle Counter (PC

-2000 Spectrex). The pure heptane was scanned first as the background.

5.4 Theory

5.4.1 Phase separation in free radical crosslinking copolymerization

Phase separation describes the phenomena of a homogeneous mixture (one phase) separating into two phases of distinct compositions. In the field of polymers, phase 144

separation concerns the polymer blends and polymer solution systems. The tendency for a polymer solution or blend to remain stable depends on the sign of free energy of mixing as defined in Flory-Huggins equation [146]:

∆퐹̅푚𝑖푥 = ∆푈̅푚𝑖푥 − 푇∆푆̅푚𝑖푥

휙 1−휙 = 푘푇 [ ln 휙 + ln(1 − 휙) + 휒휙(1 − 휙)] (5-6) 푁퐴 푁퐵 where k is Boltzmann constant, T absolute temperature, φ the polymer volume fraction of component A, 푁퐴 and 푁퐵 represent the degree of polymerization for components A and B.

The first two terms originate from entropic contribution which always promotes mixing while the last term has energetic origins and can oppose mixing (positive), zero in the case of ideal mixtures or promotes mixing (negative). The sign of the last term depends on the

Flory interaction parameter χ. When the components attract each other, χ is negative and promotes mixing. More often χ is found to be positive between species, for example between polymer and solvent. A mixture of polymers will be in thermodynamically equilibrium state when its free energy is the minimum of all states. Over the whole composition range (φ), the dependence of the free energy Fmix(φ) can be either convex or concave as illustrated in Figure 5-1, which determines if the free energy of the mixed state

Fmix is higher or lower than that of a separated state F0. If the system with overall composition ϕ0 is in a state with two phases, with volume fraction of A component in the

α phase ϕα and the fraction of A component in the β phase ϕβ, the relative amounts of each phase are determined from the lever rule [146]. The local stability of such mixture depends on the second derivative of free energy with respect to composition ϕ.

145

Figure 5-1 Composition dependence of free energy: (left) unstable and (right) stable.

Local stability is determined by the sign of the second derivative of free energy with

respect to composition[146].

휕2퐹 푚푖푥 < 0 푢푛푠푡푎푏푙푒 (5-7) 휕휙2

휕2퐹 푚푖푥 > 0 푙표푐푎푙푙푦 푠푡푎푏푙푒 (5-8) 휕휙2

2 휕 퐹푚푖푥 In the Fmix(ϕ) curve, by solving the inflection point where = 0 gives the spinodal 휕휙2 curve which is the boundary between the unstable and metastable regions.

2 휕 퐹푚푖푥 1 1 2 = 푘푇 [ + − 2휒] = 0 (5-9) 휕휙 푁퐴휙 푁퐵(1−휙)

The lowest point on this spinodal curve corresponds to the critical point and the solution is

√푁퐵 휙푐 = (5-10) √푁퐴+√푁퐵

ퟏ ퟏ ퟏ ퟐ 흌풄 = ( + ) (5-11) ퟐ √푵푨 √푵푩

146

For a strongly asymmetric binary system like the polymer solution (푁퐴 = 푁 and 푁퐵 = 1 ), the critical composition is low and the critical χ close to 0.5.

1 1 휙 = ≅ (5-12) 푐 √푁+1 √푁

1 1 1 1 1 휒 = + + ≅ + (5-13) 푐 2 √푁 2푁 2 √푁

Phase separation can occur not only in a polymer blend or solution, but also can appear in three-dimensional polymers during their formation, especially in the presence of diluents.

The outcome of such phase separation is either the formation of a gel or that of a dispersion from the network and separate phases. An analogy of phase separation can be made between the case of polymer solution and the free-radical copolymerization crosslinking

(FCC) process. While the polymer solution deals with the interaction of the solvent with a polymer having chain length of N units, the FCC process also can be viewed as the interaction between the solvent and a polymer having chain length of N between two adjacent crosslink points. Naturally, the crosslinking density or the average molecular weight between crosslink points are crucial for the phase separation process.

At the beginning of the polymerization, the system is a homogeneous mixture of solvent and monomers. Since the monomers usually have a low molecular weight, the system can be approximated as an ideal mixture which will remain homogeneous over a wide range of composition. As the free radical chain grows and before it reaches the gel point, the size difference between solvent/monomer becomes bigger and the mixture consists of solvent

(monomer can be treated as solvent for the polymer) and dissolved polymer chain. If the mixture phase separates before the gel point, the outcome will be dispersed polymer beads in the solvent. Passing the gel point, the mixture consists of a crosslinked polymer network, 147

solvent (including monomers) and dissolved polymer. Flory equilibrium swelling theory can be applied to this mixture and that is: the swelling solvent tries to mix into the chain between the crosslinking points while this free energy of mixing is balanced by the elastic free energy of the network and electrostatic interaction.

As long as the growing network is able to absorb all the available monomers and solvent, it will remain homogeneous. When the crosslinking reaction proceeds and the crosslinking density of the network increases, a critical point is passed where the equilibrium degree of swelling of the network in the presence of solvent equals to its degree of dilution. In other words, at this point the swollen network can no longer absorb the excess dilution due to the crosslink density, so the mixture will separate into two phases: network and separated phases.

Dusek [147] first studied the phase separation during an ongoing free radical crosslinking copolymerization (FCC) process in the presence of inert diluents. The assumptions are: the network and separated phase are in thermodynamic equilibrium, isotropy of the network and the assumptions of Flory theory of swelling equilibria and rubber elasticity hold. It was found that as the initial concentration of the diluent increases, the critical crosslink density of the network decreases. Later on, models proposed by Okay [148] take into account the distribution of soluble polymers in both phases and the volume contraction during the reaction. In comparison with experimental work using styrene and m-divinylbenzene (m-

DVB) it was found that a porous network appears after a critical diluent concentration or a crosslinker (DVB) concentration. Also as the crosslinking density decrease or the diluent concentration increases, the swelling ratio of the networks in toluene increases.

148

5.4.2 Swelling of crosslinked foam after curing

The phase separation mechanism during a FCC process was briefly discussed in the previous section. It was shown that as the initial diluent concentration increases, the crosslinking density decreases and porous network results from separated phases. By replacing the diluent (good solvent) with poor solvent by solvent exchange, the polymer chains will coil and the structure will undergo shrinkage. The swelling-deswelling phenomena is briefly introduced and discussed here.

The free energy change ∆F involved in the swelling of an initially dry pure polymeric network by a pure solvent consists of two components: the free energy of mixing ∆퐹푚, and the elastic free energy ∆퐹푒푙 which is due to the expansion of the network structure.

∆F = ∆퐹푚 + ∆퐹푒푙 (5-14)

The free energy of mixing was given in equation 5-5. The elastic free energy is given in the following equation:

2 3 ∆퐹푒푙 = (푘푇푣푒/2)(3훼푠 − 3 − ln 훼푠 ) (5-15) where 푣푒 is the effective number of chains in the network, 훼푠 is the linear swelling ratio and is related to the volumetric swelling ratio in the following fashion:

3 푉 1 훼푠 = = (5-16) 푉0 휙 where V is the swollen volume and V0 is the dry volume.

At the swelling equilibrium, we get the Flory-Rhener equation:

2 1⁄ − ln(1 − 푣) + 푣 + 휒푣 = 푉푠푁(푣 3 − 푣/2) (5-17)

149

where N is the crosslinking density (mol/m3), 푣 is the polymer volume fraction in the swelling network or inverse of volume swelling ratio, χ represents the interaction parameter and 푉푠 is the molar volume of the solvent.

5.4.3 Foam swelling ratio and polymer swelling ratio

The classical Flory-Rhener equation was developed based on solid rubber swelling experiment. In this section, it will be demonstrated that the overall volume increase of a porous foam due to swelling should be the same as the swelling ratio of the foam strut polymer. In a simplified 2-D representation of a foam which consists of square shaped voids (cell), the dimensions of the void before and after swelling are demonstrated in Figure

5-2 where δ is half the struts thickness, l the void length and λ the swelling ratio of the polymer assuming isotropic swelling condition which is the case for random crosslinked polymer foams. The porosity of the foam before and after swelling is calculated as:

푙2 Porosity before swelling 휀 = (5-18) 0 (푙+2훿)2

(휆푙)2 푙2 Porosity after swelling ε = = (5-19) (휆푙+2휆훿)2 (푙+2훿)2

150

Figure 5-2 schematic illustration of the relationship between foam and strut swelling

ratio.

