PROCESSING AND CHARACTERIZATION OF POLYCARBONATE FOAMS WITH SUPERCRITICAL

CO2 AND 5-PHENYL-1H-TETRAZOLE

Thomas Cloarec, B.S.

Thesis Prepared for the Degree of

MASTER OF SCIENCE

UNIVERSITY OF NORTH TEXAS

May 2015

APPROVED:

Witold Brostow, Major Professor

Samir Aouadi, Committee Member

Nandika A. D’Souza, Committee Member

Richard F. Reidy, Committee Member Cloarec, Thomas. Processing and Characterization of Polycarbonate Foams with

Supercritical Co2 and 5-Phenyl-1h-Tetrazole. Master of Science (Materials Science and

Engineering), May 2015, 100 pp., 11 tables, 59 figures, references, 5 titles.

Since their discovery in the 1930s, polymeric foams have been widely used in the industry for a variety of applications such as acoustical and thermal insulation, filters, absorbents etc. The reason for this ascending trend can be attributed to factors such as cost, ease of processing and a high strength to weight ratio compared to non-foamed polymers.

The purpose of this project was to develop an “indestructible” material made of polycarbonate (PC) for industrial applications. Due to the high price of polycarbonate, two foaming methods were investigated to reduce the amount of material used. Samples were foamed physically in supercritical CO2 or chemically with 5-phenyl-1H-tetrazole.

After thermal characterization of the foams in differential scanning calorimetry (DSC), we saw that none of the foaming methods had an influence on the glass transition of polycarbonate. Micrographs taken in scanning electron microscopy (SEM) showed that foams obtained in physical and chemical foaming had different structures. Indeed, samples foamed in supercritical CO2 exhibited a microcellular opened-cell structure with a high cell density and a homogeneous cell distribution. On the other hand, samples foamed with 5-phenyl-1H- tetrazole had a macrocellular closed-cell structure with a much smaller cell density and a random cell distribution. Compression testing showed that polycarbonate foamed physically had a compression modulus a lot greater.

Then, XLPE mesh 35 or 50 and wollastonite were added to the polymeric matrices to enhance the foaming process and the mechanical properties. DSC experiments showed that the addition of fillers changed the thermal properties of polycarbonate for both foaming methods by inducing a shift in glass transition. SEM revealed that fillers lowered the average cell diameter and increased the cell density. This phenomenon increased the compression modulus for polycarbonate foamed in supercritical CO2. However, mechanical properties decreased for samples foamed with 5-phenyl-1H-tetrazole due to their relative brittleness and the propagation of microcracks. Copyright 2015

by

Thomas Cloarec

ii ACKNOWLEDGMENTS

First, I would like to express my sincere gratitude to my adviser, Dr. Witold Brostow, who gave me the opportunity to investigate polymers during my studies toward a Master degree in Materials Science and Engineering. His guidance, knowledge and support through funding and tuition waivers were with no doubt an immeasurable chance that made my research a once in a lifetime experience.

Next, I would like to thank my colleague and friend Zachary Hoyt and Dr. Richard F.

Reidy for their support, help and many discussions throughout my research.

Finally, I would like to thank and acknowledge Dr. Nandika A. D’Souza, Dr. Marcus

Young and the Center for Advanced Research & Technology (CART) from the University of

North Texas for allowing me use their equipment, without which I could not have done my research.

Most importantly, I would like to thank my family, my lovely fiancé Hollye Cody and my sweet little Finn for their love, encouragement and support during my education.

iii TABLE OF CONTENTS

ACKNOWLEDGEMENTS…………………………………………………………………………………………………………3

LIST OF TABLES……………………………………………………………………………………………………………………..7

LIST OF FIGURES……………………………………………………………………………………………………………………8

Chapter 1………………………………………………………………………………………………………………………….…12

INTRODUCTION……………………………………………………………………………………………………….12

1.1 Polymeric foams………………………………………………………………………………………………..12

1.2 Polycarbonate foams…………………………………………………………………………………………13

1.3 Scope………………………………………………………………………………………………………………..14

1.4 References………………………………………………………………………………………………………..16

Chapter 2……………………………………………………………………………………………………………………………17

PROCESSING AND CHARACTERIZATION OF POLYCARBONATE FOAMS WITH SUPERCRITICAL

CO2 ……………………………..………………………………………………………………………………………………….17

2.1 Literature review……………………………………………………………………………………………….17

2.1.1 Supercritical fluids……………………………………………………………………………….17

2.1.2 Physical foaming via supercritical CO2 ……………………………………………….. 20

2.2 Materials for supercritical foaming in CO2 ……………………………………………………….. 25

2.3 Experimental……………………………………………………………………………………………………..25

2.3.1 Samples preparation……………………………………………………………………………25

2.3.2 Foaming process………………………………………………………………………………….25

2.3.3 Characterization methods…………………………………………………………………...27

iv 2.4 Results and discussion……………………………………………………………………………………….29

2.4.1 Differential Scanning Calorimetry (DSC)………………………………………………29

2.4.2 Scanning Electron Microscopy (SEM)…………………………………………………..33

2.4.3 Compression testing……………………………………………………………………………39

2.5 Conclusions……………………………………………………………………………………………………….42

2.6 References………………………………………………………………………………………………………..43

Chapter 3…………………………………………………………………………………………………………………………...45

PROCESSING AND CHARACTERIZATION OF POLYCARBONATE FOAMS WITH CHEMICAL

FOAMING AGENT 5-PHENYL-1H-TETRAZOLE……………………………………………………………45

3.1 Literature review………………………………………………………………………………………………45

3.1.1 Chemical foaming agents……………………………………………………………………45

3.1.2 Foaming via extrusion………………………………………………………………………..46

3.1.3 Foaming via injection molding……………………………………………………………48

3.2 Materials for chemical foaming in with 5-phenyl-1H-tetrazole…………………………50

3.3 Experimental…………………………………………………………………………………………..…….…51

3.3.1 Samples preparation…………………………………………………………………..….….51

3.3.2 Foaming process…………………………………………………………………………...…..52

3.3.3 Characterization methods…………………………………………………..……………..53

3.4 Results and discussion……………………………………………………………………………………...54

3.4.1 Differential Scanning Calorimetry (DSC)……………………………………………..54

3.4.2 Scanning Electron Microscopy (SEM)………………………………………………….55

3.4.3 X-ray tomography micro-CT………………………………………………………………..63

3.4.4 Compression testing……………………………………………………………….……….….65

3.5 Conclusions……………………………………………………………………………………………………….68

v 3.6 References……………………………………………………………………………………………………….69

Chapter 4………………………………………………………………………………………………………………………….71

PROCESSING AND CHARACTERIZATION OF POLYCARBONATE FOAMS WITH FILLERS………..71

4.1 Literature review……..…………………………………………………………………………………….71

4.2 Materials for physical and chemical foaming………………………………………………….72

4.3 Experimental………………………………………………………………………………………………….72

4.3.1 Samples preparation and foaming………………………………………………..…72

4.3.2 Characterization methods……………………………………………………………….73

4.4 Results and discussion…………….……………………………………………………………………..76

4.4.1 Fillers characterization……………………………………………………………………76

4.4.2 Polycarbonate mixes foamed in supercritical CO2 ………………………….82

4.4.3 Polycarbonate mixes foamed chemically with 5-phenyl-1H-

tetrazole…………………………………………………………………………………………………96

4.5 Conclusions……….…………………………………………………………………………………………107

4.6 References……………………………………………………………………………………………………108

Chapter 5…………………………………………………………………………………………………………………….109

CONCLUSIONS……………………………………………………………………………………………………………..109

vi LIST OF TABLES

Table 1: General properties of supercritical fluid, gas and liquid…………………………………………18

Table 2: Critical point of common solvents………………………………………………………………………….18

Table 3: Saturated solubility and diffusion coefficient of CO2 in PC for various temperatures and pressures…………………………………..…………………………………………………………………………………23

Table 4: Overall properties of PC foamed in supercritical CO2 …………………………………………….34

Table 5: Density and porosity for PC foamed in supercritical CO2 ……………………………………….38

Table 6: Percentage as a function of 5-phenyl-1H-terazole………………………………………………….52

Table 7: Overall properties for PC foamed with 5-phenyl-1H-terazole…………………………………61

Table 8: Porosity of chemically foamed samples obtained in micro-CT……………………………….64

Table 9: Overall properties for PC mixes foamed in supercritical CO2 …………………………………91

Table 10: Density and porosity for PC mixes foamed in supercritical CO2 …………………………..93

Table 11: Overall properties for PC mixes foamed chemically……..……………………………………103

vii LIST OF FIGURES

Figure 1: Chemical reaction for the synthesis of polycarbonate…………………………………………..13

Figure 2: General phase diagram for a supercritical fluid…………………………………………………….19

Figure 3: Phase diagram of CO2 ………………………………………………………………………………………….20

Figure 4: DSC curves for samples foamed in supercritical CO2 …………………………………………….30

Figure 5: DSC curves for different saturation times in supercritical CO2 after depressurization ……………………………………………………………………………………………………………………………………………32

Figure 6: DSC curves for different saturation times in supercritical CO2 after erasing the stress and the thermal history…………………………………………………………………………………………….32

Figure 7: SEM micrographs for PC foamed in supercritical CO2 for 24 h with magnification 100x (A), 300x (B) and 1000x (C)………………………………………………………………………………………….33

Figure 8: SEM micrographs for PC foamed in supercritical CO2 for 48 h with magnification 100x (A), 300x (B) and 1000x (C)………………………………………………………………………………………….35

Figure 9: SEM micrographs for PC foamed in supercritical CO2 for 72 h with magnification 100x (A), 300x (B) and 1000x (C)………………………………………………………………………………………….36

Figure 10: Cell diameter distribution for sample foamed 48 h in supercritical CO2 ...... 36

Figure 11: Cell diameter distribution for sample foamed 72 h in supercritical CO2 ……………..37

Figure 12: Volume for samples foamed in supercritical CO2 ……………………………………………..37

Figure 13: Compressive testing strain-stress curves for solid polymers and PC foamed in supercritical CO2 …………………………………………………………………………………………………………………40

Figure 14: Compression modulus for solid polymers and PC foamed in supercritical CO2 ……41

Figure 15: Schematic illustration of an extruder………………………………………………………………….47

Figure 16: Schematic illustration of an injection molding machine………………………………………49

Figure 17: Schematic illustration of an injection mold…………………………………………………………49

Figure 18: Schematic structure of 5-phenyl-1H-tetrazole…………………………………………...50

viii Figure 19: TGA curve for 5-phenyl-1H-terazole……………………………………………………………………51

Figure 20: DSC curves for samples foamed with 5-phenyl-1H-terazole………………………………..55

Figure 21: SEM micrographs for PC foamed chemically and quenched in water with 3 % (A, C) and 5 % (B, D) of 5-phenyl-1H-terazole with magnification 50x (A, B) and 150x (C, D)…………57

Figure 22: Cell size distribution for sample foamed with 3 % 5-phenyl-1H-terazole and quenched in water………………………………………………………………………………………………………………58

Figure 23: Cell size distribution for sample foamed with 5 % 5-phenyl-1H-terazole and quenched in water………………………………………………………………………………………………………………58

Figure 24: SEM micrographs for PC foamed chemically and quenched in liquid nitrogen with 3 % (A, C) and 5 % (B, D) of 5-phenyl-1H-terazole with magnification 50x (A, B) and 150x (C, D)………………………………………………………………………………………………………………………………………..60

Figure 25: Cell size distribution for sample foamed with 3 % 5-phenyl-1H-terazole and quenched in liquid nitrogen………………………………………………………………………………………………..60

Figure 26: Cell size distribution for sample foamed with 5 % 5-phenyl-1H-terazole and quenched in liquid nitrogen…………………………………………………………………………………………………61

Figure 27: Volume for samples foamed with 5-phenyl-1H-tetrazole…………………………………62

Figure 28: Three-dimensional models for samples foamed chemically and quenched in water with 3 % (A, B) and 5 % (C, D) of 5-phenyl-1H-tetrazole………………………………………………………64

Figure 29: Three-dimensional models for samples foamed chemically and quenched in liquid nitrogen with 3 % (A, B) and 5 % (C, D) of 5-phenyl-1H-tetrazole………………………………………..65

Figure 30: Compressive testing strain-stress curves for solid polymers and PC foamed in supercritical CO2 or 5-phenyl-1H-tetrazole………………………………………………………………………….66

Figure 31: Compression modulus for solid polymers and PC foamed in supercritical CO2 or 5- phenyl-1H-tetrazole…………………………………………………………………………………………………………….67

Figure 32: FTIR curves for XLPE mesh 35 and 50………………………………………………………………….77

Figure 33: FTIR curve for wollastonite…………………………………………………………………………………77

Figure 34: DSC curves for wollastonite and XLPE mesh 35 and 50……………………………………….79

Figure 35: DSC curves for unfoamed PC mixes…………………………………………………………………….80

Figure 36: Compressive testing strain-stress curves for solid polymers and PC mixes………….81

ix Figure 37: Compression modulus for solid polymers and unfoamed PC mixes…………………….82

Figure 38: DSC curves for PC mixes with 10-20% of XLPE mesh 35 or 50 and foamed in supercritical CO2 …………………………………………………………………………………………………………………83

