© 2019 Nicholas Teo ALL RIGHTS RESERVED

Home , Aerogel

NANO, MICRO AND MACRO SCALE CONTROL OF POROUS AEROGEL MORPHOLOGY

A Dissertation Presented to The Graduate Faculty of the University of Akron

In Partial Fulfillment Of the Requirements for the Degree Doctor of Philosophy

Nicholas Teo May 2019

NANO, MICRO AND MACRO SCALE CONTROL OF POROUS AEROGEL MORPHOLOGY

Nicholas Teo Dissertation

Approved: Accepted:

s s Advisor Department Chair Dr. Sadhan C. Jana Dr. Sadhan C. Jana

s s Committee Member Interim Dean of College Dr. Bryan D. Vogt Dr. Ali Dhinojwala

s s Committee Member Dean of Graduate School Dr. Younjin Min Dr. Chand K. Midha

s s Committee Member Date Dr. Coleen Pugh

s Committee Member Dr. George Chase

ABSTRACT

This research centers on combining different disparate technologies with the aerogel fabrication process to the control of nano, micro and macrostructure of aerogel morphology. It is envisioned that this control over structure at different length scales will enable aerogels to be used for various applications such as drug delivery, nanoparticle filtration or oil/water separation.

Aerogels are a class of highly porous structures (>90% porosity) with inherently small pores with size typically in the range of 2-200 nm. These small pores are achieved through a combination of both the gel formation and supercritical drying steps, differentiating aerogels from other porous material counterparts (e.g. foams). The control of morphology is accomplished through manipulation of phase growth, introduction of dispersed phase liquids and templating of structures via 3D printing in conjunction with the aerogel fabrication process to produce a variety of aerogel structural forms such as foams, microparticles and mechanical metamaterials. In this work, two different polymeric material systems were studied, namely syndiotactic polystyrene and polyimide.

Syndiotactic polystyrene was selected as it forms physically crosslinked, thermo-reversible gels which allow for on-demand gelation, as well as compatibility in water-in-oil emulsion systems. Polyimide was selected as a condensation sol-gel system provides increased flexibility in aerogel mechanical and chemical properties through different monomer selection.

iii

It was identified that chemical reaction kinetics, solvent effects, interfacial conditions and kinetics of phase separation all impact and control the structure-property relationships of aerogel materials. The body of work presented in this dissertation covers a wide variety of topics such as new synthesis of aerogel monoliths, microparticles and foams. The fabrication of these aerogel structures required thorough understanding of water-in-oil and oil-in-oil emulsion systems, solvent-monomer/polymer interactions, microfluidics and 3D printing. It was important for this work to adapt and incorporate these different topics within the restrictions and confines of the aerogel fabrication process.

The effects of solvent properties (electron accepting capability and viscosity) on the tuning of aerogel pore structures were first evaluated. Both changes in solvent properties affected the gel times of the polyimide sol, thus enabling an increase in the time gap between thermodynamic phase transitions, leading to coarsening of polyimide strands. The gel time and the reaction rate were varied through two methods. First, the solvent acidity/basicity was selected to influence the forward reaction rate and conversion of the crosslinking reaction. Second, the solvent viscosity was varied through the addition of a viscosity modifier; to delay reactant diffusion rates to the reaction sites. Both these methods resulted in a shift in pore size distribution from predominantly mesopores to macropores.

In the second part of the work, a microparticle formation process was developed that utilized a cheap, easily-assembled microfluidic device. This allowed the successful synthesis of monodisperse aerogel particles (95 % porosity) using a surfactant-free, oil-in- oil emulsion process. Extension of this microfluidic device with the addition of another flow enabled the creation of core-shell hollow microspheres with diameter of 500 µm and

iv wall thickness of 5 µm. To our knowledge, these hollow aerogel microspheres are the first of its kind reported in literature.

The third part of the study was focused on development of a set of hierarchical porous structures known as an aerogel foam, combining the properties of polymer foams (with micrometer size pores) and aerogels (mesoporous structure). These hierarchical structures were produced via emulsion-templating, followed by gelation of the continuous phase.

Both water-in-oil emulsion and oil-in-oil emulsion systems were considered to account for the polymer systems that undergo respectively thermo-reversible gelation and sol-gel transition via chemical crosslinking. The resultant aerogel foams exhibited increased porosity due to the inclusion of the macrovoids.

Finally, a new technique was developed to create complex aerogel shapes through the incorporation of 3D printing tools in the aerogel synthesis process. In this study, 3D printing was used to create sacrificial hollow molds that allowed the injection of the sol prior to gelation. The sol was allowed to cure in the hollow mold and removed from the mold through selective dissolution of the molds using a specific solvent pairing. This allowed for the creation of aerogels with curved surfaces, recessed spaces and intricate geometries. This method also allowed for the inclusion of an additional level of porosity to an inherently porous structure, resulting in a set of hierarchical porous structures.

DEDICATION

This dissertation is dedicated to my parents, who have brought me up to be the person that

I am today. Particularly, to my mother who has allowed me to chase my dreams halfway across the world.

ACKNOWLEDGEMENTS

Firstly, I would like to express my deepfelt gratitude and thanks to my advisor, Dr. Sadhan

C. Jana, for his leadership, mentorship and guidance during my graduate studies. He has contributed significantly in my development as a scientist, through insightful comments, engaging technical discussions and collaboration in the development of numerous research related ideas. However, more importantly, he has helped me to grow as a person in the areas of leadership, communication, interpersonal skills and critical thinking. It was through his many words of affirmation and encouragement that I can present this work today. He truly is a mentor and advisor in all aspects of these roles.

I would also like to thank my committee members: Dr. Bryan D. Vogt, Dr. Younjin Min,

Dr. Coleen Pugh and Dr. George Chase for their helpful suggestions, comments and encouragement. In addition, I would also like to thank Dr. Eric Amis and Piljae Joo for their help in the 3D printing aspect of this work.

Thanks also goes out to all the past and present research group members that have helped me along this journey of my Ph.D. studies. This was done through numerous technical discussions in the office, help with instrumentation and daily interactions both in the office and the labs.

Lastly, a special thank you to all my friends that I have made in the department, university and the Akron area.

vii

TABLE OF CONTENTS

Page

LIST OF FIGURES xiii

LIST OF TABLES xxiv

CHAPTER

1. INTRODUCTION ………………………………………………….…… 1

2. BACKGROUND AND LITERATURE REVIEW

2.1 Aerogels …………………………………………………………. 6

2.1.1. Overview of aerogels ……………………………………. 6

2.1.2. Silica aerogels …………………………………...... 8

2.1.3. Polyimide aerogels ………………………………………. 12

2.1.4. Syndiotactic polystyrene aerogels ………………………... 17

2.2 Foams ……………………………………………………………. 22

2.2.1 Overview of foams .……………………………………. 22

2.2.2 Polyurethane foams ….………………………………… 24

2.3 PolyHIPEs ………………………………………………………. 25

2.3.1 Overview of polyHIPEs ...………………………………. 25

2.3.2 High internal phase emulsions (HIPEs) ……………….... 27

2.3.3 Polymerization of HIPEs ………………………………. 30

2.3.4 Applications of polyHIPEs ……………………………... 34

viii

Page

2.4 Microfluidic Droplet Generation ….……………………………... 36

2.4.1 Overview of microfluidics ……………………………… 36

2.4.2 Passive droplet generation ……………………………… 37

2.4.3 Passive droplet generation geometry …………………… 40

2.4.4 Active droplet generation .…………………………….... 45

2.5 Additive Manufacturing …………………………………………. 47

2.5.1 Overview of additive manufacturing …………………… 47

2.5.2 Material types ...... 48

2.5.3 Applications ……………………………………………. 52

3. SOLVENT EFFECTS ON TUNING PORE STRUCTURES IN

POLYIMIDE AEROGELS ……………………………………………... 54

3.1 Introduction ………………………………………………………. 55

3.2 Experimental section ………………………………...... 58

3.2.1 Materials ………………………………………………. 58

3.2.2 Fabrication of polyimide aerogels ……………………... 58

3.2.3 Characterization of aerogel materials …………………. 60

3.3 Results and discussion ……………………………………………. 62

3.3.1 Control of pore size distribution through solvent

composition …………………………………………… 62

3.3.2 Control of pore size distribution through addition of

surfactant ……………………………………………. 74

3.4 Conclusion ………………………...... 79

Page

4. A SURFACTANT-FREE PROCESS FOR FABRICATION OF

POLYIMIDE AEROGEL MICROPARTICLES ...... …………………… 81

4.1 Introduction ………………………………………………………. 82

4.2 Experimental section …………………………………...... 85

4.2.1 Materials ………………………………………………. 85

4.2.2 Fabrication of droplet generator ...…….………………. 86

4.2.3 Preparation of dispersed phase solution ………………. 90

4.2.4 Fabrication of polyimide aerogel microparticles ……… 91

4.2.5 Characterization of aerogel microparticles ……………. 92

4.3 Results and discussion ……………………………………………. 94

4.3.1 Properties of aerogel monoliths ………………………. 94

4.3.2 Morphology of gel and aerogel microparticles ………. 96

4.3.3 Effect of temperature on microparticles ………………. 103

4.3.4 Morphology of core-shell hollow microspheres ………... 110

4.4 Conclusion …………………………………...... 113

5. OPEN CELL AEROGEL FOAMS VIA EMULSION-TEMPLATING .. 115

5.1 Introduction ………………………………………………………. 116

5.2 Experimental section …………...... 119

5.2.1 Materials ………………………………………………. 119

5.2.2 Fabrication of aerogel foams …....……………………. 119

5.2.3 Characterization of water-in-oil emulsions ……………. 120

5.2.4 Characterization of aerogel foams ……………………... 121

Page

5.3 Results and discussion ……………………………………………. 122

5.3.1 Emulsion stability ……………………………………… 122

5.3.2 Effect of temperature on emulsion formation .…….…... 123

5.3.3 Effect of temperature on preformed emulsions .………. 124

5.3.4 Effect of surfactant concentration on emulsions ………. 127

5.3.5 Morphology of aerogel foam …………………………. 131

5.3.6 Macrovoid size distribution …………………………… 132

5.3.7 Open cell structure ……………………………………. 136

5.3.7 Macroporous polymer skin structure …………………. 139

5.4 Conclusion ……………...... 144

5.5 Appendix ……………………………………….…………………. 145

6. POLYIMIDE-BASED AEROGEL FOAMS VIA EMULSION-

TEMPLATING …………………………………………………………. 148

6.1 Introduction ………………………………………………………. 149

6.2 Experimental section ……………………...... 151

6.2.1 Materials ………………………………………………. 151

6.2.2 Fabrication of aerogel foams …....……………………... 152

6.2.3 Characterization of oil-in-oil emulsions ………………. 154

6.2.4 Characterization of aerogel foams ……………………... 155

6.3 Results and discussion ……………………………………………. 157

6.3.1 Emulsion formation ………….………………………… 157

6.3.2 Morphology of aerogel foam …………………………. 161

Page

6.3.3 Macrovoid size distribution …………………………… 170

6.3.4 Mechanical properties …………………………………. 172

6.3.5 Oil and water absorption ………………………………. 174

6.4 Conclusion ……………………………………...... 176

7. FACILE SYNTHESIS OF BICONTINUOUS GYROID

STRUCTURED AEROGELS USING SACRIFICIAL MOLDS MADE

BY FUSED DEPOSITION MODELLING ……………………………. 177

7.1 Introduction ………………………………………………………. 178

7.2 Experimental section ……………………...... 181

7.2.1 Materials ………………………………………………. 181

7.2.2 Fabrication of sacrificial mold ....…………………..….. 182

7.2.3 Preparation of polyimide sol …………………………... 183

7.2.4 Synthesis of polyimide aerogels .………………….…… 184

7.2.5 Characterization of polyimide aerogels ………………... 186

7.3 Results and discussion ……………………………………………. 187

7.3.1 Gyroid structure ………………………………………. 187

7.3.2 Auxetic structure ……………………………………… 193

7.4 Conclusion ……………...... 195

8. OVERALL SUMMARY AND RECOMMENDATIONS FOR

FURTHER STUDY ……………………………………………………. 196

REFERENCES …………………………………………………………………. 201

xii

LIST OF FIGURES

Figure Page

2.1 General sol-gel reaction scheme consisting of (a) hydrolysis and (b)

water condensation or (c) alcohol condensation …………………… 8

2.2 (a) Acid and (b) base catalyzed hydrolysis …………...... 9

2.3 (a) Acid and (b) base catalyzed condensation …...... 9

2.4 (a) SEM of cross-section of silica aerogel and (b) schematic of “string

of pearls” ……………………………………………………………. 10

2.5 Imide formation through a two-step process using phthalic anhydride

and aniline as model compounds ………………………...... 13

2.6 Catalyst effects on polyamic acid formation ………………………... 14

2.7 (a) Flexible polyimide thin films and (b) monolithic polyimide

aerogels that can withstand the weight of a car ……………………. 17

2.8 Syndiotactic polystyrene in (a) trans-planar and (b) s (2/1) helical

conformation ………………………………………………………. 18

2.9 Temperature-concentration phase diagram of sPS in toluene. The C1

phase corresponds to the helical conformation phase ………………. 20

2.10 Schematic of spinodal decomposition followed by crystallization of

the polymer-rich phase (highlighted in red) ………………………... 21

2.11 (a) Open and (b) close cell foam structures ………………………… 24

2.12 (a) Urethane, (b) amine and (c) urea formation ……………………. 25

2.13 (a) PolyHIPE structure and (b) close up of a strut ……...... 27

xiii

Figure Page

2.14 Polystyrene polyHIPEs that are (a) surfactant-stabilized, (b)

Pickering-stabilized and (c) Pickering and surfactant-stabilized …... 29

2.15 Closed and open cell polyHIPEs due to different locus of initiation ... 31

2.16 Monomers used in fabrication of polyHIPEs with thiol-ene and thiol-

yne chemistry ……………………………………………………….. 31

2.17 Schematic of crosslinking in polyHIPEs utilizing (a) conventional

radical polymerization and (b) living radical polymerization ……… 33

2.18 Multiple release cycles of polystyrene colloid loading and release in

PNIPAM polyHIPE. (a) mass percentage of PS colloids and (b)

percentage of PS colloids ……...... 35

2.19 Schematic of droplet generation using passive and active methods 37

2.20 Passive droplet generation using cross-flow, co-flow and flow-

focusing ……………………………………………………………. 38

2.21 Schematic of various microfluidic device geometries. (a) Cross-flow,

(b) co-flow, (c) flow-focusing, (d) step emulsification, (e)

microchannel emulsification and (f) membrane emulsification ……. 41

2.22 Scheme of rapid prototyping of microfluidic channels using soft

lithography method ...... 42

2.23 Generation of monodisperse triple emulsions. (a) Schematic of co-

flow device, (b) first, (c) second and (d) third emulsification stages.

(e) Triple emulsions formed at different flow rates ………………... 43

xiv

Figure Page

2.24 (a) Flow-focusing microcapillary flow and (b) flow-focusing and co-

flow microcapillary flow ………………………………………...... 44

2.25 (a) Schematic of the PLDG device and (b) time-resolved images of

droplet generation …………………………………………………... 46

2.26 Formation of Janus droplets and microparticles using a microfluidic

device ………………………………………………………………. 47

2.27 (a) DIW of hydrogel scaffold and (b) scaffold of cells ……………. 49

2.28 (a) Selective laser melting (SLM) and (b) laser metal deposition

(LMD) ...... 50

2.29 Examples of additive manufactured aerospace components. (a)

Compressor support case for gas turbine, (b) turbine blades with

cooling channels, (c) turbine blades and (d) engine housing ………. 52

3.1 Reaction scheme for synthesis of polyimide crosslinked networks ... 59

3.2 Polyimide aerogels with varying DMF/DMAc solvent compositions

(vol%). DMAc vol% increases from left to right …………………... 63

3.3 Solid state 13C NMR spectrum of polyimide synthesized with 100%

DMAc, 100% NMP and 100% DMF ………………………………. 64

3.4 (a) IR spectrum and (b) TGA curves of polyimide synthesized with

100% DMAc, 100% NMP and 100% DMF ………………………... 65

3.5 SEM images of polyimides prepared using (a) DMF, (b) NMP and

Figure Page

3.6 BET isotherms of polyimide aerogels with increasing (a) NMP and

(b) DMAc content in the synthesis solvent ………………………… 71

3.7 Pore size distribution of aerogels with (a) NMP and (b) DMAc

content in the synthesis solvent ……………………………………. 72

3.8 Representative compressive stress/strain curves of polyimide

aerogels synthesized with 100% DMF, 100% NMP and 100% DMAc 74

3.9 Effect of surfactant concentration on DMF viscosity. Error bars are

located within the symbols. The chemical structure of F127®

surfactant is also provided in the top left corner …………………… 75

3.10 Strands of polyimide aerogels with F127® surfactant concentration

of (a) 0 vol%, (b) 0.5 vol%, (c) 2.5 vol% and (d) 5.0 vol%. Images

were taken of the surface of the aerogel monolith to facilitate

measurement of strand size ………………………………………… 76

3.11 BET isotherms of polyimide aerogels with various F127® surfactant

concentrations ………………………………………………………. 77

3.12 Maximum bubble pressure curves for neat polyimide aerogel and

polyimide gel with surfactant ………………………………………. 78

4.1 Droplet generator components (a) before and (b) after assembly. (c)

Schematic of transfer of polyimide sol droplets into heated silicone

oil bath. The inset in (c) shows a picture of gel microparticles in the

heated silicone oil bath ……………………………………………. 86

xvi

Figure Page

4.2 Droplet generation in the jetting regime. The above sequence of

images was captured using a high-speed camera at 1000 frames per

second. Dispersed phase flowrate (Qd) = 0.1 mL/min and continuous

flowrate (Qc) = 2 mL/min …………………………………………… 87

4.3 Droplet formation at varying continuous phase flowrates (Qd).

Dispersed phase flow rate (Qd) was kept constant at 0.1 mL/min …. 89

4.4 Evolution of storage (G’) and loss (G”) moduli of polyimide sol over

a period of time. The inset shows the crossover point at 1570 s (26

mins) ………………………………………………………………... 95

4.5 Optical microscope images of (a) gel and (b) aerogel microparticles.

4.6 SEM images showing (a) skin layer and internal structure of a

microparticle through sectioning, (b) porous internal structure of a

microparticle through a tear in the skin layer, (c) skin layer on the

surface of a microparticle, (d) cross-section of a microparticle, (e)

skin layer of the monolith and (f) cross-section of the monolith .…. 97

4.7 (a) IR spectra and (b) TGA traces of polyimide aerogel monoliths and

microparticles ………………………………………………………. 98

4.8 Gel and aerogel microparticle size distributions. The dispersed phase

(Qd) and continuous phase (Qc) flow rates are indicated in each chart.

Red bars correspond to gel microparticles while green bars

correspond to aerogel microparticles ………………………………. 99

xvii

Figure Page

4.9 Comparison of experimental droplet size with model prediction of

equation (6). (a) Continuous phase flow rate 2-8 mL/min, dispersed

phase flowrate 0.1 mL/min. (b) Dispersed phase flowrate 0.1-0.4

mL/min, continuous phase flowrate 8 mL/min ……...... 102

4.10 Gel and aerogel microparticle size distributions as function of

silicone oil bath temperature. Dispersed and continuous phase flow

rates were 0.3 and 6 mL/min ………………………………………. 104

4.11 BET isotherms for aerogel microparticles fabricated with varying

silicone oil bath temperatures ………………………………………. 106

4.12 Skin layer of aerogel microparticles synthesized at various silicone

oil bath temperature of (a) 60 °C, (b) 70 °C, (c) 80 °C, (d) 90 °C and

(e) 100 °C …………………………………………………………… 107

4.13 DMF/silicone oil two-phase system at various temperatures ………. 108

4.14 (a) Droplet generator for core-shell hollow microspheres, (b) optical

microscope images of hollow gel microspheres, (c) SEM image of

fractured hollow aerogel microsphere and (d) shell thickness of

hollow aerogel microsphere ………………………………………... 110

4.15 Gel and aerogel hollow microspheres size distributions for different

flow rates. The number indicated above each graph represents the

inner/shell/outer flow rates in mL/min. Red bars correspond to gel

microparticles while green bares correspond to aerogel

microparticles ………………………………………………………. 111

xviii

Figure Page

5.1 Droplet size distribution of emulsions prepared at (a) 50 C, (b) 60

C, and (c) 70 C with 1.5 vol% SPAN 80®. The emulsion prepared

at 70 C formed phase separated bilayer after 3 minutes; no dispersed

droplets could be recorded for this case, as in Figure 5.1c …………. 124

5.2 Optical microscope images of water (lighter area) emulsions with 1.5

vol% SPAN 80® at (a) 50 C, (b) 60C, and (c) 70C. Droplet size

distribution of the corresponding emulsions are presented in (d)

50C, (e) 60C, and (f) 70C ………………………………………... 125

5.3 Effect of temperature on interfacial tension of water/toluene (filled

square) and water/toluene/1.5 vol% SPAN 80® (filled triangle) …. 126

5.4 Effects of SPAN 80® surfactant concentration at 60 C: (a,d) 1.5

vol%, (b,e) 7.5 vol% and (c,f) 15 vol%. (a-c) Optical microscope

images, (d-f) droplet size distribution of the corresponding emulsions 128

5.5 Volume fraction of emulsion phase for SPAN 80® concentrations of

1.5, 7.5, and 15 vol% with time after stirring was stopped ………….. 130

5.6 SEM images of aerogel foam with (a) macrovoids and interconnects

and (b) macroporous skin layer …………………………………… 132

xix

Figure Page

5.7 SEM images of aerogel foams fabricated from emulsions with SPAN

80® concentrations of (a) 1.5 vol%, (b) 7.5 vol%, and (c) 15 vol%.

Images in (d), (e), and (f) represent higher magnification. The initial

emulsion droplet size and corresponding macrovoid size of the

aerogel foams are shown for various surfactant concentrations, (g)

1.5 vol%, (h) 7.5 vol%, and (i) 15 vol% …………………………… 133

5.8 Effect of (a) mixing speed and (b) dispersed phase content on

macrovoid size distribution. Surfactant concentration was kept

constant at 7.5 vol% ………………………………………………... 136

5.9 SEM images showing pore interconnects density for SPAN 80®

concentrations of (a) 1.5%, (b) 7.5%, and (c) 15%. (d) Macrovoid

interconnect exhibiting frustum geometry and (e) proposed

interconnect formation mechanism ………………………………… 137

5.10 SEM images of macroporous matrix of varying polymer

concentration: a) 0.02 g/mL, b) 0.04 g/mL, c) 0.06 g/mL and d) 0.08

g/mL …………………………………………………………...... 140

5.11 Skin layer formation with (a) SPAN 80 as surfactant, (b) F127 as

surfactant and c) The initial emulsion droplet size and corresponding

macrovoid size of the aerogel foams with F127 surfactant

concentration of 1.5 vol% ………………………………………… 142

5.12 Skin layer formation of monolithic sPS aerogel on a) hydrophilic

substrate and b) hydrophobic substrate ……………………………. 143

Figure Page

5.13 Interfacial elasticity (E’) and viscosity (E”) values of water in toluene

emulsions in the presence of SPAN 80®. The error bars are smaller

than the size of the symbols ...………………………………………. 146

6.1 Reaction scheme for synthesis of polyimide ……………………… 153

6.2 Droplet size distribution with (a-c) cyclohexane and (d-e) n-heptane

as the dispersed phase. Surfactant concentration are (a,d) 0.5 vol%,

(b,e) 2.5 vol% and (c,f) 5.0 vol% …………………………………... 159

6.3 Optical microscope images of n-heptane/DMF emulsion with 5 vol%

F127® surfactant concentration …………………………………….. 161

6.4 Emulsion-templated aerogel foams with cyclohexane dispersed

phase and surfactant concentration of (a) 0.5 vol%, (b) 2.5 vol% and

with n-heptane are shown in (d-g) at similar surfactant

concentrations. (g) Image of the skin layer at the macrovoid surface 162

6.5 (a) IR and (b) TGA curves of F127®, neat PI and emulsion templated

aerogel foams ...... 166

6.6 (a) Storage and loss modulus and (b) complex viscosity of neat PI

and emulsion-templated gels ………………………………………... 167

6.7 BET Isotherms of polyimide aerogel foams with varying surfactant

concentration with (a) cyclohexane and (b) n-heptane as dispersed

phase ………………………………………………………………... 169

xxi

Figure Page

6.8 Representative images of skin layers formed in emulsion-templated

aerogel foams with (a-c) cyclohexane and (d-f) n-heptane dispersed

phase with surfactant concentration (a,d) 0.5 vol%, (b,e) 2.5 vol%

and (c,f) 5 vol% ……………………………………………………. 170

6.9 Macrovoid size distribution with (a-c) cyclohexane and (d-e) n-

heptane as the dispersed phase. The surfactant concentration was

(a,d) 0.5 vol%, (b,e) 2.5 vol% and (c,f) 5.0 vol% ………………….. 171

6.10 (a) Compressive stress strain curves and (b) modulus vs bulk density

of neat polyimide, polyimide with surfactant and emulsion templated

polyimide. Graph in the insert in (a) shows the same samples at low

strains of <0.1 ……………………………………………………….. 173

6.11 Oil and Water Absorption of polyimide and emulsion templated

polyimide over time ………………………………………………… 175

7.1 Cross-section of molds and CAD models with different wall

thicknesses …………………………………………………………. 183

7.2 Procedure of polyimide aerogel synthesis …………………………. 185

7.3 Synthesis of polyimide aerogel through 3D printed hollow shells … 185

7.4 Polyimide aerogels with their hollow shells. The bottom row shows

the hollow shells, the middle row shows the aerogels fabricated from

molds with no infills and the top row shows hierarchical aerogels

fabricated from molds with infills …………………………………. 188

7.5 (a) IR spectra and (b) TGA curve of polyimide aerogel ……………. 190

xxii

Figure Page

7.6 (a) SEM image of polyimide aerogel structure and (b) BET isotherm

of polyimide aerogel ……………………………………………… 190

7.7 (a) Stress-strain curves and (b) compressive modulus vs bulk density

of polyimide aerogels ………………………………………………. 192

7.8 (a) Polyimide auxetic structure and (b-d) structure undergo tensile

deformation …………………………………………………………. 194

7.9 Stress-strain curve of auxetic structure undergoing tension ……….... 195

xxiii

LIST OF TABLES

Table Page

3.1 Elemental Analysis of polyimide synthesized with 100% DMAc, 100%

NMP and 100% DMF …………………………………………… 66

3.2 Shrinkage, bulk density, skeletal density, porosity and gel times of

polyimides synthesized with various solvent compositions …………….. 67

3.3 Surface tension, donor number (DN), acceptor number (AN) and ET(30)

of solvents. Surface tension was measured, while DN, AN and ET(30)

were obtained from Reichardt, C .………………………………………. 68

3.4 BET surface area, pore volume and micro, meso and macropore fraction

of polyimides synthesized with various solvent compositions ...... 69

3.5 Compressive modulus of polyimides synthesized with various solvent

compositions …………………………………………………………… 73

3.6 Strand diameter, BET surface area, micropore, mesopore and macropore

fraction of polyimide aerogels with F127® surfactant …………………. 77

3.7 Maximum pressure and compressive modulus of neat polyimide and

polyimide with surfactant ………………………………………………. 79

4.1 Average gel and aerogel microparticle diameter and their associated

shrinkage at different flow conditions ………………………………….. 100

xxiv

Table Page

4.2 Average gel and aerogel diameters, shrinkage, and BET surface area of

aerogel microparticles as function of silicone oil bath temperatures.

Dispersed and continuous phase flow rates were respectively 0.3 and 6

mL/min …………………………………………………………………. 105

4.3 Shrinkage, porosity and pore size distribution of monolithic aerogels

cured at various temperatures …………………………………………... 109

4.4 Average gel and aerogel hollow microspheres diameter and shell

thickness ………………………………………………………………... 111

5.1 Interfacial tension and water droplet size with 1.5 vol% SPAN 80® …. 123

5.2 Interfacial tension and droplet size with 1.5 vol% SPAN 80® …………. 125

5.3 Interfacial tension and droplet size distribution of water/toluene/SPAN

80® for different surfactant concentrations at 60 C ……………………. 127

5.4 Porosity, BET surface area, and macrovoid size distribution of emulsion-

templated aerogel foam specimens ……………………………………... 132

6.1 Interfacial tension of DMF/cyclohexane/F127® and DMF/n-

heptane/F127® systems ……………………………………………….. 158

6.2 Mean, maximum, and minimum droplet size in emulsions ……………. 159

6.3 Characteristic properties of aerogel foams as function of surfactant

concentration …………………………………………………………… 164

6.4 Average, maximum and minimum macrovoid diameters for emulsion

templated aerogel foams ………………………………………………... 172

xxv

Table Page

6.5 Compressive modulus of neat polyimide, polyimide with surfactant

concentration and emulsion templated polyimide ……………………… 173

6.6 Oil and water absorption data for neat polyimide and emulsion templated

polyimide ………………………………………………………………. 175

6.7 Micro, meso and macropore distribution of emulsion templated

polyimide …………...... 176

7.1 Bulk density and porosity of polyimide aerogel gyroids ...... 189

xxvi

CHAPTER 1

INTRODUCTION

Aerogels are a class of highly porous materials that are known for their high porosity, high surface area and small pore size distributions. Particularly, their unique pore sizes (in the range of 2 – 200 nm) are smaller than the mean free path of air molecules, meaning that convection within the highly porous structure of aerogels is virtually eliminated. This results in the particularly low thermal conductivities of aerogels, making them ideally suited as insulation materials. This property has contributed to aerogels’ wide commercial usage as thermal insulation materials in the construction industry. For example, silica aerogels have been commercialized as Spaceloft ® by Aspen Aerogels in the form of insulation blankets that can be used for building insulation or wrapped around pipes used for fluid transport. In addition to thermal insulation for the construction industry, aerogels have seen some niche applications in the aerospace industry. They are used in satellites and spacecraft in heat shielding and thermal insulation applications, such as in insulation for entry/re-entry vehicles and extravehicular activity suits. They have also seen limited use as dust collectors in satellites for capturing microscopic space dust for subsequent analysis.

This thesis research was thus focused on expansion of the use of aerogels into alternative application spaces.

To fully realize the potential of aerogels, one key limitation that needs to be overcome is the development of structure control of aerogels in the nano, micro and macro length scales, independent of the chemistry used. To date, most aerogels are produced in monolithic cylindrical forms, due to the ease of fabrication and subsequent characterization. Therefore, this work seeks to explore various structural forms that aerogels can take, to enable utilization of aerogels in other applications such as filtration, oil/water separation, drug delivery and ultrahigh stiffness, ultrahigh porosity materials.

Chapter 3 presents a method to control the pore structures of polyimide aerogel. This was achieved through modification of the solvent properties, independent of monomer and crosslinker selection. The pore structure control was achieved by manipulating the thermodynamics of phase separation, specifically, by adjusting the time gap between liquid-liquid demixing and solid-liquid demixing. This control of the time gap enabled coarsening of the polymer-rich phases which ultimately formed the fibrillar polyimide strands of the aerogel structure. The control of the time gap was achieved by manipulating the reaction rates from the solvent quality via two different mechanisms. The first method of control of reaction rates was through the solvent’s acidity, influencing the forward reaction rate and conversion. The second method was through the addition of a viscosity modifier, influencing the reaction rates physically by retarding the diffusion of monomers to the reaction site. Both these methods successfully shifted the pore size distribution of the structures from predominantly mesopores (2-50 nm) to macroporous (>50 nm). This was due to an increase of strand diameters from 9.3 nm to 30.4 nm.

In Chapter 4, a new method is presented for the fabrication of polyimide aerogel microparticles. The method circumvents typical challenges in fabrication of aerogel

2 microparticles by other methods reported in the literature, such as wide particle size distributions, non-spherical shapes and undesired agglomeration. This process utilized a simple, easily-assembled and cheap microfluidic device assembled from readily available components. This device, coupled with a heated oil bath, enabled the formation of aerogel microparticles using an oil-in-oil emulsion in a surfactant-free process. This achievement is significant as oil-in-oil emulsions are inherently unstable and are susceptible to higher rates of coalescence and Ostwald ripening, leading to undesired accelerated formation of bilayers. This process enabled the production of aerogel microparticles with average diameters of 200 µm, while retaining the high porosities characteristic of aerogels. In addition, the microfluidic flow device was improvised by incorporating an additional flow component that allowed the formation of core-shell hollow microspheres with diameters of 500 µm and shell thicknesses of 4 µm. These microparticles and hollow spheres with high porosities are ideal candidates for both drug delivery and catalyst carrier applications.

In Chapter 5, results are presented on aerogel foams developed by incorporating emulsion- templating method in conjunction with the aerogel synthesis process. The motivation for this work was to increase air permeability of the aerogel structures for filtration applications, by creating micrometer-sized voids that in turn shortens the fluid path through the structure without sacrificing filtration efficiency. A water-in-oil emulsion system was used in conjunction with a physical, thermo-reversible gelation system to incorporate micrometer-sized voids (macrovoids) into an inherently macroporous aerogel structure with typical pore dimensions in the range of 50 – 200 nm. The effects of various processing parameters such as surfactant concentration, surfactant type, mixing speed, dispersed phase volume and polymer concentration were studied. The resultant aerogel foams exhibited a

3 porosity increase from 92.5 to 97.5 %, by the incorporation of the macrovoids through emulsion-templating. In addition, various structural components such as pore interconnects and dense skin layers were investigated.

The methodology developed in Chapter 5 for emulsion-templating was extended in Chapter

6 focuses to a chemically crosslinked sol-gel system in the form of polyimides. The motivation for this research was derived from the fact that a chemically crosslinked system would impart increased flexibility the design of emulsion-templated aerogels, first due to variation of gel times and second due to adaptability of various diamines and dianhydrides.

However, the moisture sensitivity of polyimide chemistry precludes direct adaptation of water-in-oil emulsion templating as was done earlier for syndiotactic polystyrene. Instead, a separate oil-in-oil emulsion-templating system needed to be developed. This necessitated the identification of a suitable surfactant system, as well as a careful balance between emulsion stability, solvent viscosity and gel times.

In Chapter 7, the main focus was on synthesizing aerogels in various macro forms, incorporating different geometries to aerogel structures. The objective was to fabricate aerogel mechanical metamaterials mechanical properties due to geometry for a given chemistry. For example, inherently brittle aerogel structures could be rendered flexible through negative Poisson or auxetic structures, while the stiffness of aerogels can be enhanced through octet-truss or Kelvin cell geometry. To achieve the above goal, a new process of creating aerogel structures with intricate geometry, curved surfaces and recessed spaces was needed. As a result, 3D printing was brought into the process to create hollow molds, whereby the precursor sol could be injected. The sol was allowed to transition into a gel in the mold and subsequently released through dissolution of the hollow shell using

4 selected solvents. This enabled the easy release of the gel without damage to the structures.

In addition, the use of 3D printing to introduce complex geometries to the aerogel fabrication process enabled the creation of the free-standing hierarchical porous structures with porosities as high as 98.9 %. Through this process, previously brittle polyimide aerogels could achieve increased strain at break up to 17 % due to the auxetic structures.

CHAPTER 2

BACKGROUND AND LITERATURE REVIEW

2.1 Aerogels 2.1.1 Overview Aerogels are a class of highly porous material that exhibit high surface areas, high porosities, low bulk densities and small pore size distributions. The first aerogels were developed by Kistler in 1931.1 This was achieved by removing the solvent from gels (or

“jellies” as Kistler called them). However, removing the solvent from gels is no trivial matter. The pore sizes of these gels are in the range of 50 – 200 nm, and ambient solvent evaporation would lead to high capillary stresses during solvent evacuation, resulting in structure collapse and high shrinkage of the gels. This led Kistler to develop the supercritical drying process, whereby the liquid in the gel (in this case, alcohol) would be heated to above its supercritical temperature. This causes the liquid to achieve a supercritical state, eliminating interfacial tension with the solid component of the gel structure. This allows for the evacuation of the supercritical fluid without exerting capillary stresses on the porous structure. In the development of this method, Kistler managed to synthesize aerogels from a variety of materials such as alumina, nickel tartarate, stannic oxide, nitrocellulose, cellulose and even egg albumin.1

While Kistler was the first to synthesize aerogels, two challenges limited the development of these new porous materials. The first challenge was the length of time needed to synthesize these silica aerogels. Two critical developments in silica aerogel and silica chemistry allowed for widespread research and development. The first was the use of tetraethoxysilane (TEOS) instead of tetramethoxysilane (TMOS) to increase reaction rates by Teichner2, while the second development was the introduction of the two-step sol-gel preparation method by Tillotson et al.3 and Brinker et al.4 Both these developments reduced the time to make silica gels from days to hours. The second challenge limiting aerogel development was that high temperatures and pressures were required to convert alcohols

(used in silica gel synthesis) into its supercritical state. This challenge was mediated through the replacement of the fluid with liquid carbon dioxide by Tewari5 and Gowda.6

This reduced the operating and energy requirements for supercritical drying, at the cost of an additional solvent exchange step. This use of an inert liquid carbon dioxide also allowed for the development of organic aerogels that did not have high thermal stability. For example, when alcohol was used, organic gels had to be able to withstand temperatures of up to 240 °C during the supercritical drying step. In contrast, when carbon dioxide is used as a supercritical fluid, the organic gels only had to withstand temperatures of 50 °C. In addition, certain organic gels were not suited to be synthesized in alcohols due to possible undesired side reactions of the functional groups (such as carboxylic acids, amines, and anhydrides) used in step-growth polymerization chemistries.

2.1.2 Silica Aerogels

As mentioned, the first aerogels synthesized were silica aerogels. These structures are synthesized through the hydrolysis and condensation of silicon alkoxides. For example,

TEOS is reacted with water to form a metal hydroxide, which is subsequently condensed with another metal hydroxide, forming a metal oxide bond and liberating water or alcohol.

These two reactions are illustrated in Figure 2.1.

Figure 2.1: General sol-gel reaction scheme consisting of (a) hydrolysis and (b) water condensation or (c) alcohol condensation.

Both the hydrolysis and condensation reaction rates can be increased through either acid or base catalysis. The hydrolysis mechanism for both acid and base catalyzed reactions are shown in Figure 2.2, while the corresponding condensation reactions are shown in Figure

2.3.

Figure 2.2: (a) Acid and (b) base catalyzed hydrolysis

Figure 2.3: (a) Acid and (b) base catalyzed condensation

Depending on the catalysis process, different silica morphologies can be manifested. In acid catalyzed reactions, condensation reactions are preferentially conducted at terminal silicons. In addition, condensation reactions are faster than hydrolysis reactions. This results in primarily chain extension reactions, forming long, linear chains that are weakly branched or crosslinked. In contrast, base-catalyzed condensation reactions tend to occur at interior silicons, leading to branched colloidal particles in the sol. It is to be noted that an important milestone in silica aerogel chemistry was contributed by Brinker et al., who developed a two-step synthesis method, incorporating an initial acid-catalyzed step, followed by a base-catalyzed step.4 The importance of this innovation was the shortening of gel formation times by incorporating the different aspects of both catalysis step. The principal concept behind this two-step process was to promote rapid chain extension in the

9 first acid-catalyzed step, followed by a shift in focus to form crosslinked network systems through a base-catalyzed step. The introduction of the basic environment in the second step also promotes phase separation and faster gelation times.

The final structure of the silica aerogels can be said to occur in three stages, according to

Iler7: the first step involves polymerization of the monomers through hydrolysis and condensation reactions to form primary particles that are less than 1 nm in diameter. These chemistries have been described above. The second step involves agglomeration of these primary particles into secondary colloidal particles of 5 – 10 nm. These secondary particles are then connected through Si-O-Si bonds between the contact points of two particles, also known as “necks” in the third step.8 These secondary particles continue to grow and eventually link to form a chain, commonly described as a “string of pearls”.7 In these stages, reaction conditions such as temperature, pH, reaction times, catalyst type and monomer concentrations affect the final silica aerogel structure.9 The nanostructure of silica aerogels and a schematic of the “string of pearls” structure are shown in Figure 2.4.

Within the field of silica aerogels, a variety of silicon alkoxides and solvents have been used to develop aerogels with different mechanical properties and nanostructure.11 For example, monomer systems that can be used include tetramethoxysilane (TMOS),12 tetraethoxysilane(TEOS), polyethoxydisiloxane (PEDS),13 methyltriethoxysilane

(MTES)14 and methyltrimethoxysilane (MTMS).15

One drawback of silica aerogels has been their brittle nature. This is due to weakness of the “necks” in the “string of pearls” structure as the contact area between the secondary colloidal particles is small (contact between two spheres). In order to expand applications of silica aerogels, significant efforts have been undertaken to reinforce silica aerogels to better withstand mechanical stresses. Kramer et al. first started by reinforcing TEOS aerogels with hydroxy terminated polydimethylsiloxane (PDMS), achieving up to four times increase in compressive modulus compared to unreinforced silica aerogels. The resulting “aeromosils” showed increased flexibility and elongation at break.16 Leventis et al. also proceeded by introducing isocyanates to react with the residual -OH groups to form polyurethane crosslinks. This method successfully increased the compressive modulus of the silica reinforced aerogels by more than 300 times.17 Extensive work was also done by

Meador et al. in reinforcing silica aerogels. One method involved the inclusion of amine functional groups into the structure through the copolymerization of aminopropyltriethoxysilane (APTES). These amine functional groups were subsequently reacted with isocyanates to form polyurea shell coatings around the silica particles. These provided significant improvement in reinforcement compared to polyurethane functional groups.18,19 This was also attempted with epoxies as the crosslinker instead of isocyanates.20

In addition to monolithic silica aerogels, silica aerogel films have been fabricated to suit electronic applications.21 These aerogel films can be fabricated through tooling, spin coating and dip coating.22

2.1.3 Polyimide Aerogels Polyimide aerogels were first described by a patent submitted by Aspen Aerogel Inc. in

2006.23 Their high thermal, hydrolytic and radiation stability, coupled with good mechanical strength and electrical properties make them ideal materials for aerogels in demanding applications and environements.24,25 These aerogels are synthesized from a two-step polyimide synthesis process pioneered by DuPont.26 The first step involves the formation of amic acid groups by the reaction of anhydrides with amines. Addition of multifunctional monomers in this step allows for the formation of crosslinked network structures. This is subsequently followed with the second step, where the amic acid functional groups are converted to the imide functional groups through dehydration of a water molecule. Figure 2.5 shows the reaction scheme of polyimide formation using phthalic anhydride and aniline as model compounds. In addition to the two-step synthesis process outlined above, Leventis et al. also showed that polyimide aerogels can also be accomplished through a ring-opening metathesis polymerization (ROMP), eliminating the need for a second imidization step.27

Figure 2.5: Imide formation through a two-step process using phthalic anhydride and aniline as model compounds

The first reaction step is in actuality a reversible reaction, with a very fast forward reaction that is several magnitudes larger than the reverse step. The reaction rates and conversion of the first reaction depends very much on the monomers used in the synthesis. The first reaction is a nucleophilic acyl substitution reaction, whereby the nucleophilic amine group attacks the electrophilic carbon in the carbonyl group. Consequently, the more electrophilic the anhydride group is (as represented by its electron affinity value) would lead to faster amic acid formation. As a result, dianhydrides such as pyromellitic dianhydride (PMDA)

(Ea = 1.90 eV) and 3,3’,4, 4’-benzophenonetetracarboxylic dianhydride (BTDA) (Ea =

1.55) have faster reaction rates when compared to biphenyl-tetracarboxylic acid dianhydride (BPDA) (Ea = 1.38).28 Conversely, nucleophilicity of the amine similarly would also have an effect on the reaction rate.

The polyamic acid reaction rate is strongly influenced by solvent selection. Being a nucleophilic substitution reaction, polyamic acid formation is significantly sped up in dipolar aprotic solvents. The polarity of the solvents stabilizes the transition states, but its aprotic nature means that they have weaker anion solvating power and structural effects, allowing for the reaction to proceed forward.29,30 As a result, popular solvents for polyimide synthesis include tetrahydrofuran (THF), dimethyl formamide (DMF), dimethylacetamide

(DMAc) and n-methyl-2-pyrrolidone (NMP). The polyamic acid formation rate is found to

13 be higher in acidic solvents such as acetonitrile and m-cresol, compared to basic solvents such as THF.31 This reaction is also found to be autocatalytic in nature, where the acidic polyamic acid serves to accelerate the reaction. This is evident from the sigmoidal-shaped concentration versus time curve shown by Kaas.32

Inclusion of water in this reaction is detrimental to the molecular weight and mechanical properties of the resultant aerogels. The water molecules will compete with diamines to react with the dianhydrides to form dicarboxylic acids. These diacids do not react with the diamines (or at a significantly lower reaction rate at ambient temperature), resulting in overall lower molecular weight and conversion of the amic acid formation.