It can be seen from equation 5-18 and 5-19 that the porosity of the foam does not change after swelling. Let’s use V for the volume of the swollen foam, V0 for the dry foam respectively. The foam volume increase due to swelling (foam swelling ratio) is the same as the strut polymer swelling ratio as illustrated in the following:

푉 푉(1−휀) Foam swelling ratio = = 푉0 푉0(1−휀)

푆푤표푙푙푒푛 푠푡푟푢푡 푣표푙푢푚푒 = 푑푟푦 푠푡푟푢푡 푣표푙푢푚푒

= 푠푡푟푢푡 푝표푙푦푚푒푟 푠푤푒푙푙𝑖푛푔 푟푎푡𝑖표 (5-20)

151

5.5 Results and Discussion

5.5.1 CVES fabrication.

Figure 5-3a illustrates the steps of CVES fabrication. The process consists of 3 steps: emulsification, curing, and washing/drying. After drying, there is no further chemical treatment and the CVES is ready for applications. While increasing inert diluent concentration has limited influence on the template-HIPE stability, it has a significant effect on the HIPE curing step. The R19T80 sample lost its original emulsion structure evidenced by the large amount of free water at the bottom of the tube as shown in Figure

5-4. From this point on, only T0 to T60 will be further investigated as these samples show good stability during the curing process and good mechanical integrity. The presence of inert diluent in the emulsion curing procedure has a direct impact on the free radical crosslinking copolymerization process. By altering the crosslinking condition of the elastomer, the inert diluent concentration will affect the foam morphology, swelling behavior, mechanical properties, and most importantly the performance of the solvent absorption and emulsion purification. Thus the center piece of this study is the effect of inert diluent on the crosslinking density of the struts within the foams.

152

Figure 5-3 (a) Schematic illustration of polyHIPE foam preparation; (b) Foam R19T60 in compact state(left), volume expansion in methanol due to recovery of collapsed cellular

structure(middle) and in heptane due to further swelling of the strut(right); (c) Swelling

ratio and crosslinking density with toluene content in the oil phase.

153

Figure 5-4. Inert diluent effect on polyHIPE foams curing. from left to right: R19T0,

R19T20, R19T40, R19T60, R19T80.

5.5.2 Crosslinking density, swelling behavior and foam strut micromorphology.

Based on classical equilibrium swelling theory the swelling process is considered a competition between the mixing free energy of polymer chains with the swelling solvent and the elastic free energy of the polymer network[143]. The denser the polymer chains are crosslinked, the lesser the swelling due to the greater elasticity to balance the mixing of solvent and polymer chains. The relationship between swelling ratio and crosslinking density is described by Flory and Rehner [149] as in equation:

2 1⁄ − ln(1 − 푣) + 푣 + 휒푣 = 푉푠푁(푣 3 − 푣/2) (5-17).

154

The foams swelling ratio with cyclohexane as well as the crosslinking density calculated from the Flory-Rehner theory versus initial diluent concentration are plotted in Figure 5-3c.

The solid polymer swelling behavior is shown in Figure 5-5. It was found that the greater inert diluent content, the greater the swelling ratio, and as a result the smaller the crosslinking density. The foam swelling ratio stays very close to (slightly lower than) the one of the solid polymer indicating the similarity between the two materials. During the free radical crosslinking copolymerization (FCC) in the continuous oil phase, the presence of an inert diluent serves as a swelling agent for the reacting monomers system. As the polymer network starts growing, the unreacted monomers and diluent molecules are still absorbed in the swollen network, such that the overall composition remains similar to the initial state. At the same time, the crosslinking density decreases with increasing dilution degree due to the decrease in polymer volume fraction as well as intramolecular cyclization as pointed out by Dusek [147, 148]. Such a swollen network will continue growing until the point of phase separation. In the phase separated state, there coexists a solvent (diluent) rich phase with dissolved oligomers and a swollen crosslinked network phase.

155

3.0

2.8 Solid polymer Foam

2.6

2.4

2.2

2.0

1.8

Swelling ratio 1.6

1.4

1.2

1.0 0 10 20 30 40 50 60 Toluene concentration (wt%)

Figure 5-5. Foam swelling ratio and solid polymer swelling ratio from only curing the oil

phase.

The porosity of the swollen network depends on the amount of diluent and crosslinker. If a porous morphology forms, the size of the pores (from 10 Å to 1 μm) depends on the diluent’s solvation power and the glass transition temperature, and the pore volume increases as the diluent content increases[148]. In this work, the inert diluent promoted phase separation during the FCC process. Figure 5-6 shows struts’ micro-morphology characterized by high magnification FESEM. The control sample R19T0, without diluent, has a rough, solid strut surface topology which is typical for polyHIPE foams. As the diluent concentration increases in the oil phase, the solid strut surface is transformed into a porous morphology with pores in the range of 50 nm. The size of the nanopores seems to remain in the same scale while their fraction increases with increasing initial diluent

156

concentration. The emerging of such porous morphology in the strut affects the wetting properties of the foams as discussed in a later section.

Figure 5-6 Void/window morphology (scale bar is 50 µm) and struts morphology(scale bar 1 µm for a and 2 µm for b-d ) for foams prepared with dilution concentration of: 0%

(A,a), 20% (B,b), 40% (C,c) and 60% (D,d)

5.5.3 Foam mechanical behavior and volume shrinkage mechanism

Besides the impact on swelling behavior and foams strut micromorphology, the control of crosslinking by initial diluent also has a significant impact on foam’s mechanical properties and indirectly affects the foam macroscopic morphology—volume shrinkage. The foam compression behavior is shown in Figure 5-7.

157

Figure 5-7. Foams loading/unloading compression stress-strain curve in the dry state (a)

and in the swollen state (b). The inserts are compression modulus for 5 consecutive

cycles.

For porous materials, mechanical properties depend on the strut material (polymer material) and the relative density of the foam. A classical model that describes the relationship with struts material and relative density was proposed by Gibson and

Ashby[50, 150]

퐸 휌 푓 = 퐶( 푓)2 (5-21) 퐸푠 휌푠 where 퐸 and 휌 represent compressive moduli and density respectively, subscripts 푓 and 푠 stand for foam and strut, and 퐶 is a fitting constant. From this model, it can be found that the foam modulus scales with strut modulus linearly and with relative density quadratically.

As can be seen in Figure 5-8, without initial diluent, the dry foam mechanical behavior can be predicted by this model quite well such that foams compression modulus scales indeed quadratically with the relative density. In the swollen state, the foams show a significant drop in compression modulus compared with dry state. Meanwhile increasing the diluent 158

concentration causes a gradual decrease in the foams modulus both in dry and swollen state. As discussed previously, the inert diluent will cause a decrease in strut crosslinking density, thus causes a reduction in the strut mechanical property (insert citation), which

퐸 results in a reduction in the foams’ modulus as can be expected according to equation 푓 = 퐸푠

휌 퐶( 푓)2 (5-21). 휌푠

Hysteresis behavior is observed in the stress-strain curve for both dry and the swollen foams as shown in Figure 5-7. The hysteresis loss indicates the energy dissipated during the loading/unloading cycle and is plotted in Figure 5-9. Compression hysteresis is common for rubbery materials due to their viscoelastic nature. The smaller the hysteresis loss, the less the energy is lost due to internal polymer chain friction. Since the foams were prepared with the same dimensions and were compressed at the same strain rate, the change in the hysteresis loss is attributed to the change in crosslinking density in the foams. In the dry state, the hysteresis loss was found to increase with initial dilution concentration. As initial dilution concentration increases, the molecular weight between crosslinks increase from T0 to T60. This will cause an increased viscous energy dissipation during loading and unloading cycle due to the internal friction between the polymer chains as they are being deformed [151-153]. However, in the swollen state, the hysteresis loss is found to be significantly smaller than the dry state, indicating a great decrease in the energy dissipation during deformation which can be attributed to the fact that the polymer chains are in a stretched state. In contrast to the dry state, the hysteresis loss remains in the same level within error range. This might be also attributed to the fact that the network is being swollen

159

and extended such that under the strain rate in this study it is unable to detect the small variation in the hysteresis loss. After 5 cycles loading/unloading, the swollen foams exhibit good recovery in terms of compression modulus while the dry foams suffer a gradual loss in the mechanical modulus.

