Polymeric Gas Separation Membranes For

Home , Membrane gas separation

POLYMERIC GAS SEPARATION MEMBRANES FOR

CARBON DIOXIDE REMOVAL

XIA JIAN ZHONG

(B. S., Peking University, P. R. China)

A THESIS SUBMITTED FOR THE DEGREE OF

DOCTOR OF PHILOSOPHY

NUS Graduate School for Integrative Sciences and

Engineering

NATIONAL UNIVERSITY OF SINGAPORE

NOV 2012

To my parents and my wife for their understanding and support without any hesitation

Especially to my father for his selfless love until he left this world

ACKNOWLEDEGMENTS

I wish to take this opportunity to express my sincere appreciation to all the contributors during my years in the National University of Singapore and University of Texas at Austin. First of all, I am especially grateful to my supervisors, Professor

Neal Chung Tai-Shung and Professor Donald R. Paul, who have not only provided guidance during my research activities but have also given generously of their time to offer encouragement, advice and support. Their pursuance for perfection in research and publication set a great example for my professional career. I also appreciate the assistance from my TAC members – Prof. Hong Liang and Dr. Pramoda Kumai – for their valuable comments and discussions. I would like to acknowledge the NGS scholarship offered by NUS Graduate School for Integrative Sciences and

Engineering. They provide me lots of chance to attend international conferences, summer schools and even long period of research exchange. I also wish to express my recognition to NUS, A*Star and the Singapore National Research Foundation (NRF) for the financial support that enables this work to be successfully completed.

It has been pleasant to work with people both in Prof. Chung’s group in National

University of Singapore and people in Prof. Paul’s group in University of Texas at

Austin. I have enjoyed the friendships with all members of these two groups, especially Dr. Liu Songlin, Dr. Norman Horn, Dr. Xiao Youchang, Dr. Li Yi, Dr.

Rajkiran Tiwari, Ms. Wang Huan, Ms. Zhang Sui, Mr. Chen Hangzheng, Mr. Yin

Hang and many others for many good times, discussion and sharing of technical

experience. Special thanks to Ms. Chuan Irene Christina for all her kindest cooperation and help on my “2+2” exchange programme. I am also indebted to Liu Di,

Zhang Miao, Liu Jingran, Tang Zhao, Chen Xi, Yang Shengyuan for making my graduate life joyful. Finally, I must express my deepest gratefulness to my family for their endless support, especially to my dearest fiancee Yaqian for sharing my life in

Singapore.

TABLE OF CONTENTS

ACKNOWLEDEGMENTS ...... i

TABLE OF CONTENTS ...... iii

SUMMARY ...... ix

LIST OF TABLES ...... xii

LIST OF FIGURES ...... xiv

CHAPTER 1 Introduction...... 1

1.1 Membrane Technology for Gas Separations ...... 2

1.2 History of Gas Separation Membranes ...... 3

1.3 Applications Based on Gas Separation Membranes ...... 4

1.3.1 Hydrogen recovery ...... 5

1.3.2 Nitrogen Enrichment ...... 7

1.3.3 Recovery of Organic Vapor ...... 7

1.3.4 Carbon Dioxide Capture ...... 8

1.4 Materials for Gas Separation Membranes ...... 11

References ...... 17

CHAPTER 2 Background and Approaches ...... 22

iii

2.1 Permeability, Permeance and Selectivity ...... 22

2.2 Solubility ...... 24

2.3 Fractional Free volume ...... 26

2.4 Gas Transport in Rubbery Polymers ...... 27

2.5 Gas Transport in Glassy Polymers ...... 28

2.6 Effect of Temperature ...... 29

References ...... 30

CHAPTER 3 Materials and Experimental Methods ...... 33

3.1 Materials ...... 33

3.2 Preparation of Dense Membranes ...... 35

3.2.1 Preparation of Glassy Thick Membranes ...... 35

3.2.2 Preparation of Organic-Inorganic Membranes (OIMs) ...... 36

3.3 Preparation of Polymeric Thin Films ...... 39

3.4 Characterization of Physicochemical Properties ...... 41

3.4.1 Measurement of Gel Content ...... 41

3.4.2 Fourier Transform Infrared Spectrometer (FTIR) ...... 42

3.4.3 Transmission Electron Microscopy (TEM) ...... 42

3.4.4 Thermogravimetric Analysis (TGA) ...... 42

3.4.5 Wide Angle X-ray Diffraction (WAXD) ...... 43

3.4.6 X-ray Photoelectron Spectrometer (XPS) ...... 43

3.4.7 Elemental Analysis ...... 43

3.4.8 Nuclear Magnetic Resonance (NMR) ...... 44

3.4.9 Simulation Based on Molecular Dynamic ...... 44

3.4.10 Variable Angle Spectroscopic Ellipsometer ...... 46

3.5 Characterization of Gas Transport Properties ...... 47

3.5.1 Pure Gas Permeation Tests ...... 47

3.5.2 Mixed Gas Permeation Tests ...... 48

3.5.3 Pure Gas Sorption Tests ...... 48

References ...... 50

CHAPTER 4 Liquid-like Polyethylene Glycol Supported in the Organic-inorganic

Matrix for CO2 Removal ...... 53

Abstract ...... 54

4.1 Introduction ...... 55

4.2 Results and Discussion ...... 60

4.2.1 Basic Physicochemical Properties ...... 60

4.2.2 XRD Characterization ...... 67

4.2.3 The Gas Permeation Performance ...... 68

4.2.5 Effect of Testing Temperature ...... 77

4.2.5 Effect of PEGs’ Molecular Weights ...... 81

Summary ...... 85

References ...... 87

CHAPTER 5 The Effect of End Groups and Grafting on the CO2 Separation

Performance of Polyethylene Glycol Based Membranes ...... 96

Abstract ...... 97

5.1 Introduction ...... 98

5.2 Results and Discussion ...... 99

5.2.1 Basic Physicochemical Properties ...... 99

5.2.2 Gas Transport Properties of OIMs with Physical Blending ...... 101

5.2.3 Thermal Properties of GPA1100 Series ...... 105

5.2.4 Temperature Dependence of Gas Permeation Properties ...... 107

5.2.5 Thermal Grafting of PEG-azide and Characterizations ...... 112

5.2.6 Gas Permeation Properties After Thermal Grafting ...... 116

Summary ...... 119

References ...... 121

CHAPTER 6 Aging and Carbon Dioxide Plasticization of Thin Extem® XH1015

Polyetherimide Films ...... 125

Abstract ...... 126

6.1 Introduction ...... 127

6.2 Results and Discussion ...... 130

6.2.1 Aging Behavior Tracked by Gas Permeation ...... 130

6.2.2 CO2 Plasticization Pressure Curves ...... 135

6.2.3 CO2 Permeability Hysteresis ...... 140

6.2.4 CO2 Permeation Behavior for Short Exposure Times ...... 145

6.2.5 CO2 Permeation Behavior over Long Exposure Times ...... 147

Summary ...... 151

References ...... 153

CHAPTER 7 Gas Permeability Comparison of Extem® XH1015 with Polysulfone

and Ultem® via Molecular Simulation ...... 161

Abstract ...... 162

7.1 Introduction ...... 163

7.2 Results and Discussion ...... 166

7.2.1 Chain Morphology Comparison ...... 166

vii

7.2.2 Molecular Simulation ...... 169

Summary ...... 174

References ...... 175

CHAPTER 8 Conclusions and Recommendations ...... 177

8.1 Conclusions ...... 177

8.1.1 Permeability and Selectivity Enhancement by Blending PEG and its

Derivatives ...... 177

8.1.2 Permeability Enhancement by Grafting PEG-azide on the Backbone

of OIMs 178

8.1.3 Temperature Effect on Gas Permeability and Selectivity ...... 178

8.1.4 Physical Aging and Plasticization on Polymeric Thin Films ...... 179

8.2 Recommendations ...... 180

8.2.1 PEG Based Organic-Inorganic Membranes ...... 180

8.2.2 Physical Aging and Plasticization Monitored by Gas Permeability .... 181

Appendix A: Structure Determination of Extem® XH 1015 ...... 183

Results ...... 183

References ...... 188

viii

SUMMARY

Membrane technology has been considered as one of the most promising

candidates for selective removal of carbon dioxide from mixture with H2 and N2.

Conventional glassy membranes focused mainly on the size sieving ability of

polymers. Based on the different size of CO2, H2 and N2, the flux order of these

three gases in glassy polymers is usually H2 ˃ CO2 ˃ N2 and these membranes are

H2-selective. However, the separation mechanism in rubbery membranes, especially in poly (ethylene glycol) (PEG), is different due to the higher contribution on solubility selectivity, which means the size sieving effect is not the dominate factor. Therefore, the flux order of these three gases in rubber is usually

CO2 ˃ H2 ˃ N2 and these membranes are CO2-selective.

Part of this project focused on exploring the possibility of using favorable

interactions between CO2 and ethylene oxide (EO) groups to improve permeability/selectivity properties of rubbery membranes. Organic-inorganic membranes (OIMs) consisting of siloxane network and PEG segments were used as the substrate. Several PEG and PEG derivatives with different molecular weight were physically blended into the substrate before the siloxane network was formed. The membrane containing 60wt% of 1000g/mol PEG could achieve an

ultra-high CO2 permeability of 845 Barrer with CO2/H2 and CO2/N2 permselectivity around 10 and 40, respectively. A PEG derivative is blended into the substrate

followed by thermal grafting. Ultra high CO2 permeability (982 barrer at 45 ºC) is

achieved via physical blending, while extremely high CO2 permeability (1840 barrer at 45 ºC) is obtained after chemical grafting. Neither of these two

modification methods shows the loss of CO2/H2 and CO2/N2 selectivity compared to the substrate. Melting and crystallization behaviors of these PEG and PEG derivatives are believed to significantly affect the overall gas permeation performance.

Another part of this project focused on glassy membranes, which is also one

of candidates for CO2 removal in industry. However, CO2 plasticization and physical aging on glassy membranes severely reduced their chances to be further developed. Industrial glassy gas separation membranes usually have selective dense layers with thicknesses around 100 nm. It has long been assumed that these thin layers have the same properties as thick (bulk) films. However, recent research has shown that thin films with such thickness experience accelerated physical aging relative to bulk films. Thin films made from Extem® XH 1015, a new commercial polyetherimide, have been investigated by monitoring their gas permeability. The permeability of the thin films is originally greater than the thick films but eventually

decreases well below the permeability of the thick film. The CO2 plasticization of x

Extem thin films is also explored using a series of exposure protocols that indicate

CO2 plasticization is a function of film thickness, aging time, exposure time, pressure and prior history. In order to further explore the structure/property relationship of glassy polymers, some simulation works based on molecular dynamics were also conducted.

LIST OF TABLES

Table 1-1 Most important polymers used in industrial gas separation membrane [10]

...... 12

Table 1-2 Progress of membranes for the O2/N2 separation (25°C) [10] ...... 12

Table 3-1 Chemical structures of polymer used in this study ...... 34

Table 3-2 Bulk properties of polymers used in this study...... 34

Table 3-3 Atom numbers and cell dimensions for PSU, Extem and Ultem

amorphous cells ...... 46

Table 4-1 Gel content (%) of GP w/o PEG and GPP series ...... 60

Table 4-2 Thermal properties of GPP series, GP w/o PEG and pure PEGs ...... 62

Table 4-3 Average size and area fraction of silica particles in GP w/o PEG and

GPP1500-60 ...... 67

Table 4-4 Pure gas permeability and selectivity for membranes with different

compositions at 35°C ,45°C and 55°Ca ...... 70

Table 4-5 Gas permeability, solubility and diffusivity coefficient results compared

with PDMS and PEO from other sources ...... 72

Table 4-6 Mixed gas permeability and selectivity for GPP1000-60 and

GPP1500-60 at 45°C ...... 77

xii

Table 5-1 Thermal properties of GPA1100 series ...... 101

Table 5-2 Pure gas permeability and selectivity of OIMs blended with PEG-azide

...... 102

Table 5-3 Solubility and diffusivity coefficients of GPA1100 and GPP1000 series

at 45 ºC ...... 104

Table 5-4 Pure gas permeability and selectivity of GPA1100 series before and after

grafting ...... 117

Table 7-1 Gas permeation performance of flat dense PSU, Extem and Ultem

membranes ...... 167

Table 7-2 Kinetic diameters of various gases ...... 171

xiii

LIST OF FIGURES

Figure 3-1 Synthetic route of Extem XH1015 ...... 33

Figure 3-2 Synthetic route of hybrid membranes ...... 37

Figure 4-1 Concept of liquid PEGs supported in the organic-inorganic matrix ... 59

Figure 4-2 STEM images of (a) GPP without free PEG and (b) GPP1500-60 and

Figure 4-3 STEM images of (a) hybrid membranes and (b) imaginary matrix

constructed with (c) different functional groups ...... 66

Figure 4-4 STEM images of hybrid membranes after imageJ analysis: (a) GPP

w/o PEG and (b) GPP1500-60 ...... 66

Figure 4-5 XRD patterns for (a) GPP400, (b) GPP1000, (c) GPP1500 and (d)

GPP2000 at room temperature ...... 68

Figure 4-6 Selected permeability/selectivity data map for (a) CO2/H2 and (b)

CO2/N2 separation at 35°C ...... 69

Figure 4-7 CO2 permeability of GPP series and GP w/o PEG at 35°C (blank) and

45°C (checked) ...... 74

Figure 4-8 Permselectivity at 35°C (solid) and 45°C (hollow) for CO2/H2 ...... 76

xiv

Figure 4-9 DSC curves for (a) GPP400, (b) GPP1000, (c) GPP1500 and (d)

GPP2000 ...... 76

Figure 4-10 Change of CO2 permeability of GPP1500-60 from 30°C to 55°C

80*Tm2o is the onset melting temperature of PEG1500 in GPP1500-60

during 2nd heating...... 80

Figure 4-11 CO2 Sorption isotherm (a) and solubility coefficient (b) of GPP1500 at

35°C and 45°C ...... 80

Figure 4-12 Sorption isotherm (a) and solubility (b) of GPP1500, PDMS [52]

semi-crystalline PEO and amorphous PEO [28] at 35°C ...... 81

Figure 4-13 Solubility (square) and diffusivity (circle) of hybrid membranes

containing 40wt% of PEG with different molecular weights at 45°C

...... 83

Figure 4-14 Crystallization of PEO segments and PEGs with different molecular

weights ...... 84

Figure 5-1 TEM micrography of GP w/o PEG-azide at (a) a low magnitude and (b)

a high magnitude ...... 100

Figure 5-2 Glass transition temperature shifts of GPA1100 series by DSC ...... 103

Figure 5-3 DSC curves of (a) 2nd heating and (b) 2nd cooling of GPA1100 series

...... 107

Figure 5-4 The temperature dependence of CO2 permeability of GPA1100 series

...... 109

Figure 5-5 Temperature dependences of CO2 sorption isotherms of (a)

GPA1100-20 and (b) GPA1100-40 ...... 111

Figure 5-6 Temperature dependence of (a) solubility and (b) CO2 diffusivity of

GPA1100-20 and GPA1100-40 ...... 112

Figure 5-7 Temperature dependence of (a) CO2/H2 and (b) CO2/N2 selectivity of

GPA1100-20 and GPA1100-40 ...... 112

Figure 5-8 13C solid state NMR spectra of (a) GPA1100-40 pristine and (b)

GPA1100-40 after thermal grafting ...... 114

Figure 5-9 N1s XPS data before and after thermal grafting for GPA1100-40 ... 115

Figure 5-10 TGA data of GPA1100-40 before thermal grafting ...... 115

Figure 5-11 CO2 solubility (open) and diffusivity (solid) of GPA1100-40 before

and after grafting ...... 118

Figure 5-12 Wide angle XRD patterns of GP w/o PEG-azide, GPA1100-40 before

and after the grafting at ambient temperature ...... 118

Figure 5-13 2nd cooling curve of GPA1100-20 and GPA1100-40 before and after

the grafting ...... 119

xvi

Figure 6-1 DSC thermograph showing the glass-transition temperature of Extem

...... 130

Figure 6-2 Oxygen permeability of (a) Extem films with similar thickness and (b)

Matrimid, PSF and Extem films as a function of aging time ...... 131

Figure 6-3 (a) Oxygen, (b) nitrogen, (c) methane permeability of Extem films with

different thickness as a function of aging time ...... 134

Figure 6-4 O2/N2 selectivity of Extem films as a function of aging time ...... 134

Figure 6-5 CO2 sorption isotherms for thick films at 35⁰C: Matrimid [54], PSF [55],

PPO [56], butyl rubber [53] ...... 137

Figure 6-6 Normalized CO2 permeability of thin films as a function of (a) pressure

(Matrimid, PPO and PSF data are taken from the literature [18]) and (b)

CO2 concentration ...... 139

Figure 6-7 (a) CO2 permeability and (b) normalized CO2 permeability as a function

of CO2 pressure for different aging times ...... 140

Figure 6-8 CO2 permeability hysteresis curves of thin films aged (a) 25 hr, (b)

100 hr and 150 hr ...... 142

Figure 6-9 CO2 permeability as a function of time at (a) 32 atm (after increase CO2

pressure)and (b) 4 atm (after decrease CO2 pressure) during hysteresis

testing ...... 144

xvii

Figure 6-11 (a) CO2 permeability and (b) normalized CO2 permeability during

long time CO2 exposure at different pressures ...... 148

Figure 6-12 (a) CO2 permeability and (b) normalized CO2 permeability as a

function of exposure time at 8 atm for different polymer films ..... 150

Figure 6-13 (a) CO2 permeability and (b) normalized CO2 permeability as a

function of exposure time at 32 atm for different polymer films ... 151

Figure 7-1 Chemical structures of PSU, Extem and Ultem ...... 166

Figure 7-2 Chain morphologies of PSU, Extem and Ultem with 5 repeat units . 166

Figure 7-3 Morphology of two polymer chains with 5 repeat units for PSU, Extem

and Ultem ...... 169

Figure 7-4 Fractional accessible volumes and relative FAV values of PSU, Extem

and Ultem, probed with different diameters ...... 170

Figure 7-5 Correlation between gas permeability and 1/FAV ...... 172

Figure 7-6 FAV ratios of Extem/PSU and Ultem/PSU probed by different

diameters ...... 174

xviii

CHAPTER 1 Introduction

Membrane technology covers almost every aspect of separation engineering, including solid-liquid (microfiltration, ultrafiltration) [1], ion-liquid (nanofiltration

[2], reverse osmosis [3], forward osmosis [4]), liquid-liquid (pervaporation [5]), liquid-gas (gas contactor [6]) and gas-gas separation (gas separation [7]).

Compared to the convention separation processes, e.g., distillation or extraction, membrane based separations are generally cost-effective, energy efficient and environmentally friendly. Moreover, membrane separation units are modular so that they are easy to install, operate and scale up. For example, the juice concentration process in food industry is now dominated by membrane technology due to the ability of membranes to remove water at room temperature [8]. Reverse osmosis and membrane bioreactors are also very popular processes, especially, in seawater desalination and waste water treatment, because of their high reliability and efficiency [9]. However, gas separation membranes are relatively less popular than other highly competitive technologies. Developing the usage of membranes in emerging gas separation applications is a must for researchers in this field. In this introductory chapter, several applications based on gas separation membranes will be reviewed, and some potential applications in carbon dioxide related separation will be discussed. Glassy, rubbery and organic-inorganic membranes

will be involved in the membrane fabrications and discussions.

1.1 Membrane Technology for Gas Separations

Due to the extremely small size of the separation targets, gas separation membranes are usually thin selective barriers between two gas phases. The gradient of the chemical potential due to the different gas concentrations in the two phases becomes the driving force of gas diffusion across the membrane.

Today, most of gas separation membranes are in the form of hollow fiber modules, with fewer being formed in spiral-wound modules. The hollow fiber modules are usually less expensive than the spiral-wound modules, while the latter are usually considered to be more reliable in terms of easy and cheap maintenance. In principle, the permeation and separation performance of gas membranes depend on four parameters [10]: (1) the material, which determines the intrinsic permeability and separation factor; (2) the membrane structure and thickness, which determine the permeance; (3) the membrane configuration, e.g., flat sheet or hollow fiber; and (4) the module and system design. Developing a material into a commercial product takes years to evaluate and refine the aforementioned parameters.

1.2 History of Gas Separation Membranes

Long before the first commercial gas separation membranes (named Prism) were introduced, people had already noticed the potential usage of membranes as gas separation tools. In 1829, Thomas Graham discovered the law of gas diffusion by using a tube with one end sealed with plaster of Paris [11]. Three years later,

Mitchell [12] reported for the first time that different gas molecules have different tendencies to pass through rubber membranes, which means the flux of each gas is different. Since then, lots of polymers have been studied extensively to look for their potential to be gas separation membranes. H. A. Daynes and R. M. Barrer are the pioneers in performing quantitative measurements of gas permeability by using the time-lag method. A number of permeability data had been obtained from lots of potential membrane materials [13]. However, the lack of technology to produce high performance and low cost modules postponed the applications of gas separation membranes, until Loeb and Sourirajan [14] invented a novel phase inversion method to cast asymmetric cellulose acetate membranes, which enables the reduction of effective membrane thickness from several micrometers into sub-micrometer level. The invention of high-flux anisotropic membrane modules in the forms of spiral-wound and hollow fiber further facilitated the development of gas separation membranes. In 1980, Permea delivered the first generation polysulfone hollow-fiber membranes for hydrogen recovery from purge gas

steams of ammonia plants. Soon after the success of Permea, Cynara (now part of

Natco), Separex (now part of UOP), and GMS (now part of Kvaerner) had commercialized cellulose acetate membranes for removing carbon dioxide from natural gas [15]. More recently, the PolarisTM membrane developed by MTR

(Membrane Technology and Research, Inc.), which is a thin film composite

membrane in spiral wound form, shows a CO2 permeance ten times higher than the conventional cellulose membranes [16].

1.3 Applications Based on Gas Separation Membranes

Generally speaking, membrane-based gas separation has become more and more important compared to the conventional gas separation technologies such as adsorption, absorption and cryogenic distillation [15]. Many common polymer materials, such as polydimethylsiloxane (PDMS), cellulose acetate (CA), polysulfone (PSF), polyethersulfone (PES), and polyimide (PI), had been fabricated into membrane modules and applied to various gas separation processes such as hydrogen recovery, nitrogen enrichment, recovery of volatile organic compounds (VOCs), and separation of acid gases in natural gas resources and steel industries [17].

1.3.1 Hydrogen recovery

Hydrogen is one of the most demanded gases in recovery processes due to its high value and ease of separation. The cost of hydrogen is highly related to energy prices since hydrogen is mostly produced by reforming natural gas or coal. Since the oil price increases almost every month in recent years, the importance of hydrogen recovery is highlighted to reduce the running cost of refinery plants.

Hydrogenation of unsaturated hydrocarbons and hydrotreating to remove sulfur from fuels are the two major consumptions of hydrogen [18]. During these processes, purge gases containing a high partial pressure of hydrogen are needed to remove inert gases from reactors. By using a membrane separation system, most of the purge hydrogen could be recovered.

Hydrogen is a clean energy carrier that has great potential to replace gasoline, which could sufficiently reduce the production of greenhouse gases and toxic exhaust gases. By using hydrogen as the energy resource, fuel cells could directly generate electricity with water as the only exhaust. The large demand of hydrogen in the near future requires capacity upgrades of current production plants.

Nowadays, the large industrial scale production of hydrogen occurs via steam methane reforming (SMR) followed by the water-gas shift (WGS) reaction [19].

These reactions are as follows:

m Cn Hm  nH 2O  nCO  (  n)H2 (1-1) 2

CO  H2O  CO2  H2 (1-2)

Purification technologies are crucial to the a hydrogen economy because the minimum purity requirement for the fuel cell is higher than 99.99%. To date, pressure swing adsorption (PSA) and cryogenic distillation are the widely accepted methods. However, both are energy-intensive processes and only applicable for the large scale manufacture. Membrane technology offers a bright vision for the hydrogen purification market because membrane separation has the following advantages: (1) simple module design and small units, (2) comparative lower initial investment, (3) simple operation and maintenance, (4) tailored product for hydrogen purification at different scale and degree.

After SMR and WGS reaction, the major components of the stream are

hydrogen and carbon dioxide. Hydrogen enrichment can be achieved either by H2-

or CO2- selective polymeric membranes. Generally, glassy polymers, especially the

widely used polyimide membranes are H2-selective, while rubbery polymers, for

example, polyethylene oxide membranes are CO2-selective. These two types of membranes have their own advantages and disadvantages. Thus, the final choice of the membrane hinges more on the operating parameters and applications.

1.3.2 Nitrogen Enrichment

The primary utility for nitrogen is as a protecting inert atmosphere.

Membrane-generated nitrogen could be used onsite anywhere flammable materials are stored, processed or handled. Many LNG ships, offshore platforms, and chemical tank farms are supplied with membrane generated nitrogen. Onsite nitrogen generators are also used in perishable warehouses, cargo containers, sintered-metal process furnaces, and oil-well servicing [20]. Membrane based on-board inert gas generation systems are employed in supersonic aircraft, for example, USA Air Force F-22, where the nitrogen produced from the compressed air in the jet engine is purged into the fuel tank onboard [21].

1.3.3 Recovery of Organic Vapor

Recovery of organic vapor from off-gas is not only environmentally desirable but also economically favored. In a PVC (poly (vinyl chloride)) production plant, valuable compounds, like vinyl chloride monomer, can be recovered [22]. At tank farms or petrol stations, volatile organic vapors emissions can also be controlled by membrane separation systems in order to meet the governmental regulations

[23].

1.3.4 Carbon Dioxide Capture

Although there is no universal agreement on the cause of global climate change, the greenhouse gas (GHG), mainly anthropogenic carbon dioxide emission is considered as the main suspect by the public. According to a report from U.S.

Department of Energy (DOE), approximately 83% of the GHG emission in U.S. is

produced from combustion and nonfuel use of fossil fuels. The capture of CO2 would not only mitigate the global warming concern but would also have some

economic benefits because CO2 could be used for enhanced oil recovery, enhanced coal bed methane recovery, and etc.

Currently there are three major approaches to reduce CO2 emission from fossil fuel; these are pre-combustion, oxyfuel and post-combustion [24]. The main principle of pre-combustion is to supply purified oxidizer and purified fuels to the combustion turbine in order to minimize the heat loss caused by the inert gases. For example, in a typical integrated gasification combined cycle power plant (IGCC), the fuel (coal), is sent to the gasifier to produce hydrogen and carbon monoxide.