Figure 39: DSC for PC mixes with 10 or 20 % of wollastonite foamed in supercritical CO2 ….84

Figure 40: SEM micrographs for PC mixes foamed in supercritical CO2 with 10 % (A, C) and 20 % (B, D) of XLPE mesh 35 with magnification 150x (A, B) and 1000x (C, D)………………………….86

Figure 41: Cell size distribution for PC mixes with 10 % (on the left) or 20 % (on the right) of XLPE mesh 35………………………………………………………………………………………………………………………87

Figure 42: SEM micrographs for PC mixes foamed in supercritical CO2 with 10 % (A, C) and 20 % (B, D) of XLPE mesh 50 with magnification 150x (A, B) and 1000x (C, D)………………………….88

Figure 43: Cell size distribution for PC mixes with 10 % (on the left) or 20 % (on the right) of XLPE mesh 50………………………………………………………………………………………………………………………89

Figure 44: SEM micrographs for PC mixes foamed in supercritical CO2 with 10 % (A, C) and 20 % (B, D) of wollastonite with magnification 150x (A, B) and 1000x (C, D)…………………………….90

Figure 45: Cell size distribution for PC mixes with 10 % (on the left) or 20 % (on the right) of wollastonite………………………………………………………………………………………………………………………..90

Figure 46: Volume for PC mixes foamed in supercritical CO2 ……………………………………………92

Figure 47: Compressive testing strain-stress curves for solid polymers and PC mixes foamed in supercritical CO2 …………………………………………………………………………………………………………….94

Figure 48: Compression modulus for solid polymers and foamed PC mixes in supercritical CO2 …………………………………………………………………………………………………………………………………………………………………………………………………………………..95

Figure 49: DSC curves for PC mixes with 10-20% of XLPE mesh 35 or 50 and foamed with 5- phenyl-1H-tetrazole…………………………………………………………………………………………………………….96

Figure 50: DSC for PC mixes with 10 or 20 % of wollastonite and foamed with 5-phenyl-1H- tetrazole……………………………………………………………………………………………………………………………..97

Figure 51: SEM micrographs for PC mixes foamed chemically with 10 % (A, C) and 20 % (B, D) of XLPE mesh 35 with magnification 30x (A, B) and 150x (C, D)……………………………………………99

Figure 52: Cell size distribution for PC mixes with 10 % (on the left) or 20 % (on the right) of XLPE mesh 35………………………………………………………………………………………………………………………99

Figure 53: SEM micrographs for PC mixes foamed chemically with 10 % (A, C) and 20 % (B, D) of XLPE mesh 50 with magnification 30x (A, B) and 150x (C, D)…………………………………………100

x Figure 54: Cell size distribution for PC mixes with 10 % (on the left) or 20 % (on the right) of XLPE mesh 50…………………………………………………………………………………………………………………….101

Figure 55: SEM micrographs for PC mixes foamed chemically with 10 % (A, C) and 20 % (B, D) of wollastonite with magnification 30x (A, B) and 150x (C, D)……………………………………………102

Figure 56: Cell size distribution for PC mixes with 10 % (on the left) or 20 % (on the right) of wollastonite………………………………………………………………………………………………………………………103

Figure 57: Volume for PC mixes foamed chemically………………………………………………………..104

Figure 58: Compressive testing strain-stress curves for solid polymers and PC mixes foamed chemically…………………………………………………………………………………………………………………………105

Figure 59: Compression modulus for solid polymers and PC mixes foamed chemically……..106

xi Chapter 1

Introduction

1.1 Polymeric foams

Since their discovery in the 1930s, polymeric foams have been widely used in the industry for a variety of applications such as acoustical and thermal insulation, filters or absorbents 1 etc. Economists estimated that polymeric foams had a market value of $83 billion in 2012 and they expect it to reach $131 billion by 2018 2. The reason for this ascending trend can be attributed to factors such as cost, ease of processing and a high strength to weight ratio compared to non-foam polymers 3. Polyethylene (PE), polypropylene (PP), polystyrene (PS) and polyvinyl chloride (PVC) are the polymers mainly used as commercial foams 4.

Polymeric foams are classified on the basis of their structural morphology and the diameter of their voids, which can range from nanometers to millimeters. Two generic classes exist: open-cell and closed-cell foams. Open-cell foams are characterized by voids interconnected within strut, unlike closed-foams, where voids are not connected to each other 1. As a consequence of these differences in morphology, closed-cell foams usually exhibit better mechanical properties, whereas open-cell foams have a structure similar to sponges and are used for absorption applications. The average cell size, distribution, arrangement and volume fraction determine their properties 5.

Two main methods are currently used in the industry and laboratories to foam polymers: physical and chemical foaming. Two types of foams will be characterized in this thesis using

1 a physical foaming process via supercritical CO2 and chemical foaming with blowing agent 5- phenyl-1H-terazole. Both processes will be described in details in chapters 2 and 3.

1.2 Polycarbonate foams

Discovered in 1953 by Dr. H. Schnell and D.W. Fox, polycarbonate (PC) is an amorphous thermoplastic synthesized by a chemical reaction between bisphenol A and phosgene

(COCl2). The chemical reaction can be written as follows:

Figure 1: Chemical reaction for the synthesis of polycarbonate 8

PC was initially used for electronic applications such as displays and plug connections or audio-CDs 6. Due to its durability, high impact resistance and transparency, PC is also used in the automotive, construction and packaging industries etc. For example, optical reflectors, vandal proof windows and riot shields are made of PC.

Kumar, Weller and Hoffer processed polycarbonate foams for the first time in 1990

7. Since then, several laboratories have been investigating polycarbonate foams using both physical and chemical foaming techniques.

2 1.3 Scope

For the past few years, the Laboratory of Advanced Polymers & Optimized Materials

(LAPOM) has been working in collaboration with Encore Wire Corporation (EWC), a company specialized in the production of electrical wires for commercial and residential settings.

LAPOM has been involved in several projects with EWC, mostly concerned with the recycling and revalorization of polymers along with the production of new materials.

In this context, LAPOM has been asked to find a solution to improve the mechanical properties of the polypropylene (PP) spools used to store and transport electrical wires.

Indeed, due to the high density and large weight of the copper used in electrical cables, PP spools are subject to many mechanical failures. As a consequence, LAPOM started working on a new material made of PP reinforced with different fillers such as wollastonite, barium sulfate, calcium carbonate and mica. However, EWC is interested in “indestructible spools”.

Therefore, we decided to use PC (because of its high impact resistance) instead of PP. Since

PC is more expensive than PP, the idea of processing a foamed spool was conceived in order to reduce the amount of material and the manufacturing costs.

As already mentioned, two foaming processes were selected to produce PC foams: physical foaming with supercritical CO2 and chemical foaming with blowing agent 5-phenyl-

1H-terazole. After foaming, each sample was tested by differential scanning calorimetry

(DSC) to determine the influence of the foaming process on the glass transition of PC. Then, scanning electron microscopy (SEM) and X-ray tomography micro-CT were performed to characterize their structure and porosity. Finally, samples were mechanically tested in compression.

3 Afterwards, a processing method was selected for both methods and various fillers were added to the polymeric matrices to enhance their properties.

4 1.4 References

1 E. Olusegun Ogunsona. “Supercritical CO2 foamed biodegradable polymer blends of polycaprolactone and mater-bi”, Master of Science Thesis (December 2007).

2 http://www.plastemart.com

3 I. Coccorullo, L. Di Maio, S. Montesano, L. Incarnato. “Theoretical and experimental study of foaming process with chain extended recycle PET” eXPRESS Polymer Letters Vol.3, No.2

(2009): 84-96.

4 N. Mills. “Polymer foams handbook: engineering and biomechanics applications and design guide”, Butterworth-Heinemann (2007).

5 G. Gedler, M. Antunes, V. Realinho and J. I. Velasco. “Novel polycarbonate-graphene nanocomposite foams prepared by CO2 dissolution” Materials Science and Engineering 31

(2012) 012008.

6http://www.plasticseurope.org/what-is-plastic/types-of-plastics-11148/engineering- plastics/pc.aspx

7 V. Kumar. “Microcellular polymers: novel materials for the 21st century”, progress in rubber and plastics technology, Vol.9, No.1 (1993): 54-70.

8 http://www.wikipedia.org

5 Chapter 2

Processing and characterization of polycarbonate foams with supercritical CO2

2.1 Literature review

2.1.1 Supercritical fluids

A supercritical fluid (SCF) is the phase of a material or compound where its temperature and pressure are above their critical points (critical temperature Tc and critical pressure Pc) 1. As a consequence, the SCF exhibit both the properties of a liquid and a gas

2. When filling a container, a SCF will behave like a gas and fill the shape of the container.

Moreover, its molecules will have the same motion as a gas. On the other hand, its density will be close to that of a liquid, resulting in a similar dissolving effect 1. Baron Charles

Cagniard de la Tour first discovered this phenomenon in 1821 1. Since then, SCFs are commonly used in extraction separation processes and chromatography separation 2. In the past decades, several laboratories have been working on SCF in order to foam polymers.

In most cases, they represent a green and economic alternative to chemical blowing agents used in the industry due to their low cost and non-toxicity. Polymeric foams processed with a SCF are classified as microcellular foams. They are characterized by cell sizes smaller than

10 m and cell densities greater than 109 cm3. Microcellular foams generally have better mechanical properties than solid base polymers because their cell size is smaller than the flaw 3, 4. Finally, SCFs also have the advantage to have a tuneable solvent power as well as enhanced diffusion rates and a plasticization power 5.

6 Besides its critical temperature and pressure, the tuneability, which is defined as the ability of a fluid to change its density in function of the temperature and pressure, is the main property of SCFs 2. SCFs are also characterized in term of viscosity and diffusivity. The general differences in physical properties of gas, liquid and SCF are shown in table 1. Table 2 shows the critical temperature and pressure of common solvents for supercritical applications.

Table 1: General properties of supercritical fluid, gas and liquid 6.

Physical state Density (g/ml) Viscosity (g/cm.s) Diffusivity (cm2/s)

Gas 10-3 10-4 10-1

Liquid 1 10-2 10-6

Supercritical fluid 0.2 – 0.3 10-4 10-3

Table 2: Critical point of common solvents 1.

Critical temperature Critical pressure Compounds (°C) (atm)

Carbon dioxide 31.1 72.9

Ammonia 132.5 109.8

Water 374.1 218.3

Nitrous Oxide 36.5 70.6

Diethyl ether 193.6 63.8

Methane -82.1 45.8

7 Ethane 32.3 47.6

Ethylene 9.2 49.7

Propane 96.7 41.9

Pentane 196.6 33.3

Methanol 240.5 78.9

SCFs are generally characterized in term of their phase diagrams (see figure 2). In figure 2, we consider a pressure-temperature diagram for a pure material. We notice that each phase (solid, liquid and gas) are separated by phase boundaries where two different phases can coexist under given conditions. The triple point (TP) represents the temperature and pressure of a material where solid, liquid and gas phases are in equilibrium. Past the critical pressure (Pc) and temperature (Tc) of a material, properties of the gas and liquid phases are indistinguishable from each other and result in a supercritical fluid 7.

Figure 2: General phase diagram for a supercritical fluid 1.

8 2.1.2 Physical foaming via supercritical CO2

2.1.2.1 Supercritical CO2

CO2 represents a green alternative to organic foaming agents because it is chemically inert and non-toxic 2, 3, 8. Moreover, CO2 is abundant in our atmosphere, which makes it very cost efficient. It is also inflammable, has an impressive tuneability and plasticization power 2, 5. Finally, looking at its phase diagram, we notice that CO2 has a very low critical temperature at 31.1°C and critical pressure at 72.9 atm.

Figure 3: Phase diagram of CO2 9 

9 The production of polymeric foams using supercritical CO2 can be divided in 3 main steps 5,

8, 10:

1) Saturation of the sample in CO2 under its supercritical conditions for a given amount

of time. During this step, the supercritical CO2 will act as a plasticizer, what will cause

a drop in the glass transition (Tg) of the polymer 5, 8, 10. The Tg is defined as the

temperature range where the polymer will go from a glassy to a rubbery state upon

heating. Chiou et al. showed that the Tg of polystyrene (PS) dropped by 50°C under

CO2 at 25 atm (= 2.5 MPa) 10, 11. Sato and al. demonstrated that this phenomenon

is the result of CO2 in the polymer matrix, which decreases the chain mobility. This

causes a drop in the Tg temperature as well as swelling and a drop in viscosity 12.

2) Once saturated, the sample will undergo a rapid drop in pressure leading to an

oversaturation in CO2 8. Reverchon et al. observed that when the depressurization

rate increases, the cell size decreases because of a limited expansion time 5.

3) After oversaturation, cell nucleation and growth will begin. This mechanism is

thermally induced by heating the sample above its Tg. The nucleation process is

energetically favorable if the Tg of the polymer is under room temperature.

Otherwise, the sample will have to be heat to a temperature above its Tg. The

polymer needs to be in a rubbery state to foam because the matrix would be too

rigid if the polymer was still in a glassy state (below its glass transition temperature).

Finally, cell growth will stop once the matrix returns to glassy state because of a

decrease in CO2 concentration into the matrix 8.

