Development of State-Of-The-Art Interfacially Polymerized Defect-Free

Thin-Film Composite Membranes for Gas- and Liquid Separations

Dissertation by

Zain Ali (Bajwa)

In Partial Fulfillment of the Requirements

For the Degree of

Doctor of Philosophy

King Abdullah University of Science and Technology

Thuwal, Kingdom of Saudi Arabia

Zain Ali

EXAMINATION COMMITTEE PAGE

The dissertation of Zain Ali is approved by the examination committee.

Committee Chairperson: Prof. Ingo Pinnau Committee Members: Prof. Yu Han, Prof. Mohamed Eddaoudi, Prof. Sandra Kentish.

ABSTRACT

Development of State-Of-The-Art Interfacially Polymerized Defect-Free

Thin-Film Composite Membranes for Gas- and Liquid-Separations

Zain Ali

This research was undertaken to develop state-of-the-art interfacially polymerized (IP) defect-free thin-film composite (TFC) membranes and understand their structure-function- performance relationships. Recent research showed the presence of defects in interfacially polymerized commercial membranes which potentially deter performance in liquid separations and render the membranes inadequate for gas separations.

Firstly, a modified method (named KRO1) was developed to fabricate interfacially polymerized defect-free TFCs using m-phenylene diamine (MPD) and trimesoyl chloride

(TMC). The systematic study revealed the ability to heal defects in-situ by tweaking the reaction time along with considerably improving the membrane crosslinking by controlling the organic solution temperature. The two discoveries were combined to produce highly crosslinked, defect-free MPD-TMC polyamide membranes which showed exceptional performance for separating H2 from CO2. Permeance and pure-gas selectivity of the membrane increased with temperature. H2 permeance of 350 GPU and H2/CO2 selectivity of ~100 at 140 °C were obtained, the highest reported performance for this application using polymeric materials to date.

Secondly, the membranes produced using KRO1 were tested for reverse-osmosis (RO) performance which revealed significantly improved boron rejection compared to

4 commercial membranes reaching a maximum of 99% at 15.5 bar feed pressure at pH 10.

The study also unveiled direct correlations between membrane crosslinking and salt separation performance in addition to the membrane surface roughness.

Thirdly, this was followed by replacing the conventional IP TMC monomer with a large, rigid and contorted tetra-acyl chloride (TripTaC) monomer to enhance the performance of

IP TFCs. The fabricated TFCs showed considerable performance boosts especially for separating of small solutes from organic solvents such as methanol. A rise in H2 permeance was also observed compared to the conventional MPD-TMC TFCs while reaching a maximum H2/CO2 selectivity of 9 at 22 °C.

Finally, the research was completed by showing the potential of KRO1 for fabrication of defect-free TFCs using a range of aqueous diamine monomers. KRO1 enabled defect-free gas properties for all monomers used showing exceptional performance for separating H2-

CO2 and O2-N2 mixtures. It was further shown that the formulation could also improve the

RO separation of interfacially polymerized polyamide TFCs beyond those shown by commercially available TFCs.

ACKNOWLEDGMENTS

This dissertation is dedicated to my parents who have spent their lives instilling the importance of education, hard work and being a good human-being in me and my siblings.

My father who has shown me the importance of responsibility and commitment and my mother who showed us the importance of love, relationships and to fight for the betterment of human beings. My amazing sisters, Zahra and Hira, and my incredible brother, Zawar, who have filled my life with love and pride. My wife, Emilie, who has been with me with the ups and downs of research life and whose contributions to this work and my life in

KAUST have been immeasurable. And the newest addition to our life, little Mustafa, a reminder of what matters in life. Without your love, this would all be meaningless.

My friends for many lifetimes; Humza, Tayyab, Haris, Machu, Moiz, Kamal, Asghar,

Ibrahim, Fawad, Lala, Tanwiri, Imran. Those who I have gotten to meet in KAUST;

Moustapha, Joel, Balawi, Luca, Pepe, Faheem, Anton, the Budget meal, the Grape Apes.

All of you have helped me become a better person.

Eric Litwiller, the complete engineer, who has selflessly shared all his knowledge, skills, expertise. His critical thinking and application of the scientific method will serve as the gold standard for me. My colleagues and collaborators; Federico, Giuseppe, Jintang, Ramy,

Mahmood, Wojciech, Ainur, Khalid, Yasmeen, Anton, Yingge, Prof. Udo, Hakkim,

Nasser, Fahd, Rakan, Saud, Aziz, Dr. Ma, Dr. Yichang Pan, whose kindness and work have made this Ph.D. bearable in the brutal world of academia.

My academic supervisor, Prof. Ingo Pinnau, for his guidance and the incredible amount of knowledge and expertise he has shared with all our group and furthermore for the fantastic

6 team he assembled. And finally a big thanks to the committee members, Prof. Yu Han,

Prof. Mohamed Eddaoudi and, Prof. Sandra Kentish, who I am honored to have as a part of my Ph.D. defense.

Zain Ali

KAUST

April 15, 2019

TABLE OF CONTENTS

EXAMINATION COMMITTEE PAGE ...... 2

ABSTRACT ...... 3

ACKNOWLEDGMENTS ...... 5

TABLE OF CONTENTS ...... 7

LIST OF FIGURES ...... 13

LIST OF TABLES ...... 19

Chapter 1. Membrane-based Separation: Past, Present, and Future ...... 21

1.1. Early Development of Membrane Science ...... 21

1.1.1. Conceptual Development ...... 21

1.1.2. Towards Commercialization ...... 22

1.1.3. Advantages of Membrane Technology...... 27

1.2. Current State-Of-The-Art Membranes ...... 30

1.2.1. Membranes for Gas Separation ...... 30

1.2.2. Membranes for Liquid Separations ...... 38

1.3. The Future of Membrane Science ...... 43

1.3.1. Membranes for Gas Separation ...... 43

1.3.2. Future Of Liquid Separations ...... 47

1.3.3. Promising Materials ...... 53

1.4. Dissertation Goals ...... 55

1.5. References ...... 58

Chapter 2. Theory and Background ...... 67

2.1. Gas Transport through Membranes ...... 67

2.1.1. Pore-Size-Dependent Transport ...... 69

2.1.2. The Solution-Diffusion Model ...... 72

2.1.3. Characterizing Transport ...... 78

2.1.4. Permeability/Selectivity Trade-Offs...... 80

2.1.5. Process Considerations ...... 82

2.2. Liquid Transport through Membranes ...... 91

2.2.1. Solution-Diffusion ...... 91

2.2.2. Characterizing Transport ...... 95

2.2.3. Permeability/Selectivity Trade-Offs...... 96

2.2.4. Process Considerations ...... 97

2.3. Interfacial Polymerization ...... 98

2.3.1. Introduction ...... 98

2.3.2. Mechanism of Thin-Film Formation ...... 104

2.3.3. Significant Fabrication Parameters ...... 107

2.3.4. Interfacially polymerized TFCs for gas separations ...... 112

2.4. References ...... 115

Chapter 3. Materials and Methods ...... 124

3.1. Materials ...... 124

3.1.1. Synthesis of Triptycene-1,3,6,8- Tetraacetyl Chloride (TripTaC) ...... 125

3.1.2. Synthesis of 1,3,6,8- Tetramethylanthracene (TMA)...... 126

3.1.3. Synthesis of 1,3,6,8-Tetramethyltriptycene (TMT)...... 126

3.1.4. Synthesis of Triptycene-1,3,6,8- tetracarboxylic Acid (TTC) ...... 126

3.1.5. Synthesis of Triptycene-1,3,6,8- Tetraacetyl Chloride (TripTaC) ...... 126

3.2. Membrane Fabrication ...... 127

3.3. Polymer Powder Preparation ...... 128

3.4. Gas Permeation ...... 129

3.4.1. Pure-Gas Permeation ...... 129

3.4.2. Mixed-Gas Permeation ...... 130

3.4.3. Temperature Dependence Measurements ...... 131

3.5. Liquid Permeation ...... 132

3.5.1. Water Desalination ...... 132

3.5.2. Rejection and Water/Solute Flux Calculations ...... 133

3.6. Modelling ...... 134

3.7. Sorption ...... 135

3.7.1. Low-Pressure Gas Sorption ...... 135

3.7.2. High-Pressure Gas Sorption ...... 136

3.7.3. Vapor Sorption ...... 137

3.8. Extended Characterization ...... 138

3.8.1. Fourier Transform Infrared (FTIR) Spectroscopy ...... 138

3.8.2. Scanning Electron Microscopy (SEM) ...... 138

3.8.3. Atomic Force Microscopy (AFM) ...... 138

3.8.4. Wide-Angle X-ray diffraction (XRD) ...... 139

3.8.5. X-ray Photoelectron Spectroscopy (XPS) ...... 139

3.8.6. Activation Energy of Permeation ...... 139

3.8.7. Thermogravimetric Analysis ...... 140

3.8.8. Membrane Surface Charge Measurements ...... 140

3.8.9. Ellipsometry ...... 140

3.9. References ...... 142

Chapter 4. Ultra-Selective Defect-Free Interfacially Polymerized Molecular Sieve

Thin-Film Composite Membranes for H2 Purification ...... 143

4.1. Abstract ...... 143

4.2. Introduction ...... 145

4.3. Results and Discussion ...... 149

4.3.1. Pure- and mixed-gas temperature dependence...... 161

4.4. Conclusions ...... 162

4.5. References ...... 164

Chapter 5. Defect-Free Highly Selective Polyamide Thin-Film Composite Membranes for Desalination and Boron Removal ...... 168

5.1. Abstract ...... 168

5.2. Introduction ...... 170

5.3. Results and Discussion ...... 174

5.4. Conclusions ...... 193

5.5. References ...... 195

Chapter 6. Triptycene-based Interfacially Polymerized Thin-Film Composite

Polyamide Membranes for Liquid- and Gas Separations ...... 199

6.1. Abstract ...... 199

6.2. Introduction ...... 201

6.3. Results and Discussion ...... 203

6.3.1. Interfacial Polymerization using a triptycene building block ...... 203

6.3.2. Fluid Transport Behavior ...... 208

6.4. Conclusions ...... 213

6.5. References ...... 215

Chapter 7. A Generalized Method for Fabricating Ultra-Selective Defect-Free

Interfacially Polymerized Thin-Film Composite Membranes ...... 217

7.1. Abstract ...... 217

7.2. Introduction ...... 219

7.3. Results and Discussion ...... 222

7.4. Conclusions ...... 238

7.5. References ...... 241

Chapter 8: Conclusions and Recommendations ...... 243

8.1. Conclusions ...... 243

8.2. Recommendations ...... 246

8.3. References ...... 248

Appendices ...... 249

Publications ...... 249

Chapter 2 ...... 250

Monomers used for aqueous phase in interfacial polymerization ...... 250

Monomers used for organic phase in interfacial polymerization ...... 257

Additional references ...... 260

LIST OF FIGURES

Figure 1.1. Performance comparison for seawater desalination membranes at 55 bar, 25 °C. Data from Baker [1]...... 24

Figure 1.2. Global market share of desalination technologies. RO, MSF, ED and MED refer to reverse osmosis, multi-stage flash, electrodialysis and multi-effect distillation, Data from Lee et al. [24]...... 27

Figure 1.7. Simplified schematic for reverse osmosis desalination [35]...... 39

Figure 1.8. O2/N2 selectivity vs. compressor power requirements. Assumed equivalent amount of nitrogen production (99% purity). Pressure ratio = 10. Reprinted with permission from Baker [1] (Copyright © 2012 John Wiley & Sons)...... 45

Figure 1.9. Typical two-stage natural gas sweetening process. (a). Cellulose acetate (CA) vs. (b). Hypothetical high-performance membrane. Adopted from Swaidan et al. [62]. .. 46

Figure 1.10. Structure of PIM-1 ...... 53

Figure 2.1. Pore size dependent transport processes. Adopted from Swaidan et al. [3]…...68

Figure 2.2. Transport of non-condensable and condensable components through meso/micropores. Adopted from Swaidan et al. [3]...... 71

Figure 2.5. Variation of gas sorption with absolute and relative pressure: (a) O2 in PDMS [25]; (b) acetone in PDMS [26]; (c) argon in PSF [27]; (d) vinyl chloride monomer in PVC

Figure 2.6. Diffusivity variation with penetrant concentration: (a) O2 in Teflon AF1600 [30]; (b) CO2 in cross-linked poly(ethylene glycol diacrylate) [31]; (c) water in ethylcellulose [32]; (d) CO2 in polycarbonate [33]. Reprinted with permission from Matteucci et al. [11] (Copyright © 2006 John Wiley & Sons)...... 86

Figure 2.7. Variation of permeability with upstream pressure: (a) N2 in PDMS [25]; (b) CO2 in cross-linked poly(ethylene glycol diacrylate) [34]; (c) CO2 in Lexan polycarbonate [35]; (d) CO2 in poly(tetrabromophenolphthalein) [36]. Reprinted with permission from Matteucci et al. [11] (Copyright © 2006 John Wiley & Sons)...... 88