Afterwards, more hydrogen is generated by converting carbon monoxide to CO2

by WGS reaction. Before the fuel stream is sent to the gas turbine, the inert CO2 should be removed in order to achieve high electricity generation efficiency.

Because of the high temperature of the mixed steam, the thermal stability of the

membrane module becomes critical. A polybenzimidazole (PBI) membrane developed by Los Alamos National Laboratory has shown long-term hydrothermal stability at 250 degrees for 400 days [25]. Some ceramic membranes are also

competitive candidates for this high temperature CO2/H2 process [26]. However,

these ceramic membranes only allow hydrogen to pass through, and CO2 remains in the high pressure retentate stream.

Another promising technology is oxy-combustion, in which the fuel is burned with almost pure oxygen mixed with the recycled flue gas [27]. The advantage of

this technology is that the exhausts are only CO2 and water, which means the CO2 can be compressed and stored after dehydration without spending energy on other inert gases (for example, nitrogen). Thus, nitrogen has to be removed from the air to produce enriched oxygen at the beginning of this process. The conventional way

to remove N2 from air is cryogenic distillation because N2 has higher a boiling

point than O2. However, many attempts have been made to reduce the cost of oxygen generation by using membranes. Praxair [28] employed an oxygen transport membrane in the boiler so that oxygen could diffuse directly into the combustion chamber to burn. The recycled flue gases have to be sent back and mixed with oxygen and fuel because the combustion furnace cannot withstand the ultra-high temperature if the fuel was burned with pure oxygen.

Post-combustion CO2 capture has the greatest potential to be the near-term

solution for reducing CO2 emission from power plants because this technology can be retrofitted to the existing IGCC or conventional coal power plant due to the

direct removal of CO2 from flue exhaust [27]. However, lots of challenges have to be solved. First, the flue gas after combustion is at atmosphere pressure, and the content of carbon dioxide is only around 10-20%, which means the partial pressure

of CO2 is only 0.1- 0.2 atm [29]. The thermodynamic driving force for CO2 capture

is so low that additional solutions have to be applied. On the other hand, the CO2 captured by the post-combustion is in the low pressure side. Appropriate compression is needed to meet the sequestration requirements. Meanwhile, the temperature of flue gas coming from a turbine is higher than 100 ⁰C; in that case, heat exchange or additional cooling treatment is required. Amine-based absorption system is a mature and commercially available solution for the post-combustion

capture. Amines react with CO2 to form reversible chemicals that could release the

CO2 at higher temperature and lower pressure. The main advantage of the

amine-based system is that this absorption process can capture CO2 at lower pressure compared to membrane processes [29]. Nevertheless, there are other problems like scale up, amine losses, degradation and corrosion related amine absorption. Membrane technology provides a more efficient and user friendly method that could replace amine-based systems. Firstly, the membrane does not involve liquid chemicals such as amines that are difficult to handle and maintain. 10

Secondly, the footprint of a membrane separation unit is much smaller than the amine absorption tower at the same manufacturing capacity. Last but not least, membrane systems are quite easy to scale up just by employing more membrane modules. Although membrane based systems still have their drawbacks, like lower

selectivity for CO2, higher capital investment, additional recompression cost and so

on, membrane technology is still a promising solution for the post-combustion CO2 capture; because the cost of membrane modules could be reduced once the production quantity increases, and the selectivity problem could be solved by developing new membrane materials or employing multi-stage separation.

1.4 Materials for Gas Separation Membranes

Both glassy and rubbery polymers are used to produce polymeric membranes.

Glassy polymers have mechanical properties that permit them to be used in self-supported structure like hollow fibers while rubbery materials have to have a proper support substrate because of their softness. These are some specific requirements for commercial membrane materials: 1) good mechanical strength; 2) good chemical resistance; 3) high separation factor with reasonable flux; 4) high thermal stability; 5) good processability; 6) low cost and environmentally friendly.

Until now, only a few polymers are employed in gas separation membrane as listed in Table 1-1.

Table 1-1 Most important polymers used in industrial gas separation membrane [10]

Rubbery polymers Glassy polymers Poly(dimethylsiloxane) Cellulose acetate Ethylene oxide/propylene oxide-amide copolymers Polyperfluorodioxoles Polycarbonates Polyimide Poly(phenylene oxide) Polysulfone

For glassy polymers, O2/N2 selectivity is a useful index to evaluate their separation performance. Table 1-2 indicates the milestones of membrane

development for O2/N2 separation.

Table 1-2 Progress of membranes for the O2/N2 separation (25°C) [10]

year polymer O2 permeability O2/N2 selectivity (Barrer) 1970s Poly(vinyltrimethylsilane) 47 4.3 1982 Poly(4-methyl-1-pentene) 24 3.6 1985 Ethyl cellulose 15 3.4 1986 polysulfone 1.2 6.0 1989 Poly(phenylene oxide) 16 4.0 1989 Halogenated 2.3-1.4 6.4-7.4 polycarbonates (35°C) 1996 polyimides 4-0.1 6-9 2005 polyimides 18-0.079 9-19.8

In this dissertation, CO2/H2 and CO2/N2 are the targeted gas pairs to be separated, so the more interesting materials are rubbery polymers and certain

polyimides. Depending on the nature of the material, gas separation membranes

could be classified into two categories: H2 selective or CO2 selective. H2-selective membranes are common glassy polymers which inherently have better thermal

stability and mechanical property than the rubbery polymers while CO2-selective membranes are usually made of rubbery materials or organic-inorganic hybrid materials.

H2-selective membranes are preferred if the ultimate usage of the purified hydrogen is as an energy resource in a power plant operated at a low pressure of approximately several bars. The remaining carbon dioxide in the high pressure retentate stream could be storied for other usage. Therefore, the expensive

recompression of CO2 for sequestration would be reduced. However, most glassy polymers show a trade-off relationship between gas permeability and selectivity, and the absolute permeability of glassy polymers is generally smaller than that of

rubbery polymers. Furthermore, plasticization by CO2 can deteriorate the

membrane separation performance by lowering the H2/CO2 selectivity in the mixed gas compared to that expected from pure gas permeation testing. A great deal of investigations has been devoted to minimizing the plasticization caused by condensable penetrants [30-32].

CO2-selective membranes are at preferred when the purified hydrogen is to be 13

employed as a portable vehicle fuel or the feed stock for fuel cell applications. For

CO2-selective membranes, the hydrogen generated by the WGS reactions remains on the high pressure side so that it can be stored and distributed more effectively via the existing supply network without additional recompression. Moreover, because the solubility selectivity is the dominate factor in overall selectivity for

rubbery membranes, CO2/H2 selectivity and CO2 permeability can be

simultaneously improved. Nevertheless, rubbery CO2-selective membranes perform well only at low temperature, which require more substantial pre-cooling,

e.g., when used in post-combustion CO2 capture. Another potential application of

CO2-selective membranes is to upgrade biogas [33], which requires removal of

CO2 and H2S from CH4 and H2 streams. Rubbery CO2-selective membranes

usually have larger gas flux than the glassy H2-selective membranes when their

selectivity is comparable. Therefore, CO2-selective membranes will yield greater productivity for hydrogen separation at low pressure and low temperature when the same membrane area is used. Since the methane and hydrogen gas stream pressure is low in the case of bio-hydrogen purification, the relatively poor mechanical

properties of CO2-selective rubbery membranes is not a great concern.

1.4 Goals and Organization of the Dissertation

The rising concern of global climate change requires more advanced

technology to control CO2 emission world-wide. Membrane based gas separation 14

is one of the most important technologies for addressing this challenge by

capturing CO2 generated from coal fired power plants. The competition of

membrane technology for CO2/light gas separation with conventional gas separation technologies relies critically on the gas permeability and selectivity of the available membrane materials. With extensive experimental studies done on glassy materials, the structure/property relationship shows a trade-off which may

not provide separation performance good enough for CO2 removal from light gases. In this project, both rubbery (organic-inorganic hybrid membrane, but mainly rubbery) and glassy membranes are investigated. The purpose of studying rubbery membranes was to push the permeability and selectivity to a higher level without restriction of the usual trade-off relationship; while the works on glassy polymers were targeted for developing a methodology to monitor the competition

between physical aging and CO2 plasticization. In the meanwhile, some simulation works were also performed to predict the permeation performance.

Overall, the main goal of this project is to solve the problems associated with

making membrane separations a worthy technology for CO2 capture. The works done on rubbery materials focus on developing high permeability and high

selective membranes to meet the energy cost requirement of post-combustion CO2

capture when CO2 selective membranes are employed; while the works on glassy materials attempt to understand the aging and plasticization phenomenon in a very

fundamental way when H2 selective membranes are employed. 15

This dissertation is comprised of eight chapters including their introductory chapter. Chapter 2 presents the background on gas transport in polymeric membranes. Chapter 3 describes the materials and experimental techniques used in this dissertation. Chapter 4 and 5 are focused on developing new organic-inorganic hybrid membrane materials. Ethylene oxide units are

identified as the key to achieve high CO2 permeability and high CO2 light gas selectivity simultaneously. Several approaches, including physical blending, chemical grafting and refining of end groups have been used. Chapter 6 discusses the accelerated physical aging and plasticization of polymeric thin films in order to evaluate the permeation performance of glassy membranes more correctly.

Chapter 7 reports some preliminary simulation results on establishing a prediction methodology of common gas separation materials based on molecular dynamics.

Finally, Chapter 8 presents the conclusions and recommendations for future work.

References

[1] M. Cheryan, Ultrafiltration Handbook, Taylor & Francis, 1998.

[2] R. Rautenbach, A. Gröschl, Separation Potential of Nanofiltration Membranes,

Desalination, 77 (1990) 73-84.

[3] W. Byrne, Reverse Osmosis: A Practical Guide for Industrial Users, Tall Oaks

Pub., 2002.

[4] T.Y. Cath, A.E. Childress, M. Elimelech, Forward Osmosis: Principles,

Applications, and Recent Developments, Journal of Membrane Science, 281

(2006) 70-87.

[5] X. Feng, R.Y.M. Huang, Liquid Separation by Membrane Pervaporation: A

Review, Industrial & Engineering Chemistry Research, 36 (1997) 1048-1066.

[6] A. Mansourizadeh, A.F. Ismail, Hollow Fiber Gas–Liquid Membrane

Contactors for Acid Gas Capture: A Review, Journal of Hazardous Materials,

171 (2009) 38-53.

[7] B.D. Freeman, Basis of Permeability/Selectivity Tradeoff Relations in

Polymeric Gas Separation Membranes, Macromolecules, 32 (1999) 375-380.

[8] B. Jiao, A. Cassano, E. Drioli, Recent Advances on Membrane Processes for

the Concentration of Fruit Juices: A Review, Journal of Food Engineering, 63

(2004) 303-324.

[9] J.-J. Qin, K.A. Kekre, G. Tao, M.H. Oo, M.N. Wai, T.C. Lee, B. Viswanath, H.

Seah, New Option of MBR-RO Process for Production of Newater from

Domestic Sewage, Journal of Membrane Science, 272 (2006) 70-77.

[10] P. Bernardo, E. Drioli, G. Golemme, Membrane Gas Separation: A

Review/State of the Art, Industrial & Engineering Chemistry Research, 48

(2009) 4638-4663.

[11] E.A. Mason, From Pig Bladders and Cracked Jars to Polysulfones: An

Historical Perspective on Membrane Transport, Journal of Membrane Science,

60 (1987) 125-145.

[12] J.V. Mitchell, J. R. Inst., 101 (1831).

[13] D.R. Paul, Y.P. Yampolskii, Polymeric Gas Separation Membranes, CRC

Press, 1994.

[14] S. Loeb, The Loeb-Sourirajan Membrane: How It Came About, in:

Synthetic Membranes:, American Chemical Society, 1981, pp. 1-9.

[15] R.W. Baker, Future Directions of Membrane Gas Separation Technology,

Industrial & Engineering Chemistry Research, 41 (2002) 1393-1411.

[16] N. Du, H.B. Park, M.M. Dal-Cin, M.D. Guiver, Advances in High

Permeability Polymeric Membrane Materials for CO2 Separations, Energy &

Environmental Science, 5 (2012) 7306-7322.

[17] R.W. Baker, Membrane Technology and Applications, J. Wiley, 2004.

[18] S.P. Nunes, K.V. Peinemann, Membrane Technology in the Chemical Industry,

in, Wiley-VCH, Weinheim, Germany, 2006. 18

[19] L. Shao, T.S. Chung, In Situ Fabrication of Cross-Linked PEO/Silica

Reverse-Selective Membranes for Hydrogen Purification, International Journal

of Hydrogen Energy, 34 (2009) 6492-6504.

[20] D.J. Stookey, Gas-Separation Membrane Applications, in: Membrane

Technology, Wiley-VCH Verlag GmbH & Co. KGaA, 2006, pp. 119-150.

[21] V.P. Crome, A.J. Yoder, Air Separation Module Using a Fast Start Valve for

Fast Warm up of a Permeable Membrane Air Separation Module, in, Google

Patents, 2002.

[22] R.J. Lahiere, M.W. Hellums, J.G. Wijmans, J. Kaschemekat, Membrane

Vapor Separation. Recovery of Vinyl Chloride Monomer from PVC Reactor

Vents, Industrial & Engineering Chemistry Research, 32 (1993) 2236-2241.

[23] W.H. Koch, Developing Technology for Enhanced Vapor Recovery: Part 1

Vent Processors, Petroleum Equipment & Technology, (2001) 16-22.

[24] B. Metz, I.P.o.C.C.W.G. III., IPCC Special Report on Carbon Dioxide

Capture and Storage, Cambridge University Press for the Intergovernmental

Panel on Climate Change, 2005.

[25] K.A. Berchtold, Novel Polymeric-Metallic Composite Membranes for CO2

Separation at Elevated Temperatures. , in: American Filtration and Separation

Society Fall Topical Conference, Pittsburgh, PA 2006.

[26] S. Smart, C. Lin, L. Ding, K. Thambimuthu, J.C.D. Da Costa, Ceramic

Membranes for Gas Processing in Coal Gasification, Energy Environmental 19

Science, 3 (2010) 268-278.

[27] T.C. Merkel, H. Lin, X. Wei, R. Baker, Power Plant Post-Combustion Carbon

Dioxide Capture: An Opportunity for Membranes, Journal of Membrane

Science, 359 (2010) 126-139.

[28] M.M. Shah, van Hassel, B., Christie, M., Li, J., CO2 Capture by Membrane

Based Oxy-Fuel Boiler. , in: Proceedings of the 2006 Conference on Carbon

Capture and Sequestration, Alexandria, VA, 2006.

[29] E. Favre, Carbon Dioxide Recovery from Post-Combustion Processes: Can

Gas Permeation Membranes Compete with Absorption?, Journal of Membrane

Science, 294 (2007) 50-59.

[30] T.S. Chung, J. Ren, R. Wang, D. Li, Y. Liu, K. Pramoda, C. Cao, W.W. Loh,

Development of Asymmetric 6FDA-2, 6DAT Hollow Fiber Membranes for

CO2/CH4 Separation: Part 2. Suppression of Plasticization, Journal of

Membrane Science, 214 (2003) 57-69.

[31] C. Cao, T.S. Chung, Y. Liu, R. Wang, K. Pramoda, Chemical Cross-Linking

Modification of 6FDA-2, 6-DAT Hollow Fiber Membranes for Natural Gas

Separation, Journal of Membrane Science, 216 (2003) 257-268.

[32] J. Wind, C. Staudt-Bickel, D.R. Paul, W.J. Koros, The Effects of Crosslinking

Chemistry on CO2 Plasticization of Polyimide Gas Separation Membranes,

Industrial & Engineering Chemistry Research, 41 (2002) 6139-6148.

[33] S. Basu, A.L. Khan, A. Cano-Odena, C. Liu, I.F.J. Vankelecom, 20

Membrane-Based Technologies for Biogas Separations, Chemical Society

Review, 39 (2010) 750-768.

CHAPTER 2 Background and Approaches

2.1 Permeability, Permeance and Selectivity

There are three basic technical terms used to describe the performance of membranes: permeability coefficient, permeation rate, and separation factor.

Permeability and separation factor are intrinsic properties of the membrane material, while permeance is thickness dependent. Permeance is used to characterize composite membranes consisting of a porous supported layer and a non-porous thin layer whose thickness cannot be measured very accurately. The permeation rate is expressed as cm3(STP)/cm2·s·cmHg, whose units, indicate the permeation rate should be the volume of gases that pass through the membrane per area per second at unit pressure. The term permeability is mainly used here for dense, or nonporous membranes, that function by a solution-diffusion mechanism

[1, 2] to describe the transport process. In a typical gas permeation testing, the feed

gas at the high pressure (pu) side will dissolve into the upstream surface of the film, diffuse through the polymer because of the different gas partial pressures in the upstream and downstream, and will finally desorb from the downstream side of the film. This movement in one dimension could be described by the Fick’s law in the form of flux in x direction:

D dC N   l (2-1) 1 w dx

where N is the flux along the x-direction, Dl is the gas local diffusion coefficient, C is the local concentration of dissolved gas, w is the weight fraction of gas in the film.

The steady state permeability through a thin film of thickness l is defined as [3,

4]:

Nl P (2-2) pu  pd where pu and pd are the upstream pressure and downstream pressure, respectively. If

we put eq 2-1 into eq 2-2 and integrate from x = 0 (C = Cu) to x=l (C = Cd), one obtains:

C C P D u d (2-3) pu  pd where D is the concentration averaged effective diffusion coefficient in the range

from Cd to Cu:

C C 1 u Dl 1 u D  dC  Deff dC (2-4) C C Cu  Cd d 1 w Cu  Cd d and Deff is the local effective diffusion coefficient.

In a typical gas permeation test, the upstream pressure is much higher than the

downstream pressure. In this case, both pd and Cd could be ignored, and eqs 2-3

reduces to:

PDS (2-5) where S is the apparent sorption coefficient or the solubility of gas in the polymer at the upstream pressure.

C S  u (2-6) pu

The ideal selectivity of membranes for pure gases A and B is defined as follows:

PA DA SA A,B     (2-7) PB DB  SB  where overall selectivity could be divided into diffusivity selectivity and solubility selectivity.

2.2 Solubility

The solubility of a gas in a polymer can be determined directly by measuring the weight increase at certain gas pressure. The solubility of gas A in a polymer depends on the nature of the polymer, its prior history in some cases, and conditions of measurement such as the temperature. The dual mode sorption model is the most commonly used one to describe the sorption of small molecules in glassy polymer materials. It combines the independent processes that follow a simple Henry’s law form with Langmuir-type sorption. Henry’s law sorption, is observed in liquids and rubbery polymers and reflects a simple solubility

phenomenon. The Langmuir sorption is a non-equilibrium process, with a finite capacity, associated with the excess free volume of the glassy polymer. The Henry sorption can be described using the same equation as in the liquid:

CD(p)kDp (2-8)

whereCD ( p) is the gas concentration in the polymer at pressure p, kD is the Henry’s law constant.

The Langmuir sorption could be roughly classified as a “hole filling” process.

Small penetrants keep adsorbing and desorbing at a rate b and finally reach dynamic equilibrium. This movement could be depicted as: C bp C ( p)  H (2-9) H 1 bp

Overall, the sorption in the glassy polymer could be expressed as the combination of above two types of sorption:

C bp C( p)  C  C  k p  H (2-10) D H D 1 bp

When the pressure is low, there is a steep increase of the sorption isotherm curve due to the hole filling process. After these holes are saturated, the Henry’s law sorption dominates so that the slope of sorption curve becomes constant. For the rubbery polymers, like polyethylene oxide, Henry’s law describes the gas sorption isotherms.

2.3 Fractional Free volume

Polymer scientists have made significant attempts to theoretically predict the intrinsic permeability of polymers according to monomer moieties, chain structures and rigidity, and functional groups [5]. Salame [6] developed the so-called

“Permachor” method to forecast oxygen permeability in barrier-type polymers via the calculation of an empirical factor for each chemical group which contributes to the total permeability. Bicerano [7] proposed the use of packing density, cohesive energy and rotational freedom of a polymer as important parameters to predict permeability. However, the method developed by Lee [8] to correlate gas permeability with free volume has received great attention and acceptance. The free

volume concept proposed by Lee was defined as (V-Vo), where V is the specific

volume obtained experimentally, while Vo is taken as 1.3 times of Vw which is the sum of van der Waals volume of the groups calculated following the Bondi’s method [9] or Park & Paul’s method [10]. Extensive experiments suggest that the effects of polymer structure on P is much more the result of varieties in the diffusion coefficient (D) rather than solubility coefficient (S) [11].

Though many parameters affect the diffusion coefficient, the free volume plays a much more crucial role than others. Thus, a highly simplified but very effective relationship between the permeability coefficient and the free volume has been

established through experiments [12]:

P = A exp (-B/FFV) (2-11) where A and B are constants for a particular gas, while FFV is the abbreviation of fractional free volume defined as following.

FFV = (V-Vo)/V (2-12)

The FFV value can be theoretically estimated from the molecular structure with the aid of the compass force field theory [13].

2.4 Gas Transport in Rubbery Polymers

Gas permeation in rubbery polymers is well understood in terms of the solution-diffusion model. The sorption of gases in rubbery materials is the same as in low molecular weight liquids and generally obeys Henry’s law. So under a certain pressure, the permeation rate is controlled by the diffusion step and could be satisfactorily described by Fick’s law [14]. In a rubbery polymer, the segments of the polymer backbone can rotate freely so that the polymer is soft and elastic.

The thermal motion of the polymer chains controls the gas diffusion process.

The diffusion coefficients are independent of the gas concentration when the penetrants are low-sorbing permanent gases [15]. However, strong derivation from

Henry’s law expressed as non-linear sorption isotherms is observed when

condensable gases (e.g. CO2, hydrocarbons) are at sufficiently high pressure.

These phenomenon could be described by using the Flory-Huggins equation [16]:

푝 2 푙푛 푎 = 푙푛 ( ) = 푙푛∅푣 + (1 − ∅푣) + 휒(1 − ∅푣) (2-13) 푝푠 where 푎 is penetrant activity in the gas phase contiguous to the polymers, p and

ps are the partial pressure and saturation vapor pressures of the gas, respectively,

∅푣 is the volume fraction of penetrant dissolved in the polymer, and 휒 is the so-called Flory–Huggins interaction parameter. In such condition, a linear or even exponential increase in diffusivity coefficient is observed [17].

2.5 Gas Transport in Glassy Polymers

Most of the differences in transport behavior between glassy and rubbery polymers stems from the non-equilibrium nature of glassy polymers. Glassy polymers can be considered as a super-cooled liquid, thus excess free volume is created due to the lack of sufficient relaxation time for polymer chains. The diffusion coefficients strongly depend on the free volume of the polymer. From another aspect, the thermal motion of glassy polymer is limited due to the steric hindrance along the polymer backbone so that the gas diffusion coefficient is normally lower than that of rubber.

2.6 Effect of Temperature

The temperature dependence of solubility, diffusivity and permeability coefficients may be described by an appropriate combination of van’t Hoff and

Arrhenius expressions, in the absence of any thermal transitions [18, 19].

S  S exp(H / RT ) 0 s (2-14)

D  D exp(E / RT ) 0 d (2-15)

P  P exp(E / RT ) 0 P (2-16) where S0, D0 and P0 are the pre-exponential factors. Hs is the heat of sorption, Ed is

the activation energy for diffusion, and Ep is the apparent activation energy for permeation. Because the sorption could be considered as the marriage of two hypothetical thermodynamic steps, which are the condensation of the pure penetrant and the mixing of the pure penetrant with the polymer matrix, the apparent activation energy is rewritten as follows:

E  E  H  H p d cond mix (2-17)

References

[1] D.R. Paul, O.M. Ebra-Lima, Pressure-Induced Diffusion of Organic Liquids

through Highly Swollen Polymer Membranes, Journal of Applied Polymer

Science, 14 (1970) 2201-2224.

[2] D.R. Paul, The Role of Membrane Pressure in Reverse Osmosis, Journal of

Applied Polymer Science, 16 (1972) 771-782.

[3] K. Ghosal and B. D. Freeman, Gas Separation Using Polymer Membranes: An

Overview, Polym. Adv. Technol., (1994) 24.

[4] R. R. Zolandz and G. K. Fleming, Membrane Handbook, in, Chapman & Hall,

New York, 1992.

[5] M.R. Coleman, W.J. Koros, Isomeric Polyimides Based on Fluorinated

Dianhydrides and Diamines for Gas Separation Applications, Journal of

Membrane Science, 50 (1990) 285-297.

[6] M. Salame, Prediction of Gas Barrier Properties of High Polymers, Polymer

Engineering & Science, 26 (1986) 1543-1546.

[7] J. Bicerano, Prediction of Polymer Properties, Marcel Dekker, 2002.

[8] W.M. Lee, Selection of Barrier Materials from Molecular Structure, Polymer

Engineering & Science, 20 (1980) 65-69.

[9] A.A. Bondi, Physical Properties of Molecular Crystals, Liquids, and Glasses,

Wiley, 1968.

[10] J.Y. Park, D.R. Paul, Correlation and Prediction of Gas Permeability in

Glassy Polymer Membrane Materials Via a Modified Free Volume Based

Group Contribution Method, Journal of Membrane Science, 125 (1997)

23-39.

[11] M.W. Hellums, W.J. Koros, G.R. Husk, D.R. Paul, Fluorinated

Polycarbonates for Gas Separation Applications, Journal of Membrane

Science, 46 (1989) 93-112.

[12] M.R. Pixton, D.R. Paul, Gas Transport Properties of Polyarylates: Substituent

Size and Symmetry Effects, Macromolecules, 28 (1995) 8277-8286.

[13] K.-S. Chang, C.-C. Tung, K.-S. Wang, K.-L. Tung, Free Volume Analysis and

Gas Transport Mechanisms of Aromatic Polyimide Membranes: A Molecular

Simulation Study, The Journal of Physical Chemistry B, 113 (2009)

9821-9830.

[14] E. Sada, H. Kumazawa, P. Xu, H. Nishikawa, Gas Transport in Rubbery

Polymers, Journal of Applied Polymer Science, 33 (1987) 3037-3044.