10 2.1.2.2 Solubility and diffusion coefficient of CO2 into polycarbonate

The solubility coefficient S, or sorption coefficient, is defined as the amount of gas absorbed by a membrane. This coefficient depends on the interactions between the gas molecules and the membrane, in this case PC 13. The solubility coefficient is defined as followed:

C 푆 = (1) f where C is the equilibrium concentration of the gas in the polymer and f the fugacity.

The diffusion coefficient D is, on the other hand, a measure of the gas molecules mobility in a membrane 13. The diffusion coefficient depends on:

−퐸 퐷 = 퐷 exp ( 퐷 ) 0 푅푇 (2) where D0 is the pre-exponential factor, T the temperature, R the universal gas constant and

ED the activation energy of diffusion.

Work has been done to study the solubility and diffusion coefficient of CO2 when foaming a polymer. The result show that these coefficients vary with the temperature and the pressure inside the vessel, affecting the final properties of the foam. Indeed, Sumarno et al. foamed atactic polystyrene (PS) with supercritical CO2 and observed an increase in cell density as well as a decrease in cell diameter with increasing saturation pressure 14. On another hand, Arora and al had foams with larger cell size and smaller cell density at high temperatures 15. This phenomenon can be explained by the fact that the solubility of CO2 increases with high pressure and low temperature. As a result, the amount of CO2 present in

11 the matrix is higher, causing an increase in nucleation density, therefore smaller cells 5.

However, the diffusion increases with the temperature and pressure.

Ying Sun et al. studied the solubility and diffusion of CO2 in PC at saturation temperatures ranging from 75°C to 175°C and saturation pressure up to 197 atm (= 20 MPa)

10, 11. They used a magnetic suspension balance (MSB) to obtain their results, shown in table 3. Looking at the table, we observe the same behavior explained earlier for the solubility and diffusion coefficient of CO2.

Table 3: Saturated solubility and diffusion coefficient of CO2 in PC for various temperatures

and pressures 10

Solubility gas- Diffusion Temperature Pressure (MPa) amorphous coefficient x10-7 (°C) polymer (cm2/s) 75 5.0 0.0424 1.57

75 9.9 0.0726 2.66

75 15.2 0.0991 6.28

75 20.0 0.1107 8.74

100 5.1 0.0312 3.61

100 10.0 0.0585 5.64

100 15.0 0.0756 12.4

100 19.9 0.0926 13.4

125 5.0 0.0257 8.56

125 9.9 0.0490 13.2

125 15.1 0.0684 18.4

12 125 19.9 0.0846 22.9

150 5.0 0.0217 18.2

150 10.0 0.0418 34.1

150 15.1 0.0617 26.6

150 20.1 0.0733 28.8

2.1.2.3 Batch versus continuous process

Two main methods are currently used to process polymeric foams via supercritical

CO2: batch and continuous foaming. Batch process is mostly used in laboratories for research and development purposes to study and characterized foams and supercritical fluids. The foaming process occurs in a high-pressure vessel and is divided in three major steps (see section 2.1.2.1): saturation of the polymer in CO2, depressurization of the vessel and finally cell nucleation and growth 5, 8, 10.

On the other hand, continuous foaming has been developed for industry as a large scale and economically viable alternative. In this case, the foaming occurs in an extruder (see section

3.1.2) where the polymer and additives were previously introduced. During mixing, the polymer is pressurized with CO2, causing a plasticization. Therefore, the viscosity of the polymer and its glass transition decrease, allowing for processing at lower temperatures.

Finally, the polymer is depressurized at the die of the extruder, creating a foam 5, 8.

13 2.2 Materials for supercritical foaming in CO2

The polycarbonate used for this study was supplied by DOW Plastics under the trade name Calibre 303 10 TNT Polycarbonate. It has a density of 1.2 g/cm-3 and a tensile modulus of 2.4 GPa. The CO2 which was used to foam our samples under supercritical conditions was industrial grade (99.9%) supplied by Air Liquide.

2.3 Experimental

2.3.1 Samples preparation

Firstly, PC pellets were dried in a vacuum oven at 70°C for 12h to eliminate any moisture present in the polymer. Indeed, Calibre 303 10 TNT Polycarbonate has a water absorption of 0.15% at room temperature after 24h. Then, pellets were processed in a

Carver hydraulic compression mold at 220°C in order to obtain rectangular bars (2.2 mm*8.5mm*30.5mm) for thermal and structural characterization. Cylindrical samples

(4mm*18.5mm) were processed for mechanical testing. The polymer was melted for 10 minutes with a contact pressure before applying 10 tons of pressure during 5 minutes. The pressure was released and reapplied three times after 2 minutes and 30 seconds to allow trapped gases to escape and to prevent trapping bubbles. Finally, samples were cooled to room temperature with water. The same processing parameters were used throughout this study to ensure reproducibility in the sample preparation.

2.3.2 Foaming process

In order to determine the saturation time needed to foam a 2.2 mm thick sample, we assumed the CO2 sorption into PC to be unidirectional. This assumption was made to simplify our calculations.

14 -7 2 Using the Fick’s second law and a diffusion coefficient of CO2 into PC D = 2.66*10 cm /s, we get a saturation time of approximately 52 hours for a 2.2 mm thickness.

(3)

where C is the equilibrium concentration of the gas, D the diffusion coefficient and x the thickness of the sample. In this work, I will saturate our samples for 24 h, 48 h or 72 h to study the influence of the saturation time on the overall properties of the foams.

PC foams were prepared by supercritical CO2 one-step batch using a Sandri-PVT-3D chamber. Samples were placed in the high-pressure vessel at 30 atm (3 MPa) and cooled until 10°C. Looking at the phase diagram of CO2, we can see that CO2 will be in a liquid state under this conditions, allowing more CO2 to be present in the vessel since the density of a liquid is higher than the density of a gas. Then, some more gaseous CO2 was introduced into the vessel to reach 55 atm (5.5 MPa). Finally, the chamber was heated until 40°C, which is above the critical temperature, causing the pressure to reach 82 atm (8.5 MPa), which is also above the critical pressure. Once saturation was achieved, the pressure was suddenly dropped to one atmosphere in 10 seconds, leading to an oversaturation in CO2. Finally, the sample was immersed into water at 90°C for 30 seconds to initiate cell nucleation and growth. Then, cells were stabilized by cooling the sample to room temperature.

15 2.3.3 Characterization methods

2.3.3.1 Differential Scanning Calorimetry (DSC)

Differential scanning calorimetry (DSC) is a thermal analysis method where the difference in heat flow required to increase the temperature of a sample and a reference is measured as a function of temperature 16. As a result, quantitative and qualitative data on endothermic (e.g. glass transition and melting) and exothermic (e.g. crystallization) reactions are obtained.

For this project, a Netzsch DSC 204 F1 Phoenix was used to investigate the influence of the foaming process in supercritical CO2 on the glass transition of PC. For this purpose, samples between 3 to 10 mg were evenly spread into an aluminum pan before being sealed.

The top of the pan was pierced prior testing to prevent pan delamination due to eventual degassing. An empty aluminum pan was used as a reference. Then, both samples underwent two thermal cycles under helium atmosphere:

1) First, samples were held for 1 minute at 20°C before being heated to 200°C at

10°C/min and held at this temperature for an additional minute. Then, samples were

cooled down to 20°C at 10°C/min. This first cycle was performed on each sample to

eliminate their thermal history and residual stresses due to processing.

2) The second cycle performed on each sample had the same profile than the first one.

The data obtained from this cycle was used to thermally characterize our foams.

16 2.3.3.2 Scanning Electron Microscopy (SEM)

Scanning electron microscopy was performed on the foams to characterize their structure and morphology such as cell distribution, arrangement, size etc. For this purpose, a

FEI Quanta 200 Environmental Scanning Electron microscope was used. Prior to testing, samples were cryo-fractured in liquid nitrogen to avoid surface yielding, and coated with a thin layer of gold-palladium to make the sample conductive. Micrographs were analyzed with ImageJ to estimate the average cell diameter (D) by taking the cell diameter of at least

50 cells. Cell density (Nc) was also calculated from the micrographs using the following equation:

푛푀2 (4) 푁 ≈ ( )3/2 푐 퐴 where n is the number of cells, M the magnification factor and A the area of the micrograph.

The foam density f of each sample was determined according to ASTM standard

D792 based on the Archimedes’ principle. Densities were calculated by dividing the volume of water displaced by its weight. Finally, the porosity air of each foam was calculated using the following equation:

휌푓 휈푎푖푟 = 1 − 휌푝 (5) where p is the density of an unfoamed sample.

17 2.3.3.3 Compression testing

Compression testing was performed on PC foams in order to compare their compression modulus with neat PC and PP spools used by Encore Wire Corporation, which were used as references. Experiments were conducted on a hydraulic MTS 810 at a strain rate of 0.5 mm/min and a total deformation of 50 % of the initial height. Due to limitations imposed by the size of the high-pressure vessel and the saturation time needed for a given thickness, we could not process samples following ASTM standards requirements. Therefore, cylindrical samples measuring 4 mm in thickness and 18.5 mm in diameter (before foaming) were used.

2.4 Results and discussion

2.4.1 Differential Scanning Calorimetry (DSC)

In order to study the influence of the saturation time and foaming process in supercritical CO2 on the glass transition of our samples, neat PC and its foams were tested using the temperature profile described previously. The glass transition of each sample was calculated through the onset function.

Figure 4 shows the DSC curves obtained for neat PC and samples foamed for 24 h, 48 h and 72 h in supercritical CO2. Looking at the graph, we can see that the glass transition of neat PC (black curve) occurs around 145°C. Moreover, the glass transition of our sample after foaming is seen around 145°C as well. This observation implies that neither the foaming process in supercritical CO2 nor the difference in saturation time affected the glass transition of PC.

18 Endo 

Figure 4: DSC curves for samples foamed in supercritical CO2

However, we previously explained (see the literature review) that during the foaming process, supercritical CO2 acts as a plasticizer, causing a drop in the glass transition temperature 5, 8, 10. To verify this statement, we performed DSC experiments on PC after depressurization, without thermally initiating cell nucleation, cell growth and cell stabilization. Figure 5 shows the behavior after depressurization and figure 6 the behavior after erasing the thermal history and residual stress. For 24 h, we observe an endothermic transition around 70°C instead. This transition appears around 60°C for 48 h and 72 h. The fact that we thermally initiated cell nucleation and growth in water at 90°C means that the glass transition has to occur below 90°C because the polymer needs to be in a rubbery state to foam. Therefore, this endothermic transition is the glass transition of PC, so that CO2 acts as a plasticizer for the polymer at each saturation time. Then, we can see another

19 endothermic transition occurring at 145°C for t = 24 h and 110°C for t = 72 h. We believe these transitions are caused by the stress and thermal history built into the polymer during its processing. These transitions happen after the glass transition because after this temperature, the matrix is in a rubbery state, allowing the stress to be released 17.

We also have sharp endothermic peaks at 160°C for t = 24 h and 145°C for t = 48 h and t = 72 h. We can assume these peaks are the result of the CO2 escaping the polymer matrix and ending the foaming process. The fact that t = 48 h and t = 72 h exhibit the same temperatures for their glass transition and sharp peak could mean that the quantity of CO2 inside PC is the same. Finally, the foaming process for 24 h seems to affect PC differently since its thermal transitions happen at different temperatures. More experiments will be conducted in SEM to confirm this hypothesis.

Looking at run 2 for each saturation time, we notice that the glass transition occurs at 145°C

(like in figure 4) and all other endothermic peaks disappeared. Therefore, we can make the assumption that our samples are now foamed.

20 Endo 

Figure 5: DSC curves for different saturation times in supercritical CO2 after depressurization

Endo 

Figure 6: DSC curves for different saturation times in supercritical CO2 after erasing the stress and the thermal history 21 2.4.2 Scanning Electron Microscopy (SEM)

Scanning electron microscopy (SEM) was performed on our foams to study the influence of the saturation time on their structure and morphology. Images were taken at

100x, 300x and 1000x to compare the cell size, cell density and distribution in our samples.

Figure 7 shows the micrographs obtained in SEM for a saturation time of 24 h. While preparing the sample for SEM, we noticed that the sample was only foamed in surface.

Looking at figure 7, we do not notice any pores inside the polymer matrix meaning that our sample did not foam. We can explain this phenomenon by the fact that the saturation time calculated for our conditions was 52 h. Therefore, by saturating our polymer in supercritical

CO2 for 24 h, we did not give enough time to CO2 to fully penetrate the polymer matrix. As a result, the sample only foamed in surface.

500 m 200 m

A B

50 m

Figure 7: SEM micrographs for PC foamed in supercritical CO2 for 24 h with magnification 100x (A), 300x (B) and 1000x (C) C 22 Looking at figures 8 and 9, corresponding to a saturation time of 48 h and 72 h, we can see that both foams exhibit similar structures: regular pore sizes and an organized open- cell structure. Using ImageJ, we calculated an average pore size of 3.71.6 m and a cell density of 4.1 x 1018 cells/cm3 for a saturation time of 48 h. For a saturation time of 72 h, we got an average pore size of 3.71.9 m and a cell density of 3.9 x 1018 cells/cm3. These values are summarized in table 4.