Figure 2.9. Estimated tradeoff line between water/salt (NaCl) selectivity, PW/PS, and water permeability, PW, for commonly employed liquid separating polymers. Reprinted with permission from Geise et al. [37] (Copyright © 2010 Elsevier B.V.)...... 96

Figure 2.10. Nylon 6,10 via interfacial polymerization...... 99

Figure 2. 11. Fabrication of interfacially polymerized TFCs...... 100

Figure 2.12. (a) Cross-section of a generic TFC...... 101

Figure 2.13. Scanning electron micrograph of an MPDTMC-based interfacially polymerized TFC...... 102

Figure 2.14. MPD-TMC reaction at the interface...... 106

Figure 2.15. Aromatic polyamide structure via IP reaction between MPD-TMC...... 107

Figure 2.16. Fabrication parameters involved in interfacial polymerization...... 109

Figure 3.1. Constant pressure/variable volume pure-gas permeation system…………...129

Figure 3.2. Constant pressure/variable volume mixed-gas permeation system...... 131

Figure 3.3. Schematic diagram of salt rejection permeation setup [4]...... 133

Figure 3.4. Schematic of BET measurement using a Micrometrics ASAP 2020 system. Labeled components are: 1) degas line; 2) analysis line; 3) temperature control and; 4) vacuum line. Adopted from Swaidan et al. [7]...... 136

Figure 3.5. Schematic of Hiden IGA high-pressure sorption system. Adopted from Swaidan et al. [7]...... 137

Figure 4.1. Aromatic polyamide structure via interfacial polymerization reaction between MPD-TMC [30]………………………………………………………………………...148

Figure 4.2. Pure-gas separation performance of polyamide thin-film composite membranes. Effect of: a) and b) reaction time; c) and d) TMC concentration; e) and f) organic phase temperature on permeance and selectivity, respectively...... 150

Figure 4.3. Proposed in-situ pore plugging process during interfacial polymerization of thin-film composite membrane...... 151

Figure 4.4. FTIR spectra for polysulfone and TFCs in this study...... 154

Figure 4.5. XRD data for polyamide powder prepared by interfacial polymerization of trimesoyl chloride and m-phenylene diamine...... 155

Figure 4.6. Alternative gas pair selectivity data for fabricated TFCs: a) reaction time variation, b) TMC concentration variation and c) organic phase temperature variation. 156

Figure 4.7. Top surface (a, c, e, g, i, k) and cross-section (b, d, f, h, j, l) SEM images of fabricated TFCs...... 158

Figure 4.8. High magnification cross-section SEM images of fabricated TFCs for estimating polyamide layer thickness...... 159

Figure 4.9. Comparison of 300s-0.1TMC-60C and poly(p-phenylene terephthalamide) [43]: a). Pure-gas selectivity, and b) permeability data. 300s-0.1TMC-60C thickness estimated as 10 nm...... 160

Figure 4.10. (a) 300s-0.1TMC-100C pure-gas temperature dependence for H2 and CO2, and (b) Robeson plot for performance comparison of membrane studied here (300s- 0.1TMC-100C). Adapted from Robeson (2008) upper-bound assuming 1 µm films [44]. State-of-the-art data for USDOE requirements, PEBAX-coated SWC4 (PEBAX-SWC4), ZIF-8/PBI, MTR Proteus™ plotted separately from sources [14,20,31]...... 161

Figure 5.1. Proposed in-situ pore plugging process during interfacial polymerization of thin-film composite membranes………………………………………………………...173

Figure 5.2. (a) Effect on water permeance and salt rejection of: a) reaction time at fixed organic solution temperature of 20 °C, and; (b) organic solution temperature at fixed reaction time of 300 s. TMC concentration 0.1wt/vol% in Isopar G...... 177

Figure 5.3. Surface AFM images of in-house fabricated polyamide thin-film composite membranes. RMS refers to root-mean-square roughness...... 178

Figure 5.4. Surface (a, c and e) and cross-section (b, d and f); SEM images of in-house fabricated membranes for varying reaction time and organic solution temperature. The dotted line is provided to guide the eyes in order to identify the interface of the polyamide layer and the porous polysulfone support...... 180

Figure 5.5. Boron rejection as function of pH for KRO-1 and some commercial RO membranes. a) Dead-end filtration and b) crossflow filtration. The highlighted area is - B(OH)4 ion dominated region...... 181

Figure 5.6. Water over boron selectivity (Pw/PBoron) vs.: a) surface charge at pH 10, and; b- d) water permeance (Pw/l), for tested membranes at; b) pH 10; c) pH 8, and; d) pH 6. Membrane KRO-1 [CF] and UTC-80LB [CF] data were measured with crossflow system; all other membrane performance data were obtained with dead-end system...... 183

Figure 5.7. Variation of surface charge and boron selectivity at: a) pH 8, and; b) pH 6. 184

Figure 5.8. Effect of TMC concentration on water permeance and rejection...... 186

Figure 5.9. Surface (a, c and e) and cross-section (b, d and f) SEM images of in-house fabricated membranes for varying TMC concentration (0.1, 1 and 10 wt/vol%). The dotted line is provided to guide the eyes in order to identify the interface of the polyamide layer and the porous polysulfone support...... 188

Figure 5.10. Crosslinking N/O ratio vs.: a) membrane surface roughness ratio RR, and; b) water over sodium chloride selectivity (Pw/PS) for in-house fabricated membranes with fixed reaction time (300 s)...... 189

Figure 5.11. Surface SEM images of commercial membranes...... 190

Figure 5.12. Surface AFM images of commercial membranes...... 191

Figure 5.13. Roughness ratio vs. water over boron selectivity (Pw/PBoron) for KRO-1 and commercial RO membranes tested at pH 10...... 192

Figure 6.1. Characterization of interfacially polymerized polymers. (a) Reaction scheme for synthesis of crosslinked aromatic polyamide using MPD and triptycene-1,3,6,8- tetraacetyl chloride (TripTaC) monomer with internal free volume trapped between chains. (b) MPDTMC, and (c) MPDTrip, energy-minimized polymeric chains packed in an amorphous cell. Connolly surface area of MPDTMC estimated as 1900 Å3 and MPD- TripTaC as 4700 Å3. Blue color: accessible surface at probe radius of 1 Å. (d) Sorption behavior using CO2 BET measurements at 273 K up to 1 bar for bulk MPDTMC and MPDTrip powder. (e) XRD spectra of MPDTMC and MPDTrip bulk powders with average d-spacing obtained using Bragg’s Law………………………………………………….203

Figure 6.2. Pore size distribution for MPDTMC and MPDTrip powder obtained from CO2 isotherms employing the non-linear density function theory (NLDFT). TGA analysis of MPDTMC and MPDTrip powder...... 204

Figure 6.3. FTIR spectra for PAN support and TFCs fabricated in this study...... 206 Figure 6.4. Imaging of interfacially polymerized TFC membranes. (a) Cross-section SEM image of defect-free MPDTrip-100 membrane highlighting structural features. (b) High magnification SEM image of MPDTrip-100 TFC emphasizing predicted actual active layer

17 thickness and support/selective layer interface. (c) SEM top surface images depicting the TFCs produced. Samples produced using an organic solution temperature of 20 °C show two distinct top layer structures as highlighted by insets i-iv. (d) 3D visualization of MPDTMC-20 and MPDTrip-20 surface using AFM. (e) Ellipsometry images highlighting variance in surface morphology over larger sample areas, i.e., 5x5 mm for MPDTMC-20 and MPDTrip-20...... 207 Figure 6.5. Liquid separation properties of the membranes tested in this study. (a) Pure- solvent permeance through fabricated TFCs vs. the kinetic diameters of the tested solvents. (b) Rejection of Brilliant Blue R (826 g mol-1) through fabricated TFCs in water, methanol, ethanol and isopropanol. (c) Rejection of Sudan Orange (216 g mol-1) through fabricated TFCs in water, methanol, ethanol and isopropanol. (d) Structure of Sudan Orange G dye. e, Permeance of methanol vs selectivity for solutes between 210-250 g mol-1. Red stars denote MPDTrip-20 membranes from this study using Sudan Orange (216 g mol-1), blue circles denote MPDTMC membranes from this study using Sudan Orange (216 g mol-1), yellow squares denote data obtained from Solomon et al. [17] using Chrysoidine G (249 g mol-1), green triangles denote data obtained from Karan et al. [18] using 6-hydroxy- 2- naphthalenesulfonic acid sodium salt (HNSA) dye (246 g mol-1), light blue triangle denotes data obtained from Yang et al. [19] using Chrysoidine G (249 g mol-1) and gray triangles denote data obtained from Sorribas et al. [20] using polystyrene oligomers (230 g mol-1). (e) Sodium chloride rejection comparison for TFCs fabricated in this study...... 209 Figure 6.6. Vapor uptake for chosen solvents for: (a) MPDTMC powder, and (b) MPDTrip powder...... 210 Figure 6.7. (a) Pure-gas permeance of fabricated TFC membranes with respective to the kinetic diameters of the tested gases. (b) Isosteric heat of sorption obtained from CO2 sorption isotherms at 273 K and 298 K...... 213 Figure 7.1. Chemical structures of monomers used in this study………………………..222 Figure 7.2. Pure-gas permeance performance of commercial polyamide thin-film composite membranes...... 223 Figure 7.3. Pure-gas permeance performance of polyamide thin-film composite membranes using a) SRO and c) KRO-1 formulations. Pure-gas selectivity for chosen gas pairs for b) SRO and d) KRO1...... 224 Figure 7.4. Robeson plot for performance comparison of TFC membranes fabricated in this study. Adopted from Robeson (2008) upper-bound assuming 1 µm films [13]. Data for PEBAX coated SWC4 (PEBAX-SWC4), Polyamide + ZIF8 and MTR Proteus™ plotted separately from sources [6,14,15]...... 227

Figure 7.5. a) Sorption isotherms using CO2 BET measurements at 273 K up to 1 bar for bulk produced PIPTMC, PPDTMC and MPDTMC powder and b) XRD spectra of PIPTMC, PPDTMC and MPDTMC powder bulk powders with average d-spacing (using Bragg’s Law)...... 229

Figure 7.6. Salt rejection performance during BWRO (2000 ppm NaCl, 15.5 bar) and SWRO (35000 ppm NaCl, 55.5 bar) modes for MPD-, PPD- and PIP-based membranes using the: a) SRO recipe, and; b) KRO1 formulation protocol...... 230 Figure 7.7. Surface SEM images of in-house fabricated MPD (a-b), PPD (c-d) and PIP (e- f) TFC membranes using the SRO (a, c and e) and KRO1 (b, d, and f) formulation protocols...... 232 Figure 7.8. Surface AFM images of in-house fabricated MPD (a-b), PPD (c-d) and PIP (e- f) TFC membranes using the SRO (a, c and e) and KRO1 (b, d, and f) formulation protocols...... 234 Figure 7.9. Cross-sectional SEM images of in-house fabricated MPD (a-b), PPD (c-d) and PIP (e-f) TFC membranes using the SRO (a, c and e) and KRO1 (b, d, and f) formulation protocols...... 234 Figure 7.10. Crosslinking N/O ratio vs.: a) membrane surface roughness ratio, and; b) water over sodium chloride selectivity (Pw/PS). Red points denote data from this study. Blue points denote data from Ali et al. [10]...... 236

Figure 7.11. a) He vs. CO2 permeance, b) Sodium chloride (Ps/l) vs. water (A) permeance, c) He permeance vs He/CO2 selectivity and d) water over sodium chloride selectivity (Pw/PS) vs. water permeance. Red points indicate in-house fabricated TFCs. Blue points denote commercial membranes used in this study...... 237

LIST OF TABLES

Table 1.1. Energy cost comparison for desalination. Data from Pinnau et al. [25]...... 29

Table 1.2. H2/N2 separation performance for commercially used materials. Data from Baker [1]...... 32

Table 1.3. Commercial materials for nitrogen separation. Data from Baker [1]...... 34

Table 1.4. Commercially available liquid separation membranes. ISA = Integrally skinned asymmetric. HF = Hollow fibers. R and MWCO refer to rejection and molecular weight cut-off respectively...... 40

Table 4.1. USDOE specified requirements for H2/CO2 membranes [6,13,14]…………..146

Table 4.2. Membrane formation variables and sample information. Concentration = TMC concentration, temperature = organic phase temperature and ‘m’ = crosslinking degree. (N.M = not measured)...... 149

Table 4.3. Permeance data for prepared TFCs...... 151

Table 4.4. Selectivity data for prepared TFCs...... 152

Table 5.1. Boron rejection for commercially available FT-30-type membranes………171

Table 5.2. Fabricated membrane sample types...... 174

Table 5.3. Membrane water flux, permeance and sodium chloride rejection performance parameters (2000 ppm, 15.5 bar; T = 23 °C). A is water permeance in L m-2 h-1 bar-1 (commonly denoted as LMH/bar) and Pw/Ps is water over sodium chloride selectivity. 175

Table 5.4. Membrane characterization data for in-house fabricated membranes. RR and N/O measurements are made using AFM and XPS, respectively. Thickness was measured using ellipsometry. Membrane surface charge was measured using an EKA and reported for pH6...... 176

Table 5.5. Membrane characterization data for commercial membranes. RR measurements were determined using AFM...... 182

Table 6.1. Experimental conditions and properties of thin-film composite (TFC) membranes prepared in this study……………………………………………………….205

Table 6.2. Pure solvent permeation data for fabricated TFCs in LMH/bar...... 208

Table 6.3. Brilliant Blue R (826 g mol-1) rejection data for fabricated TFCs in %...... 211

Table 6.4. Sudan Orange G (216 g mol-1) rejection data for fabricated TFCs in %...... 211

Table 6.5. Salt rejection data for fabricated TFCs in %...... 212

Table 7.1. Characterizations for TFC membranes fabricated in this study. Apparent thickness measured using ellipsometry, the crosslinking ratio obtained from XPS measurements and roughness ratio calculated from AFM imaging……………………..223

Table 7.2. RO separation data for TFCs used in this study. A denotes water permeance in L m-2 h-1 bar-1 (commonly denoted as LMH/bar). BWRO experiments performed at 15.5 bar, 2000 ppm NaCl. SWRO experiments performed at 55 bar, 35000 ppm NaCl...... 231

Chapter 1. Membrane-based Separation: Past, Present, and Future

The chapter summarizes the history of membrane science including foundational theoretical observations, initial development of isotropic barrier layers and major advances and motivation towards the development of commercial thin-film gas- and liquid separation membranes.