Figure 5-8. Dry foam modulus VS foam relative density behavior without inert diluent.

0.9 Dry Swollen in heptane 0.8

0.7

0.6

0.5

0.4

0.3

Hysteresis loss ratio 0.2

0.1

0.0 0 10 20 30 40 50 60 Toluene concentration (wt%)

160

Figure 5-9. Hysteresis loss ratio for dry and swollen foams in the loading and unloading

compression cycle.

Looking at dry foams in Figure 5-10, it can be found that there is a gradual volume reduction (compact ratio) from R19T0 to R19T40 and there is a large shrinkage in volume for sample R19T60. The quantitative foam compact ratio can be found in Table 5-1. The volume reduction phenomena can be explained by two mechanisms: shrinking of the struts and collapse of the foam’s cellular structure. As the toluene was replaced by a non-swelling solvent (like methanol or water), the porous struts first shrink due to de-swelling. The nanopores on the struts further collapses due to the negative capillary pressure when solvent leaves the pores. Since the foam material is in the rubbery state and due to the relatively low porosity in the struts, the nanopore morphology will be recovered when the foams are fully dried[148]. With such a shrinkage in the struts, the bulk density would remain at the same level. Indeed, for T0 to T40 foams, the bulk density was only found to slightly decrease with increasing dilution concentration. Similar to the collapsing of nanopores in the strut, solvent leaving the voids will again exert a negative pressure to the foam. The competition of such a pressure and the foam’s compression moduli determines the overall collapse of the structure. When foam’s compression modulus is larger than such negative pressure, the foam overall structure will remain intact otherwise it will collapse during drying. This negative pressure, as can be estimated by capillary pressure in equation

5-22 scales with the liquid surface tension and inversely with the capillary radius of the foam:

161

ퟐ휸 퐏~ (5-22) 풓 where γ represents the evaporating liquid surface tension and r represents the capillary radius of the porous foam. In the scenario of drying from IPA (20 mJ/m2), the capillary pressure is found in the range of 4.7~3.1 kPa for the foams in this work which have capillary radius in the range of 8.4~13.0 μm. From the mechanical properties as listed in

Table 5-1, the compression modulus for swollen foams decrease from 60.4 kPa for T0 to

1.5 kPa for T60 which falls below the range of capillary pressure and thus foam T60 will undergo collapsing during drying process. This collapse in the macroscopic cellular structure will enable the foam to achieve a compact state, without external forces like vacuum. Once such a compact foam contacts common petroleum fuels or swelling solvents, it will expand to its original cellular structure and swell in a matter of seconds.

Such a compact-expansion-swelling mechanism is fully reversible between drying and swelling cycles, and the initial collapse of the structure allows for a great volume reduction of such foams for efficient storage and transportation.

Figure 5-10. Volume reduction of foam by increasing diluent content. Left: Dried

samples; right: R19T60 compact and expansion in heptane (dyed in red).

162

Table 5-1 Foams physical properties.

Bulk Compression Compression Porosity Compact Window Foam code density modulus dry modulus wet (%) ratio size (µm) (mg/cm3) (MPa) (kPa)

R9T0 104±1 90.1±0.1 1.1±0.1 1.10±0.15 200 ±20

R9T20 99±2 90.6±0.2 1.2±0.1 0.76±0.05 140±10

R9T40 98±2 90.7±0.2 1.4±0.1 0.44±0.04 60±9

R9T60 94±3 91.0±0.3 1.9±0.2 0.29±0.03 12±2

R19T0 51±1 95.2±0.1 1.1±0.1 0.29±0.04 60±6 8.4±0.1

R19T20 47±1 95.5±0.1 1.2±0.1 0.13±0.02 30±2 8.6±0.6

R19T40 44±1 95.8±0.1 1.5±0.2 0.08±0.02 9.4±1.5 11.7±1.0

R19T60 42±1 96.0±0.1 1.9±0.2 0.06±0.01 1.5±0.2 13.0±0.1 (expanded)

R19T60 244±27 76.8±2.6 11.3±0.5 NA NA NA (compact)

5.5.4 Wettability: contact angle, contact angle hysteresis and selectivity of oil over

water

The surface energy of the foams was measured by Fowkes’ theory using the two liquid

method, namely water and methylene iodide. Foam density and surface energy are listed 163

in Table 5-2. The surface energy is calculated to be 32 mJ/m2 for the crosslinked polymer material. It is much lower than 73 mJ/m2 of water and higher than most hydrocarbon oils and solvents. Thus in theory, water will not wet the foam while oils are able to completely wet the foams. PolyHIPE foams wettability is characterized by both water and oil contact angles. The oil chosen for the characterization is a PDMS oil because it will not swell the foam materials. The oil sessile drop spreads quickly on the solid material and is sucked into the foam immediately upon contact as can be found in Figure 5-11. This indicates the good oleophicility of the foam materials.

Table 5-2. Foam surface energy.

Contact angle with solid Surface Testing liquid polymer tension/energy(mJ/m2)

Water 100 73

Methylene iodide 30 50

Solid polymer 32

Figure 5-11. Wetting of polymer solid by PDMS oil

164

Figure 5-12 shows the advancing and receding water contact angle of all the foams prepared. The advancing contact angle for solid polymer and undiluted foam R19T0 are

100 and 145 respectively. However, the contact angle hysteresis is first enlarged from solid (CAH=30) to R19T0 (CAH=52.3). The increase in both the advancing and receding contact angle is due to the trapped air in the porous foam and decrease in three phase contact line because the existence of the windows which cut open the continuous struts.

The increased contact angle hysteresis can be explained as the increased friction pinning of the contact line during volume shrinking. For the foams prepared with toluene, the advancing contact angle increased from 145 to 155 while the receding contact angle increase from 95 to 125. As seen from the SEM micrographs, the strut surface shows grain-like roughness due to the shrinking of the strut during drying. Also seen are the pores in the nanoscale emerged from the phase separation due to toluene. It has been reported that the hierarchical scale in the surface morphology will decrease the energy barrier for the contact line to move which makes the depinning of the receding three phase line easier, thus increase the receding contact angle and a decrease in the contact angle hysteresis. To prove that the foam is both oleophilic and hydrophobic, dry foam was kept at the bottom of a beaker containing layered DI water and heptane. When immersed into the water, the foam was dry and water was prevented from wetting the foam by the trapped air. After the foam was released, it rose up and started absorbing heptane as soon as they contacted.

165

Figure 5-12. Water contact angles for R19 foam samples.

5.5.5 Oil absorption and recovery

PolyHIPE foam absorption was evaluated for both gravimetric (g/g) and volumetric capacity (ml/ml) using various solvents. If the porous sorbent doesn’t change its volume, the gravimetric capacity is affected by the liquid and sorbent density, the lighter the foam or the heavier the liquid is, the larger the gravimetric capacity will be. Meanwhile, the volumetric capacity is only affected by how much the available pore space was filled by the liquid. Thus while the reported gravimetric capacity is often in the order of hundreds, the volumetric capacity is merely unit in the best scenario (when it is fully saturated with the liquid) and often less than one when not fully saturated. However, when the porous 166

foam can expand its volume, both the gravimetric and volumetric capacity will increase depending on the ability of volume expansion. Both the volumetric and gravimetric absorption capacity for R19T0-R19T60 foams are plotted in Figure 5-13.

Figure 5-13. Gravimetric capacity (a) and volumetric capacity (b) of various solvent by

foams with increasing inert dilution content.

For non-swelling solvent absorbing like methanol and PDMS, the gravimetric and volumetric capacity is slightly less than 19 and 1 respectively for all foams. The increase in volumetric capacity for R19T60 sample is due to the expansion of the collapsed macroscopic voids. When absorbing a good solvent capable of swelling the foam materials, the foams struts volume will expand as well as the macroscopic volume of the foam. The solvent’s ability to swell the foam increases in the following order: PDMS, methanol, heptane, cyclohexane, toluene and chloroform. The Hansen solubility parameter for the solvents as well as the foams material are listed in Table 5-3. Methanol and PDMS are the furthest from the foam material, which is why they didn’t swell the foams. The other solvents solubility parameter stays quite close from 15.3 to 19 and the foam’s value is

167

calculated to be 16 which is in between. However, chloroform is less bulky compared with the other solvent molecules and it may diffuse into the foams’ crosslinked network better and causes the best swelling result.