As mentioned earlier, after the polyamic acid formation step, the amic acid groups are converted to the imide functional groups. This imidization step can be conducted either

14 thermally or chemically. The first polyimides were thermally imidized, by increasing the temperature of the system gradually to 200 °C. This closes the amic acid ring by liberation of a water molecule. However, thermal imidization brings about complications in terms of solvent volatilization and lowering of molecular weight. Both of these problems can be circumvented through chemical imidization. In this process, the cyclohydration reaction is carried out by introducing a dehydrating agent such as acetic anhydride, benzoic anhydride, or n-butyric anhydride. Catalysts such as pyridine, triethylamine, methylpyridines, and n- methylmorpholine are added to speed up the chemical imidization reactions. The most common dehydration agent and catalyst pairing is that of acetic anhydride and pyridine.33,34

The initial polyimide aerogels fabricated were found to be lacking in mechanical strength.

In an effort to address this issue, Kawagishi utilized a trifunctional amine to crosslink these linear chains made from bifunctional anhydrides and amine. This resulted in the formation of polyimide aerogels that were capable of more than 90 % porosity due to the increase in mechanical strength.35 The mechanical properties and nanostructure of polyimide aerogels can be varied through selection of monomer systems. For example, Meador et al. investigated the mechanical properties of polyimide aerogels using a variety of diamines and dianhydrides.36 In this case, monomers with flexible ether linkages in the backbone

(e.g. 4,4-oxydiphthalic anhydride (ODPA) and 4,4-oxydianiline (ODA)) provided flexibility in the aerogel structures, while monomers composed of predominantly aromatic groups (e.g. pyromellitic dianhydride (PMDA) and p-phenylenediamne (PDA)) resulted in more rigid aerogels. In addition to changing monomers in polyimide aerogel synthesis, different crosslinkers have been reported. Examples of crosslinkers used include 1,3,4- tris(4-aminophenyl)benzene (TAB),35 2,4,6-tris(4-aminophenyl)pyridine (TAPP),37

15 polyoligomeric silsesquioxane (POSS) molecules with amine functionality33 and 1,3,5- benzenetricarbonyl trichloride.38

Another approach to improving polyimide aerogel mechanical properties has been the inclusion of additives during the sol-gel process. Vivod et al.39 included carbon nanofibers in polyimide aerogels synthesized from biphenol-3,3’,4,4’-tetracarboxylic dianhydride

(BPDA) and 4,4’-oxydianiline (ODA). The presence of carbon nanofibers in the polyimide aerogel was found to decrease shrinkage, reduce density and increase toughness and flexibility of the aerogel.39 In another study, Nguyen et al. added cellulose nanocrystals extracted from tunicates (marine invertebrate animal) to BPDA and ODA polyimide systems to increase tensile modulus of the resultant aerogels.40 Another additive that was studied was clay. Wu et al. added 5 wt% of clay to a PMDA and ODA polyimide system and observed a 50 % increase in compressive modulus.41

Due to the flexibility in monomer selection of a step-growth polymerization, a variety of different physical and chemical properties can be designed into polyimide aerogels. For example, Meador et al.42 sought to reduce the dielectric constant of polyimide through the use of a fluorinated dianhydride monomer (2,2-bis(3,4-dicarboxyphenyl) hexafluoropropane dianhydride). This resulted in a decrease of relative dielectric constant from 1.30 to 1.084, making these aerogels potential candidates for antenna substrates to improve antenna gain and bandwidth.37,42,43 Owing to their superior properties, polyimide aerogels have been deployed in aerospace applications. They have been used as insulation for entry/re-entry vehicles and extravehicular suits.34 Figure 2.7 shows that depending on the material system used, polyimide aerogels can be fabricated as either flexible thin films or monolithic blocks that can withstand the weight of a car.

Figure 2.7: (a) Flexible polyimide thin films and (b) monolithic polyimide aerogels that can withstand the weight of a car. Reproduced with permission from ref 36. Copyright 2012 American Chemical Society.

2.1.4 Syndiotactic Polystyrene Aerogels

Syndiotactic polystyrene (sPS) exhibits unique polymorphism owing to its tacticity. The regular tacticity of sPS enables it to incorporate solvent molecules in a crystalline helical structure that allows for quick gelation. This is unlike other forms such as atactic polystyrene, where gels are formed at subzero temperatures due to its irregular conformation.44 Isotactic polystyrene gel formation has also been studied and its behavior straddle between the syndiotactic and atactic forms, with the ability to form stable gels at room temperature. However its long gel times of the order of days do not make it attractive for gel formation studies.45 Therefore, sPS is of particular interest in aerogel studies due to its quick gel formation kinetics.

The stereoregularity of syndiotactic polystyrene can be achieved through the use of

Ziegler-Natta catalysts such as TiCl4 and TiBr4, or metallocene catalysts such as titanium- based catalysts with mono or bis-cyclopentadienyl ligand substitutions. As typical of

17 metallocene catalyst systems, methylaluminoxane (MAO) is added to act as a catalyst activator in the polymerization of syndiotactic polystyrene.46

Due to its specific tacticity, sPS is able to achieve two different conformations. The first conformation is the all-trans planar conformation, manifested in the α and β crystalline form. These crystalline forms can be accessed through melt processing or thermal annealing. The second form is the s (2/1)2 helical conformation that can only be accessed through solvent processing. This form is manifested in the γ, δ and ε crystalline forms, with the γ form being able to host guest solvent molecules to form intercalates. Removal of the guest molecules from the γ form results in the nanoporous δ form. These two different conformations are shown schematically in Figure 2.8.

It is the latter helical form that is of interest in gel formation of sPS. These helical forms are able to crystallize quickly due to the regular tacticity, in addition to forming co-crystals with the solvent. This makes this system attractive for study due to the ease and short time frames of gel formation. As mentioned, these helical forms are attained through solvent

18 processing, and the type of solvent is important. A variety of sPS/solvent systems have been studied to form thermo-reversible gels. Solvents studied include chloroform,48 benzene,49 toluene,50 tetrahydrofuan,51 chlorobenzene52 and 1,2-dichloroethane.53

Due to its polymorphic behavior, special care must be taken to access the helical crystalline structure without locking in the planar crystalline forms. The helical conformations can be accessed through two different methods. The first is to conduct a temperature quench in a good solvent. In good solvents, the helical conformations exist as a separate thermodynamically stable phase that can be accessed from the solution through cooling. In these systems, cooling from a one-phase phase system at low polymer concentration is sufficient to effect crystallization into the helical form, thus forming the gel. Examples of good solvents include tetrahydrofuran and chloroform. An example of a phase diagram of a good solvent is shown in Figure 2.9. In bad solvents, the thermodynamically stable phase is the planar conformation. However, the helical configuration can theoretically still be accessed through rapid quenching. It is the first system (good solvent) that is of interest for sPS gel formation due to its ease in attaining the helical conformation.

Figure 2.9: Temperature-concentration phase diagram of sPS in toluene. The C1 phase corresponds to the helical conformation phase. Reproduced with permission from 50. 50 Copyright 1997 Elsevier Science Ltd

As mentioned, the helical structure can be accessed easily through the use of a good solvent. This results in the sPS/solvent system to form gels through a thermo-reversible process. To achieve this, sPS is first dissolved in a good solvent at low concentrations and heated to form a one-phase solution. For example, based on the phase diagram in Figure

2.9, this one-phase system is formed by heating to above 110 °C, at sPS concentrations of less than 0.1 g/cm3 in toluene. This system is then cooled to pass through the binodal and spinodal curves (not shown in the figure). The system undergoes liquid-liquid demixing to phase separate into polymer-rich and solvent-rich regions through spinodal decomposition.54 The system then rapidly crystallizes and drops below the glass transition temperature and vitrifies, locking in the fibrillar polymer structure.55 The vitrified structure is thereafter termed the gel. This is schematically shown in Figure 2.10, on the boxes on

20 the right highlighted in red. This gel can subsequently undergo solvent exchange and supercritically dried to form sPS aerogels.

In addition to the sPS, studies have also been conducted on sulfonated syndiotactic polystyrene (ssPS) aerogels. Wang et al.56 sulfonated 10 % of the polystyrene pendant groups in sPS by adding acetyl sulfate to the dissolved sPS and allowed the system to react at 70 °C for 3 hours. The presence of styrenesulfonic acid units in the polymer after sulfonation impedes crystallization and increase gel times of ssPS due to the formation of ionic domains within the structure. However, the successful inclusion of sulfonic acid groups increased moisture uptake and opened up the potential of sPS aerogels for supercapacitors and other applications such as fine charged particle separation.56 This increase in water uptake rate and capacity was also reported by Venditto et al.57 In this

21 study, ssPS was shown to absorb 1,2-dichloroethane by more than 3 orders of magnitude, making ssPS a candidate for absorbing VOC molecules from water.57 Another modification of sPS aerogels has been the fabrication of hybrid aerogels with silica. Wang et al.58 first prepared a sPS gel and allowed the diffusion of silica sol into the network structure. The silica sol was then allowed to form its own network structure in situ within the sPS gel.

The resultant sPS/silica hybrid aerogel exhibited 1.5 to 2 times increase in surface area due to the increased mesoporosity attributed to the introduction of the silica structure. The resulting aerogel also exhibited high compressive modulus and fast crude oil uptake.58

Owing to their ability to include guest molecules in the nanopores found in the δ and ε crystalline forms, sPS aerogels have been deployed in molecular gas separation, particularly in removing VOCs from air.59,60 One particular application of this has been the removal of ethylene and carbon dioxide from the environment to preserve the shelf life of fruits and vegetables.61 This high absorption kinetics also make them potential candidates for chemical separation and water purification applications.62 This ability for the nanoporous sites to absorb guest molecules has also been used as chemical sensors for

VOC vapors.63 Kim et al. has also recently shown that sPS aerogels can be used for nanoparticle filtration, resulting in high rejection of 75-150 nm salt particles.64–67

2.2 Foams

2.2.1 Overview of Foams

Foams are a type of porous structure developed in the early 1930s that incorporates micrometer-sized voids in a polymer continuous phase. Foams have seen success in a

22 variety of applications such as cushioning, insulation, weight-bearing structures and as absorbents.68 Foams have also been fabricated from a variety of polymeric materials such as polyurethane, polystyrene, poly(vinyl chloride), polyolefins and polycarbonate.

Depending on mechanical properties, cell morphology and material used, foams can be classified as either rigid or flexible foams. Rigid foams are typically used in applications such as building insulation, packaging and floatation devices, while flexible foams are used as bedding, cushioning, sports equipment and acoustic absorption materials. Another way to classify foams is in the dimensions of the cells within the structure. Macrocellular foams have cell dimensions of more than 100 µm, microcellular foams have cell dimensions of 1

– 100 µm while ultramicrocellular foams have cell dimensions of 0.1 – 1 µm.

In addition, foams can be categorized as either open or close cell foams. Open cell foams are formed when the lamella collapses due to bubble impingement during foam formation.

These structures are by nature interconnected, and load is born by the remaining struts.

They exhibit poor mechanical properties, feel soft and are used in cushioning applications such as mattresses and acoustic dampening materials. In contrast, close cell foams are characterized by compartmentalized voids that are not interconnected and are achieved when the intervening continuous phase lamella does not rupture during foam formation.69

Close cell foams are typically rigid and have higher density and superior insulation properties compared to open cell foams. Both these foam types can have different shape of the cell voids depending on the volume fraction of the dispersed phase. For example, at low dispersed phase volume content, the bubbles are spherical in nature and the resultant foams are close cell. At high dispersed phase volume content, neighboring bubbles impinge

23 onto each other and deform, assuming a polyhedral shape such as rhomboidal dodecahedrals or tetrakaidekahedrals as shown in Figure 2.11.70

The cells or voids in foams are formed by dispersing a gas in the polymer phase while it is still fluid. The introduced gas can be generated through a few methods: as a byproduct of chemical reaction, through decomposition of additives or vaporization of organic solvents.72 In addition to the introduction of dispersed phase gas, other additives such as nucleating agents and surfactants are also included in the foaming process to assist and stabilize bubble formation.

2.2.2 Polyurethane Foams

Polyurethane foams are the most ubiquitous foams currently used. Polyurethane foams are synthesized via the reaction of isocyanates with alcohols. In addition to the formation of urethane functional groups, water in the system reacts with isocyanates to form amine groups, liberating carbon dioxide gas in the process. The newly formed amine groups can also react with isocyanates present to form urea linkages. The carbon dioxide liberated in

24 the conversion of isocyanates to amines are used as in situ blowing agents in the formation of polyurethane foams. The reaction schemes of urethane, amine and urea formation is shown in Figure 2.12.

Figure 2.12: (a) Urethane, (b) amine and (c) urea formation

Additional blowing agents such as chlorofluorocarbons (CFCs) or volatile organic compounds (VOCs) can also be added to tune foam density for the desired application.

Surfactants are also added to stabilize the gas bubbles and prevent bubble coalescence.

2.3 PolyHIPEs

2.3.1 Overview of PolyHIPEs

Another class of porous materials are poly high internal phase emulsions (polyHIPEs).73

Unlike the previously described foams, polyHIPEs are fabricated from a templating

25 emulsion. As the name suggests, polyHIPEs are synthesized from a starting high internal phase emulsion (HIPE). HIPEs are defined as an emulsion with a dispersed phase volume content of more than 74%, dispersed in a continuous phase.74 Due to this high dispersed phase content, HIPE formation and stabilization is no trivial matter. In this respect, two factors are crucial for successful HIPE formation. The first is that the dispersed phase should be carefully added to the continuous phase to prevent unnecessary droplet jostling that could result in droplet coalescence. This is typically achieved through dropwise addition of the dispersed phase, or through the use of microfluidics. The use of microfluidics also introduces monodisperse droplets that enable easier packing to achieve high dispersed phase content. The second factor for successful HIPE stabilization is the careful selection of both the surfactant and emulsion system. In certain situations, surfactant pairs are used to improve emulsion stability of HIPEs, as shown by Yaron et al.75 Once the HIPE has been appropriately stabilized, the continuous phase is polymerized to form a solid, crosslinked structure. Subsequent evacuation of the dispersed phase results in the formation of a polyHIPE. PolyHIPEs are characterized by high porosity and the voids are interconnected by pore interconnects (also called windows). An example of a polyHIPE is shown in Figure 2.13. In addition, a variety of structural forms have been developed for polyHIPEs such as monoliths, beads76,77 and membranes.78

2.3.2 High Internal Phase Emulsion (HIPE)

As the emulsions used in the fabrication determines the resultant structure of the polyHIPE, the type of emulsion used is important. The most common emulsions used are water-in-oil emulsions, with oil-in-water emulsions gaining popularity.80 Oils are used as the continuous phase as the monomer systems selected are typically oil soluble. Water is used as the dispersed phase due to its immiscibility with the selected continuous phase and the relative ease of forming emulsions. There have also been inroads in using oil-in-oil emulsions for polyHIPE fabrication.81 An example of this includes a system with an organic continuous phase and an ionic liquid as the dispersed phase.82 In addition, miniemulsions have also been used to form dispersed phase droplets of 30 – 500 nm through a combination of high shear methods obtained from sonication and a kinetically stable emulsion using a surfactant and co-stabilizer system.83

Apart from the phases of the emulsion used, emulsion stabilization is critical for the formation of HIPEs. This emulsion stabilization can be achieved through amphiphilic materials such as surfactants or Pickering particles. The first consideration in surfactant 27 selection is Bancroft’s rule, where the surfactant should be more soluble in the desired continuous phase.84 Since most polyHIPEs utilize W/O emulsions, common surfactants used to stabilize these emulsions are nonionic in nature, with recent usage of cetryltrimethylammonium bromide (CTAB).85 It has also been reported that inclusion of co-surfactants soluble in the dispersed phase do help in stabilizing HIPEs. In these systems, two (or more) surfactants are used, with at least one surfactant that is preferentially soluble in the continuous phase and another surfactant exhibiting preferential solubility in the dispersed phase. For example, Cohen et al. successfully demonstrated this concept by adding water-soluble carbon nanotubes and carbon black to the water dispersed phase in the synthesis of polyHIPEs.86 Due to the high dispersed phase in HIPEs, these emulsions suffer from poor stability and need up to 30 wt% of surfactants to stabilize. This high surfactant loading levels could potentially lead to high costs in polyHIPE synthesis, difficulty in surfactant removal and undesired retention of surfactant in the structure leading to problems in downstream applications.87,88 This has led some researchers to explore Pickering emulsions, where HIPEs are stabilized by amphiphilic solid particles at lower loading levels of 1 – 5%.89 In surfactant-stabilized emulsions, the surfactant lowers the interfacial tension between the phases, allowing for increased interfacial area in the system. In contrast, Pickering-stabilized emulsions work on the concept of solid particles migrating to the interface between the phases, forming a rigid shell to prevent droplet coalescence. In this respect, polyHIPEs have been successfully synthesized with silica, titania and carbon nanotubes.90,91 One added benefit of Pickering HIPEs is that the nanoparticles used to stabilize the Pickering emulsions can be tuned to fit different requirements. This is achieved through functionalization of the nanoparticle through

28 techniques such as silane-modification of silica. Parameters that can be tuned include silane chemistry, extent of surface modification, nanoparticle content and mixing intensity.92 All these parameters affect the HIPE to be stabilized and polymerization kinetics, leading to different effects on the resultant polyHIPE structures. Comparatively, polyHIPE stabilized with Pickering nanoparticles exhibit larger voids in the structure (due to the size of the nanoparticles compared to surfactants) and are close celled in nature (due to formation of a nanoparticle shell). On the other hand, polyHIPE stabilized with surfactants exhibit smaller voids and are open-cell in nature. Ikem et al. showed that by using a combination of both nanoparticles and surfactant, a polyHIPE with both characteristics of surfactant and

Pickering stabilized emulsions can be found.93 Figure 2.14 shows that surfactant-stabilized polyHIPEs have pores of about 10 µm, Pickering-stabilized polyHIPEs have pores sizes of

200 µm and surfactant and Pickering-stabilized polyHIPEs have open cell pores of 100

µm.

Figure 2.14: Polystyrene polyHIPEs that are (a) surfactant-stabilized, (b) Pickering- stabilized and (c) Pickering and surfactant stabilized. Reproduced with permission from 93.93Copyright 2010, John Wiley and Sons.

2.3.3 Polymerization of PolyHIPEs

One key step of polyHIPE fabrication is the polymerization of the continuous phase. Free radical polymerization is the most commonly used, with polymer systems of polymethylmethacrylate or polystyrene.94 In this case, the locus of initiation for the polymerization reaction has a significant effect on the shape of the resultant polyHIPEs.

For example, if initiation happens at the interface, through the use of a water-soluble initiator, the resultant polyHIPE structure becomes polyhedral in shape. This is as the interface is “locked in” prior to the bulk continuous phase and the polyhedral structure of the starting HIPE is preserved. In contrast, if an organic soluble initiator is used, the bulk of the continuous phase polymerizes before the interface. This allows for rearrangement of the interface through coalescence and Ostwald ripening, leading to spherical voids.92 A study that illustrates this point was conducted by Quell et al.95 In this study, the polyHIPE was synthesized from a styrene and divinylbenzene (DVB) system with both a water- soluble (potassium persulfate, KPS) and oil-soluble (azobis(isobutyronitrile), AIBN) initiator. Figure 2.15 shows that the interfacial-initiated polymerization using KPS exhibited a closed cell structure. In addition, a polyhedral, thicker cell wall is observed.

This is attributed to an osmotic pressure gradient, which causes material distribution during polymerization. In contrast, the bulk-initiated polymerization produced open cell structures with spherical pores.

Aside from free radical polymerization, other chemistries have been used in the synthesis of polyHIPEs. One such chemistry is that of thiol-ene and thiol-yne. This was achieved through the photopolymerization of a triacrylate (trimethylolpropane triacrylate) with a trithiol (trimethylolpropane tris(3-mercaptoproprionate)). Polymerization was also carried out with octadiyne and the trithiol. Monomers used in this study are shown in Figure 2.16.

This system enables higher crosslinking in the structure, leading to improved mechanical properties such as greater strength and toughness.96

Another chemistry being used is reversible addition-fragmentation chain transfer (RAFT).

The driving force behind this study was to increase the mechanical properties of polystyrene polyHIPEs, specifically the compressive modulus and crush strength.97

Traditional free radical polymerization results in intramolecular crosslinking as the polymer chains grow. However, chain growth occurs in isolation because there is little overlap with neighboring chains. As the reaction proceeds, microgels form and these microgels crosslink intermolecularly over time. These network of microgels are comparatively weak due to the small amount of intermolecular crosslinking and contact area between the microgels. In contrast, living radical polymerization converts all the initiators into growing polymer chains at the start. These chains grow simultaneously, and a higher rate of intermolecular crosslinking happens throughout the polymerization process. This ultimately leads to a homogeneous crosslinked gel network. A schematic of the two different processes is shown in Figure 2.17. Comparatively, RAFT polymerized polyHIPEs exhibited compressive modulus 3.6 times that of regular free radical polymerized polyHIPEs. The crush strength was also improved by 3.8 times.97

Figure 2.17: Schematic of crosslinking in polyHIPEs utilizing (a) conventional radical polymerization and (b) living radical polymerization. Reproduced with permission from 97. 97. Copyright 2012 Royal Society of Chemistry

Ring-opening metathesis polymerization (ROMP) has also been attempted for polyHIPE synthesis by Deleuze et al.98 The authors seek to use ROMP for polyHIPE synthesis owing to the availability of both unsaturated constrained ring monomers that are hydrophobic and can be used in the continuous phase of W/O emulsions. Most importantly, ROMP has a living character and the terminal ends are expected to remain active, allowing for the grafting of a second polymer chain after polyHIPE formation. In this system, tetracyclo(6.,2.,1.,0)dodeca-4,9-diene (BVD) was used as the monomer with a RuCl2-

98 (PCy3)2(=CHPh) initiator that has high reactivity and good water tolerance.

In addition to the various chain growth polymerization reactions, step-growth condensation polymerization has also been attempted to synthesize polyHIPEs. David and Silverstein successfully synthesized polyurethane polyHIPEs through reaction of a diisocyanate with a trifunctional -OH terminated polycaprolactone (PCL-T). This system was chosen to

33 enable the fabrication of a biodegradable polyHIPE for potential tissue scaffolding applications. In this study, water was also used to convert the isocyanates to amines,

99 releasing CO2 to form a hierarchical porous structure.

2.3.4 Applications of PolyHIPEs

PolyHIPEs see interest in a multitude of applications, due to their high porosity and interconnected pore structure. In tissue engineering applications, porous materials are seen as ideal candidates for scaffolding applications. PolyHIPEs in particular are seen to be good candidates due to their highly interconnected porous network. As a result, biodegradable polyHIPEs synthesized from polystyrene and polycaprolactone with millimeter sized voids have been used for mouse skeletal stem cell scaffolding.100 Thermo-responsive poly(N- isopropylacrylamide) PNIPAM polyHIPEs have also been synthesized for drug release applications. These polyHIPEs absorbed large amounts of water below the lower critical solution temperature (LCST), and subsequent expel the absorbed water above the LCST, when the swollen structure shrinks. This pump-like action was tested by loading polystyrene colloids into the polyHIPE below LCST and observing the release of these particles from the structure above LCST.101 Figure 2.18 shows the release profile of the polystyrene colloids in multiple release cycles due to the pump-like action of PNIPAM polyHIPEs.

Figure 2.18: Multiple release cycles of polystyrene (PS) colloid loading and release in PNIPAM polyHIPE. (a) mass percentage of PS colloids and (b) percentage of PS colloids. Reproduced with permission from 101. 101Copyright 2010 American Chemical Society.

Another application space of porous polyHIPEs are for use as membranes. Work has been done to vary wall thicknesses (through dispersed phase content) to tune mechanical strength and gas permeability to enable polyHIPEs to be used as membranes.102 In this work, porosity was varied from 25 to 84 % through the volume of dispersed phase introduced in the emulsion formation process. Foams with low dispersed phase content

(poly low internal phase emulsions, polyLIPEs) had improved compressive modulus of

350 MPa and crush strength of 19 MPa, as expected with the increase in foam density.

These polyLIPEs still exhibited interconnected pores that allow for nitrogen and mercury permeation, but at significantly lower levels compared to poly medium internal phase emulsions (polyMIPEs) and polyHIPEs.

Tunc et al.103 synthesized poly(isodecyl acrylate) polyHIPEs for used in a chromatography columns to separate alkylbenzenes from acetonitrile.103 In this study, isodecyl acrylate and

DVB were used as the monomers for polymerization with a W/O emulsion. This HIPE was fed directly into a pretreated fused silica capillary and polymerized in situ. The resultant monolithic polyHIPE in the column had a porosity of 90% with interconnected voids. This shows the potential of polyHIPEs to replace silica in chromatography columns.

Pulko et al. used polyHIPEs attached with the organocatalyst 4-dimethylaminopyridine

(DMAP) for the acylation of methylcyclohexanol, reporting 100 % conversion.104 In this study, 4-vinylbenzyl chloride and DVB were polymerized to form the continuous phase of the polyHIPE. The unreacted benzyl chloride groups act as sites for the attachment of the

DMAP catalyst onto the polyHIPE support. This anchored, recyclable, nucleophilic catalyst was subsequently used in acylation reactions, reporting efficient conversions.

Amine functionalized PGMA-based polyHIPEs have also been investigated for absorption of metal ions such as silver and chromium.105 In this study, Mert et al.105 used polyester resin, DVB and styrene to form a polyester-based polyHIPE that had porosity of 85 %. In addition, glycidyl methacrylate was added to insert epoxy functionality to the polyHIPE structure. These epoxy groups were then functionalized with multifunctional amines such as 1,6-hexamethylenediamine, 4-aminosalicylic acid, 2-aminothiazole and 4- aminobenzothiazole with conversions between 3 – 27%. These amine-functionalized polyHIPEs were then successfully used as chelating agents to remove metal ions such as

Ag(I), Cu(II) and Cr(III).

2.4 Microfluidic Droplet Generation

2.4.1 Overview of Microfluidics

Microfluidics has emerged as a powerful tool to generate uniform droplets for a variety of applications. This is as the flows and forces in a microfluidic setup can be controlled and replicated for each individual droplet, ensuring that each droplet experiences almost identical flow conditions, resulting in monodisperse droplet distributions. This is in

36 contrast to traditional mechanically mixing, which relies on Rayleigh instabilities to form droplets due to the shearing of two immiscible phases.106 Within microfluidic droplet generation, there is a rich field of different methods and mechanisms that can be utilized.

These methods can first be split into two broad categories of passive and active droplet generation. Figure 2.19 shows the different passive and active methods in microfluidic droplet generation.

2.4.2 Passive Droplet Generation Regimes

Passive droplet generation methods are the most ubiquitous, owing to their simple geometry and ease of operation. Passive droplet generation can be categorized into five regimes of squeezing,108 dripping,109 jetting, tip-streaming110 and tip-multi-breaking.111

The transitions between these regimes are controlled by the capillary number of both the dispersed and continuous phase. Four out of the five regimes are controlled by Rayleigh- 37 plateau instability, with the exception of the squeezing regime, where confinement effect due to the geometry plays a much larger role in droplet generation. In general, droplets in the squeezing regime are generally larger than the channel dimensions. In the other four regimes, droplet generation is achieved through the interplay between interfacial, inertial and viscous forces. Figure 2.20 shows the various different regimes with three microfluidic geometries. The different microfluidic geometries of cross-flow, co-flow and flow- focusing will be discussed later.

Figure 2.20: Passive droplet generation using cross-flow, co-flow and flow-focusing. Reproduced from 112. 112Copyright 2010 American Chemical Society. Reproduced with permission from 113.113Copyright 2007 American Physical Society. Reproduced with permission from 109 109Copyright 2007 American Physical Society. Reproduced with permission from 114 114Copyright 2013 Cambridge University Press.

As the continuous phase capillary number increases, the breakup mechanism transitions from the squeezing regime to the dripping regime.115 Viscous forces drag the disperse

38 phase out, dominating over the interfacial forces. Unlike the squeezing regime, the interfacial forces break off the droplets due to large viscous forces, before the droplet can grow to the size of the channel. The droplets are generated at the tip of the nozzle (or intersection of the dispersed and continuous phase) and are transported down the channel for collection. The resultant droplets exhibit high monodispersity.

When either the continuous or disperse phases capillary number is increased (through flowrate), the regime transitions from a dripping to a jetting regime.109 This regime is characterized by an extension of the dispersed phase away from the nozzle, forming a jet.

Subsequently, Rayleigh instability occurs within the extended jet, breaking up into the droplets. Within this regime, jetting can be subdivided into the narrowing-jet and widening- jet regime. The narrowing-jet regime is controlled by a higher continuous phase velocity compared to the dispersed phase velocity. The faster flowing continuous phase extends the jet, leading to a more monodisperse droplet distribution.116 In contrast, the widening-jet regime is a result of a faster dispersed phase velocity compared to the continuous phase.

This difference in velocity results in a retardation of the dispersed phase, leading to accumulation of dispersed phase downstream, creating polydisperse droplets.117 An example of a narrow-jet and widening-jet in a co-flow droplet generation setup is shown in Figure 2.20.

The next regime is tip-streaming. In this regime, the continuous phase extends the dispersed phase jet into a conical shape, whereby the droplets break off from the apex.118,119 This results in droplets that are smaller than the original nozzle.120 Unlike the jet in the jetting regime, the jet in this regime takes on a cylindrical shape with near constant diameter.110

Four conditions need to be met in order to realize tip-streaming: (1) the geometry should

39 either be co-flow or flow-focusing,121 (2) the continuous phase should exhibit creeping flow, (3) the continuous phase flow rate should be higher than the dispersed phase flow rate, and (4) the average dispersed phase and continuous phase velocity should be similar.122

The last regime in passive droplet generation is tip-multi-breaking. Unlike the previous four regimes where droplets are generated in a continuous process, droplets in this regime are generated in a semi-batch process.123 Similar to the tip-streaming regime, the droplets are generated from the apex of the jet. However, in this case, the apex of the jet is unsteady, and droplets are detached due to Rayleigh-Plateau instability. This oscillation of the meniscus results in multi-modal droplet generation.

2.4.3 Passive Droplet Generation Geometry

In addition to the five different flow regimes described above, one must understand that device geometry is another factor that determines droplet generation. In essence, one can visualize device geometry as a means to achieve the different regimes in different ways.

Figure 2.21 shows the various different geometries that can be used in microfluidic droplet generation. The three most common device geometries encountered in microfluidic droplet geometries are cross-flow, co-flow and flow-focusing.

Figure 2.21: Schematic of various microfluidic device geometries. (a) Cross-flow, (b) co- flow, (c) flow-focusing, (d) step emulsification, (e) microchannel emulsification and (f) membrane emulsification. Reproduced with permission from 107. 107Copyright 2017 Royal Chemistry Society

The most common device geometry configuration is the cross-flow setup. The cross-flow geometry is characterized by the continuous and dispersed phase meeting at an angle, of which the T-junction is one such example.124 These devices are easily assembled with a simple geometry, typically taking a planar configuration. The variations of the cross-flow design include either the dispersed or continuous phase in the main channel and side channel respectively,125 different angles between the main and side channels126,127 and multiple/sequential T-junctions in one device.128 For example, a double T-junction setup can be used to generate alternate droplets from two different streams.129 Cross-flow geometries are easily achieved through photolithography techniques. One such technique

41 is adapted from the printed circuit board (PCB) protocol, also known as soft lithography.

In this method, a photomask is utilized to impart a desired pattern on the substrate containing various layers including photoresists. This setup then undergoes the pattern exposure, development and etching steps to produce a PCB master. Polydimethylsiloxane

(PDMS) is then cast on the PCB master and cured. This PDMS negative relief pattern is then detached from the PCB master and then applied to a glass slide to form the microfluidic channel.130 The soft lithography method is shown schematically in Figure

2.22.

Due to the multiple steps involved in the soft lithography method, alternative photolithography methods have been developed. For example, Harrison et al.131 used thiolene-based optical adhesives to function as a negative resist in the process. This cut down the number of steps significantly, resulting in a microfluidic droplet generator without masks or etching steps.131

Another popular device geometry is the co-flow geometry. In this case, both the dispersed and continuous phase flow in the same direction and in parallel.124 This geometry can be achieved by planar systems, but is most easily and commonly achieved in a 3D configuration, using annular glass capillaries.109 This geometry also provides flexibility in operating regimes as different regimes can be achieved in the same device, through the control of the dispersed and continuous flowrates. Multiple co-flow geometries in sequence can produce monodisperse multiple emulsions, as shown by Chu et al.,132 and illustrated in

Figure 2.23.

Figure 2.23: Generation of monodisperse triple emulsions. (a) Schematic of co-flow device, (b) first, (c) second and (d) third emulsification stages. (e) Triple emulsions formed at different flow rates. Reproduced with permission from 132.132Copyright 2007 John Wiley and Sons.

Flow-focusing geometries are an extension of co-flow geometry. In these setups, the parallel flows are focused typically with another glass capillary, where the arrangement and setup causes the flows to accelerate, resulting in smaller droplets.133 While there are some 2D flow-focusing devices,134 3D coaxial, axisymmetric devices work better due to the absence of wetting of channel walls. One deviation of the parallel flow-focusing geometry is the microcapillary device, where the dispersed and continuous phase are fed in opposing directions, but are still focused by the narrow orifice provided by the flow- focusing geometry.135 Combinations of both flow-focusing and co-flow geometries can also be used to generate double emulsions. Figure 2.24 shows both a flow-focusing microcapillary schematic and a combination of both microcapillary flow-focusing and co- flow geometries.

2.4.4 Active Droplet Generation

In contrast to passive droplet generation methods, active droplet generation can further be categorized into either modifying existing intrinsic forces of the system, or the input of an additional force into the system. Compared to passive droplet generation methods, active droplet generation imparts greater flexibility in control of droplet size and distribution, on demand droplet production and a shortened system response time.

One example of active droplet generation methods includes the capitalization of electrorheology. In one such work, Niu et al. used an electrical trigger signal to effect an increase in dispersed phase viscosity, allowing the formation of a droplet in flow-focusing geometry.136 In this study, electrorheological fluids (GER) droplets were generated on demand. Under an electric field signal, the viscosity of the GER fluid increased due to the formation of nanoparticle chains across of the electrodes, impeding or stopping GER fluid flow into the droplet generator.

In another electrical active droplet generation process, voltage is applied to the system to change the accumulation of charges on an interface, influencing the interaction between the two phases. For example, a DC pulse can be initiated to induce an interface into a narrow orifice that creates a droplet at the end of a jet.137,138

Another example of active droplet control is the use of laser pulses to induce cavitation in parallel immiscible streams. In this setup, Park et al. used a pulsed laser to induce bubble expansion in the dispersed phase flow stream next to a connecting orifice.139 This pulse forces rapid thermal expansion of the dispersed phase into the orifice, with the dispersed phase droplet detaching into the continuous phase. This method affords high volume, on-

45 demand droplet generation, but suffers from low volume generation. Figure 2.25 shows both the schematic of the pulsed laser-driven droplet generation (PLDG) device and time- resolved droplet generation using the device.

One application of microfluidic droplet generation is the fabrication of monodispersed solid microparticles. This is achieved by downstream polymerization of monomers downstream after droplet generation. For example, Dendukuri et al.140 polymerized a photopolymer downstream of a T-junction using UV light. By controlling the flow rates and shifting between squeezing and dripping regimes, both spherical and pill-shaped microparticles were successfully synthesized.140 In another work, Nisisako et al. used multiple cross-flow junctions to synthesize monodispersed Janus particles that could color switch upon application of a voltage between the electrode panels.141 Figure 2.26 shows the formation of the Janus droplets using such a microfluidic device.

2.5 Additive Manufacturing

2.5.1 Overview of Additive Manufacturing

Traditional established manufacturing techniques fall under the category of subtractive manufacturing. In this method, material is hewn or cut out from a solid block of material until the desired part geometry is achieved. This process can be done manually by hand or automated by machines. The advantages of this process include high throughput rate, high quality, large selection of materials and large part size. In contrast, additive manufacturing involves the successive deposition of material layer by layer to form the part. Additive manufacturing excels in intricate geometries, hollow parts and rapid prototyping requirements,142 but suffers from limited material selection, poor mechanical properties, and slow throughput rate.143 These advantages have resulted in additive manufacturing seeing widespread use in automotive,144 aerospace145 and biomedical applications, particularly in prototyping or low volume applications.146 There are multiple ways to classify the different additive manufacturing processes,147 but one helpful way is from the state of the starting material used.148

2.5.2 Material Types

The first type of starting materials are liquids, used extensively in the stereolithography

(SLA) method.149 In this method, the starting material consists of a photopolymerizable monomer system that polymerizes under exposure to UV. The photosensitive resin is placed in the vat and selectively cured with UV light one layer at a time. Upon crosslinking and polymerization of the layer, the platform ascends/descends (depending on printer design), a new layer of liquid resin fills the next layer and the subsequent layer is cured by

UV light next. This continues till the whole part is printed layer by layer. In addition to pure polymeric resins, other materials have also been adapted to this process. For example, some resins include ceramic or metallic components in the monomer resin to include additional mechanical reinforcement.150 Multi-jet modelling is another process that utilizes liquid monomer as the starting material. The UV-curable polymer is deposited on one layer through nozzles and cured immediately with UV.151 The next layer is subsequently deposited and cured, with the process repeating itself until the part is finished. As a result, the multi-jet modelling process uses less monomer material compared to SLA, where the whole vat of uncured monomer after the printing process is disposed after one print.

Fused deposition modelling (FDM) is another common additive manufacturing process.

This involves heating filaments of thermoplastic polymers above their melting point and extruding them through nozzles in a layered fashion. The nozzle is mounted on a movable head. After extrusion of the polymer thread on the first layer, the printed polymer is allowed to air cool and solidify. The head is moved up vertically and the next layer is extruded.

Most printers now are able to accommodate multiple nozzles,152 one extruding the part material and the other nozzle extruding a support material that can be washed or dissolved

48 after printing. A derivative of FDM is direct ink writing (DIW). In this instance, instead of extruding a melted thermoplastic filament, a viscous liquid is extruded through the nozzle of the movable head. After extrusion, the viscous liquid is allowed to cure, allowing the printing of gels. However, DIW has some limitations on the materials to be extruded.

Ideally, the material the be extruded needs to have low viscosity at high shear to allow extrusion (exhibiting shear-thinning properties), while having short cure times to allow for quick set times after extrusion. Other restrictions of DIW include limited applications to simple geometry, small working window between formulation, and curing and potential for structure collapse after extrusion.153–155 One such example of DIW is demonstrated by

Barry et al., combining both DIW and photocuring of acrylamide solution to form microperiodic hydrogel scaffolds for growth of fibroblasts cells.156 Figure 2.27 shows the

DIW of such hydrogel scaffold and scaffold with cells.

Powders can also be used as starting material in additive manufacturing processes. This method centers around application of material in powder form onto a stage and selectively sintering or heating the powder to form each successive layer. In selective laser sintering

(SLS), the powder to be sintered is placed in a uniform thickness to form a power bed. A laser is traced onto the powder bed in the shape or pattern of one layer of the part, to sinter the powder. The powder bed is lowered after sintering of the first layer and fresh powder is added on top and the laser is traced over the next layer. This process is repeated until the part is complete.157 SLS is one of the more versatile methods in that a wide variety of powders such as polymers, metals and ceramics can be used.158 A derivative of SLS is selectively laser melting (SLM), whereby a high-power laser beam (higher temperatures when compared to SLS) is used to melt the powder to form dense structures that have greater mechanical strength.159 Due to the high energies used to melt metal powders, this process is harder to control and results in the development of residual stresses in the part.160

Examples of metal powders used in SLM include steel, titanium and cobalt.161 Another additive manufacturing process that utilizes powder is the laser metal deposition (LMD) process. In contrast to SLS and SLM where the powder is deposited as a uniform flat bed, the powders of LMD are sprayed or deposited in a specific localized region that is heated and melted by the laser.162,163 Figure 2.28 shows schematics of both SLM and LMD.

As mentioned previously, a variety of materials such as polymers, metals, ceramics and composites can be used for SLS, SLM and LMD. However, polymers are the most commonly used material, which can be categorized into either thermoplastic polymers, or resin systems with monomers, crosslinker and UV initiators. Thermoplastic polymers that can be used for FDM, SLS and SLM include acrylonitrile-butadiene-styrene (ABS), polycarbonate, high impact polystyrene (HIPS), poly(lactic acid) (PLA), polyamide and polyester.164,165

Metal materials used in additive manufacturing can be divided into two different types.

The first method is an indirect method, where metal is introduced into the process through a polymeric binder. In the indirect method, the polymeric binder is melted to bond the metal particles. Post processing is required for this method, where the polymer binder needs to be washed out, along with thermal sintering of the metallic components to ensure structural integrity.166 Direct methods are used for SLM and LMD, where the heat from the laser is sufficient to melt the metal particles to form dense parts with sufficient structural integrity.167 Metals used in the direct method include steel, titanium and nickel alloys.

Similar to metals, ceramic powders can be used for additive manufacturing processes in both indirect and direct methods. Indirect methods predominate in ceramic additive manufacturing processes, where ceramic is incorporated in a polymer binder and fabricated using FDM, SLS or SLA. A second sintering step enables both sintering of the ceramic powder within the powder, as well as the removal of the polymer binder. Direct methods for ceramic powders are not as popular due to the high temperatures required for full sintering of zirconia and alumina into ceramic parts.168 For example, the ceramic powder

51 bed needs to be heated up to at least 1600 °C, while melting temperatures of alumina and silicon dioxide are 2000 °C and 1700 °C respectively.

2.5.3 Applications

The aerospace industry has taken advantage of additive manufacturing due to their requirement for components with complex geometries. In addition, aerospace parts require high performance materials such as titanium and nickel alloys with smaller production runs compared to the automotive or consumer product industries. Some example of aerospace parts that have been manufactured from additive manufacturing include mixing nozzles, support casings for gas turbines, turbine blades, engine housings, vents, ducts, vanes and airfoils. Some of these additive manufactured parts are shown in Figure 2.29.

Figure 2.29: Examples of additive manufactured aerospace components. (a) Compressor support case for gas turbine, (b) turbine blades with cooling channels, (c)turbine blades and (d) engine housing. Reproduced with permission from 148. 148Copyright 2013 Springer Nature.

Additive manufacturing in the automotive industry has centered around prototyping parts in the product development process. This cuts down the time required in developing new products due to the elimination of mold and part fabrication through other manufacturing methods. Interestingly, additive manufacturing has been a great benefit to motorsports, due

52 to their small part volumes and need for lightweight and high-performance alloys.

Examples of parts fabricated using additive manufacturing include gearboxes, dashboards, housings and suspension systems.

Biomedical applications are another area where additive manufacturing is used. Additive manufacturing is particularly useful in this industry due to the complex geometries required. For example, additive manufactured parts see widespread commercial use as orthodontic devices under the tradename of Invisalign™. In such instances, additive manufacturing allows for the fabrication of custom parts to suit each individual’s needs and requirements. In addition, biomedical implants (e.g. hip and acetabular cups) are being developed with designed porosity in an attempt to reduce weight as well as to match the stiffness and flexibility of the original bone to be replaced.169,170 Additive manufactured parts have also seen growth in tissue scaffolding applications due to their flexibility in designing complex porous structures. These structures allow transport of nutrients and oxygen to the cells, at the same time providing accessibility for fluids to remove waste products.171,172 Tissue scaffolds have been fabricated using FDM,173 SLA174 and SLS,175 using biocompatible materials such as polycaprolactam (PCL),176 polyether ether ketone

(PEEK), polyethylene glycol (PEG),177 poly(ethylene oxide) (PEO) and bioceramics.

There have also been limited work in developing tissue scaffolds indirectly by printing a polymer mold, followed by casting of a sol into the mold to form the tissue scaffold.178

CHAPTER 3

SOLVENT EFFECTS ON TUNING PORE STRUCTURES IN POLYIMIDE AEROGELS

Reproduced from: [Teo, N.; Jana, S.C.; Solvent Effects on Tuning Pore Structures in Polyimide Aerogels. Langmuir 2018, 34(29), 8581-8590. DOI: 10.1021/acs.langmuir.8b01513.]

Abstract

This work evaluates the effects of solvents and a block copolymer surfactant on pore structures in polyimide aerogels synthesized via sol-gel reaction process. Specifically, crosslinked polyimide gel networks are synthesized in single or mixed solvents from a combination of dimethylformamide (DMF), N-methyl-2-pyrolidone (NMP), and dimethylacetamide (DMAc) and supercritically dried to obtain aerogels. The bulk density, pore size, and mechanical properties of aerogels are determined. The results show that gel times are strongly dependent on the electron acceptance ability of the solvent system and concentration of the surfactant. At longer gel times, the polyimide strands coarsen and the pores in aerogel shift from predominantly mesoporous to macroporous state with corresponding reduction in compressive modulus. The block copolymer surfactant also slows down gelation and coarsens the polyimide strands, but only weakly affects the compressive modulus of the aerogels.