1.1. Early Development of Membrane Science

1.1.1. Conceptual Development

Principles of separation of materials using membranes started to emerge as early as the late

18th century, e.g., the term ‘osmosis’ was first used in 1748 to describe water transport through a diaphragm. Early research in the technology involved using naturally available membranes such as bladders of cattle and pigs as laboratory tools. Traube and Pfeffer made osmotic pressure calculations using membranes, which were used to develop the van ’t

Hoff equation to explain the nature of dilute solutions. The model on an ideal selective membrane was developed by Maxwell among others to formulate the kinetic theory of gases which is still widely used to explain transport of gases through membranes [1].

In the 19th century, T. Graham and J. K. Mitchell were two of the pioneers of research on permeation of gases through what were considered solid barriers [2–4]. Graham noted the expansion of isolated balloons in a CO2 atmosphere, correctly explaining the passage of

CO2 from the atmosphere through the rubber balloon walls. Mitchell made three important observations: 1) natural rubber balloons inflated with different gas sources, e.g., hydrogen and air took different amounts of time to deflate; 2) he observed that the ‘velocity’ at which

22 the gas leaves the balloon depends on the gas, and; 3) he measured an increase in the weight of the rubber in a pure-gas atmosphere, indicating that a gas could sorb in the rubber [3].

Graham followed up on Mitchell’s work by publishing a paper confirming the permeation of different gases through natural rubber. He correctly identified the material as non-porous and dense and described the transport mechanism as a three-step procedure, i.e., 1) gas sorbs on the surface of the film; 2) gas diffuses across the film due to a chemical potential difference, and; 3) gas desorbs on the downstream side. This work laid the foundation of the solution-diffusion membrane transport model. Graham also identified a major potential application: enriching oxygen from air [5].

1.1.2. Towards Commercialization

The first commercial membranes appeared in the 20th century. Small-scale research into the use of membranes for microfiltration began in Germany in the 1920s. Berchold was one of the first to reproducibly fabricate nitrocellulose membranes and characterize their pore size using a bubble test [6]. This work was followed by several researchers and by

1930s, nitrocellulose-based microfiltration membranes were commercially available. In the next two decades, microfiltration was further explored with cellulose acetate. After the

Second World War, the technology acquired its biggest application at the time: testing water quality in European cities where water supply systems were damaged and contaminated during the conflict. The application truly showed the commercial potential of membranes and the work led to the development of the Millipore Corporation, still one of the world’s largest membrane manufacturers.

Around the middle of the 20th century, van Amerongen [7] and Barrer [8] laid the fundamental scientific and mathematical basis of transport across a semipermeable barrier.

Barrer, in particular, helped make great leaps for practical use of membrane technology for separation applications, addressing essential concepts of gas and liquid flows through polymers, transport in metals and capillary systems, and material properties such as diffusivity and solubility.

Membrane-based water purification, using microfiltration, ultrafiltration and reverse osmosis, was the earliest application of the technology. This was because water quality issues were considered a critical area of research due to its ubiquitous use in human activities. By the 1950s, cellulose acetate (CA) was already identified as a potentially useful material for membrane separations, particularly RO, but low fluxes through isotropic films posed a significant challenge to the commercialization of the technology. Seawater reverse osmosis (SWRO) was the application with the biggest industrial potential, and CA- based membranes were able to provide the minimum rejection (~99.3%) for a single-step desalination process to meet the World Health Organization (WHO) requirements for potable water. However, average fabricated membranes were 10 µm in thickness and hence could not meet the flux requirements to make the process commercially attractive.

Figure 1.1. Performance comparison for seawater desalination membranes at 55 bar, 25 °C. Data from Baker [1].

Theoretically, it was well understood that decreasing membrane thickness could boost flux through the membrane, as discussed by Graham in 1886 [5], but attempts to reduce thickness led to defects in the separating layers. This is another reason for slower growth of gas separation membranes compared to liquid separation membranes: pinholes or defects in the active layer have a much more pronounced effect for gas-based systems compared to liquid systems. Even a few defects in a dense gas separation membrane can result in complete loss of selectivity performance.

A breakthrough for industrial use of membranes was made by Loeb and Sourirajan in 1961: asymmetric membranes [9]. They discovered a simple method to reproducibly fabricate a membrane with a thin active layer (< 500 nm) on a thick porous support from a single material employing a single phase separation step. The thin top layer allowed sufficient

25 flux through the membrane while the support provided mechanical stability. These membranes exhibited 20-times higher water permeance than any known membrane at the time, and reasonable salt rejection, i.e., up to 99%. With significant research funding from the United States Department of Interior Office of Saline Water (OSW), this was a major step towards large-scale commercialization of membrane technology. However, rejection was still just below the level required to produce potable water from seawater in a single pass. Another concern faced by this fabrication method was the large amount of material needed for membrane fabrication: 60-70 grams per m2 of a membrane. The process could not be translated industrially to more expensive, customized materials.

Other notable attempts were made by companies like Union Carbide and DuPont to address the issues of fabricating thin layers [10]. Union Carbide laminated multiple thin polymer layers in an attempt to achieve defect-free properties, but the complicated approach allowed production of only small membrane areas which were used to produce oxygen-rich air for medical applications [11]. DuPont used a completely different approach by modifying membrane geometry. Their hollow fibers allowed significantly higher membrane surface areas to be packed in a fixed volume (estimated around 30,000 m2/m3). This allowed higher flux to be obtained for a given module size and allowed DuPont to enter the desalination market [12].

While membrane technology started to gain significant attention in research, one of the most prominent technological breakthroughs came as the idea of a “caulking” layer which was introduced by researchers at Monsanto. They developed extremely thin polysulfone layers (approximately 100 nm) on a hollow fiber support and "caulked" the defects in the selective layer using a silicone coating. The silicone layer further acted as a protective

26 layer, and the idea of a thin-film composite (TFC) membrane was introduced. The method solved a major problem with membranes used for gas separations as the inexpensive coating ensured stable performance and minimal loss in selective properties in case of defects during the manufacturing process. The caulking material could be chosen independently depending on the application, and flux and separation ability could be optimized by controlling the thickness of the caulking layer. They further quantitatively explained the effect of the layer on the final TFC using an electrical circuit analogy, “The

Resistance Model” [13]. This development allowed for the fabrication of one of the earliest membrane-based gas separation products by Monsanto: Prism® membrane for H2/CO ratio adjustment in syngas mixtures, in the 1980s.

Arguably, one of the most significant developments in the commercialization of membrane technology came from John Cadotte who developed the interfacial polymerization (IP) method which allowed fabrication of extremely thin active layers (<100 nm) with high reproducibility using a commercially scalable method. The membranes offered very high permeance due to the extremely thin nature of the separating layer, and > 99.5% rejection of sodium chloride was achieved using specific recipes employing low-cost monomers

[14–17]. Since then, the process has been a revelation for membranes for liquid separations, dominating the water treatment and desalination membrane manufacturing market, the two biggest membrane markets, due to its ability to fabricate reproducible and inexpensive high-performance membranes on an industrial scale. Dow/Filmtec FT-30 membranes were the early industry standard and are still the most commonly studied and well-cited IP membrane type [17–21]. Figure 1.1 shows the performance comparison of early membranes for RO desalination.

1.1.3. Advantages of Membrane Technology

Chemical separation processes account for up to 15% of global energy consumption [22] with operating and capital costs rising to 40-70% of total expenses in the chemical and pharmaceutical industries [23]. With the depletion of conventional energy sources, e.g., fossil fuels, and growing understanding of the impact of industrial emissions on the planet, there is high interest in potential separation processes which could lower the global energy burden. It is estimated that a decrease in separation cost in the US alone can potentially save 4 Billion USD in energy cost as well as save 100 million tons of CO2 emissions annually.

MED ED 3% 3% Others 8%

MSF RO 23% 63%

RO can be used as a model process to estimate potential energy savings offered by membrane technology in other separation applications. Distillation represents base distillation energy requirements, but due to the use of distillation in the industry for over a century, several engineering optimizations have been made to the process, such as heat recovery, to minimize its energy burden. The two most common optimized processes are multi-stage flash distillation (MSF) and multi-effect distillation (MED). Electrodialysis

(ED) and vapor compression (VC) also hold a small share. Distillation (solvent/solute, solvent/solvent systems) involves boiling the mixture to evaporate the liquid, and collecting the condensate. Current market shares of various technologies for desalination are highlighted in Figure 1.2 with membrane-based separation predicted to dominate the market increasingly. Rough estimates of energy cost related to different desalination processes are listed in Table 1.1 [25]. Reverse osmosis membrane processes operate just above the thermodynamic limit (minimum theoretical energy of separation) of the energy requirement. With time, similar to the case for distillation, engineering optimizations, such as pressure recovery systems, have pushed RO energy requirements closer to the

29 thermodynamic limit and are a primary reason why the cost of RO desalinated water has continued to remain between 0.5 - 1 USD/m3 over the past 40 years.

Table 1.1. Energy cost comparison for desalination. Data from Pinnau et al. [25].

Membranes can achieve such low energy consumption because the separation occurs without a phase change of the feed (e.g., liquid to vapor state in distillation). The fact that it is a low-temperature, athermal process is an added benefit in several industrial separations, especially those involving biologically active systems. This, in turn, leads to greener processes with smaller carbon footprint which are quickly becoming an integral part of the legislation regarding manufacturing industries. All this coupled with low installation costs, and simple, continuous operation have made membranes a serious competitor to conventional separation methods [25].

1.2. Current State-Of-The-Art Membranes

1.2.1. Membranes for Gas Separation

The gas separation membranes market value reached 500M USD in 2010 and has continued to grow at ~8% annually [26]. Increasing demands for energy efficient processes are predicted to further accelerate this growth. Baker classified and predicted further growth based on gas separation applications as shown in Figure 1.3.

As previously discussed, Monsanto was one of the first companies to produce commercial gas separation membranes in the 1980s, using polysulfone-based membranes. CA and derivatives were, however, the primary materials used for producing commercial membranes for gas- and liquid separations. Despite hundreds of materials developed for gas separation since then, it was estimated that by the 2000s approximately ten materials

31 constitute about 90% of the gas separation market [27]. The classes of these materials are summarized in Figure 1.4.

1.2.1.1. Hydrogen Separations

Hydrogen separation from a wide variety of streams represents a significant potential application for gas separation membranes both currently and in the future. Performance of conventional materials used in hydrogen purification is shown in Table 1.2. Monsanto’s

"caulked" PRISM membranes have been used to recover hydrogen in the Haber process since their commercialization. The ammonia reactor uses hydrogen (from the steam methane reformer — SMR) and nitrogen (from the cryogenic distillation of air) to produce

ammonia, i.e., 3H2 + N2 → 2NH3. The reactor conversion rate is ~30%, and hence, hydrogen must be recovered to maximize conversion rates. Figure 1.5 shows a simplified schematic of the membrane system employed. The output stream from the ammonia reactor, at 135 bar, is sent to a condenser to remove the ammonia product and then to the membrane system with primarily hydrogen, lesser amounts of nitrogen and trace amounts of methane and ammonia. The small quantities of condensable components in the mixture are low enough to maintain required performance without significant plasticization. A two- stage membrane system is used with pressure ratios of ~2 and ~5 for the first and second stage, respectively. The input stream to the first system contains ~45% H2 and due to the pressure ratio limit, maximum enrichment of 90% can be attained. However, very high membrane selectivity is required for achieving the theoretical enrichment (diminishing returns discussed later), and hence, H2 enrichment of ~85% is achieved with 80-90% recovery [1].