Table 5-3. Materials solubility parameter

Hansen solubility Material parameter(MPa1/2)

Foam 16.0

Heptane 15.3

Cyclohexane 16.8

Toluene 18.2

Trichloromethane 19.0

PDMS oil 21.0

Methanol 29.6

When comparing foams that were made with various toluene content, volumetric capacity gradually increases as the inert diluent content increases from 0 to 60wt% in the oil phase.

In the case of R19T60, the volumetric capacity is the highest which is about 15. As discussed in previous section, the foam’s swelling ability increases due to the decrease in crosslinking density from T0 to T60. Besides the volume expansion due to swelling, the expansion from the macroscopically collapsed state further increased the volumetric capacity.

168

Figure 5-14 demonstrates the foams uptake capacity over 10 cycles of sorption-squeezing cycles. Over the cycles, the sorption capacity is maintained at high level with T0 close to

80% while T60 close to 100%. After squeezing out the solvent, the amount of liquid left inside the foam decreased from 25% for T0 to about 10% for T60. The increased capacity difference between sorption and desorption cycles from T0 to T60 indicates that larger amount of solvent can be recovered within each cycle. Moreover, the steady capacity over

10 cycles demonstrate the good reusability of the foams under loading-unloading operation.

Figure 5-14. Sorption capacity of foams over 10 cycles of saturation with heptane

followed by squeezing, normalized by dry foam weight.

169

5.5.6 Water/oil separation.

It is often found that spilled oil forms a water-in-oil type of emulsions due to either the turbulent conditions in the open water or the existence of organic substances that serve as emulsifiers. The absorption of oils from water/oil emulsions will simulate the performance of the foams under real application conditions. The water/oil emulsion was a 0.5wt% water in heptane stabilized by surfactant Span® 80 (0.1wt%). The water droplets were dispersed by ultra-sonication. Figure 5-15 shows the appearance of emulsions after being passed through the T0, T20, T40, and compact T60 foams. Emulsion treated with T0 foam has the best transparency and such a purifying effect gradually decreases from T0 to T60. The droplet size distribution of the emulsion before and after foam absorption was quantified by laser particle counter. The raw emulsion has a droplet distribution that peaks at 4 µm and has a long tail from 5-17 µm as shown in Figure 5-16. The droplet distribution of the emulsion after passing through T0-T60 foams is shown in Figure 5-17. For T0 foam, it shifts to 2 µm while the droplets larger than 10 µm disappears. For foams from T20-T60, the distributions increasingly resemble the raw emulsion and the fraction of droplets larger than 10 µm gradually increases as shown in Figure 5-18. The average window size for foam T0 to T60 gradually increases from 8.4 to 13.0 µm. Assuming the foam swells isotopically, the windows size scales with the swelling ratio and will be 11.0, 12.3, 18.7 and 30.7. Thus the increasing droplets density for T0 –T60 foams as shown in Figure 5-15 indicates that the purifying effect of water/oil emulsions is dominated by size exclusion mechanism.

170

Figure 5-15 Emulsion before and after foam absorption from left to right: raw emulsion,

and emulsion after foam adsorption T0, T20, T40 and T60.

60 1.8

1.6

50 1.4

1.2 40 1.0 0.8

Intensity (%) Intensity 0.6

30 0.4

0.2

Intensity (%) Intensity 0.0 20 10 11 12 13 14 15 16 17 Droplet size (m)

0 0 2 4 6 8 10 12 14 16 18 Droplet size (m)

Figure 5-16. Raw emulsion droplet size distribution. The insert is from 10 µm to 17 µm

and above.

171

T0 T20 80 80

70 70

60 60

50 50

40 40

Intensity(%) 30 Intensity(%) 30

20 20

10 10

0 0 0 2 4 6 8 10 12 14 16 18 0 2 4 6 8 10 12 14 16 18

T40 80 80 T60

70 70

60 60

50 50

40 40

Intensity(%) 30 Intensity(%) 30

20 20

10 10

0 0 0 2 4 6 8 10 12 14 16 18 0 2 4 6 8 10 12 14 16 18 Droplet size (m) Droplet size (m)

Figure 5-17. Emulsion droplet size distribution after foam absorption.

1.8 1.8 T0 T20 1.6 1.6

1.4 1.4

1.2 1.2

1.0 1.0

0.8 0.8

Intensity(%) 0.6 Intensity(%) 0.6

0.4 0.4

0.2 0.2

0.0 0.0 9 10 11 12 13 14 15 16 17 18 9 10 11 12 13 14 15 16 17 18 1.8 1.8 T40 T60 1.6 1.6

1.4 1.4

1.2 1.2

1.0 1.0

0.8 0.8

Intensity(%) 0.6 Intensity(%) 0.6

0.4 0.4

0.2 0.2

0.0 0.0 9 10 11 12 13 14 15 16 17 18 9 10 11 12 13 14 15 16 17 18 Droplet size (m) Droplet size (m)

Figure 5-18. Emulsion droplets distribution above 10 µm after foam absorption.

172

5.6 Conclusions

In summary, a highly porous crosslinked polymer foam was synthesized using high internal phase emulsion (HIPE) as the template. The volumetric capacity of the foams was enhanced by using an inert diluent in the HIPE continuous phase; a volumetric absorbing capacity of up to 15 ml/ml is achieved. The volumetric expansion of the compact foams is possible due to a force balance between the polymer modulus and the capillary force of drying solvent. The use of inert diluent emphasizes this effect by introducing nanopores in the strut walls by phase separation during polymerization. Control over the volumetric absorbency, oleophilicity, and ability to separate W/O emulsions is possible by varying the

HIPE internal phase fraction and inert diluent concentration. The foams also exhibit good recyclability and oil recovery.

173

CHAPTER 6

FUTURE WORK

6.1 Controlled Particle Distribution via Directional Foam Drying

In this dissertation the wetting of porous foams has been discussed in detail, however the drying of the porous media that are fully/partially saturated/wetted by a dispersion is as much of interest as the wetting process. Through the course of studying the coating of polyHIPE foams, we found great challenges to achieve a uniform coating throughout the interior strut surfaces within the foam. On the contrary, attempts to do so more often resulted in foams having a gradient of coating material. The gradient is found to be in the same direction as the evaporation occurred. Such a finding is not coincident and the reason is rooted in the capillary flow caused by drying. In a simpler situation, drying on a solid surface, the dispersed particles within a drop will undergo redistribution upon drying out.

The famous example of such phenomena is the coffee ring effect[154]. It describes the phenomena that when a spilled drop of coffee dried on a solid surface, it leaves a dense, ring like stain along the perimeter. The coffee particles which are initially dispersed homogeneously in the drop get concentrated into a tiny fraction of it. Such deposition is due to the pinning of the droplet contact line with the substrate and the evaporation as the edge of the droplet has to be replenished by the solvent from the interior. The resulting outward flow can carry virtually all the dispersed materials to the edge. Due to the distinct gradient deposition pattern, such a coffee-ring effect has significant impact in printing, washing and coating processes[98, 155, 156]. 174

Compared to the well-studied drying of a dispersion on a solid surface, drying within a porous polymer medium is poorly studied both theoretically and experimentally. In a rough picture, the process can be generalized in the following sequence: evaporation from the periphery of the porous foams; subsequent replenishment of solvent from the interior, as a result, particles will concentrate at the periphery; invasion of air into the interior and redistribution of particles from large pores towards small pores; a gradient of particles distribution forms in the end.

A few studies have touched this area recently. Hidri, etc. studied the drying of NaCl solution in a porous glass beads packing and found out that both the coffee-ring effect and the porous morphology affects the crystallization behavior of NaCl[124]. Keita, etc. studied the drying of coffee particle inside a packed assembly of glass beads and found that a gradient of concentration formed inside the porous materials after drying, and that such a gradient resulted from a receding-front effect[123].

6.1.1 Proposed future works

Directional particle gradient inside the foam will be prepared by controlled evaporation.

For example, to prepare particle gradient in the vertical direction, evaporation from the side wall of the foam will be inhibited by blocking with parafilm.

The drying of a porous foam with dispersed particles is further complicated by surfactant presented in the dispersion. As the surfactant will affect the pinning situation of solvent with the substrate. It would be interesting to investigate drying of particle dispersion with and without the presence of a surfactant. 175

Also, the foam void/window size dominates the capillary flow during drying, and it would be worthwhile to see how the foams morphology affects the distribution of particles.