3.1 Introduction

Aerogels are known for their high porosity and high specific surface area. They have traditionally been used as thermal insulation1,179 although recent reports established their potential in airborne nanoparticle filtration.64–66,180,181 Polymeric aerogels can be fabricated from physical gelation processes such as thermo-reversible gelation (as in the case of syndiotactic polystyrene182) or from the chemical sol-gel processes, such as in the cases of silica,9 polyurea,183 and polyimide33. The choice of solvents in sol-gel processes indeed influences the morphology of the resultant aerogels.184,185 Rao et al.186 reported that the choice of synthesis solvent affected the density, refractive index, surface area, pore volume, and porosity of silica aerogels. Leventis et al.187 observed various microstructures in polyurea aerogels, e.g., string-of-beads in dimethylformamide, particle clusters in dimethylsulfoxide, and fibrillar morphology in acetone. Gu et al.185 reported strong dependence of polybenzoxazine aerogel building blocks (e.g., strands or spherical aggregates) and corresponding specific surface area on choice of solvents and the reaction temperature.

A solvent affects the kinetics and the equilibrium of chemical reactions. For example, SN2 reactions can proceed at up to 109 faster rate in dipolar aprotic solvents in comparison to protic solvents with similar macroscopic properties due to weaker anion solvating power and weaker structure effects of the aprotic solvents.29,30 It is important to recognize that the macroscopic solvent properties such as surface tension, permittivity, dielectric constants, solvent polarity, or refractive index do not exert influence on molecular level chemical reactivity.188 The above macroscopic properties do not take into account solute-solvent interactions and their effects on the solvation and stabilization of the individual reactants,

55 transition states, and the products in their polar environment.189 Solvents form solvation shells around the solute molecules, which in turn lead to molecularly induced local inhomogeneities in the solvent environment, thus rendering dielectric or electrostatic approaches inadequate.190 Empirical, semi-quantitative measures were developed in an attempt to explain the solvent effects on reaction rates, such as in terms of Lewis basicity/acidity, nucleophilic/electrophilic nature, or electron donating/accepting capability. Several empirical scales are used to rank the solvents. For example, Gutmann191 proposed the donor number (DN) scale, measured based on the negative enthalpy values of the 1:1 adduct formation between SbCl5 in 1,2-dichloroethane to determine a solvent’s nucleophilic, electron donating, or cation solvation tendency. Conversely, the acceptor number (AN) determines the electron accepting ability of the solvent, which was derived from the 31P NMR measurement of triethylphosphine oxide dissolved in the desired solvent.192 Both DN and AN have been shown to correlate well with other empirical scales such as Kamlet and Taft’s β-scale for nucleophilic ability (analogous to DN), Dimroth and

193 Reichardt’s ET(30), and Kosower’s Z value for electrophilic ability (analogous to

AN).194–196

Traditionally, polyimides are synthesized using the Dupont two-step process.26 The first step - the rate determining step - involves the reactions of a dianhydride with a diamine. In this step, nucleophilic acyl substitution reactions occur where the nitrogen in the amine group attacks the carbon of the carbonyl group28 thus forming a slightly negative activated complex that proceeds to form the polyamic acid. The polyamic acid group autocatalyzes the above reactions until an equilibrium is reached. The second step involves the chemical imidization of the polyamic acid with acetic anhydride as the dehydrating agent and

56 pyridine as the catalyst.197 These reactions are typically carried out in dipolar aprotic solvents as the anions are less solvated in these solvents compared to protic solvents, thus allowing for greater effectiveness in nucleophilic attack.198

In addition to diamines and dianhydrides, a trifunctional amine is used as a crosslinker to form a three-dimensional network structure that eventually produces polyimide gels. The molecular weight of the polymer chain increases until it is no longer soluble in the solvent.

The reaction system undergoes polymerization induced phase separation as the binodal and spinodal curves approach the experimental temperature with increasing molecular weight.

The phase separation progresses until the system turns into a gel.

In this study, we investigated the effects of solvents and a block copolymer surfactant on polyimide aerogel morphology, bearing in mind that the solvents influence both the chemical kinetics of the polyimide reaction scheme and the thermodynamics of phase transition. The aim was to use the solvent environment to control and tune the pore size distribution in the aerogels. To our knowledge, the effect of surfactants on polyimide gelation has not been studied before. Such information will be useful in tuning of air permeability and airborne nanoparticle filtration efficiency of the final aerogels.64–66,181

Recently Teo and Jana199 reported open-cell aerogel foams of syndiotactic polystyrene via emulsion-templating method, whereby surfactants were used to stabilize water-in-oil emulsions and the dispersed water droplets yielded micrometer size voids in the final aerogel structures. In this context, the results presented in this paper on the effects of block copolymer surfactants may be useful in further development of emulsion-templating method for polyimide gels.

3.2 Experimental Section

3.2.1 Materials

Pyromellitic dianhydride (PMDA) was purchased from Alfa-Aesar (Haverhill, MA) and

2,2’-dimethylbenzidine (DMBZ) was purchased from Shanghai Worldyang Chemical Co.

Ltd (Shanghai, China). Tris(2-aminoethyl)amine (TREN) crosslinker and block copolymer surfactant F127® (trademark of BASF) were purchased from Sigma Aldrich (Milwaukee,

WI). Pyridine, acetic anhydride, and acetone were purchased from Fisher Scientific

(Ontario, NY). Among the solvents, N,N-dimethylformamide (DMF) was purchased from

VWR International (Radnor, PA), anhydrous 1-methyl-2-pyrrolidone (NMP) and N,N- dimethylacetamide (DMAc) were purchased from Sigma Aldrich (Milwaukee, WI).

3.2.2 Fabrication of Polyimide Aerogels.

Polyimide gels were synthesized in either a single solvent or a set of mixed solvents.

PMDA and DMBZ were dissolved separately in the selected solvent and the solutions were mixed and magnetically stirred for 2 minutes at 1200 rpm to form the polyamic acid.

Subsequently, TREN, acetic anhydride, and pyridine were added, and the solution was magnetically stirred for 3 minutes. The solution was subsequently poured into cylindrical molds of length to diameter ratio of 2:1 and allowed to gel. The gels were aged in the molds for 24 hours and solvent exchanged with acetone for at least six times to remove the solvent used in synthesis. A typical polyimide gel sample with 7.5 wt% polymer concentration can be prepared using 0.228 g of PMDA, 0.212 g of DMBZ, 0.030 g of TREN, 0.665 g of acetic anhydride, 0.625 g of pyridine, and 5.0 mL of total solvent. A higher than

58 stochiometric amount of TREN crosslinker was used in an effort to obtain aerogels with stronger mechanical properties. The reaction scheme is presented in Figure 3.1.

Figure 3.1: Reaction scheme for synthesis of polyimide crosslinked networks

The gels were solvent exchanged with liquid carbon dioxide and dried under supercritical condition of carbon dioxide at 50 ℃ and 11 MPa pressure. In this work, mixed solvents

DMF/NMP and DMF/DMAc were used with the volume percent of DMF at 0, 25, 75, and

100%.

3.2.3 Characterization of Aerogel Materials

NMR: Solid state 13C NMR cross polarization NMR spectra were collected on a Varian

NMRS 500 MHz (11.7 T) spectrometer operated at 125.62 MHz for 13C and equipped with a Varian narrow-bore triple-resonance T3 MAS NMR probe. Samples were packed into 4 mm zirconia rotors and spun about the magic angle at 15 kHz. Cross-polarization experiments were collected with a 40 kHz spectral window, a 0.02 second acquisition time, a 5 second relaxation delay, 16,832 scans, 62.5 kHz cross-polarization and TPPM decoupling fields, and a 1 ms cross-polarization period using a linear-ramped cross- polarization method. The methyl peak of hexamethylbenzene at 17.3 ppm was used as a secondary chemical shift reference.

IR: Infrared spectra was collected on a Nicolet iS50 FTIR tri-detector spectrophotometer

(Thermo Scientific, MA).

Elemental Analysis: EA analysis was conducted using a 2400 CHNS/O Series II system

(Perkin Elmer, MA).

TGA: thermogravimetric analysis was conducted under N2 with a Q50 thermogravimetric analyzer (TA Instruments, DE) using a heating rate of 20 C/min, up to 700 C.

Porosity and Pore Volume: Porosity was calculated from the values of skeletal (ρs) and bulk density (ρb) as shown in equation (1). The values of skeletal density were obtained using a helium pycnometer (AccuPyc II 1340, Micromeritics Instrument Corp., GA). Bulk density was obtained from the mass and volume of the aerogels.

𝜌 푝표푟표푠푖푡푦 = (1 − 푏) × 100% (1) 𝜌푠

Total pore volume (Vtot) was calculated from the bulk and skeletal densities, according to equation (2):

1 1 푉푡표푡 = − (2) 𝜌푏 𝜌푠

Shrinkage: Polyimide gels were synthesized and poured into cylindrical molds of 13 mm diameter. The diameter of final dried aerogel was measured and the diameter shrinkage in reference to original diameter was calculated.

Aerogel Morphology. The morphology of aerogels was studied using a scanning electron microscope (SEM, JSM5310, JEOL, MA). An accelerating voltage of 5 kV and emission current of 20 mA was used to capture the SEM images. A representative piece of fractured aerogel specimen was mounted on an aluminum stub using carbon tape, followed by sputter coating with silver (ISI-5400 Sputter Coater, Polaron, UK). Polyimide strand diameter was estimated from SEM images using the ImageJ software by considering more than 100 diameter readings for each sample.

BET surface area: Brunauer-Emmett-Teller (BET) surface area of aerogel specimens was obtained from N2 adsorption-desorption isotherms at 77 K using a Micromeritics Tristar II

3020 analyzer (Micromeritics Instrument Corp. GA). The pore volume of macropores

(Vmacropores) was deduced from total pore volume, Vtot, and the volumes of mesopores

(Vmesopores) and micropores (Vmicropores). For this purpose, CO2 adsorption-desorption isotherms at 273 K were obtained and combined with the N2 isotherms using the non-local density functional theory (NLDFT) model. The fraction of macropores, mesopores, and micropores were then subsequently calculated according to equation (3).

푉푚푎푐푟표푝표푟푒푠 푉푚푒푠표푝표푟푒푠 푉푚푖푐푟표푝표푟푒푠 휑푚푎푐푟표푝표푟푒푠 = ; 휑푚푒푠표푝표푟푒푠 = ; 휑푚푖푐푟표푝표푟푒푠 = 푉푡표푡 푉푡표푡 푉푡표푡

(3)

Compressive Modulus: Compressive modulus of aerogel specimens were obtained from compressive tests using an Instron 5567 universal testing machine (Norwood, MA) with

61 cylindrical aerogel specimens of length to diameter ratio 2:1, according to ASTM D695-

15.200 The cylindrical samples were first grinded to ensure a smooth, even, and parallel surface and then loaded onto the tensometer. A 1 kN load cell was used, with an extension rate of 1.3 mm/min. The tests were stopped when the force applied exceeded the upper limit of the load cell used. The compressive modulus of the aerogels was obtained from the slope of the stress-strain curve at low strain, typically 0.01-0.05 mm/mm.

Maximum Bubble Pressure: The mechanical integrity of the polyimide gel was evaluated using a bubble pressure rheometer setup.201,202 A 22-gauge blunt tip needle attached to a

Braintree Scientific syringe pump (Braintree, MA) was used to introduce air at a prescribed pressure. An Omega PX26 pressure transducer connected to an Omega DP25B-S-A process meter and a computer were used to capture pressure readings with time. The gels used in this test were aged for 24 hours at room temperature (20 C). The needle was inserted approximately 1 cm below the surface of the gel. Air was pumped into the gel at a flow rate of 5 mL/min and the pressure reading was recorded for 2 min. The maximum pressure was taken as the critical pressure (Pc) of the gel. Due to high modulus of the polyimide gels, the gels did not experience cavitation at Pc and instead a slight fracture of the gel surrounding the needle occurred allowing the air to escape and preventing additional increase of pressure in the system.

3.3 Results and Discussion

3.3.1 Control of Pore Size Distribution Through Solvent Composition

The aerogels produced in this work appeared amber yellow in color, typical of polyimides

(Figure 3.2). The specimens prepared in neat DMF were translucent, while those prepared

62 in NMP and DMAc appeared opaque. Figure 3.2 shows that the opacity of the aerogels increased with an increase of DMAc fraction in mixed solvent systems DMF/DMAc. A similar transition was observed in the DMF/NMP solvent system as well.

Figure 3.2: Polyimide aerogels with varying DMF/DMAc solvent compositions (vol%). DMAc vol% increases from left to right.

The solid state 13C NMR spectra of polyimide prepared in DMF, NMP and DMAc are presented in Figure 3.3. The resonances due to methyl groups (20 ppm) and aromatic carbons (120, 130 and 137 ppm) indicated the inclusion of DMBZ, while the resonances of aromatic carbons (130, 137 ppm) and imide carbonyls (165 ppm) were due to incorporation of PMDA in the polymer structure. In addition, the methylene peaks (62 ppm) confirm that the TREN trifunctional crosslinker was indeed incorporated in the polyimide aerogel structure.

Figure 3.3: Solid state 13C NMR spectrum of polyimide synthesized with the 100% DMAc, 100% NMP and 100% DMF.

Figure 3.4a shows the IR spectra of polyimide synthesized with 100% DMAc, NMP, and

DMF. All three spectra show the presence of peaks at 730, 1380, 1716, and 1778 cm-1, indicating the presence of the imide functional group. The small peaks at 1620 and 3000 cm-1 indicate the presence of small amounts of polyamic acid which did not undergo imidization. The absence of peaks at 3250-3400 cm-1 also confirm the absence of primary amine groups in imidized materials. The TGA traces of aerogels synthesized with single solvents are shown in Figure 3.4b. It is apparent that the three materials have similar TGA traces, exhibiting weight loss of 6-9 wt% until about 525 °C, after which the degradation of polyimide began. We attribute the small weight loss observed up to 525 °C to thermal degradation of the residual polyamic acid, as earlier discussed in the context of IR spectra in Figure 3.4a. The small weight loss observed before 525oC is consistent with the TGA

64 data of other polyimide aerogel materials reported in literature.203 At 700 °C, all three TGA traces show char yield of ~60 wt%.

Figure 3.4: (a) IR spectrum and (b) TGA curves of polyimide synthesized with the 100% DMAc, 100% NMP and 100% DMF.

The absence of free amine groups in imidized materials is corroborated by the data from elemental analysis as listed in Table 3.1. The molar ratio of nitrogen/carbon (N/C) for all three aerogel specimens fell in the range 0.083 - 0.085, close to the theoretical molar N/C ratio 0.083 in one PMDA/DMBZ repeat unit. This suggests that the polyimide polymer was comprised primarily of PMDA and DMBZ monomers with small contribution from

TREN. Note that TREN crosslinker has molar N/C ratio of 0.66. The close molar N/C ratio of final aerogel material to that of the polymeric repeat unit was earlier observed by

Leventis et al.27 The slightly higher N/C ratio observed for aerogels synthesized with 100%

DMAc indicates shorter oligomer chain length and slightly higher ratio of TREN crosslinker.

Table 3.1: Elemental Analysis of polyimide synthesized with the 100% DMAc, 100% NMP and 100% DMF.

Elemental Analysis (wt%) Solvent Molar N/C C H N 100 % DMF 71.15 3.37 6.87 0.083 100 % NMP 70.29 3.57 6.84 0.084 100% DMAc 70.33 3.41 6.97 0.085

The data on diameter shrinkage, bulk and skeletal density, porosity, and gel times of specimens prepared in various solvent systems are listed in Table 3.2. In the case of mixed solvents, various volume fractions of DMF with NMP and DMF with DMAc were used.

Polyimide aerogels prepared in DMF exhibited a diameter shrinkage of 13.6%, while those prepared in NMP and DMAc had diameter shrinkage of 9-10%. Such moderate shrinkage values can be attributed to capillary stress originating from the evaporation of residual synthesis solvents in the heating phase of the supercritical drying step.

The data in Table 3.2 also reflects that the skeletal density of the aerogel specimens remained relatively constant in the range 1.36 – 1.39 g/cm3, while the bulk density varied from 0.07 g/cm3 for specimens prepared in 100% DMF to 0.06 g/cm3 for 100% NMP to

0.056 g/cm3 for 100% DMAc. The bulk density values for specimens prepared in solvent mixtures fell within the range of corresponding single solvent data. The porosity of aerogel specimens fell within a narrow range of 94.9 to 96.0%.

Table 3.2: Shrinkage, bulk density, skeletal density, porosity and gel times of polyimides synthesized with various solvent compositions.

Solvent Composition Skeletal Gel Shrinkage Bulk Density Porosity (vol%) Density Time (%) (g/cm3) (%) DMF NMP DMAC (g/cm3) (min) 100 0 0 13.6 ± 0.0 0.070 ± 0.001 1.36 ± 0.04 94.9 ± 0.1 6.5 75 25 12.6 ± 0.3 0.066 ± 0.001 1.39 ± 0.04 95.2 ± 0.0 23 50 50 11.2 ± 0.0 0.063 ± 0.000 1.36 ± 0.04 95.4 ± 0.0 44 0 25 75 9.0 ± 0.0 0.056 ± 0.001 1.39 ± 0.04 96.0 ± 0.0 89 0 100 9.7 ± 0.3 0.060 ± 0.000 1.37 ± 0.04 95.6 ± 0.0 112 75 25 12.7 ± 0.0 0.066 ± 0.000 1.36 ± 0.05 95.1 ± 0.1 23 50 50 8.7 ± 0.3 0.056 ± 0.003 1.37 ±0.04 95.9 ± 0.1 56 0 25 75 9.0 ± 0.2 0.056 ± 0.001 1.36 ± 0.03 95.9 ± 0.1 70 0 100 9.2 ± 0.0 0.056 ± 0.001 1.36 ±0.03 95.9 ± 0.1 109

One key observation from the data listed in Table 3.2 is the effect of single solvent on gel time; e.g., polyimide synthesized in 100% DMF turned into a gel in about 6.5 min, while the system with the same polymer content gelled in 112 min in NMP and in 109 min in

DMAc. The gel times in the case of mixed solvents fell within the window of corresponding single solvents. At gel time, the liquid meniscus in the mold did not show movement when the mold was tilted at an angle.

Two factors are responsible for the dependence of gel times on solvents. First, higher solubility of polyimide in a solvent can delay its phase separation into solid polymer nodules that subsequently interconnect and form the strands in the gel networks as was recently established for polybenzoxazine system.204 However, such differences in solubility alone cannot explain the huge differences in gel times reported in Table 3.2. In view of this, we invoke a second factor, which is the increased reaction rates brought about by the electron accepting properties of the solvents. Adopting Gutmann’s scale of acceptor numbers (AN),191 Table 3.3 shows that the electron accepting properties of the solvents increase in the order of NMP

67 accepting capability should exhibit higher Lewis acid behavior. A more acidic solvent stabilizes the activated complex of the polyamic acid and drives the reaction forward. This is supported by other studies on polyamic acid kinetics.31,32 Kaas32 showed that the polyamic acid reaction exhibits a sigmoidal shaped concentration versus time curve indicative of a second order, equilibrium controlled, autocatalytic reaction. It was also reported that polyamic acid formation reaction is acid-catalyzed, instead of base- catalyzed.32 In a separate study, Solomin et al.31 reported higher rates of polyamic acid formation in acidic solvents such as m-cresol and acetonitrile than in more basic solvents such as THF.

Table 3.3: Surface tension, donor number (DN), acceptor number (AN) and ET(30) of solvents. Surface tension was measured, while DN, AN and ET(30) were obtained from Reichardt, C.189 Electron Donating Electron Accepting Surface Tension Solvent Donor Number Acceptor ET(30) @ 17 C (mN/m) (kcal.mol-1) Number (kcal.mol-1) DMF 37.0 ± 0.0 26.6 16 43.2 NMP 41.0 ± 0.1 27.3 13.3 42.2 DMAc 36.0 ± 0.0 27.8 13.6 42.9

An increase in gel times compared to DMF due to use of NMP, DMAc, or their mixtures with NMP also produced changes in the morphology of the corresponding aerogels. These changes will be gleaned from the SEM images (Figure 3.5) and from the BET data listed in Table 3.4.

Figure 3.5: SEM images of polyimides prepared using (a) DMF, (b) NMP, and (c) DMAc.

Table 3.4: BET surface area, pore volume and micro, meso and macropore fraction of polyimides synthesized with various solvent compositions.

BET Solvent Composition Pore Micropore Mesopore Macropore Surface (vol%) Volume Fraction Fraction Fraction Area (cm3/g) (<2 nm) (2-50 nm) (>50 nm) DMF NMP DMAC (m2/g) 100 0 0 792 ± 5 13.5 ± 0.1 0.015 0.511 0.474 75 25 760 ± 21 14.3 ± 0.1 0.010 0.155 0.835 50 50 573 ± 10 15.2 ± 0.1 0.009 0.124 0.867 25 75 - 485 ± 11 17.1 ± 0.1 0.010 0.081 0.909 457 ±16 15.9 ± 0 0.010 0.045 0.945 0 100 .2 75 25 752 ± 24 14.3 ± 0.1 0.008 0.163 0.829 50 50 552 ± 21 17.2 ± 1.2 0.007 0.041 0.952 - 25 75 474 ± 10 17.1 ± 0.4 0.005 0.039 0.956 0 100 443 ± 11 17.0 ± 0.2 0.007 0.031 0.962

Figure 3.5 shows that the aerogel specimens contained similar fibrillar structures, with the polyimide strand size increasing with an increase of gel time. Polyimides prepared with

100% DMF gelled in 6.5 min and had a corresponding strand size of 9.3 ± 1.7 nm, while those prepared with NMP and DMAc had strand sizes of 13.5 ± 3.1 and 14.8 ± 2.5 nm respectively, corresponding to higher gel times of 112 min and 109 min. The data in Table

3.2 and 3.4 shows that BET surface area reduced with an increase of gel time. In mixed solvent system of DMF with NMP, the gel time increased with NMP content from 6.5 min for 100% DMF to 112 min for 100% NMP. The corresponding aerogel specimens showed

BET surface area reduction from 792 m2/g for 100% DMF to 457 m2/g for 100% NMP.

The same trend is prevalent in systems containing DMAc; the BET surface area with 100%

DMAc is 443 m2/g.

In light of negligible changes in skeletal density of the aerogel specimens in different solvents (Table 3.2), one can attribute the reduction in specific surface area to coarsening of polymer strands. For polyimide strands of cylindrical shape as seen in Figure 3.5, the specific surface area scales inversely with the strand diameter. Recall that mean polymer strand diameter and specific surface area of polyimide aerogels synthesized in DMF were respectively 9.3 nm and 792 m2/g. In reference to these values, polyimide aerogels synthesized in NMP (mean strand diameter 13.5 nm) and DMAc (mean strand diameter

14.8 nm) should have specific surface area of 546 m2/g and 497 m2/g. The experimental data in Table 3.4 indicate surface areas of 457 and 443 m2/g respectively for the two systems, in each showing reduction.

The coarsening of polymer strands should have an impact on the volume fractions of different types of pores. Of these, micropore fraction is anticipated to be much less affected; polyimide strands do not contain inherent micropores unlike syndiotactic polystyrene systems.182 Thus, all macropores in polyimide aerogels are formed due to crossover of the adjoining strands. The data in Table 3.4 show the following trends with an increase of NMP content from 0 to 100% in the solvent system – (i) as anticipated, the micropore fraction reduced only slightly from 0.015 to 0.010, (ii) the mesopore fraction reduced significantly from 0.511 to 0.045, and (iii) the macropore fraction increased from

0.474 to 0.945. A reduction of mesopore content with an increase of NMP and DMAc fractions is gleaned from smaller area under the hysteresis loops in the BET isotherms, as

70 presented in Figure 3.6. The specimens obtained with DMF show the largest area under the hysteresis loop while those obtained with NMP or DMAc show negligible area. This shift from almost a 50:50 ratio of meso- and macropores obtained using DMF to predominantly macropores in the case of NMP and DMAc also correlates well with the shrinkage data reported in Table 3.2. Aerogel specimens with larger mesopore fractions underwent higher shrinkage due to greater capillary stress originating from smaller pores. In this context, the shift to pore sizes from primarily mesoporous state to majority macroporous state can also account for the increase of opacity of the aerogels as shown in Figure 3.2. The larger pores approaching the wavelength of light enable scattering of incident light and contribute to opacity. Corresponding pore size distributions of the aerogel specimens are presented in

Figure 3.7.

Figure 3.6: BET isotherms of polyimide aerogels with increasing (a) NMP and (b) DMAc content in the synthesis solvent.

Figure 3.7: Pore size distribution of aerogels with increasing a) NMP and b) DMAc content in the synthesis solvent

We now focus attention on why the strand size increased in systems that showed longer gel times. In polyimide systems, gelation occurs via a sequence of steps. First, the oligomer molecular weight increases with conversion. The system then undergoes polymerization induced phase separation, as the binodal line in a temperature-composition diagram gradually shifts upward toward the experimental temperature with an increase of molecular weight. Once in the binodal region, the system phase separates into a polymer-rich region

(which turns into the strands in the aerogel) and a solvent-rich region (which later turns into the pores in the aerogel). These phase-separated regions coarsen over time, before the system reaches the sol-gel transition and vitrifies due to high conversions and crosslinking densities.12 At lower rates of polymerization, i.e., at longer gel times, the sol-gel transition after phase separation is prolonged. This promotes coarsening of the polymer-rich domains in an attempt to reduce the unfavorable interfacial area with the liquid-rich domains and leads to thicker polymer strands in the gel.

The values of compressive modulus of polyimide aerogels synthesized in various solvents are listed in Table 3.5. The aerogels synthesized in DMF show the highest value of

72 compressive modulus at 44.6 MPa, while those synthesized in NMP and DMAc show compressive modulus of 8.9 and 5.8 MPa respectively. The aerogels synthesized in mixed solvents show modulus values within the bounds of materials produced in single solvents.

Table 3.5: Compressive modulus of polyimides synthesized with various solvent compositions

Solvent Composition (vol%) Compressive Modulus DMF NMP DMAc (MPa) 100 0 0 44.6 ± 1.3 75 25 23.4 ± 2.0 50 50 16.2 ±2.9 - 25 75 7.9 ± 0.4 0 100 8.9 ± 1.7 75 25 28.7 ± 1.8 50 50 13.5 ± 1.6 - 25 75 10.6 ±2.2 0 100 5.8 ± 4.4

The aerogel specimens synthesized in DMAc exhibited brittle fracture, while those synthesized in DMF and NMP show yielding behavior as gleaned from the representative stress vs. strain curves in Figure 3.8. This brittle failure can be attributed to the reduced number of crosslinks in the system due to the shift in the equilibrium to the left of the polyamic acid reaction in a more basic solvent, thus reducing the overall conversion. The stress vs. strain curves of materials synthesized in DMF and NMP are typical of porous materials, comprising of three main regions. As per Swyngedau,205 the first region involves the deformation of the aerogel matrix at a strain of 0 to 0.04 mm/mm with the applied load borne by the skeletal structure of the crosslinked polymer networks. In the second region at strain from 0.04 to 0.7 mm/mm, the skeletal structure collapses leading to densification of the pores. All pores are compacted, and the load is borne by the now compressed bulk

73 polymer in the third region at strain >0.7 mm/mm. The aerogels synthesized in DMAc exhibited brittle failure at a strain of around 0.1 mm/mm.

Figure 3.8: Representative compressive stress/strain curves of polyimide aerogels synthesized with 100% DMF, 100% NMP and 100% DMAc.

3.3.2 Control of Pore Size Distribution Through Addition of Surfactant The data discussed up to this point established that at longer gel times, the chosen solvent systems produced thicker polyimide strands, higher fractions of macropores, and inferior compressive modulus values. Therefore, it is not meaningful to control the gel times through manipulation of solvent composition if one is interested in achieving significant macropores and strong compressive properties. An alternative is to stick to an aprotic solvent with high AN (such as DMF) for synthesis of aerogels but manipulate the gel times via alteration of the viscosity of the solution and solubility by some other means. In this context, a PEO-PPO-PEO block copolymer (F217) was added as a viscosity modifier of 74 the solution. Specifically, the block copolymer was included in the initial synthesis step by first dissolving it in DMF. The viscosity vs. surfactant concentration data presented in

Figure 3.9 show an increase of viscosity by more than 2-fold at 5 vol% of the surfactant.

The idea is that an increase in viscosity of the solution would shift the reactions to the diffusion-controlled regime and allow longer times for phase separation and coarsening of polymer strands. In conjunction, a high AN solvent helps maintain the equilibrium and high conversion, thus generating aerogels with desired mechanical strength.

2.2

2.0

1.8

1.6

1.4

1.2 Viscosity(mPa.s) 1.0

0.8

0 1 2 3 4 5 Surfactant concentration (vol%)

Figure 3.9: Effect of surfactant concentration on DMF viscosity. Error bars are located within the symbols. The chemical structure of F127® surfactant is also provided in the top left corner.

The presence of surfactant did indeed increase the gel time, as shown in Table 3.6, with the highest surfactant concentration of 5 vol% increasing the gel time from 6.5 mins (for neat DMF with no surfactant) to 29 mins. The polymer strand size also increased with an

75 increase of surfactant concentration. This can be seen from SEM images in Figure 3.10 and the values of strand diameters measured from SEM images and listed in Table 3.6. The strand diameters increased from 9.3 nm for system with no surfactant to 15.5 nm with 0.5 vol% surfactant to 19.9 nm for 2.5 vol% surfactant to 30.4 nm for 5 vol% surfactant concentration. The scatter of the data increased at higher surfactant concentration. The increase of polymer strand diameter with surfactant concentration also led to reduction of

BET surface area, e.g., 792 m2/g for the case with no surfactant to 286 m2/g for the system with 5 vol% surfactant, as listed in Table 3.6. The inclusion of surfactants also led to shifting of micro, meso and macropore fractions (Table 3.6). The most significant changes occurred in the fractions of meso- and macropores. For example, the presence of 0.5 vol% surfactant caused a reduction of mesopore fraction from 0.511 to 0.077 and increase of macropore fraction from 0.474 to 0.909. Additional surfactants at 5 vol% produced aerogel systems with meso- and macropore fractions at 0.041 and 0.953 respectively. The reduction in BET surface area and mesopore fractions are evident from the BET isotherms presented in Figure 3.11.

Figure 3.10: Strands of polyimide aerogels with F127® surfactant concentration of (a) 0 vol%, (b) 0.5 vol%, (c) 2.5 vol% and (d) 5.0 vol%. Images were taken of the surface of the aerogel monolith to facilitate measurement of strand size.

Table 3.6: Strand diameter, BET surface area, micropore, mesopore and macropore fraction of polyimide aerogels with F127® surfactant.

Surfactant Strand BET Micropore Mesopore Macropore Gel Time concentration Diameter Surface Fraction Fraction Fraction (min) (vol%) (nm) Area (m2/g) (< 2 nm) (2-50 nm) (> 50 nm) 0 9.3 ± 1.7 792 ± 5 0.015 0.511 0.474 6.5 0.5 15.5 ± 3.8 544 ± 27 0.014 0.077 0.909 14 2.5 19.9 ± 3.4 426 ± 30 0.010 0.057 0.933 17 5 30.4 ±6.3 286 ± 17 0.006 0.041 0.953 29

Figure 3.11: BET isotherms of polyimide aerogels with various F127® surfactant

concentrations.

Next, the compressive modulus values of gels and corresponding aerogels were determined. The compressive modulus of gels was obtained from pressure versus time curves shown in Figure 3.12 for a set of representative polyimide gels. The maximum values of pressure (critical pressure) are listed in Table 3.7. As is intuitive, pressure first

77 increased with time, reaching a maximum pressure, called the critical pressure, then dropped slightly, and finally reached a plateau. At critical pressure, the gel structure undergoes yielding and fractures in selective locations, thus allowing high-pressure air to escape via the gel-solid interface of the container. The critical pressure can be used as an analogue of mechanical strength of the gel. From the data presented in Figure 3.12 and

Table 3.7, one can infer that the mechanical strength of the gel was weakly dependent on surfactant concentration. The same trend is evident for the data on compressive modulus of aerogels in Table 3.7, also measured using an Instron 5567 tensometer.

Neat PI 70 0.5 2.5 60 5.0

Pressure(KPa) 20

0 0 20 40 60 80 100 120 Time (s)

Figure 3.12: Maximum bubble pressure curves of neat polyimide gel and polyimide gel with surfactant.

Table 3.7: Maximum pressure and compressive modulus of neat polyimide and polyimide with surfactant.

Surfactant Concentration Critical Pressure Compressive (vol%) (KPa) Modulus (MPa) 0.0 66.8 ± 0.4 44.5 ± 1.3 0.5 66.4 ± 0.0 44.1 ± 2.6 2.5 64.5 ± 0.8 44.1 ± 1.1 5 63.2 ± 0.2 38.3 ± 2.7

The above sets of data established that the inclusion of F127® surfactant led to a slow- down of gelation, coarsening of polymer strand diameter, and reduction of mesopore fraction, but did not cause reduction of compressive modulus values, unlike in the cases where NMP and DMAc were used as the solvents. Table 3.5 shows that the compressive modulus of aerogels synthesized with 100% NMP and 100% DMAc were 8.9 and 5.8 MPa respectively, while an aerogel with 5 vol% surfactant in 100% DMF exhibited a compressive modulus of 38.3 MPa. These aerogel specimens had similar macropore fractions of around 0.95.

3.4 Conclusion The results presented in this chapter show that several properties of polyimide aerogels, such as polymer strand diameter, mesopore and macropore fractions, specific surface area, and compressive modulus can be manipulated via selection of appropriate solvents and concentration of a block copolymer surfactant during preparation of corresponding polymer gels. The time of gelation can be prolonged via selection of electron donating solvents such as DMAc and NMP or inclusion of a block copolymer surfactant that increases the viscosity of the medium. In these cases, a slow-down of gelation also led to

79 coarsening of polymer strands, which in turn adversely affected BET surface area and mesopore fractions. However, the compressive modulus of polyimide aerogels reduced significantly when NMP and DMAc were used in synthesis. On the other hand, the inclusion of surfactants did not compromise the compressive modulus of aerogels or the mechanical strength of polyimide gels.

CHAPTER 4

A SURFACTANT-FREE PROCESS FOR FABRICATION OF POLYIMIDE AEROGEL MICROPARTICLES

Reproduced from: [Teo, N.; Jana, S.C.; A Surfactant-Free Process for Fabrication of Polyimide Aerogel Microparticles. Langmuir. Accepted on 15 Jan 2019.]

Abstract

This work focuses on fabrication of polyimide aerogel microparticles of diameter 200-1000

µm from a surfactant-free, two-phase, silicone oil/dimethylformamide (DMF) oil-in-oil

(O/O) system using a simple microfluidic device. The polyimide sol prepared in DMF is turned into droplets suspended in silicone oil in the microfluidic device. The droplets are guided to a heated silicone oil bath to accelerate sol-gel transition and imidization reactions, thereby yielding spherical, discrete gel microparticles that do not undergo coalescence. The discrete gel microparticles are isolated and supercritically dried to obtain aerogel microparticles. The microparticle size distribution shows dependence on dispersed and continuous phase flowrates in the microfluidic channels. The microparticle surface morphology shows dependence on silicone oil bath temperature.

4.1 Introduction

Aerogels are known for high porosities (often exceeding 90%) and high surface areas, reaching values up to 1000 m2/g. Aerogel materials have been produced in several shapes such as monoliths,66 microparticles,206 and films and sheets.36 Among these structural forms, cylindrical monoliths are most commonly studied due to their ease of fabrication.

Kistler first reported silica aerogel monoliths in 1931.1 Aerogels of several other precursor materials were later reported, such as syndiotactic polystyrene (sPS),62,64 polyurea,183,187,207 and polyimide.27,33,180 In past work, a large majority of aerogel articles were fabricated in monolithic forms and only a handful of studies reported flexible aerogel films.36 Recently, monoliths of aerogel foams were reported with large fractions of macrovoids of typical diameter 30 µm, introduced in the structures of inherently meso- (pore diameter 2-50 nm) and macroporous (pore diameter >50 nm) aerogels.199,208,209 Gu and Jana208 synthesized aerogel foams by filling the pores of a sacrificial bi-continuous polymer blend with polyurea sol-gel precursor materials. Teo and Jana199 used water-in-oil (W/O) emulsion- templating method for introduction of macrovoids in sPS aerogel monoliths. The emulsion- templating method was extended by Teo et al.210 to oil-in-oil (O/O) emulsions to include macrovoid formation in mesoporous monolithic aerogels of moisture sensitive monomers, such as the dianhydrides used in the synthesis of polyimide aerogels. The present work investigated a surfactant-free microfluidic process to obtain spherical, mesoporous polyimide aerogel microparticles of diameter 200-1000 µm.

Aerogel microparticles of diameter 50-200 µm may be advantageous in applications involving ion transport or ion adsorption, removal of noxious liquids, or delivery of drugs where large surface areas and short diffusion paths may be beneficial. Aerogel

82 microparticles are of particular interest due to their ease of fabrication, ease of handling, and surface smoothness that prevent inflammatory responses from the body.211 Aerogel microparticles derived from several bio-based polysaccharide materials such as chitin, chitosan, alginate, agar, cellulose, and starch have been reported.211–213

Typically, aerogel microparticles are produced via emulsion polymerization of precursor sol droplets stabilized by surfactants in an immiscible continuous liquid medium. The sol droplets originate from Rayleigh instability followed by stream breakup induced by mechanical mixing of a two phase system.214 In this method, the microparticle size distribution shows dependence on mixing speed, surfactant concentration, and the dispersed phase content. Due to the wide distribution of shear rate in mechanical mixing and undesired particle agglomeration by coalescence, a wide distribution of particle sizes is typically obtained. In preparing aerogel microparticles, another concern is attaining appropriate emulsion stability, particularly for systems that need longer times for gelation.

A majority of emulsion polymerization processes utilize either an oil-in-water (O/W) or

W/O system, e.g., in the synthesis of silica microparticles.215 However, O/W or W/O systems are not suitable for fabrication of aerogel microparticles of moisture-sensitive monomer systems such as polyimides, or for systems with appreciable solubility between the water and oil phases. Gu et al.206 successfully adopted an O/O emulsion system to synthesize polyimide aerogel microparticles using a SPAN® 85 / Hypermer® 1599 surfactant pairing. This system resulted in a broad particle size distribution, in the range

10-90 µm. Other surfactant-free and non-mechanical fabrication methods for aerogel microparticles have also been reported.216 For example, Moner-Girona et al.217 injected the silica sol directly into supercritical liquids such as acetone and carbon dioxide, while

Zhang et al.218 used an ambient pressure drying process based on injection of ammonia gas into a condensed silica solution dissolved in heptane.

One avenue to achieve greater control over particle size and particle size distribution is to capitalize on the accurate metering of microfluidic droplet generators. Droplet generation using microfluidic devices takes place in 5 different regimes, such as squeezing, dripping,219 jetting,135 tip-streaming, and tip-multibreaking.111 Most microfluidic systems utilize viscous shear forces to form droplets and are built on a variety of device geometries such as cross-flow, co-flow, and flow-focusing to obtain the desired shear force fields.220

In co-flow geometry, the dispersed and continuous phase flows occur in parallel streams.

The 2D co-flow geometry is configured using soft lithography,140,221–224 while 3D co-flow devices are typically fabricated through spatial arrangements of the glass capillaries.225,226

To the best of our knowledge, a large majority of microfluidic droplet generation systems use either an O/W or W/O emulsion systems that are stabilized by the surfactants.227 The

O/O emulsion systems are inherently unstable owing to weak association of surfactants and appreciable mutual solubility of the two oil phases228, even though droplets can be more easily formed in O/O emulsion systems using microfluidic droplet generators. Also, the rates of droplet coalescence and Ostwald ripening are higher in O/O emulsions compared to W/O emulsion systems, resulting in rapid phase separation and formation of bilayer structures in O/O emulsions.229 Thus, rapid coalescence observed for O/O emulsions even at high surfactant loading levels is a deterrent to obtaining substantial amounts of gel microparticles. This limitation is particularly important for the present study on polyimide gel microparticles as the time-scale of the sol-gel transition in a polyimide system is typically longer than the time-scale of stability of an O/O emulsion system. One

84 may resort to using higher surfactant loadings to achieve greater stability in O/O emulsions, but the final aerogel microparticles may also retain greater proportions of surfactants.

This work seeks to circumvent the challenges in aerogel microparticle synthesis described above. In summary, these challenges include large particle size distributions, unstable O/O emulsion systems, high coalescence rates and the need for high surfactant loading levels.

In addition, another objective was to make this method of producing aerogel microparticles cost effective and easily accessible without prohibitively high equipment costs. To the best of our knowledge, this work is the first time that a microfluidic setup is used in the fabrication of aerogel microparticles to obtain a narrow particle size distribution. In addition, the laboratory-fabricated microfluidic co-flow droplet generator is easily assembled without the need for specialised or custom components, making it easily obtainable and disposable. Another key critical innovation of this work is that polyimide aerogel microparticles are produced using an oil-in-oil two-phase system without the use of any surfactants. This ensures that the aerogel microparticles produced do not contain any undesired surfactant that would require specific washing steps.

4.2 Experimental Section

4.2.1 Materials

Pyromellitic dianhydride (PMDA) was purchased from Alfa-Aesar (Haverhill, MA) and

2,2’-dimethylbenzidine (DMBZ) was purchased from Shanghai Worldyang Chemical Co.

Ltd (Shanghai, China). Tris(2-aminoethyl)amine (TREN) crosslinker was purchased from

Sigma Aldrich (Milwaukee, WI). Pyridine, acetic anhydride, acetone, and silicone oil were purchased from Fisher Scientific (Ontario, NY). N,N-dimethylformamide (DMF) was

85 purchased from VWR International (Radnor, PA). The components of the droplet generator, such as 27 G syringe, 0.16 cm ID Tee connector, and Tygon tubing were obtained from McMaster-Carr (Aurora, OH).

4.2.2 Fabrication of Droplet Generator

The co-flow droplet generator used in this work was adapted from the work of Li et al.230

This co-flow droplet generator was assembled by inserting a 27 G flat tip needle with inner diameter (ID) of 0.2 mm into a 1.6 mm ID Tee connector. These two components were secured using a 1.6 mm ID Tygon tubing (Figure 4.1a). The assembled microfluidic co- flow generator is shown in Figure 4.1b.

Figure 4.1: Droplet generator components (a) before and (b) after assembly. (c) Schematic of transfer of polyimide sol droplets into heated silicone oil bath. The inset in (c) presents an image from experiments showing gel microparticles floating in heated silicone oil bath.

In this work, the dispersed and continuous phases were respectively polyimide sol in DMF with room temperature viscosity ~ 1.98 mPa.s and silicone oil with room temperature viscosity ~ 48.6 mPa.s. The interfacial tension between these two liquids at room temperature was measured to be 3.0 mN/m. For droplet generation, the dispersed phase flow rate was varied in the range 0.1-0.4 mL/min, which translated into a dispersed phase

86 capillary number (Cad) of 0.015-0.059. The dispersed phase capillary number Cad is defined as the ratio of viscous and interfacial forces as shown in equation (1).

휇푣 퐶푎 = (1) 푑 𝜎

In equation (1), µ is the dynamic viscosity, v is the velocity of the dispersed phase, and σ is the interfacial tension between the continuous and dispersed phases. The Weber number

(Wed) of the dispersed phase calculated from equation (2) fell in the range of 0.15-2.34.

𝜌푣2푙 푊푒 = (2) 푑 𝜎

In equation (2), ρ is the density of the dispersed phase and l is the characteristic length. The continuous phase flow rate was varied between 2 and 8 mL/min, translating into a continuous phase capillary number, Cac, of 0.29 and 1.17 respectively for the above two flow rates using equation (1). The operating domain bounded by the above flow rates placed the droplet formation regime in this study within the jetting regime, as Cac + Wed ≥

O(1).107,111,219 The jetting mechanism is clearly shown in the sequence of images captured using a high speed camera, in Figure 4.2. It is to be noted that this flowrate pairing represents the lowest Cac and Cad in the paper. At these low flowrates, the droplet generation already exhibits jetting behavior, indicating that for other higher flow rates (and higher Ca), the mechanism consistently remains in the jetting regime.

Figure 4.2: Droplet generation in the jetting regime. The above sequence of images was captured using a high-speed camera at 1000 frames per second. Dispersed phase flowrate (Qd) = 0.1 mL/min and continuous phase flowrate (Qc) = 2 mL/min.

High speed image captures of droplet generation and image capture of droplet generation at varying flowrates are also included Figure 4.3. At all these flow rates, droplet generation sits firmly in the jetting regime, as evidenced from the fact that droplet is formed at a discrete distance from the syringe tip where the dispersed phase is introduced. As mentioned in the paper, this regime is due to the high Capillary number of both the dispersed and continuous phase. A couple of observations can be made from Figure 4.3.

Firstly, the droplet size decreases as the continuous phase flow rate increases. This is consistent with the gel and aerogel particle size reported in the main paper. Secondly, an increase in continuous phase flowrate (Qc) results in a higher throughput rate of droplet generation. For example, in a 0.10 s time window, Qc = 2 mL/min generates only one droplet while for the same time window, Qc = 8 mL/min generates 3 droplets.

Figure 4.3: Droplet formation at varying continuous phase flowrates (Qd). Dispersed phase flow rate (Qd) was kept constant at 0.1 mL/min.