Table 4: Overall properties of PC foamed in supercritical CO2

Average pore size Cell density Nc Samples (m) (cells/cm3)

PC foamed for 48 h 3.71.6 4.1 x 1018

PC foamed for 72 h 3.71.9 3.9 x 1018

Figures 10 and 11 show the cell size distribution for both saturation times. We can see that both foams have a narrow distribution where most pores have a diameter between

2 to 5 m. One can assume that the similarities in terms of structure and morphology for t =

48 h and t = 72 h is due to the fact that we reached the solubility limit of CO2 into PC after 48 h. As a consequence, the quantity of CO2 present in the polymer matrix after 48 h and 72 h is the same, resulting in the same final structure. To confirm this hypothesis, I estimated the amount of CO2 present in the matrix after oversaturation by measuring the change of volume of each sample before and after foaming. Figure 12 regroups the data obtained and we can see that t = 48 h and t = 72 h have a similar change in volume, meaning that the final amount of CO2 in the matrix was the same. Therefore, we reached the solubility limit of CO2 into PC after 48 h. This conclusion explains why thermal transitions where happening at the 23 same temperatures in DSC for t = 48 h and t = 72 h. Finally, we noticed that the change in volume for t = 24 h is much smaller than other saturation time. This phenomenon makes sense since we previously demonstrated that these samples only foamed on the surface.

500 m 200 m

A B

50 m

Figure 8: SEM micrographs for PC foamed in supercritical CO2 for 48 h with magnification 100x (A), 300x (B) and 1000x (C) C

24 500 m 200 m

A B

50 m

Figure 9: SEM micrographs for PC foamed in supercritical CO2 for 72 h with magnification 100x (A), 300x (B) and 1000x (C) C

Figure 10: Cell diameter distribution for sample foamed 48 h in supercritical CO 2 25 Figure 11: Cell diameter distribution for sample foamed 72 h in supercritical CO2

Figure 12: Volume for samples foamed in supercritical CO2 26 The resolution of x-ray tomography micro-CT being about 6 m, I could not determine the porosity of the foams using this method. Therefore, I estimated their porosity using ASTM D792 based on the Archimedes’ principle. Table 5 regroups the porosity air, calculated using the following equation:

휌푓 (5) 휈푎푖푟 = 1 − 휌푝

Table 5: Density and porosity for PC foamed in supercritical CO2

3 Samples Density 흆풇 (g/cm ) Porosity air (%)

PC foamed for 24 h 1.060.3 4.51

PC foamed for 48 h 0.970.2 122

PC foamed for 72 h 0.980.2 11.52

Looking at table 5, we first notice that the porosity for samples foamed 48 h and 72 h in supercritical CO2 is nearly the same. These results prove that we reached the limit of solubility of CO2 into PC after 48 h. Furthermore, we can see that the foaming process decreases the density of PC since neat PC had a density of 1.2 g/cm3. Finally, the total porosity is smaller than the change in volume after foaming. Therefore, another phenomenon participates in the change in volume. Two hypotheses were formulated to explain this phenomenon: the chains constituting PC swell due to the foaming process or the space between the chains increased. Further experiments would need to be performed to verify these hypotheses.

27 2.4.3 Compression testing

Using the Fick’s second law and a diffusion coefficient of CO2 into PC D = 2.66*10-7 cm2/s, we get a saturation time of approximately 150 hours for a 4 mm thick compression sample. Following the previous procedure, we would have to foam our samples for 75 h, 150 h and 225 h to match former foaming time of 24 h, 48 h and 72 h. However, the data obtained in DSC and SEM showed that a foaming time of 24 h was not enough to entirely foam the polymer. Moreover, the solubility of CO2 into PC was reached after 48 h, giving similar properties for samples foamed for 48 h and 72 h. One can assume that the same phenomenon would occur for foaming times of 75 h, 150 h and 225 h. Therefore, I decided to only foam our samples for 150 h.

Figure 13 shows the strain-stress curves obtained for the Sunoco spool, neat PC and PC foamed in supercritical CO2 for 150 h. Firstly, we notice that the curves obtained in compressive testing can be divided into three distinct regions:

1) A linear elastic region where the relationship between the strain and the stress is

linear. The compression modulus was calculated using the slope of the linear elastic

region.

2) A plastic region happening after the yield point and characterized by a plateau where

the strain increases quickly compared to the stress.

3) A strain hardening region where the strain and the stress increase sharply.

When comparing the curves, we can see that the stress required to deform the sample is greater for PC and then the foam. Moreover, we do not have a relatively flat plateau for the foam, which is a common phenomenon for high-density foams 18. Zhang, Ashby and Sun et al. demonstrated that the rise in the plateau region was the consequence of the gas

28 (present into the matrix) being compressed, overtaking the response of the foam to the compressive stress 18, 19, 20, 21.

Figure 13: Compressive testing strain-stress curves for solid polymers and PC foamed in supercritical CO2

Figure 14 regroups the compressive modulus obtained for our samples. We can see that the modulus of PC and the foam are greater than the spool. Microcellular structures can exhibit higher mechanical properties than solid base materials when the cell size is smaller than the flaw 3, 4. However, the modulus of the foam is lower than neat PC. We can assume that is phenomenon is the result of the open-cell structure of our foams, which exhibit lower mechanical properties than closed-cell structures. Moreover, the cell size has to be smaller than the flaw to enhance the mechanical properties of the foam 3, 4. In our

29 case, pores are still probably bigger than the flaw. Finally, lower mechanical properties could be attributed to the presence of micro bubbles of gas during processing in compression molding.

Figure 14: Compression modulus for solid polymers and PC foamed in supercritical CO2

30 2.5 Conclusions

After foaming PC in supercritical CO2, samples were tested and characterized in DSC,

SEM and compression. DSC experiments showed that the glass transition of PC was dropping to 60°C after oversaturation of the polymer matrix in CO2. Then, cell nucleation and growth were thermally initiated in hot water at 90°C. This last step in the foaming process caused the glass transition of PC to shift back to its original temperature at 145°C.

SEM experiments showed that the solubility of CO2 in PC was reached after 48 h, giving samples an organized opened-cell structure with small pores and a high cell density for 48 h and 72 h. On the other hand, samples saturated for 24 h did not foam. Therefore, a saturation time of 48 h was selected for the rest of this study. After testing the foams in compression testing, we saw that neat PC foams had a higher compression modulus than the spool used by Encore Wire Corporation. However, PC foams had lower mechanical properties than neat PC. In chapter 4, fillers will be added to the polymer matrices to enhance their physical and mechanical properties.

31 2.6 References

1 http://barron.rice.edu/Courses/475/475_2013/475_projects_2013/Hizir_Draft.pdf

2 E. Olusegun Ogunsona. “Supercritical CO2 foamed biodegradable polymer blends of polycaprolactone and mater-bi”, Master of Science Thesis (December 2007).

3 M.T. Liang and C.M. Wang. “Production of engineering plastics foams by supercritical

CO2” Ind. Eng. Chem. Res., Vol.39, No.12 (2000).

4 I. Tsivintzelis, A.G. Angelopoulou, C. Panayiotou. “Foaming of polymers with supercritical

CO2: an experimental and theoretical study” Polymer, Vol.48 (2007): 5928-5939.

5 E. Reverchon, S. Cardea. “Production of controlled polymeric foams by supercritical CO2”.

J. of Supercritical Fluids, Vol.40 (2007): 144-152.

6 http://www.uni-leipzig.de/~pore/files/3rd_irtg_workshop/dvoyashkin.pdf

7 http://www1.lsbu.ac.uk/water/water_phase_diagram.html

8 L. J.M. Jacobs, M. F. Kemmere, J. T.F. Keurentjes. “Sustainable polymer foaming using high-pressure carbon dioxide: a review on fundamentals, processes and applications” Green

Chem., Vol.10 (2008): 731-738.

9http://www.chegg.com/homework-help/questions-and-answers/phase-diagram- pressure-temperature-graph-shows-ranges-temperature-pressure-phase-stable-ph- q4139861

10 Y. Sun, M. Matsumoto, K. Kitashima, M. Haruki, S.I. Kihara, S. Takishima. “Solubility and diffusion coefficient of supercritical CO2 in polycarbonate and CO2 induced crystallization of polycarbonate” J. of Supercritical Fluids, Vol.95 (2014): 35-43.

11 J. S. Chiou, J.W. Barlow, D.R. Paul. “Plasticization of glassy polymers by CO2” J. Apllied

Polymer Science, Vol.30 (1985): 2633-2642.

32 12 Y. Sato, K. Fujiwara, T. Takikawa, Sumarno, S. Takishima, H. Masuoka. “Solubilities and diffusion coefficients of carbon dioxide and nitrogen in polypropylene, high density polyethylene, and polystyrene under high pressures and temperatures” Fluid Phase

Equilibria, Vol.162 (1999): 261-276.

13 C.M. Laot. “Gas transportation properties in polycarbonate: influence of the cooling rate, physical aging and orientation” PhD dissertation (October 2001).

14 A. R. Sumarno, G.S. Bernardus, A.S. Ismail, P.A. Putu Teta. “Characteristics of polystyrene microcellular plastic structure processed at elevated pressure and temperature using supercritical carbon dioxide” Proceeding of the Eight Meeting on Supercritical Fluids

(Chemical Reactivity and Material processing in Supercritical Fluids) (2002): 179.

15 K.A. Arora, A.J. Lesser, T.J. McCarthy. “Preparation and characterization of microcellular polystyrene foams processed in supercritical carbon dioxide” Macromolecules, Vol. 31

(1998): 4614-4620.

16 http://instrument-specialists.com/applications/differential-scanning-calorimetry-dsc/

17 http://www.tainstruments.co.jp/application/pdf/Thermal_Library/Applications_Briefs/

18 Z. Ma, G. Zhang, Q. Yang, X. Shi, Y. Liu. “Mechanical and dielectric properties of microcellular polycarbonate foams with unimodal or bimodal cell-size distribution” J. of

Cellular Plastics (2014): 1-21

19 H. Sun, G.S. Sur, J.E. Mark. “Microcellular foams from polyethersulfone and polyphenylsulfone: preparation and mechanical properties” Eur Polymer Journal, Vol.38

(2002): 2373-2381.

20 L.J Gibson, M.F. Ashby. “Cellular solids: structure and properties” Cambridge University

Press (1997)

21 J. Zhang, M.F. Ashby. “Mechanics of cellular materials” Master of Science Thesis (1988) 33 Chapter 3

Processing and characterization of polycarbonate foams with chemical foaming agent 5-phenyl-1H-terazole

3.1 Literature review

3.1.1 Chemical foaming agent

In the industry, chemical foaming of a polymer is generally obtained using chemical foaming agents (CFAs). Either organic or inorganic compounds, CFAs are defined as thermally unstable components, which create gas upon heating to a certain temperature 1,

2, 3. The gas created will induce a foaming process in the polymer melt. CFAs are typically used in injection and extrusion processes (both methods will be described later) to obtain medium to high-density foams. CFAs are added in small quantities, typically 1 % to 3 %, and are used to obtain microcellular to macrocellular foams with pore diameter of a few microns to millimeters.

Chlorofluorocarbons (CFCs) such as CFC11 (chemical formula: CCL3F) were used in the past due to their low boiling points, low diffusivity and non-flammability 4. However, NASA found out that CFCs were harmful for the environment since CFC11 used to foam polyurethane created a hole in stratospheric ozone layer. Since then, environmentally responsible CFAs have been developed by the industry to respect the Montreal Protocol 5.

CFAs are characterized as endothermic or exothermic, depending on their behavior.

CFAs that undergo an endothermic transition cause an absorption of energy and a release of carbon dioxide (CO2) along with moisture upon decomposition. On another hand, CFAs that undergo an exothermic transition cause an energy release and produce nitrogen 3.

34 CFAs are usually engineered for a specific polymer and application. Indeed, they must be compatible with a given polymer and their decomposition range has to match the temperature at which the polymer is processed 3.

3.1.2 Foaming via extrusion

Generally used to process thermoplastics, extrusion is a continuous manufacturing method used to produce polymer pellets and final products such as pipes, sheets, wires etc.

6, 7. Figure 15 illustrates a schematic diagram of an extruder. The polymer, usually in the form of pellets, is introduced into the extruder through the hopper to feed a heated screw inside the barrel. First, pellets are compacted by the rotation of the screw to form a bed of solid material at the feeding zone (or feeding section). Then, the polymer starts to soften and melts in the compression zone due to the heat and shear produced by the screw. Once in the metering zone, the polymer exits the extruder through the die, which is designed in function of the final shape desired, and cooled until room temperature in a water bath 6,

8.

35 Figure 15: Schematic illustration of an extruder 9

When foaming a polymer in extrusion, a few adjustment are needed on the extruder to ensure optimum processing conditions. Firstly, the extruder has to have a minimum length/diameter ratio of 24:1 and the design of the screw should allow pressure to build along its length while mixing gently. As a result, the CFA will have enough time to completely decompose and the gas created will remain in the polymer melt. Screen packs should be removed to avoid a drastic drop in pressure, causing premature foaming. Degassing vent ports must also be closed to avoid any leakage. While processing, the temperature profile along the screw should be built in a bell curve where the temperature at the feeding zone is as cold as possible to avoid premature foaming and gas leak through the hopper. The temperature should progressively increase before decreasing at the die to increase melt strength 3. The pressure at the die needs to be greater than the solubility pressure of the

CFA in the polymer and the pressure drop rate has to be high enough to increase the cell density and the expansion ratio 10. Finally, the temperature and size of the cooling bath needs to be adapted to allow the polymer to foam 3. 36 3.1.3 Foaming via injection molding

Unlike extrusion, injection molding is a non-continuous manufacturing process used to produce plastic parts with complex geometries and high quality finish such as bottles, syringes, cases etc. Figure 16 and 17 respectively illustrate a schematic diagram of an injection molding machine and a mold. We can see that an injection molding machine is composed of two mains parts: the clamping unit that holds the mold under pressure during injection and cooling and the injection unit composed by a mold and a screw 11.