Table 1.2. H2/N2 separation performance for commercially used materials. Data from Baker [1].

1.2.1.2. Air Separations

A large proportion of the membrane-based gas separation market currently revolves around the production of highly enriched nitrogen from air [27]. The nitrogen is used for applications such as fuel blanketing in airplanes employing an on-board inert gas generation system (OBIGGS). The system provides a constant flow of pure, inert nitrogen over the fuel chamber to minimize chances of combustion of oxygen/fuel mixtures and is an essential safety feature of modern aircraft [28].

The process involves compressing atmospheric air at a pressure between 8 - 10 bar and passing it over an oxygen selective membrane. The majority of oxygen (ideally) permeates to the downstream side of the membrane leaving a nitrogen-enriched stream in the retentate. The primary capital and operating costs of the process, i.e., > 50%, are associated with the compression of air [27].

The current market requirements for the nitrogen product require a purity of > 99%. TPX

[poly(4-methyl-1-pentene)] was one of the early membrane materials used for air separation. The membranes exhibited O2/N2 selectivity of ~4 and could produce N2 at 99% purity but had a relatively low recovery of 25%. This meant larger feed streams needed to be processed to obtain a given amount of N2. As stated before, a large proportion of the nitrogen production cost is associated with feed compression. Hence, low recovery translated to higher product costs.

Table 1.3. Commercial materials for nitrogen separation. Data from Baker [1].

The second generation of membrane materials constituted of polyaramids, polysulfones and polycarbonates, as summarized in Table 1.3. Although these materials exhibited 10- fold or lower permeabilities, the higher selectivities, i.e., 6-8, doubled recovery rates from

25% to 50%. This meant that larger membrane areas were required to produce amounts equivalent to the previous generation of membranes, but compressor size and power could be halved, resulting in significant (50%) energy savings for compression processes. These materials are still prevalent in the industry and are used for producing nitrogen between

1,000 - 1,000,000 scfd with purities between 95 - 99%. Above these flowrates, cryogenic distillation and pressure-swing absorption dominate the market [27]. Figure 1.6

35 summarizes appropriate technologies for separation with respect to the product amount required.

The primary reason for the success of this system is that the source, atmospheric air, is generally low in condensable components. Hence, issues like plasticization are relatively simple to avoid. This advantage is coupled with a readily available, inexpensive source supply, i.e., atmospheric air, as well as relatively large concentration of nitrogen in the feed

(~79%). Furthermore, obtaining the nitrogen product from the retentate (instead of the low- pressure permeate) eliminates recompression costs.

The permeate-side product of the nitrogen enrichment process, enriched-oxygen, is quickly gaining industrial importance and is a potentially significant emerging application for membrane technology [27]. Though some applications require high purity oxygen (85-

98%), it is more realistic for membranes to tackle applications with purity requirements between 30-70% such as oxygen for biomedical use, catalyst regeneration, and enhanced combustion applications. This translates to O2/N2 selectivity of < 10 [1]. However, membranes for oxygen enrichment applications face fierce competition from technologies such as cryogenic distillation and vacuum swing adsorption. Such technologies can produce enriched oxygen at 40 - 60 USD/ton of equivalent pure oxygen (EPO2). EPO2 is defined as the additional amount of pure oxygen required to be added to the air to produce a required concentration of oxygen in the final mixture. State-of-the-art membrane materials for this application, summarized by the Robeson (2008) trade-off, cannot beat the 40 USD/ton EPO2 cost by competing processes [29,30].

1.2.1.3. Natural Gas Sweetening Separations

After the initial successful implementation of membrane technology for water desalination and hydrogen purification, natural gas sweetening, i.e., removal of CO2 from methane, was quickly identified as an application. By the late 1980s, some smaller membrane fabrication companies including Grace, Cynara, and Separex were producing CA-based membranes for this application. In 1994, one of the first hollow fiber membranes systems was installed for the application employing polyimides [1].

Natural gas is currently the fastest growing segment of the energy production sector, doubling in production and consumption globally between 1980 to 2000 [31]. This is principally because growing environmental concerns caused by fossil fuel emissions have led to increased emission regulations for power plants, and natural gas offers significantly lower emissions (~50%) than conventional fuel sources like coal. This was well predicted

37 by Baker [1] among others, and hence, natural gas sweetening both was (over the last two decades) and is one of the biggest potential applications for membrane technology. The

Castor Underground Natural Gas Storage (UGS) project in Spain, one of the world’s largest gas separation facilities, is employing UOP’s CA-based Separex membranes. The plant is expected to process 1,000 MMscfd of natural gas [32].

Amine absorption technology currently dominates this market due to its ability to process very large feed streams and produce very high methane recoveries. High recovery of methane is of the utmost concern in the natural gas sweetening industry due to the large volumes of natural gas present and the large methane content in crude gas mixtures. Even small decreases in recovery can translate to millions in monetary losses for a company.

Membranes can offer significant advantages over such systems, for example:

a) No toxic and expensive amine use and regeneration

b) High energy efficiency and low environmental impact to meet increasing

regulations

c) Small footprint, of utmost importance in off-shore applications

d) Continuous operation allowed by membranes vs. batch separation for amine

absorption technology

Despite these advantages, membranes currently comprise only ~5% of this market. This is due to the complex feed stream mixtures involved in processing natural gas. Along with

CO2 and CH4, a natural gas feed can contain water, methanol, oil, glycol, dust particles, drilling fluids, and C2+ components, as well as additional acid gases such as H2S. Such components do not only significantly damage membrane module materials but can

seriously affect membrane performance. Primary source components, such as CO2, can cause significant plasticization of the membrane, resulting in loss of separation performance. This has been a major bottleneck to the implementation of new materials in such separations [1,33].

1.2.2. Membranes for Liquid Separations

Membrane-based liquid separations have been researched since the 1940s. Desalination, as mentioned, was understandably a primary motivation for research in the technology, where reverse osmosis (RO) served as the model application. In RO, a salt/water feed solution is pumped against an osmotic pressure gradient to obtain potable water (Figure 1.7). Over the past 50 years, RO technology has matured tremendously with currently more than 15,000 desalination plants operating worldwide, producing approximately 15 billion cubic meters of water per year. Membrane-based liquid separation processes currently hold approximately 50% of the global and 85% of the US desalination market. The US industry alone is valued at 4.2 billion USD with 8% annual growth predicted up to 2023 [34].

Figure 1.7. Simplified schematic for reverse osmosis desalination [35].

Research on membranes for liquid separations has recently shifted towards removal of micro-contaminants from a large variety of organic and inorganic solvents. This field is commonly referred to as nanofiltration (NF), organic solvent nanofiltration (OSN) or organic solvent reverse osmosis (OSRO). Research on membranes for organic solvent nanofiltration has been minimal, with the first publications around 1990, and though there is growing interest (primarily over the last decade), publications between 2000-2010 were still below 100 per year [23]. Despite this, hundreds of industrial applications have been identified where organic nanofiltration can contribute significantly as both a standalone technology and in combination with current industrial separation technologies, e.g., extraction, adsorption, distillation, and evaporation.

The most commonly used commercially available liquid-separation membranes are listed in Table 1.4.

Table 1.4. Commercially available liquid separation membranes. ISA = Integrally skinned asymmetric. HF = Hollow fibers. R and MWCO refer to rejection and molecular weight cut-off respectively.

Manufactur R MWCO Membrane Material Type Solvent Marker Ref er (%) (Da)

Cellulose ISA HL10255SI Toyobo Water NaCl 99 <200 [36] triacetate HF IP SW30/FT30 PA/PSf Dow Filmtec Water NaCl 99 <200 [17] TFC IP NF270 PA Dow Filmtec Water NaCl 55 200-400 [37] TFC IP NF200 PA Dow Filmtec Water MgSO4 3 200-400 [37] TFC IP NF90 PA Dow Filmtec Water NaCl 90 200-400 [37] TFC IP TS80 PA TriSep Water NaCl 99 <200 [37] TFC IP TS40 Polypiperazine TriSep Water NaCl 99 200 [37] TFC IP XN45 PA TriSep Water NaCl 95 500 [37] TFC IP UTC20 Polypiperazine Toray Water NaCl 60 <200 [37] TFC IP TR60 PA Toray Water NaCl 55 400 [37] TFC GE CK Cellulose acetate ISA Water MgSO4 94 2000 [37] Osmonics GE IP DK PA Water MgSO4 98 200 [37] Osmonics TFC GE IP DL PA Water MgSO4 96 150-300 [37] Osmonics TFC GE IP HL PA Water MgSO4 98 150-300 [37] Osmonics TFC IP NFX PA Synder Water MgSO4 99 150-300 [37] TFC

IP NFW PA Synder Water NaCl 97 300-500 [37] TFC IP NFG PA Synder Water MgSO4 50 600-800 [37] TFC IP TFC SR100 PA Koch Water NaCl 99 200 [38] TFC IP SR3D PA Koch Water NaCl 99 200 [38] TFC IP SPIRAPRO PA Koch Water NaCl 99 200 [38] TFC IP ESNA1 PA Nitto-Denko Water NaCl 89 100-300 [37] TFC Sulfonated NTR7450 Nitto-Denko ISA Water NaCl 50 600-800 [37] Polyethersulfone DuraMem Polyimide DuraMem ISA Acetone Styrene 99 <236 [39] 150 Azo- Ethanol 41 182 [40] benzene Starmem 122 Polyimide StarMem ISA Toluene Styrene 87 236 [39] Rose i-propanol 99.9 <1017 [41] Bengal Starmem Polyimide StarMem ISA n-heptane Styrene 90 400 [42] 240 Toluene Styrene 90 380 [42] Erythros SolSep 169 Silicone-based SolSep TFC Acetone 91 880 [43] ine B Ethyl Victoria 65 >506 [43] acetate Blue

Because industrial interest in organic nanofiltration is relatively recent, limited data are available on commercial membranes. Some further noteworthy mentions include Koch’s

SelRO [44] membranes, namely MPF-60, MPF-50, and MPF-44. The former two hydrophobic membranes have been discontinued whereas the hydrophilic MPF-44 is still available. The manufacturers claim stability over a wide range of solvents including THF, dioxane, ketones (e.g., methyl ethyl ketone), chlorinated solvents (e.g., chloroform and

42 dichloromethane), aromatics (e.g., toluene), lower hydrocarbons, ethyl acetate and alcohols among others. MPF-50 was thoroughly studied [44–51] for organic nanofiltration applications, but varying conclusions were made on performance, generally different from manufacturer’s quoted values.

SolSep is a new player to the membrane industry with manufacturing beginning only a decade ago [43]. Their membranes are gaining attention due to their high chemical and thermal stabilities coupled with high permeances and narrow rejections. Currently, the company manufactures membranes with molecular weight cut-offs (MWCOs) between

300-750 g mol-1 which are stable in esters, alcohols, ketones, aromatics, and chlorinated solvents. They are shown to be stable at operating conditions up to 40 bar and 150 °C [52].

DuraMem, PuraMem, PoroGen, GMT, PolyAn and AMS membrane technologies are also significant companies producing membranes for nanofiltration, but information on composition and performance is limited.

1.3. The Future of Membrane Science

1.3.1. Membranes for Gas Separation

1.3.1.1. Hydrogen Separations

Although the use of hydrogen separation membranes has been primarily applied to ammonia production to date, there is growing interest in separation of hydrogen from a wide variety of resources. This is driven by both growing need for hydrogen in the petrochemical sector, to produce essential industrial solvents such as methanol, as well as for refinery operations such as hydrocracking, hydro-treating, and desulfurization to lower refinery emissions [53–55].

Further applications include power generation, specifically using Integrated Gasification

Combined Cycle (IGCC) systems [56,57]. Merkel et al. [56] discussed the potential use of

H2/CO2 selective membranes for use with IGCC plants and made a detailed comparison with pulverized coal (PC) plants. Though currently, IGCC plants produce electricity at 1.25 times the cost of subcritical PC plants, IGCC technology offer advantages such as higher power generation efficiency, i.e., 45% compared to 35% average in PC plants, as well as cleaner operation, i.e., lower emissions along with improved potential for carbon capture post- and pre-combustion [58]. Furthermore, as the tax on CO2 emissions increases, IGCC plants become far more competitive than PC plants due to the relatively easier removal of

CO2. Merkel et al. [56] argue that membranes which offer H2/CO2 performances could help

IGCC achieve the required costs of electricity while sequestering 90% of the CO2 produced

[59]. USDOE [60] (United States Department of Energy) specified performance targets

with H2 permeance of 200 GPU and selectivity of approximately 12 over CO2 at operating temperatures around 150 °C for membranes required for IGCC operations.