The jetting regime was chosen due to a few considerations. Firstly, the objective was to obtain as small a particle size for this setup. This made the jetting regime ideal, as at high

Cac, droplet sizes approach the size of the dispersed phase outlet. Secondly, high flowrates were desired to obtain a high throughput rate of droplet generation, considering the time constraints that the sol-gel transition imposed (~26 mins). Thirdly, there was a lower limit

89 to the flowrate of the dispersed phase for the generation of droplets. Sufficient viscous force (through control of dispersed phase velocity) was required to enable projection of the dispersed phase into the highly viscous silicone oil of the continuous phase. This minimal velocity coupled with the low interfacial tension of the system results in large Cad that places the flows in the jetting regime. Lastly, as this method of microparticle generation does not utilize surfactants, special measures must be taken to ensure that the droplets do not coalesce prematurely due to contract prior to the sol-gel transition. One such way is to have a high continuous phase velocity to ensure that the droplets are sufficiently spaced during the sol-gel translation. As a result, the high Cac also aids in placing the flows in the jetting regime.

4.2.3 Preparation of Dispersed Phase Solution

The dispersed phase polyimide precursor solutions (henceforth sol) were prepared at room temperature by mixing PMDA, DMBZ, and TREN in DMF, as per the process outlined by

Teo and Jana.231 Briefly, PMDA and DMBZ, dissolved separately in DMF, were mixed with a magnetic stirrer at 1000 rpm for 2 minutes. Subsequently, TREN, acetic anhydride, and pyridine were added to trigger crosslinking reactions and to allow for the chemical imidization of the polyamic acid. The solution was magnetically stirred for an additional

1.5 minutes. A typical polyimide sol sample with 5.6 wt% polymer concentration was prepared using 0.160 g PMDA, 0.159 g DMBZ, 0.015 g TREN, 0.50 g acetic anhydride,

0.47 g pyridine, and 5.0 mL of DMF. A higher than stoichiometric amount of TREN crosslinker was used to obtain appropriate gelation times.231 After stirring, the polyimide

90 sol was immediately transferred into a syringe pump for injection into the droplet generator.

4.2.4 Fabrication of Polyimide Aerogel Microparticles

The dispersed phase (polyimide sol) was injected at the specified flow rates through the 27

G needle (Figure 4.1a), while the continuous phase (silicone oil) was injected into the Tee connector through the Tygon tubing (Figure 4.1a). Both phases were delivered through

Chemyx syringe pumps (Stafford, TX) at varying flowrates, e.g., 0.1-0.4 mL/min for the dispersed phase and 2-8 mL/min for the continuous phase. The generated droplets were guided into a heated silicone oil bath for gelation and subsequent collection as shown schematically in Figure 4.1c. Actual images of gel droplets floating in silicone oil are presented in the inset of Figure 4.1c.

The silicone oil bath was heated to 80 °C to enable fast gelation of the polyimide sol droplets. The quick gelation (in less than 10 s) of polyimide sol at this condition prevented coalescence of the droplets, thereby forming discrete, spherical polyimide gel microparticles. The gel microparticles were aged further in silicone oil for 24 h, removed from silicone oil, and washed with chloroform. The gel microparticles were subsequently solvent exchanged sequentially with solvent mixtures consisting of 25 vol% acetone/75 vol% DMF, 50 vol% acetone/50 vol% DMF, 75 vol% acetone/25 vol% DMF, and finally with 100 vol% acetone each at 12 h intervals. In addition, the gels were further washed with 100 vol% acetone for an additional 5 times at 12 h intervals to remove as much of the

DMF, silicone oil, and chloroform from the gels as possible. The gels were subsequently solvent exchanged with liquid carbon dioxide in an autoclave by washing with 100 vol%

91 liquid carbon dioxide for 6 times at 1.5 h intervals. The liquid carbon dioxide filled gels were subsequently dried under supercritical condition of carbon dioxide at 50 °C and 11

MPa pressure to yield aerogel microparticles.

4.2.5 Characterization of Aerogel Microparticles

Interfacial Tension Measurements. The interfacial tension between the dispersed (DMF) and continuous (silicone oil) liquid phases was measured using a Du Noy tensiometer

(Interfacial Tensiometer 70545, Central Scientific Co., VA) with data taken in triplicate.

Gel Time. The gel time of polyimide solution at room temperature (19-20 C) was obtained from the crossover point of the storage and loss moduli of the sol measured using an ARES

G2 Rheometer (TA Instruments, New Castle, DE). This also allowed for the collection of complex viscosity data. The polyimide sol was poured into a solvent trap and loaded into the rheometer fitted with a 50 mm cone. The rheometer was operated at a constant angular frequency of 1 rad/s and at 10 % strain.

Gel and Aerogel Microparticle Size Distribution. The size distributions of gel and aerogel microparticles were studied using an Olympus BX51 optical microscope (OM). The images of a population of particles were collected and analyzed using the ImageJ software.

Typically, the sizes of more than 100 particles were considered in each case.

Aerogel Morphology. The morphology of aerogels was studied using a scanning electron microscope (SEM, JSM5310, JEOL, MA) at an accelerating voltage of 5 kV and emission current of 20 mA. A representative piece of fractured aerogel specimen was mounted on an aluminum stub using carbon tape, followed by sputter coating with silver (ISI-5400

Sputter Coater, Polaron, UK).

IR: Infrared spectra were collected on a Nicolet iS50 FTIR tri-detector spectrophotometer

(Thermo Scientific, MA).

TGA: Thermogravimetric analysis (TGA) was conducted under N2 with a Q50 thermogravimetric analyzer (TA Instruments, DE) using a heating rate of 20 °C/min, up to

700 °C.

Porosity and Pore Volume: Porosity was calculated from the values of skeletal (ρs) and bulk density (ρb) as shown in equation (3). The values of skeletal density were obtained using a helium pycnometer (AccuPyc II 1340, Micromeritics Instrument Corp., GA). Bulk density was obtained from the mass and volume of the aerogel specimens.

𝜌 푝표푟표푠푖푡푦 = (1 − 푏) × 100% (3) 𝜌푠

Total pore volume (Vtot) was calculated from the values of bulk and skeletal density, according to equation (4):

1 1 푉푡표푡 = − (4) 𝜌푏 𝜌푠

Note that only the bulk density of aerogel monoliths was obtained. The bulk density of aerogel microparticles could not be determined due to difficulty in isolating single particles for measurement of their diameter and weight.

Diameter Shrinkage: Bulk monolithic polyimide gels were synthesized in cylindrical form in a mold of 13 mm in diameter and 26 mm in length. Diameter shrinkage of supercritically dried aerogel specimens was obtained from the diameter of the final dried aerogel and the diameter of the gel. For microparticles, the apparent diameter shrinkage was calculated from the average diameter of the gel and aerogel particles.

Brunauer-Emmett-Teller (BET) surface area: BET surface areas of aerogel specimens were obtained from N2 adsorption-desorption isotherms at 77 K, using a Micromeritics

Tristar II 3020 analyzer (Micromeritics Instrument Corp. GA). For monolithic aerogels, the pore volume of macropores (Vmacropores) was deduced from total pore volume, Vtot and the volumes of mesopores (Vmesopores) and micropores (Vmicropores). For this purpose, CO2 adsorption-desorption isotherms at 273 K were obtained and combined with the N2 isotherms using the non-local density functional theory model. The fraction of macropores, mesopores, and micropores were subsequently calculated according to equation (5).

푉푚푎푐푟표푝표푟푒푠 푉푚푒푠표푝표푟푒푠 푉푚푖푐푟표푝표푟푒푠 휑푚푎푐푟표푝표푟푒푠 = ; 휑푚푒푠표푝표푟푒푠 = ; 휑푚푖푐푟표푝표푟푒푠 = (5) 푉푡표푡 푉푡표푡 푉푡표푡

4.3 Results and Discussion

4.3.1 Properties of Aerogel Monoliths

We first characterized the aerogel monoliths as benchmark materials. The aerogel monoliths were synthesized following the same recipe of polyimide sol as was used in the synthesis of gel microparticles. The gel was cast in 13 mm cylindrical polypropylene molds at room temperature and supercritically dried after several steps of solvent exchange as described in the experimental section.

The polyimide aerogel monoliths exhibited a diameter shrinkage of 10.8 %, skeletal density of 1.31 ± 0.01 g/cm3, and bulk density of 0.053 ± 0.004 g/cm3. These properties yielded a porosity of 96.0 ± 0.3 % using equation (3) and pore volume of 18.2 ± 1.5 cm3/g using equation (4).

Time Window for Droplet Generation. The gel time inferred from the crossover of the storage (G’) and loss moduli (G”) of polyimide aerogel monoliths at room temperature was found to be 1570 s (~26 min), as shown in Figure 4.4. The gel time provided a time window

94 for droplet generation from the sol without clogging of the droplet generator. In this work, droplets were generated within 5 min of preparation of the sol. The viscosity of the sol remained relatively constant over a 5 min period, varying from 1.98 mPa.s right after sol preparation to 2.03 mPa.s after 5 min. This guaranteed constant value of viscosity ratio of the continuous and dispersed phases during droplet generation. The generated droplets were conveyed into a bath of hot silicone oil as depicted schematically in Figure 4.1c. In the process, the sol droplets turned into gel microparticles due to faster crosslinking and chemical imidization reactions promoted by the hot silicone oil. An estimated gel time, as inferred from the time for the droplet to descend to the bottom of the silicone oil bath, was

~10 s. At the bottom of the oil bath, the particles remained discrete and did not coalesce, indicating that they had turned into crosslinked gel microparticles. Recall that the temperature of silicone oil bath was 80 C.

Figure 4.4: Evolution of storage (G’) and loss moduli (G”) of polyimide sol at room temperature. The inset shows the crossover point at 1570 s (~26 mins)

4.3.2 Morphology of Gel and Aerogel Microparticles

The process described in the experimental section allowed successful formation of discrete aerogel microparticles. The OM and SEM images of representative gel and aerogel microparticles are presented in Figure 4.5. Both gel (Figure 4.5a) and aerogel microparticles (Figure 4.5b) were discrete and of spherical shapes. The SEM image in

Figure 4.5c of a single aerogel microparticle shows a smooth surface and almost perfect spherical shape.

Figure 4.5: Optical microscope images of (a) gel and (b) aerogel microparticles, (c) SEM image of an aerogel microparticle.

A set of representative SEM images in Figure 4.6 of fractured aerogel microparticles reveals the following trends. First, the aerogel microparticles had highly porous internal structures as evident from the images in Figure 4.6b and 4.6d. Second, the aerogel microparticles contained distinct skin morphology (Figure 4.6a,c), much denser than the internal structures. Third, a comparison of the microparticle skin layer in Figure 4.6c with that of monolith skin layer in Figure 4.6e reveals that aerogel monoliths had much more open pore structures in their skin layers than in the case of aerogel microparticles. Fourth, one can envision the skin layers of both the monolith and microparticles as two- dimensional organizations of polyimide strands. Fifth, the monolith skin layers had thicker

96 polyimide strands (Figure 4.6e) compared to the microparticles (Figure 4.6c). Note that the gel time was about 26 min for the monoliths while it was estimated to be 10 s for the microparticles, as discussed earlier. The longer gel time in the case of monoliths also allowed much longer time for coarsening of the polymer domains during liquid-liquid demixing process.231

Figure 4.6: SEM images showing (a) skin layer and internal structure of a microparticle through sectioning, (b) porous internal structure of a microparticle through a tear in the skin layer, (c) skin layer on the surface of a microparticle, (d) cross-section of a microparticle,(e) skin layer of the monolith, and (f) cross-section of the monolith.

The IR and TGA data presented in Figure 4.7 present evidence that both the monolith and microparticles had similar chemical properties. Figure 4.7a shows IR spectra with the peaks at 730, 1100, 1380, 1716, and 1778 cm-1, confirming the presence of the imide functional group. The 1716 and 1778 cm-1 absorbance bands correspond to the symmetrical and asymmetrical stretching of the C=O group respectively, while the absorbance band at 730 cm-1 is due to the C=O bending. The absorbance bands at 1380 cm-1 correspond to C-N stretching in the imide rings, while the absorbance band at 1100 cm-1 is due to the imide

97 ring deformation. The absence of significant peaks at 1620 and 3000 cm-1 indicates that the majority of polyamic acid groups were imidized in the final aerogel materials. The TGA data in Figure 4.7b show that both monolith and microparticles exhibited good thermal stability at high temperatures. The monolith showed degradation curves typical of polyimides, with a weight loss of 3.5 wt% at 525 °C. The microparticles, however, showed an earlier onset of degradation, with a weight loss of 9.5 wt% at 525 °C. This 6 % higher weight loss at 525 °C observed in the case of microparticles can be attributed to residual polyamic acid in the structures that was not fully imidized. We infer that TGA traces show higher sensitivity than IR spectroscopy at detecting low levels of the polyamic acid functional groups in the microparticles. However, at 700 °C, both materials exhibited close values of char yield, e.g., 67.7 wt% for monolith and 67.6 wt% for microparticles. The

TGA traces presented in Figure 4.7b are characteristic of polyimide aerogels as reported in literature.203

Figure 4.7: (a) IR spectra and (b) TGA traces of polyimide aerogel monoliths and microparticles.

Microparticle Size Distribution. The microparticle size distribution was studied as functions of the dispersed and continuous phase flow rates. The flow rates were varied following three protocols: (1) continuous phase flow rate (Qc) was varied for a fixed dispersed phase flow rate (Qd), (2) Qd was varied for a given Qc, and (3) Qc and Qd were varied while keeping Qd/Qc ratio constant. The data presented in Figure 4.8 and Table 4.1 show the effects of various flow conditions on resultant gel and corresponding aerogel microparticle size distributions. The mean volume of an ensemble of aerogel microparticles shows 6-11% shrinkage in reference to the ensemble average volume of the corresponding gel microparticles. These values compare well with the shrinkage data of polyimide aerogel monoliths, ~10.8 % as discussed earlier.

Figure 4.8: Gel and aerogel microparticle size distributions. The dispersed phase (Qd) and continuous phase (Qc) flow rates are indicated in each chart. Red bars correspond to gel microparticles while green bars correspond to aerogel microparticles.

Table 4.1: Average gel and aerogel microparticle diameter and their associated shrinkage at different flow conditions. Dispersed phase Continuous phase Average Average gel aerogel Shrinkag Flowrate Flowrate microparticl Velocit Velocit microparticl e (mL/min (mL/min e diameter y (m/s) y (m/s) e diameter (%) ) ) (µm) (µm) 0.1 0.048 2 0.018 778 ± 106 713 ± 108 8.4 0.1 0.048 4 0.036 387 ± 50 345 ± 61 10.9 0.1 0.048 6 0.054 319 ± 37 293 ± 50 8.2 0.1 0.048 8 0.072 197 ± 27 182 ± 44 7.6 0.2 0.096 8 0.072 323 ± 33 287 ± 24 11.1 0.3 0.145 8 0.072 374 ± 37 350 ± 49 6.4 0.4 0.193 8 0.072 383 ± 41 342 ± 34 10.7 0.2 0.096 4 0.036 627 ± 88 572 ± 96 8.7 0.3 0.145 6 0.054 441 ± 67 401 ± 51 9.1

It is apparent from Figure 4.8 and the data presented in Table 4.1 that microparticles fabricated exhibited polydispersity in microparticle diameters. This is consistent with the jetting regime of droplet generation.107 In the jetting regime, a jet of dispersed phase extends from the outlet of the droplet generator and breaks up due to Rayleigh instabilities at a distance away from the outlet. Jetting was observed in the setup used in the present work due to high viscosity (48.6 mPa.s) of the silicone oil acting as the continuous phase compared to a viscosity of 1.98 mPa.s of the sol and the low interfacial tension of 3.0 mN/m between the two phases. The polydispersity in microparticle size arises from capillary perturbations of the extended jet, leading to uneven droplet formation.107 In addition, the average microparticle size is often larger than the diameter of the needle (200 µm). This is indicative of the widening-jet regime, 107 where the dispersed phase velocity is larger than the continuous phase velocity, leading to deceleration of the dispersed phase jet as it extends out into the continuous phase. This results in local buildup of the dispersed phase fluid in the thread undergoing breakup, resulting in uneven droplet sizes.

100

An exception to the above observation was microparticles fabricated with a dispersed phase flow rate of 0.1 mL/min and the continuous phase flow rate of 8.0 mL/min. This case exhibited an average particle diameter of 197 µm, very close to the diameter of the needle

(200 µm) used to deliver the dispersed phase. At these flow conditions, the continuous phase velocity (0.072 m/s) was found to be greater than the dispersed phase velocity (0.048 m/s), indicating that this flow condition had transitioned into the narrowing-jet regime.

This set of flow conditions in the narrowing-jet regime also shows the smallest particle size distribution, as a narrowing-jet (thinner jet) would suppress the capillary instability, thus producing more monodisperse droplets.124

The data presented in Figure 4.8a-d indicate that an increase of the continuous phase flowrate from 2 to 8 mL/min for a given dispersed phase flow rate of 0.1 mL/min helped reduce the average droplet size from 778 to 197 µm and also produced narrower microparticle size distribution, as reflected in the reduction of standard deviation values from 106 to 27 µm. The reduction of microparticle size can be attributed to an increase of the viscous stress with an increase of the continuous phase flow rate. Note that the interfacial force remained constant in all these experiments as no surfactant was added.

Also, the temperature was not varied during the droplet generation process. The reduction in microparticle size was predicted by Zhu and Wang,107 expounding on earlier work by

Cubaud and Mason232 as presented in equation (6):

퐷 = 푔(푘∗)(휑)0.5 (6) 푤표

In equation (6), D is the droplet diameter, Wo is the width of the outer channel, g(k*) is a function of the wavenumber k* with g(k*) ~ 1.43, and ϕ is the ratio of the dispersed and

101 continuous phase flow rates, Qd/Qc. A comparison of the experimental and predicted droplet size based on equation (6) is shown in Figure 4.9a.

The relationship in equation (6) developed for the narrow-jetting regime also shows good correlation with the experimental droplet diameter data, particularly for higher continuous phase flow rates if the continuous phase velocity was higher or similar to the dispersed phase velocity. However, at a low continuous phase flow rate of 2 mL/min, a significant deviation of the experimental droplet size occurred from the prediction of equation (6). The continuous phase velocity of 0.07 m/s and the dispersed phase velocity of 0.19 m/s in this case, belong to the widening-jet regime, where equation (6) is no longer valid. This aspect, however, will be considered in a future study.

Figure 4.9: Comparison of experimental droplet size with model prediction of equation (6). (a) Continuous phase flow rate 2-8 mL/min, dispersed phase flowrate 0.1 mL/min. (b) Dispersed phase flowrate 0.1-0.4 mL/min, continuous phase flowrate 8 mL/min.

The effects of dispersed phase flowrate are now analyzed. It is apparent from the data presented in Figure 4.8e-h that an increase of the dispersed phase flow rate from 0.1 to 0.4 mL/min also produced an increase of the average diameter of gel microparticles from 197

102

± 27 to 383 ± 41 µm. The experimentally obtained mean droplet diameter values were also compared to the prediction of equation (6) as presented in Figure 4.9b. A good correlation is found for continuous phase velocity higher or equal to the dispersed phase velocity, for example, when the dispersed phase flow rate was less than 0.3 mL/min. At higher dispersed phase flow rates such as 0.4 mL/min, the system once again transitioned to a widening-jet regime, resulting in a deviation from the prediction of equation (6).

Finally, the effects of an increase of both the dispersed and continuous phase flow rates were investigated by varying the values of Qc and Qd while keeping the ratio Qd/Qc constant, e.g., Qc=0.1 mL/min; Qd=2 mL/min to Qc=0.4 mL/min; Qd=8 mL/min. The data presented in Figure 4.8i-l and Table 4.1 shows that the gel microparticle size reduced from

778±106 µm to 383±41 µm with an increase of the flow rates in the dispersed and continuous phases. It is apparent from such data that the higher the continuous phase flow rate for a given ratio Qd/Qc, the smaller the average gel microparticle diameter and the narrower the distribution of gel microparticle size. For the above flow conditions, the dispersed phase velocity was 2.67 times the continuous phase velocity, leading to significant deviations from the predicted droplet size as in equation (6). In view of this, a comparison with the narrowing-jet regime model was not attempted.

4.3.3 Effect of Temperature on Microparticles.

Recall that silicone oil in the oil bath was heated to expedite crosslinking and gelation of the sol droplets. At this point, it is unknown if the temperature of the oil bath exerted an effect on the microparticle morphology. In view of this, the silicone oil bath temperature

103 was varied in the range 60-100 C. The morphology of the aerogel microparticles produced with dispersed and continuous phase flow rates of respectively 0.3 and 6 mL/min was analyzed. The gel and aerogel microparticle size distributions are presented in Figure 4.10.

Table 4.2 lists the average size of gel and aerogel microparticles along with data on diameter shrinkage and BET surface area. It is evident that gel microparticle diameter reduced from 563±140 µm to 395±44 µm with an increase of the temperature of the oil bath from 60 to 100 C. The OM images in the insets of Figure 4.10 present visual impression that with an increase of the oil bath temperature, the gel microparticles experienced higher shrinkage and correspondingly higher density, the latter inferred from an increase of the opacity of the microparticles.

Figure 4.10: Gel and aerogel microparticle size distributions as function of silicone oil bath temperature. Dispersed and continuous phase flow rates were respectively 0.3 and 6 mL/min.

104

Table 4.2: Average gel and aerogel diameter, shrinkage, and BET surface area of aerogel microparticles as function of silicone oil bath temperatures. Dispersed and continuous phase flow rates were respectively 0.3 and 6 mL/min. Average BET Average gel aerogel Diameter Surface Temperature microparticle microparticle shrinkage area (m2/g) (°C) diameter diameter (%) (µm) (µm) 60 563 ± 140 546 ± 129 3.0 484 70 516 ± 109 494 ± 102 4.3 431 80 492 ± 93 452 ± 68 8.1 412 90 441 ± 67 405 ± 56 8.2 409 100 395 ± 44 376 ± 52 4.8 369

The hysteresis loops seen in BET isotherms in Figure 4.11 indicate significant mesoporosity (pore diameter 2-50 nm) in aerogel microparticles. The surface area data listed in Table 4.2 indicate that the aerogel microparticles had high surface area, 360-484 m2/g. The surface area of corresponding monolithic aerogels was higher, 537 m2/g. The relatively lower BET surface area in the case of aerogel microparticles is due to significant fraction of skin layers that formed on the particle surfaces. Note that aerogel microparticles reported in Table 4.2 had significantly higher surface area to volume ratio (11-15 1/mm) than the monoliths (0.4 1/mm). The surface area to volume ratio of cylindrical aerogel monoliths was calculated based on actual diameter of 11.6 mm and length of 22.6 mm.

The BET surface area reported in Table 4.2 reduced with an increase of bath temperature, e.g., 484 m2/g at 60 °C to 369 m2/g at 100 °C. This is commensurate with a reduction of area under the hysteresis loops formed by the adsorption and desorption isotherms in the

BET curves (Figure 4.11) indicating reduction of mesopore fractions.

105

3000 60 70 80

90 /g STP) /g 3 100 2000

1000 QuantityAbsorbed (cm 0 0.0 0.2 0.4 0.6 0.8 1.0

Relative Pressure (P/P0)

Figure 4.11: BET isotherms for aerogel microparticles fabricated with varying silicone oil bath temperature.

We now examine the morphology of the surfaces of aerogel microparticles. Figure 4.10a shows that the microparticles fabricated at 60 °C exhibit porous skin layer morphology similar to what was observed for monolithic aerogels, e.g., in Figure 4.6e. Figure 4.12a also shows that the skin layer of the microparticles was comprised of fibrillar strands of 10

– 20 nm diameter. The particle surface remained porous and retained fibrillar morphology as the temperature of the silicone oil bath was increased; however, the skin layers developed wrinkles. The degree of wrinkling increased with an increase of the temperature from 70 to 90 °C, as shown in Figure 4.12b-d. Such wrinkles can be attributed to outward flow of DMF from inside the droplets caused by higher solubility of DMF in silicone oil at higher temperature. This also caused a reduction of the volume of gel particles, measured from reduction of particle diameter as presented in Table 4.2. However, at 100 °C, the

106 fibrous morphology of the particle surface was lost (Figure 4.12e), as the miscibility of

DMF with silicone oil reached a point where the outflow of DMF was so high that the polyimide precipitated out of the solution within a thin shell near the surface. This is also reflected in OM images in Figure 4.10, where the microparticles appeared dense and dark compared to the translucent microparticles obtained at other temperatures.

Figure 4.12: Skin layer of aerogel microparticles synthesized at various silicone oil bath temperature of (a) 60 °C, (b) 70 °C, (c) 80 °C, (d) 90 °C and (e) 100 °C.

We anticipated that the use of high silicone oil bath temperature would also increase the rate of polyamic acid formation and expedite imidization reactions, thus promoting faster gel formation. In view of the aerogel microparticle morphology discussed above in conjunction with Figure 4.12, we now accept that higher silicone oil temperature also increased the solubility between DMF and silicone oil. In this context, one can describe the fate of a polyimide sol droplet as it enters hot silicone oil. First, it experiences a temperature spike in a thin shell within the droplets. Second, the shell gels almost immediately (< 10 s) thus forming a solid porous skin around the droplet. Third, almost simultaneously, higher solubility of DMF in silicone oil results in a net fluid flow out of the liquid droplets into the continuous phase, thus resulting in shrinkage and wrinkling of the already formed shell.

The increase in solubility of DMF in silicone oil with an increase of temperature was inferred qualitatively in a separate experiment as presented Figure 4.13.

107

Figure 4.13: DMF/silicone oil two-phase system at various temperatures.

As temperature decreases, two effects can be seen. First, the volume fraction of DMF (top liquid) increases slightly. Second, the silicone oil fraction (bottom liquid) becomes increasingly cloudy. This increase in opacity is attributed to previously dissolved DMF precipitating out of the silicone oil at lower temperature. Both these observations show that the solubility of DMF in silicone oil is higher at higher temperatures.

In view of the strong temperature effects on aerogel microparticle morphology, a separate set of experiments were conducted on polyimide gel monolith to determine if higher imidization temperature also had any effect on skin morphology, bulk morphology, and porosity of the monolithic aerogels. For this purpose, monolithic polyimide gels were cured in an oven at 20, 40, 60, 80 and 100 °C followed by supercritical drying to obtain corresponding aerogel specimens. Table 4.3 shows the effect of temperature on properties of monolithic polyimide aerogels.

108

Table 4.3: Shrinkage, porosity and pore size distribution of monolithic aerogels cured at various temperatures Temperature Diameter Shrinkage Porosity Micropore Mesopore Macropore (°C) (mm) (%) (%) percent percent percent 20 11.2 ± 0.0 13.9 ± 0.0 97.0 ± 0.0 0.5 2.1 97.4 40 11.3 ± 0.0 13.3 ± 0.2 97.2 ± 0.0 0.7 2.7 96.6 60 11.5 ± 0.2 12.4 ± 0.2 97.6 ± 0.0 0.6 2.4 97.0 80 11.7 ± 0.1 10.4 ± 0.3 98.0 ± 0.0 0.5 2.0 97.5 100 11.7 ± 0.1 10.0 ± 0.3 97.6 ± 0.0 0.5 2.9 96.6

It is seen that an increase of imidization temperature led to a reduction of shrinkage of the aerogels. For monoliths synthesized at 60 °C, the average diameter of the final aerogels was 11.5 mm, while monoliths synthesized at 100 °C had an average diameter of 11.7 mm.

Comparing these two monoliths, a 1.7 % increase in aerogel diameter is observed as the curing temperature was increased from 60 to 100 °C. Under the same thermal treatment, as discussed earlier, the aerogel microparticles experienced a 30 % reduction in average microparticle diameter from 563 to 395 µm as the silicone oil bath temperature was increased from 60 to 100 °C. The data in Table 4.3 also show that calculated porosities of the monoliths are within a narrow range of 97.0 to 98.0 %. The pore size distribution also shows that the monoliths cured at various temperature are predominantly macroporous, with macropore fraction in the range of 96.6 to 97.5 %, indicating that internal strand structure did not change significantly at higher temperature. The data presented in Table

4.3 confirm that wrinkling of aerogel microparticle skin layer was caused by shrinkage of the outer layer of microparticles due out flow of DMF into silicone oil promoted by higher solubility.

109

4.3.4 Morphology of core-shell hollow microspheres

With the success of fabricating aerogel microparticles, we next shifted our efforts to fabricate core-shell hollow aerogel microspheres. This was done by modifying the droplet generator to add another annular flow into the system. This increased the cost of the microfluidic device by 100 %, to a total cost of USD 2.00. This setup is shown in Figure

4.14a. In this setup, silicone oil was used in the inner flow, DMF with gel precursors in the shell flow and silicone oil in the outer flow. Similar to the fabrication of the microparticles, once the hollow microspheres were generated, they were transferred to a heated silicone oil bath (70 °C) for gelation. The subsequent fabrication steps to make aerogels were identical to the microparticles described above. Figure 4.14 shows that we successfully fabricated both core-shell hollow gel and aerogel microspheres with a particle size of 774 to 1071 µm and 778 to 1014 µm respectively, with shell thicknesses of 4 to 93 µm. The shells of the hollow microspheres continue to retain high porosity, possessing fibrillar strand morphology, similar to the aerogel microparticles described above.

Figure 4.14: (a) Droplet generator for core-shell hollow microspheres, (b) optical microscope images of hollow gel microspheres, (c) SEM image of fractured hollow aerogel microsphere and (d) shell thickness of hollow aerogel microsphere.

In this study, we also investigated the effect of different flow rates on the final particle size and shell thickness. Throughout this portion of the study, the inner flow was kept constant at 0.08 mL/min to minimize the flow rates of the shell and outer flow required for stable

110 droplet generation. The shell flow rate was varied from 0.2 to 0.5 mL/min and the outer flow rate was varied from 1 to 4 mL/min. The effect of flow rate on gel and aerogel microsphere size distributions is shown in Figure 4.15 and Table 4.4.

Figure 4.15: Gel and aerogel hollow microspheres size distributions for different flow rates. The number indicated above each graph represents the inner/shell/outer flow rates in mL/min. Red bars correspond to gel microparticles while green bars correspond to aerogel microparticles.

Table 4.4: Average gel and aerogel hollow microspheres diameter and shell thickness Inner Shell Outer Average Average Shell Flow Flow Flow gel aerogel Shrinkage Thickness Rate Rate Rate diameter diameter (%) (µm) (mL/min) (mL/min) (mL/min) (µm) (µm) 1071 ± 1014 ± 5.3 1 35.6 123 125 0.2 2 812 ± 54 828 ± 82 12.9 - 2.0 3 804 ± 24 808 ± 43 7.9 - 0.5 0.08 4 774 ± 30 778 ± 52 4.1 - 0.5 0.3 847 ± 58 830 ± 111 25.4 5.9 0.4 2 789 ± 90 810 ± 118 46.7 2.7 0.5 847 ± 137 833 ± 123 93.4 1.7 .

111

Increasing the outer flow rate from 1 to 4 mL/min decreases both the average gel and aerogel microspheres diameter from 1071 to 774 µm and from 1014 to 778 µm respectively. This decrease is also shown in the hollow microsphere size distribution shown in Figure 4.15a to 4.15d. Similar to the microparticles, the increase in outer flow rate increase the continuous phase velocity, of which dominates the determination of the hollow microsphere size, as discussed earlier. The narrowing of size distribution with increasing outer flow rate can once again be explained by the transition from widening-jet to narrowing-jet. Increasing the outer flow rate also decreases the shell thickness from 35.6 to 4.1 µm. This decrease in shell thickness can be attributed to the increase in viscous stress from increasing outer flow rate, which result in faster nipping off of the droplets from the jet tip. As the core droplet of silicone oil was generated upstream and not susceptible to deformation, the only way overall droplet size could be reduced due to increasing viscous stress was through the reduction of the shell flows around the silicone oil droplet core.

By increasing the shell flow rate, the average gel microsphere diameters remained relatively constant, but the size distribution broadened considerably, as seen from Figure

4.15e to 4.15h. Increasing shell flow rate increases the dispersed phase (inner and shell) velocity, once again shifting the system to the widening-jet regime, leading to increased polydispersity. This trend is similar to what was reported by the microparticle system described earlier. The increasing shell thickness from 12.9 to 93.4 µm can be attributed to the increased in mass flow rate of the shell flow, comprising of gel precursors in the DMF which subsequently crosslink to form the shell.

Table 4.4 also shows that there is negligible shrinkage of the gels upon supercritical drying into aerogels, as shown by the close overlap of the red (gel) and green (aerogel) bars in the

112 microsphere size distribution. This low shrinkage is to be expected as shrinkage comes from the gel component of the hollow microspheres, which in this case only account for a thin core section that is > 4% of the total diameter of the microsphere. The negative values of shrinkage in Table 4.4 do not indicate that the aerogels expanded upon supercritical drying, but rather the difference in the size distributions of the gel and corresponding aerogels

4.4 Conclusion

The results presented in this paper establish that polyimide gel and aerogel microparticles can be successfully synthesized using a co-flow microfluidic set up from surfactant-free oil-in-oil systems. The rapid sol to gel transition is achieved using a hot silicone oil bath decoupled from the droplet generation system, thus avoiding coalescence of sol droplets and expediting transformation of sol droplets into discrete gel droplets of spherical structure. The use of a microfluidic droplet generator enables the control of microparticle size through both the dispersed and continuous phase flowrates, resulting in the synthesis of aerogel microparticles with average diameters of 197 µm and narrow microparticle size distribution. The results also show significant wrinkling of the aerogel microparticle surfaces attributed to out flow of DMF into hot silicone oil and shrinkage of the thin shell near the microparticle surfaces.

In addition, this process can be translated to fabricate aerogel core-shell hollow microspheres. It is our belief that this is the first time this morphology has been successfully achieved. These core-shell hollow microspheres have average diameters of 800 µm, with

113 shell thicknesses that can be as low as 4 µm. This fabrication method of aerogel microparticles and hollow microspheres can easily be adapted to other material systems through optimization of gel times, temperature, flow rates and selection of disperse and continuous phases. This synthesis of size controllable microparticles and core-shell hollow microspheres opens up these aerogel structural forms for potential drug delivery applications.

114

CHAPTER 5

OPEN CELL AEROGEL FOAMS VIA EMULSION-TEMPLATING

Reproduced from: [Teo, N.; Jana, S.C.; Open Cell Aerogel Foams via Emulsion-Templating. Langmuir 2017. 33. 12729-12738. DOI: 10.1021/acs.langmuir.7b03139]

Abstract Water-in-oil emulsion-templating method is used in this work for fabrication of open cell aerogel foams from syndiotactic polystyrene (sPS). A surfactant-stabilized emulsion is prepared at 60 C-100 C by dispersing water in a solution of sPS in toluene. The sPS gel, formed upon cooling of the emulsion to room temperature, locks the water droplets inside the gel. The gel is solvent exchanged in ethanol and then dried under supercritical condition of carbon dioxide to yield the aerogel foams. The fabricated aerogel foams show significant fraction of macropores (with diameters in the micrometer range) defined as macrovoids that originated from the emulsified water droplets. In conjunction, customary macropores of diameter 50-200 nm are also present from the sPS gel structure. The macrovoids add additional openness to the aerogel structures. This paper evaluates the structural characteristics of the macrovoids, such as diameter distribution, macrovoid interconnect density, and skin layer density in conjunction with the final aerogel foam properties.

115

5.1 Introduction

Porous materials have been used in a variety of industrial applications such as membranes for gas and liquid filtration,233,234 catalyst supports,235 substrates for electrical components,236 gas sequestration reservoirs237 and absorbents for chemical and oil spills.238

The versatility of porous materials in undertaking multiple roles stems from their high porosity, large surface areas, and low density, all the while offering sufficient mechanical integrity to be free-standing. Polymeric foams were first reported in patent literature in

1935239 and covered extensively in a number of authoritative monographs.68,240,241 Foams are a class of widely used porous materials with micrometer-sized, air-filled voids organized in either open-celled or closed-celled forms inside a host polymer. Two key methods dominate fabrication of polymer foams: (1) in-situ chemical foaming of thermoset polymers, e.g., polyurethanes242 and (2) foaming via the use of physical blowing agents in conjunction with extrusion, e.g., foaming of polystyrene using hydrocarbons.243

Attempts have been made in literature to incorporate hierarchical porosity into material structures to improve their properties, mimicking natural systems such as wood, bamboo, and bone.244 For example, Wong et al.245 fabricated hierarchical polyHIPEs consisting of surfactant-stabilized droplets (~15 µm) and particle-stabilized Pickering emulsion droplets

(>100 µm), while Elsing et al. 246 produced hierarchical porous structures in polystyrene foams via foaming of styrene-in-water emulsions. Gu and Jana208 recently fabricated open cell polyurea aerogel foams by using the templates obtained from a co-continuous immiscible polymer blend system. These authors dissolved one polymer phase from a co- continuous blend to create a template material with micrometer-sized cavities in which polyurea gel was synthesized via sol-gel chemistry. The second polymer phase was then

116 removed by using a solvent and the gel was supercritically dried to yield aerogel foam structures with significant mesopores and micrometer size voids. It was found that the small amount of residual template polymers remaining in the aerogel foam structure affected the bulk density and the total porosity. To circumvent the above issue and to develop a more versatile method of fabrication of hierarchical polymer aerogel foam structures, an emulsion-templating method was used in this work in conjunction with a thermo-reversible gelation system.

Aerogels, first reported in 1931,1 are obtained by replacing the liquids in the gel structures with air. The precursor gels are synthesized either through sol-gel reactions9 or a thermo- reversible gelation process.247 Kistler pioneered the process of silica aerogel fabrication that found widespread use as thermal insulation1,248 and in aerospace applications.249,250

Over time, aerogels have been synthesized from polymeric materials such as polyurea,183 polyimide,33 and polystyrene59 to overcome the mechanical limitations of silica aerogels.251

One advantage of polymeric aerogels is that a large variety of different structural morphologies can be accessed either from using an array of different monomers undergoing step-growth polymerization reactions207, or through thermodynamic phase transitions following different kinetic pathways of gelation.187,252

The aerogel foams in this work were fabricated by water-in-oil emulsion templating79,253–

255 of a polymer sol that eventually turned into a gel via a thermo-reversible sol-gel transition.64 Syndiotactic polystyrene (sPS) solution in toluene was considered as the continuous phase in this work due to its fast crystallization kinetics, enabling rapid thermo- reversible gelation, which relaxes the constraints on time limits of emulsion stability.50 The regular tacticity of sPS enables the incorporation of solvent molecules and allows

117 crystallization into the helical δ-from intercalate. A solution of sPS and toluene prepared at high temperature and subsequently cooled under ambient conditions passes through crystallization, binodal, and spinodal curves and undergoes liquid-liquid demixing via spinodal decomposition.54 The polymer-rich phase then crystallizes via nucleation and growth process. The system is vitrified as the temperature drops below the glass transition temperature of the polymer, thus locking in the fibrillar polymer structure55 and resulting in a physically cross-linked network of polymer strands (average diameter of 50 nm). The sol-gel transition typically takes place at above 50 C, making the gel structure experimentally accessible at room temperatures. In contrast, atactic polystyrene is only able to form gels at sub-zero temperatures and its crystallization is significantly harder due to its irregular conformation.44 Isotactic polystyrene is able to form stable gels at room temperature, but the long gelation times of up to two weeks makes the system impractical due to the issues of emulsion stability in the emulsion templating step.45

The aqueous phase in the emulsion system used in this study is responsible for creating large fractions of macropores with typical diameters of 20 µm, defined in this work as macrovoids. In conjunction, the 50-200 nm diameter macropores that customarily form in sPS gels, lead to a structure reminiscent of open cell polymer foams with porous skin layers. The additional openness due to macrovoids may lead to possible adaptation of aerogels in air or liquid filtration applications.

118

5.2 Experimental Section

5.2.1 Materials

Syndiotactic polystyrene (Mw ~ 300,000 g/mol) was acquired from Scientific Polymer

Producers Inc. (Ontario, NY). Surfactants SPAN 80® (trademark of Croda Inc) and F127®

(trademark of BASF), chlorophenylsilane, and toluene were purchased from Sigma Aldrich

(Milwaukee, WI). Ethanol was acquired from Decon Laboratories Inc. (King of Prussia,

PA).

5.2.2 Fabrication of Aerogel Foams

A water-in-oil (W/O) emulsion-templating method was used in this work to obtain micrometer size water droplets dispersed in an organic sol. The sol transitions into an organic gel, thus locking the emulsified water droplets within the gel structure. This process resembles the fabrication of polyHIPEs 73,256,257, with the exception that the oil-phase in this study is an organic sol which later transitions into a gel with its inherent pores filled by the organic solvent.

The continuous phase organic sol was prepared by dissolving syndiotactic polystyrene

(sPS) in toluene at 100 C at a solid concentration of 0.06 g/mL. A typical sample with dispersed phase content of 33 vol% and surfactant concentration of 1.5 vol% of the continuous phase was prepared as follows. Toluene (2.67 mL), 1.33 mL of water, 0.04 mL of SPAN 80®, and 0.16 g sPS pellets were added into a sealed vial of internal diameter

1.92 cm. The amounts were adjusted accordingly to obtain samples of varying dispersed phase content (25, 33 and 50 vol%) and surfactant concentrations (1.5, 7.5 and 15 vol%).

119

Next, the vial was heated in an oil bath at 100 C for one hour to enable dissolution of sPS in toluene. The resultant material was removed from the oil bath and left to cool under ambient conditions while the components were continuously mixed using a magnetic stir bar of length 1.27 cm and diameter 3.175 mm. In the process, a water-in-oil emulsion was formed. It was found that sPS gelation occurred at around 50 C. In view of this, the emulsion was allowed to cool to 60 C in approximately 2 min under ambient cooling conditions and poured into glass molds. This facilitated material handling before the viscosity increased appreciably near the gelation point. The glass molds were immediately quenched in an ice-water bath and left to stand for 5 hours. The resultant gels were soaked successively in ethanol to remove water and surfactants and then in liquid carbon dioxide to replace ethanol. The interconnected macropores of sPS and the macrovoids facilitated removal of water and surfactants and exchange of ethanol with liquid carbon dioxide. The gel was subsequently dried under supercritical condition of carbon dioxide at 50 C and 11

MPa.

In the rest of the paper, the numbers in specimen designation “sPS_x_y_z” indicate a dispersed phase content of x vol%, surfactant concentration of y vol%, and a magnetic stirrer mixing speed of z rpm. Recall that numbers x and y represent the vol% with respect to the volume of the continuous phase.

5.2.3 Characterization of Water-in-Oil Emulsions Emulsion Droplet Size. The water droplet size in W/O emulsions prepared without sPS and with varying surfactant concentrations was measured using an Olympus BX51 optical

120 microscope fitted with a heating stage controlled by an Instec STC200 temperature controller. A drop of the emulsion was first placed onto a microscope slide with a cover slip and inserted into the heating stage at the required temperatures of 50, 60, and 70 C.

Optical images were recorded at prescribed time periods. The droplet size distribution was obtained from the analysis of the optical images using the ImageJ software. In each case, more than 200 droplets were considered.

Interfacial Tension Measurements. The interfacial tension between water and toluene was measured using a Du Noy tensiometer (Interfacial Tensiometer 70545, Central Scientific

Co., VA) at several temperatures. First, the surfactant was dissolved in toluene or water for

30 minutes, depending on its hydrophilic lipophilic balance (HLB) value. Next, 20 mL of the aqueous solution was added into a clean glass container and the toluene solution was gently added onto the top of the water phase by pouring the toluene solution down a clean glass slide to prevent any emulsification or droplet formation. The glass container was then heated to 80 C, below the boiling points of both water and toluene, and transferred to the tensiometer for measurement of interfacial tension values at temperatures of 50, 60, and 70

C.

5.2.4 Characterization of Aerogel Foams Aerogel Foam Morphology and Macrovoid Size Distribution. The morphology of the aerogel foams was studied using a scanning electron microscope (JSM5310, JEOL, MA).

An accelerating voltage of 2 kV and emission current of 20 mA was used to capture the

SEM images. The aerogel foams were fractured, and a representative piece was mounted on an aluminum stub using carbon tape, followed by sputter coating with silver (ISI-5400

121

Sputter Coater, Polaron, UK). The macrovoid size distribution was obtained from the analysis of SEM images using ImageJ software. Typically, more than 150 macrovoids were considered for determining the macrovoid size distribution for each specimen.

Porosity: Porosity was calculated from the values of skeletal (ρs) and bulk density (ρb) as shown in equation (1). The values of skeletal densities were obtained using a helium pycnometer (AccuPyc II 1340, Micromeritics Instrument Corp., GA). Bulk density of the aerogel foams was calculated from the mass and volume of the foam specimens.

𝜌 푝표푟표푠푖푡푦 = (1 − 푏) × 100% (1) 𝜌푠

BET surface area: Brunauer-Emmett-Teller (BET) surface area of the aerogel foams was obtained from N2 adsorption-desorption isotherms at 77 K using a Micromeritics Tristar

II 3020 analyzer (Micromeritics Instrument Corp. GA).

5.3 Results and Discussion 5.3.1 Emulsion Stability Emulsion stability in a 10-minute window was important in this work as sPS sol turned into gel in approximately 3 minutes from the time the emulsion was poured into the glass mold. In view of this, emulsion stability over a period of time and corresponding water droplet size distribution at various temperatures and surfactant concentrations were investigated.