6.2 HIPE Curing Mechanism

On one hand, emulsifiers affect HIPE stability by reducing oil/water interfacial tension; on the other hand, they can influence the curing process through controlling the diffusion of free radicals through the interface.

Experimentally, it was found that by increasing one emulsifier (PGS) concentration, the

HIPE stability gets improved. Figure 6-1 shows HIPE curing kinetics during chemorheology measurement. HIPE made with 2.5 wt% PGS shows first a decrease in the storage modulus before curing starts, which indicates that the emulsion undergoes coalescence at such emulsifier concentration. Such a decrease prior to curing disappears when emulsifier concentration increases to 6.5 wt% and above. The curing rate can be estimated from the slope of the curing and it remains in the same proximity with emulsifier concentration increasing. However, replacing part of PGS with the emulsifier DTDMAMS while keeping the overall concentration of emulsifiers the same, the rate of curing as well as the final magnitude of storage modulus both increase. Such an increase tells us that the addition of a co-emulsifier (DTDMAMS is a cationic surfactant) has a significant influence on the emulsion curing process.

176

10000

1000

Storage modulus (Pa) modulus Storage 2.5% PGS 100 6.5% PGS 7.4% PGS 6.5% PGS+0.9% DTDMAMS

0 200 400 600 800 1000 1200 1400 Time (s)

Figure 6-1. Emulsifier concentration & composition effect on HIPE curing process.

The classical HIPE curing mechanism (water/oil type) involves the following steps: 1) oil monomers diffuse through the interfacial film into the aqueous phase and there they meet the initiators; 2) in the aqueous phase, radicals form, grow in molecular weight, and diffuse back to the oil phase through the interfacial film; 3) polymerization in the bulk oil phase.

What can be found from this mechanism is that there is substantial diffusion across the interfacial film back and forth. We postulate that the addition of a second emulsifier (an ionic surfactant) might facilitate the diffusion of radical across the interfacial film. To be specific, the cationic emulsifier will attract the negative charged aqueous initiator

(persulfate) at the interface and it will result in an increased rate of radical diffusion through the interfacial film.

177

6.2.1 Proposed future work

To get the full picture of the curing mechanism, oil based initiator can be used for oil-phase initiated curing. By doing this, the radicals will only form in the bulk oil phase which will test if the diffusion of radicals is the cause of the variations in the curing process. Emulsions with various emulsifier composition and initiator configuration will be studied with chemorheology, mechanical compression measurement, dynamic mechanical analysis, and thermal gravimetric analysis.

6.3 CNC Crosslinking with Polyethylenimine

One drawback of coating polyHIPE foam with CNC or GO particles is that there is lack of a permanent chemical bonds linking those functional particles with the foam. For CNC coatings, the intrinsic hydrophilicity nature will be prone to swelling and destabilize in aqueous environment, thus limiting the application of CNC coated foams in aqueous filtration processes.

6.3.1 Proposed future work

One possible solution is the crosslinking of the CNCs with PEI (Polyethylenimine). The crosslinked CNC-PEI network will be insoluble in aqueous environment and endure the shearing effect of liquids flowing through the foams like in filtration process. Also, both the PEI and CNC have ability of absorbing metal ions.

178

6.4 Hybrid CNC-rGO Coating for Controlling Conductive Foam Sensitivity

The conductivity- strain behavior (in chapter 4) of GO coated foams is found to rely on the

GO coating concentration and strain history. Here we propose a mechanism of controlling the conductivity-strain behavior by using conductive and nonconductive hybrid fillers coating protocol. The CNC is a non-conductive nanomaterial as previously introduced. By inserting CNC nanocrystal into the rGO network, the conductive path formed will be interrupted. During deformation of the coating layer, the cracks formation in both CNC and GO network can help enhance the conductive response to the strain.

6.4.1 Proposed experimental work

The conductivity-strain behavior of CNC-RGO hybrid coated foams will be investigated over the CNC and RGO ratio and overall concentration.

179

REFERENCES

1. Liu, P.S. and K.M. Liang, Review Functional materials of porous metals made by P/M, electroplating and some other techniques. Journal of Materials Science, 2001. 36(21): p. 5059-5072. 2. Banhart, J., Manufacture, characterisation and application of cellular metals and metal foams. Progress in Materials Science, 2001. 46(6): p. 559-632. 3. Ohji, T., Chapter 11.2.2 - Porous Ceramic Materials, in Handbook of Advanced Ceramics (Second Edition), S. Somiya, Editor. 2013, Academic Press: Oxford. p. 1131-1148. 4. Hammel, E.C., O.L.R. Ighodaro, and O.I. Okoli, Processing and properties of advanced porous ceramics: An application based review. Ceramics International, 2014. 40(10, Part A): p. 15351-15370. 5. Smolders, C.A., et al., Microstructures in phase-inversion membranes. Part 1. Formation of macrovoids. Journal of Membrane Science, 1992. 73(2): p. 259-275. 6. Wu, D., et al., Design and Preparation of Porous Polymers. Chemical Reviews, 2012. 112(7): p. 3959-4015. 7. Hrubesh, L.W., Aerogel applications. Journal of Non-Crystalline Solids, 1998. 225: p. 335- 342. 8. Thomas, A., F. Goettmann, and M. Antonietti, Hard Templates for Soft Materials: Creating Nanostructured Organic Materials. Chemistry of Materials, 2008. 20(3): p. 738-755. 9. Tebboth, M., et al., Polymerised high internal phase emulsions for fluid separation applications. Current Opinion in Chemical Engineering, 2014. 4(0): p. 114-120. 10. Silverstein, M.S., Emulsion-templated porous polymers: A retrospective perspective. Polymer, 2014. 55(1): p. 304-320. 11. Taylor, G., The formation of emulsions in definable fields of flow. Proceedings of the Royal Society of London. Series A, 1934. 146(858): p. 501-523. 12. Rumscheidt, F.D. and S.G. Mason, Particle motions in sheared suspensions XII. Deformation and burst of fluid drops in shear and hyperbolic flow. Journal of Colloid Science, 1961. 16(3): p. 238-261. 13. Grace, H.P., Dispersion Phenomena in High-Viscosity Immiscible Fluid Systems and Application of Static Mixers as Dispersion Devices in Such Systems. Chemical Engineering Communications, 1982. 14(3-6): p. 225-277. 14. Emulsion Formation, Stability, and Rheology, in Emulsion Formation and Stability. 15. Lissant, K.J., Geometry of High-Internal-Phase-Ratio Emulsions. Journal of Colloid and Interface Science, 1966. 22(5): p. 462-&. 16. Lissant, K.J. and K.G. Mayhan, Study of Medium and High Internal Phase Ratio Water- Polymer Emulsions. Journal of Colloid and Interface Science, 1973. 42(1): p. 201-208. 17. Lissant, K.J., et al., Structure of High Internal Phase Ratio Emulsions. Journal of Colloid and Interface Science, 1974. 47(2): p. 416-423. 18. Aronson, M.P. and M.F. Petko, Highly Concentrated Water-in-Oil Emulsions - Influence of Electrolyte on Their Properties and Stability. Journal of Colloid and Interface Science, 1993. 159(1): p. 134-149.