[15] A.Y. Alentiev, V.P. Shantarovich, T.C. Merkel, V.I. Bondar, B.D. Freeman, Y.P.

Yampolskii, Gas and Vapor Sorption, Permeation, and Diffusion in Glassy

Amorphous Teflon Af1600, Macromolecules, 35 (2002) 9513-9522.

[16] J.H. Petropoulos, Mechanisms and Theories for Sorption and Diffusion of

Gases in Polymers, in: D.R. Paul, Yampol'skii, Y. P. (Ed.) Polymeric Gas

Separation Membranes, CRC Press, Boca Raton, FL, 1994, pp. 17-81. 31

[17] H. Lin, B.D. Freeman, Gas and Vapor Solubility in Cross-Linked

Poly(Ethylene Glycol Diacrylate), Macromolecules, 38 (2005) 8394-8407.

[18] T.H. Kim, W.J. Koros, G.R. Husk, Temperature Effects on Gas Permselection

Properties in Hexafluoro Aromatic Polyimides, Journal of Membrane Science,

46 (1989) 43-56.

[19] W.J. Koros, W.C. Madden, Transport Properties, John Wiley & Sons, Inc.,

2002.

CHAPTER 3 Materials and Experimental Methods

3.1 Materials

Extem® XH 1015 was obtained from SABIC Innovative Plastics (formerly GE

Plastics) and was dried at 120 ºC overnight in a vacuum oven before use. It is synthesized from bisphenol-A dianhydride (BPADA) and diamino diphenyl sulfone

(DDS) via condensation polymerization as shown in Figure 3-1.

Figure 3-1 Synthetic route of Extem XH1015

The glass-transition temperature of Extem® 1015 was determined at a heating rate of 10 ºC/min by a PerkinElmer DSC6000, defined as the midpoint of change in heat capacity during the second heating. Matrimid 5218, a thermoplastic polyimide made from the monomers 3,3’,4,4’-benzophenone tetracarboxylicdianhydride

(BTDA) and diaminophenylindane (DAPI), was used as received from Huntsman

Advanced Materials. Polysulfone made from bisphenol A was also used. Their chemical structures are summarized in Table 3-1. Some bulk properties of these polymers (i.e., gas permeability, density, etc.) are given in Table 3-2.

Table 3-1 Chemical structures of polymer used in this study

Polymer name Structure

O O Polyetherimide, O N N S n Extem® 1015 O O O O O O O O Polyimide,

N N Matrimid® 5218 n O O O Polysulfone, PSF O S O n O

Table 3-2 Bulk properties of polymers used in this study

Polymer Density, Glass-transition Permeability, P (barrera) Selectivity

3 ρ (g/cm ) temperature, (°C) O2 N2 CH4 CO2 O2/N2

Polyetherimide, Extem [1]. 1.31 260 (267b) 0.81 0.13 0.13 3.3 6.2

Polyimide, Matrimid [2] 1.20 310 1.7 0.25 0.19 6.5 6.8

Polysulfone, PSF [3] 1.24 186 1.4 0.25 0.25 5.6 5.6 a 1 barrer = 1 × 10-10 cm3 (STP) cm cm-2 s-1 cm-Hg-1 b Glass-transition temperature provided by supplier [1].

Jeffamine®ED-2003 (O,O′-Bis(2-aminopropyl) polypropylene glycol-block-polyethylene glycol-block-polypropylene glycol, Mw: 2000g/mol), ethanol (AR grade), 3-glycidyloxypropyltrimethoxysilane (GOTMS) were purchased from Sigma-Aldrich and used as received. Polyethylene glycol (400,

1000, 1500 and 2000g/mol) were purchased from Sigma-Aldrich.

Methoxy-PEG-azide (Mw: 350 and 1100g/mol.) were obtained from Creative

PEGWorks and used without further treatment. Concentrated hydrochloric acid from Fisher Scientific was employed as the catalyst for the hydrolysis of GOTMS.

Cyclohexanone, the solvent, was purchased from Fisher Scientific and dried with molecular sieves before use. Dicholoromethane (DCM) was purchased from

Merck and utilized without further purification.

The gases, N2, O2, CH4 and CO2 were provided by Matheson Tri-gas and were

at least 99.99% pure. High purity compressed H2 (99.999%) and CO2/N2 (50:50) mixed gases were purchased from SOXAL, Singapore.

3.2 Preparation of Dense Membranes

3.2.1 Preparation of Glassy Thick Membranes

Dense Extem membranes were fabricated using a solution casting method described elsewhere26. In short, a solution was prepared by dissolving 2 wt% of the polymer sample in a solvent of DCM. The solution was then filtered through a 1µm filter (Whatman) and cast onto a Si wafer with a metal ring on top. The whole plate was leveled horizontally and covered with a glass plate. A small gap was left to 35

allow slow solvent evaporation to form a dense membrane. The nascent membrane was further dried at 250°C under vacuum for 48h to remove residual solvents.

3.2.2 Preparation of Organic-Inorganic Membranes (OIMs)

The synthetic route as shown in Figure 3-2 is similar to the report by Shao and

Chung [4]. The molar ratio of GOTMS, ethanol, water and hydrochloric acid in the hydrolysis solution was 1:1.13:3.2:0.05. After stirring for 60 min at room temperature, the alkoxysilane solution was transferred to the mixed solvent of water/ethanol (30/70 wt%) containing 2 wt% of Jaffamine® ED-2003. The molar ratio of GOTMS to Jaffamine® ED-2003 was kept at 4:1 in all samples. The solution was then heated to 60 ºC and while stirring for 1 hour to facilitate the epoxy-amine reaction and preliminary condensation. Afterwards, the predetermined amount of PEG was added into the mixture and stirred for 10 minutes, followed by filtration through a 5µm Waterman filter. The solution was poured into a homemade Teflon petri dish which was placed level in an oven at 30

ºC. The Teflon dish was covered with an aluminum foil with small holes to control the evaporation rate of solvent. After drying at 30 ºC for 1 day, the oven temperature was increased to 40 ºC for 2 days. Subsequently, the membranes were moved into a vacuum oven at 70 ºC for 1 day to complete the condensation reaction and remove the residual solvents. The membranes were sealed into a zip-lock bag and put into a refrigerator (-20 ºC ) for 2 hours to induce crystallization of the free PEGs which

were dispersed in the GOTMS-Jaffamine matrix. At the end, the membrane samples were peeled off the petri dish and stored in a desiccator for testing and characterization. The thermal history of these samples was kept the same for easy comparison of various properties.

Figure 3-2 Synthetic route of hybrid membranes

The molar ratio of GOTMS to Jaffamine® ED-2003 was maintained at 4:1 for all the matrices. The gas permeation performance of the hybrid membranes in other ratios could be found in a previous work [4]. Four PEGs (all with two hydroxyl end groups) of different molecular weights were blended into the

GOTMS-PEO (also referred as GP w/o PEG) matrices. The blended hybrid 37

membranes are represented by GPPX-Y-Z, in which GPP represents the

GOTMS-PEO-PEG blend, X represents the molecular weight of blended PEG, Y represents the weight percentage of the blended PEG, and Z represents the temperature. For example, GPP1000-40-35 means the matrix contains 40 wt% of free PEG1000 tested at 35 ºC.

For hybrid membranes with Methoxy-PEG-azide, the preparation procedure is similar. Methoxy-PEG-azide was used to replace PEG in the physical blending step. Figure 3-3 shows the scheme of membrane synthesis involving

Methoxy-PEG-azide.

Figure 3-3 Synthetic route of OIMs containing PEG-azide and the subsequent thermal reactions

In the last step, these OIMs blended with PEG-azide were peeled off and stored in a desiccator before any characterization. Thermal grafting was performed in a vacuum furnace by ramping the temperature from ambient to 140

ºC at 1 ºC/min and holding for 12 hours. The thickness of the obtained membranes ranged from 150 μm to 300 μm. The blended hybrid OIMs are labeled by

GPAX-Y, in which GPA represents the GOTMS-PEO PEG-azide blend, X represents the molecular weight of blended PEG, and Y represents the weight percentage of the blended PEG. For example, GPA1100-20 means the OIM contains 20 wt% of PEG-azide 1100g/mol.

3.3 Preparation of Polymeric Thin Films

Thin films of Extem were made by spin-coating polymer solutions onto silicon wafers. The solution was prepared by dissolving dried Extem in cyclohexanone at 80 ºC and slowly cooling to room temperature. The solution was filtered through 0.45μm, 0.2μm and 0.1μm Whatman PTFE membranes to eliminate any particles that could create pinholes in the thin films. Film thickness was controlled by adjusting the solution concentration from 3 wt% to 5 wt% and the spin speed from 1000 to 1400 rpm. For different polymers, these parameters vary due to different solution viscosity and volatility. A layer of highly permeable poly(dimethylsiloxane) (PDMS) was spin-coated on top of the Extem thin films in order to block possible pin holes and provide extra ease of handling. According to

the series resistance model [5], the permeability of the composite membrane is given by the following equation,

푙 푙 푙 composite = PDMS + Extem (3-1) 푃composite 푃PDMS 푃Extem where lcomposite, lPDMS and lExtem are the thickness of the composite membrane, the

PDMS and the Extem layers, respectively; Pcomposite, PPDMS and PExtem are the permeability coefficients of the composite membrane, the PDMS and the Extem

layers. The PDMS coating layer is approximately 3μm thick and the typical CO2 permeability is 3800 barrer [6], while the Extem layer is 160 nm thick and the

CO2 permeability is only 3 barrer. The contribution of the PDMS layer to the total resistance is less than 2% so that it has negligible effect on the measured permeation rate. On the other hand, the PDMS layer does not undergo physical aging itself and appears to have little or no influence on the behavior of the underlying glassy layer as reported by Rowe et al. [7]. In short, the effect of the

PDMS layer on the physical aging and CO2 plasticization behavior of the Extem films can be ignored. The wafer with the two polymer layers was then placed on a hot plate at 110 ºC for 15 min to crosslink the PDMS and removal all the residual solvent. More details about the PDMS coating technique can be found elsewhere

[8]. The bi-layer film was lifted from the wafer by using water and mounted on a square copper wire frames [9]. Samples were dried in an oven at 100 ºC overnight to remove the residual water and solvent.

All the samples were heated above the bulk Tg of Extem for 15 min in a N2 atmosphere and then rapidly quenched to room temperature immediately. This process erases the prior thermal history of the glassy polymer layer and relaxes any chain orientation imparted during spin-coating. The point at which the film is quenched is defined as the zero time in the aging experiment. Previous work has shown that prior history of a thin film can have an important effect on its aging behavior [10]. On the other hand, care has to be taken to avoid polymer degradation

while annealing above Tg. Extem is a relatively new commercial polymer , and published information about this material is quite limited [11, 12]. In this work, the

Extem films were annealed at 282 ºC to ensure completed removal of prior thermal history.

3.4 Characterization of Physicochemical Properties

3.4.1 Measurement of Gel Content

To examine the extent of polycondensation in the sol-gel system, the nascent membranes were subjected to water immersion for 1 week. The remaining insoluble samples were dried in the vacuum at 70 ºC for 24 hours to remove water before weighing again. The gel content was calculated by equation 3-2.

% gel content = m/m0 ×100% (3-2)

where m and m0 are the weights of the insoluble fraction and the original weight 41

of the membrane, respectively.

3.4.2 Fourier Transform Infrared Spectrometer (FTIR)

The FTIR-ATR measurements were performed using a Perkin-Elmer FT-IR

Spectrometer Spectrum 2000 with scan range from 4000 cm-1 to 600 cm-1. The

Extem thin film used in FTIR-ATR was made by ring casting.

3.4.3 Transmission Electron Microscopy (TEM)

Ultra-thin films for a combined scanning transmission electron microscope

(STEM/EDX) were prepared by a solution method following the procedure used for membrane casting.

3.4.4 Thermogravimetric Analysis (TGA)

The weight loss of OIMs containing Methoxyl-PEG-azide during thermal treatment was characterized by thermogravimetric analysis (TGA) with a TGA

2050 Themogravimetric Analyzer (TA Instruments). The analysis was carried out with a ramp of 10 °C/min in the temperature range from 50 to 800 °C. The purge

gas was N2 and its flow rate was controlled at 50 ml/min.

3.4.5 Wide Angle X-ray Diffraction (WAXD)

A wide-angle X-ray diffractometer (Bruker D8) was utilized to determine the

structures of the membranes at room temperature (25 ºC ). Ni-filtered Cu Kα radiation with a wave length of 1.54Å was used in the XRD experiments. The d-spacing was calculated based on the Bragg’s law nλ = 2dsinθ, where n is the integral number, λ is the wavelength, d is the dimension spacing and 2θ is the diffraction angle.

3.4.6 X-ray Photoelectron Spectrometer (XPS)

The nitrogen 1s bonding energy change after thermal grafting was monitored by an XPS spectrometer (Kratos Analytical Ltd., England) under ultrahigh vacuum. All spectra were obtained at a photoelectron take-off angle vertical to the sample.

3.4.7 Elemental Analysis

The elemental analysis was carried out using a Flash EA 1112 Elemental

Analyzer manufactured by Thermo Electron. The analyzer was able to determine

CHNS (Carbon, hydrogen, nitrogen and sulfur) on dried samples. The oxygen content was obtained by the subtracting the ash % plus the total for carbon,

hydrogen, nitrogen and sulfur.

3.4.8 Nuclear Magnetic Resonance (NMR)

NMR analyses were conducted using a Bruker 400 FT-NMR spectrometer at a resonance frequency of 400.132 MHz for 1H and 100.621MHz for 13C, while 1H

NMR spectra (1D, homonuclear decoupling) were obtained from an Extem® XH

1015 pellet sample dissolved in DMSO-d6 using a 5 mm BBO (broadband observe)

probe. Other spectra were performed in CDCl3 and chemical shifts are given in ppm unit.

The chemical structure changes of OIMs with PEG-azide after thermal grafting were characterized by solid state 13C NMR (100.6 MHz) with magic angle spinning

(MAS) at 7.5 kHz performed on a Bruker DRX 400 spectrometer.

3.4.9 Simulation Based on Molecular Dynamics

Molecular dynamic simulations were conducted using Material Studio 4.4 from Accelrys. Polymer chains consisting of 10 repeat units of polysulfone (PSU),

Ultem or Extem were constructed via a polymer build function, followed by energy minimization separately. Initiator and terminator were unknown because of

limited publications or information on these commercial polymers. The isotactic configuration with random torsion and head-to-tail orientation was assumed prior to the amorphous cell construction. Besides, polymer chains consisting of 5 repeat units were also constructed in order to discuss the chain packing morphology. Two polymer chains of PSU, Extem and Ultem were put together and minimized in order to discuss structural effects on chain packing. The initial density of the polymer periodic cell was set as 0.1 g/cm3 in order to eliminate aromatic ring catenation and chain scission during amorphous cell construction as suggested by

Heuchel and Hofmann [13]. Four polymer chains were employed for amorphous cell construction based on compass force field calculations by Amorphous Cell module. Conformational optimization and fine convergence with a maximum iteration of 10,000 was performed before molecular dynamics simulations to proceed. The detailed information for cell models built is listed in Table 3-3. The equilibrium stage temperature was set to 308 K when MD (molecular dynamics) by Discover module was launched. This is because all the pure gas permeability data were collected at 35°C. A set of MD under isothermal-isobaric (NPT) mode were performed to compress amorphous cell density from 0.1 g/cm3 to the experimental density. The simulated densities are listed in Table 3-2. The total MD time for each polymer was 500 ps in order to allow the amorphous cell to reach equilibrium.

Table 3-3 Atom numbers and cell dimensions for PSU, Extem and Ultem amorphous cells

The fractional accessible volume (FAV) was calculated from the free volume and occupied volume simulated by the Connolly task. The FAV distribution of each polymer was plotted by using different diameter probes to detect available volume in the amorphous cell. For one particular diameter there is one unique FAV, which means only particles smaller than the probe could go through or occupy the volume under this definition. Thus, this so called Connolly task corresponds to the kinetic diameters of gases employed. In this study, we choose probe diameters ranging from 2.0 Å to 5.0 Å, not only because even the diameter of smallest helium molecules is bigger than 2.0 Å, but also the number of cavities larger than

5.0 Å is very small and can be ignored, compared to that of the smaller cavities.

3.4.10 Variable Angle Spectroscopic Ellipsometer

A variable angle spectroscopic ellipometer (J.A. Woollam Co., model 2000D) was used to characterize the thickness of thin films on a silicon substrate.

Simultaneous measurement of more than 500 wavelengths ranged from 193 to

1000 nm could be achieved by using two different types of lamps (deuterium lamp

and quartz tungsten halogen lamp). A three-layer model consisting of Si, SiO2 and

Cauchy layer was used to fit the thickness of thin polymeric films. The Cauchy model as follows is valid since the film was transparent over the measured spectral range:

퐵 퐶 푛(λ) = 퐴 + + (3-3) 휆2 휆4 where n is the refractive index, λ is the wavelength of the sodium (Na) D under room temperature, A, B and C are constants obtained from the regression analysis.

3.5 Characterization of Gas Transport Properties

3.5.1 Pure Gas Permeation Tests

A constant volume method was employed to obtain the pure gas permeability

[14] and the solution-diffusion mechanism was used to explain the gas transport properties through dense membranes [15]. The measured rate of pressure increase

(dp/dt) at steady state was used to calculate the gas permeability according to the following equation.

273.151010 Vl dp  P    760 AT ( p 76 /14.7) dt   (3-4) where P is the gas permeability in barrer (1 barrer = 1×10−10 cm3 (STP) cm/cm2﹒ s ﹒cmHg), V is the volume of downstream chamber (cm3), l is the membrane

thickness (cm), A is the effective area of membrane (cm2), T is the operating temperature (K) and p is the upstream operating pressure (psia).

3.5.2 Mixed Gas Permeation Tests

The mixed gas permeation properties for these hybrid membranes were obtained with a lab-made mixed gas permeation cell as described in the work of

Xia et al. [16]. OIMs samples were tested with a CO2/N2 (50%/50% mole fraction) binary gas and anlysised by gas chromatography at 45 ºC with a total pressure of 4

atm. The gas permeability of CO2 and N2 is calculated by following equations:

10 y Vl 273.1510 CO2 dp  PCO    2 760 AT (76 /14.7)(x p) dt  CO2 (3-5) 10 (1y )Vl 273.1510 CO2 dp  PN    2 760 AT (76 /14.7)[(1x ) p] dt  CO2 (3-6) where yCO2 is the mole fraction of CO2 in the feed gas, and xCO2 is the mole

fraction of CO2 in the permeate.

3.5.3 Pure Gas Sorption Tests

Sorption measurements from 30 ºC to 50 ºC were conducted by using the gravimetric sorption technique (Cahn D200 microbalance sorption cell) which could be found elsewhere [17]. The sorption cell loaded with approximately 70

mg samples was evacuated overnight prior to sorption testing. From the weight gain, the amount of gas absorbed could be obtained after buoyancy correction.

Therefore, the solubility coefficient could be obtained by fitting the slope of the sorption isotherms. The diffusivity of gas penetrants was then calculated from solubility and permeability.

References

[1] S. Lee, Extreme Performance - or Processability?: New Tp Polyimide Offers

Both, in: Plastics Technology, January 2007.

[2] P.S. Tin, T.S. Chung, Y. Liu, R. Wang, S.L. Liu, K.P. Pramoda, Effects of

Cross-Linking Modification on Gas Separation Performance of Matrimid

Membranes, Journal of Membrane Science, 225 (2003) 77-90.

[3] J.S. McHattie, W.J. Koros, D.R. Paul, Gas Transport Properties of

Polysulphones: 1. Role of Symmetry of Methyl Group Placement on

Bisphenol Rings, Polymer, 32 (1991) 840-850.

[4] L. Shao, T.S. Chung, In Situ Fabrication of Cross-Linked Peo/Silica

Reverse-Selective Membranes for Hydrogen Purification, International

Journal of Hydrogen Energy, 34 (2009) 6492-6504.

[5] J.J. Shieh, T.S. Chung, D.R. Paul, Study on Multi-Layer Composite Hollow

Fiber Membranes for Gas Separation, Chemical Engineering Science, 54

(1999) 675-684.

[6] T.C. Merkel, V.I. Bondar, K. Nagai, B.D. Freeman, I. Pinnau, Gas Sorption,

Diffusion, and Permeation in Poly(dimethylsiloxane), Journal of Polymer

Science Part B-Polymer Physics, 38 (2000) 415-434.

[7] B.W. Rowe, B.D. Freeman, D.R. Paul, Physical Aging of Ultrathin Glassy

Polymer Films Tracked by Gas Permeability, Polymer, 50 (2009) 5565-5575.

[8] N.R. Horn, D.R. Paul, Carbon Dioxide Plasticization of Thin Glassy Polymer

Films, Polymer, 52 (2011) 5587-5594.

[9] Y. Huang, D.R. Paul, Experimental Methods for Tracking Physical Aging of

Thin Glassy Polymer Films by Gas Permeation, Journal of Membrane

Science, 244 (2004) 167-178.

[10] B.W. Rowe, B.D. Freeman, D.R. Paul, Influence of Previous History on

Physical Aging in Thin Glassy Polymer Films as Gas Separation Membranes,

Polymer, 51 (2010) 3784-3792.

[11] N. Peng, T.S. Chung, M.L. Chng, W. Aw, Evolution of Ultra-Thin

Dense-Selective Layer from Single-Layer to Dual-Layer Hollow Fibers Using

Novel Extem (R) Polyetherimide for Gas Separation, Journal of Membrane

Science, 360 (2010) 48-57.

[12] J. Xia, S. Liu, P.K. Pallathadka, M.L. Chng, T.S. Chung, Structural

Determination of Extem XH 1015 and Its Gas Permeability Comparison with

Polysulfone and Ultem Via Molecular Simulation, Industrial & Engineering

Chemistry Research, 49 (2010) 12014-12021.

[13] M. Heuchel, D. Hofmann, Molecular Modelling of Polyimide Membranes for

Gas Separation, Desalination, 144 (2002) 67-72.

[14] W.H. Lin, R.H. Vora, T.S. Chung, The Gas Transport Properties of

6FDA-Durene/1,4 Phenylenediamine (PPDA) Copolyimides, J. Polym. Sci.

Polym. Phys., 38 (2000) 2703-2713. 51

[15] W.J. Koros, W.C. Madden, Transport Properties, John Wiley & Sons, Inc.,

2002.

[16] J. Xia, S. Liu, C.H. Lau, T.S. Chung, Liquid-Like Polyethylene Glycol

Supported in the Organic-Inorganic Matrix for CO2 Removal,

Macromolecules, 44 (2011) 5268-5280.

[17] T.S. Chung, S.S. Chan, R. Wang, Z. Lu, C. He, Characterization of

Permeability and Sorption in Matrimid/C60 Mixed Matrix Membranes,

Journal of Membrane Science, 211 (2003) 91-99.

CHAPTER 4 Liquid-like Polyethylene Glycol Supported

in the Organic-inorganic Matrix for CO2 Removal

This chapter was published as a journal paper:

J. Xia, S. Liu, C.H. Lau, T.S. Chung, Liquid-like polyethylene glycol supported in the organic-inorganic matrix for CO2 removal, Macromolecules, 44 (2011) 5268-5280

Abstract

This chapter introduces a novel method to improve CO2 permeability and

CO2/light gas permselectivity by adding polyethylene glycol (PEG) into the

PEG-silica organic-inorganic hybrid matrix. The matrix is cross-linked by in situ formation of silica nano particles with diameters ranging from 1 to 5 nm. These particles disperse homogenously in the organic phase according to the morphology studied by scanning TEM. Many small particles were evolved after blending PEG into the matrix due to a high fraction of bi-functional and tri-functional silicon containing groups. Four PEGs with different molecular weights (Mw = 400, 1000,

1500, 2000g/mol) are added into the matrix prior to the matrix formation.

Permeability coefficients of three pure gases (H2, N2 and CO2) and one mixed gas

(CO2/N2) are measured to explore the effects of PEG content and molecular weight on the gas transport properties. The membrane containing 60 wt% of 1000 g/mol

PEG achieved an ultra-high CO2 permeability of 845 Barrer with CO2/H2 and

CO2/N2 permselectivity around 10 and 40, respectively. Fundamental studies on the effect of temperature on gas diffusivity and solubility are also carried out.

Molecular weights of PEGs have been proved to play a crucial role in determining the permeability and diffusivity.

4.1 Introduction

The failure of the 2010 United Nations Climate Change Conference in Cancun clearly indicated the increasing difficulty in forcing governments around the world to control carbon dioxide emission [1]. In this context, clean energy technology emerges as an important solution to overcome the dilemma between the economic development and the fight against global warming [2]. Reports show that more than

30% of CO2 produced by human activities are attributed to the running of coal-fired

power stations [3]. In this sense, the capture of CO2 in the coal-fired power

generation industry is especially significant in worldwide CO2 control. Meanwhile,

compared with other small individual sources of CO2 generation, large and centralized coal-fired power stations provide more convenience in the monitoring

and control of CO2 emission. Therefore, the technology of CO2 removal from large coal-fired power plants is highly demanded.

The IGCC (Integrated Gasification Combined Cycle) power plant program, which is the core technology of FutureGen sponsored by the U.S. Department of

Energy, requires extensive applications of high efficient CO2 removal technology

[4]. Currently there are three major approaches to reduce CO2 emission: pre-combustion capture, oxyfuel combustion and post-combustion capture [5].

Several technologies have already been developed for CO2 separation from light gases, including amine adsorption [6], pressure-swing adsorption [7] and membrane separation [4, 8-18]. Each technology has its own pros and cons.

Membrane separation systems may have drawbacks at the current stage, such as

relatively lower permselectivity for CO2 over light gases and additional recompression cost compared to amine based adsorption systems; however,

membrane based CO2 capture business appears to be more promising because new

materials with ultra-high CO2 permeability or permeance have been continuously developed [5].