The injection molding process is divided in six major steps 11, 12, 13. Once this procedure is over, this method will be repeated over and over:

1) Clamping: the two halves of the mold are held together under pressure before

injecting the polymer into the mold cavity.

2) Injection: in this operation, polymer pellets are fed into the barrel of the injection

machine through the hopper, where the raw material is melted and mixed by a

rotating screw at a given temperature and pressure. Once the quantity of material

needed to fill the print has accumulated in front of the screw, the polymer is injected

into the mold through the nozzle. This step is called the shot.

3) Dwelling: after injecting the polymer, a given pressure is applied on the mold to

ensure all cavities of the mold are filled.

4) Cooling: the melted polymer cools down and solidifies due to cold water circulating

into the mold. Moreover, the temperature gradient between the polymer and the

mold also induces cooling.

5) Mold opening: the pressure on the clamps is released to open the mold once the

polymer is fully solidified. 37 6) Ejection: the final product is ejected from the mold by the ejection system (ejector

plate and pins)

Figure 16: Schematic illustration of an injection molding machine 14

Figure 17: Schematic illustration of an injection mold 13

Foaming in injection molding is mostly used in the industry to avoid sink marks (local depression of the polymer causing a surface defect) and reduce the weight of the final

38 product. As a consequence, the amount of material injected (polymer + CFA) into the mold during the shot operation has to be reduced to allow the polymer to foam and fill the device.

The formation of sink marks can be avoid by the expansion of the polymer induced by the foaming process. Finally, the injection speed has to be high enough to ensure uniform expansion of the polymer 3.

3.2 Materials for supercritical foaming with chemical foaming agent 5-phenyl-1H-terazole

The polycarbonate used for this study was supplied by DOW Plastics under the trade name Calibre 303 10 TNT Polycarbonate. It has a density of 1.2 g/cm3 and a tensile modulus of 2.4 GPa. PetroChemTrade supplied the CFA 5-phenyl-1H-terazole (chemical formula:

C7H6N4) in a powder form. 5-phenyl-1H-terazole is an endothermic CFA with a theoretical decomposition temperature at 216°C.

Figure 18: Schematic structure of 5-phenyl-1H-tetrazole 17

A thermogravimetric analysis (TGA) was performed on 5-phenyl-1H-terazole to precisely determine its decomposition temperature. Figure 18 shows the curve obtained for a thermal cycle from 50°C to 600°C at 20°C/min. Samples were held for 1 minute at 50°C and

600°C. Looking at figure 19, we can see that 5-phenyl-1H-terazole starts to decompose at about 215°C, which corresponds to the temperature given by the supplier.

39 Figure 19: TGA curve for 5-phenyl-1H-terazole

3.3 Experimental

3.3.1 Samples preparation

PC pellets were dried in a vacuum oven at 70°C for 12h to eliminate any moisture present in the polymer since Calibre 303 10 TNT Polycarbonate has a water absorption of

0.15% at room temperature for 24h. The same processing parameters were used throughout this study to ensure reproducibility in the sample preparation.

40 3.3.2 Foaming process

In order to reproduce the foaming conditions which occur during injection molding machine, 5 g of PC pellets were placed into an aluminum cylinder (15.2mm*37.8mm) with a given percentage per mass of CFA. Then, another aluminum recipient was used to enclose the polymer. Finally, samples were placed into an oven for 10 minutes at 220°C. After 10 minutes, samples were quenched into water or liquid nitrogen (LN) for 30 seconds to stop the foaming process and stabilize the foams. Samples obtained with both quenching methods were characterized to determine their influence on the foaming process. Table 6 shows the different percentages of 5-phenyl-1H-terazole used and the results obtained.

Table 6: Percentage as a function of 5-phenyl-1H-terazole

Percentage per mass of 5-phenyl-1H- Final result terazole (%)

1 Polymer did not melt and no foaming

3 Polymer melted and foamed

5 Polymer melted and foamed

Polymer melted but the foaming was not 7 controlled

Polymer melted but the foaming was not 10 controlled

In this work, only results for samples with 3 % and 5 % of 5-phenyl-1H-terazole are reported. As seen in table 6, samples mixed with 1 % of CFA did not foam and samples mixed with 7 % and 10 % had a very random and uncontrolled foaming process. 41 3.3.3 Characterization methods

3.3.3.1 Differential Scanning Calorimetry (DSC)

For this project, a Netzsch DSC 204 F1 Phoenix apparatus was used to investigate the influence of the foaming process with 5-phenyl-1H-terazole on the glass transition of PC. For this purpose, cut samples between 3 to 10 mg were evenly spread into an aluminum pan before being sealed. The top of the pan was pierced prior testing to prevent pan delamination due to eventual degassing. An empty aluminum pan was used as a reference.

Then, both samples underwent two thermal cycles under helium atmosphere:

1) First, samples were held for 1 minute at 20°C before being heated to 200°C at

10°C/min and held at this temperature for an additional minute. Then, samples were

cooled down to 20°C at 10°C/min. This first cycle was performed on each sample to

eliminate their thermal history and residual stresses due to processing.

2) The second cycle performed on each sample had the same profile than the first one.

The data obtained from this cycle was used to thermally characterize our foams.

3.3.3.2 Scanning Electron Microscopy (SEM)

Scanning electron microscopy was used to characterize the structure and morphology of the foams. For this purpose, a FEI Quanta 200 Environmental Scanning

Electron Microscope was utilized. Micrographs were analyzed with ImageJ software to estimate the average cell diameter (D) by taking the cell diameter of at least 10 cells. Cell density (Nc) was also calculated from the micrographs using the following equation:

푛푀2 (4) 푁 ≈ ( )3/2 푐 퐴 where n is the number of cells, M the magnification factor and A the area of the micrograph.

42 3.3.3.3 X-ray tomography micro-CT

A SkyScan 1172 scanner was used in this work to determine the porosity of the foams. Micro-CT is a non-destructive technique which uses x-rays to obtain a three- dimensional model for detailed analysis. The scanner emits x-rays through the sample, which is rotating at a preset speed. The raw data is a simple image file of the side view if the sample during rotation. The reconstruction program then takes of these images and creates

“slices” of the material. Finally, a specified modeling program adds these “slices” together to create a three-dimensional model. Information such as total porosity, number of pores and geometrical content can be obtained with this characterization method. Tests were performed under a voltage of 61 kV and a current of 163 A.

3.3.3.4 Compression testing

Compression testing was performed on PC foams in order to compare their compression modulus with neat PC and PP spools used by Encore Wire Corporation, which were used as references. Experiments were conducted on a hydraulic MTS 810 at a strain rate of 0.5 mm/min and a total deformation of 50 % of the initial height.

3.4 Results and discussion

3.4.1 Differential Scanning Calorimetry (DSC)

Figure 20 shows the DSC curves obtained for neat PC and foams processed with different percentages of 5-phenyl-1H-terazole. Looking at the graph, we can see that the glass transition of neat PC (black curve) occurs around 145°C. Then, we notice that for each percentages of CFA, the glass transition of the foams did not change and remains in the

43 same temperature range. This observation demonstrates that neither the foaming process nor the quantity of CFA and the quenching method influence the glass transition of PC.

Endo 

Figure 20: DSC curves for samples foamed with 5-phenyl-1H-terazole

3.4.2 Scanning Electron Microscopy (SEM)

Images were taken at 50x and 150x to compare the cell size, cell density and distribution in our samples.

Figure 21 shows the micrographs obtained in SEM for samples foamed with 5-phenyl-

1H-terazole and quenched in water. Figures 22 and 23 show the cell size distribution for 3 % and 5 % of CFA. Unlike samples foamed in supercritical conditions, both foams exhibit a

44 closed-cell structure with a random cell size distribution. We believe this phenomenon occurred because we could not control the foaming process and the formation of the voids.

The average cell size is a lot greater since we have an average diameter of 0.150.1 mm for 3

% of CFA and 0.350.2 mm for 5 % of CFA. We also notice that the diameter of the pores is about twice as big for 5 % compared to 3 %. This can be explained by the fact that more CO2 was produced with the decomposition of 5 % of 5-phenyl-1H-terazole, inducing more foaming.

Finally, we calculated a cell density of 3.7 x 109 cells/cm3 for 3 % of CFA and 2.6 x 109 cells/cm3 for 5 % of CFA. The cell density obtained in chemical foaming for samples quenched in water is much smaller than the ones obtained in supercritical CO2. We can assume that this phenomenon is due to the larger average cell size and random distribution of the pores obtained in chemical foaming.

45 1 mm 400 m

A B

1 mm 400 m

D C Figure 21: SEM micrographs for PC foamed chemically and quenched in water with 3 % (A, C) and 5 % (B, D) of 5-phenyl-1H-terazole with magnification 50x (A, B) and 150x (C, D)

46 Figure 22: Cell size distribution for sample foamed with 3 % 5-phenyl-1H-terazole and quenched in water

Figure 23: Cell size distribution for sample foamed with 5 % 5-phenyl-1H-terazole and quenched in water

47 Figure 24 shows the micrographs obtained in SEM for samples foamed with 5-phenyl-

1H-terazole and quenched in liquid nitrogen. Figures 25 and 26 show the cell size distribution for 3 % and 5 % of CFA. Looking at these figures, we see that foams quenched in liquid nitrogen also exhibit a closed-cell structure with a large cell size distribution. However, they have a smaller average cell size with respectively 0.20.1 mm and 0.250.1 mm for 3 % and 5 % of 5-phenyl-1H-terazole. We can assume that the diminution in diameter compared to foams quenched in water is due to the fact that they thermally contract more during quenching because the temperature gradient with the oven is greater for liquid nitrogen than for water. Therefore, the diameter of the pores decreases, causing a general decrease in volume. On another hand, the cell size distribution is narrower in this case for both percentages of CFA. Finally, we calculated a cell density of 1.4 x 109 cells/cm3 for 3 % of CFA and 9.6 x 109 cells/cm3 for 5 % of CFA. The cell density obtained in chemical foaming for samples quenched in liquid nitrogen remains much smaller than the ones obtained in supercritical CO2 but higher than samples quenched in water. For both quenching methods, we can see that the cell density is greater for 3 % of 5-phenyl-1H-terazole because the average pore size is smaller, allowing more pores to develop into the polymer matrix.

48 1 mm 400 m

A B

1 mm 400 m

C D Figure 24: SEM micrographs for PC foamed chemically and quenched in liquid nitrogen with 3 % (A, C) and 5 % (B, D) of 5-phenyl-1H-terazole with magnification 50x (A, B) and 150x (C, D)

49 Figure 25: Cell size distribution for sample foamed with 3 % 5-phenyl-1H-terazole and quenched in liquid nitrogen Figure 26: Cell size distribution for sample foamed with 5 % 5-phenyl-1H-terazole and quenched in liquid nitrogen

Table 7 details the overall properties of PC foamed chemically with 5-phenyl-1H- terazole and quenched in water or liquid nitrogen (LN):

Table 7: Overall properties for PC foamed with 5-phenyl-1H-terazole

Samples Average pore size (mm) Cell density Nc (cells/cm3)

3 % CFA quenched in LN 0.20.2 1.4 x 109

3 % CFA quenched in water 0.150.1 3.7 x 109

5 % CFA quenched in LN 0.250.2 9.6 x 109

5 % CFA quenched in water 0.350.2 2.6 x 109

50 We calculated the change in volume of the foams after foaming in order to determine if this phenomenon is related to the porosity. Figure 27 shows the results obtained. Considering incertitude, the change in volume occurring after foaming in chemical for both quenching method is the same. Therefore, quenching samples in water or liquid nitrogen has no influence on the final volume of the foams.

Figure 27: Volume for samples foamed with 5-phenyl-1H-tetrazole

51 3.4.3 X-ray tomography micro-CT

X-ray tomography micro-CT was performed on our foams to estimate their porosity.

Figure 28 and 29 respectively show the three-dimensional models obtained for samples foamed with 5-phenyl-1H-terazole quenched in water and in liquid nitrogen. Looking at the figures, we clearly see that the foams obtained with both quenching methods have a random foaming process since we do not notice any organized distribution of the pores.

Moreover, we can also see very large voids and pores with very different diameters. Table 8 regroups the porosity of the foams. Firstly, we notice that the porosity obtained is greater than for samples foamed in supercritical CO2. Then, for both quenching methods, the porosity obtained after chemical foaming is the same when considering incertitude.

Looking at the data previously obtained for the change in volume of the samples after foaming, we can see that the porosity is not linearly related to this property. Indeed, the porosity determined after foaming is smaller than the change in volume. Therefore, the change in volume is partly due to the pores created. Two hypotheses were formulated to explain this phenomenon: the chains constituting PC swell due to the foaming process or the space between the chains increased. Further experiments would have to be performed to verify these hypotheses.