180

19. Kabalnov, A.S. and E.D. Shchukin, Ostwald ripening theory: applications to fluorocarbon emulsion stability. Advances in Colloid and Interface Science, 1992. 38(0): p. 69-97. 20. Kizling, J. and B. Kronberg, On the formation and stability of concentrated water-in-oil emuslions, aphrons. Colloids and Surfaces, 1990. 50(0): p. 131-140. 21. Williams, J.M., A.J. Gray, and M.H. Wilkerson, Emulsion stability and rigid foams from styrene or divinylbenzene water-in-oil emulsions. Langmuir, 1990. 6(2): p. 437-444. 22. Williams, J.M. and D.A. Wrobleski, Spatial distribution of the phases in water-in-oil emulsions. Open and closed microcellular foams from cross-linked polystyrene. Langmuir, 1988. 4(3): p. 656-662. 23. Williams, J.M., High internal phase water-in-oil emulsions: influence of surfactants and cosurfactants on emulsion stability and foam quality. Langmuir, 1991. 7(7): p. 1370-1377. 24. Princen, H.M. and A.D. Kiss, Rheology of foams and highly concentrated emulsions: III. Static shear modulus. Journal of Colloid and Interface Science, 1986. 112(2): p. 427-437. 25. Derjaguin, B., Die elastischen Eigenschaften der Schäume. Kolloid-Zeitschrift, 1933. 64(1): p. 1-6. 26. Gurevitch, I. and M.S. Silverstein, Nanoparticle-Based and Organic-Phase-Based AGET ATRP PolyHIPE Synthesis within Pickering HIPEs and Surfactant-Stabilized HIPEs. Macromolecules, 2011. 44(9): p. 3398-3409. 27. Clayton, A.M., Chemorheology of Thermosetting Polymers. ACS Symposium Series. Vol. 227. 1982: AMERICAN CHEMICAL SOCIETY. 348. 28. Foudazi, R., et al., Chemorheology of Poly(high internal phase emulsions). Macromolecules, 2013. 46(13): p. 5393-5396. 29. Langmuir, I., Experiments with oil on water. Journal of Chemical Education, 1931. 8(5): p. 850. 30. Langmuir, I., THE ADSORPTION OF GASES ON PLANE SURFACES OF GLASS, MICA AND PLATINUM. Journal of the American Chemical Society, 1918. 40(9): p. 1361-1403. 31. Langmuir, I., THE CONSTITUTION AND FUNDAMENTAL PROPERTIES OF SOLIDS AND LIQUIDS. II. LIQUIDS.1. Journal of the American Chemical Society, 1917. 39(9): p. 1848- 1906. 32. Langmuir, I., THE CONSTITUTION AND FUNDAMENTAL PROPERTIES OF SOLIDS AND LIQUIDS. PART I. SOLIDS. Journal of the American Chemical Society, 1916. 38(11): p. 2221- 2295. 33. Liggieri, L., et al., Adsorption Kinetics of Alkylphosphine Oxides at Water/Hexane Interface: 2. Theory of the Adsorption with Transport across the Interface in Finite Systems. Journal of Colloid and Interface Science, 1997. 186(1): p. 46-52. 34. Ferrari, M., et al., Adsorption Kinetics of Alkylphosphine Oxides at Water/Hexane Interface: 1. Pendant Drop Experiments. Journal of Colloid and Interface Science, 1997. 186(1): p. 40-45. 35. M., H., S. A., and B.P. M., Adsorption‐desorption barrier, diffusional exchanges and surface instabilities of longitudinal waves for aperiodic regimes. AIChE Journal, 1981. 27(6): p. 1002-1008. 36. Ravera, F., M. Ferrari, and L. Liggieri, Adsorption and partitioning of surfactants in liquid– liquid systems. Advances in Colloid and Interface Science, 2000. 88(1–2): p. 129-177.

181

37. Joos, P. and G. Serrien, Adsorption kinetics of lower alkanols at the air/water interface: Effect of structure makers and structure breakers. Journal of Colloid and Interface Science, 1989. 127(1): p. 97-103. 38. Hansen, R.S., The theory of diffusion controlled absorption kinetics with accompanying evaporation. The Journal of Physical Chemistry, 1960. 64(5): p. 637-641. 39. Van Hunsel, J. and P. Joos, Steady-state dynamic interfacial tensions of 1-alkanols during mass transfer across the hexane/water interface. Langmuir, 1987. 3(6): p. 1069-1074. 40. Fainerman, V.B., S.A. Zholob, and R. Miller, Adsorption Kinetics of Oxyethylated Polyglycol Ethers at the Water−Nonane Interface. Langmuir, 1997. 13(2): p. 283-289. 41. Shimbashi, T., The Mass-transfer Rate through the Liquid-Liquid Interface. V. Diffusion through an Interface Considering the Variation in the Amount of Adsorption. Bulletin of the Chemical Society of Japan, 1975. 48(2): p. 626-629. 42. Mansfield, W., The spontaneous emulsification of mixtures of oleic acid and paraffin oil in alkaline solutions. Australian Journal of Chemistry, 1952. 5(2): p. 331-338. 43. Rubin, E. and C.J. Radke, Dynamic interfacial tension minima in finite systems. Chemical Engineering Science, 1980. 35(5): p. 1129-1138. 44. Princen, H.M., Rheology of Foams and Highly Concentrated Emulsions .2. Experimental- Study of the Yield Stress and Wall Effects for Concentrated Oil-in-Water Emulsions. Journal of Colloid and Interface Science, 1985. 105(1): p. 150-171. 45. Brunauer, S., P.H. Emmett, and E. Teller, Adsorption of gases in multimolecular layers. Journal of the American chemical society, 1938. 60(2): p. 309-319. 46. Sing, K., The use of nitrogen adsorption for the characterisation of porous materials. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 2001. 187: p. 3-9. 47. Grace†, H.P., DISPERSION PHENOMENA IN HIGH VISCOSITY IMMISCIBLE FLUID SYSTEMS AND APPLICATION OF STATIC MIXERS AS DISPERSION DEVICES IN SUCH SYSTEMS. Chemical Engineering Communications, 1982. 14(3-6): p. 225-277. 48. van Aken, G.A. and F.D. Zoet, Coalescence in Highly Concentrated Coarse Emulsions. Langmuir, 2000. 16(18): p. 7131-7138. 49. Princen, H.M., Rheology of foams and highly concentrated emulsions: I. Elastic properties and yield stress of a cylindrical model system. Journal of Colloid and Interface Science, 1983. 91(1): p. 160-175. 50. Gibson, L., et al., The mechanics of two-dimensional cellular materials. Proceedings of the Royal Society of London. A. Mathematical and Physical Sciences, 1982. 382(1782): p. 25- 42. 51. Gibson, L.J. and M.F. Ashby, The Mechanics of Three-Dimensional Cellular Materials. Proceedings of the Royal Society of London. Series A, Mathematical and Physical Sciences, 1982. 382(1782): p. 43-59. 52. Khelifa, F., et al., Effect of cellulosic nanowhiskers on the performances of epoxidized acrylic copolymers. Journal of Materials Chemistry, 2012. 22(38): p. 20520-20528. 53. Tarley, C.R.T., et al., Preparation of new ion-selective cross-linked poly(vinylimidazole-co- ethylene glycol dimethacrylate) using a double-imprinting process for the preconcentration of Pb2+ ions. Journal of Colloid and Interface Science, 2015. 450: p. 254- 263.

182

54. Stang, M., H. Karbstein, and H. Schubert, Adsorption kinetics of emulsifiers at oil—water interfaces and their effect on mechanical emulsification. Chemical Engineering and Processing: Process Intensification, 1994. 33(5): p. 307-311. 55. Stokes, G.G., On the effect of the internal friction of fluids on the motion of pendulums. Vol. 9. 1851: Pitt Press Cambridge. 56. Richardson, J. and W. Zaki, Sedimentation and fluidisation: Part I. Chemical Engineering Research and Design, 1954. 75: p. S82-S100. 57. Di Felice, R. and R. Kehlenbeck, Sedimentation Velocity of Solids in Finite Size Vessels. Chemical Engineering & Technology, 2000. 23(12): p. 1123-1126. 58. Bell, J.M. and F.K. Cameron, The Flow of Liquids through Capillary Spaces. The Journal of Physical Chemistry, 1905. 10(8): p. 658-674. 59. Lucas, R., Ueber das Zeitgesetz des kapillaren Aufstiegs von Flüssigkeiten. Kolloid- Zeitschrift, 1918. 23(1): p. 15-22. 60. Washburn, E.W., The Dynamics of Capillary Flow. Physical Review, 1921. 17(3): p. 273- 283. 61. Cai, J., et al., Generalized Modeling of Spontaneous Imbibition Based on Hagen–Poiseuille Flow in Tortuous Capillaries with Variably Shaped Apertures. Langmuir, 2014. 30(18): p. 5142-5151. 62. Law, Y.Y., D.L. Feke, and I. Manas-Zloczower, Method for probing the microstructure of particle beds using infiltration behavior. Powder Technology, 2013. 237: p. 427-431. 63. Fries, N. and M. Dreyer, The transition from inertial to viscous flow in capillary rise. Journal of Colloid and Interface Science, 2008. 327(1): p. 125-128. 64. Hanžič, L., L. Kosec, and I. Anžel, Capillary absorption in concrete and the Lucas–Washburn equation. Cement and Concrete Composites, 2010. 32(1): p. 84-91. 65. Fries, N. and M. Dreyer, An analytic solution of capillary rise restrained by gravity. Journal of Colloid and Interface Science, 2008. 320(1): p. 259-263. 66. Li, K., et al., Criteria for Applying the Lucas-Washburn Law. Scientific Reports, 2015. 5: p. 14085. 67. Zhmud, B.V., F. Tiberg, and K. Hallstensson, Dynamics of Capillary Rise. Journal of Colloid and Interface Science, 2000. 228(2): p. 263-269. 68. Gilman, J.J., Direct Measurements of the Surface Energies of Crystals. Journal of Applied Physics, 1960. 31(12): p. 2208-2218. 69. Rhee, S.K., Surface energies of silicate glasses calculated from their wettability data. Journal of Materials Science, 1977. 12(4): p. 823-824. 70. Eral, H.B., D.J.C.M. ’t Mannetje, and J.M. Oh, Contact angle hysteresis: a review of fundamentals and applications. Colloid and Polymer Science, 2013. 291(2): p. 247-260. 71. O’Loughlin, M., et al., Capillary rise dynamics of aqueous glycerol solutions in glass capillaries: A critical examination of the Washburn equation. Journal of Colloid and Interface Science, 2013. 411: p. 257-264. 72. Rudzinski, W. and B. Wojciechowski, Introduction. Effects of Surface Heterogeneity in Adsorption and Catalysis on Solids. Langmuir, 1993. 9(10): p. 2483-2484. 73. de Gennes, P.G., Wetting: statics and dynamics. Reviews of Modern Physics, 1985. 57(3): p. 827-863. 74. Tanner, L.H., The spreading of silicone oil drops on horizontal surfaces. Journal of Physics D: Applied Physics, 1979. 12(9): p. 1473. 183