In H2-selective glassy membranes, H2 gas molecules are smaller and thus diffuse

faster than CO2 [19-21]. Usually, H2-selective membranes are made of glassy polymers, which have better thermal stability and mechanical property than rubbery

polymers [22]. However, most H2-selective membranes have a trade-off relation between gas permeability and permselectivity, which leads them more inferior to

rubbery membranes for CO2 removal [23]. Reverse-selective membranes made from rubbery materials, like silicon rubber [24, 25] and polyethylene oxide (PEO)

[15, 26-36], could be promising materials for CO2 removal, because larger CO2 molecules, due to higher solubility, can pass through these membranes faster than

H2 and N2. In recent years, many PEO or polyethylene glycol (PEG) based CO2 reversible selective membranes have been developed because of their high ethylene 56

oxygen concentrations, including PEO copolymers [16, 30, 37], PEO blends [8, 9,

30, 33, 34, 37-40] and cross-linked PEO [11, 15, 33, 41-44]. However, PEO is easy to crystallize, resulting in reduced free volume and lower permeability [28].

Many approaches have been reported to reduce the crystallization tendency of

PEO/PEG. They may be briefly summarized as follows based on their own unique systems. a) PEO copolymer: The crystallization of ethylene oxide (EO) segments can be

prevented if the EO segments are too short to line up with the neighboring EO

chains [39]. b) Cross-linked PEO: Lin et al. [15, 26] have developed a series of membranes

with CO2 permeability of 400 Barrer at 35 ºC by cross-linking poly (ethylene

glycol diacrylate). Shao and Chung [10] have synthesized a series of

cross-linked organic-inorganic membranes by in situ formation of nano silica

particles. The CO2 permeability of these membranes reached 300 to 492 Barrer

with ideal permselectivity of 8-9. c) PEO Blends: Liquid PEG (Mw = 300 g/mol) was directly impregnated into the

microporous region of a cellulose membrane by Kawakami et al. [40].

Surprisingly, the CO2 permeability was only 82 Barrer at 25 ºC due to the

microporous tortuosity and the less permeable cellulose substrate. In order to

further improve the CO2 permeability and permselectivity, Car et al. [34, 37, 38] 57

and Yave et al. [8, 9, 30] have incorporated liquid PEG into thermoplastic poly

(amide-b-ethylene oxide) and poly (ethylene oxide)-poly (butylenes

terephthalate) multi-block copolymers via physical blending. However, the high

percentage of hard amide segments in these copolymers still has a negative

effect on permeability enhancement.

Ideally, liquid PEG should be the best selective medium for CO2 removal because of its high solubility and diffusivity, as well as the high permselectivity of

CO2/light gas. However, liquid PEG itself cannot form a free standing membrane.

Thus, a proper selection of matrix materials for hosting PEG is the key factor to effectively use its superior gas separation properties towards the separation of

CO2/light gas.

In this chapter, organic-inorganic membranes (OIMs) developed in our laboratory [10] are employed as the matrix to support the liquid PEG. In situ formed silica nano particles connect the poly (propylene oxide)-b-poly (ethylene oxide)-b-poly (propylene oxide) (PPO-PEO-PPO) chains, thus forming a dense membrane with sufficient mechanical strength. Low molecular weight PEGs (Mw

= 400, 1000, 1500 and 2000 g/mol) are directly blended in the matrix as shown in

Figure 4-1. This technique has a few advantages: (1) high capacity to host free

PEGs (up to 60 wt%); (2) a similar structure between the matrix and the free PEGs; 58

(3) permeability could be tailored by changing the content or the molecular weight

of the PEGs. CO2 permeability ranging from 55.6 to 1040 Barrer is achieved by

using different PEGs at temperatures from 35 ºC to 55 ºC. CO2/H2 and CO2/N2 permselectivity varies from 6.0 to 10.7 and 26.2 to 52.4, respectively. The effect of temperature is so significant that the permeability will increase 15 times with a temperature increment of 25 ºC. The elimination of PEG crystals is the major

reason for the permeability jump. CO2 sorption study reveals that CO2 solubility increases when the PEG melts. Among all these PEGs, PEG1000 is found to be the best candidate to maximize both permeability and permselectivity. Interestingly,

CO2 diffuses faster if the molecular weight of PEGs is comparable to the PEO segments in the matrix.

Figure 4-1 Concept of liquid PEGs supported in the organic-inorganic matrix

4.2 Results and Discussion

4.2.1 Basic Physicochemical Properties

All the membrane samples show crosslinked characteristics in gel content experiments as summarized in Table 4-1. For the GP w/o PEG sample, 96% of the reactants are cross-linked. For other samples with blended PEGs, the organic-inorganic matrices are insoluble while the blended PEGs can be completely washed out. On the whole, the organic-inorganic matrix is successfully fabricated regardless of the PEG loading.

Table 4-1 Gel content (%) of GP w/o PEG and GPP series

GP w/o PEG 96±1

20 wt (%) 40 wt (%) 60 wt (%)

GPP400 75±1 53±2 36±2

GPP1000 76±1 56±1 34±1

GPP1500 77±1 55±2 33±2

GPP2000 77±1 58±1 36±2

The GP w/o PEG and GPP400 membranes are transparent because their melting points are below room temperature as shown in Table 4-2, while GPP1000,

GPP1500 and GPP2000 are translucent due to the existence of PEG crystals.

Transparency of GPP400 series indicates that the silica nano particles are dispersed homogeneously.

Good dispersion of silica nano particles in the matrix is also observed in the

STEM graphs given in Figure 4-2a and 4-2b. The EDX spectrum in Figure 4-2c reveals that these nano-particles are composed of silica. The regular shape and sharp edges of nano particles in Figure 4-2a shows that these particles are in the crystal form. The distance between two particles varies from 5 nm to 50 nm, most of which are larger than the fully extended PPO-PEO-PPO polymer chain

(approximately 10 nm estimated from the bond length). Thus, incomplete cross-linking may exist.

Table 4-2 Thermal properties of GPP series, GP w/o PEG and pure PEGs

Tm2o Tm2p Tc 2o Tc 2p -ΔHc -ΔH’c

a b c d e f (°C) (°C) (°C) (°C) (J/g) (J/g)

GP w/o PEG 14.9 24.5 -20.1 -24.5 51 N/A

Jaffamine 29.4 37.2 15.6 11.2 114 N/A ED-2003

GPP400-20 9.7 20.5 -6.8 -13.8 58 66

GPP400-40 7.2 19.3 -7.7 -14.3 77 80

GPP400-60 6.4 20.0 -5.7 -13.3 88 95

PEG400 -9.5 7.7 -15.1 -21.2 123 N/A

GPP1000-20 -8.1 35.2 12.5 4.5 74 73

GPP1000-40 16 41.7 20.8 14.2 92 94

GPP1000-60 21.6 45.0 24.9 15.7 120 116

PEG1000 32.3 40.7 26.7 22.3 159 N/A

GPP1500-20 32.3 43.3 10.9 1.5 75 73

GPP1500-40 42.2 46.8 22.9 17.3 87 94

GPP1500-60 42.3 48.5 26.7 23.5 115 115

PEG1500 37.7 47.0 24.0 19.5 158 N/A

GPP2000-20 35.8 46.2 15.9 7.2 70 76

GPP2000-40 47.2 52.3 29.3 26.0 102 101

GPP2000-60 51.7 54.7 32.8 28.0 123 125

PEG2000 48.2 56.2 34.3 29.7 175 N/A a Onset melting temperature of the 2nd heating curve b Peak melting temperature of the 2nd heating curve c Onset crystallization temperature of the 2nd cooling curve d Peak crystallization temperature of the 2nd cooling curve e ΔHc is the heat of crystallization obtained experimentally from DSC f ’ ΔH c is the theoretical value calculated by equation 4-1

In this sol-gel system, any intermediates species, like [-Si-(OH)2OCH3] or

[-Si-(OCH3)2OH] would be considered as the result of partial hydrolysis and condensation. In addition, two partially hydrolyzed species may link together to form a siloxane [Si-O-Si] bond as illustrated by structure B in Figure 4-3c. In other words, not all the silicon portion of the precursors will participate in the construction of silica nanoparticles. Instead, some of them may only contribute one silicon-containing group while the rest remains flexible. The white dashed lines in Figure 4-3a are drawn for the illustration of possible network topology. In

Figure 4-3b, the white solid lines schematically represent the possible polymer alignment. Four kinds of groups with different degrees of hydrolysis (i.e., A, B, C

& D in the bottom of Figure 4-3c) play different roles in constructing the network.

The un-hydrolyzed groups (A) prevent the precursor from crosslinking and functions as the end groups as illustrated in Figure 4-3b. The mono-functional groups (B) bridge two crosslinking precursors and extending the polymer chain length between two nano-particles. Bi-functional (C) and Tri-functional (D) groups may crosslink together and form the silica nano particles. The silicon signal of un-hydrolyzed (A) and mono-functional (B) groups cannot be observed by STEM due to the resolution limitation.

Above hypotheses are supported by the analyses done by ImageJ [45] as shown

in Figure 4-4a and Table 4-3. In the image analyses, black and white are inversed for better particles identification. The average particle size (nm2) and the total area fraction are calculated by using different STEM pictures to achieve statistical results. Based on the 2-D area fraction information, relatively less silica content is observed in GP w/o PEG in Figure 4-4a than that in GPP1500-60 in Figure 4-4b by comparing with their theoretical weight percentages, although we cannot calculate the volume fraction from the area fraction directly. The missing silicon-containing components in GP w/o PEG are assigned to un-hydrolyzed and mono-functional groups (i.e., A & B in Figure 4-3c), which are too small to be captured by STEM. Conversely, a considerable portion of much smaller silica particles in GPP1500-60 can be observed in Figure 4-2b and 4-4b, although fewer bigger silica particles exist. These tiny nano dots (3.0 nm2) are located around the agglomerated patterns, and all the particles in GPP1500-60 do not exhibit a regular sharp edge as observed in GP w/o PEG. The distance from one dot to another is less than 5 nm, which is in the length range of PPO-PEO-PPO segments of the matrix. These are the strong evidences to prove that the cross-linking in

GPP1500-60 involves more structure C and D rather than structure A and B in

Figure 4-3c.

The hydrophilic PEGs absorb the water released from the condensation reactions. Thus, the reaction equilibrium moves to the product side. In other words, 64

more silica nano particles will be formed instead of being individual silicon containing functional groups. This is also consistent with the result of particle area analyses in Table 4-3. Strong affinity of PEG with water molecules promotes the condensation reaction and the formation of nano-silica nuclei. Another reason of the morphological difference is that the high viscosity of PEG molecules prevents the growth of nano silica nuclei from forming bigger particles.

Figure 4-2 STEM images of (a) GPP without free PEG and (b) GPP1500-60 and (c) EDX analysis of silica particles

Figure 4-3 STEM images of (a) hybrid membranes and (b) imaginary matrix constructed with (c) different functional groups

Figure 4-4 STEM images of hybrid membranes after imageJ analysis: (a) GPP w/o PEG and (b) GPP1500-60

Table 4-3 Average size and area fraction of silica particles in GP w/o PEG and GPP1500-60

Count Average Area Silica wt. size (nm2) Fraction (%) (%)

GP w/o PEG 106±4 44±2 7.5±2 11.1 GPP1500-60 1370±70 3.0±0.2 4.7±1 4.4

4.2.2 XRD Characterization

The smooth broad peak in Figure 4-5a shows that GP w/o PEG is amorphous at room temperature because freezing at -20ºC cannot induce crystallization as its Tc is -24.4ºC. GPP400 series are also amorphous because their melting points are below room temperature. GPP1000, GPP1500 and GPP2000 series show two characteristic peaks of PEO in Figure 4-5b, c and d. The peak at scattering angle 2θ

= 19.2° (d=4.62Å) corresponds to the reflections from (120) planes of the monoclinic PEG crystal. Another peak at 2θ = 23.3° (d=3.81Å) contains the overlapped reflections from (032), (132 ), (112), ( 212 ), (124 ), ( 204 ), and (004) planes [46].

Figure 4-5 XRD patterns for (a) GPP400, (b) GPP1000, (c) GPP1500 and (d) GPP2000 at room temperature

4.2.3 The Gas Permeation Performance

CO2 permeability and CO2/light gas permselectivity of PEO based membranes strongly depends on the EO group concentration [23]. Due to the existence of PEG

crystals, which are thought to be impermeable, CO2 permeability and CO2/light gas

permselectivity are compromised. By rising the temperature, CO2 permeability and

CO2/light gas permselectivity could ramp up because of more available EO groups after melting. In this matrix, the molecular weight of PEGs also plays a key role in determining the performance for the reason of interactions at molecular level.

Above factors will be described in more detail in the following sections. The best

data point with CO2 permeability of 696 barrer and CO2/H2 selectivity at 10.6 at

35°C almost reaches the empirical limit compared to the upper bound [47], as

shown in Figure 4-6a, while the CO2/N2 separation performance already exceeds the upper bound [26] as shown in Figure 4-6b.

Figure 4-6 Selected permeability/selectivity data map for (a) CO2/H2 and (b)

CO2/N2 separation at 35°C

The CO2, H2 and N2 permeability of GPP400, 1000, 1500 and 2000 series at different temperatures are depicted in Table 4-4. For easy comparison, gas permeation performance of GP w/o PEG and pure semi-crystalline PEO are also listed in Table 4-4.

Table 4-4 Pure gas permeability and selectivity for membranes with different compositions at 35°C ,45°C and 55°Ca

d d d PH2 PN2 PCO2 CO2/H2 CO2/N2 b c Sample T (°C) Tm2p (°C) (Barrer) (Barrer) (Barrer) selectivity selectivity

35 33.4 6.25 275 8.2 44.0 GP w/o PEG 14.9 45 49.2 9.90 344 7.0 34.7 35 36.0 7.29 333 9.3 45.7 GPP400-20 20.5 45 53.1 11.7 413 7.8 35.3 35 37.4 7.54 363 9.7 48.1 GPP400-40 19.3 45 53.5 11.9 455 8.5 38.2 35 40.0 8.20 427 10.7 52.4 GPP400-60 20.0 45 53.8 12.9 479 8.9 37.1 35 41.5 8.40 383 9.2 45.8 GPP1000-20 35.2 45 61.0 13.4 474 7.8 35.4 35 55.0 11.2 540 9.8 48.2 GPP1000-40 41.7 45 79.5 18.3 662 8.3 36.2 35 65.9 14.3 696 10.6 48.7 GPP1000-60 45.0 45 86.6 20.4 845 9.8 41.4 35 25.2 4.30 178 7.1 41.4 GPP1500-20 43.3 45 67.5 14.4 509 7.5 35.3 35 20.4 3.32 141 6.9 42.5 GPP1500-40 46.8 45 70.3 16.0 566 8.1 35.4 35 9.30 1.38 55.6 6.0 40.3 GPP1500-60 48.5 45 96.0 22.3 809 8.4 36.3 35 20.1 3.31 139 6.9 42.0 GPP2000-20 45 46.2 38.1 8.70 274 7.2 31.4 55 97.0 22.7 604 6.3 26.6 35 18.0 3.01 128 7.1 42.4 GPP2000-40 45 52.3 38.9 8.50 285 7.3 33.7 55 111 23.0 749 6.7 32.6 35 11.5 1.70 68.0 5.9 40.0 GPP2000-60 45 54.7 25.8 4.73 163 6.3 34.5 55 143 34.1 1040 7.3 30.5 Semi-crystalline 35 1.8 0.24 13 7.2 54 28 68.0 PEO 45 6.4 0.99 40 6.3 40 70

a 55°C was only employed for GPP2000 series b GPPX-Y represents the sample code, where X is the molecular weight and Y is the weight percentage of PEG blended. c Peak melting temperatures of the 2nd heating curve d Permeability, 1 Barrer = 10-10 cm3 (STP).cm/cm2·s·cmHg

The gas permeation performance of GP w/o PEG is comparable to the previous

work [10]. The reason of the higher H2 and CO2 permeability compared to the pure semi-crystalline PEO is that the crystallinity of PEO segments in GP w/o PEG is greatly suppressed by the cross-linking.

For GPP400 and GPP1000 series, both permeability and permselectivity are improved at 35 ºC and 45 ºC , which is contradictory to the trade-off relationship of glassy polymers. These membranes take the advantage of the higher solubility

selectivity of CO2 over H2 and N2 introduced by EO groups. As mentioned

previously, the EO group concentration determines the CO2/light gas selectivity. By

increasing the weight percentage of PEG400, the CO2/H2 permselectivity increases

from 8.2 (no PEG) to 10.7 (60 wt% PEG400) at 35 ºC . The CO2 permeability increases by more than 50% at 35 ºC with 60 wt% of PEG400 in the matrix. The diffusivity increment is the major driving force for the permeability improvement,

while the CO2 solubility shows little change with the addition of PEG400 and

PEG1000 as shown in Table 4-5. The liquid PEGs in the matrix greatly facilitates

the transportation of CO2, leading to an ultra-high diffusivity than the conventional solid membranes.

Table 4-5 Gas permeability, solubility and diffusivity coefficient results compared with PDMS and PEO from other sources

b 6 PCO2 SCO2 DCO2×10 (Barrer) (cm3 STP/cm3 atm) (cm2/s) Samplesa

35ºC 45ºC 35ºC 45ºC 35ºC 45ºC

GP w/o PEG 275 344 1.59 1.16 1.32 2.26

GPP400-20 333 413 1.47 1.15 1.72 2.73

GPP400-40 363 455 1.48 1.25 1.87 3.00

GPP400-60 427 479 1.50 1.19 2.17 3.05

GPP1000-20 387 474 1.51 1.17 1.93 3.07

GPP1000-40 540 662 1.65 1.27 2.48 3.97

GPP1000-60 696 845 1.65 1.25 3.20 5.14

GPP1500-20 178 509 1.14 1.24 1.19 3.12

GPP1500-40 141 566 0.82 1.27 1.31 3.39

GPP1500-60 55.6 810 0.54 1.31 0.78 4.68

GPP2000-20 139 274 0.99 1.19 1.07 1.76

GPP2000-40 128 285 0.74 1.22 1.32 1.77

GPP2000-60 68 163 0.46 1.29 1.13 0.96

Semi-crystalline 13 40 0.37 N/A 0.27 N/A PEO28

Amorphous 143 N/A 1.3 N/A 0.86 N/A PEO-3528

PDMS-3525 3800 N/A 1.3 N/A 22.2 N/A

a GPPX-Y represents the sample code, where X is the molecular weight and Y is the weight percentage of PEG blended. b 1 Barrer = 10-10 cm3 (STP).cm/cm2·s·cmHg

Upon surpassing the melting temperatures, GPP1500 and 2000 show tremendous permeability jumps due to significant increments in both diffusivity and solubility as listed in Table 4-5. Because of the existence of PEG crystals, the permeability of all gases for GPP1500 series is smaller than that for GP w/o PEG at 35 ºC.

Nevertheless, the CO2 permeability rises by a factor of 10 after the crystals melt.

The CO2 permeability of GPP1500-60 increases from 55.6 Barrer at 35 ºC to 809

Barrer at 45 ºC. Within our expectation, the permeability of GPP2000 series is still lower than GP w/o PEG even at 45 ºC, because PEG crystals do not fully melt. In order to investigate the permeation properties of GPP2000 series in a liquid state, the permeation cell is heated up to 55 ºC to facilitate the melting of PEG2000. A similar permeability jump (from 68 Barrer at 35 ºC to 1040 Barrer at 55 ºC) is observed. The melting point measured by DSC is slightly higher than that observed from the permeation jumping. This difference could be caused by the temperature lag in the dynamic DSC measurement.

Normally, the gas permeability will increase with increasing PEG content. On the contrary, the permeability declines with increasing PEG content when PEG is in a semi-crystalline form as shown in Figure 4-7. The main reason of this phenomenon 73

is that the PEG crystals are considered as impermeable domains for gases. In addition, the shape, size and size distribution of the crystallites affect the tortuosity and the chain mobility of the amorphous phase. Thus, the diffusivity decreases when the impermeable crystals exist. Furthermore, semi-crystalline membranes have less sorption sites than amorphous membranes. Sorption measurements reveal

that the CO2 solubility of GPP1500 and GPP2000 is smaller in the semi-crystalline state as shown in Table 4-5. In conclusion, increasing PEG content in the semi-crystalline membranes will lead to less amorphous regions and create more complicated tortuosity for gas diffusion. Thus, the permeability decreases because of the lower diffusivity and the lower solubility.

Figure 4-7 CO2 permeability of GPP series and GP w/o PEG at 35°C (blank) and 45°C (checked)

The permselectivity varies a lot when PEG content or temperature changes. For

membranes with liquid PEGs, the CO2/H2 permselectivity increases in pace with the PEG content as shown in Figure 4-8 because of the increased EO concentration.

For membranes with PEG crystals, the CO2/H2 permselectivity decreases with increasing PEG content, because the size sieving effect is enhanced by the increasing tortuosity. In rubbery materials, the permselectivity usually weighs

heavier on solubility selectivity. Although H2 diffuses faster than CO2, the

permeability of CO2 is still much larger than H2 because of the higher solubility.

Diffusivity selectivity of CO2/H2 is usually below 1 due to the size sieving effect.

However, the CO2 diffusivity gains a lot upon traversing the melting point, leading

to a relative smaller diffusivity difference between CO2 and H2. Thus, the overall selectivity is improved by minimizing the size sieving effect by melting the crystals.

As shown in Figure 4-8, the CO2/H2 permselectivity of GPP1500 series after PEG crystals melting is improved to different extents depending on the PEG content in the blends. GPP1500 with 60wt% PEG shows the greatest improvement from 6.0 to

8.4, while GPP2000 series show less CO2/H2 permselectivity improvement because

PEG2000 crystals only partially melt as indicated by the small melting peaks at lower temperatures in Figure 4-9d.

Figure 4-8 Permselectivity at 35°C (solid) and 45°C (hollow) for CO2/H2

Figure 4-9 DSC curves for (a) GPP400, (b) GPP1000, (c) GPP1500 and (d) GPP2000

Table 4-6 shows the CO2/N2 mixed gas results. A considerable CO2/N2 selectivity decrease from 41.4 to 31.3 for GPP1000-60 and from 36.3 to 32.0 for GPP1500-60

at 45 ºC is observed. CO2 permeability is comparable with the pure gas permeability

for both samples. However, the N2 permeability increases from 20.4 to 26.8 Barrer for GPP1000-60 and from 22.3 to 26.2 Barrer for 1500-60, respectively. These

phenomena are probably caused by the CO2 plasticization in the solubility selectivity membranes [48].

Table 4-6 Mixed gas permeability and selectivity for GPP1000-60 and GPP1500-60 at 45°C

50:50 CO2/N2 mixed gas CO /N Sample 2 2 PCO2 PN2 selectivity (Barrer) (Barrer)

31.3±0.6 GPP1000-60 839 (845)* 26.8 (20.4) (41.4)

32.0±0.5 GPP1500-60 838 (809) 26.2 (22.3) (36.3)

*Number between brackets is the permeability and permselectivity tested by pure gases.

4.2.5 Effect of Testing Temperature

The effect of testing temperature on CO2 permeability of GPP1500-60 is

illustrated in Figure 4-10. The CO2 permeability increases from 26 Barrer at 30ºC to

860 Barrer at 55ºC. The temperature dependence of permeability without thermal

transition can be described by an Arrhenius type expression P  P0 exp(EP / RT ) .

These linear relationships are only observed at lower temperatures and higher temperatures. In between these two regions, a jump of permeability is observed near the onset melting temperature, as shown in Figure 4-10. Mogri and Paul [49, 50] have also observed a change in permeability of two orders of magnitude when the side chains of poly(octadecyl acrylate) melt. Hirayama et al.[33] and Kim et al.[51] have also observed an abrupt increase of permeability when PEG melt in their PEG based materials. Nonlinear relationships appear with the thermal transition, because

Ep is changed, as well as the three parameters shown in equation (2-17). The decrease of the activation energy for diffusion could be caused by the change of diffusion mechanism from the hopping diffusion to the liquid-like diffusion. The diffusivity of gases in the polymer melt could reach the diffusive limit on a time scale similar to that of the polymer melt. Meanwhile, the tortuosity will be minimized, and the chain motion in the amorphous region will be no longer affected by the crystallites. There is a 6 times increment in diffusivity of GPP1500-60 at

45ºC compared to 35ºC, as shown in Table 4-5. The absolute value of condensation enthalpy will not change because it is determined by the penetrants themselves.

However, the mixing enthalpy of CO2 and PEO based polymers will be more

negative because less energy is needed to allow CO2 molecules penetrate into the matrix. Thus, the solubility increases when a thermal transition happens. 78

Afterwards the solubility will decrease when the temperature further increases.

The solid lines in Figure 4-11a and 4-11b represent the CO2 sorption isotherms and the solubility coefficients of GPP1500 series and GP w/o PEG at 35ºC, while the dashed lines represent the sorption at 45 ºC, respectively. The solubility coefficients increase from 0.54, 0.82 and 1.14 to 1.31, 1.27 and 1.24 cm3

(STP)/cm3atm for GPP1500-60, GPP1500-40 and GPP1500-20 after melting, respectively. Being the membrane with the highest crystal content, GPP1500-60 shows the lowest solubility coefficient close to that of the semi-crystalline PEO at

35ºC as shown in Figure 4-12b. On the contrary, the solubility coefficient of GP w/o

PEG decreases from 1.59 to 1.16 cm3 (STP)/cm3atm when temperature increases because there is no thermal transition. For the same reason, GPP1000 series, in which PEG1000 crystals already melt at both temperatures shows a solubility loss, as shown in Table 4-5. However, the diffusivity increment overweighs the slight decrease of solubility. Therefore, the permeability still shows considerable

enhancement at 45ºC . CO2 permeability of GPP1000-60 increases from 696 Barrer to 845 Barrer despite its solubility decreases from 1.65 to 1.25 cm3 (STP)/cm3atm.

The solubility of GPP1000 series at 35 ºC could reach 1.5~1.6 cm3 STP/cm3 atm, which is comparable with the solubility of amorphous PEO [28] (1.3 cm3 STP/cm3 atm) as shown in Figure 4-12b and the crosslinked PEO rubber [26] (1.5 cm3

STP/cm3 atm). The solubility of GPP2000 series increases from 0.46~0.99 to 79

1.19~1.29 cm3 (STP)/cm3atm due to the partial melting as indicated by the small melting peaks as shown in Figure 4-9d.

Figure 4-10 Change of CO2 permeability of GPP1500-60 from 30°C to 55°C nd *Tm2o is the onset melting temperature of PEG1500 in GPP1500-60 during 2 heating.