1.3.1.2. Air Separations

As previously discussed, the majority of cost for air separations using membranes is associated with the compressor. High recovery results in significant cost decreases as a smaller feed-volume of gas is required for a given volume of product. Recovery is highly dependent on membrane selectivity, and Figure 1.8 below shows the relationship between compressor power requirements and membrane selectivity. The graph shows that beyond

O2/N2 selectivity of 8, significant increases in selectivity provide little to no reduction in compressor costs. Because the current generation of air-separating membranes already provide these selectivities, little improvement is required. However, increasing permeabilities can help decrease membrane surface area required for the separation and result in membrane capital cost decreases. Hence, the next generation of membranes for this application need materials that combine higher permeabilities with the selectivity of current materials [29,30].

1.3.1.3. Natural Gas Separations

As discussed previously, plasticization creates a major problem for membranes in natural gas applications. Plasticization results in decreased gas-pair selectivity, resulting in increased loss of the expensive methane product. Membranes offering higher permselectivity performance, by providing higher selectivity in the presence of condensable materials than current (CA-based) membranes, can have significant direct monetary advantages coupled with other advantages highlighted previously [61].

Swaidan et al. [62] built upon a comparison previously made by Baker to highlight the impact of replacing current membranes for natural gas sweetening with a hypothetical high-performance membrane. The two membranes were compared using a two-stage membrane configuration, proposed for maximizing recovery, as shown in Figure 1.9. The

46 comparison showed how replacing current CA-based membranes with high-performance membranes can significantly reduce both capital costs (via reduction of membrane area) and operating costs (by reducing compression costs and methane loss). The hypothetical membrane must be able to produce the given performance in industrial conditions with realistic mixed-gas streams.

Figure 1.9. Typical two-stage natural gas sweetening process. (a). Cellulose acetate (CA) vs. (b). Hypothetical high-performance membrane. Adopted from Swaidan et al. [62].

1.3.2. Future Of Liquid Separations

Desalination membranes have reached industrial maturity, and significantly improving their properties can yield negligible cost and energy savings for the RO desalination process. Such membranes could potentially be enhanced by increasing fouling resistance and improving free chlorine stability to increase useful membrane lifetime, but due to the low cost of membranes involved, there is little industrial emphasis [63–67]. One potential improvement gaining interest is the removal of small molecular weight trace components from water streams, e.g., urea, arsenic, boron, etc [22]. Increased separation capabilities for removal of boron, for example, can potentially lower desalination costs by eliminating the need for a second-pass during the reverse osmosis desalination process in regions such as the Middle East. Such separations, though, have proven challenging for membrane systems [34,68].

The scope of the technology can be divided into three main categories: i) concentration; ii) solvent exchange, and; iii) purification. Concentration refers to concentrating the solute in either the upstream or downstream of the process. This can be used as a pretreatment step to purification for both removing a high-value solute and solvent regeneration. Membrane materials for concentration processes are relatively more straightforward to design as low rejections, i.e., 80 - 95% (for use with downstream purification processes), very low rejections, i.e., < 10% (with low solvent flux for upstream concentration) and negative rejections can be employed. Solvent exchange refers to exchanging a solute carrying solvent A with a different solvent, either to regenerate the solvent or to simplify subsequent separations.

Purification involves high rejection separations of a solute from binary or tertiary mixtures of solutes in a solvent. Using an example of a binary solute system, i.e., solute A and B in solvent C, the membrane must possess a sharp cut-off between A and B. This understandably becomes difficult with solutes of similar size and/or similar interactions with the membrane. Demand for very high rejections, generally coupled with high yields, makes membranes for purification challenging to fabricate. This is because of both limited knowledge of separation mechanisms in liquid-based separation systems, as it is difficult to relate pore size or free volume to the rejection of solutes, as well as lack of understanding of polymer swelling behavior which causes variance in pore size or free volume with varying solvents.

The global chemical industry is currently valued at well over a trillion dollars [69], with purification accounting for a large proportion of all manufacturing related costs [70].

Because of the large number of applications associated with organic nanofiltration, this section will only very briefly cover current and future potential uses of the technology. The

European Chemical Industry Council (CEFIC) [71] classifies chemical industries in four sectors: (i) consumer chemicals, chemical products supplied directly to consumers; (ii) specialty chemicals, supplied as industrial raw materials; (iii) pharmaceuticals, and; (iv) base chemicals which cover mainly petrochemicals and their derivatives. Each industrial sector will be discussed according to the broad organic nanofiltration applications, i.e., concentration, solvent exchange, and purification.

1.3.2.1. Concentration

Research in medicine over the past decades has identified several compounds present in natural plants and herbs which could offer health benefits, from strengthening the immune system to improving hair growth. These compounds can range from, but are not limited to, carotenoids, catechins, terepenoids, flavonoids, antioxidants, and vitamins. Athermal, low- cost separation of these sensitive compounds using organic nanofiltration is a potentially huge market. Sereewatthanawut [72] used organic nanofiltration to enrich a stream of γ- oryzanol, known for its antioxidant properties, from rice bran oil. Rosmarinic acid was recovered by Peshev et al. [73] using DuraMem200. They were able to obtain rejections over 99% along with flux and recovery higher than current industrial techniques. Both authors reported no change in their respective compounds’ properties. Recovery of solvents from cooking oil is of major importance as output retentate streams tend to contain 75% toxic solvents compared to the 25% product. These are generally removed by energy- intensive distillation. Firman [74] used organic nanofiltration for simultaneous deacidification and n-hexane removal from crude soybean oil using TFCs with cellulose acetate (CA) and polydimethylsiloxane (PDMS) layers. Reasonable fluxes of up to 1

LMH/bar (liters m-2 h-1 bar-1) were obtained with 80% oil rejection.

Ionic liquids are currently a hot topic of research showing unique properties such as immiscibility with both organic solvents and water as well as the ability to be tailored specifically for applications [75]. Han [76] used a two-step nanofiltration process using

Starmem 122 to separate the ionic liquid Cyphos 101 with 100% yield and 74% purity.

Furthermore, Van Doorslaer [77], Abels [78] and Hazarika [79] used commercial nanofiltration membranes to separate the ionic liquids monomethyl azelate, 1,3-

50 dimethylimidazolium, and 1-butyl-3-methylpyridinium. This could serve as a very lucrative market for organic nanofiltration.

In pharmaceutical applications, nanofiltration has enormous potential in antibiotic concentration and recovery. Cao [80] used nanofiltration to recover 6-aminopenicillanic acid which is an active pharmaceutical ingredient (API) for antibiotic production. Evonik

[81] applied nanofiltration to concentrate an API waste stream from 1 to 10% resulting in a reported profit increase of > 1 million USD per year. High product recovery (> 95%) for both led to a payback time of less than a year coupled with long-term energy savings. Large amounts of solvents used in the pharmaceutical industry can also be effectively recovered using nanofiltration. Potential for ethyl acetate, isopropanol, methanol and ethanol recovery have been shown [82], which could be used in the production of APIs such as

Alprazolam, Donepezil, Atenolol and Riluzole, resulting in both monetary saving and improved environmental sustainability. Rundquist [83] showed the ability of nanofiltration to separate isopropyl acetate from a multiple-impurity-containing (more than 40 different compounds) mother liquor to purities high enough to use directly in a subsequent crystallization process.

Lube oil dewaxing is an extremely solvent intensive process generally using toluene and methyl ethyl ketone (MEK). Data from a single ExxonMobil refinery [84] shows that it generates more than 300,000 m3/year of solvent requiring recovery. This is generally done by flash distillation, using relatively large amounts of energy. PEI-based TFCs containing

UZM-5 nanoparticles have been shown [85] to significantly lower the energy burden of the process while maintaining yield and purity requirements.

1.3.2.2. Solvent Exchange

The mechanism of solvent exchange was explained before. This strategy can be efficient in removing toxic and high boiling point solvents from those offering better environmental credentials and low energy requirements for downstream removal using distillation (or other) processes.

Sheth [86] used Koch MPF-50 and MPF-60 to exchange an API in ethyl acetate with methanol. An interesting study [87] was performed employing a comprehensive set of commercially available nanofiltration membranes for the solvent exchange of industrial waste streams containing cyclohexane and n-hexane to renewable, inexpensive and less toxic solvents like ethanol. The study showed potential savings using already available commercial membranes with a single-step nanofiltration process. Rundquist [88] showed how nanofiltration could be integrated with counter-current chromatography to remove solvents used upstream to trace levels.

1.3.2.3. Purification

Nanofiltration can also be used to produce high purity sugars such as glucose, fructose and raffinose among others. Vanneste [89] applied simple stage cascade systems employing commercially available nanofiltration membranes to set up high yield/purity systems for their separation. Further potential in industry was reported by Nwuha [90] for catechins,

Tsibranska [91] for flavonoids, and for fatty acids by Gupta et al. [92].

Many chemical reactions, especially involving large, complex molecules, are sensitive to how the atoms are arranged in a molecule, i.e., stereochemistry. Different forms of stereoisomers are especially difficult to separate due to their similar physical properties.

Recently, nanofiltration was employed in a single-unit stereoisomer synthesis and separation system [93,94] to study the mechanism for chiral selectivity in α- phenylethylamine salt. With such nanofiltration-assisted chemical synthesis, a particular species permeates much faster through the membrane compared to its counterpart. Another example was shown by Ghazali [95], who used a precipitation agent to insolubilize the (S)- enantiomer of phenethyl alcohol while permeating the (R)-enantiomer. Nanofiltration can also serve as a useful tool for continuous removal of homogeneous catalysts from product streams [96], simultaneously driving the reaction forward (Le-Chatelier’s principle) while preventing catalyst loss. Starmem 122 membranes were used [97] to purify large dendrimeric macromolecules with extremely high rejections. Such macromolecules are currently used extensively for research purposes and are priced at > US$1000 per gram, showing good future potential for nanofiltration.

For API separation, Geens [70] covers a wide range of APIs currently being separated using nanofiltration as well as the potential for hundreds of others. Because of the complex nature of APIs as well as heat and pH sensitivity, separations using current technologies such as crystallization and chromatography are complex and expensive. Szekely [98] performed a detailed comparison of currently available purification technologies in the pharmaceutical industry for impurity removal and compared them to nanofiltration. He concluded that an average 100-fold reduction in energy costs could potentially be obtained by employing nanofiltration, but also highlighted the issue of sensitive API species reacting with the membrane materials.

1.3.3. Promising Materials

The total volume of a polymeric material can be divided into “occupied volume”, i.e., volume occupied by atoms in the material, and "free volume", i.e., void space between the atoms, also referred as pores [99]. On the basis of this pore size, polymers can be classified in three categories: (i) macroporous with pore size > 500 Å; (ii) mesoporous with pore size between 20–500 Å, and; (iii) microporous with pore size < 20 Å.

Inorganic microporous materials, primarily zeolites, and activated carbons have shown great potential for use in membrane-based separations [100]. However, due to their poor mechanical properties (brittleness), and the difficulty of producing continuous, defect-free membranes, their application as membrane materials has been limited. Therefore, attempts have been made to combine the low-cost and easy processability of common polymers with the high permselectivity of microporous Figure 1.10. Structure of PIM-1 materials. This was achieved by producing polymers from monomers containing very contorted and rigid groups. This created free microporous regions in the polymers as the rigid polymer chains cannot occupy the interchain free space, unlike more flexible ones.

Such polymers of intrinsic microporosity (PIMs) are defined as polymers containing “a continuous network of interconnected intermolecular voids which form as a direct consequence of the shape and rigidity of the component macromolecules” [101].

PIM materials exhibit very high diffusion coefficients and have already shown great potential in gas separations, based on isotropic film test results [102–111]. Thin films are

54 generally a pre-requisite for testing liquid-based separations and hence, minimal data are available for such separations. Tsarkov et al. carried out a preliminary examination using thick PIM-1 (Figure 1.10) and showed high ethanol permeance [112]. Fritsch et al. [42] and Anokhina et al. [113] further tested PIM-I for nanofiltration and reported high solute rejections coupled with high fluxes. Gorgojo et al. [101] were able to fabricate very thin

PIM-1 films (comparable to those produced via IP) and reported solvent permeances at least an order of magnitude higher than currently available commercial membranes. All membranes in the above examples were prepared using solution casting, spin-coating or dip-coating. An alternative fabrication process to manufacture PIM-based thin-film composite membranes using interfacial polymerization will be investigated in this study and, if successful, could be a major contribution to the membrane industry.

1.4. Dissertation Goals

Interfacially polymerized thin-film composite polyamide membranes have revolutionized the global desalination industry [114]. Although, research has suggested the presence of micro-defects in polyamide TFC RO membranes such as the highly rated m-phenylene diamine/trimesoyl chloride chemistry. The defects potentially reduce separation performance in liquid separations and render the membranes inadequate for gas separation applications. Removal of defects from such membranes can likely improve their liquid separation performance as well as allow their use in the gas separation industry. This work is aimed at the removal of defects from interfacially polymerized thin-film composite polyamide membranes and utilizing the lessons learned to produce defect-free state-of-the- art membranes incorporating novel building blocks. The research further aims to bridge the gap between gas- and liquid separations with intent to provide comprehensive gas and liquid transport characterization for interfacially polymerized TFCs under identical fabrication conditions.