122

5.3.2 Effect of Temperature on Emulsion Formation.

Emulsions with surfactant concentration of 1.5 vol% SPAN 80® were prepared at 50, 60 and 70 C. This was achieved by first heating the emulsion to 100 C in an oil bath, mimicking the gel formation process. The emulsion was then removed from the oil bath and allowed to air cool, while maintaining mixing. Once the emulsion reached appropriate temperature (50, 60 or 70 C), a droplet was placed on the glass slide, covered with a cover slip, and the assembly was allowed to reach desired equilibrium temperature (in less than a minute). Optical images of the droplets were taken as soon as the emulsion was first placed on the slide and after three minutes.

It was seen that emulsions prepared at higher temperatures led to a shift of droplet size distribution to larger size. This was exhibited by an increase of the average droplet size registered right after placement of the emulsion droplet from 11.8 to 13.9 to 19.4 µm respectively for emulsion temperatures of 50, 60, and 70 C, as shown in Table 5.1.

Table 5.1: Interfacial tension and water droplet size with 1.5 vol% SPAN 80® Interfacial Average Maximum Minimum Temperature Tension with Droplet Droplet Droplet (°C) surfactant of 1.5 Diameter Diameter Diameter vol% (mN/m) (µm) (µm) (µm) 50 2.0 11.8 ± 9.0 58.0 2.3 60 3.6 13.9 ± 8.8 56.1 3.0 70 5.7 19.4 ± 17.7 163.7 3.7

In addition, emulsions prepared at lower temperature were found to be more stable. For example, droplet size shifted to higher values after 3 minutes for emulsions prepared at higher temperature as shown in Figure 5.1.

123

Figure 5.1: Droplet size distribution of emulsions prepared at (a) 50 C, (b) 60 C, and (c) 70 C with 1.5 vol% SPAN 80®. The emulsion prepared at 70 C formed phase separated bilayer after 3 minutes; no dispersed droplets could be recorded for this case, as in Figure 5.1c.

5.3.3 Effect of Temperature on Preformed Emulsions

A second set of experiments was conducted with slightly different procedures to investigate the effect of temperature on droplet size distribution. In this case, emulsions were prepared at 20 C and placed onto the glass slide with the cover slip. The emulsions were subsequently heated to 50 C, allowed to equilibrate for 1 min, and optical image of the droplets was captured. The emulsion was then then heated to 60 and 70 C and respective images were taken. This set of experiments allowed study of the effect of gradual temperature change on the emulsion properties, isolating the effect of mixing or agitation.

The images in Figure 5.2(a-c) show that water droplets grew in size with an increase of the temperature. Table 5.2 lists the values of maximum, minimum, and average droplet diameter at each temperature inferred from the distributions presented in Figure 5.2(d-f).

124

Figure 5.2: Optical microscope images of water (lighter area) emulsions with 1.5 vol% SPAN 80® at (a) 50 C, (b) 60C, and (c) 70C. Droplet size distribution of the corresponding emulsions are presented in (d) 50C, (e) 60C, and (f) 70C.

The data in Table 5.2 show that the average droplet size at three temperatures remained almost constant at 13-14 µm, while the standard deviation grew with temperature and the maximum droplet size increased from 63.8 µm at 50 C to 247.0 µm at 70 C. The minimum droplet size changed from 1.4 to 2.9 µm with a 20 C increase in temperature from 50 C. The above trends indicate that droplet coalescence was favored at higher temperature.

Table 5.2: Interfacial tension and droplet size with 1.5 vol% SPAN 80®. Average Maximum Minimum Interfacial Temperature Droplet Droplet Droplet Tension at 60 C (°C) Diameter Diameter Diameter (mN/m) (µm) (µm) (µm) 50 2.0 12.7 ± 10.9 63.8 1.4 60 3.6 12.6 ± 13.8 113.4 2.2 70 5.7 13.8 ± 29.8 247.0 2.9

125

In an effort to understand why coalescence events increased with an increase of temperature, the interfacial tension values were measured in the case of water/toluene and water/toluene/SPAN 80® system at 50-75 C and presented in Figure 5.3.

Figure 5.3: Effect of temperature on interfacial tension of water/toluene (filled square) and water/toluene/1.5 vol% SPAN 80®(filled triangle)

The data in Figure 5.3 presents two interesting trends. First, for water/toluene system without the surfactant, interfacial tension reduced with an increase of temperature, e.g., from 33.5 mN/m at 50 C to 30.8 mN/m at 75 C. This can be attributed to an increase of the mutual solubility between toluene and water at higher temperature. Second, as expected, the addition of SPAN 80® surfactant lowered the value of interfacial tension from 33.5 mN/m for water/toluene system to 2 mN/m for water/toluene/1.5 vol% SPAN

80® system, both measured at 50 C. However, the interfacial tension increased several

126 folds with temperature in the presence of SPAN 80®, e.g., from 2 mN/m at 50 C to 7.6 mN/m at 73 C. One can attribute this to loss of solubility in water of the nonionic surfactant SPAN 80® with an increase of temperature and simultaneous increase of solubility in the organic phase; this also resulted in a reduction of HLB value of the surfactant at higher temperature.258 The hydrophilic heads of the surfactant that bind to the aqueous phase dehydrates at higher temperature due to weakening of hydrogen bonding.259,260 As a consequence, the surfactant molecules migrate away from the water/toluene interface into the bulk toluene phase and an increase of the interfacial tension ensues as is evident in Figure 5.3. In this scenario, water droplets experience higher frequency of coalescence events and larger water droplets are obtained.

5.3.4 Effect of Surfactant Concentration on Emulsions

The data in Table 5.3 taken for emulsions prepared at 60 C indicate that an increase of surfactant concentration led to formation of smaller droplets and narrowing of the droplet size distribution, as shown in Figure 5.4. In addition, the emulsions were found to be more stable at higher surfactant concentration and experienced a smaller shift in droplet size distribution over a 3-minute period.

Table 5.3: Interfacial tension and droplet size distribution of water/toluene/SPAN 80® for different surfactant concentrations at 60 C. Average Maximum Minimum Surfactant Interfacial Droplet Droplet Droplet Concentration Tension at 60oC Diameter Diameter Diameter (vol%) (mN/m) (µm) (µm) (µm) 1.5 3.5 20.6 ± 14.7 84.6 3.7 7.5 1.0 13.0 ± 7.6 46.1 3.3 15 0.8 12.1 ± 9.0 53.3 1.6

127

Figure 5.4: Effects of SPAN 80® surfactant concentration at 60 C: (a,d) 1.5 vol%, (b,e) 7.5 vol% and (c,f) 15 vol%. (a-c) Optical microscope images, (d-f) droplet size distribution of the corresponding emulsions.

As is expected, the interfacial tension value reduced with an increase of surfactant concentration (Table 5.3), which favored creation of additional interfacial area between water and toluene phases as evidenced by the creation of smaller water droplets as shown in Figure 5.4. The diameter distribution is quite broad with an average diameter of 20.6 µm and standard deviation 14.7 µm for surfactant concentration of 1.5 vol% (Table 5.3). The diameter distributions became much narrower with average diameter of 13.0 µm and 12.1

µm for surfactant concentration of 7.5 and 15 vol% respectively, also shown in Table 5.3.

The maximum and minimum droplet size also showed strong dependence on surfactant concentration, e.g., both became smaller with an increase of surfactant concentration

(Table 5.3). The smaller size droplets also settled at much slower rate as presented in Figure

5.5, thus contributing to the stability of the emulsions. Water droplets in emulsion with 1.5

128 vol% SPAN 80® settled appreciably within a minute; in view of this, emulsions with 1.5 vol% SPAN 80® were not considered for preparation of aerogel foams.

The emulsion stability was also investigated visually by monitoring the fraction of the original emulsion taken in a vial that remained in emulsified state in a period of 35 min.

This is represented by the ratio of height (h) of the emulsion layer and the total height (h0) of the vial. Such data show a definitive trend for SPAN 80® concentrations of 1.5, 7.5 and

15 vol%, as seen in Figure 5.5. The settling rate, inferred from the slope of h/h0 vs. time curves, was the lowest for emulsions with surfactant concentrations of 15 vol%; in this case smaller droplets settled much more slowly. In contrast, the emulsion with surfactant concentrations of 1.5 vol% showed much higher settling rate, with up to 40% of the liquid from the original emulsion settled within the first minute after stirring was discontinued.

The present work benefited from emulsions that settled slowly, thus enabling an even distribution of water droplets in the emulsion volume. In view of this, emulsions of 1.5 vol% SPAN 80® concentration were deemed unsatisfactory for further study.

129

1.5 vol% SPAN 80 1.0 7.5 vol% SPAN 80 15 vol% SPAN 80 0.9

0.8 0

0.7 h/h

0.6

0.5

0.4

0 5 10 15 20 25 30 35 Time (mins)

Figure 5.5: Volume fraction of emulsion phase for SPAN 80® concentrations of 1.5, 7.5, and 15 vol% with time after stirring was stopped.

Effect of Surfactant Type. Up to this point, the data produced with SPAN 80® have been discussed. In a number of cases, emulsions were prepared with F127® as the surfactant.

The presence of F127®, a PEO-PPO-PEO block copolymer surfactant, produced much higher reduction of interfacial tension at 60 C compared to SPAN 80®, e.g., 0.6 mN/m for F127® vs. 3.5 mN/m for SPAN 80® for the same 1.5 vol% surfactant. The average water droplet diameter with 1.5 vol% F127® was 2.5 ± 1.6 µm. The effect F127® on water droplet size distribution will be discussed later.

130

5.3.5 Morphology of Aerogel Foam

The fabrication method adopted in this work yielded aerogel foam structures with typical macrovoid diameters of ~20 µm embedded in the inherently produced aerogel structure of the host sPS polymer. These macrovoids were occupied by the dispersed water droplets in

W/O emulsion and locked in the sPS gel structure. As is apparent from a representative

SEM image presented in Figure 5.6, the macrovoids were interconnected. One representative macrovoid and one interconnect are labelled in Figure 5.6a for clarity. The materials between adjoining macrovoids were due to macroporous sPS aerogel, as shown in Figure 5.6b. The macroporous sPS, the macrovoids, and the macrovoid interconnects in conjunction produced the open cell aerogel foam structures.

The W/O interfaces in the emulsion turned into porous skin layers after sPS gelation. The

SEM image presented in Figure 5.6b shows that the skin layer had much higher density than the bulk material. The macroporous nature of sPS domains between two adjoining macrovoids distinguishes the aerogel foams from the open cell conventional foam structures reported in literature.261 The skin materials in traditional polymer foams do not show inherent, volumetric porous networks, as exhibited in Figure 5.6b. The aerogel foams reported in this work consistently exhibit porosity values greater than 95%, e.g., a dispersed phase content of 50 vol% in the templating emulsion possessed the highest overall porosity of around 97.5% (sPS_50_7.5_1600), as listed in Table 5.4.

131

Figure 5.6: SEM images of aerogel foam with (a) macrovoids and interconnects and (b) macroporous skin layer.

Table 5.4: Porosity, BET surface area, and macrovoid size distribution of emulsion- templated aerogel foam specimens. BET Macrovoid Size Distribution (%) Macrovoid Porosity surface Specimen Mean size Between 20 (%) area < 20 µm > 50 µm (µm) and 50 µm (m2/g) sPS_0_0_1600 92.5 ± 0.2 267 ± 6 - - - - sPS_33_1.5_1600 95.4 ± 0.4 247 ± 8 61.5 ± 61.6 36 24 40 sPS_33_7.5_1600 96.0 ± 0.1 222 ± 2 14.3 ± 10.8 77 22 1 sPS_33_15_1600 95.5 ± 0.3 208 ± 9 11.9 ± 8.8 85 15 0 sPS_33_7.5_800 95.5 ± 0.1 225 ± 4 22.6 ± 26.4 68 17 15 sPS_33_7.5_1200 95.6 ± 0.4 243 ± 6 20.2 ± 17.5 68 24 8 sPS_25_7.5_1600 95.5 ± 0.2 262 ± 4 12.3 ± 8.7 86 13 1 sPS_50_7.5_1600 97.5 ± 0.1 204 ± 6 17.8 ± 11.5 66 33 1

5.3.6 Macrovoid Size Distribution

The emulsion-templated aerogel foams showed significant degree of tunability of macrovoid size via changes in processing parameters, such as dispersed phase content, mixing speed, and surfactant concentration.

Effect of Surfactant Concentration. Macrovoid size showed the strongest dependence on surfactant concentration. For example, the macrovoid mean size of the specimens sPS_33_1.5_1600, sPS_33_7.5_1600, and sPS_33_15_1600 were ~61.5, 14.3, and 11.9

132

µm respectively for surfactant concentrations of 1.5, 7.5, and 15 vol% (Table 5.4). The macrovoid size distribution became narrower with an increase of the surfactant concentration, as shown in Figure 5.7. For some specimens, e.g., sPS_33_1.5_1600, the standard deviation of the macrovoid diameter distribution was greater than the mean values, indicating a large spread. In view of this, the macrovoid size distribution data was split into three fractions before analysis, e.g., those with diameter less than 20 µm, between

20 and 50 µm, and greater than 50 µm, as listed in Table 5.4.

Figure 5.7: SEM images of aerogel foams fabricated from emulsions with SPAN 80® concentrations of (a) 1.5 vol%, (b) 7.5 vol%, and (c) 15 vol%. Images in (d), (e), and (f) represent higher magnification. The initial emulsion droplet size and corresponding macrovoid size of the aerogel foams are shown for various surfactant concentrations, (g) 1.5 vol%, (h) 7.5 vol%, and (i) 15 vol%.

133

We now examine the trend of macrovoid size distribution. First, the SEM images in Figure

5.7a-c indicate that the macrovoid size reduced with an increase of surfactant concentration, similar to what was found for water droplet size distribution discussed using the data presented in Table 5.3 and Figure 5.4. Second, the higher magnification SEM images in Figure 5.7d-f reveal that the population density of macrovoid interconnects increased with an increase of surfactant concentration.

The water droplet size distributions of the starting emulsions are compared with the resultant macrovoids in Figure 5.7g-i. Two distinct observations can be made. First, the macrovoid size is larger than that of dispersed water droplets. The average water droplet size for emulsion systems with 1.5, 7.5, and 15 vol% SPAN 80® concentration was 20.6,

13.0, and 12.1 µm respectively (Table 5.3). The corresponding average macrovoid sizes are 61.5, 14.3, and 11.9 µm (Table 5.4). Second, the gap in starting emulsion water droplet size and macrovoid size reduced with an increase of surfactant concentration.

At lower surfactant concentration and for higher interfacial tension values, the emulsion system is more prone to coalescence events, as discussed earlier. The time for sPS gelation, typically 3 minutes, was much longer than about 1 minute to observe appreciable coalescence and settling in these cases (Figure 5.5) and as a result, coalescence events were more abundant. In contrast, at higher surfactant concentrations, the emulsions were much more stable and resistant to coalescence events. Thus, in these cases, the resultant macrovoid size distributions closely mirrored the starting emulsion droplet size distribution shown in Figure 5.7i.

It is to be noted that for polyHIPE formation, fabrication of micrometer sized macrovoids were typically achieved with high SPAN 80® surfactant concentrations of 20 vol%.262–264

134

There was one reported case of polyHIPEs formed with surfactant concentration of 5 vol%, but these resulted in macrovoids of 500 µm to 8mm, orders of magnitude larger than macrovoids achieved in this study.265

Effect of Mixing Speed. The data listed in Table 5.4 showed that the fraction of macrovoids with diameter greater than 50 µm reduced with an increase of the mixing speed. For example, for specimens sPS_33_7.5_800, sPS_33_7.5_1200, and sPS_33_7.5_1600, the fractions with diameter greater than 50 µm were respectively 14%, 8%, and 1% for stirring speeds 800, 1200, and 1600 rpm. The fractions in diameter range 20-50 µm increased from

17% to 24% with an increase of stirring speed from 800 to 1200 rpm, while the fractions with diameter smaller than 20 µm increased from 68% to 77% with an increase of mixing speed from 1200 to 1600 rpm. This indicated that at higher mixing speeds, a greater proportion of smaller macrovoids were formed, as shown in Figure 5.8a. This trend can be explained as follows. At higher rotational speed of the magnetic stir bar, effective shear rate was higher. As a consequence, the shear stress of the continuous phase outweighed the restoring interfacial stresses on water droplets. Accordingly, larger water droplets underwent break up into smaller ones; these in turn produced smaller size macrovoids in the resultant aerogel foam. However, one cannot increase the stirring speed indefinitely as associated shear heating may result in boiling of the liquids and promote liquid to solid phase transition of the emulsion system.

Effect of Dispersed Phase Content. The macrovoid size distribution was shown to be weakly influenced by the dispersed phase content, as shown in Table 5.4 and Figure 5.8b.

For example, the specimen sPS_25_7.5_1600 with 25% water phase and sPS_50_7.5_1600 with 50% water phase showed close mean macrovoid diameter values of 12.3 ± 8.7 µm

135 and 17.8 ± 11.5 µm respectively. This is an expected trend as there was no change in either the shear stress or the interfacial stress with the increase of water content in the system. A small increase in mean macrovoid diameter can be attributed to higher frequency of coalescence events in the presence of greater fractions of water.

Figure 5.8: Effect of (a) mixing speed and (b) dispersed phase content on macrovoid size distribution. Surfactant concentration was kept constant at 7.5 vol%.

The polydisperse nature of size distributions of droplet and macrovoid presented in Figure

5.1, 5.2, 5.4, 5.7 and 5.8 are due to a dynamic equilibrium of breakup and coalescence events in emulsion preparation process achieved via mechanical agitation. One can, however, obtain monodisperse macrovoids from other emulsion-templating methods, e.g., via using microfluidics, as demonstrated by Quell et al.227,266

5.3.7 Open Cell Structure

It was alluded to earlier that the macrovoids, macroporous sPS domains, and the macrovoid interconnects are responsible for producing an open cell structure. The degree of openness

136 can be inferred from the pore interconnect density and from the total projected surface area of the pore interconnects, as apparent in SEM images. The SEM images presented in Figure

5.9a-c show that the degree of openness of the cellular structures increased due to an increase of the surfactant concentration. An increase in the volume fraction of the dispersed phase also resulted in a more open cell structure, although, as inferred earlier in reference to Figure 5.8b, the macrovoid size was only weakly dependent on the dispersed phase content.

Figure 5.9: SEM images showing pore interconnects density for SPAN 80® concentrations of (a) 1.5%, (b) 7.5%, and (c) 15%. (d) Macrovoid interconnect exhibiting frustum geometry and (e) proposed interconnect formation mechanism.

The pore interconnect structures seen in Figure 5.6a and 5.9a-c can be interpreted as follows. We invoke similarity with pore interconnect formation observed in polyHIPEs. In polyHIPEs, predominantly open cell structures were reported, although the polymer skins, unlike in this work, remained non-porous. The interconnects in polyHIPEs originate either from the shrinkage of the polymer skin due to density difference between the polymer and the precursor monomer or from the rupture of the thin polymer films during the dispersed

137 phase extraction process.267 However, these two mechanisms cannot adequately explain pore interconnect formation in aerogel foam systems described in this paper. First, a pre- made polymer was used in this work. Second, the customary volumetric shrinkage of the monolithic sPS gels during the supercritical drying step remained small, ~4%. The polymer aerogel monoliths did not show fracture at such low levels of shrinkage during the drying process.

A closer examination of interconnects in Figure 5.9 also revealed no angular tears caused by stress-cracking. Instead, interconnects were found sandwiched between two larger macrovoids with a circular cross-sectional shape exhibiting a spherical frustum geometry, as shown in Figure 5.9d. The unique geometry of interconnects highlighted in Figure 5.9, and not found in PolyHIPEs, leads us to suggest that a different mechanism was at play for their genesis in the current study. It is apparent that interconnects are of much smaller in size than the macrovoids. It is also apparent that sPS strands wrapped around the interconnects, creating close to spherical hole patterns. Such regular structures quite possibly developed during gelation of sPS and we believe were due to phase separated droplets of water formed upon cooling of the emulsion to room temperature. This is supported by the data from Jou and Mather268 who indicated that mass fraction of dissolved water in toluene varies from 0.02 at 100 °C to 0.001 at 15 °C.

Based on the above observation, we present that interconnects reported in Figure 5.9 were kinetically controlled. Recall that the emulsion templated sol system has four components

- sPS, toluene, surfactant, and water. In view of data of Jou and Mather,268 water had higher solubility in toluene at emulsion preparation temperature of 100 C than at room temperature. As the temperature was reduced prior to gel formation, a polymer-rich phase

138 was formed via liquid-liquid demixing. In conjunction, water dissolved in toluene also lost its solubility and formed new water droplets inside the organic liquid. The increasing viscosity of the gel with time and the soluble surfactants in the organic phase stabilized the newly phase-separated water droplets, as schematically presented in Figure 5e. The polymer-phase vitrified via thermo-reversible gelation and surrounded the newly formed water droplets in the midst of coalescence, resulting in the spherical frustum structure as seen in Figure 5.9d and schematically presented in Figure 5.9e.

5.3.8 Macroporous polymer skin structure

The concentration of sPS in the continuous phase is another parameter that can be used to tune the open structures in aerogel foams. The interstitial space between the polymer strands are considered macroporous, with an average pore size of 200 nm. An increase in polymer concentration increases the volume fraction of the polymer-rich region during spinodal decomposition, and correspondingly, increases the diameter of the polymer strands. This, in turn, reduces the pore size. Such an effect is evident from the SEM images in Figure 5.10, where sPS concentration in the continuous phase was increased from 0.02 to 0.08 g/mL. Kim et al.64 explored the effect of sPS concentration on permeability of sPS aerogel monoliths and showed that at low polymer concentrations of 0.01 g/mL, air permeability was 9.74 x 10-10 m2, which reduced to 5.29 x 10-10 m2 at polymer concentration of 0.08 g/mL. This reduction of permeability is attributed to thicker polymer strands, smaller pore size, and lower pore volume fraction.

139

Figure 5.10: SEM images of macroporous matrix of varying polymer concentration: a) 0.02 g/mL, b) 0.04 g/mL, c) 0.06 g/mL and d) 0.08 g/mL.

We now revisit the issue of skin layers. As briefly mentioned earlier, skin layers formed at two different locations in the aerogel foam structure- first, at the interfaces between the sPS sol and water droplets and second, at the surfaces of the container of the mold. These skin layers are regions where sPS strands were aggregated together, sealing off the macropores. In both cases, heterogeneous surfaces were involved. Gu and Jana16 also observed dense skin layers in polyurea aerogel foams attributed to heterogeneous nucleation of polyurea domains on PEO surfaces. In this case, the skin layer thickness was measured to be 62 nm, which is the same order of magnitude of the diameter of one, typical sPS strand (50 nm).

140

The BET surface area values presented in Table 5.4 are directly affected by skin layer formation. For example, monolithic aerogel specimen sPS_0_0_1600 produced with no emulsion-templating shows a BET surface area of 267 m2/g compared to 204 m2/g for an emulsion-templated aerogel specimen sPS_50_7.5_1600 with 50% dispersed water phase.

Such a reduction in BET surface area with the addition of dispersed water is indicative of the formation of denser skin layers that are not readily accessible for adsorption of nitrogen gas. This also suggests that in the skin layers, sPS strands underwent higher degree of aggregation within a given volume compared to the bulk.

The above scenario was further examined by controlling the interfacial tension between the two phases. In the first study, different types of surfactants were used in emulsion formation. Surfactants which causes largest reduction of interfacial energy of the water- toluene interface also should result in thinnest possible skin layers at the macrovoid boundaries yielding an open porous structure very much akin to the bulk phase. This was verified using aerogel foam specimens prepared with SPAN® 80 and F127®. Note that

F127® is a PEO-PPO-PEO triblock copolymer with average Mw of 12,600 g/mol. For a

1.5 vol% surfactant concentration, SPAN® 80 provided higher interfacial tension value of

3.5 mN/m compared to 0.6 mN/m for F127®, both at 60oC. The F127® triblock copolymer lowers the interfacial tension more than SPAN 80® due to its increased solubility of the

PEO and PPO blocks in the aqueous and organic phases of the emulsion respectively. For example, SPAN 80® is insoluble in water at room temperature, while the measured cloud point of F127® in water was measured at 102 °C.

Specimens prepared with SPAN® 80 exhibited a denser skin layer compared to the skin layers of macrovoids prepared with F127® surfactants, as shown in Figure 5.11a and 5.11b

141 respectively. Such an observation presents an interesting scenario for future work to delineate the relationship between interfacial tension and skin layer density.

Figure 5.11: Skin layer formation with (a) SPAN 80 as surfactant, (b) F127 as surfactant and c) The initial emulsion droplet size and corresponding macrovoid size of the aerogel foams with F127 surfactant concentration of 1.5 vol%.

The greater reduction of interfacial tension due to F127® surfactant also shifted the macrovoid diameter distribution to smaller sizes, due to the smaller dispersed phase droplets formed in the emulsion, as shown in Figure 5.11c. The macrovoids obtained with

F127® as the surfactant had diameters in the narrow range of 0-15 µm.

A second study was conducted to investigate the effect of surface energy of the glass mold surface on the properties of the skin layer formed. Two glass slides with different surface energies were prepared as substrates for gel formation. The first glass slide was plasma treated to produce a hydrophilic surface (water contact angle of 18.2O ± 2O), while a second glass slide was coated with chlorophenylsilane to produce a more hydrophobic surface

(water contact angle of 67.7 O ± 1.1O) that has high compatibility with both sPS and toluene solvent. SEM images in Figure 5.12 show the skin layers formed during gel formation in direct contact with the glass substrate. It can be seen that aerogels formed on the hydrophilic surface (Figure 5.12a) show a denser skin layer compared to gels formed on the hydrophobic surface (Figure 5.12b). This reinforces the hypothesis that sPS 142 preferentially nucleates at high energy interfaces, forming a denser skin layer. This also shows that both internal and external skin layer morphology can be manipulated through the interfacial energy between the two phases of the emulsion and substrate respectively.

Figure 5.12: Skin layer formation of monolithic sPS aerogel on a) hydrophilic substrate and b) hydrophobic substrate.

The two phenomena discussed previously, i.e., pore interconnect density and skin layer formation, are good avenues to control the accessibility of the structures to fluid flow. For example, to create a hierarchical structure impervious to fluid flow, one can use a system with low surfactant concentration, which in turn can reduce the number of pore interconnects and the surface area, as well as create a denser skin layer with smaller pore size. In contrast, a reduction of interfacial energy in the system results in a more open and interconnected network of macrovoids with less dense skin layers at the aerogel surfaces and macrovoid walls. In addition, polymer concentration can also be another design factor to increase or decrease the permeability of the macroporous networks.

143

5.4 Conclusions

The emulsion-templating method was successfully implemented to prepare open cell aerogel foams. These structures show significant flexibility to tune the macrovoid size, macroporous nature of the skin, and the skin layer structure and morphology found at the interfaces between the voids and the polymer. This flexibility opens up this structure for a variety of applications. For example, these aerogels, or their corresponding gels, can be used as a membrane for gas and liquid filtration, where selectivity can be controlled through both the skin layer structure and the density of the macroporous matrix.

Permeability of these membranes can also be improved through the inclusion of the macrovoids and the openness of the foam cell structure. In another possible application, these aerogels can potentially be used to attenuate elastic waves. Their low density, high porosity, and critical ability to tailor macrovoid dimensions allow these hierarchical aerogels to be a potential candidate for acoustic and blast wave attenuation applications.

144

5.5 Appendix

In addition to lower interfacial tension values, the increased emulsion stability at higher surfactant concentration was inferred from the data on interfacial rheological parameters as presented in Figure 5.13. The interfacial elasticity and viscosity values were obtained from the pendant drop method. For these measurements, a water droplet (dispersed phase) was dispensed in toluene solution with the respective surfactant concentration (continuous phase). Subsequently, the droplet shape was subjected to oscillation at a frequency of 0.1

Hz.269 As the drop shape (and hence the interfacial area) changed, the associated change in interfacial tension was recorded using the axisymmetric drop shape analysis (ADSA) method. The amplitude of the area, interfacial tension oscillations, and the phase shift due to the perturbation were extracted from the signals and the interfacial elasticity and viscosity were calculated. Both the control of the oscillations in drop shape and interfacial rheology measurements were conducted by the DROPimage Advanced software.270 All measurements were conducted in triplicate. Such data established that the interfaces were stable to perturbations and fluctuations thus deterring coalescence events.

145

Figure 5.13: Interfacial elasticity (E’) and viscosity (E”) values of water in toluene emulsions in the presence of SPAN 80®. The error bars are smaller than the size of the symbols.

The data in Figure 5.13 show a reduction of the value of interfacial elasticity (E’) and interfacial viscosity (as inferred from E”) with an increase of the surfactant concentration during low frequency (0.1 Hz) deformations of the interfacial area. The interfacial viscosity, inferred from the value of E”, did not show as much reduction as the interfacial elasticity. At high bulk concentrations, the surfactants undergo rapid adsorption (during extension) and desorption (during compression) onto the interfaces, thus immediately smoothing out any interfacial tension gradients experienced by the deformed interfaces.271,272 This enables the interface to withstand deformations without rupturing, for example, during potential coalescence events. Consequently, the overall frequency of successful coalescence events is reduced at high surfactant concentrations and stable emulsions with smaller droplets result.

However, the interfacial rheological parameters discussed above was based on two key assumptions. First, it was assumed that the trend of decreasing interfacial elasticity with

146 increasing surfactant concentration at low frequency would invert at higher frequencies.

This was observed in the case of a study conducted by Stubenrauch and Miller.272 However, without access to high frequency tensiometers, a single frequency measurement can show only one part of the trend. The second assumption made was that the frequency relevant for film stability was considered to be low, and in the region of 0.1 Hz. This, however, may not be the case, as highlighted by Santini and Stubenrauch.271,272 Due to these two assumptions, this data was not included in the original study.

147

CHAPTER 6

POLYIMIDE-BASD AEROGEL FOAMS VIA EMULSION-TEMPLATING

Reproduced from: [Teo, N.; Gu, Z.; Jana, S.C.; Polyimide-Based Aerogel Foams via Emulsion-Templating. Polymer 2018.157.95-102. DOI: 10.1016/j.polymer.2018.10.030]

Abstract

An oil-in-oil emulsion-templating method is used to fabricate polyimide aerogel foam materials. These materials contain micrometer size voids (macrovoids) in conjunction with inherently produced meso- and macropores in polyimide gels. Polyamic acid is first synthesized from diamines and dianhydrides and then chemically imidized to obtain a sol.

An immiscible oil-type dispersed phase is introduced in the sol via emulsification and the sol is subsequently allowed to transition into a gel, thereby locking the dispersed phase droplets within the structure. The gel is subsequently dried under supercritical conditions to obtain aerogel foams. This paper evaluates the stability of the oil-in-oil emulsion used for templating with reference to the gel times of the continuous phase. Specifically, the effects of surfactant concentration on macrovoid size, mesopore size, and mechanical properties of the aerogel foams are investigated. In addition, water and oil absorption behavior of the aerogel foams are studied.

148

6.1 Introduction

Polyimides were first reported in 1908, although its usage increased after the late 1950s once high molecular weight polyimides were successfully prepared.197 These materials possess excellent properties such as high thermal, hydrolytic, and radiation stability, as well as good mechanical strength and electrical properties at elevated temperatures.273,25

These properties, in conjunction with its gel forming capability, have qualified polyimides as an appropriate candidate for aerogel fabrication. Polyimide aerogels and their hybrids have been reported in literature.35 For example, the network structure of silica aerogels were coated with a layer of polyimide to impart structural integrity and improve load bearing capabilities.274 The step-growth nature of polyimide synthesis also accommodates a variety of monomers for tailoring the mechanical properties, shrinkage, and surface energy of polyimide aerogels.33,36 Traditionally, polyimides are fabricated using the

Dupont two-step process. First, polyamic acids are obtained by reacting dianhydrides and diamines. The second step involves chemical imidization using acetic anhydride as a dehydrating agent and pyridine as a catalyst, whereby the amic acid functional group is converted to the imide functional group.26 Other synthesis routes for polyimide aerogels have also been reported, such as ring opening polymerization27 and substituting isocyanates for amines.203

Aerogels are frequently fabricated in a cylindrical monolith shape, although recent studies have reported both flexible films and microparticle forms.206,216 The high thermal stability and the close proximity of aerogel pore sizes to the mean free path of air molecules impart excellent thermal insulation properties in polyimide aerogels. This has seen polyimide aerogels deployed as insulation for entry/re-entry vehicles and extravehicular activity

149 suits.34 The inherent low dielectric properties of polyimide aerogels also make them suitable for fabrication of antenna42,43 and membrane separators for batteries.37 In addition, the open pore structures of polyimide aerogels, with both meso- (diameter 2-50 nm) and macropores (diameter >50 nm), have shown to be useful in filtering airborne nanoparticles.

These aerogels have been shown to achieve a high separation efficiency of 99.99%, while maintaining high air permeability values ~1 x 10-10 m2.180

The work of Zhai and Jana180 established strong mesopore fraction vs. filtration efficiency and macropore fraction vs. air permeability relationships. This study’s takeaway was that high air permeability and high filtration efficiency require high fractions of macropores and mesopores respectively. However, aerogel monoliths prepared by the methods reported in literature does not allow independent control of both meso- and macropore fractions. This necessitated another method of incorporating macropore fractions in aerogel materials without altering the mesopore content, bearing in mind that the final structures should be one-component, free-standing hierarchical structures. This served as the motivation for the development of aerogel foams, fabricated via inclusion of micrometer sized voids, defined as macrovoids, into inherently porous aerogel structures.

The macrovoids allow quick transfer of fluids through the aerogel structures, thus allowing higher fluid permeability.

Aerogel foams have previously been fabricated using a number of methods. Gu and Jana208 had used a solid template of co-continuous polymer systems to synthesize polyurea aerogel foams. Wang and Jana209 used a solution of polyethylene oxide to obtain micrometer size voids in syndiotactic polystyrene (sPS) aerogels. Recently, Teo and Jana199 successfully used a water-in-oil emulsion-templating method to introduce spherical, ~20 µm sized

150 macrovoids in macroporous sPS aerogel structures. In this work,199 sPS and a nonionic surfactant were first dissolved in toluene at high temperature and the solution was emulsified using deionized water. The sPS solution turned into a gel when cooled to room temperature, thus locking in the emulsified water droplets within the vitrified structure.

The gel was supercritically dried to yield sPS aerogel foam materials. However, this water- in-oil emulsion templating method cannot be extended to water-sensitive monomers such as those encountered in polyimide systems. On the other hand, aerogel foams produced from high temperature polyimides has the potential for expanded applications such as in hot air filtration. In view of this, the present work details an oil-in-oil (O/O) emulsion- templating method for incorporation of significant fractions of macrovoids in polyimide aerogel structures. This yields polyimide aerogel foams with macrovoids (~30 µm) seen in foams, as well as mesopores found inherently in polyimide aerogels.

6.2 Experimental Section

6.2.1 Materials

Pyromellitic dianhydride (PMDA) was purchased from Alfa-Aesar (Haverhill, MA) and

2,2’-dimethylbenzidine (DMBZ) was purchased from Shanghai Worldyang Chemical Co.

Ltd (Shanghai, China). Tris(2-aminoethyl)amine (TREN) and surfactant F127®

(trademark of BASF) were purchased from Sigma Aldrich (Milwaukee, WI). Pyridine, acetic anhydride, cyclohexane, and acetone were purchased from Fisher Scientific

(Ontario, NY). N,N-dimethylformamide (DMF) was purchased from VWR International

(Radnor, PA) and n-heptane was purchased from EMD Millipore (Billerica, MA).

151

6.2.2 Fabrication of Aerogel Foams

An O/O emulsion-templating method was used in this work to obtain micrometer size oil droplets dispersed in the sol. The sol transitioned into a gel, thus locking the emulsified oil droplets within the gel structures. This process resembles the fabrication of polyHIPEs,73,256,257 with the exception that the continuous oil-phase in this study was initially an organic sol, which later turned into a gel with its inherent pores filled by the organic solvent.

All materials were stored in desiccators and used as purchased, with the exception of

F127® surfactant, which was dried in a vacuum oven for at least 30 minutes at a temperature of 35 C prior to dissolution to mitigate the effect of absorbed moisture. The continuous oil phase was first prepared by dissolving PMDA and F127® surfactant in DMF over a period of at least 30 minutes, to ensure full dissolution of the surfactant. DMBZ was dissolved in DMF, added to the PMDA and surfactant solution, and magnetically stirred for 2 minutes at 1600 rpm to form the polyamic acid solution. TREN, acetic anhydride, and pyridine were added at the same time together to the polyamic acid solution to serve respectively as the crosslinker, dehydrating agent, and catalyst. These three reagents had to be added all at once as the fast reaction kinetics of the crosslinker (TREN) would lead to immediate precipitation of the crosslinked materials if added separately. The reaction was conducted at room temperature of 19 °C in a sealed glass vial under air, with the reaction scheme shown in Figure 6.1. The final reaction mixture was stirred magnetically for 3 minutes followed by addition of the dispersed phase (cyclohexane or n-heptane) and stirred for an additional 5 minutes. This allowed for the viscosity of the solution to increase with conversion, imparting increased stability to the O/O emulsion. For neat polyimide

152 samples without any surfactant or dispersed phase, this step involved further stirring of only 1.5 minutes, to allow transfer to the molds before gelation. A typical sample with 30 vol% dispersed phase and 0.5 vol% surfactant based on the continuous phase was prepared from 0.114 g PMDA, 0.017 g F127®, 0.106 g DMBZ, 0.015 g TREN, 0.360 g acetic anhydride, 0.307 g pyridine, 2.5 mL DMF, and 1.67 mL dispersed phase (cyclohexane or n-heptane). The above recipe produced a polymer concentration of 7.4 wt% in the solution.

Figure 6.1: Reaction scheme for synthesis of polyimide.

153

The final mixture was poured into cylindrical molds and allowed to gel. The selection of monomers, crosslinker content, and overall polymer concentration in the sol allowed fast gelation and vitrification of the emulsion while facilitating pouring of the sol in the mold.

The gels were aged in the molds for 48 hours before demolding. The presence of surfactant delayed gelation of the system. Accordingly, the gels were left in the molds for 48 hrs to ensure completion of crosslinking and imidization reactions. The demolded gels were solvent exchanged sequentially with 25 vol% acetone / 75 vol% DMF, 50 vol% acetone /

50 vol% DMF, 25 vol% acetone / 75 vol% DMF, and finally with 100 vol% acetone at 12- hour intervals. In addition, the gels were further washed with 100 vol% acetone for an additional 5 times at 12-hour intervals to remove as much of the DMF as possible. The gels were subsequently solvent exchanged with liquid carbon dioxide in an autoclave by washing with 100 vol% liquid carbon dioxide for 6 times at 1.5 hours intervals. The liquid carbon dioxide infused gels were subsequently dried under supercritical condition of carbon dioxide at 50 C and 11 MPa. The surfactant concentration was varied between 0.5 vol% and 5 vol% to impart stability of the O/O emulsion against creaming and coalescence before gelation of the continuous phase.

As reference materials, polyimide aerogels were also synthesized with surfactant without dispersed phase, and without surfactant and without dispersed phase (neat PI).

6.2.3 Characterization of Oil-in-Oil (O/O) Emulsions

Interfacial Tension Measurement. The interfacial tension between the dispersed

(cyclohexane or n-heptane) and the continuous (DMF) phases was measured using a Du

Noy tensiometer (Interfacial Tensiometer 70545, Central Scientific Co., VA). First, the

154 surfactant was dissolved in DMF at concentrations of 0.5, 2.5, and 5.0 vol%. Next, 20 mL of the dispersed phase was gently added to the top of the surfactant solution in DMF by pouring the liquid over a clean glass slide to prevent emulsification and droplet formation during pouring. Interfacial tension was then recorded in triplicate by placing the Du Noy ring at the interface of these two phases.

Emulsion Droplet Size. The droplet size distribution of O/O emulsions (prepared by mixing surfactant, DMF, and cyclohexane/n-heptane) was studied using Olympus BX51 optical microscope (Center Valley, PA). For this purpose, a drop of emulsion was placed on a microscope slide with a depression and its optical images were recorded. The images were analyzed using ImageJ software to yield droplet size distributions from the diameter data of more than 100 droplets.

6.2.4 Characterization of Aerogel Foams

Aerogel Foam Morphology and Macrovoid Size Distribution. The morphology of aerogel foams was obtained using a scanning electron microscope (JSM5310, JEOL, MA). An accelerating voltage of 5 kV and emission current of 20 mA was used to capture the SEM images. For this purpose, a representative piece of fractured aerogel foam specimen was mounted on an aluminum stub using carbon tape, followed by sputter coating with silver

(ISI-5400 Sputter Coater, Polaron, UK). The macrovoid size distribution was obtained from the analysis of SEM images using ImageJ software. For each specimen, more than

100 macrovoids were considered for determining the macrovoid size distribution. The size of polyimide strands in SEM images was measured using ImageJ software.

155

IR: Infrared spectra was recorded on a Nicolet iS50 FTIR tri-detector spectrophotometer

(Thermo Scientific, MA).

TGA: Thermogravimetric analysis was conducted under N2 with a Q50 thermogravimetric analyzer (TA Instruments, DE) using a heating rate of 20 C/min, up to 800 C.

Porosity and Pore Volume: Porosity was calculated from the values of skeletal (ρs) and bulk density (ρb) using equation (1). The values of skeletal density were obtained using a helium pycnometer (AccuPyc II 1340, Micromeritics Instrument Corp., GA). Bulk density of the aerogel foams was calculated from the values of mass and volume of the foam specimens.

𝜌 푝표푟표푠푖푡푦 = (1 − 푏) × 100% (1) 𝜌푠

Total pore volume (Vtot) was calculated from the bulk and skeletal density according to equation (2):

1 1 푉푡표푡 = − (2) 𝜌푏 𝜌푠

Shrinkage: The diameter (d0) of cylindrical plastic molds for polyimide gels was 13 mm.

The diameter shrinkage was calculated from the values of diameter (d) of dried aerogel and d0.

BET surface area: Brunauer-Emmett-Teller (BET) surface area of the aerogels and aerogel foams were obtained from N2 adsorption-desorption isotherms at 77 K using a

Micromeritics Tristar II 3020 analyzer (Micromeritics Instrument Corp. GA).

Gel times and Viscosity: The gel time of the polyimide solution at room temperature was obtained from the crossover point of the storage (G’) and loss (G”) moduli measured using an ARES G2 Rheometer (TA Instruments, New Castle, DE). For this purpose, the final reaction mixture was poured into a solvent trap and loaded into the rheometer fitted with a

156

50 mm cone and plate set up. The rheometer was operated at a constant angular frequency of 1 rad/s at 10 % strain.

Compressive Modulus: Compressive modulus of aerogel and aerogel foam specimens were obtained as per ASTM D695-15 method from compressive tests using an Instron 5567 tensometer (Norwood, MA) using a 1 kN load cell and a strain rate of 1.3 mm/min. Aerogel and aerogel foam specimens were molded into cylindrical shapes with height to diameter ratio of 2:1. The final aerogel specimens were grinded to ensure smooth and parallel surfaces. The compressive modulus value was obtained from the slope of stress vs. strain curves at low strain, typically ~0.01 mm/mm.

Contact Angle: The values of contact angle of deionized water were measured using a

Rame-hart Model 590 Advanced Automated Goniometer/Tensiometer using the

DROPimage Advanced software (Succasunna, NJ). For this purpose, aerogel specimens were first compressed to remove all the pores before placing water drops on them.

6.3 Results and Discussion

6.3.1 Emulsion Formation

The O/O emulsions prepared from the surfactant and the continuous and dispersed phase liquids were examined to obtain a first approximation of the dispersed phase droplet size distribution in such systems. This was done without any polyimide precursor monomers in the system. Both DMF/cyclohexane and DMF/n-heptane systems exhibited intrinsically low interfacial tension values, respectively at 3.6 and 4.0 mN/m (Table 6.1). The interfacial tension values reduced to 2.0 and 3.5 mN/m respectively for DMF/cyclohexane and

DMF/n-heptane systems with the addition of F127® surfactant. The surfactant

157 concentration above 0.01 vol% had no additional effect on interfacial tension, indicating that surfactant concentrations 0.5, 2.5, and 5.0 vol% were above the critical micelle concentration (CMC). This is in line with other CMC studies of Pluronic® block copolymers in aqueous systems.275

Table 6.1: Interfacial tension of DMF/cyclohexane/F127® and DMF/n-heptane/F127® systems. Continuous Dispersed Surfactant concentration Interfacial tension phase phase (vol% of DMF) (mN/m) 0 3.6 ± 0.1 0.01 2.0 ± 0.0 cyclohexane 0.5 2.0 ± 0.0 2.5 2.0 ± 0.1 DMF 5.0 2.0 ± 0.1 0 4.0 ± 0.0 0.01 3.5 ± 0.0 n-heptane 0.5 3.5 ± 0.1 2.5 3.4 ± 0.1 5.0 3.4 ± 0.1

A continuous phase of the emulsion system was also prepared by mixing DMF, acetic anhydride, pyridine, and F127 ® in the same proportion as were used later in polyimide gel fabrication. The droplet size distributions for such a system are presented in Figure 6.2 as function of surfactant concentration. The values of maximum, minimum, and mean droplet size are listed in Table 6.2.