Figure 4-11 CO2 Sorption isotherm (a) and solubility coefficient (b) of GPP1500 at 35°C and 45°C

Figure 4-12 Sorption isotherm (a) and solubility (b) of GPP1500, PDMS [52] semi-crystalline PEO and amorphous PEO [28] at 35°C

4.2.5 Effect of PEGs’ Molecular Weights

Figure 4-13 shows the solubility and diffusivity of blended membranes

containing 40 wt% PEGs at 45 ºC. The CO2 solubility for these membranes varies

little as they have the same PEG content, while the CO2 permeability changes dramatically. GPP1000-40 has the highest diffusivity while GPP2000-40 is the lowest. For other samples with different contents of PEGs, the trend is almost the same. The lowest diffusivity of GPP2000-40 can be ascribed to the partial melting of the PEG crystals. However, the diffusivity difference among GPP400-40,

GPP1000-40 and GPP1500-40 could only be ascribed to the molecular weight difference, because these PEGs are already in liquid states. Due to the technical limitation, there is no suitable characterization method available to monitor the effect of molecular weight on the diffusivity coefficient above the melting

temperature. An alternative method, by analyzing the crystallization and melting behavior of the blend membranes, is an important way to understand this phenomenon. Although melting is observed from DSC, we could still assume that liquid PEG molecules in the matrix only vibrate violently in the confined environment rather than free flowing. Extensive studies have been done in the past decades to understand the crystallization behavior of low molar mass PEGs. It is well known that PEG with a low molecular weight (Mw ≤3000 g/mol) crystallizes exclusively with extended chains under normal cooling rate [53, 54]. In this scenario, we propose a possible crystallization mechanism as depicted in Figure

4-14. PEGs with different molecular weights will have different orientations and alignments in the confined environment. During membrane fabrication, PEGs are heated to 70 ºC to surpass their melting temperatures so that the equilibrium could be reached between the PEO segments of the matrix and the liquid PEG molecules.

The compatibility between the PEO segments and the PEG molecules is fairly good so that they crystallize together upon cooling. All the membranes present one crystallization peak during second cooling, although some of the membranes have two melting peaks as shown in Figure 8a-d. Analysis on the heat of crystallization

in Table 2 also supports the assumption of co-crystallization. ΔHc is the heat of

’ crystallization obtained experimentally from the integration of DSC peak area. ΔH c is the theoretical value calculated from

Hc ' Hc (PEG) wt%(PEG)  Hc (matrix)(1 wt%(PEG)) (4-1) 82

’ ΔHc and ΔH c values of GPP1000, 1500 and 2000 series are almost the same. For

’ GPP400 series, ΔHc is smaller than ΔH c, revealing that GPP400 series is easier to crystallize. Crystallization temperatures of GPP400 blends, which are higher than both GP w/o PEG and PEG400, also provide strong evidence for the conclusion that

GPP400 series has the strongest interactions with the PEO segments in the matrix.

Thus, the diffusivity of GPP400 series should be the lowest due to its highest tendency to crystallize. This correlates well with the result obtained from sorption experiments. The high concentration of hydrogen bonding of PEG400 may be the reason of their strong crystallization tendency.

Figure 4-13 Solubility (square) and diffusivity (circle) of hybrid membranes containing 40 wt% of PEG with different molecular weights at 45°C

Figure 4-14 Crystallization of PEO segments and PEGs with different molecular weights

In GP w/o PEG, PEO segments cannot fully crystallize due to cross-linking. A shoulder peak, rather than a symmetric exothermic peak in the cooling curve is observed, indicating that a considerable amount of imperfect PEO crystals are formed because of the confinement. The crystallization peaks of GPP1000, 1500 and 2000 are located in between the peaks of the GP w/o PEG and the respective pure PEGs. The fully extended PEG chains may facilitate the PEO segments of the matrix to crystallize by reducing the extra free energy required to align chain segments. Thus, the crystallization temperature increases with PEG content.

Crystal defects and lots of chain ends are created during crystallization because of different chain lengths as illustrated in Figure 4-14. Several PEG400 molecules

could be aligned along one PEO segment (Mw≈1300). However, only one

PEG1000 chain is allowed to align along the PEO segment because of their similar 84

molecular weights and chain lengths. Once one PEG1000 chain is aligned with the

PEO segment, the remaining portion of the PEO segment may not arrange regularly, thus leading to more defects. Therefore, the diffusivity of GPP1000 series is the highest. The PEO segments of the matrix could be fully coupled with PEG1500 because of their similar chain lengths. One folding may exist when PEG2000 is employed, because its chain length almost doubles that of the PEO segment. The diffusivity of GPP1500 is slightly lower than that of GPP1000 at 45 ºC , because fewer defects are created.

Summary

In this chapter, a novel method of employing liquid PEG as a gas separation medium is introduced. The PEO-silica organic-inorganic network synthesized by the sol-gel process has been proved to be an effective matrix for holding liquid PEG.

The permeability has been improved more than twice without losing permselectivity when PEG with a proper molecular weight is employed. With the

loading of PEG molecules, CO2/H2 and CO2/N2 permselectivity is found to increase by more than 10%. A significant permeability jump is observed when the testing temperature surpasses the melting point of PEG. The diffusivity increment is considered as the major contributor to the permeability jump, while the solubility shows little change if there is no thermal transition. A considerable increase of

solubility coefficient is observed when solid PEG becomes liquid. Four PEGs with different molecular weights have been investigated. PEG1000, which has a similar molecular weight to the PEO segments in the PEO-silica organic-inorganic matrix,

is proved to be the best candidate to improve both CO2 permeability and CO2/light gas permselectivity.

References

[1] E. Kintisch, A. Regalado, Climate Negotiations Cancun Delegates See the Trees

through a Forest of Hot Air, Science, 330 (2010) 1597-1597.

[2] D.P. Schrag, Preparing to Capture Carbon, Science, 315 (2007) 812-813.

[3] R.E.H. Sims, H.H. Rogner, K. Gregory, Carbon Emission and Mitigation Cost

Comparisons between Fossil Fuel, Nuclear and Renewable Energy Resources

for Electricity Generation, Energy Policy, 31 (2003) 1315-1326.

[4] K. Damen, M. van Troost, A. Faaij, W. Turkenburg, A Comparison of Electricity

and Hydrogen Production Systems with Co2 Capture and Storage. Part A:

Review and Selection of Promising Conversion and Capture Technologies,

Progress in Energy and Combustion Science, 32 (2006) 215-246.

[5] T.C. Merkel, H. Lin, X. Wei, R. Baker, Power Plant Post-Combustion Carbon

Dioxide Capture: An Opportunity for Membranes, Journal of Membrane

Science, 359 (2010) 126-139.

[6] G.T. Rochelle, Amine Scrubbing for Co2 Capture, Science, 325 (2009)

1652-1654.

[7] S.I. Yang, D.Y. Choi, S.C. Jang, S.H. Kim, D.K. Choi, Hydrogen Separation by

Multi-Bed Pressure Swing Adsorption Of synthesis Gas, Adsorption, 14 (2008)

583-590.

[8] W. Yave, A. Car, K.V. Peinemann, M.Q. Shaikh, K. Rätzke, F. Faupel, Gas

Permeability and Free Volume in Poly(Amide-B-Ethylene

Oxide)/Polyethylene Glycol Blend Membranes, Journal of Membrane Science,

339 (2009) 177-183.

[9] W. Yave, A. Car, S.S. Funari, S.P. Nunes, K.V. Peinemann, Co2-Philic Polymer

Membrane with Extremely High Separation Performance, Macromolecules, 43

(2010) 326-333.

[10] L. Shao, T.S. Chung, In Situ Fabrication of Cross-Linked Peo/Silica

Reverse-Selective Membranes for Hydrogen Purification, International

Journal of Hydrogen Energy, 34 (2009) 6492-6504.

[11] N.P. Patel, J.M. Zielinski, J. Samseth, R.J. Spontak, Effects of Pressure and

Nanoparticle Functionality on Co2-Selective Nanocomposites Derived from

Crosslinked Poly(Ethylene Glycol), Macromolecular Chemistry and Physics,

205 (2004) 2409-2419.

[12] N. Patel, Morphological, Mechanical and Gas-Transport Characteristics of

Crosslinked Poly(Propylene Glycol): Homopolymers, Nanocomposites and

Blends, Polymer, 45 (2004) 5941-5950.

[13] K.C. O’Brien, G. Krishnan, K.A. Berchtold, S. Blum, R. Callahan, W. Johnson,

D.-L. Roberts, D. Steele, D. Byard, J. Figueroa, Towards a Pilot-Scale

Membrane System for Pre-Combustion Co2 Separation, Energy Procedia, 1

(2009) 287-294.

[14] D. Munoz, E. Maya, J. Deabajo, J. Delacampa, A. Lozano, Thermal Treatment 88

of Poly(Ethylene Oxide)-Segmented Copolyimide Based Membranes: An

Effective Way to Improve the Gas Separation Properties, Journal of Membrane

Science, 323 (2008) 53-59.

[15] H. Lin, E.V. Wagner, J.S. Swinnea, B.D. Freeman, S.J. Pas, A.J. Hill, S.

Kalakkunnath, D.S. Kalika, Transport and Structural Characteristics of

Crosslinked Poly(Ethylene Oxide) Rubbers, Journal of Membrane Science,

276 (2006) 145-161.

[16] D. Husken, T. Visser, M. Wessling, R.J. Gaymans, Co2 Permeation Properties

of Poly(Ethylene Oxide)-Based Segmented Block Copolymers, Journal of

Membrane Science, 346 (2010) 194-201.

[17] A. Brunetti, F. Scura, G. Barbieri, E. Drioli, Membrane Technologies for Co2

Separation, Journal of Membrane Science, 359 (2010) 115-125.

[18] Y.C. Xiao, T.S. Chung, Grafting Thermally Labile Molecules on

Cross-Linkable Polyimide to Design Membrane Materials for Natural Gas

Purification and Co2 Capture, Energy & Environmental Science, 4 (2011)

201-208.

[19] B.T. Low, Y. Xiao, T.S. Chung, Amplifying the Molecular Sieving Capability

of Polyimide Membranes Via Coupling of Diamine Networking and Molecular

Architecture, Polymer, 50 (2009) 3250-3258.

[20] L. Shao, T.S. Chung, S. Goh, K. Pramoda, Polyimide Modification by a Linear

Aliphatic Diamine to Enhance Transport Performance and Plasticization 89

Resistance, Journal of Membrane Science, 256 (2005) 46-56.

[21] P.S. Tin, T.S. Chung, Y. Liu, R. Wang, S.L. Liu, K.P. Pramoda, Effects of

Cross-Linking Modification on Gas Separation Performance of Matrimid

Membranes, Journal of Membrane Science, 225 (2003) 77-90.

[22] J. Xia, S. Liu, P.K. Pallathadka, M.L. Chng, T.S. Chung, Structural

Determination of Extem Xh 1015 and Its Gas Permeability Comparison with

Polysulfone and Ultem Via Molecular Simulation, Industrial & Engineering

Chemistry Research, 49 (2010) 12014-12021.

[23] H. Lin, B.D. Freeman, Materials Selection Guidelines for Membranes That

Remove Co2 from Gas Mixtures, Journal of Molecular Structure, 739 (2005)

57-74.

[24] I. Pinnau, Z. He, Pure- and Mixed-Gas Permeation Properties of

Polydimethylsiloxane for Hydrocarbon/Methane and Hydrocarbon/Hydrogen

Separation, Journal of Membrane Science, 244 (2004) 227-233.

[25] T.C. Merkel, V.I. Bondar, K. Nagai, B.D. Freeman, I. Pinnau, Gas Sorption,

Diffusion, and Permeation in Poly(Dimethylsiloxane), Journal of Polymer

Science Part B: Polymer Physics, 38 (2000) 415-434.

[26] H. Lin, E. Van Wagner, B.D. Freeman, L.G. Toy, R.P. Gupta,

Plasticization-Enhanced Hydrogen Purification Using Polymeric Membranes,

Science, 311 (2006) 639-642.

[27] H. Lin, B.D. Freeman, S. Kalakkunnath, D.S. Kalika, Effect of Copolymer 90

Composition, Temperature, and Carbon Dioxide Fugacity on Pure- and

Mixed-Gas Permeability in Poly(Ethylene Glycol)-Based Materials: Free

Volume Interpretation, Journal of Membrane Science, 291 (2007) 131-139.

[28] H. Lin, B.D. Freeman, Gas Solubility, Diffusivity and Permeability in

Poly(Ethylene Oxide), Journal of Membrane Science, 239 (2004) 105-117.

[29] H.Y. Zhao, Y.M. Cao, X.L. Ding, M.Q. Zhou, J.H. Liu, Q. Yuan, Poly(Ethylene

Oxide) Induced Cross-Linking Modification of Matrimid Membranes for

Selective Separation of Co2, Journal of Membrane Science, 320 (2008)

179-184.

[30] W. Yave, A. Car, K.V. Peinemann, Nanostructured Membrane Material

Designed for Carbon Dioxide Separation, Journal of Membrane Science, 350

(2010) 124-129.

[31] S.R. Reijerkerk, M.H. Knoef, K. Nijmeijer, M. Wessling, Poly(Ethylene

Glycol) and Poly(Dimethyl Siloxane): Combining Their Advantages into

Efficient Co2 Gas Separation Membranes, Journal of Membrane Science, 352

(2010) 126-135.

[32] V.A. Kusuma, B.D. Freeman, M.A. Borns, D.S. Kalika, Influence of Chemical

Structure of Short Chain Pendant Groups on Gas Transport Properties of

Cross-Linked Poly(Ethylene Oxide) Copolymers, Journal of Membrane

Science, 327 (2009) 195-207.

[33] Y. Hirayama, Y. Kase, N. Tanihara, Y. Sumiyama, Y. Kusuki, K. Haraya, 91

Permeation Properties to Co2 and N2 of Poly(Ethylene Oxide)-Containing and

Crosslinked Polymer Films, Journal of Membrane Science, 160 (1999) 87-99.

[34] A. Car, C. Stropnik, W. Yave, K.V. Peinemann, Peg Modified

Poly(Amide-B-Ethylene Oxide) Membranes for Co2 Separation, Journal of

Membrane Science, 307 (2008) 88-95.

[35] Y. Li, T.S. Chung, Molecular-Level Mixed Matrix Membranes Comprising

Pebax® and Poss for Hydrogen Purification Via Preferential Co2 Removal,

International Journal of Hydrogen Energy, 35 (2010) 10560-10568.

[36] H. Chen, Y. Xiao, T.S. Chung, Synthesis and Characterization of Poly

(Ethylene Oxide) Containing Copolyimides for Hydrogen Purification,

Polymer, 51 (2010) 4077-4086.

[37] A. Car, C. Stropnik, W. Yave, K.V. Peinemann, Tailor-Made Polymeric

Membranes Based on Segmented Block Copolymers for Co2separation,

Advanced Functional Materials, 18 (2008) 2815-2823.

[38] A. Car, C. Stropnik, W. Yave, K.V. Peinemann, Pebax®/Polyethylene Glycol

Blend Thin Film Composite Membranes for Co2 Separation: Performance

with Mixed Gases, Separation and Purification Technology, 62 (2008)

110-117.

[39] D. Husken, J. Feijen, R.J. Gaymans, Hydrophilic Segmented Block

Copolymers Based on Poly(Ethylene Oxide) and Monodisperse Amide

Segments, Journal of Polymer Science Part A: Polymer Chemistry, 45 (2007) 92

4522-4535.

[40] M. Kawakami, H. Iwanaga, Y. Hara, M. Iwamoto, S. Kagawa, Gas

Permeabilities of Cellulose Nitrate/Poly(Ethylene Glycol) Blend Membranes,

Journal of Applied Polymer Science, 27 (1982) 2387-2393.

[41] R.D. Raharjo, H.Q. Lin, D.E. Sanders, B.D. Freeman, S. Kalakkunnath, D.S.

Kalika, Relation between Network Structure and Gas Transport in Crosslinked

Poly(Propylene Glycol Diacrylate), Journal of Membrane Science, 283 (2006)

253-265.

[42] V.A. Kusuma, S. Matteucci, B.D. Freeman, M.K. Danquah, D.S. Kalika,

Influence of Phenoxy-Terminated Short-Chain Pendant Groups on Gas

Transport Properties of Cross-Linked Poly(Ethylene Oxide) Copolymers,

Journal of Membrane Science, 341 (2009) 84-95.

[43] G. Kang, Y. Cao, H. Zhao, Q. Yuan, Preparation and Characterization of

Crosslinked Poly(Ethylene Glycol) Diacrylate Membranes with Excellent

Antifouling and Solvent-Resistant Properties, Journal of Membrane Science,

318 (2008) 227-232.

[44] S. Kalakkunnath, D.S. Kalika, H. Lin, R.D. Raharjo, B.D. Freeman, Molecular

Relaxation in Cross-Linked Poly(Ethylene Glycol) and Poly(Propylene Glycol)

Diacrylate Networks by Dielectric Spectroscopy, Polymer, 48 (2007) 579-589.

[45] M.D. Abramoff, Magelhaes, P.J., Ram, S.J., Image Processing with Imagej,

Biophotonics International, 11 (2004) 36-42. 93

[46] H. Wang, J.K. Keum, A. Hiltner, E. Baer, Impact of Nanoscale Confinement on

Crystal Orientation of Poly(Ethylene Oxide), Macromolecular Rapid

Communications, 31 (2010) 356-361.

[47] L.M. Robeson, The Upper Bound Revisited, Journal of Membrane Science,

320 (2008) 390-400.

[48] S.R. Reijerkerk, K. Nijmeijer, C.P. Ribeiro Jr, B.D. Freeman, M. Wessling, On

the Effects of Plasticization in Co2/Light Gas Separation Using Polymeric

Solubility Selective Membranes, Journal of Membrane Science, 367 (2011)

33-44.

[49] Z. Mogri, D.R. Paul, Gas Sorption and Transport in Side-Chain Crystalline and

Molten Poly(Octadecyl Acrylate), Polymer, 42 (2001) 2531-2542.

[50] Z. Mogri, D.R. Paul, Membrane Formation Techniques for Gas Permeation

Measurements for Side-Chain Crystalline Polymers, Journal of Membrane

Science, 175 (2000) 253-265.

[51] J.H. Kim, S.Y. Ha, Y.M. Lee, Gas Permeation of Poly(Amide-6-B-Ethylene

Oxide) Copolymer, Journal of Membrane Science, 190 (2001) 179-193.

[52] T.C. Merkel, V.I. Bondar, K. Nagai, B.D. Freeman, I. Pinnau, Gas Sorption,

Diffusion, and Permeation in Poly(Dimethylsiloxane), Journal of Polymer

Science Part B-Polymer Physics, 38 (2000) 415-434.

[53] D. Pfefferkorn, S.O. Kyeremateng, K. Busse, H.-W. Kammer, T.

Thurn-Albrecht, J.r. Kressler, Crystallization and Melting of Poly(Ethylene 94

Oxide) in Blends and Diblock Copolymers with Poly(Methyl Acrylate),

Macromolecules, 44 (2011) 2953-2963.

[54] Y.K. Godovsky, G.L. Slonimsky, N.M. Garbar, Effect of Molecular Weight on

the Crystallization and Morphology of Poly (Ethylene Oxide) Fractions,

Journal of Polymer Science Part C: Polymer Symposia, 38 (1972) 1-21.

CHAPTER 5 The Effect of End Groups and Grafting on the CO2 Separation Performance of Polyethylene Glycol

Based Membranes

This chapter was published as a journal paper:

J. Xia, S. Liu, T.S. Chung, The effect of end groups and grafting on the CO2 separation performance of polyethylene glycol based membranes, Macromolecules, 44 (2011) 7727-7736

Abstract

Rubbery membranes have attracted significant attention for the CO2/light gas separation due to their high permeability and high solubility selectivity. This

chapter reports universal strategies to tailor CO2 permeability and CO2/light gas permselectivity via blending or grafting of poly (ethylene glycol) (PEG) derivatives into or onto organic-inorganic membrane (OIM) substrates. A PEG derivative is blended into the substrate, which contains inorganic siloxane networks and poly(ethylene oxide) segments, followed by thermal grafting. Ultra

high CO2 permeability (982 barrer at 45 ºC) was obtained via physical blending,

while an extremely high CO2 permeability (1840 barrer at 45 ºC) was obtained after chemical grafting. Neither of these two modification methods shows a loss of

CO2/H2 and CO2/N2 selectivity compared to the substrate. Melting and crystallization behavior of this PEG derivative are believed to significantly affect the overall gas permeation performance. All the features make these OIMs

promising membrane materials for CO2 capture.

5.1 Introduction

In chapter 4, the separation performance of OIMs was successfully enhanced by using a physical blending method [1].These OIMs with PEG up to 60 wt%

demonstrated CO2 permeability of 845 barrer with CO2/H2 and CO2/N2 permselectivity around 10 and 40, respectively. However, their performance could still be further improved by eliminating hydrogen bonding using non-hydroxyl groups as the end groups according to the trend reported by Yave at al. [2]. By modifying OIMs with O-zone initiated polymerization of poly(ethylene glycol

methacrylate) (PEGMA), Lau et al. [3] have achieved a CO2 permeability of 1920 barrer in mixed gas permeation tests. Thus, in this chapter two kinds of methoxy-PEG-azide (Mw = 350 and 1100) have been employed to replace PEG for comparison of end groups. Furthermore, azide functional groups provide the freedom to investigate the effect of PEG grafting on the overall gas permeation performance because they are reactive to the C-H bond. Due to the high crystallization tendency of higher molecular weight PEG, OIMs with

PEG-azide1100 are not amorphous at ambient temperature. Therefore, the investigation of temperature dependence of gas permeation has also been carried out.

5.2 Results and Discussion

5.2.1 Basic Physicochemical Properties

These reverse-selective OIMs are constructed by employing the in situ formed siloxane inorganic networks as the crosslinking junctions. The organic segments containing PEO chains are immobilized by these inorganic contents, as shown in

Figure 3-3. The inorganic components (black spots) are dispersed homogenously in the organic phase as indicated by the TEM micrograph in Figure 5-1a. The lattice features inside of the black spots as shown in Figure 5-1b prove that the inorganic siloxane networks have been successfully synthesized. The micromorphology analysis in details regarding to the shape and the size distribution could be found in our previous work [1]. PEG-azides of Mw350 and Mw1000 were blended into the pristine substrate (also referred as GP w/o PEG-azide). The blended hybrid OIMs are labeled by GPAX-Y, in which GPA represents the GOTMS-PEO PEG-azide blend, X represents the molecular weight of blended PEG, and Y represents the weight percentage of the blended PEG. For example, GPA1100-20 means the OIM contains 20 wt% of PEG-azide 1100g/mol.

Figure 5-1 TEM micrography of GP w/o PEG-azide at (a) a low magnitude and (b) a high magnitude

It is well known that PEGs and their derivatives have a strong crystallization tendency due to the highly polar ethylene oxide moieties. Meanwhile, the melting temperatures and crystallization temperatures of these PEG based materials are close to the ambient temperature, which makes the thermal history of these OIMs very unclear. Therefore, all the OIMs employed in this study were cooled to -20ºC first and then naturally warmed up to ambient temperature prior to characterization.

Because of the cooling induced crystals, GPA1100 series are translucent even at ambient temperature due to the relative high melting temperature of

PEG-azide1100 (above 25 ºC), as shown in Table 5-1. The membranes of GP w/o

PEG-azide and GPA350 series are transparent due to their lower melting temperatures (below 25 ºC).

100

Table 5-1 Thermal properties of GPA1100 series

a b c d Samples Tg (°C) Tm2p (°C) Tc 2p (°C) -ΔHc (J/g) -ΔH’c (J/g)

GP w/o PEG-azide N/A 24.5 -24.5 51 N/A

GPA1100-20 -51.9 41.1 2.5 76 71

GPA1100-40 -54.2 42.1 16.1 95 91

PEG-azide1100 NA 45.1 22.5 153 N/A

[a] Peak melting temperature of the 2nd heating curve [b] Peak crystallization temperature of the 2nd cooling curve

[c] ΔHc is the heat of crystallization obtained experimentally from DSC ’ [d] ΔH c is the theoretical value calculated by equation 5-1

5.2.2 Gas Transport Properties of OIMs with Physical Blending

Previous studies show that both CO2 permeability and CO2/H2 selectivity can

be enhanced by adding a low molecular weight PEG into an existing CO2 selective

material [1, 2, 4-9]. However, this kind of enhancement reaches a limit when

hydroxyl terminated PEG (normal PEG) is employed. The high cohesive energy

caused by the hydrogen bonding between the hydroxyl groups and the EO groups

inhibits further improvement on the CO2 diffusivity. In this work, PEGs terminated

with methoxy group and azide group are used to replace the normal PEG.

According to the previous work [1], the molecular weight of the blended PEG has a

significant effect on the gas permeation performance. Therefore, only PEG-azides

101

with similar molecular weights are used. GPA350 and GPA1100 series (here A

represents PEG-azide) are compared with GPP400 and GPP1000 series in our

previous work (the last P represents PEG), respectively. For easy comparison, only

the permeability data when the membranes are in amorphous status are listed

(Table 5-2).

Table 5-2 Pure gas permeability and selectivity of OIMs blended with PEG-azide

H2 N2 CO2 Temperature CO2/H2 CO2/N2 Sample permeability permeability permeability (°C) selectivity selectivity (barrer)a (barrer) (barrer) GP w/o 35 33 6.3 275 8.2 44 PEG-azide 45 49 9.9 344 7.0 35 35 54.5 (36)b 12.6 (7.3) 530 (333) 9.5 (9.3) 41 (46) GPA350-20 45 77.5 (53) 19.3 (11.7) 634 (413) 8.2 (7.8) 33 (35) 35 70.7 (37) 20.4 (7.5) 750 (363) 10.6 (9.7) 37 (48) GPA350-40 45 101 (54) 29.8 (11.9) 913 (455) 9.0 (8.5) 31 (38) GPA1100-20 45 62 (61) 13.9 (13.4) 468 (474) 7.5 (7.8) 34 (35) GPA1100-40 45 91 (80) 22.6 (18.3) 982 (662) 10.8 (8.3) 43 (36)

[a] 1 Barrer = 10-10 cm3 (STP).cm/cm2·s·cmHg [b] data in the blanket are from chapter 4 where PEG 400 and 1000 were used. In other words, GPA350 and GPA1100 series are compared with GPP400 and GPP1000 series, respectively.