The dissertation is divided into 8 chapters. The first chapter deals with introductory knowledge to membrane science: its emergence, initial commercialization as well as current state-of-the-art membranes along with future perspectives utilizing polymers of intrinsic microporosity for gas- and liquid separations.

Chapter 2 provides the necessary theoretical background for transport through membranes for gases and liquids including fundamental thermodynamics as well as essential process considerations when designing membrane systems. The chapter also introduces interfacial polymerization in detail.

Chapter 3 deals with specific materials and methods employed in experimental research in the following chapters.

Chapter 4 discusses the variation of fabrication parameters for m-phenylene diamine/ trimesoyl chloride-based interfacially polymerized thin-film composite polyamide membranes to enable in-situ self-healing of defects. The removal of defects, by simple tweaking of the reaction time, from the membrane allowed gas-separation performance characterized using pure-gas permeation measurements. It was observed that further modification of the commercially utilized recipe by increasing organic solution temperature significantly improved crosslinking of the polyamide layer enabling exceptional separation properties for hydrogen from components including carbon dioxide, nitrogen, and methane. High temperature mixed-gas permeation was performed using

H2/CO2 mixtures to unveil exceptional performance.

Chapter 5 analyses the properties of defect-free interfacially polymerized thin-film composite polyamide membranes in liquid separation systems. The membranes are tested under brackish water reverse osmosis conditions to understand the variation of sodium chloride removal properties with fabrication parameters. The membranes are characterized using scanning electron microscopy, atomic force microscopy, ellipsometry, surface charge measurements, X-ray photoelectron spectroscopy, and contact angle measurements to establish structure-property relationships. The highest performing membranes are tested for removal of boron from water sources and compared to available commercial membranes for comparison.

Chapter 6 discusses potential effects of replacing currently available monomers for interfacial polymerization with highly kinked and contorted PIM-like building blocks to fabricate crosslinked polyamide TFCs. In this study, trimesoyl chloride (used in conventional interfacial polymerization process) is substituted with a rigid triptycene-

1,3,6,8-tetraacetyl chloride moiety. The high separation performance of the resulting TFCs is characterized using a wide range of aqueous-, organic- and gas-separation systems to show the potential use of the novel high-performance materials in diverse applications.

Measurement and characterization techniques including high/low-pressure gas sorption, organic vapor sorption, and molecular dynamics studies are carried out to establish structure-property relationships.

Chapter 7 analyzes the potential to use the modified interfacial polymerization fabrication method for facile fabrication of defect-free TFCs using commercial monomers. Trimesoyl chloride is reacted with m-phenylene diamine, p-phenylene diamine, and piperazine. The effect of varying diamine chemistry is studied using gas permeation and salt rejection measurements. Surface and bulk characterizations are made on the prepared samples to correlate structure-property relationships.

Chapter 8 discusses concluding remarks and future recommendations.

1.5. References

[1] R.W. Baker, Membrane Technology and Applications, 3rd ed, Wiley, Chichester, UK., 2012.

[2] T. Graham, Notice of the singular inflation of a bladder, J. Membr. Sci. 100 (1995) 9.

[3] J.K. Mitchell, On the penetrativeness of fluids, J. Membr. Sci. 100 (1995) 11–16.

[4] T. Graham, XI. On the law of the diffusion of gases, Trans. R. Soc. Edinburgh 12 (1834) 222–258.

[5] T. Graham, XVIII. On the absorption and dialytic separation of gases by colloid septa, Philos. Trans. R. Soc. London 156 (2006) 399–439.

[6] H. Bechhold, Kolloidstudien mit der Filtrationsmethode, Zeitschrift für Elektrotechnik und Elektrochemie 13 (1907) 527–533.

[7] R.M. Barrer, Diffusion in elastomers, Kolloid-Zeitschrift 120 (1951) 177–190.

[8] R. M. Barrer, Diffusion in and through Solids, J. Phys. Chem. 46 (1942) 533–534.

[9] S. Loeb, S. Sourirajan, The preparation of high-flow semi-permeable membranes for separation of water from saline solutions, US 3,133,132, 1964.

[10] R.M. Salemme, Preparation of microporous polycarbonate resin membranes, 1976.

[11] W.J. Koros, G.K. Fleming, Membrane-based gas separation, J. Membr. Sci. 83 (1993) 1–80.

[12] C.R. Anlonson, R.J. Gardner, C.F. King, D.Y. Ko, Analysis of gas separation by permeation in hollow fibers, Ind. Eng. Chem. Process Des. Dev. 16 (1977) 463–469.

[13] J.M.S. Henis, M.K. Tripodi, Composite hollow fiber membranes for gas separation: the resistance model approach, J. Membr. Sci. 8 (1981) 233–246.

[14] J.E. Cadotte, Interfacially synthesized reverse osmosis membrane, US 4,277,344 A, 1981.

[15] J.E. Cadotte, R.J. Petersen, R.E. Larson, E.E. Erickson, A new thin-film composite seawater reverse osmosis membrane, Desalination 32 (1980) 25–31.

[16] J.E. Cadotte, R.S. King, R.J. Majerle, R.J. Petersen, Interfacial synthesis in the preparation of reverse osmosis membranes, J. Macromol. Sci. Part A - Chem. 15 (1981) 727–755.

[17] R.E. Larson, J.E. Cadotte, R.J. Petersen, The FT-30 seawater reverse osmosis

membrane--element test results, Desalination 38 (1981) 473–483.

[18] P. Lipp, R. Gimbel, F.H. Frimmel, Parameters influencing the rejection properties of FT30 membranes, J. Membr. Sci. 95 (1994) 185–197.

[19] M.J. Kotelyanskii, N.J. Wagner, M.E. Paulaitis, Atomistic simulation of water and salt transport in the reverse osmosis membrane FT-30, J. Membr. Sci. 139 (1998) 1–16.

[20] S.Y. Kwak, S.G. Jung, S.H. Kim, Structure-motion-performance relationship of flux-enhanced reverse osmosis (RO) membranes composed of aromatic polyamide thin films, Environ. Sci. Technol. 35 (2001) 4334–4340.

[21] J.E. Cadotte, R.J. Petersen, Synthetic membranes:, American Chemical Society, Washington, D.C., 1981.

[22] D.S. Sholl, R.P. Lively, Seven chemical separations to change the world, Nature 532 (2016) 6–8.

[23] P. Marchetti, M.F. Jimenez Solomon, G. Szekely, A.G. Livingston, Molecular separation with organic solvent nanofiltration: A critical review, Chem. Rev. 114 (2014) 10735–10806.

[24] K.P. Lee, T.C. Arnot, D. Mattia, A review of reverse osmosis membrane materials for desalination-Development to date and future potential, J. Membr. Sci. 370 (2011) 1–22.

[25] I. Pinnau, F.A. Pacheco, E. Litwiller, D. Jintang, Elucidation of the microstructure of interfacially polymerized reverse osmosis membranes, in: 8th Sino-US Jt. Conf. Chem. Eng., Shanghai, China, 2015.

[26] P. Bernardo, E. Drioli, G. Golemme, Membrane gas separation: A review/state of the art, Ind. Eng. Chem. Res. 48 (2009) 4638–4663.

[27] R.W. Baker, Future directions of membrane gas separation technology, Ind. Eng. Chem. Res. 41 (2002) 1393–1411.

[28] D.F. Sanders, Z.P. Smith, R. Guo, L.M. Robeson, J.E. McGrath, D.R. Paul, B.D. Freeman, Energy-efficient polymeric gas separation membranes for a sustainable future: A review, Polymer 54 (2013) 4729–4761.

[29] B.D. Bhide, S.A. Stern, A new evaluation of membrane processes for the oxygen- enrichment of air. II. Effects of economic parameters and membrane properties, J. Membr. Sci. 62 (1991) 37–58.

[30] B.D. Bhide, S.A. Stern, A new evaluation of membrane processes for the oxygen- enrichment of air. I. Identification of optimum operating conditions and process configuration, J. Membr. Sci. 62 (1991) 13–35.

[31] Minqi Li, World Energy 2017-2050: Annual Report, Salt Lake City, 2017.

[32] D. Bessarabov, UGS project uses UOP technology to remove impurities from gas, Membr. Technol. 2010 (2010) 2.

[33] R.W. Baker, K. Lokhandwala, Natural gas processing with membranes: An overview, Ind. Eng. Chem. Res. 47 (2008) 2109–2121.

[34] S.S. Shenvi, A.M. Isloor, A.F. Ismail, A review on RO membrane technology: Developments and challenges, Desalination 368 (2015) 10–26.

[35] P. Deka, Large-Area Graphene Oxide Membranes for Water Purification, RMIT University, 2017.

[36] J. Duan, Liquid and gas permeation studies on the structure and properties of polyamide thin-film composite membranes, Ph.D. thesis, King Abdullah University of Science and Technology, Saudi Arabia, 2014.

[37] A.W. Mohammad, Y.H. Teow, W.L. Ang, Y.T. Chung, D.L. Oatley-Radcliffe, N. Hilal, Nanofiltration membranes review: Recent advances and future prospects, Desalination 356 (2015) 226–254.

[38] Koch, Koch Membrane Products, (2016).

[39] M.F. Jimenez Solomon, Y. Bhole, A.G. Livingston, High flux hydrophobic membranes for organic solvent nanofiltration (OSN)-Interfacial polymerization, surface modification and solvent activation, J. Membr. Sci. 434 (2013) 193–203.

[40] S. Karan, S. Samitsu, X. Peng, K. Kurashima, I. Ichinose, Ultrafast viscous permeation of organic solvents through diamond-like carbon nanosheets, Science 335 (2012) 444–447.

[41] X. Li, P. Vandezande, I.F.J. Vankelecom, Polypyrrole modified solvent resistant nanofiltration membranes, J. Membr. Sci. 320 (2008) 143–150.

[42] D. Fritsch, P. Merten, K. Heinrich, M. Lazar, M. Priske, High performance organic solvent nanofiltration membranes: Development and thorough testing of thin film composite membranes made of polymers of intrinsic microporosity (PIMs), J. Membr. Sci. 401–402 (2012) 222–231.

[43] SolSep, SolSep BV Products, (2016).

[44] P. Silva, S. Han, A.G. Livingston, Solvent transport in organic solvent nanofiltration membranes, J. Membr. Sci. 262 (2005) 49–59.

[45] L.E.M. Gevers, G. Meyen, K. De Smet, P. Van De Velde, F. Du Prez, I.F.J. Vankelecom, P.A. Jacobs, Physico-chemical interpretation of the SRNF transport mechanism for solutes through dense silicone membranes, J. Membr. Sci. 274

(2006) 173–182.

[46] E. Gibbins, M. D’Antonio, D. Nair, L.S. White, L.M. Freitas dos Santos, I.F.J. Vankelecom, A.G. Livingston, Observations on solvent flux and solute rejection across solvent resistant nanofiltration membranes, Desalination 147 (2002) 307– 313.

[47] J. Geens, B. Van Der Bruggen, C. Vandecasteele, Transport model for solvent permeation through nanofiltration membranes, Sep. Purif. Technol. 48 (2006) 255– 263.

[48] D. Bhanushali, D. Bhattacharyya, Advances in Solvent-Resistant Nanofiltration Membranes, Ann. N. Y. Acad. Sci. 984 (2003) 159–177.

[49] Y. Zhao, Q. Yuan, A comparison of nanofiltration with aqueous and organic solvents, J. Membr. Sci. 279 (2006) 453–458.

[50] J. Geens, K. Peeters, B. Van Der Bruggen, C. Vandecasteele, Polymeric nanofiltration of binary water-alcohol mixtures: Influence of feed composition and membrane properties on permeability and rejection, J. Membr. Sci. 255 (2005) 255– 264.

[51] D. Bhanushali, S. Kloos, D. Bhattacharyya, Solute transport in solvent-resistant nanofiltration membranes for non-aqueous systems: Experimental results and the role of solute-solvent coupling, J. Membr. Sci. 208 (2002) 343–359.

[52] F.P. Cuperus, Recovery of organic solvents and valuable components by membrane separation, Chemie-Ingenieur-Technik 77 (2005) 1000–1001.

[53] P. Li, Z. Wang, Z. Qiao, Y. Liu, X. Cao, W. Li, J. Wang, S. Wang, Recent developments in membranes for efficient hydrogen purification, J. Membr. Sci. 495 (2015) 130–168.

[54] C.M. Kalamaras, A.M. Efstathiou, Hydrogen production technologies: Current state and future developments, Conf. Pap. Energy 2013 (2013) 1–9.

[55] United States Department of Energy, Basic research needs for the hydrogen economy. Workshop on hydrogen production, storage and use., Lemont, 2003.