158

Figure 6.2: Droplet size distribution with (a-c) cyclohexane and (d-e) n-heptane as the dispersed phase. Surfactant concentration are (a,d) 0.5 vol%, (b,e) 2.5 vol% and (c,f) 5.0 vol%.

Table 6.2: Mean, maximum, and minimum droplet size in emulsions. Surfactant Maximum Minimum Dispersed Mean droplet concentration droplet droplet phase diameter (µm) (vol%) diameter (µm) diameter (µm) 0.5 61.7 ± 34.5 191.9 18.6 2.5 cyclohexane 50.2 ± 33.5 179.2 8.3 5 33.5 ± 13.0 77.1 9.1 0.5 81.6 ± 67.3 267.6 14.5 2.5 n-heptane 58.0 ± 24.9 135.9 14.7 5 34.7 ± 27.7 141.2 5.8

Several trends became apparent from the data presented in Figure 6.2. The dispersed phase droplets were generally smaller in the case of cyclohexane. It is noted from the data in

Table 6.2 that the maximum droplet diameter for cyclohexane with 0.5 and 2.5 vol%

159 surfactant concentration were 191.9 and 77.1 µm respectively, while the corresponding maximum droplet diameter for the n-heptane system was 267.6 and 141.2 µm respectively.

Such a difference in size can be attributed to lower values of interfacial tension in the case of DMF/cyclohexane system, as listed in Table 6.1. It is intuitive that higher surfactant concentration also led to stabilization of larger interfacial area between the polar and the non-polar phases, resulting in smaller dispersed phase droplets. The mean droplet diameter in the case of DMF/cyclohexane system were 61.7 ± 34.5 µm, 50.2 ± 33.5 µm, and 33.5 ±

13.0 µm respectively for surfactant concentration of 0.5, 2.5, and 5 vol%. A similar trend, i.e., the mean droplet size reduced with an increase of surfactant concentration, is apparent in the case of DMF/n-heptane system.

It is recognized that O/O emulsions have poor stability due to weaker associations of the traditional surfactants at the O/O interface and higher mutual solubility between the dispersed and continuous phases.276 These inherent characteristics lead to rapid flocculation, coagulation, and ultimately phase separation in O/O emulsion systems, with more significant Ostwald ripening than in aqueous emulsions.229 This was prevalent in the experimental system considered in this work, as presented in Figure 6.3 for the DMF/n- heptane/F127® emulsion with 5 vol% surfactant concentration. The images in Figure 6.3 show significant coalescence of the dispersed phase droplets in just 6 seconds after the emulsion was placed on the optical microscope stage.

160

Figure 6.3: Optical microscope images of n-heptane/DMF emulsion with 5 vol% F127® surfactant concentration.

The presence of surfactants also increased the viscosity of the continuous phase. Teo and

Jana231 observed an increase of viscosity in solution of F127® surfactant in DMF, e.g., the viscosity increased from 0.87 mPa-s for DMF to 2.3 mPa-s with 5 vol% F127® in solution.

Such an increase in viscosity has potential impacts on the diffusion kinetics and the gel time of the polyimide system. The presence of polyimide precursor materials also increases the viscosity of the continuous phase. The higher viscosity of the continuous phase can affect the size of the dispersed phase droplets in the emulsion and the macrovoids in the gel structures. In view of the above, the morphology of aerogel foam structures is discussed next.

6.3.2 Morphology of Aerogel Foam

The aerogel foams primarily consist of macrovoids (as in foams) in a porous polymer domain (as in aerogels). The porous polymer domains in turn are formed by the networks

161 of polymer strands with embedded inherent pore structures. The macrovoids presented in

Figure 6.4 originated from the dispersed phase liquid droplets in the starting O/O emulsions. Recall that the dispersed phase liquid was replaced by acetone in the solvent exchange step and finally by liquid carbon dioxide in the supercritical drying step. The images presented in Figure 6.4 confirm that the macrovoids produced in polyimide aerogel foams had a certain uniqueness. They were not connected by open pores as in polyHIPEs267 or in emulsion-templated syndiotactic polystyrene (sPS) aerogel foams.199 However, this type of aerogel foams cannot be strictly classified as closed cell foams as the domains that separate the macrovoids are mesoporous in nature (Figure 6.4g). A closer look at Figure

6.4g reveals that polyimide strands organized more densely at the macrovoid surface layers than in the bulk. As will be seen later, such dense organization of polymer strands at the macrovoid surface had strong ramification on specific surface area.

Figure 6.4: Emulsion-templated aerogel foams with cyclohexane dispersed phase and surfactant concentration of (a) 0.5 vol%, (b) 2.5 vol% and (c) 5 vol%. The corresponding emulsion-templated aerogel foams with n-heptane are shown in (d-g) at similar surfactant concentrations. (g) Image of the skin layer at the macrovoid surface.

Table 6.3 lists the data on porosity, pore volume, shrinkage, bulk density, and surface area of aerogel foam materials. Recall that the polymer content in DMF solution was kept constant in all cases at 7.4 wt% and the dispersed phase content was kept constant at 30

162 vol%, while the concentration of F127® surfactant and type of dispersed phase was varied.

It is apparent from the data presented in Table 6.3 that emulsion-templating with cyclohexane led to an increase of the porosity from 93.7 to 95.5 % and corresponding increase in pore volume from 9.3 m3/g to 15.0 m3/g. However, both porosity and total pore volume reduced from 95.5 to 91.4 % and from 15.0 to 7.2 m3/g respectively with an increase of surfactant concentration from 0.5 vol% to 5 vol%. We attribute this to diameter shrinkage especially at high surfactant concentrations. The aerogel diameter shrinkage was

11.0% for polyimide with no surfactant compared to 19.2% for an emulsion-templated polyimide aerogel foam with 5 vol% surfactant. Another immediate effect of shrinkage is higher bulk density. The bulk density increased from 0.065 to 0.127 g/cm3 as the surfactant concentration was increased from 0.5 vol% to 5 vol%. The n-heptane-templated aerogel foams exhibited similar trends (Table 6.3).

163

Surfactant Total Pore Bulk BET Strand Dispersed Porosity Shrinkage Gel Time concentration Volume Density Surface Diameter phase (%) (%) (mins) (vol%) (cm3/g) (g/cm3) Area (m2/g) (nm) 0 0.081 ± - 93.7 ± 0.0 9.3 ± 0.0 11.0 ± 0.3 5.2 ± 0.0 812 ± 13 9.3 ± 1.7 (neat PI) 0.000 0.065 ± 0.5 95.5 ± 0.2 15.0 ± 0.5 11.1 ± 1.0 7.7 ± 0.3 456 ± 6 15.5 ± 3.9 0.002 0.074 ± 2.5 Cyclohexane 94.5 ± 0.2 13.1 ± 0.1 15.3 ± 0.5 8.6 ± 0.6 363 ± 6 21.1 ± 3.7 0.002 0.127 ± 5 91.4 ± 0.2 7.2 ± 0.1 19.2 ± 0.0 12.4 ± 0.3 285 ± 25 28.1 ± 6.3 0.001 0.056 ± 0.5 96.0 ± 0.1 17.2 ± 0.0 7.5 ± 0.0 10.5 ± 0.5 521 ± 36 15.7 ± 3.0 0.001 0.067 ± 2.5 n-heptane 95.1 ± 0.1 14.2 ± 0.2 10.9 ± 0.1 13.2 ± 0.3 377 ± 21 27.8 ± 4.8 0.001 0.094 ± 5 92.7 ± 0.2 9.8 ± 0.4 13.3 ± 0.3 15.2 ± 0.3 232 ± 3 35.4 ± 6.4 0.002 Table 6.3: Characteristic properties of aerogel foams as function of surfactant concentration.

164

The DMF and acetone residues in emulsion-templated polyimide gel may be responsible for higher diameter shrinkage observed in the case of aerogel foams. As will be discussed next, the supercritically dried aerogel foams contained residues of surfactants even after several solvent exchange steps. The residual surfactant in turn possibly retained DMF and acetone in the gel, which yielded capillary stress during the supercritical drying step, thus causing collapse of the pore structures and shrinkage of the aerogel foam.

The IR spectra of the neat PI aerogel, aerogel foams and F127® surfactants are presented in Figure

6.5a. The neat PI aerogel (blue spectra) shows that the imide functional groups were successfully formed as inferred from the presence of absorbance bands at 1716 and 1776 cm-1, corresponding to the symmetrical and asymmetrical stretching of the C=O group, respectively. In addition, the band at 1365 cm-1 is associated with the C-N stretching of the imide rings in the structure. The absence of significant peaks at 2926 and 3273 cm-1, indicative of the COOH and CONH functional groups respectively, confirm that the amic acid functional groups were chemically imidized. The

IR spectra for the F127® block copolymer surfactant is indicated by the black curve in Figure 5a.

The surfactant is characterized by the peaks at 2880 cm-1 and 1100 cm-1, corresponding to the alkane and ether groups respectively. For aerogel foams, an increase in the absorbance peaks at

2880 and 1100 cm-1 with increasing surfactant concentration indicates retention of surfactant in the aerogel foam structure. Note that the C-O-C peak of the F127 surfactant at 1100 cm-1 overlaps with the imide ring deformation band at the same wavenumber.

The TGA traces presented in Figure 6.5b corroborate such a finding. It is noted that degradation of F127® surfactant started at around 350 C (5 % weight loss at 393 °C) with no char residue at

800 C. In contrast, neat PI aerogel monoliths started degradation at 525 C (5 % weight loss at

553 °C) and had a char yield of 58 wt%. The char yield from emulsion-templated aerogel foams

165 at 800 C varied with surfactant concentration used in the synthesis step. The char yield reduced with an increase of surfactant concentration, e.g., 57, 52, and 44 wt% for surfactant concentrations of 0.5, 2.5, and 5.0 vol% respectively. The above data indicate that emulsion-templated aerogel foams contained residual surfactant despite repeated washings during the solvent exchange step.

The TGA traces in Figure 6.5b of the aerogel foams (green, red and yellow curves) do not exhibit distinct weight loss regimes attributed to the residual F127. Instead, they exhibit a distributed departure from the neat PI curve (blue curve) at 450 °C. This is to be expected as polyimide is a good thermal conductor and that some time is required for heat transfer to the surfactant embedded within the structure.

Figure 6.5: (a) IR and (b) TGA curves of F127®, neat PI and emulsion templated aerogel foams.

The data in Table 6.3 also indicate that the presence of surfactants delayed gelation of polyimide in emulsion-templated systems. The gelation time was inferred from the crossover between the storage and loss modulus values (Figure 6.6a). The neat PI itself turned into a gel in 5.2 min, while the emulsion-templated systems took 7.7, 8.6, and 12.4 min at surfactant concentrations of 0.5,

2.5, and 5 vol% respectively. As alluded to earlier, the sol-gel transition is a function of the cross-

166 linking reactions. Teo and Jana231 showed that the gel times in polyimides can be manipulated through two mechanisms. First, a change in the solvent environment can result in a change in the reaction equilibrium and ultimately the conversion in step-growth reactions. It was shown that a more basic solvent environment delays gel times and affects the crosslinking conversion as evident from the mechanical properties. Second, gelation can be prolonged by the addition of a viscosity modifier, such as the F127® surfactant used in the current work, that also increases the viscosity, thus inhibiting the diffusion rates of the reactants to the reaction sites. This second approach did not affect the reaction equilibrium and instead helped retain the superior mechanical properties.231

In the context of the present work, therefore, we assume that the reaction equilibrium was not altered by the use of F127® surfactant.

The complex viscosity of all reaction systems increased with time (Figure 6.6b) due to crosslinking reactions proceeding in the continuous phase. However, the viscosity increased comparatively slower in emulsion-templated systems.

Figure 6.6: (a) Storage and loss modulus and (b) complex viscosity of neat PI and emulsion- templated gels.

The BET surface area of aerogel foams reduced with an increase of surfactant concentration in the reaction mixture (Table 6.3). This can also be seen in the BET isotherms in Figure 6.7. The BET

167 surface area reduced from 812 m2/g for neat PI aerogel monolith to 285 m2/g for cyclohexane- templated and 232 m2/g for n-heptane-templated aerogel foam materials at 5 vol% surfactant concentration. The lower surface area in aerogel foams can be attributed to higher density macrovoid skin layers. The macrovoid skin layers were 1-2 strand diameters thick, as evident from

Figure 4g. It was reported earlier for sPS aerogel foams199 that the skin layers of higher density originate from preferential nucleation of polymer strands at the high energy interfaces, very much akin to expedited heterogeneous separation. The polyimide strands also thickened in the presence of surfactants, resulting in a reduction of specific surface area. This latter factor is supported by the data on polyimide strand diameter (Table 6.3) gleaned from the SEM images presented in

Figure 6.8. The mean polyimide strand diameter increased from 9.3 nm for neat PI monoliths to

15.5 nm, 21.1 nm, and 28.1 nm for cyclohexane-templated aerogel foams synthesized with surfactant concentrations of respectively 0.5, 2.5, and 5.0 vol%. The n-heptane-templated aerogel foam systems exhibited even higher diameter polymer strands, e.g., 15.7 nm, 27.8 nm, and 35.4 nm at surfactant concentrations of 0.5, 2.5 and 5.0 vol% respectively. Such a thickening of polymer strands is attributed to a larger gap between the binodal line of liquid-liquid demixing phase separation and the sol-gel transition caused by a reduction in reaction rates due to higher solvent viscosity.204,277

168

Figure 6.7: BET Isotherms of polyimide aerogel foams with varying surfactant concentration with (a) cyclohexane and b) n-heptane as dispersed phase.

In this polyimide reaction system, the system first undergoes liquid-liquid demixing, forming both polymer-rich and solvent-rich regions.4 These regions gradually coarsen to reduce the free energy of the system through reduction of interfacial area between the two phases. As the crosslinking reaction proceeds, the system encounters a sol-gel transition, thus locking the network structure in place. If the reaction rates are reduced through increased viscosity of the system (in this case, via introduction of the surfactant) there is a greater propensity of the two phases to continue coarsening prior to undergoing sol-gel transition, resulting in thicker strands. This was elaborated earlier by

Teo and Jana.231

169

Figure 6.8: Representative images of skin layers formed in emulsion-templated aerogel foams with (a-c) cyclohexane and (d-f) n-heptane dispersed phase with surfactant concentration (a,d) 0.5 vol%, (b,e) 2.5 vol% and (c,f) 5 vol%.

6.3.3 Macrovoid Size Distribution

The final macrovoid size distribution in the aerogel foams resulted from the dispersed phase liquid droplet size distribution in the parent emulsion. Recall that dispersed phase droplets created the macrovoids in the gel and aerogel foams. Therefore, it was expected that the dispersed phase droplet size distributions presented in Figure 6.2 would also map well for macrovoid size distribution in the aerogel foams. From the data in Figure 6.9, it is inferred that macrovoid size reduced with an increase of surfactant concentration. The macrovoids found in cyclohexane- templated aerogel foams were consistently smaller than those produced n-heptane-templated system (Table 6.4). It is worth noting that the macrovoid sizes presented in Figure 6.9 and Table

6.4 are consistently smaller than the emulsion droplets presented in Figure 6.2 and Table 6.2. This can be attributed to the differences in viscosity of the two systems. As explained earlier, the presence of polyimide precursors increased the viscosity of the system as the crosslinking reactions

170 proceeded, as shown in Figure 6.6b. This significant increase in viscosity reduced creaming, aggregation, and coalescence rates. In addition, shrinkage up to 19.2% experienced during supercritical drying by aerogel foams is a contributing factor for smaller macrovoid sizes in aerogel foams. It is also noted that the heat from the microscope light used in the characterization of emulsion droplet size in Figure 6.2 could also quite possibly facilitate coalescence of droplets in the emulsion systems (without polyimide precursors) at a much faster rate.

Figure 6.9: Macrovoid size distribution with (a-c) cyclohexane and (d-e) n-heptane as the dispersed phase. The surfactant concentration was (a,d) 0.5 vol%, (b,e) 2.5 vol% and (c,f) 5.0 vol%.

171

Table 6.4: Average, maximum and minimum macrovoid diameters for emulsion templated aerogel foams. Surfactant Average Maximum Minimum Dispersed concentration macrovoid macrovoid macrovoid phase (vol%) diameter (µm) diameter (µm) diameter (µm) 0.5 39.5 ± 39.0 214 4.6 2.5 cyclohexane 30.5 ± 20.1 113 2.9 5 16.5 ± 11.9 59.1 3.3 0.5 72.9 ± 49.6 228 10.8 2.5 n-heptane 36.5 ± 24.7 154 7.1 5 30.8 ± 19.2 95.0 3.7

6.3.4 Mechanical Properties

The compressive stress vs. strain diagrams of neat polyimide aerogel monolith, polyimide aerogel monolith with surfactant, and emulsion-templated polyimide aerogel monolith are presented

Figure 6.10a. The compressive modulus values are listed in Table 6.5. All 3 compressive stress vs. strain curves exhibit the same shape, with three broad regions. As per Swyngedau,205 the first region (strain 0-0.04 mm/mm) represents the deformation of the original matrix with the applied load borne by the skeletal structure of the crosslinked polymer networks. The second region with strain from 0.04 to 0.7 mm/mm represents the collapse of the skeletal structure and densification of the pores. The third region at strains >0.7 mm/mm represents almost complete compaction of the pores and the load is now borne by the compressed bulk polymer. The compressive modulus of the aerogel structure was calculated from the first region of the data; representative stress-strain curves at low strains are provided in the inset in Figure 6.10a.

172

Figure 6.10: (a) Compressive stress strain curves and (b) modulus vs bulk density of neat polyimide, polyimide with surfactant and emulsion templated polyimide. Graph in the insert in (a) shows the same samples at low strains of <0.1.

Table 6.5: Compressive modulus of neat polyimide, polyimide with surfactant concentration and emulsion templated polyimide. Surfactant concentration Compressive Modulus Dispersed Phase (vol%) (MPa) 0 44.5 ± 1.3 0.5 44.1 ± 2.6 No 2.5 44.1 ± 1.1 5 38.3 ± 2.7 0.5 12.8 ± 0.5 2.5 Yes 16.2 ± 1.4 5 18.9 ± 0.7

The data presented in Table 6.5 show that the inclusion of macrovoids through the emulsion- templating process led to a drop of compressive modulus values from 44 MPa to 12.8 MPa owing to reduction of the bulk density of the aerogel foam materials. This trend is also reflected in Figure

6.10b. In Figure 6.10b, both polyimide with surfactant and emulsion-templated aerogels display similar trends, but at different modulus values. In both cases, the bulk density increases with an increase of surfactant concentration but the compressive modulus is not significantly affected as was earlier reported by Teo and Jana.231 As discussed earlier, the amount of surfactant retained in

173 the aerogel structures also increased in proportion to the amount of surfactant taken in the gel precursor materials. We note here that the residual surfactant was dispersed well within the structure and did not provide load bearing capability at low strain regions.

6.3.5 Oil and Water Absorption

The macrovoids and the mesoporous structures of aerogel foams contributed to oil absorption capacity. The data presented in Figure 6.11 and Table 6.6 indicate that emulsion templated aerogels were able to absorb up to two times the weight of the oil-phase, i.e., n-heptane compared to the corresponding aerogel monolith. However, aerogel foams produced at higher surfactant concentration absorbed less oil, e.g., 11.2 mL/g at surfactant concentration of 0.5 vol% to 7.7 mL/g at surfactant concentration of 5.0 vol%. We attribute this to greater shrinkage and corresponding reduction in total pore volume from 15.0 to 7.2 m3/g (Table 6.3). The oil uptake rate first showed an initial jump from 0.16 to 0.35 mL/g-s between polyimide aerogel monolith and emulsion- templated aerogel with 0.5 vol% surfactant concentration. This jump in oil uptake rate can be corroborated with an increase of the water contact angle values from 68.7 to 84.9o (Table 6.6), attributed to polypropylene hydrophobic groups of the surfactant aggregating on the interfaces and pointing outward, thus increasing the hydrophobic nature of the surface. Note that water contact angle on a melt-pressed film of F127® was 52.4 ± 0.3 o. Thus, the use of hydrophobic liquids such as cyclohexane in the dispersed phase during emulsion-templating increased the affinity of PPO segments of F127® molecules to remain at the macrovoid interfaces. This, in turn increased the water contact angle values on compressed aerogel foam specimens. The contact angle values, however, did not change at higher surfactant concentrations.

174

Figure 6.11: Oil and Water Absorption of polyimide and emulsion templated polyimide over time

The presence of macrovoids increased the total water uptake slightly from 12.4 to 14.3 mL/g. In a similar vein, an increase in surfactant concentration in the aerogel foams resulted in a reduction of total water uptake due to increased shrinkage.

Table 6.6: Oil and water absorption data for neat polyimide and emulsion templated polyimide

Oil Absorption Water Absorption Surfactant Dispersed Contact Total Total Concentration Initial rate Initial rate Phase Angle (o) Absorbed Absorbed (vol%) (mL/g.s) (mL/g.s) (mL/g) (mL/g) 0.0 - 68.7 ± 0.8 5.6 ± 0.1 0.16 ± 0.02 12.4 ± 0.2 0.09 ± 0.00 0.5 84.9 ± 0.9 11.2 ± 0.2 0.35 ± 0.00 14.3 ± 0.2 0.11 ± 0.01 2.5 Cyclohexane 85.1 ± 2.4 7.7 ± 0.5 0.17 ± 0.02 10.6 ± 1.4 0.08 ± 0.00 5.0 84.2 ± 0.9 7.2 ± 0.3 0.16 ± 0.01 9.1 ± 0.5 0.08 ± 0.03

Both the oil and water uptake rates reduced with an increase of surfactant concentration. For example, the oil uptake rate for aerogel foams reduced from 0.35 to 0.16 mL/g-s with increasing surfactant concentration. This can be attributed to reduction of micro and mesopores associated with the thickening of the polymer strands as shown in Table 6.7.231 This shift to larger pore sizes would result in a lower capillary pressure, thus reducing the fluid uptake rate. The water uptake

175 rate was also consistently lower than the oil uptake rate due to inherent hydrophobicity of the aerogel foams.

Table 6.7: Micro, meso and macropore distribution of emulsion templated polyimide Surfactant Micropore Mesopore Macropore Dispersed Concentration Volume Volume Volume Phase % % % (vol%) (cm3/g) (cm3/g) (cm3/g) 0.0 - 0.20 1.48 6.93 51.4 6.42 47.4 0.5 0.11 0.73 0.52 3.45 14.34 95.82 2.5 Cyclohexane 0.08 0.62 0.49 3.77 12.51 95.61 5.0 0.05 0.62 0.34 4.67 6.85 94.70

In summary, the macrovoids aid in increasing the total pore volume and act as reservoirs for fluid storage, while the nanometer sized pores of the polyimide aerogel continuous phase provide the capillary pressure for fluid absorption, thus determining the fluid uptake rate.

6.4 Conclusion

This paper reports successful adaptation of emulsion-templating method for fabrication of aerogel foams from water-sensitive monomers, such as polyimides, using an oil-in-oil emulsion system.

The polyimide aerogel foam systems reported in this work exhibited micrometer size voids along with meso and macropores inherent to polyimide aerogels. The block copolymer surfactant stabilized the O/O emulsion system, produced smaller size dispersed phase droplets, and thickened the polyimide strands, but did not alter the compressive mechanical properties. The final aerogel foam structures displayed improved oil absorption capabilities due to an increase of pore volume coupled with an increase of hydrophobicity derived from the residual surfactants.

176

CHAPTER 7

FACILE SYNTHESIS OF BICONTINUOUS GYROID STRUCTURED AEROGELS USING SACRIFICIAL MOLDS MADE BY FUSED DEPOSITION MODELLING

Abstract

A fused deposition modelling (FDM) process is used to synthesize aerogels with a bicontinuous gyroid structure. These enabled the incorporation of ultrahigh porosity in these structures, higher than traditional aerogel monoliths. The sacrificial molding process utilized in this study enabled the easy release of the synthesized gels from the molds, without any damage to the intricate geometry of these molds. This process grants flexibility in the geometrical forms that the gels can take, enabling the fabrication of free-standing polyimide aerogels with up to 98.9 % porosity and low bulk density values of 0.0146 g/cm3. In addition, the incorporation of the bicontinuous gyroid structure allows for previously brittle aerogels to achieve elasticity and increased elongation at break. This new process is particularly attractive for sol-gel systems as it does not have stringent processing restrictions compared to other gel printing processes. This process opens up aerogels for new potential applications such as metastructured damping applications, as well as formation of complex membrane structures with both macro and mesopores.

177

7.1 Introduction

Aerogels are a class of highly porous materials that are renowned for their high porosity (> 90 %) and high surface areas (up to 1000 m2/g). Kistler first synthesized silica aerogels by removing the solvent in gels through supercritical drying.1 These silica aerogels were known for their small pore size distributions (50 – 200 nm), resulting in low thermal conductivity as convection is prevented within the structure. As a result, silica aerogels have seen widespread use as thermal insulation in construction and aerospace applications. However, silica aerogels are typically brittle and are prone to collapse in the presence of moisture. These drawbacks have led to the development of polymeric aerogels that continue to exhibit high porosity and high surface areas yet possessing improved mechanical and chemical properties such as high stiffness, chemical resistance and hydrophobicity. Polymeric aerogels have been synthesized from a variety of materials such as polystyrene,66 PEEK,278 polyurea,183 polyurethane279 and polyimide.231 Most aerogels described are synthesized from the sol-gel process, with some materials synthesized from a thermo-reversible gelation process (e.g. polystyrene and PEEK). The majority of aerogels in the literature are fabricated in the form of monoliths, cylinders, films36 or foams199,208,210 with a small proportion reporting aerogel microparticles.206 Complex geometrical shapes have been avoided due to the high cost of tooling and molding, as well as difficulty in removing gels from intricate molds without damage to the structure.

While polymeric aerogels are still used primarily as thermal insulation, there has been ongoing development to use these porous materials in other applications. For example, polyimide aerogels have been used as antenna substrates,42 and recently in filtration. Zhai and Jana180 and Kim and

Jana64 have shown that aerogels are excellent candidates for nanoparticle filtration, achieving

99.9% rejection of nanoparticles in the 75-150 nm range, owing to the small pore distributions of

178 both polyimide and polystyrene aerogels. It is envisioned that allowing aerogels to adopt increasingly complex geometrical configurations would assist in the expansion of the potential applications that aerogels can take.

Injection molding and lithography have been the conventional technologies to achieve 3D structural forms. However, these technologies are not able to fabricate complex 3D structures such as the bicontinuous gyroid structures due to limitations in mold removal and 3D patterning.

Additive manufacturing is a promising technology allowing one to create extremely complex geometries since it prints objects to a layer-by-layer manner. This technology grants immense flexibility in achieving different geometries, particularly in the creation of porous structures. In the fabrication of porous structures, 3D printing allows a greater degree of control over pore structure, pore size distribution and porosity over traditional porous structure fabrication methods such as emulsion, porogen and salt leaching techniques.280,281 In addition, 3D printing can be combined with the above mentioned techniques to introduce macro-level porosity in addition to micro and nano level porosity to form hierarchical porous structures. For example, hierarchical porous structures can be fabricated through powder-based 3D printing. In this method, liquid is sprayed onto a powder bed that undergoes either binding, reaction or crystallization. The final hierarchical structure is comprised of the macrostructure determined by the layer-by-layer injection, while micrometer-sized pores are found in the interstitial spaces of the powder particles that comprise the 3D-printed strands.282 This method however suffers from poor resolution and poor mechanical properties and are restricted by parameters such as powder flowability, stability, wettability and reactivity to achieve stable free-standing structures.283

Another pathway to create hierarchical porous structures has been 3D printing of gels, also known as direct ink writing (DIW), where a paste or liquid is extruded and subsequently dried, gelled or

179 sintered. In this case, the macrostructure porosity can be controlled and introduced through the programmed movement of the nozzle, while the gel would possess inherent macro or mesoporosity. For example, mesoporous bioactive glass (MBG) scaffolds have been successfully synthesized by Wu et al.284 These hierarchical porous structures shows superior ion and drug release profiles, owing to the mesoporosity and high surface areas attributed to the structure formed from the crosslinking of the MBG powder with PVA. In another study, Minas et al. combined emulsion and foam templating into DIW, introducing micrometer sized pores in the DIW scaffold, creating a hierarchical structure with two levels of hierarchy.153 While DIW aids in introducing nano and micrometer sized pores into a structure with millimetre sized porosity, there are severe limitations to the materials that can be used. Ideally, the material to be printed needs to have low viscosity at high shear to enable printing, while possessing quick set times after printing to ensure dimensional integrity of the resultant structure.155 This leads to typical problems experienced by

DIW such as its limited applications to simple geometry, small working window between formulation and crosslinking/setting and potential for structure collapse after extrusion.153,155

To counteract the limitations of DIW, the concept of 3D printing sacrificial molds have been investigated. Sacrificial molds are printed through SLA of a water or alkaline soluble photopolymer.285,286 Subsequently, biomaterial, silicone or urethane elastomers can be injected into the mold to cure. Finally, the cured material can be released form the mold through dissolution in NaOH solution.287 However, the highly basic mold release solvent may be detrimental to gel structures, causing degradation of the desired structures. In addition, other limitations such as viscosity control and swelling during and after irradiation caused by monomer influences could potentially negatively impact resolution and mechanical properties of the generated sacrificial molds. Therefore, development of thermoplastic sacrificial molds would be a benefit to circumvent

180 the challenges faced by SLA methods. It is also envisioned that these thermoplastic sacrificial molds can be easily released with organic solvents without damage to the desired internal structure.

In this study, we introduce a facile method to fabricate hierarchical porous structures comprising of both micrometer and nanometer-sized porosity. This is achieved using a sacrificial molding process with the molds made of high impact polystyrene printed by fused deposition modelling

(FDM). We contend that this method may well serve as a platform technology for fabrication of complex gel and aerogel structures not previously accessible through injection molding or lithography. We successfully fabricated polyimide mesoporous structures within a bicontinuous gyroid mold. The mold was subsequently removed without damaging the internal gel structures.

This method adopted in this work is superior to other sol-gel printing methods as it is not restricted to simplified geometries or gel times. The method may also enable synthesis of aerogel metamaterials whereby both the macrostructure geometry and the chemistry can be independently used as levers to design desired mechanical properties.288–290 In addition, the structures developed in this work shows promise in developing flame-retardant materials without the need for additives.

7.2 Experimental Section

7.2.1 Materials

Pyromellitic dianhydride (PMDA) was purchased from Alfa-Aesar (Haverhill, MA) and 2,2’- dimethylbenzidine (DMBZ) was purchased from Shanghai Worldyang Chemical Co. Ltd

(Shanghai, China). Tris(2-aminoethyl)amine (TREN) crosslinker was purchased from Sigma

Aldrich (Milwaukee, WI). Pyridine, acetic anhydride and acetone were purchased from Fisher

Scientific (Ontario, NY). Dimethylformamide (DMF) and dimethyl sulfoxide (DMSO) were

181 purchased from VWR International (Radnor, PA). High impact polystyrene (HIPS) dissolvable filament was purchased from MatterHacker (Foothill Ranch, CA).

7.2.2 Fabrication of Sacrificial mold

A bicontinuous gyroid structure was generated from an implicit equation (1) in K3DSurf:291

푠푖푛푥푐표푠푦 + 푠푖푛푦푐표푠푧 + 푠푖푛푧푐표푠푥 = 0 (1)

The generated bicontinuous gyroid surfaces were subsequently imported into Meshlab and converted into volume models.292 The wall thickness of the volumes were varied at this step, varying from 3.4 to 6.0 cm. The models were then imported to Simplify3D, where the original filled model was converted into the hollow and infilled gyroid objects. This was achieved by controlling the density of infill of the model in Simplify3D. For the hollow molds, a single extruder was used, while a double extruder was used for the molds with infill. In the latter case, the secondary extruder was used to print infill strands in an alternating manner, ensuring that there was always a gap between strands which could be subsequently filled by the input sol. In addition, to prevent leakage of the sol in the molds, the extrusion multiplier was set at 1.35 (distance between strands) and the printing speed set at 3600 mm/min. The molds were printed with a Flashforge

Creator Pro (Zhejiang, China) using HIPS filaments. HIPS filaments were specifically chosen as they were soluble in organic solvents such as DMF, but resistant to dissolution in solvents used for polyimide synthesis, namely DMSO. Representative cross-section molds and their respective

CAD models with and without infill are shown in Figure 7.1.

182

Figure 7.1: Cross-section of molds and CAD models with different wall thicknesses.

7.2.3 Preparation of Polyimide Sol

The polyimide precursor solution (sol) was prepared at room temperature by mixing PMDA,

DMBZ, and TREN in DMSO as the solvent, as per the process outlined by Teo and Jana.231 These dianhydride and diamine monomers were selected for two reasons. Firstly, the presence of aromatic rings in these monomers, along with the absence of flexible ether linkages is expected to produce greater mechanical strength in the resultant polymer backbone. Secondly, these monomers exhibit greater reactivity compared to other typical dianhydrides and diamines used in polyimide synthesis thereby reducing gel times.28,293 Briefly, PMDA and DMBZ, dissolved individually in

DMSO, were mixed with a magnetic stirrer at 1000 rpm for 2 minutes. This produced PMDA end- capped polyamic acid oligomers. Subsequently, TREN, acetic anhydride, and pyridine were added, and the solution was magnetically stirred for an additional 1.5 minutes. The addition of these chemicals promoted both the crosslinking and chemical imidization reaction. A typical polyimide sol sample with 6.7 wt% polymer concentration was prepared using 0.42 g PMDA, 0.42 g DMBZ,

0.080 g TREN, 1.33 g acetic anhydride, 1.25 g pyridine, and 10.0 mL of DMSO. A higher than

183 stochiometric amount of TREN crosslinker was used to obtain appropriate gelation times and stiffer gels.231

7.2.4 Synthesis of Polyimide Aerogels

Once the sol was prepared, it was transferred into a disposable syringe and injected into the 3D- printed hollow shells. Typically, the sol takes 27 mins to transition to a gel. In this work, the sol was left to cure for an additional 24 hours at room temperature to ensure that there was sufficient time for the majority of the crosslinking reaction to take place. After gelation, the gel and hollow shell was washed successively with DMF to dissolve and remove the HIPS exterior shell. For more intricate geometries, e.g., structures with infill or thin wall thicknesses, sonication was required to effectively dissolve the HIPS from the crevices in the gel structures. In the next step, DMF solvent was removed from the inherently porous structure without collapse of the structure due to capillary stresses. Supercritical drying using liquid carbon dioxide was adopted in this work for the purpose.

However, DMF is immiscible with liquid carbon dioxide. Therefore, acetone was used as the bridging solvent due to its miscibility with both DMF and liquid carbon dioxide. The polyimide gels released from the shell were solvent-exchanged with acetone/DMF solvent mixtures with an increasing acetone content at 12-hour intervals. At the end, the samples were washed with 100 % acetone for an additional 5 times to ensure that residual DMF was removed from the gel structure.

The acetone-filled gels were subsequently solvent-exchanged with liquid carbon dioxide and dried under supercritical condition of carbon dioxide at 50 °C and 11 MPa pressure. The process described above is illustrated schematically in Figure 7.2 and shown pictorially in Figure 7.3.

184

Figure 7.2: Procedure of polyimide aerogel synthesis

Figure 7.3: Synthesis of polyimide aerogel through 3D printed hollow shells

Significant preliminary work was done to find the optimal pairing of synthesis solvent, mold release solvent, bridging solvent, and the materials for fabrication of the sacrificial molds. The solvent selection for synthesis was crucial. For example, the synthesis solvent had to enable synthesis of the rigid gels, without causing dissolution or swelling of the gyroid shell. The mold release solvent, on the other, had to be miscible with the synthesis solvent to reduce the interfacial

185 stress and to prevent structural collapse and shrinkage of the gel network in addition to its primary function to rapidly dissolve the shell materials. The bridging solvent had to be miscible with all the solvents used in the process, namely, the synthesis solvent, the mold release solvent, and liquid carbon dioxide. The mutual solubility of liquid carbon dioxide with the bridge solvent and mold release solvent ensures efficient removal of the latter solvents in the solvent exchange steps from the structures prior to supercritical drying process. Any residual synthesis or mold release solvents in the gel subjected to supercritical drying would develop capillary stress, causing local collapse of the porous aerogel structure.

Characterization of Polyimide Aerogels

IR: Infrared spectra was collected on a Nicolet iS50 FTIR tri-detector spectrophotometer (Thermo

Scientific, MA).

TGA: Thermogravimetric analysis (TGA) was conducted under N2 with a Q50 thermogravimetric analyzer (TA Instruments, DE) using a heating rate of 20 °C/min, up to 800 °C.

Porosity and Pore Volume: Porosity was calculated from the values of skeletal (ρs) and bulk density (ρb) as shown in equation (2). The values of skeletal density were obtained using a helium pycnometer (AccuPyc II 1340, Micromeritics Instrument Corp., GA). Bulk density was obtained from the mass and volume of the aerogels.

𝜌 푝표푟표푠푖푡푦 = (1 − 푏) × 100% (2) 𝜌푠

186

Brunaur-Emmett-Teller (BET) surface area: BET surface area of aerogel specimens were obtained from N2 adsorption-desorption isotherms at 77 K using a Micromeritics Tristar II 3020 analyzer

(Micromeritics Instrument Corp. GA).

Compressive Modulus. Compressive modulus of the aerogel samples was measured using an

Instron 5567 tensometer (Norwood, MA). A 1 kN load cell was used, with an extrusion rate of 1.3 mm/min. The compressive modulus of the aerogels was obtained from the slope of the stress-strain curve at a low strain, typically from 0.01 – 0.05 mm/mm.

7.3 Results and Discussion 7.3.1 Gyroid Structure Polyimide aerogels were synthesized using the experimental method described above. First, the method adopted in this work enabled the creation of complex aerogel geometry with curved surfaces and recessed spaces. The method also allowed easy release of the gel specimens from the hollow shell molds without any damage to its relatively fragile structure. Second, the incorporation of 3D-printing with the aerogel synthesis process allowed creation of hierarchical porous structures with increased porosity. Figure 7.4 shows representative images of aerogels from their respective

3D printed hollow shells.

187

Figure 7.4: Polyimide Aerogels with their hollow shells. The bottom row shows the hollow shells, the middle row shows the aerogels fabricated from molds with no infills and the top row shows hierarchical aerogels fabricated from molds with infills.

The aerogel specimens produced in this work exhibited both high porosity and low density, characteristic of aerogels, as shown in Table 7.1. For example, polyimide aerogel gyroids with wall thickness of 3.4 mm had a bulk density of 0.0308 g/cm3 and a porosity of 97.7 %. This bulk density increased nearly linearly to 0.0534 g/cm3 as the wall thickness increased to 6.0 mm. This also resulted in a concomitant reduction in porosity to 96.0 % for the same increase in wall thickness. The gyroids with infill produced lower density and higher porosity than the normal gyroids. This was expected as the infill strands of HIPS added additional porosity to the structure after dissolution in DMF. This same trend in porosity and bulk density is reflected for the gyroids with infills, with the bulk density increasing from 0.0146 g/cm3 to 0.0322 g/cm3 and porosity

188 decreasing from 98.9 to 97.6 % for the same increase in wall thickness from 3.4 mm to 6.0 mm.

The skeletal density was found to be highly consistent across all samples, registering a value of

1.34 ± 0.056 g/cm3. Once again, this was expected, as the skeletal density is determined by the chemical composition and thermodynamics of gelation, independent of the macrostructure geometry.

Table 7.1: Bulk density and porosity of polyimide aerogel gyroids.

Wall Thickness 3 Infill Bulk Density, ρb (g/cm ) Porosity (%) (mm) Control - 0.0749 94.4 3.4 0.0308 97.7 4.7 No 0.0431 96.8 6.0 0.0534 96.0 3.4 0.0146 98.9 4.7 Yes 0.0225 98.3 6.0 0.0322 97.6

Figure 7.5a shows the IR spectrum of the polyimide synthesized in this work. The presence of peaks at 730, 1380, 1716 and 1778 cm-1 indicate the presence of imide groups, while the lack of significant peaks at 1620 and 3000 cm-1 indicate that the majority of polyamic acid was chemically converted into the imide functional groups. The TGA curve in Figure 7.5b shows that synthesized aerogels exhibit high thermal stability, characteristic of polyimides. The synthesized polyimide exhibits a 5 % weight loss at 443 °C, with the majority of degradation occurring at 550 °C. At 800

°C, the char yield was 55 wt%.

189

Figure 7.5: (a) IR spectra and (b) TGA curve of polyimide aerogel

The synthesized aerogels also exhibited a high surface area of 579 m2/g. The high BET surface area and high porosity of the structure are attributed to the fibrillar structure of the polyimide strands shown in Figure 7.6a. These strands were measured to have a thickness of 14.3 ± 2.7 nm as estimated from the SEM images. The BET isotherm of the synthesized polyimide is shown in

Figure 7.6b.

Figure 7.6: (a) SEM image of polyimide aerogel structure and (b) BET isotherm of polyimide aerogel

190

We next looked at the mechanical properties of these gyroid structures. Most of the aerogels’ exhibited similar stress-strain curves typical of porous materials, as described by Swyngedau.205

To elucidate this behavior, we turn to the control aerogel sample, marked as the green curve in

Figure 7.7a. This sample is a polyimide aerogel sample with no complex geometry imparted by

3D-printing. The compression stress strain curve of the aerogels can be divided into 3 broad regions. The first region, defined as the region from 0 – 0.1 mm/mm strain, involves the deformation of the aerogel structure, with the applied load borne by the skeletal structure of the polyimide strands. The second region is described by a plateau, characterized by the region from

0.1 – 0.6 mm/mm strain, whereby the skeletal structure has collapsed, and the pores of the aerogel structure is being compressed. Once all the pores have been compacted, the stress strain curve enters the third region. In the region of > 0.6 mm/mm strain, the curve increases drastically as the applied load is now borne by the compressed bulk polymer.

Comparing the stress-strain curves of the control sample with that of the gyroid samples, we noticed some differences in the stress-strain curves as shown in Figure 7.7. First, the gyroid structures with wall thicknesses of 4.7 and 6.0 mm exhibit the same stress-strain behavior as the control aerogel sample, with the 3 broad regions described above. However, the gyroid structure with the wall thickness of 3.4 mm exhibited brittle fracture at 0.2 mm/mm strain. This can be attributed to low bulk density of 0.0308 cm3/g and porosity of 97.7 % of the structure. For this reason, the compression tests were not conducted on the samples with the infill, as their bulk density values were similar or lower than that of this sample. The next trend that we observed was that the stress-strain curves of specimens with 4.7 and 6.0 mm wall thickness showed initial brittle fracture of the structure after the first region. This is characterized by a drop in the stress value at the 0.06 mm/mm strain mark, before entering the second region and subsequently plateauing out.

191

This phenomenon was also observed visually, with crack formation and propagation in certain areas of the structure during the compression tests. This partial brittle fracture can be attributed to the geometry introduced through the gyroid structure, specifically to the collapse of the macro interstitial spaces between the curved volumes of the gyroid. The second and third region of these structures are similar to that exhibited by the control samples.

Figure 7.7: (a) Stress-strain curves and (b) compressive modulus vs bulk density of polyimide aerogels

We next looked at the effect of wall thickness on mechanical properties. Compression modulus of the aerogel samples were measured at low strain, in the first region of the stress-strain curve.

Figure 7.7b shows that the modulus varies linearly with bulk density of the aerogel sample. For example, the control sample exhibited a compressive modulus of 5.85 MPa with a corresponding bulk density of 0.0749 g/cm3, while the gyroid structure with a wall thickness of 3.4 mm had a compressive modulus of 0.626 MPa with a corresponding bulk density of 0.0308 g/cm3. The compressive modulus of specimens with gyroid structures with increasing wall thickness show linear relationship between the above two values. This result is interesting in itself. For porous materials, the compressive modulus typically scales exponentially with the bulk density, according

192 to the Gibson and Ashby model.71 In particular, porous materials follow a power-law relationship between bulk density and compressive modulus, with foams exhibiting an exponent of 2, while aerogels typically report a higher exponent of 3.6.294 In our result, the almost linear dependence of modulus with bulk density translates into an exponent value of close to 1, significantly lower than the reported values of between 2 and 3.6. This results suggest that the gyroid geometry does impart some advantage for load bearing applications.295

7.3.2 Auxetic Structure

In addition to gyroid structures, we applied this fabrication process to create auxetic polyimide aerogel structures.296–298 The auxetic structure fabricated measured 9 cm x 3.2 cm x 5.4 mm, with a density of 0.0228 g/cm3 and is shown in Figure 7.8a. Introducing this auxetic geometry into the aerogel synthesis process gave the final structure surprising elongation at break. Polyimide aerogels are typically brittle with negligible elongation at break (0.06 mm/mm), particularly when synthesized with monomers with aromatic groups and the absence of ether linkages.36 In contrast, auxetic polyimide aerogels showed up to 0.173 mm/mm strain at break under tension (Figure 7.9), due to the deformation of the individual macro strands shown in Figure 7b-d.