With the addition of PEG-azide, the permeability of CO2 as well as other gases

shows surprising increments of up to 100%, compared to those blended with

normal PEG. The permeability improvement increases with more loading. Sorption

102

studies show that solubility is enhanced with more PEG-azide loading. Diffusivity calculated based on the solution-diffusion mechanism shows improvement as well.

The solubility enhancement could be attributed to a higher EO group density, while the diffusivity increment could be ascribed to the increasing chain flexibility, as indicated by the decreasing glass transition temperature as shown in Figure 5-2.

Figure 5-2 Glass transition temperature shifts of GPA1100 series by DSC

The permeability increment of samples with lower molecular weights

(GPP400 to GPA350 series) is greater than that of the higher molecular weight samples (GPP1000 to GPA1100 series) after replacing hydroxyl groups with azide groups. This is due to the fact that the end group concentrations are higher in

GPA350 and GPP400 than those in GPA1100 and GPP1000 at the same loading. 103

Without hydrogen bonding, the chain flexibility increases, thus the diffusivity of

CO2 in GPA350 and GPA1100 increases, as shown in Table 5-3. The CO2 solubility of GPA1100 series also increases slightly compared to that of GPP1000 series,

because more EO moieties are able to interact with CO2 after eliminating hydrogen

bonding. The CO2/H2 permselectivity increases because of the increased solubility selectivity. On the other hand, the size sieving effect has also been weakened because the polymer chains are more flexible after eliminating hydrogen bonding.

The slight decrease of CO2/N2 selectivity except GPA1100-40 is ascribed to the

weakened size sieving effect, because the kinetic diameter of CO2 (3.3Å) is smaller

than that of N2 (3.64 Å). The solubility and diffusivity of OIMs blended with 20 wt%

PEG and PEG-azide are almost the same because the effect of replacing the end group is not so obvious due to their low PEG loadings.

Table 5-3 Solubility and diffusivity coefficients of GPA1100 and GPP1000 series at 45 ºC

SCO2 DCO2 3 3 2 6 PEG blended (%) (cm STP/cm atm) (cm /s)x10

GPP1000 GPA1100 GPP1000 GPA1100

0 1.16 1.16 2.26 2.26

20 1.17 1.17 3.07 3.05

40 1.27 1.29 3.97 5.79

GPA1100-40 is also evaluated by a binary gas mixture of CO2/N2 (50:50 mole

104

fraction) at 45 ºC. Similar to our previous work [1], CO2 permeability (998 barrer)

is comparable to that of the pure gas (982 barrer), while the N2 permeability

increases from 22.6 barrer to 31.5 barrer. Therefore, the CO2/N2 selectivity

decreases from 43 to 32, which is caused by the CO2 plasticization [10].

5.2.3 Thermal Properties of GPA1100 Series

Thermal properties of polymeric membranes are critical to the gas permeation performance. This is especially true for the PEO-based rubbery membranes, because they are highly crystallizable materials and their melting temperatures are close to the normal permeation testing temperatures, 35 ºC. Because of their structure similarity, PEG-azide and PEO segments of GP w/o PEG-azide could crystallize with each other as revealed by the heat of crystallization analysis. In

Table 1, ΔHc is the heat of crystallization experimentally obtained from DSC, while

’ ΔH c is the theoretical value calculated using Equation 6 by assuming co-crystallization.

Hc ' Hc (a) wt%(PEG - azide)  Hc (b)(1 wt%(PEG - azide)) (5-1)

where ΔHc(a) is the heat of crystallization of pure PEG-azide, ΔHc(b) is the heat of crystallization of GP w/o PEG-azide. The experimental value is comparable to the theoretical value. Figure 5-3a and 5-3b are the 2nd heating and 2nd cooling DSC curves of GPA1100 series. GP w/o PEG-azide and pure PEG-azide1100 are also

105

included for comparison. A melting point depression (MPD) is observed for

GPA1100 series as shown in Table 5-1. Similar to the crystallization behavior of crystallizable block copolymers, the MPD in our case could be caused by the confined crystallization of PEG-azide in the OIMs [11]. The extending of the

PEG-azide crystals is restricted by the cross-linked network, which would result in a decrease of the melting temperature. The crystallization peaks of GPA1100 series are located between the peaks of GP w/o PEG-azide and pure PEG-azide. That the lowest crystallization temperature belongs to GP w/o PEG-azide indicates its lowest crystallization tendency, because the PEO segments in the GP w/o

PEG-azide are immobilized by crosslinking. When PEG-azide1100, which has a similar molecular weight to the PEO segments in the matrix, is added into the matrix, co-crystallization happens. PEG-azide1000 plays a role of crystallization intermediate between two PEO segments so that the membrane becomes easy to crystallize. When more PEG-azide is added, the crystallization tendency increases, thus resulting in a higher crystallization temperature as seen in Figure 5-3b.

106

Figure 5-3 DSC curves of (a) 2nd heating and (b) 2nd cooling of GPA1100 series

5.2.4 Temperature Dependence of Gas Permeation Properties

For a temperature range that does not involve thermal transition, there is a logarithmic relationship between permeability P and 1/T, which could be described by a van’t Hoff-Arrhenius equation (equation 2-16). Only few researchers had done their research in the temperature range that involves the thermal transition. Mogri and Paul [12, 13] developed a system to monitor the permeability change before and after side chain melting in acrylates. They found that both diffusivity and solubility increased above the melting temperature due to lower diffusion activation energy in the molten state and an increase in sorption sites. Okamoto et al. [14],

Hirayama et al. [15], and Kim et al. [16] have individually observed the discontinuity in the lnP versus 1/T curve in their PEO-based polymer systems.

Abrupt points around the melting temperatures indicate that the permeability jumps could be attributed to the melting of PEO crystals. However, detailed analyses on

107

the temperature dependence of the selectivity, especially for the CO2/H2 selectivity, have not been addressed.

Figure 5-4 shows the temperature dependence of permeability measured at

30-50 ºC and 2 atm. A similar permeability discontinuity is observed, especially for those with a higher PEG-azide loading. Below the melting temperature, PEG-azide crystals are impermeable physical obstacles that impede gas transport. In addition, the existence of crystalline domains restricts the chain motion in amorphous regions. Thus, the permeability declines with increased crystallinity. GPA1100-40 shows the lower permeability below the melting temperature, while GPA1100-20 has a higher permeability. After the crystals melt, the permeability increases with increasing EO group concentration. Therefore, the permeability trend above the

melting temperature is reversed. For GPA1100-40, the CO2 permeability could reach as high as 1052 barrer at 2 atm and 50 ºC. Permeability jumps happen between 35 and 40 ºC; however, the melting points measured by DSC are larger than 40 ºC as shown in Table 1. This may be caused by the temperature lag in the dynamic DSC measurement, as well as the facilitated melting at the presence of

compressed CO2 (discussed later) [17].

108

Figure 5-4 The temperature dependence of CO2 permeability of GPA1100 series

Figure 5-5a and b present the temperature dependence of CO2 sorption isotherms of GPA1100-20 and GPA1100-40. Most of isotherm curves are linear, except for GPA1100-40 at 35 ºC, where the sorption isotherm curves show a nonlinear trend when the pressure increases. The nolinear sorption isotherm may

be explained by the CO2 induced plasticization [10, 17]. The dissolved CO2 in the polymer matrix disassociates polymer chains, thus lowering the melting temperature of PEG-azide. Simultaneously, more effective EO groups are released

because of the melting, which further increases the sorption of CO2. When the loading of PEG-azide is low, this phenomenon is not obvious. However, when the

PEG-azide content increases, plasticization causes sorption increases so that the

109

GPA1000-40 isotherm shows an upward trend at 35 ºC. The solubility coefficient for the linear sorption isotherm is obtained by fitting the slope, while dC/dp at 2 atm is used to calculate the solubility for the nonlinear isotherms. Figure 5-6a shows the

temperature dependence of CO2 solubility and N2 solubility. Typical “∧”shaped

curves of solubility are observed for CO2, while the N2 solubility always increases with increasing temperature [15]. Below the melting temperature, the solubility is proportional to the volume fraction of amorphous regions, the solubility therefore decreases with more PEO-azide loading (i.e., higher overall crystallinity). After the crystals melt, this trend is reversed because OIMs with more liquid PEO-azide have

more EO groups favorable for CO2 sorption. When the temperature further

increases, CO2 solubility decreases because the heat of sorption ( H s ) is negative in

PEO based membranes [15, 16]. N2 solubility of these OIMs always increases despite of the thermal transition because is always positive for N2 [15, 16]. H2 sorption in this kind of material is too low to be measured accurately. However, we

believe that the H2 solubility would not change too much when the temperature

varies. Figure 5-6b shows the temperature dependence of CO2 diffusivity in

GPA1100-20 and GPA1100-40. The diffusivity below the melting points is comparable. Nevertheless, the diffusivity of GPA1100-40 increases tremendously after PEG-azide melts because of the high PEG-azide content. The temperature

dependence of the solubility and diffusivity may be further related to the CO2/H2

and CO2/N2 permselectivity as shown in Figure 5-7a and b. GPA1100-20 shows the 110

highest CO2/H2 permselectivity at 40ºC, at which the temperature the highest CO2 solubility is obtained. It is believed that the highest permeability selectivity comes from the highest solubility selectivity, because the selectivity in rubbery materials is governed by the solubility selectivity. With more EO groups, the solubility selectivity increases, thus resulting in a high selectivity of GPA1100-40 (12.3) than

that of GPA1100-20 (8.5). CO2/N2 permselectivity always declines with rising

temperature due to the decreasing CO2/N2 solubility selectivity and the weakened size sieving effect.

Figure 5-5 Temperature dependences of CO2 sorption isotherms of (a) GPA1100-20 and (b) GPA1100-40

111

Figure 5-6 Temperature dependence of (a) solubility and (b) CO2 diffusivity of GPA1100-20 and GPA1100-40

Figure 5-7 Temperature dependence of (a) CO2/H2 and (b) CO2/N2 selectivity of GPA1100-20 and GPA1100-40

5.2.5 Thermal Grafting of PEG-azide and Characterizations

Azido groups are linear 1, 3-dipolar structures that could easily form nitrenes, which are highly reactive and short-lived nitrogen intermediates, with the presence of ultraviolet radiation or heat. Nitrenes have four nonbonded electrons so that they

112

are able to insert into C-H bonds, following the reactivity order of tertiary > secondary > primary C-H bonds [18, 19]. The most possible thermal grafting mechanism is shown in Figure 3-3. The activated nitrenes attack the tertiary C-H bonds of PPO segments and form C-N bonds at 140 ºC under vacuum. 13C solid state NMR spectra in Figure 5-8 show the chemical environment change after thermal grafting. By comparing Figure 5-8a (GPA1100-40 pristine) and Figure 5-8b

(GPA1100-40 after thermal grafting), small peaks belonging to -CH2-N3 (g; δ =

51ppm) [20] disappear, indicating that azide groups are fully reacted. This correlates well with the nitrogen 1s binding energies change characterized by XPS as shown in Figure 5-9. Before thermal grafting, two distinct peaks (405 eV and

401 eV) with a peak areas ratio of 1:2, are assigned to the central, electron-deficient nitrogen and the other two nitrogen atoms in the azide group, respectively [21]. A small shoulder peak at 399 eV corresponding to the organic C-N bond [21] is also observed due to the existence of amine groups during matrix synthesis. Reaction of nitrenes with C-H bonds via insertion reduces the 401 eV and 405 eV peak to the level of noise. However, the intensity or the peak area of the C-N bond increases

after thermal grafting, thus indicating the successful conversion of C-N3 groups into

C-NH groups. Peak c (δ = 70.3 ppm) in Figure 5-8a and b is normally assigned to

the CH2 of PEG, while peak d (δ = 74.8 ppm) and e (δ = 71.7 ppm) are assigned to

CH2 and CH of poly (propylene glycol) [22]. Peak a (δ = 9.07 ppm), peak f (δ = 17.0 ppm) and peak b (δ = 22.9 ppm) at a lower δ value are assigned to the carbons of the 113

aliphatic part as shown in Figure 5-8a and b. After PEG-azide grafting, the relative intensity of the CH moiety (peak e) decreases and a small shoulder peak at a higher

δ value is observed. Part of e-type carbons are connected to the nitrogen atoms, which are more electronegative than the hydrogen atoms. Thus, a chemical shift of the carbon atoms at the grafting point moves to a higher δ value. Other types of grafting have not been seen in 13C-NMR spectra, which might be caused by their relative low reactivity and the low sensitivity of solid state 13C NMR. At 140 ºC , only grafting reactions are activated. The self-decomposition of PEG-azide1000 starts at 240 ºC as seen from TGA in Figure 5-10.

Figure 5-8 13C solid state NMR spectra of (a) GPA1100-40 pristine and (b) GPA1100-40 after thermal grafting

114

Figure 5-9 N1s XPS data before and after thermal grafting for GPA1100-40

Figure 5-10 TGA data of GPA1100-40 before thermal grafting

115

5.2.6 Gas Permeation Properties After Thermal Grafting

Gas permeability is almost doubled and the selectivity only slightly decreases

after thermal grafting as depicted in Table 5-4. The highest CO2 permeability

obtained is 1840 barrer with CO2/H2 and CO2/N2 selectivty of 8.3 and 36 at 2 atm and 45 ºC. Compared to the pristine substrate (GP w/o PEG-azide), the permeability is almost improved by 5-fold and the selectivity even slightly increases. With PEG grafting on the backbone of OIMs, the crystallization tendency is greatly depressed because PEG-azide is partially immobilized. The polymer chains of the blended

PEG cannot be rearranged easily and will create lots of chain ends and voids within the “mesh” of the OIMs. Therefore, the diffusivity will increase due to more free

volume created by these irregular chain ends and voids. Figure 5-11 shows the CO2

diffusivity and solubility changes after thermal grafting of GPA1100-40. The CO2

diffusivity shows a great improvement at any temperature. However, the CO2 solubility shows enhancement only below the melting temperature in pristine

GPA1100-40. Above the melting point, the solubility is slightly lower than that before grafting. This could be ascribed to the partial immobilization of PEG-azide.

Fewer EO groups are involved in the CO2 sorption because some of the EO groups are not available due to the steric hindrance. The increment of solubility at 30 and

35 ºC could be attributed to the amorphous nature of GPA1100-40 after grafting.

Figure 5-12 shows the WXRD patterns at ambient temperature of GP w/o

116

PEG-azide, GPA1100-40 before and after grafting. GP w/o PEG-azide is amorphous because of the cross-linked structure. GPA1100-40 is a semi-crystalline membrane at room temperature, which has two characteristic crystal peaks of PEO located at 2θ = 19.2 and 23.3 [23]. With the PEG grafting, GPA1100-40 does not shows obvious crystal peaks. DSC cooling curves have also revealed the suppressed crystallization tendency after grafting as shown in Figure 5-13. The crystallization temperatures of GPA1100-20 and GPA1100-40 drop from 2.5 ºC and 16.1 ºC to

-13.4 ºC and 8.5 ºC after grafting, respectively. Exclusion from crystallization at

ambient temperature makes these OIMs more valuable for CO2 separation in a broad temperature range. These preliminary results have shown a promising

method to further improve CO2 permeability of the OIMs.

Table 5-4 Pure gas permeability and selectivity of GPA1100 series before and after grafting

H2 N2 CO2 Temperature CO2/H2 CO2/N2 Sample permeability permeability permeability (°C) selectivity selectivity (barrer)[a] (barrer) (barrer) 35 36.7 7.54 321 8.7 43 GPA1100-20 45 62.3 13.9 468 7.5 34

GPA1100-20 35 70.4 13.6 595 8.5 44 after grafting 45 106 19.4 695 6.6 36 35 22.0 4.62 247 11.2 53 GPA1100-40 45 91.0 22.6 982 10.8 43 GPA1100-40 35 61.1 10.5 513 8.4 49 after grafting 45 222 50.5 1840 8.3 36

[a] 1 Barrer = 10-10 cm3 (STP).cm/cm2·s·cmHg 117

Figure 5-11 CO2 solubility (open) and diffusivity (solid) of GPA1100-40 before and after grafting

Figure 5-12 Wide angle XRD patterns of GP w/o PEG-azide, GPA1100-40 before and after the grafting at ambient temperature

118

Figure 5-13 2nd cooling curve of GPA1100-20 and GPA1100-40 before and after the grafting

Summary

This chapter explored strategies to increase the CO2 permeability without

sacrificing the selectivity for the CO2 selective OIMs. PEG terminated with one

azide and one methoxy group tends to enhance the CO2 permeability by a factor of 2 compared to hydroxyl end groups. A permeability jump is observed when the temperature surpasses the melting point of PEG-azide. The temperature dependences of permeability, solubility, diffusivity and selectivity are discussed in

detail. Maxima of CO2 solubility and CO2/H2 permselectivity are obtained in a given temperature range. Attempt to graft PEG chains onto the backbone of the

119

OIMs via azide reaction has been made. Solid state 13C-NMR, XPS and TGA

indicate that the PEG-azide has been successfully grafted. The CO2 permeability increases by 5-fold for grafted membranes as compared with the un-grafted substrate. The crystallization tendency of the grafted membrane has been greatly depressed. Diffusivity increment is believed to be the major reason for the

permeability enhancement. Extremely high CO2 permeability makes these OIMs

great potential as commercial membrane materials for CO2 capture. However, it is still a long way to fabricate these materials into hollow fibers or asymmetric membranes with a separation layer thickness in the range of 100 nm. Studies on making these materials into multilayer thin film composite membranes by continuous coating [24, 25] are ongoing in our group. Because PEO-based materials are hydrophilic and mechanically weak compared to the conventional glassy polymers, future studies focused on the long-term stability under water vapor exposure and various gas mixture feeds are critical to the next step of development.

120

References

[1] J. Xia, S. Liu, C.H. Lau, T.S. Chung, Liquid-Like Polyethylene Glycol

Supported in the Organic-Inorganic Matrix for Co2 Removal,

Macromolecules, 44 (2011) 5268-5280.

[2] W. Yave, A. Car, S.S. Funari, S.P. Nunes, K.V. Peinemann, Co2-Philic Polymer

Membrane with Extremely High Separation Performance, Macromolecules,

43 (2010) 326-333.

[3] C.H. Lau, S. Liu, D.R. Paul, J. Xia, Y.C. Jean, H. Chen, L. Shao, T.S. Chung,

Nanohybrid Membranes for Green Hydrogen Purification, Adv. Energy

Mater., 1 (2011) 634-642.

[4] A. Car, C. Stropnik, W. Yave, K.V. Peinemann, Peg Modified

Poly(Amide-B-Ethylene Oxide) Membranes for Co2 Separation, Journal of

Membrane Science, 307 (2008) 88-95.

[5] A. Car, C. Stropnik, W. Yave, K.V. Peinemann, Tailor-Made Polymeric

Membranes Based on Segmented Block Copolymers for Co2separation,

Advanced Functional Materials, 18 (2008) 2815-2823.

[6] A. Car, C. Stropnik, W. Yave, K.V. Peinemann, Pebax®/Polyethylene Glycol

Blend Thin Film Composite Membranes for Co2 Separation: Performance

with Mixed Gases, Separation and Purification Technology, 62 (2008)

110-117.

121

[7] M. Kawakami, H. Iwanaga, Y. Hara, M. Iwamoto, S. Kagawa, Gas

Permeabilities of Cellulose Nitrate/Poly(Ethylene Glycol) Blend Membranes,

Journal of Applied Polymer Science, 27 (1982) 2387-2393.

[8] W. Yave, A. Car, K.V. Peinemann, Nanostructured Membrane Material

Designed for Carbon Dioxide Separation, Journal of Membrane Science, 350

(2010) 124-129.

[9] W. Yave, A. Car, K.V. Peinemann, M.Q. Shaikh, K. Rätzke, F. Faupel, Gas

Permeability and Free Volume in Poly(Amide-B-Ethylene

Oxide)/Polyethylene Glycol Blend Membranes, Journal of Membrane

Science, 339 (2009) 177-183.

[10] S.R. Reijerkerk, K. Nijmeijer, C.P. Ribeiro Jr, B.D. Freeman, M. Wessling,

On the Effects of Plasticization in Co2/Light Gas Separation Using

Polymeric Solubility Selective Membranes, Journal of Membrane Science,

367 (2011) 33-44.

[11] B. Nandan, J.Y. Hsu, H.L. Chen, Crystallization Behavior of

Crystalline-Amorphous Diblock Copolymers Consisting of a Rubbery

Amorphous Block, Polymer Reviews, 46 (2006) 143 - 172.

[12] Z. Mogri, D.R. Paul, Membrane Formation Techniques for Gas Permeation

Measurements for Side-Chain Crystalline Polymers, Journal of Membrane

Science, 175 (2000) 253-265.

[13] Z. Mogri, D.R. Paul, Gas Sorption and Transport in Side-Chain Crystalline 122

and Molten Poly(Octadecyl Acrylate), Polymer, 42 (2001) 2531-2542.

[14] K.i. Okamoto, M. Fuji, S. Okamyo, H. Suzuki, K. Tanaka, H. Kita, Gas

Permeation Properties of Poly(Ether Imide) Segmented Copolymers,

Macromolecules, 28 (1995) 6950-6956.

[15] Y. Hirayama, Y. Kase, N. Tanihara, Y. Sumiyama, Y. Kusuki, K. Haraya,

Permeation Properties to Co2 and N2 of Poly(Ethylene Oxide)-Containing

and Crosslinked Polymer Films, Journal of Membrane Science, 160 (1999)

87-99.

[16] J.H. Kim, S.Y. Ha, Y.M. Lee, Gas Permeation of Poly(Amide-6-B-Ethylene

Oxide) Copolymer, Journal of Membrane Science, 190 (2001) 179-193.

[17] J.S. Wang, Y. Kamiya, Y. Naito, Effects of Co2 Conditioning on Sorption,

Dilation, and Transport Properties of Polysulfone, Journal of Polymer

Science Part B: Polymer Physics, 36 (1998) 1695-1702.

[18] B.T. Low, T.S. Chung, H. Chen, Y.C. Jean, K.P. Pramoda, Tuning the Free

Volume Cavities of Polyimide Membranes Via the Construction of

Pseudo-Interpenetrating Networks for Enhanced Gas Separation

Performance, Macromolecules, 42 (2009) 7042-7054.

[19] C.J. Moody, G.B. Whitham, Reactive Intermediates, Oxford University Press,

New York, 1992.

[20] M. Lammens, J. Skey, S. Wallyn, R. O'Reilly, F. Du Prez, Polymeric Ligands

as Homogeneous, Reusable Catalyst Systems for Copper Assisted Click 123

Chemistry, Chemical Communications, 46 (2010) 8719-8721.

[21] J.P. Collman, N.K. Devaraj, T.P.A. Eberspacher, C.E.D. Chidsey, Mixed

Azide-Terminated Monolayers: A Platform for Modifying Electrode

Surfaces, Langmuir, 22 (2006) 2457-2464.

[22] A.J. Brandolini, D.D. Hills, Nmr Spectra of Polymers and Polymer Additives,

Marcel Dekker, New York, 2000.

[23] H. Wang, J.K. Keum, A. Hiltner, E. Baer, Impact of Nanoscale Confinement

on Crystal Orientation of Poly(Ethylene Oxide), Macromolecular Rapid

Communications, 31 (2010) 356-361.

[24] T.S. Chung, E.R. Kafchinski, R.S. Kohn, P. Foley, R.S. Straff, Fabrication of

Composite Hollow Fibers for Air Separation, Journal of Applied Polymer

Science, 53 (1994) 701-708.

[25] J.J. Shieh, T.S. Chung, D.R. Paul, Study on Multi-Layer Composite Hollow

Fiber Membranes for Gas Separation, Chemical Engineering Science, 54

(1999) 675-684.

124

CHAPTER 6 Aging and Carbon Dioxide Plasticization of

Thin Extem® XH1015 Polyetherimide Films

This chapter was published as a journal paper:

J. Xia, T.S. Chung, P. Li, N. Horn, D.R. Paul, Aging and carbon dioxide plasticization in thin polyetherimide films, Polymer, 53 (2012) 2099-2108

125

Abstract

Industrial gas separation membranes have selective dense layers with thicknesses around 100 nm. It has long been assumed that these thin layers have the same properties as thick (bulk) films. However, recent research has shown that thin films with such thickness experience accelerated physical aging relative to bulk films and, thus, their permeation properties can differ significantly from the bulk.

Thin films made from Extem® XH 1015, a new commercial polyetherimide, have been investigated by monitoring their gas permeability. The permeability of the thin films is originally greater than the thick films but eventually decreases well below

the permeability of the thick film. The CO2 plasticization of Extem thin films is

explored using a series of exposure protocols that indicate CO2 plasticization is a function of film thickness, aging time, exposure time, pressure and prior history.

126

6.1 Introduction

Glassy amorphous polymers are of interest as gas separation membrane because of their good balance of permselectivity versus permeability [1]. During the past three decades, hundreds of commercial or lab synthesized glassy polymers have been screened for certain separation applications by evaluating their gas permeability in the form of thick dense films [2]. This is the very first step to select an appropriate material to develop a commercial membrane, e.g., asymmetric hollow fibers. From the perspective of productivity, the thickness of the dense thin selective layer of hollow fiber membranes should be no more than a few hundred nanometers. Due to the limitation of characterization techniques, the thickness of this dense layer of asymmetric membrane is usually obtained by dividing the gas permeability measured on a thick film by the permeance of the fabricated membrane rather than by direct measurement. This assumes that the dense thin selective layers (~100 nm) at the surface of the hollow fiber have the same permeability as thick dense films (~20 μm or more). However, a good deal of work from this laboratory over the past decade has shown that this assumption is not valid since the gas permeability of thin films is not only a function of thickness but also a function of time [3-21]. As a result, thick dense film data do not give a full picture of material performance as gas separation membranes.