52 Table 8: Porosity of chemically foamed samples obtained in micro-CT

3 % CFA 5 % CFA 3 % CFA 5 % CFA

Samples quenched in quenched in quenched in quenched in

water water liquid nitrogen liquid nitrogen

Porosity (%) 47  9 50  15 43.5  13 47.5  17

A B

C D Figure 28: Three-dimensional models for samples foamed chemically and quenched in water with 3 % (A, B) and 5 % (C, D) of 5-phenyl-1H-tetrazole

53 A B

C D Figure 29: Three-dimensional models for samples foamed chemically and quenched in liquid nitrogen with 3 % (A, B) and 5 % (C, D) of 5-phenyl-1H- tetrazole

3.4.4 Compression testing

Figure 30 shows the strain-stress curves obtained for the Sunoco spool, neat PC and

PC foamed both physically and chemically. For foams foamed with 5-phenyl-1H-terazole, we can see the same regions (linear elastic region, plastic region and strain hardening) observed on the strain-stress curves for PC foamed in supercritical CO2 (see section 2.4.3). However, samples foamed chemically did not reach a strain of 0.5 mm/mm, except for samples with 5

% CFA and quenched in liquid nitrogen. Their structures were experiencing mechanical

54 failures around 0.3 mm/mm because samples were very brittle. Finally, we can see that the stress required to deform the sample is smaller than for the spool from Sunoco.

Figure 30: Compressive testing strain-stress curves for solid polymers and PC foamed in supercritical CO2 or 5-phenyl-1H-tetrazole

Figure 31 presents the compressive modulus obtained for the foams. We can see that the compression modulus values for samples foamed with 5-phenyl-1H-terazole are much smaller than for neat PC and PC foamed in supercritical CO2. We can assume that this phenomenon is due to the fact that pores are very large, the cell density is lower than supercritical samples and the cell distribution is entirely random. Indeed, Jacobs et al. demonstrated that the mechanical strength of a foam decreases when its cell size increases

55 2. Moreover, Sun et al. showed that foams with lower relative densities were more fragile and subject to microcracks propagating more easily in the matrix 16.

Figure 31: Compression modulus for solid polymers and PC foamed in supercritical CO2 or 5-phenyl-1H-tetrazole

56 3.5 Conclusions

After chemically foaming PC with 5-phenyl-1H-tetrazole, samples were tested and characterized in DSC, SEM and compression. DSC experiments showed that the glass transition of PC was not affected by the foaming process and remained around 145°C. Unlike

PC foamed in supercritical CO2, SEM experiments showed that chemical foaming resulted in a very disorganized closed-cell structure with very large pores. Furthermore, one observes a smaller change in volume and porosity for samples quenched in liquid nitrogen. Mechanical testing in compression showed that PC foams had very poor mechanical properties due to their low cell densities and large pores. The method using 3 % of CFA and water quenching was selected for the next part of the study due to its slightly better properties and cheaper cost. Indeed, less CFA was used and quenching the foams in liquid nitrogen would be too expensive on a large scale operation. In chapter 4, results for fillers added to the polymer matrix to enhance its physical and mechanical properties will be reported.

57 3.6 References

1 E. Reverchon, S. Cardea. “Production of controlled polymeric foams by supercritical CO2”.

J. of Supercritical Fluids, Vol.40 (2007): 144-152.

2 L. J.M. Jacobs, M. F. Kemmere, J. T.F. Keurentjes. “Sustainable polymer foaming using high-pressure carbon dioxide: a review on fundamentals, processes and applications” Green

Chem., Vol.10 (2008): 731-738.

3 http://www.ptonline.com/articles/how-to-mold-extrude-using-chemical-foaming-agents

4 N. Mills. “Polymer foams handbook: engineering and biomechanics applications and design guide”, Butterworth-Heinemann (2007).

5 M. Nar. “Structural, thermal, and acoustic performances of polyurethane foams for green buildings”, PhD dissertation UNT (August 2014).

6 https://www.tut.fi/ms/muo/tyreschool/moduulit/moduuli_6/hypertext/3/3_2.html

7 http://www.substech.com/dokuwiki/doku.php?id=extrusion_of_polymers

8 http://www.polydynamics.com/Overview_Polymer_Processing.pdf

9http://www.google.fr/imgres?imgurl=http%3A%2F%2Fpatentimages.storage.googleapis.c om%2FUS20100303883A1%2FUS20100303883A1-20101202- D00000.png&imgrefurl=http%3A%2F%2Fwww.akitarescueoftulsa.com%2Fscrew-extruder- diagram%2F&h=951&w=1877&tbnid=xcAKHuqUDg4qRM%3A&zoom=1&docid=LdRGxXNcKQ owRM&ei=vJ_jVK3sBoaGyQSrpYDIAw&tbm=isch&iact=rc&uact=3&dur=376&page=2&start= 13&ndsp=18&ved=0CFUQrQMwEQ

10 J. W.S. Lee, K. Wang, C. B. Park. “Challenge to extrusion of low-density microcellular polycarbonate foams using supercritical carbon dioxide” Ind. Eng. Chem. Res., Vol.44 (2005):

92-99

11 http://www.patterson-rothwell.co.uk/services/plastic-injection-moulding- explained.htm

12 http://www.custompartnet.com/wu/InjectionMolding

58 13 http://www.polyplastics.com/en/support/mold/outline/

14 http://elitemachinerysystems.com/plastic_injection_molding.php

15 Z. Ma, G. Zhang, Q. Yang, X. Shi, Y. Liu. “Mechanical and dielectric properties of microcellular polycarbonate foams with unimodal or bimodal cell-size distribution” J. of

Cellular Plastics (2014): 1-21

16 H. Sun, G.S. Sur, J.E. Mark. “Microcellular foams from polyethersulfone and polyphenylsulfone: preparation and mechanical properties” Eur Polymer Journal, Vol.38

(2002): 2373-2381.

17 http://www.sigmaaldrich.com/catalog/product/aldrich/347744?lang=en®ion=US

59 Chapter 4

Processing and characterization of polycarbonate foams with fillers

4.1 Literature review

Fillers are widely used in laboratories and in the industry to enhance the overall properties and reduce the manufacturing cost of a polymer. This philosophy is also applied to polymeric foams with the incorporation of nanoparticles into the polymer matrix.

The most common nanoparticles used are silica, carbon nanofibers or montmorillonite

(MMT). Several laboratories demonstrated that the addition of nanoparticles into polymeric foams enhances their mechanical and physical properties by reinforcing the cell walls 1.

Lee et al. observed an increased in tensile modulus of their polystyrene (PS) foamed in supercritical CO2 of respectively 28 % and 45 % with 1 % or 5 % per weight of carbon nanofibers 2. Zhai et al. showed that nanoparticles were acting as very good heterogeneous nucleation sites, resulting in smaller pores, higher cell density and more homogeneous cell size distribution 3. They arrived at this conclusion by adding 1 % to 9 % per weight of silica to their foams.

60 4.2 Materials for supercritical foaming in CO2

10 % and 20 % per mass of fillers were added to the polycarbonate and foaming agents (supercritical CO2 or 5-phenyl-1H-terazole) previously used to enhanced the mechanical properties of the foams and decrease their manufacturing price.

Instead of using nanoparticles, microparticles were used in this work to study their effects on the foam and reduce manufacturing cost. Scraps of cross-linked polyethylene (XLPE) were supplied by Encore Wire Corporation with two mesh sizes: 35 ( = 500 m) and 50 ( = 297

m). XLPE scraps came from miscellaneous products. Moreover, we also used wollastonite due to impressive results in previous studies. Wollastonite (CaSiO3) is a calcium inosilicate mineral supplied by MasterBond under a powder form with an average particle size of 8 m.

4.3 Experimental

4.3.1 Samples preparation

Firstly, PC pellets were dried in a vacuum oven at 70°C for 12h to eliminate any moisture present in the polymer. Indeed, Calibre 303 10 TNT Polycarbonate has a water absorption of 0.15% at room temperature for 24h. Then, pellets were processed in a twin screw C.W Brabender at 220°C and 50 rpm. 40 g of material (PC pellets + 10 or 20 % of filler) was progressively introduced into the chamber during 2 minutes to ensure a homogeneous melting and mixing of the materials. Then, the material was mixed for 4 minutes to complete the mixing process. Finally, mixes were pelletized with a Fritsch pelletizer and processed following both methods described previously (see section 2.3.2 for supercritical foaming and section 3.3.2 for chemical foaming). Samples were foamed for 48 h in supercritical CO2 at

40°C and 82 atm. Samples foamed chemically were foamed with 3 % per mass of 5-phenyl-

1H-tetrazole at 220°C for 10 minutes and quenched in water. 61 4.3.3 Characterization methods

4.3.3.1 Fourier Transform Infrared Spectroscopy (FTIR)

Fourier transform infrared spectroscopy (FTIR) was performed on each filler to determine their composition. A Nicolet 6700 spectrometer from Thermo Electron was used in this work.

FTIR is a characterization technique used to identify organic and inorganic compounds under liquid, solid or gaseous phase. This method measures the absorption of infrared radiation by a sample for various wavelengths. The specific wavelengths absorbed by a given materials are characteristics to its structure, allowing its detection.

4.3.3.2 Differential Scanning Calorimetry (DSC)

For this project, a Netzsch DSC 204 F1 Phoenix was used to investigate the influence of the foaming process and fillers on the glass transition of PC. For this purpose, samples between 3 to 10 mg were evenly spread into an aluminum pan before being sealed. The top of the pan was pierced prior testing to prevent pan delamination due to eventual degassing.

An empty aluminum pan was used as a reference. Then, both samples underwent two thermal cycles under helium atmosphere:

1) First, samples were held for 1 minute at 20°C before being heated to 200°C at

10°C/min and held at this temperature for an additional minute. Then, samples were

cooled down to 20°C at 10°C/min. This first cycle was performed on each sample to

eliminate their thermal history and residual stresses due to processing.

2) The second cycle performed on each sample had the same profile than the first one.

The data obtained from this cycle was used to thermally characterize our foams.

62 4.3.3.3 Scanning Electron Microscopy (SEM)

Scanning electron microscopy was performed on the foams to characterize the influence of the fillers on their structure. For this purpose, a FEI Quanta 200 Environmental

Scanning Electron microscope was used. Prior to testing, samples were cryo-fractured in liquid nitrogen to avoid surface yielding, and coated with a thin layer of gold-palladium to make the sample conductive. Micrographs were analyzed with ImageJ to estimate the average cell diameter (D) by taking the cell diameter of at least 50 cells. Cell density (Nc) was also calculated from the micrographs using the following equation:

푛푀2 (4) 푁 ≈ ( )3/2 푐 퐴 where n is the number of cells, M the magnification factor and A the area of the micrograph.

The foam density f of each sample was determined according to ASTM standard

D792 based on the Archimedes’ principle. Densities were calculated by dividing the volume of water displaced by its weight. Finally, the porosity air of each foam was calculated using the following equation:

휌푓 휈푎푖푟 = 1 − 휌푝 (5) where p is the density of an unfoamed sample.

63 4.3.3.4 X-ray tomography micro-CT

A SkyScan 1172 scanner was used in this work to determine the porosity of the foams with fillers. Micro-CT is a non-destructive technique using x-rays to obtain a three- dimensional model for detailed analysis. The scanner emits x-rays through the sample, which is rotating at a preset speed. The raw data is a simple image file of the side view if the sample during rotation. The reconstruction program then takes of these images and creates

“slices” of the material. Finally, a specified modeling program adds these “slices” together to create a three-dimensional model. Information such as total porosity, number of pores, geometrical content etc. can be obtained with this characterization method.

Tests were performed under a voltage of 61 kV and a current of 163 A.

4.3.3.5 Compression testing

Compression testing was performed on PC foams with fillers in order to compare their compression modulus with neat PC, PP spools and the foams without fillers.

Experiments were conducted on a hydraulic MTS 810 at a strain rate of 0.5 mm/min and a total deformation of 50 % of the initial height. Due to limitations imposed by the size of the high-pressure vessel and the saturation time needed for a given thickness, we could not process samples following ASTM standards requirements. Therefore, cylindrical samples measuring 4 mm in thickness and 18.5 mm in diameter (before foaming) were used.

64 4.4 Results and discussion

4.4.1 Fillers characterization

Fourier transform infrared spectroscopy (FTIR), differential scanning calorimetry

(DSC), scanning electron microscopy (SEM) and compression testing were performed on the fillers and PC mixes in order to characterize the influence of the fillers. Indeed, understanding the behavior of each filler into the PC matrix will allow a better interpretation of the results once mixes are foamed. It will also allow us to differentiate the effect of both foaming methods on our samples.

4.4.1.1 Fourier Transform Infrared Spectroscopy (FTIR)

Firstly, both XLPEs and wollastonite were tested in fourier transform infrared spectroscopy (FTIR) to determine their composition. Figures 32 and 33 respectively show the data obtained for XLPEs and wollastonite.

Looking at figure 32 for XLPEs, we can see that both XLPE have similar curves, which means that they have the same composition. Indeed, both of them are mainly composed of low density polyethylene (LDPE), triacontane and ethylene vinyl acetate polymer (which were used to cross-link the polymer), and two natural vegetable waxes from Brazil and

Mexico. However, XLPE mesh 35 has a reddish color and XLPE mesh 50 a greenish color. We believe this phenomenon is due to different concentrations in waxes.