193

Figure 7.8: (a) Polyimide auxetic structure and (b-d) structure undergo tensile deformation.

The stress-strain curve of the auxetic structure under tension in Figure 7.9 shows four distinct brittle fractures, corresponding to the breaking of four strands as tension was applied to the structure. The stress-strain curve also shows that the structure had 4 different tensile moduli, decreasing from 0.286, to 0.0653, to 0.0179, to 0.0127 MPa. This decrease in modulus can be explained by the fact that after each strand broke, more stress is concentrated to the remaining strands and the reduction in the number of strands lead to earlier and easier fracture of the surviving strands.

194

Figure 7.9: Stress-strain curve of auxetic structure undergoing tension

7.4 Conclusion

The work presented in this paper supports that additive manufacturing tools can be easily implemented in generation of aerogel structures with complicated shapes, such as the bicontinuous gyroid aerogel structures with curved surfaces and recessed spaces. The compatibility of sacrificial molding process enables quick fabrication of these structures through sol-gel chemistry that allows for easy extraction of the gel without damage to the structure. The 3D-printed aerogel structures show similar porosity and bulk density as the monoliths, but the compressive modulus varies proportionally with the bulk density in contrast to bulk density exponent of 2.0 for foams and 3.6 for aerogel monoliths.

195

CHAPTER 8

OVERALL SUMMARY AND RECOMMENDATIONS FOR FURTHER STUDY

This work focused on the nano, micro and macro structure control of aerogel morphology through the combination of a variety of different technologies with the aerogel synthesis process. The inclusion of physical chemistry principles, microfluidics, emulsion-templating and 3D printing processes allowed the formation of a variety of structural forms such as monoliths, microparticles, hollow microspheres, foams and mechanical metamaterials.

Chapter 3 showed that the nanostructure of polyimide aerogels could be controlled through changing solvent properties, independent from monomer selection. This was achieved through changing solvent properties. Increasing the basicity of the solvent reduced the forward reaction rate, resulting in a greater separation between the liquid-liquid and solid-liquid phase transitions.

This allowed for the coarsening of the polymer-rich domains before gelation, resulting in thicker polyimide strands and a predominantly macroporous structure. This contrasts with polyimide aerogels synthesized in a more acidic solvent, which resulted in a mesoporous structure. One drawback of using solvent basicity to control pore size distribution is that basicity affects not only the reaction rate, but conversion as well. Use of a basic solvent led to increasingly brittle aerogels that fracture easily compared to aerogels synthesized in more acidic solvents. This lead to the

196 development of controlling reaction rates through a physical process, rather than an equilibrium process, namely, through the viscosity of the solvent. The addition of a viscosity modifier to increase viscosity of the solvent enabled a similar time gap between the phase separations, without interfering with the reaction conversions. This led to increase in strand thickness from 10 to 30 nm, without any detrimental impact to compressive modulus values.

In Chapter 4, a new process to synthesize aerogel microparticles was developed, capitalizing on the accurate metering of microfluidic droplet generators. The combination of an easily assembled microfluidic device with a heated oil bath allowed for the generation of monodisperse gel microparticles whereby particle size and distribution could be controlled through continuous and disperse phase flow rates. Careful control of oil bath temperature and residence time allowed for the elimination of surfactants in the microfluidic synthesis process. The resultant microparticles were discrete and spherical, without the problems of coalescence and agglomeration typically seen in other microparticle fabrication processes. The addition of another flow enabled the fabrication of core-shell hollow microspheres, which are the first of its kind reported in literature.

Chapter 5 and 6 detailed the development of aerogel foams using the emulsion-templating method.

In Chapter 5, a physically crosslinked, thermo-reversible gelation system was selected as a proof- of-concept for the incorporation of emulsion-templating into the aerogel fabrication process. In addition, this system enabled the use of a water-in-oil emulsion system, which is significantly easier to stabilize. One area of concern was that the temperature changes required for this thermo- reversible system had a deleterious effect on the surfactant’s amphiphilic ability, leading to emulsion instability. Therefore, fine-tuning of emulsion stability, temperature and gel times were needed to ensure that the emulsion was incorporated in the gel structure without any undesired

197 creaming or bilayer formation. In this study, a variety of processing parameters were studied, with surfactant concentration identified as the largest contributor to macrovoid size distribution.

In Chapter 6, the lessons learnt in Chapter 5 was transferred to a chemically crosslinked, sol-gel system. The shift to a polyimide sol-gel system necessitated the use of oil-in-oil emulsions. The first part of this study involved the characterization of these emulsions through interfacial tension measurements and droplet size distributions. A block copolymer surfactant was used to impart a measure of stability to these emulsions to ensure that these emulsions were successfully incorporated in the gel. Characterization of the resultant aerogel foams indicated that the block copolymer surfactant was retained in the aerogel foam structure. This increased the surface energy of the polyimide aerogel foams due to the orientation of the hydrophobic groups of the surfactant at the emulsion interfaces. This translated to a preferential absorption of oil over water, resulting in an almost two-fold increase in oil carrying capacity with similar water carrying capability.

In Chapter 7, a new process was developed to enable aerogels to assume complex, intricate shapes with curved surfaces and recessed spaces. This was achieved through 3D printing of sacrificial hollow molds into which the precursor sol could be injected in. In the development of this process, two key challenges had to be overcome. The first was the optimal pairing of synthesis solvent, mold release solvent, bridging solvent and material used for fabrication of the sacrificial molds.

Each solvent selected had to fulfill its designated purpose at its designated stage in the process without interference in the other steps. The other key challenge was the optimization of printing parameters to ensure that the injected sol would not leak out of the hollow shell, as well as enabling the hollow shell to be dissolved quickly during the mold release step. This technique enabled the fabrication of free-standing, hierarchical porous structures with up to 98.9 % porosity.

198

The research presented in this work lays the foundation in the development of new processing techniques to control the nano, micro and macro structure of aerogels. However, these techniques are still in their infancy, and a few suggestions are provided for future work:

Tuning Pore Structure Through Solvent Effects. This work can be used to tune aerogel structures for nanoparticle filtration purposes, specifically to control both selectivity and permeability. In addition, this technique can be applied to not only polyimide aerogels, but other material systems such as polyurea, polyurea and polyurethane.

Aerogel Microparticles. The aerogel microparticle synthesis process can be adapted for the cellulose aerogel system. Cellulose aerogel microparticles would be an ideal candidate for drug delivery applications due to its high biocompatibility and biodegradable properties. Another area of interest would be the production of microparticles in other shapes beside the typical spherical shape. For example, pill shaped microparticles can potentially be fabricated through controlled coalescence of droplets prior to gelation, or through pulsed control of dispersed phase flowrates in the droplet generator.

Aerogel Core-Shell Hollow Spheres. As mentioned, these structures are very unique in the sense that they are hollow in the core but have a porous shell that could permit transmission of fluids into and out of the core. This would make these structures ideal as catalyst carriers, whereby catalysts can be inserted into the hollow core and would not be allowed to leave the core due to the small pore size distribution of the shells. However, the permeable shells could potentially allow the diffusion of reagents in, and products out of the core after interaction with the embedded catalysts. An alternative to this would be to allow diffusion of monomers into the core, followed by polymerization in the core, and subsequent release from the spheres through fracture.

199

Aerogel Foams. This hierarchical porous structure has the potential for multiple different applications. The first potential application is for oil/water separation. The macrovoids in these structures would reduce the total path length of fluid flow, resulting in higher permeability.

Therefore, filtration experiments should be conducted to determine both flow rate and rejection capabilities of these aerogel foams. Another area of interest would be to use Pickering emulsions to increase emulsion stability at elevated temperatures (particularly for the sPS system) to enable the incorporation of larger dispersed phase volumes. Another potential study would be incorporate emulsion-templating into a film casting process, enabling the fabrication of emulsion-templated sheets, which could be easier to manipulate as a membrane. One last study would be to use these aerogel foams as tissue scaffolding.

Aerogel Mechanical Materials. This work developed a process using 3D printed sacrificial molds to enable aerogels to take complex structure. While initial work has been done to investigate the effect of mechanical metamaterial geometry, there exist a whole host of geometries that can be explored. Additional work can also be done to independently vary chemistry (e.g. monomer selection, crosslinker ratio, solvent properties) and geometrical parameters (e.g. strand thickness, periodicity). In addition, dynamic origami-like structures can also be explored, with aerogel structures that can expand and contract repeatedly.

200

REFERENCES (1) Kistler, S. Coherent Expanded Aerogel Jellies. Nature 1931, 127, 741. (2) Nicolaon, G.; Teichner, S. On a New Process of Preparation of Silica Xerogels and Aerogels and Their Textural Properties. Bull. Soc. Chim. Fr. 1968, 5, 1900. (3) Tillotson, T. M.; Hrubesh, L. W.; Thomas, I. M. Partially Hydrolyzed Alkoxysilanes as Precursors for Silica Aerogels. MRS Online Proc. Libr. Arch. 1988, 121. https://doi.org/10.1557/PROC-121-685. (4) Brinker, C. J.; Keefer, K. D.; Schaefer, D. W.; Assink, R. A.; Kay, B. D.; Ashley, C. S. Sol-Gel Transition in Simple Silicates II. J. Non-Cryst. Solids 1984, 63 (1), 45–59. https://doi.org/10.1016/0022-3093(84)90385-5. (5) Tewari, P. H.; Hunt, A. J.; Lofftus, K. D. Ambient-Temperature Supercritical Drying of Transparent Silica Aerogels. Mater. Lett. 1985, 3 (9), 363–367. https://doi.org/10.1016/0167-577X(85)90077-1. (6) Nguyen, D.; Gowda, G. Preparation of Silica Aerogel Insulating Material. J. Can. Ceram. Soc. 54 (40), 1985. (7) Iler, R. Chemistry of Silica - Solubility, Polymerization, Colloid and Surface Properties, and Biochemistry; Wiley: New York, 1979; pp 100–120. (8) Dieudonné, P.; Alaoui, A. H.; Delord, P.; Phalippou, J. Transformation of Nanostructure of Silica Gels during Drying. J. Non-Cryst. Solids 2000, 262 (1), 155–161. https://doi.org/10.1016/S0022-3093(99)00687-0. (9) Brinker, C. J.; Scherer, G. Sol-Gel Science: The Physics and Chemistry of Sol-Gel Processing; Academic Press. (10) Zhang, G.; Dass, A.; Rawashdeh, A.-M. M.; Thomas, J.; Counsil, J. A.; Sotiriou- Leventis, C.; Fabrizio, E. F.; Ilhan, F.; Vassilaras, P.; Scheiman, D. A.; et al. Isocyanate- Crosslinked Silica Aerogel Monoliths: Preparation and Characterization. J. Non-Cryst. Solids 2004, 350, 152–164. https://doi.org/10.1016/j.jnoncrysol.2004.06.041. (11) Soleimani Dorcheh, A.; Abbasi, M. H. Silica Aerogel; Synthesis, Properties and Characterization. J. Mater. Process. Technol. 2008, 199 (1), 10–26. https://doi.org/10.1016/j.jmatprotec.2007.10.060. (12) Nakanishi, K.; Minakuchi, H.; Soga, N.; Tanaka, N. Structure Design of Double-Pore Silica and Its Application to HPLC. J. Sol-Gel Sci. Technol. 1998, 13 (1), 163–169. https://doi.org/10.1023/A:1008644514849. (13) Wagh, P. B.; Begag, R.; Pajonk, G. M.; Rao, A. V.; Haranath, D. Comparison of Some Physical Properties of Silica Aerogel Monoliths Synthesized by Different Precursors. Mater. Chem. Phys. 1999, 57 (3), 214–218. https://doi.org/10.1016/S0254- 0584(98)00217-X. (14) Harreld, J. H.; Ebina, T.; Tsubo, N.; Stucky, G. Manipulation of Pore Size Distributions in Silica and Ormosil Gels Dried under Ambient Pressure Conditions. J. Non-Cryst. Solids 2002, 298 (2), 241–251. https://doi.org/10.1016/S0022-3093(01)01051-1. (15) Cao, B.; Zhu, C. Sol–Gel Derived Self-Supporting Film. J. Non-Cryst. Solids 1999, 246 (1), 34–38. https://doi.org/10.1016/S0022-3093(99)00047-2. (16) Kramer, S. J.; Rubio-Alonso, F.; Mackenzie, J. D. Organically Modified Silicate Aerogels, “Aeromosils.” MRS Online Proc. Libr. Arch. 1996, 435. https://doi.org/10.1557/PROC-435-295.

201

(17) Leventis, N.; Sotiriou-Leventis, C.; Zhang, G.; Rawashdeh, A.-M. M. Nanoengineering Strong Silica Aerogels. Nano Lett. 2002, 2 (9), 957–960. https://doi.org/10.1021/nl025690e. (18) Capadona, L. A.; Meador, M. A. B.; Alunni, A.; Fabrizio, E. F.; Vassilaras, P.; Leventis, N. Flexible, Low-Density Polymer Crosslinked Silica Aerogels. Polymer 2006, 47 (16), 5754–5761. https://doi.org/10.1016/j.polymer.2006.05.073. (19) Katti, A.; Shimpi, N.; Roy, S.; Lu, H.; Fabrizio, E. F.; Dass, A.; Capadona, L. A.; Leventis, N. Chemical, Physical, and Mechanical Characterization of Isocyanate Cross- Linked Amine-Modified Silica Aerogels. Chem. Mater. 2006, 18 (2), 285–296. https://doi.org/10.1021/cm0513841. (20) Meador, M. A. B.; Fabrizio, E. F.; Ilhan, F.; Dass, A.; Zhang, G.; Vassilaras, P.; Johnston, J. C.; Leventis, N. Cross-Linking Amine-Modified Silica Aerogels with Epoxies: Mechanically Strong Lightweight Porous Materials. Chem. Mater. 2005, 17 (5), 1085–1098. https://doi.org/10.1021/cm048063u. (21) Mulders, N.; Chan, M. H. W. Heat Capacity Measurements of $^{3}\mathrm{He}$- $^{4}\mathrm{He}$ Mixtures in Aerogel. Phys. Rev. Lett. 1995, 75 (20), 3705–3708. https://doi.org/10.1103/PhysRevLett.75.3705. (22) Kim, G.-S.; Hyun, S.-H. Synthesis of Window Glazing Coated with Silica Aerogel Films via Ambient Drying. J. Non-Cryst. Solids 2003, 320 (1), 125–132. https://doi.org/10.1016/S0022-3093(03)00027-9. (23) Wendell, R.; Wang, J.; Begag, R. Polyimide Aerogels, Carbon Aerogels, Metal Carbide Aerogels and Methods of Making Same. (24) Sroog, C. E.; Endrey, A. L.; Abramo, S. V.; Berr, C. E.; Edwards, W. M.; Olivier, K. L. Aromatic Polypyromellitimides from Aromatic Polyamic Acids. J. Polym. Sci. Part Polym. Chem. 1996, 34 (11), 2069–2086. https://doi.org/10.1002/pola.1996.846. (25) Feng, J.; Wang, X.; Jiang, Y.; Du, D.; Feng, J. Study on Thermal Conductivities of Aromatic Polyimide Aerogels. ACS Appl. Mater. Interfaces 2016, 8 (20), 12992–12996. https://doi.org/10.1021/acsami.6b02183. (26) M, E. W.; Ivan, M. R. Polyimides of Pyromellitic Acid. US2710853 A, June 14, 1955. (27) Leventis, N.; Sotiriou-Leventis, C.; Mohite, D. P.; Larimore, Z. J.; Mang, J. T.; Churu, G.; Lu, H. Polyimide Aerogels by Ring-Opening Metathesis Polymerization (ROMP). Chem. Mater. 2011, 23 (8), 2250–2261. https://doi.org/10.1021/cm200323e. (28) Ghosh, M.; Mittal, K. L. Synthesis. In Polyimides: Fundamentals and Applications; Marcel Dekker Inc.: New York, 1996; pp 7–187. (29) Buncel, E.; Stairs, R. A. Dipolar Aprotic Solvents. In Solvent Effects in Chemistry; John Wiley & Sons, Inc, 2015; pp 140–154. https://doi.org/10.1002/9781119044307.ch6. (30) Miller, J.; Parker, A. J. Dipolar Aprotic Solvents in Bimolecular Aromatic Nucleophilic Substitution Reactions1. J. Am. Chem. Soc. 1961, 83 (1), 117–123. (31) Solomin, V.; Kardash, I.; SNAGOVSKII, I.; Messerle, P.; Zhubanov, B.; Pravednikov, A. FEATURES OF KINETICS OF AMINE ACYLATION BY ANHYDRIDES OF AROMATIC AND ALICYCLIC CARBOXYLIC-ACIDS. Dokl. Akad. Nauk SSSR 1977, 236 (1), 139–142. (32) Kaas, R. L. Autocatalysis and Equilibrium in Polyimide Synthesis. J. Polym. Sci. Polym. Chem. Ed. 1981, 19 (9), 2255–2267. https://doi.org/10.1002/pol.1981.170190911. (33) Guo, H.; Meador, M. A. B.; McCorkle, L.; Quade, D. J.; Guo, J.; Hamilton, B.; Cakmak, M.; Sprowl, G. Polyimide Aerogels Cross-Linked through Amine Functionalized

202

Polyoligomeric Silsesquioxane. ACS Appl. Mater. Interfaces 2011, 3 (2), 546–552. https://doi.org/10.1021/am101123h. (34) Guo, H.; Meador, M. A. B.; McCorkle, L.; Quade, D. J.; Guo, J.; Hamilton, B.; Cakmak, M. Tailoring Properties of Cross-Linked Polyimide Aerogels for Better Moisture Resistance, Flexibility, and Strength. ACS Appl. Mater. Interfaces 2012, 4 (10), 5422– 5429. https://doi.org/10.1021/am301347a. (35) Kawagishi, K.; Saito, H.; Furukawa, H.; Horie, K. Superior Nanoporous Polyimides via Supercritical CO2 Drying of Jungle-Gym-Type Polyimide Gels. Macromol. Rapid Commun. 2007, 28 (1), 96–100. https://doi.org/10.1002/marc.200600587. (36) Meador, M. A. B.; Malow, E. J.; Silva, R.; Wright, S.; Quade, D.; Vivod, S. L.; Guo, H.; Guo, J.; Cakmak, M. Mechanically Strong, Flexible Polyimide Aerogels Cross-Linked with Aromatic Triamine. ACS Appl. Mater. Interfaces 2012, 4 (2), 536–544. https://doi.org/10.1021/am2014635. (37) Shen, D.; Liu, J.; Yang, H.; Yang, S. Intrinsically Highly Hydrophobic Semi-Alicyclic Fluorinated Polyimide Aerogel with Ultralow Dielectric Constants. Chem. Lett. 2013, 42 (10), 1230–1232. https://doi.org/10.1246/cl.130623. (38) Meador, M. A. B.; Alemán, C. R.; Hanson, K.; Ramirez, N.; Vivod, S. L.; Wilmoth, N.; McCorkle, L. Polyimide Aerogels with Amide Cross-Links: A Low Cost Alternative for Mechanically Strong Polymer Aerogels. ACS Appl. Mater. Interfaces 2015, 7 (2), 1240– 1249. https://doi.org/10.1021/am507268c. (39) Vivod, S.; Meador, M. A. B.; Cakmak, M.; Guo, H. Carbon Nanofiber/Polyimide Aerogel Thin Film Composites. Polym Prep 2011, 52 (2), 127–128. (40) Nguyen, B. N.; Cudjoe, E.; Douglas, A.; Scheiman, D.; McCorkle, L.; Meador, M. A. B.; Rowan, S. J. Polyimide Cellulose Nanocrystal Composite Aerogels. Macromolecules 2016, 49 (5), 1692–1703. https://doi.org/10.1021/acs.macromol.5b01573. (41) Wu, W.; Wang, K.; Zhan, M.-S. Preparation and Performance of Polyimide-Reinforced Clay Aerogel Composites. Ind. Eng. Chem. Res. 2012, 51 (39), 12821–12826. https://doi.org/10.1021/ie301622s. (42) Meador, M. A. B.; Wright, S.; Sandberg, A.; Nguyen, B. N.; Van Keuls, F. W.; Mueller, C. H.; Rodríguez-Solís, R.; Miranda, F. A. Low Dielectric Polyimide Aerogels As Substrates for Lightweight Patch Antennas. ACS Appl. Mater. Interfaces 2012, 4 (11), 6346–6353. https://doi.org/10.1021/am301985s. (43) Meador, M. A. B.; McMillon, E.; Sandberg, A.; Barrios, E.; Wilmoth, N. G.; Mueller, C. H.; Miranda, F. A. Dielectric and Other Properties of Polyimide Aerogels Containing Fluorinated Blocks. ACS Appl. Mater. Interfaces 2014, 6 (9), 6062–6068. https://doi.org/10.1021/am405106h. (44) Tan, H.; Moet, A.; Hiltner, A.; Baer, E. Thermoreversible Gelation of Atactic Polystyrene Solutions. Macromolecules 1983, 16 (1), 28–34. https://doi.org/10.1021/ma00235a006. (45) Aubert, J. Isotactic Polystyrene Phase Diagrams and Physical Gelation. Macromolecules 1988, 21 (12), 3468–3473. (46) Schellenberg, J. Syndiotactic Polystyrene: Synthesis, Characterization, Processing, and Applications; John Wiley & Sons, 2009. (47) Albunia, A. R.; Musto, P.; Guerra, G. FTIR Spectra of Pure Helical Crystalline Phases of Syndiotactic Polystyrene. Polymer 2006, 47 (1), 234–242. https://doi.org/10.1016/j.polymer.2005.10.135.

203

(48) Kobayashi, M.; Yoshioka, T.; Imai, M.; Itoh, Y. Structural Ordering on Physical Gelation of Syndiotactic Polystyrene Dispersed in Chloroform Studied by Time-Resolved Measurements of Small Angle Neutron Scattering (SANS) and Infrared Spectroscopy. Macromolecules 1995, 28 (22), 7376–7385. https://doi.org/10.1021/ma00126a015. (49) Daniel, C.; Deluca, M. D.; Guenet, J.-M.; Brûlet, A.; Menelle, A. Thermoreversible Gelation of Syndiotactic Polystyrene in Benzene. Polymer 1996, 37 (7), 1273–1280. https://doi.org/10.1016/0032-3861(96)80854-8. (50) Daniel, C.; Menelle, A.; Brulet, A.; Guenet, J.-M. Thermoreversible Gelation of Syndiotactic Polystyrene in Toluene and Chloroform. Polymer 1997, 38 (16), 4193– 4199. https://doi.org/10.1016/S0032-3861(96)01005-1. (51) Daniel, C.; Avallone, A.; Guerra, G. Syndiotactic Polystyrene Physical Gels: Guest Influence on Structural Order in Molecular Complex Domains and Gel Transparency. Macromolecules 2006, 39 (22), 7578–7582. https://doi.org/10.1021/ma061124k. (52) Roels, T.; Deberdt, F.; Berghmans, H. Solvent Quality and Phase Stability in Syndiotactic Polystyrene-Solvent Systems. Macromolecules 1994, 27 (21), 6216–6220. https://doi.org/10.1021/ma00099a042. (53) Daniel, C.; Mandanis, A.; Guerra, G.; Musto, P. 1,2-Dichloroethane Conformation and Molecular Organization in Syndiotactic Polystyrene Gels. Macromol. Symp. 2001, 166 (1), 109–116. https://doi.org/10.1002/1521-3900(200103)166:1<109::AID- MASY109>3.0.CO;2-Z. (54) Guerra, G.; Vitagliano, V. M.; De Rosa, C.; Petraccone, V.; Corradini, P. Polymorphism in Melt Crystallized Syndiotactic Polystyrene Samples. Macromolecules 1990, 23 (5), 1539–1544. (55) Cheng, S. Z. D. Phase Transitions in Polymers: The Role of Metastable States; Elsevier, 2008. (56) Wang, X.; Zhang, H.; C. Jana, S. Sulfonated Syndiotactic Polystyrene Aerogels: Properties and Applications. J. Mater. Chem. A 2013, 1 (44), 13989–13999. https://doi.org/10.1039/C3TA13099A. (57) Venditto, V.; Pellegrino, M.; Califano, R.; Guerra, G.; Daniel, C.; Ambrosio, L.; Borriello, A. Monolithic Polymeric Aerogels with VOCs Sorbent Nanoporous Crystalline and Water Sorbent Amorphous Phases. ACS Appl. Mater. Interfaces 2015, 7 (2), 1318–1326. https://doi.org/10.1021/am507116e. (58) Wang, X.; Jana, S. C. Synergistic Hybrid Organic–Inorganic Aerogels. ACS Appl. Mater. Interfaces 2013, 5 (13), 6423–6429. https://doi.org/10.1021/am401717s. (59) Daniel, C.; Sannino, D.; Guerra, G. Syndiotactic Polystyrene Aerogels: Adsorption in Amorphous Pores and Absorption in Crystalline Nanocavities. Chem. Mater. 2008, 20 (2), 577–582. https://doi.org/10.1021/cm702475a. (60) Guerra, G.; Mensitieri, G.; Venditto, V.; Reverchon, E.; Daniel, C. Nanoporous and Microporous Manufacts Based on Syndiotactic Polystyrene and Processes for Their Preparation. US20060229373A1, October 12, 2006. (61) Romina Albunia, A.; Minucci, T.; Guerra, G. Ethylene Removal by Sorption from Polymeric Crystalline Frameworks. J. Mater. Chem. 2008, 18 (9), 1046–1050. https://doi.org/10.1039/B715144F. (62) Daniel, C.; Alfano, D.; Venditto, V.; Cardea, S.; Reverchon, E.; Larobina, D.; Mensitieri, G.; Guerra, G. Aerogels with a Microporous Crystalline Host Phase. Adv. Mater. 2005, 17 (12), 1515–1518. https://doi.org/10.1002/adma.200401762.

204

(63) Mensitieri, G.; Venditto, V.; Guerra, G. Polymeric Sensing Films Absorbing Organic Guests into a Nanoporous Host Crystalline Phase. Sens. Actuators B Chem. 2003, 92 (3), 255–261. https://doi.org/10.1016/S0925-4005(03)00273-9. (64) Kim, S. J.; Chase, G.; Jana, S. C. Polymer Aerogels for Efficient Removal of Airborne Nanoparticles. Sep. Purif. Technol. 2015, 156 (Part 2), 803–808. https://doi.org/10.1016/j.seppur.2015.11.005. (65) Kim, S. J.; Raut, P.; Jana, S. C.; Chase, G. Electrostatically Active Polymer Hybrid Aerogels for Airborne Nanoparticle Filtration. ACS Appl. Mater. Interfaces 2017, 9 (7), 6401–6410. https://doi.org/10.1021/acsami.6b14784. (66) Kim, S. J.; Chase, G.; Jana, S. C. The Role of Mesopores in Achieving High Efficiency Airborne Nanoparticle Filtration Using Aerogel Monoliths. Sep. Purif. Technol. 2016, 166, 48–54. https://doi.org/10.1016/j.seppur.2016.04.017. (67) Kim, S. J.; Jana, S. C. Effects of Skin Layers on Air Permeability in Macroporous Polymer Aerogels. Polymer 2017, 126, 432–436. https://doi.org/10.1016/j.polymer.2017.03.039. (68) Klempner, D.; Sendijarevic, V. Polymeric Foams and Foam Technology, 2nd ed.; Hanser Publishers: Munich, 2004. (69) Lee, L. J.; Zeng, C.; Cao, X.; Han, X.; Shen, J.; Xu, G. Polymer Nanocomposite Foams. Compos. Sci. Technol. 2005, 65 (15), 2344–2363. https://doi.org/10.1016/j.compscitech.2005.06.016. (70) Lissant, K. J. The Geometry of High-Internal-Phase-Ratio Emulsions. J. Colloid Interface Sci. 1966, 22 (5), 462–468. https://doi.org/10.1016/0021-9797(66)90091-9. (71) Gibson, L. J.; Ashby, M. F. Cellular Solids: Structure and Properties; Cambridge University Press, 1999. (72) Jahani, D.; Ameli, A.; Jung, P. U.; Barzegari, M. R.; Park, C. B.; Naguib, H. Open-Cell Cavity-Integrated Injection-Molded Acoustic Polypropylene Foams. Mater. Des. 2014, 53, 20–28. https://doi.org/10.1016/j.matdes.2013.06.063. (73) Silverstein, M. S. PolyHIPEs: Recent Advances in Emulsion-Templated Porous Polymers. Prog. Polym. Sci. 2014, 39 (1), 199–234. https://doi.org/10.1016/j.progpolymsci.2013.07.003. (74) Lissant, K. J.; Mayhan, K. G. A Study of Medium and High Internal Phase Ratio Water/Polymer Emulsions. J. Colloid Interface Sci. 1973, 42 (1), 201–208. https://doi.org/10.1016/0021-9797(73)90025-8. (75) Yaron, P. N.; Scott, A. J.; Reynolds, P. A.; Mata, J. P.; White, J. W. High Internal Phase Emulsions under Shear. Co-Surfactancy and Shear Stability. J. Phys. Chem. B 2011, 115 (19), 5775–5784. https://doi.org/10.1021/jp2005919. (76) Zhang, H.; Hardy, G. C.; Rosseinsky, M. J.; Cooper, A. I. Uniform Emulsion-Templated Silica Beads with High Pore Volume and Hierarchical Porosity. Adv. Mater. 2003, 15 (1), 78–81. https://doi.org/10.1002/adma.200390017. (77) Zhang, H.; Hardy, G. C.; Khimyak, Y. Z.; Rosseinsky, M. J.; Cooper, A. I. Synthesis of Hierarchically Porous Silica and Metal Oxide Beads Using Emulsion-Templated Polymer Scaffolds. Chem. Mater. 2004, 16 (22), 4245–4256. https://doi.org/10.1021/cm0492944. (78) Zhao, C.; Danish, E.; Cameron, N. R.; Kataky, R. Emulsion-Templated Porous Materials (PolyHIPEs) for Selective Ion and Molecular Recognition and Transport: Applications in

205

Electrochemical Sensing. J. Mater. Chem. 2007, 17 (23), 2446–2453. https://doi.org/10.1039/B700929A. (79) Cameron, N. R. High Internal Phase Emulsion Templating as a Route to Well-Defined Porous Polymers. Polymer 2005, 46 (5), 1439–1449. https://doi.org/10.1016/j.polymer.2004.11.097. (80) Cameron, N. R.; Sherrington, D. C. Synthesis and Characterization of Poly(Aryl Ether Sulfone) PolyHIPE Materials. Macromolecules 1997, 30 (19), 5860–5869. https://doi.org/10.1021/ma961403f. (81) Mezzenga, R.; Fredrickson, G. H.; Kramer, E. J. Tailoring Morphologies in Polymeric High Internal Phase Emulsions by Selective Solvent Casting. Macromolecules 2003, 36 (12), 4457–4465. https://doi.org/10.1021/ma021196l. (82) Shirshova, N.; Bismarck, A.; Steinke, J. H. G. Ionic Liquids as Internal Phase for Non- Aqueous PolyHIPEs. Macromol. Rapid Commun. 2011, 32 (23), 1899–1904. https://doi.org/10.1002/marc.201100472. (83) Luo, Y.; Wang, A.-N.; Gao, X. Miniemulsion Template Polymerization to Prepare a Sub- Micrometer Porous Polymeric Monolith with an Inter-Connected Structure and Very High Mechanical Strength. Soft Matter 2012, 8 (29), 7547–7551. https://doi.org/10.1039/C2SM25836F. (84) Bancroft, W. D. The Theory of Emulsification, V. J. Phys. Chem. 1912, 17 (6), 501–519. https://doi.org/10.1021/j150141a002. (85) Zhang, S.; Chen, J.; Perchyonok, V. T. Stability of High Internal Phase Emulsions with Sole Cationic Surfactant and Its Tailoring Morphology of Porous Polymers Based on the Emulsions. Polymer 2009, 50 (7), 1723–1731. https://doi.org/10.1016/j.polymer.2008.11.004. (86) Cohen, N.; Samoocha, D. C.; David, D.; Silverstein, M. S. Carbon Nanotubes in Emulsion-Templated Porous Polymers: Polymer Nanoparticles, Sulfonation, and Conductivity. J. Polym. Sci. Part Polym. Chem. 2013, 51 (20), 4369–4377. https://doi.org/10.1002/pola.26851. (87) Williams, J. M.; Wrobleski, D. A. Spatial Distribution of the Phases in Water-in-Oil Emulsions. Open and Closed Microcellular Foams from Cross-Linked Polystyrene. Langmuir 1988, 4 (3), 656–662. https://doi.org/10.1021/la00081a027. (88) Williams, J. M.; Gray, A. J.; Wilkerson, M. H. Emulsion Stability and Rigid Foams from Styrene or Divinylbenzene Water-in-Oil Emulsions. Langmuir 1990, 6 (2), 437–444. https://doi.org/10.1021/la00092a026. (89) Menner, A.; Ikem, V.; Salgueiro, M.; Shaffer, M. S. P.; Bismarck, A. High Internal Phase Emulsion Templates Solely Stabilised by Functionalised Titania Nanoparticles. Chem. Commun. 2007, 0 (41), 4274–4276. https://doi.org/10.1039/B708935J. (90) Ikem, V. O.; Menner, A.; Bismarck, A. High Internal Phase Emulsions Stabilized Solely by Functionalized Silica Particles. Angew. Chem. Int. Ed. 2008, 47 (43), 8277–8279. https://doi.org/10.1002/anie.200802244. (91) Ikem, V. O.; Menner, A.; Bismarck, A. High-Porosity Macroporous Polymers Sythesized from Titania-Particle-Stabilized Medium and High Internal Phase Emulsions. Langmuir 2010, 26 (11), 8836–8841. https://doi.org/10.1021/la9046066. (92) Gurevitch, I.; Silverstein, M. S. Polymerized Pickering HIPEs: Effects of Synthesis Parameters on Porous Structure. J. Polym. Sci. Part Polym. Chem. 2010, 48 (7), 1516– 1525. https://doi.org/10.1002/pola.23911.

206

(93) Ikem, V. O.; Menner, A.; Horozov, T. S.; Bismarck, A. Highly Permeable Macroporous Polymers Synthesized from Pickering Medium and High Internal Phase Emulsion Templates. Adv. Mater. 2010, 22 (32), 3588–3592. https://doi.org/10.1002/adma.201000729. (94) Kimmins, S. D.; Wyman, P.; Cameron, N. R. Photopolymerised Methacrylate-Based Emulsion-Templated Porous Polymers. React. Funct. Polym. 2012, 72 (12), 947–954. https://doi.org/10.1016/j.reactfunctpolym.2012.06.015. (95) Quell, A.; de Bergolis, B.; Drenckhan, W.; Stubenrauch, C. How the Locus of Initiation Influences the Morphology and the Pore Connectivity of a Monodisperse Polymer Foam. Macromolecules 2016, 49 (14), 5059–5067. https://doi.org/10.1021/acs.macromol.6b00494. (96) Lovelady, E.; Kimmins, S. D.; Wu, J.; Cameron, N. R. Preparation of Emulsion- Templated Porous Polymers Using Thiol–Ene and Thiol–Yne Chemistry. Polym. Chem. 2011, 2 (3), 559–562. https://doi.org/10.1039/C0PY00374C. (97) Luo, Y.; Wang, A.-N.; Gao, X. Pushing the Mechanical Strength of PolyHIPEs up to the Theoretical Limit through Living Radical Polymerization. Soft Matter 2012, 8 (6), 1824– 1830. https://doi.org/10.1039/C1SM06756G. (98) Deleuze, H.; Faivre, R.; Herroguez, V. Preparation and Functionalisation of Emulsion- Derived Microcellular Polymeric Foams (PolyHIPEs) by Ring-Opening Metathesis Polymerisation (ROMP)Electronic Supplementary Information (ESI) Available: Experimental Section. See Http://Www.Rsc.Org/Suppdata/Cc/B2/B208832k/. Chem. Commun. 2002, No. 23, 2822–2823. https://doi.org/10.1039/b208832k. (99) David, D.; Silverstein, M. S. Porous Polyurethanes Synthesized within High Internal Phase Emulsions. J. Polym. Sci. Part Polym. Chem. 2009, 47 (21), 5806–5814. https://doi.org/10.1002/pola.23624. (100) Lumelsky, Y.; Zoldan, J.; Levenberg, S.; Silverstein, M. S. Porous Polycaprolactone−Polystyrene Semi-Interpenetrating Polymer Networks Synthesized within High Internal Phase Emulsions. Macromolecules 2008, 41 (4), 1469–1474. https://doi.org/10.1021/ma7027177. (101) Grant, N. C.; Cooper, A. I.; Zhang, H. Uploading and Temperature-Controlled Release of Polymeric Colloids via Hydrophilic Emulsion-Templated Porous Polymers. ACS Appl. Mater. Interfaces 2010, 2 (5), 1400–1406. https://doi.org/10.1021/am100049r. (102) Manley, S. S.; Graeber, N.; Grof, Z.; Menner, A.; Hewitt, G. F.; Stepanek, F.; Bismarck, A. New Insights into the Relationship between Internal Phase Level of Emulsion Templates and Gas–Liquid Permeability of Interconnected Macroporous Polymers. Soft Matter 2009, 5 (23), 4780–4787. https://doi.org/10.1039/B900426B. (103) Tunç, Y.; Gölgelioğlu, Ç.; Hasirci, N.; Ulubayram, K.; Tuncel, A. Acrylic-Based High Internal Phase Emulsion Polymeric Monolith for Capillary Electrochromatography. J. Chromatogr. A 2010, 1217 (10), 1654–1659. https://doi.org/10.1016/j.chroma.2010.01.020. (104) Pulko, I.; Wall, J.; Krajnc, P.; Cameron, N. R. Ultra-High Surface Area Functional Porous Polymers by Emulsion Templating and Hypercrosslinking: Efficient Nucleophilic Catalyst Supports. Chem. – Eur. J. 2010, 16 (8), 2350–2354. https://doi.org/10.1002/chem.200903043.

207

(105) Mert, E. H.; Kaya, M. A.; Yıldırım, H. Preparation and Characterization of Polyester– Glycidyl Methacrylate PolyHIPE Monoliths to Use in Heavy Metal Removal. Des. Monomers Polym. 2012, 15 (2), 113–126. https://doi.org/10.1163/156855511X615001. (106) GRACE†, H. P. Dispersion Phenomena in High Viscosity Immiscible Fluid Systems and Application of Static Mixers as Dispersion Devices in Such Systems. Chem. Eng. Commun. 1982, 14 (3–6), 225–277. https://doi.org/10.1080/00986448208911047. (107) Zhu, P.; Wang, L. Passive and Active Droplet Generation with Microfluidics: A Review. Lab. Chip 2017, 17 (1), 34–75. https://doi.org/10.1039/C6LC01018K. (108) Garstecki, P.; Stone, H. A.; Whitesides, G. M. Mechanism for Flow-Rate Controlled Breakup in Confined Geometries: A Route to Monodisperse Emulsions. Phys. Rev. Lett. 2005, 94 (16), 164501. https://doi.org/10.1103/PhysRevLett.94.164501. (109) Utada, A. S.; Fernandez-Nieves, A.; Stone, H. A.; Weitz, D. A. Dripping to Jetting Transitions in Coflowing Liquid Streams. Phys. Rev. Lett. 2007, 99 (9), 094502. https://doi.org/10.1103/PhysRevLett.99.094502. (110) Suryo, R.; Basaran, O. A. Tip Streaming from a Liquid Drop Forming from a Tube in a Co-Flowing Outer Fluid. Phys. Fluids 2006, 18 (8), 082102. https://doi.org/10.1063/1.2335621. (111) Zhu, P.; Kong, T.; Kang, Z.; Tian, X.; Wang, L. Tip-Multi-Breaking in Capillary Microfluidic Devices. Sci. Rep. 2015, 5, 11102. https://doi.org/10.1038/srep11102. (112) Zagnoni, M.; Anderson, J.; Cooper, J. M. Hysteresis in Multiphase Microfluidics at a T- Junction. Langmuir 2010, 26 (12), 9416–9422. https://doi.org/10.1021/la1004243. (113) Guillot, P.; Colin, A.; Utada, A. S.; Ajdari, A. Stability of a Jet in Confined Pressure- Driven Biphasic Flows at Low Reynolds Numbers. Phys. Rev. Lett. 2007, 99 (10), 104502. https://doi.org/10.1103/PhysRevLett.99.104502. (114) Gordillo, J. M.; Sevilla, A.; Campo-Cortés, F. Global Stability of Stretched Jets: Conditions for the Generation of Monodisperse Micro-Emulsions Using Coflows. J. Fluid Mech. 2014, 738, 335–357. https://doi.org/10.1017/jfm.2013.600. (115) Menech, M. D.; Garstecki, P.; Jousse, F.; Stone, H. A. Transition from Squeezing to Dripping in a Microfluidic T-Shaped Junction. J. Fluid Mech. 2008, 595, 141–161. https://doi.org/10.1017/S002211200700910X. (116) Cubaud, T.; Mason, T. G. Capillary Threads and Viscous Droplets in Square Microchannels. Phys. Fluids 2008, 20 (5), 053302. https://doi.org/10.1063/1.2911716. (117) Castro-Hernández, E.; Gundabala, V.; Fernández-Nieves, A.; Gordillo, J. M. Scaling the Drop Size in Coflow Experiments. New J. Phys. 2009, 11 (7), 075021. https://doi.org/10.1088/1367-2630/11/7/075021. (118) Martz, T. D.; Bardin, D.; Sheeran, P. S.; Lee, A. P.; Dayton, P. A. Microfluidic Generation of Acoustically Active Nanodroplets. Small 2012, 8 (12), 1876–1879. https://doi.org/10.1002/smll.201102418. (119) Jeong, W.-C.; Lim, J.-M.; Choi, J.-H.; Kim, J.-H.; Lee, Y.-J.; Kim, S.-H.; Lee, G.; Kim, J.-D.; Yi, G.-R.; Yang, S.-M. Controlled Generation of Submicron Emulsion Droplets via Highly Stable Tip-Streaming Mode in Microfluidic Devices. Lab. Chip 2012, 12 (8), 1446–1453. https://doi.org/10.1039/C2LC00018K. (120) Gañán-Calvo, A. M.; González-Prieto, R.; Riesco-Chueca, P.; Herrada, M. A.; Flores- Mosquera, M. Focusing Capillary Jets Close to the Continuum Limit. Nat. Phys. 2007, 3 (10), 737–742. https://doi.org/10.1038/nphys710.

208

(121) Tseng, Y.-H.; Prosperetti, A. Local Interfacial Stability near a Zero Vorticity Point. J. Fluid Mech. 2015, 776, 5–36. https://doi.org/10.1017/jfm.2015.246. (122) Marín, A. G.; Campo-Cortés, F.; Gordillo, J. M. Generation of Micron-Sized Drops and Bubbles through Viscous Coflows. Colloids Surf. Physicochem. Eng. Asp. 2009, 344 (1), 2–7. https://doi.org/10.1016/j.colsurfa.2008.09.033. (123) Zhu, P.; Kong, T.; Lei, L.; Tian, X.; Kang, Z.; Wang, L. Droplet Breakup in Expansion- Contraction Microchannels. Sci. Rep. 2016, 6, 21527. https://doi.org/10.1038/srep21527. (124) Anna, S. L. Droplets and Bubbles in Microfluidic Devices. Annu. Rev. Fluid Mech. 2016, 48 (1), 285–309. https://doi.org/10.1146/annurev-fluid-122414-034425. (125) Zhao, C.-X.; Middelberg, A. P. J. Two-Phase Microfluidic Flows. Chem. Eng. Sci. 2011, 66 (7), 1394–1411. https://doi.org/10.1016/j.ces.2010.08.038. (126) Abate, A. R.; Weitz, D. A. Air-Bubble-Triggered Drop Formation in Microfluidics. Lab. Chip 2011, 11 (10), 1713–1716. https://doi.org/10.1039/C1LC20108E. (127) Steegmans, M. L. J.; Schroën, K. G. P. H.; Boom, R. M. Characterization of Emulsification at Flat Microchannel Y Junctions. Langmuir 2009, 25 (6), 3396–3401. https://doi.org/10.1021/la8035852. (128) Frenz, L.; Blouwolff, J.; Griffiths, A. D.; Baret, J.-C. Microfluidic Production of Droplet Pairs. Langmuir 2008, 24 (20), 12073–12076. https://doi.org/10.1021/la801954w. (129) Zheng, B.; Tice, J. D.; Ismagilov, R. F. Formation of Droplets of Alternating Composition in Microfluidic Channels and Applications to Indexing of Concentrations in Droplet-Based Assays. Anal. Chem. 2004, 76 (17), 4977–4982. https://doi.org/10.1021/ac0495743. (130) McDonald, J. C.; Whitesides, G. M. Poly(Dimethylsiloxane) as a Material for Fabricating Microfluidic Devices. Acc. Chem. Res. 2002, 35 (7), 491–499. https://doi.org/10.1021/ar010110q. (131) Harrison, C.; Cabral, J. T.; Stafford, C. M.; Karim, A.; Amis, E. J. A Rapid Prototyping Technique for the Fabrication of Solvent-Resistant Structures. J. Micromechanics Microengineering 2004, 14 (1), 153. https://doi.org/10.1088/0960-1317/14/1/021. (132) Chu, L.-Y.; Utada, A. S.; Shah, R. K.; Kim, J.-W.; Weitz, D. A. Controllable Monodisperse Multiple Emulsions. Angew. Chem. Int. Ed. 2007, 46 (47), 8970–8974. https://doi.org/10.1002/anie.200701358. (133) Gañán-Calvo, A. M. Generation of Steady Liquid Microthreads and Micron-Sized Monodisperse Sprays in Gas Streams. Phys. Rev. Lett. 1998, 80 (2), 285–288. https://doi.org/10.1103/PhysRevLett.80.285. (134) Anna, S. L.; Bontoux, N.; Stone, H. A. Formation of Dispersions Using “Flow Focusing” in Microchannels. Appl. Phys. Lett. 2003, 82 (3), 364–366. https://doi.org/10.1063/1.1537519. (135) Utada, A. S.; Chu, L.-Y.; Fernandez-Nieves, A.; Link, D. R.; Holtze, C.; Weitz, D. A. Dripping, Jetting, Drops, and Wetting: The Magic of Microfluidics. MRS Bull. 2007, 32 (9), 702–708. https://doi.org/10.1557/mrs2007.145. (136) Niu, X.; Zhang, M.; Wu, J.; Wen, W.; Sheng, P. Generation and Manipulation of “Smart” Droplets. Soft Matter 2009, 5 (3), 576–581. https://doi.org/10.1039/B816553J. (137) He, M.; Kuo, J. S.; Chiu, D. T. Electro-Generation of Single Femtoliter- and Picoliter- Volume Aqueous Droplets in Microfluidic Systems. Appl. Phys. Lett. 2005, 87 (3), 031916. https://doi.org/10.1063/1.1997280.