75. Lopez, J., C.A. Miller, and E. Ruckenstein, Spreading kinetics of liquid drops on solids. Journal of Colloid and Interface Science, 1976. 56(3): p. 460-468. 76. Cazabat, A. and M.C. Stuart, Dynamics of wetting: effects of surface roughness. The Journal of Physical Chemistry, 1986. 90(22): p. 5845-5849. 77. Vincent, O., A. Szenicer, and A.D. Stroock, Capillarity-driven flows at the continuum limit. Soft Matter, 2016. 12(31): p. 6656-6661. 78. Gruener, S., et al., Capillary rise of water in hydrophilic nanopores. Physical Review E, 2009. 79(6): p. 067301. 79. Shen, A., et al., A model for capillary rise in nano-channels with inherent surface roughness. Applied Physics Letters, 2017. 110(12): p. 121601. 80. Haneveld, J., et al., Capillary filling of sub-10nm nanochannels. Journal of Applied Physics, 2008. 104(1): p. 014309. 81. Ishino, C., et al., Wicking within forests of micropillars. EPL (Europhysics Letters), 2007. 79(5): p. 56005. 82. Sbragaglia, M., et al., Surface Roughness-Hydrophobicity Coupling in Microchannel and Nanochannel Flows. Physical Review Letters, 2006. 97(20): p. 204503. 83. Wenzel, R.N., Resistance of solid surfaces to wetting by water. Industrial & Engineering Chemistry, 1936. 28(8): p. 988-994. 84. Cassie, A. and S. Baxter, Wettability of porous surfaces. Transactions of the Faraday society, 1944. 40: p. 546-551. 85. Extrand, C.W., Contact Angles and Hysteresis on Surfaces with Chemically Heterogeneous Islands. Langmuir, 2003. 19(9): p. 3793-3796. 86. Gao, L. and T.J. McCarthy, Wetting 101°. Langmuir, 2009. 25(24): p. 14105-14115. 87. Krumpfer, J.W. and T.J. McCarthy, Contact angle hysteresis: a different view and a trivial recipe for low hysteresis hydrophobic surfaces. Faraday Discussions, 2010. 146(0): p. 103- 111. 88. Gao, L. and T.J. McCarthy, “Artificial Lotus Leaf” Prepared Using a 1945 Patent and a Commercial Textile. Langmuir, 2006. 22(14): p. 5998-6000. 89. Öner, D. and T.J. McCarthy, Ultrahydrophobic Surfaces. Effects of Topography Length Scales on Wettability. Langmuir, 2000. 16(20): p. 7777-7782. 90. Dimitrov, D.I., A. Milchev, and K. Binder, Capillary Rise in Nanopores: Molecular Dynamics Evidence for the Lucas-Washburn Equation. Physical Review Letters, 2007. 99(5): p. 054501. 91. Robinson, J.L., et al., Achieving Interconnected Pore Architecture in Injectable PolyHIPEs for Bone Tissue Engineering. Tissue Engineering Part A, 2014. 20(5-6): p. 1103-1112. 92. Moglia, R.S., et al., Injectable Polymerized High Internal Phase Emulsions with Rapid in Situ Curing. Biomacromolecules, 2014. 93. Moglia, R.S., et al., Fabrication of injectable and high porosity polyMIPE scaffolds for soft tissue regeneration. Abstracts of Papers of the American Chemical Society, 2013. 245. 94. Zhang, T. and Q. Guo, Continuous preparation of polyHIPE monoliths from ionomer- stabilized high internal phase emulsions (HIPEs) for efficient recovery of spilled oils. Chemical Engineering Journal, 2017. 307: p. 812-819. 95. Pulko, I., et al., Emulsion templated open porous membranes for protein purification. Journal of Chromatography A, 2011. 1218(17): p. 2396-2401. 96. Lake, L.W., Enhanced oil recovery. 1989. 184

97. Morrow, N.R., Wettability and Its Effect on Oil Recovery. 98. Still, T., P.J. Yunker, and A.G. Yodh, Surfactant-Induced Marangoni Eddies Alter the Coffee- Rings of Evaporating Colloidal Drops. Langmuir, 2012. 28(11): p. 4984-4988. 99. Cameron, N., et al., Study of the formation of the open-cellular morphology of poly (styrene/divinylbenzene) polyHIPE materials by cryo-SEM. Colloid and Polymer Science, 1996. 274(6): p. 592-595. 100. Sevsek, U., et al., Post polymerisation hypercrosslinking of styrene/divinylbenzene poly(HIPE)s: Creating micropores within macroporous polymer. Polymer, 2014. 55(1): p. 410-415. 101. Xu, H., et al., Interconnected Porous Polymers with Tunable Pore Throat Size Prepared via Pickering High Internal Phase Emulsions. Langmuir, 2016. 32(1): p. 38-45. 102. Gruner, S., Rheology and dynamics of simple and complex liquids in mesoporous matrices. 2010: Logos Verlag Berlin GmbH. 103. Beeckman, J.W., Mathematical description of heterogeneous materials. Chemical Engineering Science, 1990. 45(8): p. 2603-2610. 104. Hager, W.H., Wilfrid Noel Bond and the Bond number. Journal of Hydraulic Research, 2012. 50(1): p. 3-9. 105. Hamraoui, A. and T. Nylander, Analytical Approach for the Lucas–Washburn Equation. Journal of Colloid and Interface Science, 2002. 250(2): p. 415-421. 106. Wu, S., Org. Coat. Plast. Chem.,, 1971. 31(27). 107. Bico, J., U. Thiele, and D. Quéré, Wetting of textured surfaces. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 2002. 206(1–3): p. 41-46. 108. Reiter, G., Dewetting of thin polymer films. Physical Review Letters, 1992. 68(1): p. 75-78. 109. Sharma, A. and G. Reiter, Instability of Thin Polymer Films on Coated Substrates: Rupture, Dewetting, and Drop Formation. Journal of Colloid and Interface Science, 1996. 178(2): p. 383-399. 110. Kim, H., et al., Controlled Uniform Coating from the Interplay of Marangoni Flows and Surface-Adsorbed Macromolecules. Physical Review Letters, 2016. 116(12): p. 124501. 111. Simmons, B.A., et al., Microstructure Determination of AOT + Phenol Organogels Utilizing Small-Angle X-ray Scattering and Atomic Force Microscopy. Journal of the American Chemical Society, 2001. 123(10): p. 2414-2421. 112. R. Cameron, N. and D. C. Sherrington, Preparation and glass transition temperatures of elastomeric PolyHIPE materials. Journal of Materials Chemistry, 1997. 7(11): p. 2209- 2212. 113. Y., S.A., et al., Polymerized high internal‐phase emulsions: Properties and interaction with water. Journal of Applied Polymer Science, 2002. 84(11): p. 2018-2027. 114. Livshin, S. and M.S. Silverstein, Enhancing hydrophilicity in a hydrophobic porous emulsion-templated polyacrylate. Journal of Polymer Science Part A: Polymer Chemistry, 2009. 47(18): p. 4840-4845. 115. Gitli, T. and M.S. Silverstein, Bicontinuous hydrogel-hydrophobic polymer systems through emulsion templated simultaneous polymerizations. Soft Matter, 2008. 4(12): p. 2475- 2485. 116. Isogai, A., T. Saito, and H. Fukuzumi, TEMPO-oxidized cellulose nanofibers. Nanoscale, 2011. 3(1): p. 71-85.