[56] T.C. Merkel, M. Zhou, R.W. Baker, Carbon dioxide capture with membranes at an IGCC power plant, J. Membr. Sci. 389 (2012) 441–450.

[57] M. Kanniche, R. Gros-Bonnivard, P. Jaud, J. Valle-Marcos, J.-M. Amann, C. Bouallou, Pre-combustion, post-combustion and oxy-combustion in thermal power plant for CO2 capture, Appl. Therm. Eng. 30 (2010) 53–62.

[58] DOE/NETL, Advanced Carbon Dioxide Capture R&D Program: Technology Update, 2010.

[59] United States Department of Energy, Cost and performance baseline for fossil energy plants: Volume 1: Bituminous coal and natural gas to electricity, Report Number DOE/NETL-2007/1281, 2007.

[60] U.S. Department Of Energy (USDOE), Advanced carbon dioxide capture R&D program: Pre-combustion membranes, 2013.

[61] M. Wessling, S. Schoeman, T. van der Boomgaard, C.A. Smolders, Plasticization of gas separation membranes, Gas Sep. Purif. 5 (1991) 222–228.

[62] R. Swaidan, Intrinsically microporous polymer membranes for high performance gas separation, Ph.D. thesis, King Abdullah University of Science and Technology (KAUST), Saudi Arabia, 2014.

[63] J.S. Louie, Fouling-resistant coatings for reverse osmosis membranes: gas and liquid permeation studies on morphology and mass transport effects, Ph.D.thesis, Stanford University, USA, 2008.

[64] Q. Cheng, Y. Zheng, S. Yu, H. Zhu, X. Peng, J. Liu, J. Liu, M. Liu, C. Gao, Surface modification of a commercial thin-film composite polyamide reverse osmosis membrane through graft polymerization of N-isopropylacrylamide followed by acrylic acid, J. Membr. Sci. 447 (2013) 236–245.

[65] J. Sun, L.P. Zhu, Z.H. Wang, F. Hu, P. Bin Zhang, B.K. Zhu, Improved chlorine resistance of polyamide thin-film composite membranes with a terpolymer coating, Sep. Purif. Technol. 157 (2016) 112–119.

[66] S. Hong, I.C. Kim, T. Tak, Y.N. Kwon, Interfacially synthesized chlorine-resistant polyimide thin film composite (TFC) reverse osmosis (RO) membranes, Desalination 309 (2013) 18–26.

[67] J. Glater, S. kwan Hong, M. Elimelech, The search for a chlorine-resistant reverse osmosis membrane, Desalination 95 (1994) 325–345.

[68] R. Bernstein, S. Belfer, V. Freger, Toward improved boron removal in RO by membrane modification: feasibility and challenges., Environ. Sci. Technol. 45 (2011) 3613–20.

[69] A. Tullo, Chemical firms sell businesses in wave of divestitures, Chem. Eng. News (2013) 13.

[70] J. Geens, B. De Witte, B. Van Der Bruggen, Removal of API’s (active pharmaceutical ingredients) from organic solvents by nanofiltration, Sep. Sci. Technol. 42 (2007) 2435–2449.

[71] American Chemical Council, Chemical industry profile, Am. Chem. Counc. (2015).

[72] I. Sereewatthanawut, I.I.R. Baptista, A.T. Boam, A. Hodgson, A.G. Livingston,

Nanofiltration process for the nutritional enrichment and refining of rice bran oil, J. Food Eng. 102 (2011) 16–24.

[73] D. Peshev, L.G. Peeva, I.I.R. Baptista, A.T. Boam, Application of organic solvent nanofiltration for concentration of antioxidant extracts of rosemary (Rosmarinus officiallis L.), Chem. Eng. Res. Des. 89 (2011) 318–327.

[74] L.R. Firman, N.A. Ochoa, J. Marchese, C.L. Pagliero, Deacidification and solvent recovery of soybean oil by nanofiltration membranes, J. Membr. Sci. 431 (2013) 187–196.

[75] J.S. Wilkes, J.A. Levisky, R.A. Wilson, C.L. Hussey, Dialkylimidazolium chloroaluminate melts: a new class of room-temperature ionic liquids for electrochemistry, spectroscopy and synthesis, Inorg. Chem. 21 (1982) 1263–1264.

[76] S. Han, H.T. Wong, A.G. Livingston, Application of organic solvent nanofiltration to separation of ionic liquids and products from ionic liquid mediated reactions, Chem. Eng. Res. Des. 83 (2005) 309–316.

[77] C. Van Doorslaer, D. Glas, A. Peeters, A. Cano-Odena, I. Vankelecom, K. Binnemans, P. Mertens, D. De Vos, Product recovery from ionic liquids by solvent- resistant nanofiltration: application to ozonation of acetals and methyl oleate, Green Chem. 12 (2010) 1726.

[78] C. Abels, C. Redepenning, A. Moll, T. Melin, M. Wessling, Simple purification of ionic liquid solvents by nanofiltration in biorefining of lignocellulosic substrates, J. Membr. Sci. 405–406 (2012) 1–10.

[79] S. Hazarika, N.N. Dutta, P.G. Rao, Dissolution of lignocellulose in ionic liquids and its recovery by nanofiltration membrane, Sep. Purif. Technol. 97 (2012) 123–129.

[80] X. Cao, X. Wu, T. Wu, K. Jin, B.K. Hur, Concentration of 6-aminopenicillanic acid from penicillin bioconversion solution and its mother liquor by nanofiltration membrane, Biotechnol. Bioprocess Eng. 6 (2001) 200–204.

[81] G. Baumgarten, OSN at Evonik, in: 3rd Int. Conf. Org. Solvent Nanofiltration, 2010.

[82] S. Darvishmanesh, L. Firoozpour, J. Vanneste, P. Luis, J. Degrève, B. van der Bruggen, Performance of solvent resistant nanofiltration membranes for purification of residual solvent in the pharmaceutical industry: Experiments and simulation, Green Chem. 13 (2011) 3476–3483.

[83] E.M. Rundquist, C.J. Pink, A.G. Livingston, Organic solvent nanofiltration: a potential alternative to distillation for solvent recovery from crystallisation mother liquors, Green Chem. 14 (2012) 2197.

[84] R.M. Gould, L.S. White, C.R. Wildemuth, Membrane separation in solvent lube dewaxing, Environ. Prog. 20 (2001) 12–16.

[85] M. Namvar-Mahboub, M. Pakizeh, S. Davari, Preparation and characterization of UZM-5/polyamide thin film nanocomposite membrane for dewaxing solvent recovery, J. Membr. Sci. 459 (2014) 22–32.

[86] J.P. Sheth, Y. Qin, K.K. Sirkar, B.C. Baltzis, Nanofiltration-based diafiltration process for solvent exchange in pharmaceutical manufacturing, J. Membr. Sci. 211 (2003) 251–261.

[87] S. Darvishmanesh, T. Robberecht, P. Luis, J. Degrève, B. van der Bruggen, Performance of nanofiltration membranes for solvent purification in the oil industry, JAOCS, J. Am. Oil Chem. Soc. 88 (2011) 1255–1261.

[88] E. Rundquist, C. Pink, E. Vilminot, A. Livingston, Facilitating the use of counter- current chromatography in pharmaceutical purification through use of organic solvent nanofiltration, J. Chromatogr. A 1229 (2012) 156–163.

[89] J. Vanneste, S. De Ron, S. Vandecruys, S.A. Soare, S. Darvishmanesh, B. Van Der Bruggen, Techno-economic evaluation of membrane cascades relative to simulated moving bed chromatography for the purification of mono- and oligosaccharides, Sep. Purif. Technol. 80 (2011) 600–609.

[90] V. Nwuha, Novel studies on membrane extraction of bioactive components of green tea in organic solvents: Part I, J. Food Eng. 44 (2000) 233–238.

[91] I.H. Tsibranska, B. Tylkowski, Concentration of ethanolic extracts from Sideritis ssp. L. by nanofiltration: Comparison of dead-end and cross-flow modes, Food Bioprod. Process. 91 (2013) 169–174.

[92] A. Gupta, N.B. Bowden, Separation of cis-fatty acids from saturated and trans -fatty acids by nanoporous polydicyclopentadiene membranes, ACS Appl. Mater. Interfaces 5 (2013) 924–933.

[93] D. Kozma, H. Simon, C. Kassai, Z. Madarász, E. Fogassy, Investigation of the physicochemical basis of enantiomeric enrichment: The example of α- phenylethylamine with achiral dicarboxylic acids, Chirality 13 (2001) 29–33.

[94] N.F. Ghazali, D. a Patterson, A.G. Livingston, Elucidation of the mechanism of chiral selectivity in diastereomeric salt formation using organic solvent nanofiltration., Chem. Commun. (2004) 962–3.

[95] N.F. Ghazali, F.C. Ferreira, A.J.P. White, A.G. Livingston, Enantiomer separation by enantioselective inclusion complexation-organic solvent nanofiltration, Tetrahedron Asymmetry 17 (2006) 1846–1852.

[96] P. van der Gryp, A. Barnard, J.P. Cronje, D. de Vlieger, S. Marx, H.C.M. Vosloo, Separation of different metathesis Grubbs-type catalysts using organic solvent nanofiltration, J. Membr. Sci. 353 (2010) 70–77.

[97] J.T. Rundel, B.K. Paul, V.T. Remcho, Organic solvent nanofiltration for microfluidic purification of poly(amidoamine) dendrimers, J. Chromatogr. A 1162 (2007) 167–174.

[98] G. Szekely, M. Gil, B. Sellergren, W. Heggie, F.C. Ferreira, Environmental and economic analysis for selection and engineering sustainable API degenotoxification processes, Green Chem. 15 (2013) 210–225.

[99] P.M. Budd, N.B. McKeown, D. Fritsch, Free volume and intrinsic microporosity in polymers, J. Mater. Chem. 15 (2005) 1977–1986.

[100] M. Heuchel, D. Fritsch, P.M. Budd, N.B. McKeown, D. Hofmann, Atomistic packing model and free volume distribution of a polymer with intrinsic microporosity (PIM-1), J. Membr. Sci. 318 (2008) 84–99.

[101] P. Gorgojo, S. Karan, H.C. Wong, M.F. Jimenez-Solomon, J.T. Cabral, A.G. Livingston, Ultrathin polymer films with intrinsic microporosity: Anomalous solvent permeation and high flux membranes, Adv. Funct. Mater. 24 (2014) 4729– 4737.

[102] R. Swaidan, B. Ghanem, I. Pinnau, Fine-tuned intrinsically ultramicroporous polymers redefine the permeability/selectivity upper bounds of membrane-based air and hydrogen separations, ACS Macro Lett. 4 (2015) 947–951.

[103] M.A. Abdulhamid, H.W.H. Lai, Y. Wang, Z. Jin, Y.C. Teo, X. Ma, I. Pinnau, Y. Xia, Microporous Polyimides from Ladder Diamines Synthesized by Facile Catalytic Arene-Norbornene Annulation as High-Performance Membranes for Gas Separation, Chem. Mater. (2019) acs.chemmater.8b05359.

[104] X. Ma, M. Abdulhamid, X. Miao, I. Pinnau, Facile synthesis of a hydroxyl- functionalized Tröger’s base diamine: A new building block for high-performance polyimide gas separation membranes, Macromolecules 50 (2017) 9569–9576.

[105] B.S. Ghanem, M. Hashem, K.D.M. Harris, K.J. Msayib, M. Xu, P.M. Budd, N. Chaukura, D. Book, S. Tedds, A. Walton, N.B. McKeown, Triptycene-based polymers of intrinsic microporosity: Organic materials that can be tailored for gas adsorption, Macromolecules 43 (2010) 5287–5294.

[106] F. Alghunaimi, B. Ghanem, N. Alaslai, R. Swaidan, E. Litwiller, I. Pinnau, Gas permeation and physical aging properties of iptycene diamine-based microporous polyimides, J. Membr. Sci. 490 (2015) 321–327.

[107] B.S. Ghanem, F. Alghunaimi, Y. Wang, G. Genduso, I. Pinnau, Synthesis of highly gas-permeable polyimides of intrinsic microporosity serived from 1,3,6,8- tetramethyl-2,7-diaminotriptycene, ACS Omega 3 (2018) 11874–11882.

[108] R. Swaidan, B. Ghanem, M. Al-Saeedi, E. Litwiller, I. Pinnau, Role of intrachain rigidity in the plasticization of intrinsically microporous triptycene-based polyimide

membranes in mixed-Gas CO2/CH4 separations, Macromolecules 47 (2014) 7453– 7462.

[109] R. Swaidan, M. Al-Saeedi, B. Ghanem, E. Litwiller, I. Pinnau, Rational design of intrinsically ultramicroporous polyimides containing bridgehead-substituted triptycene for highly selective and permeable gas separation membranes, Macromolecules 47 (2014) 5104–5114.

[110] B.S. Ghanem, R. Swaidan, E. Litwiller, I. Pinnau, Ultra-microporous triptycene- based polyimide membranes for high-performance gas separation, Adv. Mater. 26 (2014) 3688–3692.