Wollastonite is mainly composed of calcium and silicon oxide.

65 Figure 32: FTIR curves for XLPE mesh 35 and 50

Figure 33: FTIR curve for wollastonite

66 4.4.1.2 Differential Scanning Calorimetry (DSC)

Fillers were also tested in differential scanning calorimetry to determine their influence on the glass transition of neat PC.

Figure 34 shows the DSC curves for both XLPEs and wollastonite. Looking at it, we can see an endothermic peak around 110°C for both fillers. Theses peaks correspond to the melting point of XLPEs 4, 5. Therefore, XLPEs are not entirely cross-linked. For XLPE mesh

35, we notice small endothermic bumps, or shoulders, around 65°C and 80°C corresponding to the thermal history of the sample 4. Finally, there is another endothermic peak at 125°C for XLPE mesh 50, which does not appear for XLPE mesh 35. This peak was previously observed for neat XLPE mesh 50.

In the temperature range considered, we do not see any thermal transitions for wollastonite.

67 Endo 

Figure 34: DSC curves for wollastonite and XLPE mesh 35 and 50

Figure 35 shows the DSC curves obtained for unfoamed PC mixes. For samples mixed with 10 or 20 % of XLPE mesh 35, we notice an endothermic peak at 110°C. We explained earlier that this peak corresponds to the melting point of XLPE. Then, we can see the glass transition of PC occurring at 145°C. For samples mixed with 10 or 20 % of XLPE mesh 50, we observe the same endothermic peak at 110°C. Moreover, we can see a small bump at 125°C corresponding to the peak previously observed for XLPE mesh 50. Once again, the glass transition of PC is visible at 145°C. Finally, we can se the glass transition of PC at 145°C for samples mixed with 10 % of wollastonite and 135°C for samples with 20 % of wollastonite.

Therefore, adding 20 % of wollastonite to the PC matrix modifies its thermal properties.

68 Endo 

Figure 35: DSC curves for unfoamed PC mixes

4.4.1.3 Compression testing

PC mixes were tested in compression to study the effect of the fillers on the samples before foaming.

Figure 36 shows the strain-stress curves obtained for 10 and 20 % of each filler. Once again, we can see the same regions (linear elastic region, plastic region and strain hardening) observed on the strain-stress curves for PC foamed in supercritical CO2 and chemically with

5-phenyl-1H-terazole. (see section 2.4.3). Except mixes with XLPE mesh 35 and 20 % wollastonite, most mixes require more stress than the spool from Sunoco to reach 0.5 mm/mm. However, they all exhibit lower mechanical properties than neat PC.

69 Figure 36: Compressive testing strain-stress curves for solid polymers and PC mixes

Figure 37 shows the compression modulus obtained for PC mixes. We can see that the compression moduli are similar for every mixes. For both XLPEs, the percentage in filler inside the polymer matrix does not influence the mechanical properties in compression since samples with 10 or 20 % have the same modulus. This observation does not concur with our expectations since we thought mechanical properties would increase with the percentage of

XLPE. Finally, samples mixed with 10 % wollastonite have the best modulus. This value drastically drops when the percentage of wollastonite is increased to 20 %. The general drop in modulus compared to neat PC can be attributed to several factors. Firstly, we believe fillers did not melt (because the melting point of wollastonite is much higher than 220°C and

XLPEs are cross-linked) and blend in the matrix, creating flaws and weak points. Then, some

70 air bubbles might have formed around the fillers during processing in compression molding, making the material weaker.

Figure 37: Compression modulus for solid polymers and unfoamed PC mixes

4.4.2 Polycarbonate mixes foamed in supercritical CO2

4.4.2.1 Differential Scanning Calorimetry (DSC)

Figure 38 shows the DSC curves obtained for PC mixes with 10-20% of XLPE mesh 35 or 50 and foamed in supercritical CO2. After physical foaming, the same endothermic peaks are visible for both XLPEs at 110°C. For XLPE mesh 50, there is once again a small bump at

125°C. However, we cannot see the glass transition of PC, which was previously occurring at

71 145°C, in the temperature range considered. DSC experiments with a larger temperature range would have to be performed to determine the temperature range where the glass transition is now occurring. This observation means that the combination of XLPEs and the foaming process in supercritical CO2 modified the thermal properties of PC. Indeed, we were previously able to see the glass transition of PC at 145°C when neat PC was foamed physically or when PC was mixed with XLPE mesh 35 and 50 without foaming.

Endo 

Figure 38: DSC curves for PC mixes with 10-20% of XLPE mesh 35 or 50 and foamed in supercritical CO2

Figure 39 shows the DSC curves for PC mixes with 10 or 20 % of wollastonite and foamed in supercritical CO2. Once again, the combination of wollastonite and physical foaming in supercritical CO2 modified the thermal properties of PC. Indeed, we can see a

72 shift in glass transition for both mixes. Now, the glass transition occurs at 120°C for 10 % wollastonite, instead of 145°C before foaming. For 20 % wollastonite, the glass transition shifted from 135°C to 100°C. The bump occurring at 170°C for 20 % of wollastonite is due to the thermal history of the sample.

Endo 

Figure 39: DSC for PC mixes with 10 or 20 % of wollastonite foamed in supercritical CO2

73 4.4.2.2 Scanning Electron Microscopy (SEM)

Scanning electron microscopy (SEM) was performed on our foams to study the influence of the fillers on their structure and morphology. Images were taken at 150x and

1000x to compare the cell size, cell density and distribution in our samples.

Figure 40 shows the micrographs obtained in SEM for samples mixed with 10-20 % of

XLPE mesh 35 and foamed in supercritical CO2. Figure 40 shows the cell distribution for both mixes. Samples foamed in supercritical CO2 with XLPE mesh 35 exhibit a very different structure than neat PC foams. Indeed, we do not have an opened-cell structure anymore.

The average cell size is much smaller since we have an average diameter of 0.650.2 m for

10 % of XLPE mesh 35 and 0.550.3 m for 20 % of filler. Looking at figure 41, we can see that the cell size distribution is larger than neat PC foams. Finally, we calculated a cell density of 1.4 x 1019 cells/cm3 for 10 % of XLPE mesh 35 and 1.2 x 1019 cells/cm3 for 20 % of

XLPE mesh 35. As explained in the literature, the presence of fillers, here XLPE mesh 35, into the polymer matrix decreased pore diameter and increased the cell density 1.

74 400 m 400 m

A B

50 m 50 m

C D

Figure 40: SEM micrographs for PC mixes foamed in supercritical CO2 with 10 % (A, C) and 20 % (B, D) of XLPE mesh 35 with magnification 150x (A, B) and 1000x (C, D)

Figure 41: Cell size distribution for PC mixes with 10 % (on the left) or 20 % (on the right) of XLPE mesh 35

75 Figure 42 shows the micrographs obtained in SEM for samples mixed with 10-20 % of

XLPE mesh 50 and foamed in supercritical CO2. Figure 43 shows the cell distribution for both mixes. Like XLPE mesh 35, foams with XLPE mesh 50 exhibit a closed-cell structure. The average cell size is also smaller since we have an average diameter of 0.700.45 m for 10 % of XLPE mesh 50 and 0.450.3 m for 20 % of filler. Looking at figure 43, we can see that the cell size distribution is larger than neat PC foams. Finally, we calculated a cell density of 5.0 x

1019 cells/cm3 for 10 % of XLPE mesh 50 and 7.4 x 1019 cells/cm3 for 20 % of XLPE mesh 50.

Once again, the presence of fillers resulted into a smaller pore size and a greater cell density compared to neat PC foams and foams with XLPE mesh 35. This trend can be explained by the fact that XLPE mesh 50 has a smaller diameter than XLPE mesh 35.

For both XLPEs, the average pore size decreases with increasing percentage of filler.

Then, around some particles of both XLPE mesh 35 and 50, we can see a void that could have been created by a gas bubble during processing in compression molding or supercritical CO2 during the foaming process. These voids could have an impact on the mechanical properties of the foams. Indeed, such voids could represent a weak point in the polymeric matrices and drastically decrease their compression modulus.

76 400 m 400 m

A B

50 m 50 m

C D

Figure 42: SEM micrographs for PC mixes foamed in supercritical CO2 with 10 % (A, C) and 20 % (B, D) of XLPE mesh 50 with magnification 150x (A, B) and 1000x (C, D)

77 Figure 43: Cell size distribution for PC mixes with 10 % (on the left) or 20 % (on the right) of XLPE mesh 50

Figure 44 shows the micrographs obtained in SEM for samples mixed with 10-20 % of wollastonite and foamed in supercritical CO2. Figure 45 shows the cell distribution for both mixes. Firstly, we notice that wollastonite particles have a rectangular shape, unlike XLPEs which have a spherical shape. Foams with wollastonite also exhibit a closed-cell structure with slightly bigger pores. The average cell size is 0.850.4 m for 10 % of wollastonite and

0.550.2 m for 20 % of wollastonite. Looking at figure 45, we can see that the cell size distribution is larger than the one for previous foams. Finally, we calculated a cell density of

2.4 x 1019 cells/cm3 and 8.8 x 1018 cells/cm3 for respectively 10 % and 20 % of wollastonite.

Wollastonite particles being smaller than XLPEs particles, we expected a smaller pore size, a greater cell density and a more homogeneous distribution. None of the expected properties occurred. We believe this phenomenon is the result of the shape of wollastonite particles.

However, every mix follows the same trend: the addition of filler decreases pore size and increases cell density.

78 400 m 400 m

A B

50 m 50 m

C D

Figure 44: SEM micrographs for PC mixes foamed in supercritical CO2 with 10 % (A, C) and 20 % (B, D) of wollastonite with magnification 150x (A, B) and 1000x (C, D)

Figure 45: Cell size distribution for PC mixes with 10 % (on the left) or 20 % (on the right) of wollastonite

79 Table 9 summarizes all the data previously described for PC mixes foamed in supercritical CO2:

Table 9: Overall properties for PC mixes foamed in supercritical CO2

Samples Average pore size (m) Cell density Nc (cells/cm3)

Neat PC 48h 3.71.58 4.1 x 1018

10 % XLPE mesh 35 0.650.2 1.4 x 1019

20 % XLPE mesh 35 0.550.3 1.2 x 1019

10 % XLPE mesh 50 0.700.4 5.0 x 1019

20 % XLPE mesh 50 0.450.3 7.4 x 1019

10 % wollastonite 0.850.4 2.4 x 1019

20 % wollastonite 0.550.2 8.8 x 1018

Figure 46 shows the change in volume of the samples after foaming. The overall change in volume after foaming process is smaller compared to neat PC foams. This phenomenon is the result of a smaller pore size. Moreover, the change in volume for every mixes and concentrations are similar.

80 Figure 46: Volume for PC mixes foamed in supercritical CO2

The resolution of x-ray tomography micro-CT being about 6 m, we could not determine the porosity of the foams using this method. Therefore, we estimated their porosity using ASTM D792 based on the Archimedes’ principle. Table 10 shows the porosity

air, calculated using the following equation:

휌푓 휈푎푖푟 = 1 − (5) 휌푝

81 Table 10: Density and porosity for PC mixes foamed in supercritical CO2

3 Samples Density 흆풇 (g/cm ) Porosity air (%)

10% XLPE mesh 35 1.030.3 8.11.1

20% XLPE mesh 35 1.090.5 6.90.8

10% XLPE mesh 50 1.030.2 6.31.2

20% XLPE mesh 50 1.040.3 5.61.5

10% wollastonite 1.020.1 10.10.5

20% wollastonite 1.110.4 8.30.7

Looking at table 10, we first notice that the porosity after foaming for PC mixes is smaller than neat PC foams saturated for 48 h. These results correlate with the diminution in volume previously observed after foaming.

4.4.2.3 Compression testing

Figure 47 shows the strain-stress curves obtained for PC mixes. Once again, we can see the same regions (linear elastic region, plastic region and strain hardening) observed on the strain-stress curves for PC foamed in supercritical CO2 and chemically with 5-phenyl-1H- terazole. Compared to unfoamed PC mixes, we can see that besides 20 % XLPE mesh 35, every mixes require a greater stress than the spool to reach a strain of 0.5 mm/mm.

82 Figure 47: Compressive testing strain-stress curves for solid polymers and PC mixes foamed in supercritical CO2

Figure 48 shows the compression modulus obtained for PC mixes foamed in supercritical CO2. As observed on the previous figure, the foaming process enhanced the mechanical properties of the samples since the compression modulus is greater after foaming (see figure 36). Indeed, the modulus of each sample increased, except for 10 % wollastonite. However, the mechanical properties of foams with fillers remain lower than the ones for neat PC foams. We assume this phenomenon is the result of the particle size.

Indeed, the addition of nanoparticles into polymeric foams should enhance their mechanical and physical properties by reinforcing the cell walls. In our case, we used microparticles.

Therefore, fillers were too big to go into the pores and reinforce the walls. Moreover, we

83 noticed on the micrographs obtained in SEM that some fillers were surrounded by voids, which could have been created by a gas bubble during processing in compression molding or supercritical CO2 during the foaming process. These voids could be a weakness in the material and have an impact on the mechanical properties of the foams.