185

117. Marcano, D.C., et al., Improved Synthesis of Graphene Oxide. ACS Nano, 2010. 4(8): p. 4806-4814. 118. DONG, X.M., J.-F. REVOL, and D.G. GRAY, Effect of microcrystallite preparation conditions on the formation of colloid crystals of cellulose. Cellulose, 1998. 5(1): p. 19-32. 119. Marchessault, R., F. Morehead, and N. Walter, Liquid crystal systems from fibrillar polysaccharides. Nature, 1959. 184(4686): p. 632-633. 120. Beck-Candanedo, S., M. Roman, and D.G. Gray, Effect of Reaction Conditions on the Properties and Behavior of Wood Cellulose Nanocrystal Suspensions. Biomacromolecules, 2005. 6(2): p. 1048-1054. 121. Saito, T., et al., Homogeneous Suspensions of Individualized Microfibrils from TEMPO- Catalyzed Oxidation of Native Cellulose. Biomacromolecules, 2006. 7(6): p. 1687-1691. 122. Saito, T., et al., Cellulose Nanofibers Prepared by TEMPO-Mediated Oxidation of Native Cellulose. Biomacromolecules, 2007. 8(8): p. 2485-2491. 123. Keita, E., et al., MRI evidence for a receding-front effect in drying porous media. Physical Review E, 2013. 87(6): p. 062303. 124. Hidri, F., et al., Porous medium coffee ring effect and other factors affecting the first crystallisation time of sodium chloride at the surface of a drying porous medium. Physics of Fluids, 2013. 25(12): p. 127101. 125. Xu, L., et al., Dynamics of Drying in 3D Porous Media. Physical Review Letters, 2008. 101(9): p. 094502. 126. Laurindo, J.B. and M. Prat, Numerical and experimental network study of evaporation in capillary porous media. Drying rates. Chemical Engineering Science, 1998. 53(12): p. 2257- 2269. 127. Laurindo, J.B. and M. Prat, Numerical and experimental network study of evaporation in capillary porous media. Phase distributions. Chemical Engineering Science, 1996. 51(23): p. 5171-5185. 128. Carswell, C., Unique oil spill in East China Sea frustrates scientists. Nature, 2018. 554(7690): p. 17-18. 129. Sanchi oil spill continues; impacts still unclear. C&EN Global Enterprise, 2018. 96(5): p. 16- 16. 130. Ivshina, I.B., et al., Oil spill problems and sustainable response strategies through new technologies. Environmental Science: Processes & Impacts, 2015. 17(7): p. 1201-1219. 131. Al-Majed, A.A., A.R. Adebayo, and M.E. Hossain, A sustainable approach to controlling oil spills. Journal of Environmental Management, 2012. 113: p. 213-227. 132. Teas, C., et al., Investigation of the effectiveness of absorbent materials in oil spills clean up. Desalination, 2001. 140(3): p. 259-264. 133. Wang, J., Y. Zheng, and A. Wang, Superhydrophobic kapok fiber oil-absorbent: Preparation and high oil absorbency. Chemical Engineering Journal, 2012. 213: p. 1-7. 134. Deschamps, G., et al., Oil Removal from Water by Selective Sorption on Hydrophobic Cotton Fibers. 1. Study of Sorption Properties and Comparison with Other Cotton Fiber- Based Sorbents. Environmental Science & Technology, 2003. 37(5): p. 1013-1015. 135. Choi, H.M. and R.M. Cloud, Natural sorbents in oil spill cleanup. Environmental science & technology, 1992. 26(4): p. 772-776.

186

136. Junping, Z. and S. Stefan, Polyester Materials with Superwetting Silicone Nanofilaments for Oil/Water Separation and Selective Oil Absorption. Advanced Functional Materials, 2011. 21(24): p. 4699-4704. 137. Liu, Y., et al., Cost-Effective Reduced Graphene Oxide-Coated Polyurethane Sponge As a Highly Efficient and Reusable Oil-Absorbent. ACS Applied Materials & Interfaces, 2013. 5(20): p. 10018-10026. 138. Choi, S.-J., et al., A Polydimethylsiloxane (PDMS) Sponge for the Selective Absorption of Oil from Water. ACS Applied Materials & Interfaces, 2011. 3(12): p. 4552-4556. 139. Pham, V.H. and J.H. Dickerson, Superhydrophobic Silanized Melamine Sponges as High Efficiency Oil Absorbent Materials. ACS Applied Materials & Interfaces, 2014. 6(16): p. 14181-14188. 140. Stolz, A., et al., Melamine-derived carbon sponges for oil-water separation. Carbon, 2016. 107: p. 198-208. 141. Gui, X., et al., Carbon Nanotube Sponges. Advanced Materials, 2010. 22(5): p. 617-621. 142. Sun, H., Z. Xu, and C. Gao, Multifunctional, Ultra‐Flyweight, Synergistically Assembled Carbon Aerogels. Advanced Materials, 2013. 25(18): p. 2554-2560. 143. Flory, P.J., Principles of polymer chemistry. 1953. 144. Ono, T., et al., Lipophilic polyelectrolyte gels as super-absorbent polymers for nonpolar organic solvents. Nature Materials, 2007. 6: p. 429. 145. Zhang, A., et al., Poly(dimethylsiloxane) Oil Absorbent with a Three-Dimensionally Interconnected Porous Structure and Swellable Skeleton. ACS Applied Materials & Interfaces, 2013. 5(20): p. 10201-10206. 146. Rubinstein, M. and R. Colby, Polymers Physics. 2003: Oxford. 147. Dušek, K., Phase separation during the formation of three‐dimensional polymers. Journal of Polymer Science Part C: Polymer Symposia, 1967. 16(3): p. 1289-1299. 148. Okay, O., Phase separation in free-radical crosslinking copolymerization: formation of heterogeneous polymer networks. Polymer, 1999. 40(14): p. 4117-4129. 149. Flory, P.J. and J.R. Jr., Statistical Mechanics of Cross‐Linked Polymer Networks I. Rubberlike Elasticity. The Journal of Chemical Physics, 1943. 11(11): p. 512-520. 150. Maiti, S.K., L.J. Gibson, and M.F. Ashby, Deformation and energy absorption diagrams for cellular solids. Acta Metallurgica, 1984. 32(11): p. 1963-1975. 151. McKenna, G.B., K.M. Flynn, and Y. Chen, Swelling in crosslinked natural rubber: experimental evidence of the crosslink density dependence of χ. Polymer, 1990. 31(10): p. 1937-1945. 152. Thomas, O., et al., Effect of cross-link density on the morphology, thermal and mechanical properties of flexible molded polyurea/urethane foams and films. Journal of Polymer Science Part B: Polymer Physics, 1994. 32(13): p. 2155-2169. 153. George, S.C., M. knörgen, and S. Thomas, Effect of nature and extent of crosslinking on swelling and mechanical behavior of styrene–butadiene rubber membranes. Journal of Membrane Science, 1999. 163(1): p. 1-17. 154. Deegan, R.D., et al., Capillary flow as the cause of ring stains from dried liquid drops. Nature, 1997. 389(6653): p. 827-829. 155. Han, W. and Z. Lin, Learning from “Coffee Rings”: Ordered Structures Enabled by Controlled Evaporative Self-Assembly. Angewandte Chemie International Edition, 2012. 51(7): p. 1534-1546. 187

156. Zhang, Z., et al., Controlled Inkjetting of a Conductive Pattern of Silver Nanoparticles Based on the Coffee-Ring Effect. Advanced Materials, 2013. 25(46): p. 6714-6718.

188