127

A thermodynamic equilibrium state is not established when polymers are

cooled below their glass-transition temperature (Tg). The super-cooled polymer

“liquid” possesses a relatively high free volume because of the “frozen in” polymer segment packing. However, these materials slowly progress towards an equilibrium state [22] by the process called physical aging. The observed aging rate becomes much faster as the film becomes thinner for reasons that are still not well understood, but growing evidence suggests there is relative of high segmental mobility at free surface than in the bulk [23]. Because permeability is very sensitive to free volume, gas permeability is a convenient method for tracking the aging rate of thin films. This permeability decrease is of great significance for high flux membranes and has been observed for asymmetric hollow fiber membranes [24-26]; e.g., Lin and Chung [26] reported that the flux (or the permeance) of polyimide hollow fiber membranes was reduced by nearly 40% after aging for 720 hours. Due to the complicated nature of the hollow fiber membranes, the aging mechanism was not well recognized at that time. The integration of thin film aging data into the analysis of high flux membrane development has become possible via some recently developed methodology [6-20]. In this approach, thin films are first spin

coated onto a silicon wafer and then thermally annealed above Tg as free standing films, i.e., in an unconfined environment. To date, the focus of this work have been on polysulfone, Matrimid® 5218 and poly(2,6-dimethyl-1,4-phenylene oxide)

(PPO), which are the most extensively used polymers for commercial membranes 128

[6-10, 15-18] and a series of experimental hexafluoropropane dianhydride (6FDA) based copolyimides [11-14, 19, 20].

It is important to explore the thin film aging properties of other glassy amorphous polymer materials, especially commercially available products of potential interest as membrane materials, to better understand the factors that affect aging behavior and to guide selection of new membrane materials in a more holistic way. Polyetherimides have been extensively studied during past decades because they potentially have attractive properties and processability [27-31]. The first commercial polyethermide was Ultem® 1000, developed by General Electric

Plastics [32]; however, this polymer has so far not been amenable to our spin-coating techniques using common solvents. Hence, we focus on another polyetherimide known commercially as Extem® XH 1015, which has an even

higher Tg and better thermal properties, that can be successfully fabricated into free-standing thin films. This chapter explores the aging and carbon dioxide plasticization behavior of this new polyethermide and compares the results with those for PSF and Matrimid® 5218 from the previous work [18]. Particular attention

is devoted to the permeation response of thin films for a variety of CO2 conditions

due to the current strong interest in CO2 related separations, such as the removal of

CO2 from natural gas [33-37] and flue gases [38-41].

129

6.2 Results and Discussion

6.2.1 Aging Behavior Tracked by Gas Permeation

By knowing the glassy transition temperature of Extem as shown in Figure

6-1, the thin Extem films are heating above glass transition temperature for 15°C

under N2 atmosphere in order to define the “zero” time point.

Figure 6-1 DSC thermograph showing the glass-transition temperature of Extem

The aging behavior of Extem thin films was monitored by observing the

change in the gas permeability of oxygen, nitrogen and methane. Carbon dioxide was not used in the aging study because it will affect the aging process as described later due to its high level of solubility in the polymers. Preparation of thin glassy films requires great care and it is essential to ensure that any pinhole defects no

130

matter how small are sealed by the PDMS coating [42]. As a check on reproducibility, several Extem thin films were prepared from different batches but using the same protocol. The thicknesses were slightly different; however, as shown

by the O2 permeability results shown in Figure 6.2a, the data are quite reproducible.

Figure 6-2 Oxygen permeability of (a) Extem films with similar thickness and (b) Matrimid, PSF and Extem films as a function of aging time

131

Figure 6-2b compares the aging behavior of Extem thin films, as tracked by the

O2 permeability, with that of Matrimid and PSF; the latter data were taken from

previous work [15]. After 1000 hr of aging, the O2 permeability of Matrimid and

Extem is reduced by a factor of 2 compared to their initial values, while PSF shows somewhat less sensitivity to aging. However, for thick films only about a 10% reduction in permeability occurs over 1000 hours [7]. At this point, we cannot

interpret the different shapes of the aging curves observed. The initial O2 permeability values at short time are all larger than those for thick films as shown in

Table 3-2. It is believed that the rapid quenching from above Tg to room temperature leads to a somewhat higher free volume for the thin films. The initial

values of O2, N2 and CH4 permeability for 160 nm and 320 nm Extem films are larger than that of thick films as shown in Figures 6-3a, b and c. The relatively smaller oxygen gas molecule shows the least decline in permeability with aging, while nitrogen and methane show larger responses. In all cases, the 156 nm film has lower gas permeability coefficients and shows somewhat more rapid aging effects

than the 320 nm film. Figure 6-4 shows that the O2/N2 selectivity is initially lower than the bulk value but eventually reaches and exceeds the bulk value as aging proceeds. The thinner film has a somewhat higher selectivity than the 320 nm film.

These trends are consistent with the results observed by Huang and Paul [7] and

Kim et al. [11].

132

133

Figure 6-3 (a) Oxygen, (b) nitrogen, (c) methane permeability of Extem films with different thickness as a function of aging time

Figure 6-4 O2/N2 selectivity of Extem films as a function of aging time

The greater aging rate of the thinner film is consistent with the proposal of enhanced mobility at the free polymer surface [15, 23]. The initially lower

134

permeability for the thinner film is believed to be caused by aging occurring in the first few minutes prior permeation measurements based on the modeling by Rowe et al. [15].

6.2.2 CO2 Plasticization Pressure Curves

Plasticization of polymer membranes by CO2 is well known in the literature,

due to the high solubility of CO2 in the polymer, and can cause severe deterioration

of the CO2/light gas selectivity by increasing the permeability of all gas species,

especially at high CO2 partial pressure. There have been many reports on such behavior based on studies using dense thick films [34, 43-47]. However, there is growing evidence that this behavior can be quite different in thin films with thicknesses of the order of 100 nm; the latter are believed to better reveal how high flux commercial membranes respond. In thin films, physical aging of the glassy polymer is much more rapid than in thick films and may compete with the plasticization process. Zhou et al. have observed enhancement of plasticization in thin polyimide membranes [48], while Wessling et al. found similar behavior in composite membranes [49]. Horn and Paul [18] have reported a series of

experimental protocols for characterizing these competing effects of aging and CO2 plasticization for Matrimid, PSF and PPO thin films. They found that thin films are

more sensitive to changes in CO2 upstream pressure in permeation tests and

135

undergo more serious plasticization at high pressures. Film thickness, CO2 pressure, aging time, exposure time and prior history are evidently important factors influencing thin film plasticization behavior. Here, we compare the responses of

Extem to the protocols of CO2 exposure reported recently for Matrimid, PPO and

PSF. To facilitate their comparison, it is useful to understand the level of CO2

sorption in each of these polymers. As a guide for this, Figure 6-5 compares CO2 sorption isotherms for each of the polymers obtained on thick film samples. It would be useful to have sorption data for thin films; however, obtaining such data is a rather tedious process beyond the scope of this report. However, the sorption curve can be obtained by using an ellipsometric technique [50]. The solubility of

CO2 in Matrimid, PPO and PSF increases as the glass transition temperature of the polymer increases as expected from prior studies [51-53]; however, Extem deviates

somewhat from this trend. Sorption of CO2 in butyl rubber is shown for comparison.

136

Figure 6-5 CO2 sorption isotherms for thick films at 35⁰C: Matrimid [54], PSF [55], PPO [56], butyl rubber [53]

Figure 6-6a shows so-called CO2 plasticization curves for thin films for Extem

plus Matrimid, PPO and PSF; to facilitate comparison, the CO2 permeability for each polymer has been normalized by the permeability of that polymer at 2 atm. All the films are of similar thickness (160-200 nm) and aged for 100 hr at 35 ºC prior to

the CO2 test in order to achieve a standardized result. The films was first exposed to

2 atm of CO2 and then the pressure was incremented by 2 atm until reaching 20 atm, followed by increasing by 4 atm until reaching 40 atm. The films were held for 3 min (much longer than the time lag) at each pressure. The permeability data were collected during the final 60 s. The time interval has to be precisely controlled

because the CO2 permeability is strongly affected by the previous CO2 exposure

137

history. Matrimid undergoes the greatest relative change of CO2 permeability, while

PSF shows the least. The effect of CO2 concentration on plasticization can be

presented more directly by plotting the normalized permeability versus the CO2 concentration in the polymer (thick film data) at the upstream surface as shown in

Figure 6-6b. No universal trend is observed contrary to previous studies [46, 47]; this suggests that the plasticization is related to the nature of the polymer in addition to its sorption isotherm.

138

Figure 6-6 Normalized CO2 permeability of thin films as a function of (a) pressure

(Matrimid, PPO and PSF data are taken from the literature [18]) and (b) CO2 concentration

The observed CO2 plasticization curves for thin films are also a function of prior aging time as shown for Extem in Figures 6-7a and b. Two thin films were aged for 100 hr and 400 hr before plasticization evaluation. The film aged 100 hr seems to experience greater plasticization at high pressures as made evident when the permeability is normalized (Figure 6-7b) by using the value at 2 atm as the standard. At low pressures (less than 15 atm), the curves for the films aged 100 hr and 400 hr have almost the same shapes. However, the film aged 100 hr experiences a nearly 60% increase in permeability when the pressure rises to 40 atm while the

permeability of the film aged 400 hr only increases by 20%. Thus, CO2

plasticization strongly depends on the aging time of thin films. The reduction in free 139

volume during aging may be the key issue.

Figure 6-7 (a) CO2 permeability and (b) normalized CO2 permeability as a function

of CO2 pressure for different aging times

6.2.3 CO2 Permeability Hysteresis

In order to further explore the dynamics of plasticization, a protocol similar to

140

one used previously [18, 19] was employed for thin Extem films. The protocol consists of four parts: (1) pressurization from 2 atm to 32 atm with 10 min intervals at each pressure; (2) hold at 32 atm for 4 hr and record the permeability change; (3) depressurization from 32 atm to 4 atm with 10 min intervals at each pressure; (4) hold at 4 atm overnight and record the permeability change with time. During each pressurization stage, the permeability is measured at 2, 5 and 9 min. During each depressurization stage, the permeability is measured at 5 and 9 min. The purpose of this frequent measurement is to track changes over short exposure times. Three thin films with similar thicknesses but exposed to different prior aging times (25, 100 and 150 hr) were tested using this protocol.

Figures 6-8a and b show the CO2 hysteresis curves obtained for the thin Extem films; to make a clear comparison, the scales are the same in both figures. It is clear that the film aged 25 hr experienced the greatest plasticization during pressurization and hysteresis during depressurization. The extent of these effects follows the order

25 hr > 100 hr > 150 hr. The pressure where permeability is at a minimum, often called the “plasticization pressure” in thick film studies, shifted from 12 atm for the film aged for 25 hr to 20-24 atm for the film aged 150 hr. During the 10 minute observation interval, there are noticeable increases in permeability at all pressures for the film aged for 25 hr while such changes are only observed at 16 atm and beyond for the film aged for 100 hr. However, for the film aged for 150 hr, such 141

changes only occur at 28 atm of CO2 and beyond. It appears that aging, or the

associated densification, stabilizes these thin films against plasticization by CO2.

Figure 6-8 CO2 permeability hysteresis curves of thin films aged (a) 25 hr, (b) 100 hr and 150 hr

During the 4 hr holding at 32 atm of CO2, as shown in Figure 6-9a, the

142

permeability of the film aged for 25 hr increases significantly but eventually decreases with time. The film aged for 100 hr showed an even larger increase in permeability during this period before declining. The film aged for 150 hr showed the least increase during this time scale with no decline over the time frame studied.

We believe these effects are due to the competing dynamics of plasticization and physical aging. This will become clearer from observations over a longer time of

CO2 exposure as discussed later.

143

Figure 6-9 CO2 permeability as a function of time at (a) 32 atm (after increase CO2

pressure)and (b) 4 atm (after decrease CO2 pressure) during hysteresis testing

The response of the Extem thin films to depressurization is notably different from that of Matrimid and PPO studied previously [18]. The permeability keeps increasing and reaches the maximum point at 4 atm. This behavior does not mean that the film undergoes greater plasticization at lower pressures but a longer time

144

scale is needed for Extem to respond to the decreased CO2 content or to relax. With sufficient relaxation time, as shown later in Figure 6-9b, the films held at 4 atm of

CO2 overnight show a monotonic decline of permeability.

6.2.4 CO2 Permeation Behavior for Short Exposure Times

The previous discussions about CO2 permeation behavior are all based on

complicated protocols, where the CO2 pressure and exposure time are changed frequently. In this section, Extem thin films are exposed to 8, 16, 24 and 32 atm of

CO2 for 2 hr each. Matrimid and PSF data using the same procedure [18] are also included in Figures 6-10a and b for comparison. As discussed in the previous study

[18], Matrimid begins increasing in permeability upon each change in pressure, while PSF shown an opposite trend. Extem behaves more or less the same as

Matrimid. Under 8 atm of CO2, although the pressure is lower than the

“plasticization pressure”, the Extem thin films already experience plasticization.

145

Figure 6-10 (a) CO2 permeability and (b) normalized CO2 permeability in 2 hr CO2 exposure at programming higher pressures

146

6.2.5 CO2 Permeation Behavior over Long Exposure Times

The aforementioned experiments represent relatively short periods of exposure

to a CO2 feed while industrial membranes may be exposed over numerous years.

Thus, it is important to explore longer term effects of CO2 plasticization on these thin films. Three Extem thin films aged 200 hr were exposed to 8, 16 and 32 atm of

CO2 for roughly 200 hr as shown in Figures 6-11a and b. As might be expected, the

CO2 permeability of each film increases with time at the beginning; however, as reported previously, the permeability reaches a maximum and then declines. At 32 atm, the permeability occurs within 2 hr and reaches a value that is 50% greater than the initial value. At 8 atm, the film takes 100 hr to reach the maximum, and this maximum value is only 10% higher than the initial permeability. The film exposed

to 16 atm of CO2 shows intermediate behavior. As discussed in elsewhere [18], physical aging appears to be the dominate effect after the film has reached the maximum permeability. There is some reason to believe from these long term

exposure experiments that higher pressures of CO2 may accelerate the physical aging process.

147

Figure 6-11 (a) CO2 permeability and (b) normalized CO2 permeability during long

time CO2 exposure at different pressures

The long time exposure effects for Extem are compared with that of Matrimid

and PSF [18] in Figures 6-12a a~b and Figures 6-13a~b. At 8 atm of CO2, Matrimid shows the greatest relative maximum, while PSF shows a slight, but broad, 148

“turnover”. The exposure time for Extem to reach the maximum appears to be

longer than that of Matrimid. At 32 atm of CO2, Extem experiences the highest relative change in permeability at an exposure time of 2 hr. The difference between

Matrimid and PSF has been attributed to the differences in CO2 sorption. However, the data for Extem does not appear to fit this expectation, suggesting that these

effects are more complex than simply the amount of CO2 sorbed by the polymer.

However, one must remember that this is based on thick film data. More information about aging and plasticization behavior of other polymers may be needed to establish any universal rules.

149

Figure 6-12 (a) CO2 permeability and (b) normalized CO2 permeability as a function of exposure time at 8 atm for different polymer films

150

Figure 6-13 (a) CO2 permeability and (b) normalized CO2 permeability as a function of exposure time at 32 atm for different polymer films

Summary

The aging properties and CO2 plasticization behavior of a new polyetherimide

(Extem® XH 1015) have been systematically explored using different protocols developed in a previous paper from this laboratory [18]. The aging properties follow the same general trend as reported for a few other polymers of interest, including polysulfone and the polyimide known as Matrimid. Based on available

data for CO2 sorption in thick films made from these polymers, it appears that the

level of CO2 sorption itself may not be the unifying factor to explain the different

response to the CO2 exposure observed for these polymers. For a particular polymer,

CO2 plasticization is a function of film thickness, CO2 pressure, exposure time,

151

aging time and prior history. Indeed, one might speculate that the fractional free

volume and its change with time and CO2 exposure may be an important factor in this picture.

152

References

[1] R.W. Baker, Future Directions of Membrane Gas Separation Technology,

Industrial & Engineering Chemistry Research, 41 (2002) 1393-1411.

[2] L.M. Robeson, The Upper Bound Revisited, Journal of Membrane Science,

320 (2008) 390-400.

[3] M.S. McCaig, D.R. Paul, Effect of Uv Crosslinking and Physical Aging on the

Gas Permeability of Thin Glassy Polyarylate Films, Polymer, 40 (1999)

7209-7225.

[4] M.S. McCaig, D.R. Paul, Effect of Film Thickness on the Changes in Gas

Permeability of a Glassy Polyarylate Due to Physical Agingpart I.

Experimental Observations, Polymer, 41 (2000) 629-637.

[5] M.S. McCaig, D.R. Paul, J.W. Barlow, Effect of Film Thickness on the

Changes in Gas Permeability of a Glassy Polyarylate Due to Physical

Agingpart Ii. Mathematical Model, Polymer, 41 (2000) 639-648.

[6] Y. Huang, D.R. Paul, Experimental Methods for Tracking Physical Aging of

Thin Glassy Polymer Films by Gas Permeation, Journal of Membrane Science,

244 (2004) 167-178.

[7] Y. Huang, D.R. Paul, Physical Aging of Thin Glassy Polymer Films Monitored

by Gas Permeability, Polymer, 45 (2004) 8377-8393.

[8] Y. Huang, D.R. Paul, Physical Aging of Thin Glassy Polymer Films Monitored

153

by Optical Properties, Macromolecules, 39 (2006) 1554-1559.

[9] Y. Huang, D.R. Paul, Effect of Film Thickness on the Gas-Permeation

Characteristics of Glassy Polymer Membranes, Industrial & Engineering

Chemistry Research, 46 (2007) 2342-2347.

[10] Y. Huang, X. Wang, D.R. Paul, Physical Aging of Thin Glassy Polymer Films:

Free Volume Interpretation, Journal of Membrane Science, 277 (2006)

219-229.

[11] J.H. Kim, W.J. Koros, D.R. Paul, Physical Aging of Thin 6fda-Based

Polyimide Membranes Containing Carboxyl Acid Groups. Part I. Transport

Properties, Polymer, 47 (2006) 3094-3103.

[12] J.H. Kim, W.J. Koros, D.R. Paul, Effects of Co(2) Exposure and Physical

Aging on the Gas Permeability of Thin 6fda-Based Polyimide Membranes -

Part 1. Without Crosslinking, Journal of Membrane Science, 282 (2006)

21-31.

[13] J.H. Kim, W.J. Koros, D.R. Paul, Effects of Co(2) Exposure and Physical

Aging on the Gas Permeability of Thin 6fda-Based Polyimide Membranes -

Part 2. With Crosslinking, Journal of Membrane Science, 282 (2006) 32-43.

[14] J.H. Kim, W.J. Koros, D.R. Paul, Physical Aging of Thin 6fda-Based

Polyimide Membranes Containing Carboxyl Acid Groups. Part Ii. Optical

Properties, Polymer, 47 (2006) 3104-3111.

[15] B.W. Rowe, B.D. Freeman, D.R. Paul, Physical Aging of Ultrathin Glassy 154

Polymer Films Tracked by Gas Permeability, Polymer, 50 (2009) 5565-5575.

[16] B.W. Rowe, B.D. Freeman, D.R. Paul, Influence of Previous History on

Physical Aging in Thin Glassy Polymer Films as Gas Separation Membranes,

Polymer, 51 (2010) 3784-3792.

[17] B.W. Rowe, S.J. Pas, A.J. Hill, R. Suzuki, B.D. Freeman, D.R. Paul, A

Variable Energy Positron Annihilation Lifetime Spectroscopy Study of

Physical Aging in Thin Glassy Polymer Films, Polymer, 50 (2009) 6149-6156.

[18] N.R. Horn, D.R. Paul, Carbon Dioxide Plasticization of Thin Glassy Polymer

Films, Polymer, 52 (2011) 5587-5594.

[19] L. Cui, W. Qiu, D.R. Paul, W.J. Koros, Responses of 6fda-Based Polyimide

Thin Membranes to Co2 Exposure And physical Aging as Monitored by Gas

Permeability, Polymer, 52 (2011) 5528-5537.

[20] L. Cui, W. Qiu, D.R. Paul, W.J. Koros, Physical Aging of 6fda-Based

Polyimide Membranes Monitored by Gas Permeability, Polymer, 52 (2011)

3374-3380.

[21] T.M. Murphy, D.S. Langhe, M. Ponting, E. Baer, B.D. Freeman, D.R. Paul,

Physical Aging of Layered Glassy Polymer Films Via Gas Permeability

Tracking, Polymer, 52 (2011) 6117-6125.

[22] J.A. Forrest, K. Dalnoki-Veress, The Glass Transition in Thin Polymer Films,

Advances in Colloid and Interface Science, 94 (2001) 167-195.

[23] B. Jerome, J. Commandeur, Dynamics of Glasses Below the Glass Transition, 155

Nature, 386 (1997) 589-592.

[24] T.S. Chung, E.R. Kafchinski, Aging Phenomenon of

6fda-Polyimide/Polyacrylonitrile Composite Hollow Fibers, Journal of

Applied Polymer Science, 59 (1996) 77-82.

[25] T.S. Chung, S.K. Teoh, The Ageing Phenomenon of Polyethersulphone

Hollow Fibre Membranes for Gas Separation and Their Characteristics,

Journal of Membrane Science, 152 (1999) 175-188.

[26] W.H. Lin, T.S. Chung, The Physical Aging Phenomenon of 6fda-Durene

Polyimide Hollow Fiber Membranes, Journal of Polymer Science Part

B-Polymer Physics, 38 (2000) 765-775.

[27] K. Kneifel, K.V. Peinemann, Preparation of Hollow Fiber Membranes from

Polyetherimide for Gas Separation, Journal of Membrane Science, 65 (1992)

295-307.

[28] J. Kurdi, A.Y. Tremblay, The Influence of Casting Solution Structure on the

Microporosity of Polyetherimide Gas Separation Membranes Prepared by the

Coagulation Post-Leaching Method, Journal of Membrane Science, 184 (2001)

175-186.

[29] J.J. Shieh, T.S. Chung, R. Wang, M.P. Srinivasan, D.R. Paul, Gas Separation

Performance of Poly(4-Vinylpyridine)/Polyetherimide Composite Hollow

Fibers, Journal of Membrane Science, 182 (2001) 111-123.

[30] D.L. Wang, K. Li, W.K. Teo, Preparation and Characterization of 156

Polyetherimide Asymmetric Hollow Fiber Membranes for Gas Separation,

Journal of Membrane Science, 138 (1998) 193-201.

[31] D.L. Wang, K. Li, W.K. Teo, Preparation of Asymmetric Polyetherimide

Hollow Fibre Membrane with High Gas Selectivities, Journal of Membrane

Science, 208 (2002) 419-426.

[32] W.K. Greer, Ultem Polyetherimide Resin and Its Automotive Uses, Plastics

Engineering, 39 (1983) 34-34.

[33] Y.C. Xiao, B.T. Low, S.S. Hosseini, T.S. Chung, D.R. Paul, The Strategies of

Molecular Architecture and Modification of Polyimide-Based Membranes for

Co(2) Removal from Natural Gas-a Review, Progress in Polymer Science, 34

(2009) 561-580.

[34] J.D. Wind, D.R. Paul, W.J. Koros, Natural Gas Permeation in Polyimide

Membranes, Journal of Membrane Science, 228 (2004) 227-236.

[35] J.D. Wind, C. Staudt-Bickel, D.R. Paul, W.J. Koros, The Effects of

Crosslinking Chemistry on Co2 Plasticization of Polyimide Gas Separation

Membranes, Industrial & Engineering Chemistry Research, 41 (2002)

6139-6148.

[36] W.L. Qiu, C.C. Chen, L.R. Xu, L.L. Cui, D.R. Paul, W.J. Koros, Sub-T(G)

Cross-Linking of a Polyimide Membrane for Enhanced Co(2) Plasticization

Resistance for Natural Gas Separation, Macromolecules, 44 (2011)

6046-6056. 157

[37] H. Lin, E. Van Wagner, R. Raharjo, B.D. Freeman, I. Roman,

High-Performance Polymer Membranes for Natural-Gas Sweetening,

Advanced Materials, 18 (2006) 39-44.

[38] D. Aaron, C. Tsouris, Separation of Co2 from Flue Gas: A Review, Separation

Science and Technology, 40 (2005) 321-348.

[39] H.Q. Lin, E. Van Wagner, B.D. Freeman, L.G. Toy, R.P. Gupta,

Plasticization-Enhanced Hydrogen Purification Using Polymeric Membranes,

Science, 311 (2006) 639-642.

[40] J.Z. Xia, S.L. Liu, T.S. Chung, Effect of End Groups and Grafting on the

Co(2) Separation Performance of Poly(Ethylene Glycol) Based Membranes,

Macromolecules, 44 (2011) 7727-7736.

[41] C.H. Lau, S.L. Liu, D.R. Paul, J.Z. Xia, Y.C. Jean, H.M. Chen, L. Shao, T.S.

Chung, Silica Nanohybrid Membranes with High Co(2) Affinity for Green

Hydrogen Purification, Advanced Energy Materials, 1 (2011) 634-642.

[42] J.M.S. Henis, M.K. Tripodi, The Developing Technology of Gas Separating

Membranes, Science, 220 (1983) 11-17.

[43] L. Zhang, Y. Xiao, T.S. Chung, J. Jiang, Mechanistic Understanding of

Co2-Induced Plasticization of a Polyimide Membrane: A Combination of

Experiment and Simulation Study, Polymer, 51 (2010) 4439-4447.

[44] J.S. Chiou, D.R. Paul, Effects of Co2 Exposure on Gas Transport Properties

of Glassy Polymers, Journal of Membrane Science, 32 (1987) 195-205. 158

[45] E.S. Sanders, Penetrant-Induced Plasticization and Gas Permeation in Glassy

Polymers, Journal of Membrane Science, 37 (1988) 63-80.

[46] A. Bos, I.G.M. Pünt, M. Wessling, H. Strathmann, Co2-Induced

Plasticization Phenomena in Glassy Polymers, Journal of Membrane Science,

155 (1999) 67-78.

[47] J.D. Wind, S.M. Sirard, D.R. Paul, P.F. Green, K.P. Johnston, W.J. Koros,

Carbon Dioxide-Induced Plasticization of Polyimide Membranes:

Pseudo-Equilibrium Relationships of Diffusion, Sorption, and Swelling,

Macromolecules, 36 (2003) 6433-6441.

[48] C. Zhou, T.S. Chung, R. Wang, Y. Liu, S.H. Goh, The Accelerated Co2

Plasticization of Ultra-Thin Polyimide Films and the Effect of Surface

Chemical Cross-Linking on Plasticization and Physical Aging, Journal of

Membrane Science, 225 (2003) 125-134.