Figure 48: Compression modulus for solid polymers and foamed PC mixes in supercritical CO2

84 4.4.3 Polycarbonate mixes foamed chemically with 5-phenyl-1H-tetrazole

4.4.3.1 Differential Scanning Calorimetry (DSC)

Figure 49 shows the DSC curves obtained for PC mixes with 10-20% of XLPE mesh 35 or 50 and foamed chemically with 5-phenyl-1H-tetrazole. The DSC curves are similar to the ones obtained physical foaming. Indeed, we can see the same endothermic peaks for both

XLPEs at 110°C and at 125°C for XLPE mesh 50. Moreover, we cannot see the glass transition of PC in the temperature range considered (see section 4.4.2.1).

Endo 

Figure 49: DSC curves for PC mixes with 10-20% of XLPE mesh 35 or 50 and foamed with 5-phenyl-1H-tetrazole

85 Figure 50 shows the DSC curves PC mixes with 10 or 20 % of wollastonite and foamed chemically with 5-phenyl-1H-tetrazole. Once again, the DSC curves are very similar to the ones obtained in physical foaming. We can see a shift in glass transition for both mixes. Now, the glass transition occurs at 120°C for 10 % wollastonite, instead of 145°C before foaming.

For 20 % wollastonite, the glass transition shifted from 135°C to 100°C (see section 4.4.2.1).

Endo 

Figure 50: DSC for PC mixes with 10 or 20 % of wollastonite and foamed with 5- phenyl-1H-tetrazole

The fact that the DSC curves for PC mixes foamed chemically or physically are similar should have been expected. Indeed, when foamed without any filler, the foaming process did not affect the thermal properties of PC. However, we observed a change in thermal

86 properties when unfoamed PC was mixed with fillers. Therefore, fillers induce a change in the physical properties of PC.

4.4.3.2 Scanning Electron Microscopy (SEM)

Images were taken at 30x and 150x to compare the cell size, cell density and distribution in our samples.

Figure 51 shows the micrographs obtained in SEM for samples mixed with 10-20 % of

XLPE mesh 35 and foamed chemically. Figure 52 shows the cell distribution for both mixes.

Samples foamed chemically with XLPE mesh 35 exhibit a very similar structure compared to neat PC foams with 5-phenyl-1H-tetrazole. Indeed, both foams have a closed-cell structure with a large cell distribution. However, the average cell size is smaller. For 10 % of XLPE mesh 35, the average cell size is 63.538m and 57.539 m for 20 % of filler. Finally, we calculated a cell density of 4.3 x 1011 cells/cm3 for 10 % of XLPE mesh 35 and 5.5 x 1011 cells/cm3 for 20 % of XLPE mesh 35. As explained in the literature, the presence of fillers, here XLPE mesh 35, into the polymer matrix decreased pores diameter, and increased the cell density 1.

87 2 mm 2 mm

A B

400 m 400 m

C D

Figure 51: SEM micrographs for PC mixes foamed chemically with 10 % (A, C) and 20 % (B, D) of XLPE mesh 35 with magnification 30x (A, B) and 150x (C, D)

Figure 52: Cell size distribution for PC mixes with 10 % (on the left) or 20 % (on the right) of XLPE mesh 35

88 Figure 53 shows the micrographs obtained in SEM for samples mixed with 10-20 % of

XLPE mesh 50 and foamed in supercritical CO2. Figure 54 shows the cell distribution for both mixes. Like XLPE mesh 35, foams with XLPE mesh 50 exhibit a closed-cell structure with a random cell distribution. The average cell size is 9865 m for 10 % of XLPE mesh 50 and

65.543 m for 20 % of filler. For 10 % XLPE mesh 50, the cell density is 4.7 x 1011 cells/cm3 and 3.9 x 1011 cells/cm3 for 20 % of XLPE mesh 50. Once again, the presence of fillers resulted into a smaller pore size and a greater cell density compared to neat PC foams.

2 mm 2 mm

A B

400 m 400 m

C D

Figure 53: SEM micrographs for PC mixes foamed chemically with 10 % (A, C) and 20 % (B, D) of XLPE mesh 50 with magnification 30x (A, B) and 150x (C, D)

89 Figure 54: Cell size distribution for PC mixes with 10 % (on the left) or 20 % (on the right) of XLPE mesh 50

Figure 55 shows the micrographs obtained in SEM for samples mixed foamed chemically with 10-20 % of wollastonite. Figure 56 shows the cell distribution for both mixes.

Foams with wollastonite also exhibit a closed-cell structure with a large cell distribution. The average cell size is similar to foams with XLPEs with an average diameter of 106.565 m for

10 % of wollastonite and 85.548 m for 20 % of wollastonite. The cell density for 10% wolastonite is 8.1 x 1011 cells/cm3 and 9.4 x 1011 cells/cm3 for 20 % of filler. Wollastonite particles being smaller than XLPEs particles, we expected a smaller pore size, a greater cell density. However, the average cell size is similar. We believe this phenomenon is the result of the shape of wollastonite particles. This trend was also observed for PC mixes foamed in supercritical CO2.

Finally, we can see that wollastonite particles are present around and inside the pores. This phenomenon could potentially increase the compression modulus of the foams since they could reinforce the walls of the pores 1.

90 When comparing these results with samples foamed with supercritical CO2, we notice the same trends. Indeed, fillers decreased the average cell diameter and increased the cell density compared to regular foams. However, pore size is a lot larger and cell density a lot smaller for samples foamed chemically.

2 mm 2 mm

A B

400 m 400 m

C D

Figure 55: SEM micrographs for PC mixes foamed chemically with 10 % (A, C) and 20 % (B, D) of wollastonite with magnification 30x (A, B) and 150x (C, D)

91 Figure 56: Cell size distribution for PC mixes with 10 % (on the left) or 20 % (on the right) of wollastonite

Table 10 regroups all the data previously described for PC mixes foamed chemical with 5- phenyl-1H-tertazole:

Table 11: Overall properties for PC mixes foamed chemically

Samples Average pore size (m) Cell density Nc (cells/cm3)

Neat PC 48h 13090 3.7 x 109

10 % XLPE mesh 35 63.538 4.3 x 1011

20 % XLPE mesh 35 57.539 5.5 x 1011

10 % XLPE mesh 50 98.65 4.7 x 1011

20 % XLPE mesh 50 65.543 3.9 x 1011

10 % wollastonite 106.565 8.1 x 1011

20 % wollastonite 85.548 9.4 x 1011

92 Figure 57 shows the change in volume of the samples after foaming. The overall change in volume after foaming process is smaller compared to neat PC foams. This phenomenon is the result of a smaller pore size. Moreover, the change in volume similar for every mixes.

Figure 57: Volume for PC mixes foamed chemically

93 4.4.3.3 Compression testing

Figure 58 shows the strain-stress curves obtained for PC mixes foamed chemically.

Once again, we can see the same regions (linear elastic region, plastic region and strain hardening) observed on the strain-stress curves for PC foamed in supercritical CO2 and chemically with 5-phenyl-1H-terazole. Moreover, we do not have a flat plateau (see section

2.4.3). Compared to PC foamed with 3% of 5-phenyl-1H-tetrazole quenched in water, we can see that every PC mixes exhibit lower mechanical properties. However, foams are less brittle are the addition of fillers and a strain of 0.5 mm/mm can be reached.

Figure 58: Compressive testing strain-stress curves for solid polymers and PC mixes foamed chemically

94 Figure 59 shows the compression modulus obtained for PC mixes foamed chemically.

Unlike PC mixes foamed chemically, the addition of fillers to the polymeric matrices did not enhance their mechanical properties. Indeed, we can see that the compression modulus drastically dropped compared to neat PC foamed chemically. After characterizing our samples in SEM, an increase in modulus was expected due to the decrease in cell size and increase in cell density induced by the addition of fillers. We believe this phenomenon is the result of the large and random cell distribution (diameter and position) of the foams.

Moreover, the processing conditions in the oven caused the formation of large voids in the polymer, resulting in lower mechanical properties.

Figure 59: Compression modulus for solid polymers and PC mixes foamed chemically

95 4.5 Conclusions

Before testing PC mixes foams, each filler was tested in FTIR and DSC to study its behavior and understand its effect on PC. Firstly, FTIR showed that XLPE mesh 35 and XLPE mesh 50 had the same composition.

When mixing PC with 10-20% of fillers, we observed a change in physical properties.

Indeed, DSC experiments showed a shift in glass transition for samples mixed with wollastonite. Furthermore, for samples mixed with XLPEs, characteristics endothermic peaks of XLPE were observed, as well as the glass transition of PC at 145°C. Finally, we noticed a large drop in compression modulus compared to the unfoamed neat PC.

After foaming PC mixes physically or chemically, they showed a larger shift in glass transition for samples mixed with wollastonite. Moreover, the glass transition of XLPE mixes could not be observed in the range of temperature considered. SEM micrographs showed that for both foaming methods, the addition of fillers caused a decrease in pore size and an increase in cell density, leading to a lower porosity and change in volume. Finally, PC mixes foamed in supercritical CO2 showed a higher compression modulus than the unfoamed samples. However, the mechanical properties remain lower than expected due to the size of the fillers. On the other hand, PC mixes foamed chemically exhibit very low mechanical properties due to processing issues.

96 4.6 References

1 L. J.M. Jacobs, M. F. Kemmere, J. T.F. Keurentjes. “Sustainable polymer foaming using high-pressure carbon dioxide: a review on fundamentals, processes and applications” Green

Chem., Vol.10 (2008): 731-738.

2 L. J. Lee, C. Zeng, X. Cao, X. Han, J. Shen, G. Xu. “Polymer nanocomposite foams” Compos.

Sci. Technol., Vol. 65 (2005): 2344-2363

3 W. Zhai, J. Yu, L. Wu, W. Ma, J. He. “Heterogeneous nucleation uniformizing cell size distribution in microcellular nanocomposite foams” Polymer, Vol.47 (2006): 7580-7589

4 L. Boukezzi, A. Boubakeur, C. Laurent, M. Lallouani. “ DSC study of artificial thermal aging of XLPE insulation cables” 2007 International Conference on Solid Dielectrics, Winchester

(UK)

5 J. A. Diego, J. Belena, J. Orrit, J. Sellarès, M. Mudarra, J. C. Canadas. “TDSC study of XLPE recrystallization effects in the melting range of temperatures” J. of Applied Physics, Vol. 39

(2006): 1932-1938

97 Chapter 5

Conclusions

The purpose of this project was to develop an “indestructible” material made of polycarbonate (PC) for industrial purposes. In order to reduce the amount of material and the manufacturing costs, samples were foamed either physically in supercritical CO2 or chemically with 5-phenyl-1H-tetrazole. For physical foaming, three saturation times were studied. Two quenching methods and different percentages of chemical foaming agent (CFA) were studied for chemical foaming.

The first stage of this work was to study the influence of the foaming process on PC.

Samples were characterized in DSC, SEM and compression. DSC experiments showed that none of the foaming methods had an influence on the glass transition of the material once foamed. However, we saw that during the foaming process in supercritical CO2, the glass transition dropped to 60°C after oversaturation of the sample in CO2. After thermally initiating cell nucleation and growth, the glass transition shifted back to 145°C. Micrographs taken in SEM showed that samples foamed physically and chemically had very different structures. Indeed, samples foamed in supercritical CO2 exhibited a microcellular opened- cell structure with a high cell density and a homogeneous cell distribution. We also demonstrated that the limit of CO2 into PC was reached after 48 h. On the other hand, samples foamed with 5-phenyl-1H-tetrazole had a macrocellular closed-cell structure with a smaller cell density and a random cell distribution. As a consequence, supercritical samples had a much higher compression modulus. However, the mechanical properties remained lower than neat PC, which was used as a reference. We believe this phenomenon is the 98 result of the pore size, which is larger than the flaw, and micro bubbles created in the polymeric matrices during processing in compression molding. Once characterized, specific parameters were selected for both methods: a saturation time of 48 h for physical foaming and an addition of 3 % of CFA with a quenching in water for chemical foaming.

The second stage of this work consisted in the addition of fillers to the polymeric matrices to enhance their physical and mechanical properties. XLPE mesh 35 and 50 were from miscellaneous products supplied by Encore Wire Corporation. Wollastonite was also selected due to impressive results in previous studies. Firstly, each filler was tested in FTIR and DSC. FTIR showed that XLPE mesh 35 and XLPE mesh 50 had a similar composition. Then,

10 to 20% of fillers were mixed to unfoamed PC. DSC results showed a change in physical properties caused by the fillers. PC mixes with wollastonite experienced a shift in glass transition and samples mixed with XLPEs exhibited characteristic endothermic peaks previously observed for XLPEs. Finally, we noticed a large drop in compression modulus compared to unfoamed neat PC.

The last stage of this work consisted in the foaming of PC mixes. DSC experiments showed an even more important shift in glass transition for wollastonite mixes. Moreover, the glass transition for XLPEs mixes could not be observed in the range of temperature considered. SEM micrographs showed that for both foaming method, the addition of fillers induced a general decrease in pore size, an increase in cell density and a lower porosity.

Finally, samples were tested in compression. An increase in compression modulus was observed for PC mixes after foaming in supercritical CO2. However, the mechanical properties remained lower than the ones for foamed neat PC due to the micro size of the

99 fillers. On the other hand, a drastic drop in compression modulus was observed for samples foamed with 5-phenyl-1H-tetrazole. We believe this phenomenon is the result of the processing in the oven where large voids formed in the matrix.

100