209

(138) He, M.; Kuo, J. S.; Chiu, D. T. Effects of Ultrasmall Orifices on the Electrogeneration of Femtoliter-Volume Aqueous Droplets. Langmuir 2006, 22 (14), 6408–6413. https://doi.org/10.1021/la060259g. (139) Park, S.-Y.; Wu, T.-H.; Chen, Y.; Teitell, M. A.; Chiou, P.-Y. High-Speed Droplet Generation on Demand Driven by Pulse Laser-Induced Cavitation. Lab. Chip 2011, 11 (6), 1010. https://doi.org/10.1039/c0lc00555j. (140) Dendukuri, D.; Tsoi, K.; Hatton, T. A.; Doyle, P. S. Controlled Synthesis of Nonspherical Microparticles Using Microfluidics. Langmuir 2005, 21 (6), 2113–2116. https://doi.org/10.1021/la047368k. (141) Nisisako, T.; Torii, T.; Takahashi, T.; Takizawa, Y. Synthesis of Monodisperse Bicolored Janus Particles with Electrical Anisotropy Using a Microfluidic Co-Flow System. Adv. Mater. 2006, 18 (9), 1152–1156. https://doi.org/10.1002/adma.200502431. (142) Pham, D. T.; Dimov, S. S. Rapid Prototyping and Rapid Tooling—the Key Enablers for Rapid Manufacturing. Proc. Inst. Mech. Eng. Part C J. Mech. Eng. Sci. 2003, 217 (1), 1– 23. https://doi.org/10.1243/095440603762554569. (143) Ambrosi, A.; Pumera, M. 3D-Printing Technologies for Electrochemical Applications. Chem. Soc. Rev. 2016, 45 (10), 2740–2755. https://doi.org/10.1039/C5CS00714C. (144) Song, Y.; Yan, Y.; Zhang, R.; Xu, D.; Wang, F. Manufacture of the Die of an Automobile Deck Part Based on Rapid Prototyping and Rapid Tooling Technology. J. Mater. Process. Technol. 2002, 120 (1), 237–242. https://doi.org/10.1016/S0924- 0136(01)01165-7. (145) Thomas, C. L.; Gaffney, T. M.; Kaza, S.; Lee, C. H. Rapid Prototyping of Large Scale Aerospace Structures. In 1996 IEEE Aerospace Applications Conference. Proceedings; 1996; Vol. 4, pp 219–230 vol.4. https://doi.org/10.1109/AERO.1996.499663. (146) Giannatsis, J.; Dedoussis, V. Additive Fabrication Technologies Applied to Medicine and Health Care: A Review. Int. J. Adv. Manuf. Technol. 2009, 40 (1), 116–127. https://doi.org/10.1007/s00170-007-1308-1. (147) Standard Terminology for Additive Manufacturing Technologies; 10. (148) Guo, N.; Leu, M. C. Additive Manufacturing: Technology, Applications and Research Needs. Front. Mech. Eng. 2013, 8 (3), 215–243. https://doi.org/10.1007/s11465-013- 0248-8. (149) Jacobs, P. F. Rapid Prototyping & Manufacturing: Fundamentals of Stereolithography; Society of Manufacturing Engineers, 1992. (150) Chartier, T.; Chaput, C.; Doreau, F.; Loiseau, M. Stereolithography of Structural Complex Ceramic Parts. J. Mater. Sci. 2002, 37 (15), 3141–3147. https://doi.org/10.1023/A:1016102210277. (151) Chua, C. K.; Leong, K. F.; Lim, C. S. Rapid Prototyping: Principles and Applications (with Companion CD-ROM) Third Edition; World Scientific Publishing Company, 2010. (152) Khalil, S.; Nam, J.; Sun, W. Multi‐nozzle Deposition for Construction of 3D Biopolymer Tissue Scaffolds. Rapid Prototyp. J. 2005, 11 (1), 9–17. https://doi.org/10.1108/13552540510573347. (153) Minas, C.; Carnelli, D.; Tervoort, E.; Studart, A. R. 3D Printing of Emulsions and Foams into Hierarchical Porous Ceramics. Adv. Mater. 2016, 28 (45), 9993–9999. https://doi.org/10.1002/adma.201603390. (154) Smay, J. E.; Gratson, G. M.; Shepherd, R. F.; Cesarano, J.; Lewis, J. A. Directed Colloidal Assembly of 3D Periodic Structures. Adv. Mater. 2002, 14 (18), 1279–1283.

210

https://doi.org/10.1002/1521-4095(20020916)14:18<1279::AID-ADMA1279>3.0.CO;2- A. (155) Hong, S.; Sycks, D.; Chan, H. F.; Lin, S.; Lopez, G. P.; Guilak, F.; Leong, K. W.; Zhao, X. 3D Printing of Highly Stretchable and Tough Hydrogels into Complex, Cellularized Structures. Adv. Mater. 2015, 27 (27), 4035–4040. https://doi.org/10.1002/adma.201501099. (156) Barry, R. A.; Shepherd, R. F.; Hanson, J. N.; Nuzzo, R. G.; Wiltzius, P.; Lewis, J. A. Direct-Write Assembly of 3D Hydrogel Scaffolds for Guided Cell Growth. Adv. Mater. 2009, 21 (23), 2407–2410. https://doi.org/10.1002/adma.200803702. (157) Kumar, S. Selective Laser Sintering: A Qualitative and Objective Approach. JOM 2003, 55 (10), 43–47. https://doi.org/10.1007/s11837-003-0175-y. (158) Levy, G. N.; Schindel, R.; Kruth, J. P. RAPID MANUFACTURING AND RAPID TOOLING WITH LAYER MANUFACTURING (LM) TECHNOLOGIES, STATE OF THE ART AND FUTURE PERSPECTIVES. CIRP Ann. 2003, 52 (2), 589–609. https://doi.org/10.1016/S0007-8506(07)60206-6. (159) Abe, F.; Osakada, K.; Shiomi, M.; Uematsu, K.; Matsumoto, M. The Manufacturing of Hard Tools from Metallic Powders by Selective Laser Melting. J. Mater. Process. Technol. 2001, 111 (1), 210–213. https://doi.org/10.1016/S0924-0136(01)00522-2. (160) Kruth, J. P.; Froyen, L.; Van Vaerenbergh, J.; Mercelis, P.; Rombouts, M.; Lauwers, B. Selective Laser Melting of Iron-Based Powder. J. Mater. Process. Technol. 2004, 149 (1), 616–622. https://doi.org/10.1016/j.jmatprotec.2003.11.051. (161) Lu, L.; Fuh, J. Y. H.; Chen, Z. D.; Leong, C. C.; Wong, Y. S. In Situ Formation of TiC Composite Using Selective Laser Melting. Mater. Res. Bull. 2000, 35 (9), 1555–1561. https://doi.org/10.1016/S0025-5408(00)00339-1. (162) Gasser, A.; Backes, G.; Kelbassa, I.; Weisheit, A.; Wissenbach, K. Laser Additive Manufacturing. Laser Tech. J. 2010, 7 (2), 58–63. https://doi.org/10.1002/latj.201090029. (163) Lewis, G. K.; Schlienger, E. Practical Considerations and Capabilities for Laser Assisted Direct Metal Deposition. Mater. Des. 2000, 21 (4), 417–423. https://doi.org/10.1016/S0261-3069(99)00078-3. (164) Caulfield, B.; McHugh, P. E.; Lohfeld, S. Dependence of Mechanical Properties of Polyamide Components on Build Parameters in the SLS Process. J. Mater. Process. Technol. 2007, 182 (1), 477–488. https://doi.org/10.1016/j.jmatprotec.2006.09.007. (165) Odell, D.; Wright, P. K.; Montero, M.; Roundy, S.; Ahn, S. Anisotropic Material Properties of Fused Deposition Modeling ABS. Rapid Prototyp. J. 2002, 8 (4), 248–257. https://doi.org/10.1108/13552540210441166. (166) Marcus, H.; Barlow, J.; Beaman, J.; Bourell, D.; Agarwala, M. Direct Selective Laser Sintering of Metals. Rapid Prototyp. J. 1995, 1 (1), 26–36. https://doi.org/10.1108/13552549510078113. (167) Kruth, J.-P.; Levy, G.; Klocke, F.; Childs, T. H. C. Consolidation Phenomena in Laser and Powder-Bed Based Layered Manufacturing. CIRP Ann. 2007, 56 (2), 730–759. https://doi.org/10.1016/j.cirp.2007.10.004. (168) Hagedorn, Y.; Meiners, W.; Wissenbach, K.; Wilkes, J. Additive Manufacturing of ZrO2‐Al2O3 Ceramic Components by Selective Laser Melting. Rapid Prototyp. J. 2013, 19 (1), 51–57. https://doi.org/10.1108/13552541311292736.

211

(169) Harrysson, O. L. A.; Cansizoglu, O.; Marcellin-Little, D. J.; Cormier, D. R.; West, H. A. Direct Metal Fabrication of Titanium Implants with Tailored Materials and Mechanical Properties Using Electron Beam Melting Technology. Mater. Sci. Eng. C 2008, 28 (3), 366–373. https://doi.org/10.1016/j.msec.2007.04.022. (170) Bandyopadhyay, A.; Krishna, B. V.; Xue, W.; Bose, S. Application of Laser Engineered Net Shaping (LENS) to Manufacture Porous and Functionally Graded Structures for Load Bearing Implants. J. Mater. Sci. Mater. Med. 2008, 20 (1), 29. https://doi.org/10.1007/s10856-008-3478-2. (171) Peltola, S. M.; Melchels, F. P. W.; Grijpma, D. W.; Kellomäki, M. A Review of Rapid Prototyping Techniques for Tissue Engineering Purposes. Ann. Med. 2008, 40 (4), 268– 280. https://doi.org/10.1080/07853890701881788. (172) Cooke, M. N.; Fisher, J. P.; Dean, D.; Rimnac, C.; Mikos, A. G. Use of Stereolithography to Manufacture Critical-Sized 3D Biodegradable Scaffolds for Bone Ingrowth. J. Biomed. Mater. Res. B Appl. Biomater. 2003, 64B (2), 65–69. https://doi.org/10.1002/jbm.b.10485. (173) Zein, I.; Hutmacher, D. W.; Tan, K. C.; Teoh, S. H. Fused Deposition Modeling of Novel Scaffold Architectures for Tissue Engineering Applications. Biomaterials 2002, 23 (4), 1169–1185. https://doi.org/10.1016/S0142-9612(01)00232-0. (174) Melchels, F. P. W.; Feijen, J.; Grijpma, D. W. A Review on Stereolithography and Its Applications in Biomedical Engineering. Biomaterials 2010, 31 (24), 6121–6130. https://doi.org/10.1016/j.biomaterials.2010.04.050. (175) Kolan, K. C. R.; Leu, M. C.; Hilmas, G. E.; Velez, M. SELECTIVE LASER SINTERING OF 13-93 BIOACTIVE GLASS BONE SCAFFOLDS. 7. (176) Chim, H.; Hutmacher, D. W.; Chou, A. M.; Oliveira, A. L.; Reis, R. L.; Lim, T. C.; Schantz, J.-T. A Comparative Analysis of Scaffold Material Modifications for Load- Bearing Applications in Bone Tissue Engineering. Int. J. Oral Maxillofac. Surg. 2006, 35 (10), 928–934. https://doi.org/10.1016/j.ijom.2006.03.024. (177) Arcaute, K.; Mann, B. K.; Wicker, R. B. Stereolithography of Three-Dimensional Bioactive Poly(Ethylene Glycol) Constructs with Encapsulated Cells. Ann. Biomed. Eng. 2006, 34 (9), 1429–1441. https://doi.org/10.1007/s10439-006-9156-y. (178) Lin, L.; Ju, S.; Cen, L.; Zhang, H.; Hu, Q. Fabrication of Porous β-TCP Scaffolds by Combination of Rapid Prototyping and Freeze Drying Technology. In 7th Asian-Pacific Conference on Medical and Biological Engineering; Peng, Y., Weng, X., Eds.; IFMBE Proceedings; Springer Berlin Heidelberg, 2008; pp 88–91. (179) Yoldas, B. E.; Annen, M. J.; Bostaph, J. Chemical Engineering of Aerogel Morphology Formed under Nonsupercritical Conditions for Thermal Insulation. Chem. Mater. 2000, 12 (8), 2475–2484. https://doi.org/10.1021/cm9903428. (180) Zhai, C.; Jana, S. C. Tuning Porous Networks in Polyimide Aerogels for Airborne Nanoparticle Filtration. ACS Appl. Mater. Interfaces 2017, 9 (35), 30074–30082. https://doi.org/10.1021/acsami.7b09345. (181) Kim, S. J.; Jana, S. C. Effects of Skin Layers on Air Permeability in Macroporous Polymer Aerogels. Polymer 2017, 126, 432–436. https://doi.org/10.1016/j.polymer.2017.03.039. (182) Daniel, C.; Dammer, C.; Guenet, J.-M. On the Definition of Thermoreversible Gels: The Case of Syndiotactic Polystyrene. Polymer 1994, 35 (19), 4243–4246. https://doi.org/10.1016/0032-3861(94)90604-1.

212

(183) Shinko, A.; C. Jana, S.; Ann Meador, M. Crosslinked Polyurea Aerogels with Controlled Porosity. RSC Adv. 2015, 5 (127), 105329–105338. https://doi.org/10.1039/C5RA20788F. (184) Meador, M. A. B.; Malow, E. J.; He, Z. J.; McCorkle, L.; Guo, H.; nauyen, B. Synthesis and Properties of Nanoporous Polyimide Aerogels Having a Covalently Bonded Network Structure. Polym Prep 2010, 51, 265. (185) Gu, S.; Li, Z.; Miyoshi, T.; Jana, S. C. Polybenzoxazine Aerogels with Controllable Pore Structures. RSC Adv. 2015, 5 (34), 26801–26805. https://doi.org/10.1039/C5RA02635K. (186) Venkateswara Rao, A.; Pajonk, G. M.; Parvathy, N. N. Effect of Solvents and Catalysts on Monolithicity and Physical Properties of Silica Aerogels. J. Mater. Sci. 1994, 29 (7), 1807–1817. https://doi.org/10.1007/BF00351300. (187) Leventis, N.; Chidambareswarapattar, C.; Bang, A.; Sotiriou-Leventis, C. Cocoon-in- Web-Like Superhydrophobic Aerogels from Hydrophilic Polyurea and Use in Environmental Remediation. ACS Appl. Mater. Interfaces 2014, 6 (9), 6872–6882. https://doi.org/10.1021/am500685k. (188) Parker, A. J. Protic-Dipolar Aprotic Solvent Effects on Rates of Bimolecular Reactions. Chem. Rev. 1969, 69 (1), 1–32. https://doi.org/10.1021/cr60257a001. (189) Makitra, R. G. Reichardt, C., Solvents and Solvent Effects in Organic Chemistry, Weinheim: Wiley-VCH, 2003, 630 P. Russ. J. Gen. Chem. 2005, 75 (4), 664–664. https://doi.org/10.1007/s11176-005-0294-y. (190) Reichardt, C. Solvents and Solvent Effects: An Introduction. Org. Process Res. Dev. 2007, 11 (1), 105–113. https://doi.org/10.1021/op0680082. (191) Gutmann, V. Empirical Parameters for Donor and Acceptor Properties of Solvents. Electrochimica Acta 1976, 21 (9), 661–670. https://doi.org/10.1016/0013- 4686(76)85034-7. (192) Mayer, U.; Gutmann, V.; Gerger, W. The Acceptor Number — A Quantitative Empirical Parameter for the Electrophilic Properties of Solvents. Monatshefte Für Chem. Chem. Mon. 1975, 106 (6), 1235–1257. https://doi.org/10.1007/BF00913599. (193) Kosower, E. M. The Effect of Solvent on Spectra. I. A New Empirical Measure of Solvent Polarity: Z-Values. J. Am. Chem. Soc. 1958, 80 (13), 3253–3260. https://doi.org/10.1021/ja01546a020. (194) Schmid, R. Effect of Solvent on Chemical Reactions and Reactivity. 110. (195) Fawcett, W. R. Acidity and Basicity Scales for Polar Solvents. J. Phys. Chem. 1993, 97 (37), 9540–9546. https://doi.org/10.1021/j100139a045. (196) Marcus, Y. The Effectivity of Solvents as Electron Pair Donors. J. Solut. Chem. 1984, 13 (9), 599–624. https://doi.org/10.1007/BF00650369. (197) Wilson, D.; Stenzenberger, H. .; Hergenrother, P. M. Polyimides; Chapman and Hall: New York, 1990. (198) Alexander, R.; Ko, E. C. F.; Parker, A. J.; Broxton, T. J. Solvation of Ions. XIV. Protic- Dipolar Aprotic Solvent Effects on Rates of Bimolecular Reactions. Solvent Activity Coefficients of Reactants and Transition States at 25°. J. Am. Chem. Soc. 1968, 90 (19), 5049–5069. https://doi.org/10.1021/ja01021a002. (199) Teo, N.; Jana, S. C. Open Cell Aerogel Foams via Emulsion Templating. Langmuir 2017, 33 (44), 12729–12738. https://doi.org/10.1021/acs.langmuir.7b03139. (200) Standard Test Method for Compressive Properties of Rigid Plastics; 15.

213

(201) Fei, P.; Wood, S. J.; Chen, Y.; Cavicchi, K. A. Maximum Bubble Pressure Rheology of Low Molecular Mass Organogels. Langmuir 2015, 31 (1), 492–498. https://doi.org/10.1021/la503832r. (202) Zimberlin, J. A.; Crosby, A. J. Water Cavitation of Hydrogels. J. Polym. Sci. Part B Polym. Phys. 2010, 48 (13), 1423–1427. https://doi.org/10.1002/polb.21968. (203) Chidambareswarapattar, C.; Larimore, Z.; Sotiriou-Leventis, C.; T. Mang, J.; Leventis, N. One-Step Room-Temperature Synthesis of Fibrous Polyimide Aerogels from Anhydrides and Isocyanates and Conversion to Isomorphic Carbons. J. Mater. Chem. 2010, 20 (43), 9666–9678. https://doi.org/10.1039/C0JM01844A. (204) Li, H.; Gu, S.; Thomas, S.; Liu, T.; Jana, S. C. Investigation of Polybenzoxazine Gelation Using Laser Light Scattering. J. Appl. Polym. Sci. 2018, 135 (3), 45709. https://doi.org/10.1002/app.45709. (205) Swyngedau, S.; Nussinovitch, A.; Peleg, M. Models for the Compressibility of Layered Polymeric Sponges. Polym. Eng. Sci. 1991, 31 (2), 140–144. https://doi.org/10.1002/pen.760310212. (206) Gu, S.; Zhai, C.; Jana, S. C. Aerogel Microparticles from Oil-in-Oil Emulsion Systems. Langmuir 2016, 32 (22), 5637–5645. https://doi.org/10.1021/acs.langmuir.6b01043. (207) Lee, J. K.; Gould, G. L.; Rhine, W. Polyurea Based Aerogel for a High Performance Thermal Insulation Material. J. Sol-Gel Sci. Technol. 2009, 49 (2), 209–220. https://doi.org/10.1007/s10971-008-1861-6. (208) Gu, S.; Jana, S. C. Open Cell Aerogel Foams with Hierarchical Pore Structures. Polymer 2017, 125, 1–9. https://doi.org/10.1016/j.polymer.2017.07.085. (209) Wang, X.; Jana, S. C. Tailoring of Morphology and Surface Properties of Syndiotactic Polystyrene Aerogels. Langmuir 2013, 29 (18), 5589–5598. https://doi.org/10.1021/la400492m. (210) Teo, N.; Gu, Z.; Jana, S. C. Polyimide-Based Aerogel Foams, via Emulsion-Templating. Polymer 2018, 157, 95–102. https://doi.org/10.1016/j.polymer.2018.10.030. (211) García-González, C. A.; Alnaief, M.; Smirnova, I. Polysaccharide-Based Aerogels— Promising Biodegradable Carriers for Drug Delivery Systems. Carbohydr. Polym. 2011, 86 (4), 1425–1438. https://doi.org/10.1016/j.carbpol.2011.06.066. (212) Quignard, F.; Valentin, R.; Renzo, F. D. Aerogel Materials from Marine Polysaccharides. New J. Chem. 2008, 32 (8), 1300–1310. https://doi.org/10.1039/B808218A. (213) Alnaief, M.; Alzaitoun, M. A.; García-González, C. A.; Smirnova, I. Preparation of Biodegradable Nanoporous Microspherical Aerogel Based on Alginate. Carbohydr. Polym. 2011, 84 (3), 1011–1018. https://doi.org/10.1016/j.carbpol.2010.12.060. (214) GRACE†, H. P. Dispersion Phenomena in High Viscosity Immiscible Fluid Systems and Application of Static Mixers as Dispersion Devices in Such Systems. Chem. Eng. Commun. 1982, 14 (3–6), 225–277. https://doi.org/10.1080/00986448208911047. (215) Alnaief, M.; Smirnova, I. In Situ Production of Spherical Aerogel Microparticles. J. Supercrit. Fluids 2011, 55 (3), 1118–1123. https://doi.org/10.1016/j.supflu.2010.10.006. (216) Kwon, J.; Kim, J.; Yoo, T.; Park, D.; Han, H. Preparation and Characterization of Spherical Polyimide Aerogel Microparticles. Macromol. Mater. Eng. 2014, 299 (9), 1081–1088. https://doi.org/10.1002/mame.201400010. (217) Moner-Girona, M.; Roig, A.; Molins, E.; Llibre, J. Sol-Gel Route to Direct Formation of Silica Aerogel Microparticles Using Supercritical Solvents. J. Sol-Gel Sci. Technol. 2003, 26 (1–3), 645–649. https://doi.org/10.1023/A:1020748727348.

214

(218) Zhang, Y.; Wang, J.; Zhang, X. Surfactant-Free Synthesis of Silica Aerogel Microspheres with Hierarchically Porous Structure. J. Colloid Interface Sci. 2018, 515, 1–9. https://doi.org/10.1016/j.jcis.2018.01.010. (219) Cramer, C.; Fischer, P.; Windhab, E. J. Drop Formation in a Co-Flowing Ambient Fluid. Chem. Eng. Sci. 2004, 59 (15), 3045–3058. https://doi.org/10.1016/j.ces.2004.04.006. (220) Christopher, G. F.; Anna, S. L. Microfluidic Methods for Generating Continuous Droplet Streams. J. Phys. Appl. Phys. 2007, 40 (19), R319. https://doi.org/10.1088/0022- 3727/40/19/R01. (221) Vladisavljević, G. T.; Khalid, N.; Neves, M. A.; Kuroiwa, T.; Nakajima, M.; Uemura, K.; Ichikawa, S.; Kobayashi, I. Industrial Lab-on-a-Chip: Design, Applications and Scale-up for Drug Discovery and Delivery. Adv. Drug Deliv. Rev. 2013, 65 (11), 1626– 1663. https://doi.org/10.1016/j.addr.2013.07.017. (222) Lian, M.; Collier, C. P.; Doktycz, M. J.; Retterer, S. T. Monodisperse Alginate Microgel Formation in a Three-Dimensional Microfluidic Droplet Generator. Biomicrofluidics 2012, 6 (4), 044108. https://doi.org/10.1063/1.4765337. (223) Yeh, C.-H.; Chen, Y.-C.; Lin, Y.-C. Generation of Droplets with Different Concentrations Using Gradient-Microfluidic Droplet Generator. Microfluid. Nanofluidics 2011, 11 (3), 245–253. https://doi.org/10.1007/s10404-011-0791-8. (224) Xu, Q.; Nakajima, M. The Generation of Highly Monodisperse Droplets through the Breakup of Hydrodynamically Focused Microthread in a Microfluidic Device. Appl. Phys. Lett. 2004, 85 (17), 3726–3728. https://doi.org/10.1063/1.1812380. (225) Umbanhowar, P. B.; Prasad, V.; Weitz, D. A. Monodisperse Emulsion Generation via Drop Break Off in a Coflowing Stream. Langmuir 2000, 16 (2), 347–351. https://doi.org/10.1021/la990101e. (226) Zhang, J. M.; Li, E. Q.; Thoroddsen, S. T. A Co-Flow-Focusing Monodisperse Microbubble Generator. J. Micromechanics Microengineering 2014, 24 (3), 035008. https://doi.org/10.1088/0960-1317/24/3/035008. (227) Quell, A.; Elsing, J.; Drenckhan, W.; Stubenrauch, C. Monodisperse Polystyrene Foams via Microfluidics - A Novel Templating Route: Monodisperse Polystyrene Foams via Microfluidics. Adv. Eng. Mater. 2015, 17 (5), 604–609. https://doi.org/10.1002/adem.201500040. (228) Crespy, D.; Landfester, K. Making Dry Fertile: A Practical Tour of Non-Aqueous Emulsions and Miniemulsions, Their Preparation and Some Applications. Soft Matter 2011, 7 (23), 11054–11064. https://doi.org/10.1039/C1SM06156A. (229) Marqusee, J. A.; Ross, J. Kinetics of Phase Transitions: Theory of Ostwald Ripening. J. Chem. Phys. 1983, 79 (1), 373–378. https://doi.org/10.1063/1.445532. (230) Li, T.; Zhao, L.; Liu, W.; Xu, J.; Wang, J. Simple and Reusable Off-the-Shelf Microfluidic Devices for the Versatile Generation of Droplets. Lab. Chip 2016, 16 (24), 4718–4724. https://doi.org/10.1039/C6LC00967K. (231) Teo, N.; Jana, S. C. Solvent Effects on Tuning Pore Structures in Polyimide Aerogels. Langmuir 2018, 34 (29), 8581–8590. https://doi.org/10.1021/acs.langmuir.8b01513. (232) Cubaud, T.; Mason, T. G. Capillary Threads and Viscous Droplets in Square Microchannels. Phys. Fluids 2008, 20 (5), 053302. https://doi.org/10.1063/1.2911716. (233) Cheremisinoff, N. P. Liquid Filtration, 2nd ed.; Butterworth-Heinemann, 1998. (234) Murkes, J.; Carlsson, C. G. Crossflow Filtration; John Wiley & Sons, 1988.

215

(235) Yang, Y.; Chiang, K.; Burke, N. Porous Carbon-Supported Catalysts for Energy and Environmental Applications: A Short Review. Catal. Today 2011, 178 (1), 197–205. https://doi.org/10.1016/j.cattod.2011.08.028. (236) Schaefer, D. W. Engineered Porous Materials. MRS Bulletin. 1994, pp 14–17. (237) Férey, G.; Serre, C.; Devic, T.; Maurin, G.; Jobic, H.; L. Llewellyn, P.; Weireld, G. D.; Vimont, A.; Daturi, M.; Chang, J.-S. Why Hybrid Porous Solids Capture Greenhouse Gases? Chem. Soc. Rev. 2011, 40 (2), 550–562. https://doi.org/10.1039/C0CS00040J. (238) Adebajo, M. O.; Frost, R. L.; Kloprogge, J. T.; Carmody, O.; Kokot, S. Porous Materials for Oil Spill Cleanup: A Review of Synthesis and Absorbing Properties. J. Porous Mater. 2003, 10 (3), 159–170. https://doi.org/10.1023/A:1027484117065. (239) Scheirs, J.; Priddy, D. Modern Styrenic Polymers: Polystyrenes and Styrenic Copolymers; John Wiley & Sons, 2003. (240) Mills, N. Polymer Foams Handbook: Engineering and Biomechanics Applications and Design Guide; Butterworth-Heinemann, 2007. (241) M. Jacobs, L. J.; F. Kemmere, M.; F. Keurentjes, J. T. Sustainable Polymer Foaming Using High Pressure Carbon Dioxide: A Review on Fundamentals, Processes and Applications. Green Chem. 2008, 10 (7), 731–738. https://doi.org/10.1039/B801895B. (242) Ashida, K. Polyurethane and Related Foams: Chemistry and Technology; CRC Press, 2006. (243) Lee, S. T.; Park, C. B.; Ramesh, N. S. Polymeric Foams; Taylor & Francis, 2007. (244) Lakes, R. Materials with Structural Hierarchy. Nature 1993, 361, 511–515. https://doi.org/10.1038/361511a0. (245) Wong, L. L. C.; Ikem, V. O.; Menner, A.; Bismarck, A. Macroporous Polymers with Hierarchical Pore Structure from Emulsion Templates Stabilised by Both Particles and Surfactants. Macromol. Rapid Commun. 2011, 32 (19), 1563–1568. https://doi.org/10.1002/marc.201100382. (246) Elsing, J.; Stefanov, T.; D. Gilchrist, M.; Stubenrauch, C. Monodisperse Polystyrene Foams via Polymerization of Foamed Emulsions: Structure and Mechanical Properties. Phys. Chem. Chem. Phys. 2017, 19 (7), 5477–5485. https://doi.org/10.1039/C6CP06612G. (247) Daniel, C.; Dammer, C.; Guenet, J.-M. On the Definition of Thermoreversible Gels: The Case of Syndiotactic Polystyrene. Polymer 1994, 35 (19), 4243–4246. https://doi.org/10.1016/0032-3861(94)90604-1. (248) Yoldas, B. E.; Annen, M. J.; Bostaph, J. Chemical Engineering of Aerogel Morphology Formed under Nonsupercritical Conditions for Thermal Insulation. Chem. Mater. 2000, 12 (8), 2475–2484. https://doi.org/10.1021/cm9903428. (249) Akimov, Y. K. Fields of Application of Aerogels (Review). Instrum. Exp. Tech. 2003, 46 (3), 287–299. (250) Jones, S. M. Aerogel: Space Exploration Applications. J. Sol-Gel Sci. Technol. 2006, 40 (2–3), 351–357. https://doi.org/10.1007/s10971-006-7762-7. (251) Wei, T.-Y.; Lu, S.-Y.; Chang, Y.-C. Transparent, Hydrophobic Composite Aerogels with High Mechanical Strength and Low High-Temperature Thermal Conductivities. J. Phys. Chem. B 2008, 112 (38), 11881–11886. https://doi.org/10.1021/jp804855v. (252) Kanamori, K.; Aizawa, M.; Nakanishi, K.; Hanada, T. Elastic Organic–Inorganic Hybrid Aerogels and Xerogels. J. Sol-Gel Sci. Technol. 2008, 48 (1–2), 172–181. https://doi.org/10.1007/s10971-008-1756-6.

216

(253) Kircher, L.; Theato, P.; Cameron, N. R. Functionalisation of Porous Polymers from High-Internal-Phase Emulsions and Their Applications. In Functional Polymers by Post- Polymerisation Modification: Concepts, Guidelines and Applications; 2013; pp 333–352. (254) Imhof, A.; Pine, J. H. Ordered Macroporous Materials by Emulsion Templating. Nature 1997, 389 (6654), 948–951. (255) Menner, A.; Haibach, K.; Powell, R.; Bismarck, A. Tough Reinforced Open Porous Polymer Foams via Concentrated Emulsion Templating. Polymer 2006, 47 (22), 7628– 7635. https://doi.org/10.1016/j.polymer.2006.09.022. (256) Silverstein, M. S.; Cameron, N. R. PolyHIPEs—Porous Polymers from High Internal Phase Emulsions. Encycl. Polym. Sci. Technol. 2010. (257) Tebboth, M.; Menner, A.; Kogelbauer, A.; Bismarck, A. Polymerised High Internal Phase Emulsions for Fluid Separation Applications. Curr. Opin. Chem. Eng. 2014, 4, 114–120. https://doi.org/10.1016/j.coche.2014.03.001. (258) Bourrel, M.; Salager, J. L.; Schechter, R. S.; Wade, W. H. A Correlation for Phase Behavior of Nonionic Surfactants. J. Colloid Interface Sci. 1980, 75 (2), 451–461. https://doi.org/10.1016/0021-9797(80)90470-1. (259) Mitra, R. K.; Paul, B. K. Effect of NaCl and Temperature on the Water Solubilization Behavior of AOT/Nonionics Mixed Reverse Micellar Systems Stabilized in IPM Oil. Colloids Surf. Physicochem. Eng. Asp. 2005, 255 (1), 165–180. https://doi.org/10.1016/j.colsurfa.2004.12.037. (260) Shinoda, K. Thermodynamic Aspects of Nonionic Surfactant—Water Systems. J. Colloid Interface Sci. 1970, 34 (2), 278–282. https://doi.org/10.1016/0021- 9797(70)90179-7. (261) Gong, L.; Kyriakides, S.; Jang, W.-Y. Compressive Response of Open-Cell Foams. Part I: Morphology and Elastic Properties. Int. J. Solids Struct. 2005, 42 (5–6), 1355–1379. https://doi.org/10.1016/j.ijsolstr.2004.07.023. (262) Barbetta, A.; Cameron, N. R. Morphology and Surface Area of Emulsion-Derived (PolyHIPE) Solid Foams Prepared with Oil-Phase Soluble Porogenic Solvents: Span 80 as Surfactant. Macromolecules 2004, 37 (9), 3188–3201. https://doi.org/10.1021/ma0359436. (263) Barbetta, A.; Cameron, N. R.; Cooper, S. J. High Internal Phase Emulsions (HIPEs) Containing Divinylbenzene and 4-Vinylbenzyl Chloride and the Morphology of the Resulting PolyHIPE Materials. Chem. Commun. 2000, 0 (3), 221–222. https://doi.org/10.1039/A909060F. (264) Cameron, N. R.; Barbetta, A. The Influence of Porogen Type on the Porosity,Surface Area and Morphology of Poly(Divinylbenzene) PolyHIPE Foams. J. Mater. Chem. 2000, 10 (11), 2466–2471. https://doi.org/10.1039/B003596N. (265) Williams, J. M.; Wrobleski, D. A. Spatial Distribution of the Phases in Water-in-Oil Emulsions. Open and Closed Microcellular Foams from Cross-Linked Polystyrene. Langmuir 1988, 4 (3), 656–662. https://doi.org/10.1021/la00081a027. (266) Quell, A.; Sottmann, T.; Stubenrauch, C. Diving into the Finestructure of Macroporous Polymer Foams Synthesized via Emulsion Templating: A Phase Diagram Study. Langmuir 2017, 33 (2), 537–542. https://doi.org/10.1021/acs.langmuir.6b03762. (267) Cameron, N. R.; Sherrington, D. C.; Albiston, L.; Gregory, D. P. Study of the Formation of the Open-Cellular Morphology of Poly(Styrene/Divinylbenzene) PolyHIPE Materials

217

by Cryo-SEM. Colloid Polym. Sci. 1996, 274 (6), 592–595. https://doi.org/10.1007/BF00655236. (268) Jou, F.-Y.; Mather, A. E. Liquid−Liquid Equilibria for Binary Mixtures of Water + Benzene, Water + Toluene, and Water + p-Xylene from 273 K to 458 K. J. Chem. Eng. Data 2003, 48 (3), 750–752. https://doi.org/10.1021/je034033g. (269) Leser, M. E.; Acquistapace, S.; Cagna, A.; Makievski, A. V.; Miller, R. Limits of Oscillation Frequencies in Drop and Bubble Shape Tensiometry. Colloids Surf. Physicochem. Eng. Asp. 2005, 261 (1), 25–28. https://doi.org/10.1016/j.colsurfa.2004.11.043. (270) Freer, E. M.; Svitova, T.; Radke, C. J. The Role of Interfacial Rheology in Reservoir Mixed Wettability. J. Pet. Sci. Eng. 2003, 39 (1), 137–158. https://doi.org/10.1016/S0920-4105(03)00045-7. (271) Santini, E.; Ravera, F.; Ferrari, M.; Stubenrauch, C.; Makievski, A.; Krägel, J. A Surface Rheological Study of Non-Ionic Surfactants at the Water–Air Interface and the Stability of the Corresponding Thin Foam Films. Colloids Surf. Physicochem. Eng. Asp. 2007, 298 (1), 12–21. https://doi.org/10.1016/j.colsurfa.2006.12.004. (272) Stubenrauch, C.; Miller, R. Stability of Foam Films and Surface Rheology: An Oscillating Bubble Study at Low Frequencies. J. Phys. Chem. B 2004, 108 (20), 6412– 6421. https://doi.org/10.1021/jp049694e. (273) Sroog, C. E.; Endrey, A. L.; Abramo, S. V.; Berr, C. E.; Edwards, W. M.; Olivier, K. L. Aromatic Polypyromellitimides from Aromatic Polyamic Acids. J. Polym. Sci. A 1965, 3 (4), 1373–1390. https://doi.org/10.1002/pol.1965.100030410. (274) Guo, J.; N. Nguyen, B.; Li, L.; B. Meador, M. A.; A. Scheiman, D.; Cakmak, M. Clay Reinforced Polyimide/Silica Hybrid Aerogel. J. Mater. Chem. A 2013, 1 (24), 7211– 7221. https://doi.org/10.1039/C3TA00439B. (275) Batrakova, E.; Lee, S.; Li, S.; Venne, A.; Alakhov, V.; Kabanov, A. Fundamental Relationships Between the Composition of Pluronic Block Copolymers and Their Hypersensitization Effect in MDR Cancer Cells. Pharm. Res. 1999, 16 (9), 1373–1379. https://doi.org/10.1023/A:1018942823676. (276) Crespy, D.; Landfester, K. Making Dry Fertile: A Practical Tour of Non-Aqueous Emulsions and Miniemulsions, Their Preparation and Some Applications. Soft Matter 2011, 7 (23), 11054–11064. https://doi.org/10.1039/C1SM06156A. (277) Nakanishi, K.; Soga, N. Phase Separation in Silica Sol-Gel System Containing Polyacrylic Acid I. Gel Formaation Behavior and Effect of Solvent Composition. J. Non Cryst. Solids 1992, 139, 1–13. https://doi.org/10.1016/S0022-3093(05)80800-2. (278) Talley, S. J.; Yuan, X.; Moore, R. B. Thermoreversible Gelation of Poly(Ether Ether Ketone). ACS Macro Lett. 2017, 6 (3), 262–266. https://doi.org/10.1021/acsmacrolett.7b00109. (279) Chidambareswarapattar, C.; McCarver, P. M.; Luo, H.; Lu, H.; Sotiriou-Leventis, C.; Leventis, N. Fractal Multiscale Nanoporous Polyurethanes: Flexible to Extremely Rigid Aerogels from Multifunctional Small Molecules. Chem. Mater. 2013, 25 (15), 3205– 3224. https://doi.org/10.1021/cm401623h. (280) Hollister, S. J. Porous Scaffold Design for Tissue Engineering. Nat. Mater. 2005, 4 (7), 518–524. https://doi.org/10.1038/nmat1421. (281) Yoo, D. New Paradigms in Hierarchical Porous Scaffold Design for Tissue Engineering. Mater. Sci. Eng. C 2013, 33 (3), 1759–1772. https://doi.org/10.1016/j.msec.2012.12.092.

218

(282) Butscher, A.; Bohner, M.; Hofmann, S.; Gauckler, L.; Müller, R. Structural and Material Approaches to Bone Tissue Engineering in Powder-Based Three-Dimensional Printing. Acta Biomater. 2011, 7 (3), 907–920. https://doi.org/10.1016/j.actbio.2010.09.039. (283) Cox, S. C.; Thornby, J. A.; Gibbons, G. J.; Williams, M. A.; Mallick, K. K. 3D Printing of Porous Hydroxyapatite Scaffolds Intended for Use in Bone Tissue Engineering Applications. Mater. Sci. Eng. C 2015, 47, 237–247. https://doi.org/10.1016/j.msec.2014.11.024. (284) Wu, C.; Luo, Y.; Cuniberti, G.; Xiao, Y.; Gelinsky, M. Three-Dimensional Printing of Hierarchical and Tough Mesoporous Bioactive Glass Scaffolds with a Controllable Pore Architecture, Excellent Mechanical Strength and Mineralization Ability. Acta Biomater. 2011, 7 (6), 2644–2650. https://doi.org/10.1016/j.actbio.2011.03.009. (285) Liska, R.; Schwager, F.; Maier, C.; Cano‐Vives, R.; Stampfl, J. Water-Soluble Photopolymers for Rapid Prototyping of Cellular Materials. J. Appl. Polym. Sci. 2005, 97 (6), 2286–2298. https://doi.org/10.1002/app.22025. (286) Park, J. H.; Jung, J. W.; Kang, H.-W.; Cho, D.-W. Indirect Three-Dimensional Printing of Synthetic Polymer Scaffold Based on Thermal Molding Process. Biofabrication 2014, 6 (2), 025003. https://doi.org/10.1088/1758-5082/6/2/025003. (287) Jiang, Y.; Wang, Q. Highly-Stretchable 3D-Architected Mechanical Metamaterials. Sci. Rep. 2016, 6, 34147. https://doi.org/10.1038/srep34147. (288) A. Zadpoor, A. Mechanical Meta-Materials. Mater. Horiz. 2016, 3 (5), 371–381. https://doi.org/10.1039/C6MH00065G. (289) Christensen, J.; Kadic, M.; Kraft, O.; Wegener, M. Vibrant Times for Mechanical Metamaterials. MRS Commun. 2015, 5 (3), 453–462. https://doi.org/10.1557/mrc.2015.51. (290) Zheng, X.; Lee, H.; Weisgraber, T. H.; Shusteff, M.; DeOtte, J.; Duoss, E. B.; Kuntz, J. D.; Biener, M. M.; Ge, Q.; Jackson, J. A.; et al. Ultralight, Ultrastiff Mechanical Metamaterials. Science 2014, 344 (6190), 1373–1377. https://doi.org/10.1126/science.1252291. (291) Schoen, A. Infinite Periodic Minimal Surfaces Without Self-Intersections; Technical Note TN D-5541; NASA: Washington, D.C., 1970; pp 1–91. (292) Cignoni, P.; Callieri, M.; Corsini, M.; Dellepiane, M.; Ganovelli, F.; Ranzuglia, G. MeshLab: An Open-Source Mesh Processing Tool. Eurographics Ital. Chapter Conf. 2008, 8. (293) Harris, F. W. Synthesis of Aromatic Polyimides from Dianhydrides and Diamines. In Polyimides; Chapman and Hall: New York, 1990. (294) Fan, H.; Hartshorn, C.; Buchheit, T.; Tallant, D.; Assink, R.; Simpson, R.; Kissel, D. J.; Lacks, D. J.; Torquato, S.; Brinker, C. J. Modulus–Density Scaling Behaviour and Framework Architecture of Nanoporous Self-Assembled Silicas. Nat. Mater. 2007, 6 (6), 418–423. https://doi.org/10.1038/nmat1913. (295) Qin, Z.; Jung, G. S.; Kang, M. J.; Buehler, M. J. The Mechanics and Design of a Lightweight Three-Dimensional Graphene Assembly. Sci. Adv. 2017, 3 (1), e1601536. https://doi.org/10.1126/sciadv.1601536. (296) Kolpakov, A. G. Determination of the Average Characteristics of Elastic Frameworks. J. Appl. Math. Mech. 1985, 49 (6), 739–745. https://doi.org/10.1016/0021-8928(85)90011- 5.

219

(297) Lakes, R. Advances in Negative Poisson’s Ratio Materials. Adv. Mater. 1993, 5 (4), 293– 296. https://doi.org/10.1002/adma.19930050416. (298) Evans, K. E. Auxetic Polymers: A New Range of Materials. Endeavour 1991, 15 (4), 170–174. https://doi.org/10.1016/0160-9327(91)90123-S.

220