[49] M. Wessling, M. Lidon Lopez, H. Strathmann, Accelerated Plasticization of

Thin-Film Composite Membranes Used in Gas Separation, Separation and

Purification Technology, 24 (2001) 223-233.

[50] N.R. Horn, D.R. Paul, submitted to Macromolecules, (2012).

[51] S. Kanehashi, K. Nagai, Analysis of Dual-Mode Model Parameters for Gas

Sorption in Glassy Polymers, Journal of Membrane Science, 253 (2005)

117-138.

[52] K. Toi, G. Morel, D.R. Paul, Gas Sorption and Transport in Poly(Phenylene 159

Oxide) and Comparisons with Other Glassy Polymers, Journal of Applied

Polymer Science, 27 (1982) 2997-3005.

[53] D.R. Paul, in: Drioli E, Giorno L (Eds.) Comprehensive Membrane Science

and Technology, Oxford: Academic Press, 2010, pp. 75-90.

[54] W.C. Madden, D. Punsalan, W.J. Koros, Age Dependent Co2 Sorption in

Matrimid® Asymmetric Hollow Fiber Membranes, Polymer, 46 (2005)

5433-5436.

[55] Y. Maeda, D.R. Paul, Effect of Antiplasticization on Gas Sorption and

Transport. I. Polysulfone, Journal of Polymer Science Part B: Polymer Physics,

25 (1987) 957-980.

[56] Y. Maeda, in, University of Texas at Austin, 1985, pp. p. 413.

160

CHAPTER 7 Gas Permeability Comparison of Extem®

XH1015 with Polysulfone and Ultem® via Molecular

Simulation

This chapter was published as a journal paper:

J. Xia, S. Liu, P.K. Pallathadka, M.L. Chng, T.-S. Chung, Structural determination of Extem XH 1015 and its gas permeability comparison with polysulfone and ultem via molecular simulation, Industrial & Engineering Chemistry Research, 49 (2010) 12014-12021

161

Abstract

Extem® XH 1015 is a new brand of polyetherimide with good thermal, mechanical properties and processability. The gas permeability of this new polymer is reported for the first time. Polysulfone (PSU) and Ultem® are employed as reference samples for the elucidation of permeability and selectivity differences among them because of their structural similarities. In addition to qualitative comparison of chain rigidity and packing with gas transport properties, computational simulations powered by Material Studio are performed at a molecular level to quantitatively investigate the relationship between the fractional accessible volume (FAV) and gas permeability. The FAV differences among these polymers increase with an increase in gas molecules diameters; thus these polymers have similar permeability for small gas molecules but diverse for large gas molecules. Their selectivity differences are also discussed in terms of FAV ratio.

The FAV concept is proved to be more effective than free volume to analyse and predict gas separation performance.

162

7.1 Introduction

Throughout history, human beings have witnessed many evolutions involving materials. The emergence of engineered plastics is of interest. Polyimide is one of these high performance engineering plastics aiming at replacing metals as well as targeting other specialty applications such as electronics, automotives, and membranes. However, not all the polyimide materials meet industrial specifications because of their poor processability rendered by high chain rigidity and high glass transition temperatures. One of the exceptions is PEI (polyetherimide), which is a particular class of polyimides that not only retains high temperature characteristics but also possesses good melt processability. It has balanced physicochemical properties between performance and practicability due to the introduction of ether group. One of the commercialized PEI was developed by GE (general electric)

Plastic in 1986 under the trademark of Ultem. It was synthesized from bisphenol-A dianhydride (BPADA) and meta-phenylene diamine (MPD) through condensation polymerization [1]. This polyetherimide has already been employed in many applications, such as healthcare applications, pharmaceuticals and membrane separations. Gas separation membrane made from Ultem in Paul’s group [2] has shown impressively high selectivity for many gas pairs, especially for helium/methane.

163

In recent years, Extem XH1015, a new brand of polyetherimide, was developed by GE, and has been commercialized in Europe since 2007. This so called amorphous thermoplastic polymer offers a higher Tg (267°C), exceptional dimensional stability, high strength, stiffness, and creep resistance at elevated temperatures, outstanding flame, smoke and low smoke-toxicity performance without additives. Even though the Extem XH resin is not the strongest material at room temperature, it outperforms many other high-performance amorphous and semi-crystalline thermoplastics at temperatures exceeding 240°C by remaining most of its tensile strength and creep resistance [3]. Unlike other semi-crystalline thermoplastic polyimides, this fully amorphous Extem XH does not need any post heat treatment to achieve its desired properties.

Based on our previous understanding on polyetherimide [2], this new amorphous material may have potential to be a membrane material due to its higher

Tg as compared to Ultem. However, the dense Extem membrane does not show impressive gas separation performance. Further endeavor is hindered by the lack of information on its chemical structure to explain its structure relationship with gas permeability and to explore methodologies for chemical and physical modifications to enhance its application potentials. A bundle of patents [4-8] filed by GE provided few clues on this new PEI, but the detailed structure under the trademark of Extem had never been released. In appendix A, the chemical structure of Extem XH1015 164

has been determined by using NMR, elemental analysis and FTIR.

Interestingly, Extem’s main chemical repeating unit is quite similar to another grade of commercial PEI, named Ultem XH6050, of which the chemical structure also consists of a sulfone group [9]. This so-called PEIS (polyetherimide sulfone) combines the desired properties of polyetherimide and polysulfone into a single resin, which include good processability, high Tg and good thermal stability at high temperatures. The good melt processability may be derived from the unique characteristics of polyetherimde and polysulfone. Gallucci and Malinoski [10] reported that the end group of PEIS plays an important role on its Tg and thermal stability. PEIS, which has the same structure with Ultem XH6050, has a higher Tg

(258°C) by employing more anhydride as the end groups. This suggests that the same strategy may also be employed to improve the Tg of Extem XH1015. Their similar physical properties [3] indirectly support our hypothesis.

Since polysulfone (PSU) and Ultem contain part of the repeating unit in Extem as illustrated in Figure 7-1, the objectives of this chapter are to correlate its gas permeation properties with its chemical structure via both experiments and molecular simulation, and to further investigate its potential as a material for gas separation.

165

Figure 7-1 Chemical structures of PSU, Extem and Ultem

7.2 Results and Discussion

7.2.1 Chain Morphology Comparison

Figure 7-2 Chain morphologies of PSU, Extem and Ultem with 5 repeat units

Figure 7-2 shows the single chain morphologies of PSU, Extem and Ultem containing 5 repeating units at the completely extended state. Although the 166

morphologies are far from the real conformation in our membranes, it is still valuable for the qualitative discussion to correlate the membrane gas permeation performance with their chemical structures. As we can see in this figure, Ultem, being the most rigid polymer, takes a much extended chain conformation; while

PSU, the softest one, exhibits a zigzag like morphology. Usually, an increase in chain stiffness is accompanied with a decrease in segmental mobility. As a result, from PSU, Extem to Ultem, the thermal motions generating transient gaps for penetrants to diffuse decrease. These calculated chain morphologies correlate well with their gas selectivity as shown in Table 7-1.

Table 7-1 Gas permeation performance of flat dense PSU, Extem and Ultem membranes

a 1 Barrer = 1×10−10 cm3(STP) cm/cm2 s cmHg = 7.5005×10−18 m2 s−1 Pa−1; b Density simulated from Material Studio; c FFV Calculated from Bondi-Park & Paul’s method; d FFV calculated from Material Studio.

The sequence of chain rigidity in these polymers can also be elucidated from their monomers. PSU and Extem both have bisphenol A and diamino diphenyl sulfone (DDS) moieties. The additional heterocyclic imide group in Extem induces 167

a planar and relatively larger ring in the backbone, which increases the chain rigidity. The introduction of a heterocyclic imide group also enhances the inter-chain attractions. As shown in Figure 7-3, where two Extem polymer chains are constructed at fully extended state, two heterocyclic imide rings are partially aligned parallel to each other to minimize the total energy. This inter-chain attraction effect increases the chain packing efficiency, thus leading to a lower permeability. Ultem also has the same phenomenon, while PSU polymer chains could not be aligned in parallel, which correlates well with its high permeability.

Although the discussion is only valid in the ideal case, the polymer chain packing will still follow the principle of minimum total potential energy in the real materials.

Comparing Extem with Ultem, the only difference is that the meta-phenylene diamine (MPD) moiety in Ultem is replaced by DDS in Extem. Since the DDS structure can reduce the effective chain packing due to tetrahedral configuration of

the -SO2- group, Extem has a higher permeability and a lower selectivity than

Ultem.

168

Figure 7-3 Morphology of two polymer chains with 5 repeat units for PSU, Extem and Ultem

7.2.2 Molecular Simulation

The FFV values calculated by both the Bondi-Park & Paul method [11] and

Material studio were listed in Table 7-1 in order to compare their gas separation performance quantitatively. PSU has the largest FFV, which correlates well with experimental data of the highest permeability. However, there is a contradiction when it comes to Extem and Ultem. The FFV value of Extem calculated by the

Bondi-Park & Paul method is smaller than that of Ultem, while the permeability of

Extem is much larger than that of Ultem. Obviously, the FFV concept alone cannot interpret this discrepancy because FFV just gives a sum of the total free volume with no differentiation for different penetrants with different dimensions. In order to

169

explain this contradiction, the fractional accessible volume (FAV) distributions of

PSU, Extem and Ultem, as well as their relative FAV values raised by Chang et al.

[12] in terms of PSU/Ultem and Extem/Ultem were calculated and listed in Figure

7-4. In our case, the relative FAV values were defined, respectively, as follows:

FAV (PSU )  FAV (Ultem) 100% (7-1) FAV (Ultem) and

FAV (Extem)  FAV (Ultem) 100% (7-2) FAV (Ultem)

Figure 7-4 Fractional accessible volumes and relative FAV values of PSU, Extem and Ultem, probed with different diameters

The selection of the horizontal axis values is based on the kinetic diameter of the probe gas molecule, as shown in Table 7-2 for easy interpretation with permeation data. 170

Table 7-2 Kinetic diameters of various gases

For one particular gas, three FAV values were calculated from these three polymers. It can be seen that the FAV of PSU is always the largest, which correlates well with its highest permeability of all gases as presented in Table 7-1. While the

FAV values of Ultem are always the smallest, correlating well with its lowest gas permeability. Furthermore, for all the three polymers, gas permeability and the reciprocal of FAV follow linear relationships for relatively larger gas molecules, especially for methane and nitrogen as shown in Figure 7-5, even though their polymer structures are different. In Table 7-1, if can be seen that when the gas molecule diameter is small, the difference in their gas permeability is small, while when the gas diameter increases, this difference increases. For example, the helium permeability coefficients of PSU, Extem, and Ultem are 13, 11.1 and 9.4 barrers with a ratio of 1.38:1.18:1, while the methane permeability coefficients become

0.25, 0.13 and 0.036 barrers with a ratio of 6.94:3.61:1. These phenomena can be well explained using the relative FAV concept. As plotted in Figure 7-4, the relative

FAV values of PSU/Ultem and Extem/Ultem increase when the probe diameter increases. In other words, the FAV difference between PSU and Ultem becomes larger when the probe diameter increases. This indicates that the fractional

171

accessible volume is a more precise tool in analyzing the gas permeability than the total free volume.

Figure 7-5 Correlation between gas permeability and 1/FAV

The gas selectivity of polymer dense membranes shows a good correlation with the FAV ratio, which was defined for Extem and Ultem, as follows:

FAV (Extem) 100% (7-3) FAV (PSU ) and

FAV (Ultem) 100% (7-4) FAV (PSU )

As shown in Figure 7-6, the FAV ratio of Ultem/PSU is about 76% for helium, while it is only 63% for methane. This means, using PSU as a reference, that the fractional accessible volume for both helium and methane transport in Ultem is less than in

PSU. However, the decrease of FAV for methane is more severe than that for helium.

172

Thus, the helium/methane selectivity increases from 56 for PSU to 261 for Ultem.

The same approach could be used to explain the helium/methane selectivity increase in the Extem membrane. Another method is to investigate the slope of a straight line connecting any two data points of interest on the FAV curve. For any gas pair in one particular polymer, if the slope of the line between the selected two data points is negative, the selectivity of this gas pair for this particular polymer is higher than the reference polymer. For example, if helium and methane are chosen in the Ultem/PSU FAV ratio curve, the slope of line between these two points is definitely negative. Therefore, the selectivity of helium/methane is better in Ultem than in PSU. When comparing the gas selectivity of more than two polymers, this approach is still validate by just choosing one reference polymer. For example, if there are two straight lines linking helium and methane in the Ultem/PSU and

Extem/PSU FAV ratio curves, the absolute value of the slope for Ultem/PSU is obviously larger than that for Extem/PSU. This demonstrates that the helium/methane selectivity difference between PSU (56) and Ultem (261) is larger than that between PSU (56) and Extem (85). This method could explain other gas

pairs as well, like O2/N2. The absolute value of the slope between O2 and N2 in the

Extem/PSU FAV ratio curve is smaller than that in Ultem/PSU. This correlates well with their gas selectivity difference between PSU (5.6)/Extem (6.2) and PSU

(5.6)/Ultem (8.0), respectively.

173

Figure 7-6 FAV ratios of Extem/PSU and Ultem/PSU probed by different diameters

Summary

Gas permeability of Extem dense membranes was reported for the first time.

Due to the structural similarities, PSU and Ultem are used for comparison of fractional free volume (FFV), fractional accessible volume (FAV), gas permeability and selectivity. The effect of different monomer structure in their gas separation performance is also discussed in detail. Computational simulation powered by Material Studio, especially the FAV value simulation was shown to be a more accurate method to analyse and predict gas separation performance.

174

References

[1] Heath, D.; Wirth, J. Polyetherimides, US Patent 3847867, 1974.

[2] Barbari, T.A.; Koros, W.J.; Paul, D.R. Polymeric membranes based on

bisphenol-A for gas separations, J. Membr. Sci. 1989, 42, 69

[3] Lee, S. Extreme performance--or processability? new TP polyimide offers both,

Plastics Technology Magazine, January 2007,

http://www.ptonline.com/articles/200701fa6.html.

[4] Klopfer, H. Polyetherimides with high thermal stability and solvent resistance

and precursors therefor, US Patent 4565858, 1986.

[5] Takekoshi, T.; Klopfer, H. Method for making polyetherimides using

carboxylic acid salts of alkali metals or zinc as catalysts, US Patent 4293683,

1981.

[6] Gallucci, R.; Odle, R. Polyimide sulfone, method and article made therefrom,

US Patent 7041773, 2006.

[7] Evans, T.; Grade, M. Novel sulfur-containing polyetherimides, US Patent

4429102, 1984.

[8] Peters, E.; Bookbinder, D.; Cella, J. Very high heat thermoplastic

polyetherimides containing aromatic structure, US Patent 4965337, 1990.

[9] Odle, R.; Gallucci, R. New high heat polyetherimide resins, 61st SPE Antec

Tech. Conf. 2003, 1853.

175

[10] Gallucci, R.; Malinoski, J. Stabilization of polyetherimide sulfones, US Patent

7411014, 2008.

[11] Park, J.Y.; Paul, D.R. Correlation and prediction of gas permeability in glassy

polymer membrane materials via a modified free volume based group

contribution method. J. Membr. Sci. 1997, 125, 23.

[12] Chang, K.S.; Tung, C.C.; Wang, K.S.; Tung, K.L. Free volume analysis and

gas transport mechanisms of aromatic polyimide membranes: a molecular

simulation study, J. Phys. Chem. B. 2009, 113, 9821.

[13] McHattie, J.S.; Koros, W.J.; Paul, D.R. Gas transport properties of

polysulfones: 1. Role of symmetry of methyl group placement on bisphenol

rings. Polymer 1991, 32, 840.

[14] Barbari, T.A. Ph.D. Thesis. The University of Texas at Austin, 1986.

176

CHAPTER 8 Conclusions and Recommendations

8.1 Conclusions

Gas separation membranes have been evaluated for removing CO2 from gas

mixtures, mainly CO2/H2 and CO2/N2. This thesis shows that blending poly

(ethylene glycol) (PEG) and its derivatives into organic-inorganic membranes

(OIMs) is an efficient approach to increase the ethylene oxide content and

decrease the crystallinity of PEG. Thus, high CO2 permeability and CO2/light gas selectivity could be achieved simultaneously. By monitoring the gas permeability and selectivity of glassy thin films, accelerated physical aging and its competition

with CO2 plasticization was observed. This finding greatly enhances our understanding on the flux decrease of hollow fiber modules along with the time.

The preliminary results obtained from simulation works also reveal a possibility of replacing experiments by performing computational simulations.

8.1.1 Permeability and Selectivity Enhancement by Blending PEG and its

Derivatives

Organic-inorganic membranes (OIMs) based on PEG and siloxane is a great matrix for modification with low molecular weight PEG and its derivatives.

177

Within the confined space, PEG and its derivatives behave like a liquid in a stable state. The overall permeability is greatly enhanced by the extraordinary increase

of CO2 diffusivity caused by addition of PEG. The selectivity is improved slightly by adding more ethylene oxide groups, in other words, by increasing the content of blended PEG. The end groups of PEG derivatives also play an important role to determine the extent of permeability enhancement. Hydrogen bondings formed between hydroxyl groups of PEG prevent the further increase of gas diffusivity.

8.1.2 Permeability Enhancement by Grafting PEG-azide on the Backbone of OIMs

The crystallization tendency of PEG is greatly depressed after PEG-azide is partially immobilized on the backbone of OIMs. In addition, a high amount of chain ends and packing defects would possibly be created by this chemical grafting. Therefore, the diffusivity of gases increases tremendously following this modification.

8.1.3 Temperature Effect on Gas Permeability and Selectivity

A significant permeability jump is observed when the testing temperature surpasses the melting point of PEG or PEG-azide. Both diffusivity and solubility

178

are increased due to the melt of crystals. The crystals are impermeable because gas diffusivity and solubility in these domains are theoretically zero. When crystals melt, there is more material through which permeation can occur and the torture path around the crystal is eliminated. Thus permeability takes a large jump.

8.1.4 Physical Aging and Plasticization on Polymeric Thin Films

For the films with thickness in nanometer range, an accelerated physical aging is observed for several common glassy polymers. This phenomenon could be attributed to the higher mobility of surface compared to the bulk or the diffusion of free volumes from bulk to the surface. However, these two explanations have not been proved by experimental methods till now. For a

particular polymer, CO2 plasticization is a function of film thickness, CO2 pressure, exposure time, aging time and prior history. The aging rate increases when the film

becomes thinner. The higher the CO2 pressure imposed on the film, the faster the plasticization happens. With longer aging time prior to plasticization experiment, the film is harder to be plasticized. Indeed, one might speculate that the fractional

free volume and its change with time and CO2 exposure may be an important factor in this picture. With thinner films,

179

8.2 Recommendations

8.2.1 PEG Based Organic-Inorganic Membranes

In this dissertation, a series of poly (ethylene glycol) based organic-inorganic membranes have been developed, and the results appears quite promising.

However, the effect of water vapour, which is also one of the major components in the flue gas, has not been explored on this preliminary investigation. In fact, these

PEG based membranes may swell due to the existence of water vapour. Several single testings containing small amount of water vapor have been performed.

However, no conclusion could be reached at this time. Further works should explore more towards the effect of water vapour on these membranes.

All the OIMs were tested in thick film form. Therefore, it is a challenge to apply these materials as the selective dense layer of thin film composite membranes. Some preliminary efforts to coat these materials on polysulfone hollow fibers have been made. However, more work in this area will be conducted to convert these OIMs into high flux composite membranes. Severe intrusion of coating solution into ploysulfone support made these membranes less permeable.

Complicated preparation process of these OIMs and their shrinkage during crosslinking process may be other reasons preventing these composite membranes to be defect-free. The hydrolysis and condensation time may be the key to tailor

180

the viscosity of the coating solution.

8.2.2 Physical Aging and Plasticization Monitored by Gas Permeability

All the works on physical aging and CO2 plasticization are now based on pure gas permeation testing. Preliminary result shows that aging is a function of film

thickness, CO2 pressure, prior thermal and CO2 exposure history. The mechanism and dynamic of physical aging of polymers are not fully understood at current stage. It is believed that glassy transition temperature of thin films plays an important role to determine their aging characteristics from a review of basic polymer physics. Lots of literatures show a change of glass transition temperature when the thickness of films deceases into nano-meter range. However, there is no agreement till now.

Diffusivity and solubility are the two basic factors to determine the gas transport properties of a membrane. There is no mature technique to monitor the diffusivity for thin film right now. However, elliposometry had been proved as a pioneer technique to measure the solubility of gases in thin film. By knowing solubility, diffusivity can be obtained by a back calculation. This information can be extremely useful to understand the plasticization phenomenon in thin film.

Furthermore, it will be very interesting if the work could be extended to mixed

181

gases monitoring. Some transport properties of thin film, like solubility and diffusivity, already show differences with that of thick film, so the mixed gases permeation experiments may have different behaviours as well. In reality, the thickness of the dense selective layer of a gas separation membrane is only few nano-meters. Thus, the mixed gases experiments made on thin film will be much more meaningful to lead a discussion.

182

Appendix A: Structure determination of Extem® XH 1015

The structure of Extem® XH 1015 was unknown when it was purchased from supplier since it was newly commercialized. In order to make proper discussion on work using this material, the structure of this polymer was determined by using NMR, elemental analysis and FT-IR. This structure was determined in 2010 and verified by their company one year later.

Results

The strategy of determining the chemical structure Extem XH 1015 is to combine the information from patents and experiments. NMR, especially 2-D NMR was used to determine the protons’ position and their relationship with carbon. The 1H

NMR and 13C NMR spectra of Extem XH 1015 are illustrated in Figure A-1 and

Figure A-2, respectively.

Figure A-1 H-NMR spectrum of Extem XH1015 183

Figure A-2 C-NMR spectrum of Extem XH1015

The 2D-NMR spectra used to assign each carbon and proton in the aromatic region are given in Figure A-3 and Figure A-4.

Figure A-3 COSY spectrum of Extem XH1015

184

Figure A-4 HETCOR spectrum of Extem XH1015

In addition to NMR spectra and FTIR-ATR shown in Figure A-5, the elemental analysis results of this polyetherimide sulfone shown in Table A-1 also generally agree with the theoretical values for the proposed structure. Very minor differences are found due to experimental error and/or different element compositions of the polymer end group. As discussed previously, dianhydrate may be used to cap the polymer chain. As a result, the overall nitrogen and sulfur content is a little bit lower than the theoretical value.

185

Figure A-5 The FTIR-ATR spectrum of Extem XH1015 film

Table A-1 Experimental and theoretical elemental analysis

The aliphatic parts of 1H- and 13C-NMR spectra of Extem can be completely interpreted with the aid of the 1H-13C-HETCOR spectrum (not shown here because its position is too far away from the aromatic region). The adsorption of

C-12 carbon at 30.9 ppm of the 13C-NMR spectrum correlates well with a very sharp singlet of H-a protons at 1.64ppm of the 1H-NMR spectrum, revealing the existence of methyl group. C-13 at 42.5 ppm is also in the aliphatic region due to a relative low chemical shift compared to these carbons in the aromatic region

186

(110-170ppm). However, the 1H-13C-HETCOR spectrum does not show any proton connected to this carbon. The above evidences all strongly support a 2,2- substituted propane structure in the Extem’s backbone.

Although the area integration curve of all proton peaks is not shown in Figure A-1, the peak area ratio of a~h protons could be determined roughly to be 3:2:1:2:1:2:1:2, which is well coincident with the theoretical protons ratio based on the chemical structure of Extem. Fortunately, aromatic protons, except H-g, all belong to the AX two spin system [1]. Each peak of these protons is observed as a small doublet that

1 1 is split by the proton nearby. The H– H correlation (COSY) performed in CDCl3 is shown in Figure A-3, revealing that H-b (7.04-7.06 ppm) is split by H-d (7.28-7.30 ppm) through a 3-bond coupling. So do H-h (8.13-8.15 ppm) and H-f (7.72-7.74 ppm) pair, as well as H-e (7.63-7.65 ppm) and H-c (7.15-7.17 ppm) pair. The H-g at

7.18 ppm is assigned as an isolated proton because of its absence from the COSY spectrum, although it is overlapped by H-f.

Figure A-2 shows the signal assignment for a 13C NMR spectrum with 1H noise decoupled during acquisition. These carbon atoms at the high frequency are mainly attached to heteroatoms, making them valuable in the determination of monomers, especially diamino monomer. Carbon atom C-1 (165.9 ppm) is assigned to the imide carbonyl group. C-5 (164.2 ppm) and C-8 (152.7 ppm) constituting the 187

BPADA ether segment are assigned due to the direct connection to the oxygen atom.

A similar carbon chemical shift is also found for C-17 (147.5 ppm) in the polysulfone 13C-NMR spectrum [2]. The assignment of each peak marked in Figure

3 is assisted by the correlations shown in 1H-1H COSY (Figure A-3) and 1H-13C

HETCOR (Figure A-4) spectra to fit the structure of polyetherimide sulfone.

-1 In the case of FTIR-ATR spectra as shown in Figure A-5, bands at around 1781 cm

(attributed to C=O asymmetric stretch of imide groups), 1717cm-1 (attributed to

C=O symmetric stretch of imide groups) and 1360 (attributed to C-N stretch of imide groups) are characteristic imide peaks as indicated in Shao et al.’s work [3].

-1 Although parts of the band at 1320 cm (attributed to -SO2- asymmetric stretch) are overlapped by strong adsorption of the imide group at 1360cm-1, the existence of

-1 sulfone group can still be proved by the strong peak at 1150cm (attributed to -SO2- symmetric stretch ) [4].

References

[1] Friebolin, H. Basic one and two dimensional NMR spectroscopy 3rd ed.;

Weinheim: New York, 1998.

[2] Pham, Q.; Pétiaud, R. Proton and carbon NMR spectra of polymers 5th ed.;

Wiley: New York, 2003.

188

[3] Shao, L.; Chung, T.S.; Goh, S.; Pramoda, K. Transport properties of

cross-linked polyimide membranes induced by different generations of

diaminobutane (DAB) dendrimers. J. Membr. Sci. 2004, 238, 153.

[4] Bolong, N.; Ismail, A.F.; Salim, M.R.; Rana, D.; Matsuura, T. Development and

characterization of novel charged surface modification macromolecule to

polyethersulfone hollow fiber membrane with polyvinylpyrrolidone and water.

J. Membr. Sci. 2009, 331, 40.

189