[111] G. Genduso, B. Ghanem, Y. Wang, I. Pinnau, G. Genduso, B.S. Ghanem, Y. Wang, I. Pinnau, Synthesis and gas-permeation characterization of a novel high-surface area polyamide derived from 1,3,6,8-tetramethyl-2,7-diaminotriptycene: Towards polyamides of intrinsic microporosity (PIM-PAs), Polymers 11 (2019) 361.

[112] S. Tsarkov, V. Khotimskiy, P.M. Budd, V. Volkov, J. Kukushkina, A. Volkov, Solvent nanofiltration through high permeability glassy polymers: Effect of polymer and solute nature, J. Membr. Sci. 423–424 (2012) 65–72.

[113] T.S. Anokhina, A.A. Yushkin, P.M. Budd, A.V. Volkov, Application of PIM-1 for solvent swing adsorption and solvent recovery by nanofiltration, Sep. Purif. Technol. (2015).

[114] N. Fridman-Bishop, V. Freger, What makes aromatic polyamide membranes superior: New insights into ion transport and membrane structure, J. Membr. Sci. 540 (2017) 120–128.

Chapter 2. Theory and Background

This chapter deals with the fundamental theories and models explaining the transport of penetrants through membranes coupled with the understanding of membrane performance in gas- and liquid systems. Relevant process conditions and their potential effects are discussed. Finally, the chapter reviews the potential advantages of the interfacial polymerization process and its basic concepts.

2.1. Gas Transport through Membranes

Baker defines a membrane as, “nothing more than a discrete, thin interface that moderates the permeation of chemical species in contact with it” while Mulder describes it as, "a selective barrier between two phases” [1,2]. Essentially a membrane is a semi-permeable barrier allowing permeation of one component over another. Membranes come in a wide variety of thicknesses and morphological structures. The thickness of initially developed membranes ranged between 10 to 100 µm. With the development of membrane science, membrane active layers are currently on the order of ≤ 1 µm and hence are referred to as

"thin-films”. Because such films are difficult to handle mechanically, the thin-films are generally fabricated on support materials and the final composition is known as a thin-film composite (TFC) membrane.

On the basis of pore size, polymeric membranes can be classified in three categories: (i) macroporous with pore size > 500 Å; (ii) mesoporous with pore size between 20–500 Å, and; (iii) microporous with pore size < 20 Å, depicted in Figure 2.1. In polymeric membranes, pore size represents a pore size distribution (PSD) rather than a rigid singular value. Macroporous and mesoporous membranes are generally classed as “porous”

68 membranes. Transport of components through such membranes is highly dependent on the shape and size of the penetrants involved and their ability to pass through the pores in the polymer matrix, rather than depending on inherent polymer properties. In ideal cases, any particles larger than the pore size of the membrane are completely rejected, while any particles smaller than the membrane pore size are entirely permeated. Particles with sizes close to the membrane PSD are partially rejected. Such membranes are used primarily in liquid separation processes for both soluble and insoluble components in a stream, e.g., ultrafiltration for separating viruses, proteins, and enzymes from water or pharmaceutical streams and microfiltration for removing bacteria and organic matter from wastewater streams.

Figure 2.1. Pore size dependent transport processes. Adopted from Swaidan et al. [3].

Membranes with PSDs < 20 Å are generally classified as ‘dense’ membranes as no visible pores can be observed directly. For a component to permeate through such membranes, the component must diffuse through the polymer due to a chemical potential difference

(generally a pressure or concentration gradient). Hence, separation depends entirely on the transport rate of the component within the membrane material. This transport is determined by the diffusivity and solubility of the component in the membrane material. Industrial processes such as pervaporation, reverse osmosis, and gas separation utilize non-porous dense membranes [1].

2.1.1. Pore-Size-Dependent Transport

For large pore sizes, i.e., > 100 nm, gas permeates through a membrane by convective or viscous flow. As seen in Figure 2.1, for large pores, gas molecules have relatively little interaction with the walls of the pores. Hence, energy is primarily lost by intermolecular interactions, defined as the viscosity of a mixture. All molecules flow at the same rate, and thus, no separation of gases can be achieved [4].

As pore sizes decrease, i.e., between 2 to 100 nm (mesoporous membranes), the mean free path of gas molecules becomes comparable to the size of the pores. The mean free path of a gas is defined as the average distance traveled by a gas molecule before colliding with another gas molecule. It can be described [5] as;

푘푇 휆 = (Eq. 2.1) 휋푑푔푎푠푝√2 where, λ = mean free path length, T = temperature, k = Boltzmann constant, dgas = gas molecular diameter and p = hydrostatic pressure.

For such pore sizes, interactions between the flowing molecules and the pore walls become more dominant than intermolecular interactions and hence, separation can occur depending on the component’s interaction with the pore walls (known as Knudsen diffusion). The flux through such a membrane is described [6] as;

휋푛푟2퐷푘∆푝 퐽 = (Eq. 2.2) 푅푇휏푙 where, n = number of pores in membrane, r = pore radius, τ = tortuosity and l = membrane thickness. Dk can further be described as;

8푅푇 퐷푘 = 0.66푟√ (Eq. 2.3) 휋푀푊

where, R = universal gas constant and MW = molecular weight of the penetrant.

Transport via Knudsen diffusion is inversely proportional to the molecular weight of the penetrant, and hence, selectivity is proportional to the square root of the ratio of the molecular weights of the components,

푀푘 훼푗,푘 ∝ √ (Eq. 2.4) 푀푗

As the equation of transport states, Knudsen diffusion is dependent on process conditions such as temperature and pressure and membranes with large pore sizes (> 2 nm) can be used given low pressure or high temperature operating conditions. Similarly, process conditions can be used to maximize potential permselectivity for required applications,

235 238 e.g., porous membranes were used for the separation of U F6 and U F6 despite Knudsen

71 selectivity of only ≈ 1.0043. Hundreds of stages were used to obtain reasonable enrichment

[4].

As pore size is further reduced to < 2 nm (microporous range), it starts to reach the size of molecular dimensions of gases and transport mechanisms become more complex.

Transport is highly dependent on the intrinsic PSD of the material as well as the size

(primarily affecting diffusion) and chemical nature (primarily affecting sorption) of the penetrants. Pores in the microporous range (< 2 nm) are further divided into supermicropores (0.7 to 2 nm) and ultramicropores (< 0.7 nm). Transport in the supermicroporous range occurs by a combination of mechanisms including molecular sieving, surface diffusion and capillary condensation depending on the condensability of the components; see Figure 2.2. For ultramicropores, penetrants permeate by the solution- diffusion model.

Figure 2.2. Transport of non-condensable and condensable components through meso/micropores. Adopted from Swaidan et al. [3].

2.1.2. The Solution-Diffusion Model

The foundations of the solution-diffusion model were established by Graham in the nineteenth century but started to gain significant attention in the 1940s to describe the transport of components through dense barrier layers [7,8]. Before this, pore flow-based models were dominant in the literature due to their relatively simplistic explanation of transport phenomena. Over the last 50 years, the solution-diffusion model has established itself as the dominant model used to explain transport through dense materials due to its ability to describe a wide range of membrane-based processes including gas separation and reverse osmosis [9].

In simplistic terms, the solution-diffusion model describes transport through a dense material in three steps;

a. Penetrant sorbs on the upstream side of the membrane

b. Penetrant diffuses through the membrane material across a chemical potential

gradient

c. Penetrant desorbs on the downstream side of the membrane

The mathematical basis of the model begins with the assumption that the passage of a penetrant through the membrane is a function of the chemical potential gradient across the membrane (see Figure 2.3). Hence, the flux, J, of a component through a membrane can be expressed as:

휕휇 퐽 = −퐿 (Eq. 2.5) 휕푥 where, dµ/dx = chemical potential gradient and L is a proportionality constant.

Chemical potential is a generic expression which represents a driving force, such as pressure, concentration or temperature gradient, or a mixture of multiple driving forces combined depending on the specific process involved. If the driving force is restricted to only pressure and concentration, the chemical potential can be expressed as:

휕휇 = 푅푇. 휕 푙푛(훾푐) + 푣휕푝 (Eq. 2.6) where, γ = activity coefficient and c = molar concentration (mol/mol) and v = molar volume of component.

For incompressible fluids, Eq. 2.6 can be integrated remembering molar volume changes with pressure and can be described as;

휇 = 휇표 + 푅푇푙푛(훾푐) + 푣(푝 − 푝표) (Eq. 2.7) where, µ0 = chemical potential of the pure component at pressure p0.

For compressible gases, Eq. 2.6 can be integrated remembering molar volume changes with pressure and can be described as;

푝 휇 = 휇표 + 푅푇푙푛(훾푐) + 푅푇푙푛 (Eq. 2.8) 푝표

0 where, p is described as the saturation vapor pressure of component, psat, Eq. 2.7 can be written for incompressible liquids as:

표 휇 = 휇 + 푅푇푙푛(훾푐) + 푣푖(푝 − 푝푠푎푡) (Eq. 2.9) whereas for compressible gases, it can be written as:

푝 휇 = 휇표 + 푅푇푙푛(훾푐) + 푅푇푙푛 (Eq. 2.10) 푝푠푎푡

The primary assumption for the solution-diffusion model is that on both sides of the membrane, membrane material interfaces are in equilibrium with the fluids, i.e., no variation in the driving force. This implies a constant chemical potential gradient from one membrane side to the other and further leads to the assumption that the diffusion rate through the membrane is much lower than sorption and desorption rates at the membrane interface. SD model also assumes that (contrary to the pore flow model) the pressure through the membrane is constant at the higher-pressure value. This implies that pressure is transmitted similarly to liquids in a solution-diffusion membrane [9,10].

Eq. 2.5 and Eq. 2.6 can be combined to express flux as:

−푅푇퐿 휕푐 퐽 = . (Eq. 2.11) 푐 휕푥

For gas permeation, RTL/c can be replaced by D, the diffusivity or diffusion coefficient, and flux of a gas across a concentration gradient can be defined according to Fick’s First

Law:

휕푐 퐽 = −퐷 (Eq. 2.12) 휕푥

D can be assumed constant if there is no interaction between the penetrants themselves or with the polymeric material. Fick’s First Law can be integrated across the membrane thickness (l) and written as:

(푐 − 푐 ) 퐽 = −퐷 0 1 (Eq. 2.13) 푙

where, c0 and c1 are concentrations of the penetrant on the upstream and downstream side of the membrane, respectively. Eq. 2.13 above describes the steady-state transport of penetrant through a membrane represented schematically in Figure 2.3.

In gas permeation, a gas (or mixture) is pressurized at pressure p0 and permeated through the membrane to a lower downstream pressure, p1. To determine the gas transport equation, the chemical potential on the upstream bulk (µ0) and upstream of the membrane (µ0(m)) must be equated. These are described by Eq. 2.9 and 2.10. Equating the two equations gives:

푝 휇표 + 푅푇푙푛(훾 푐 ) + 푅푇푙푛 = 휇 0 0 푝 푠푎푡 (Eq. 2.14) 표 = 휇 + 푅푇푙푛(훾0(푚)푐0(푚)) + 푣(푝0 − 푝푠푎푡)

This can be simplified to:

훾 푝표 −푣(푝표 − 푝푠푎푡) 푐표(푚) = . . 푐표. 푒푥푝 ( ) (Eq. 2.15) 훾표(푚) 푝푠푎푡 푅푇

The exponential term in the Eq. 2.15 is very close to 1, and so the equation can be reduced to:

훾표 푝표 푐표(푚) = . . 푐표 (Eq. 2.16) 훾표(푚) 푝푠푎푡

Since, co x po equals the partial pressure of the gas (pio) in the feed, Eq. 2.16 can be written as:

훾 푝푖표 푐표(푚) = . (Eq. 2.17) 훾표(푚) 푝푠푎푡

γo/(γo(m)*psat) is described as the sorption coefficient, S, of the component and describes the solubility of a gas in a polymer, when in equilibrium with the partial pressure of the gas.

Hence, the concentration of the component at the feed/membrane interface can be described as:

푐표(푚) = 푆. 푝표 (Eq. 2.18) whereas, on the permeate side can be described as:

푐1(푚) = 푆. 푝1 (Eq. 2.19)

Eq. 2.12 and Eq. 2.13 can be combined with Eq. 2.9 to give:

(푝 − 푝 ) 퐽 = 퐷푆 표 1 (Eq. 2.20) 푙

D x S is defined as the permeability coefficient, P, and so the flux can be defined as:

(푝 − 푝 ) 퐽 = 푃. 표 1 (Eq. 2.21) 푙

Eq. 2.21 is widely used to understand the gas permeation properties of materials as it highlights the assumptions of the model which are summarized as: i). There is a concentration gradient within the membrane whereas the pressure remains constant. ii). Sorption of a component in the membrane is independent of the total gas pressure and is a function of its activity in the adjacent gas.

Eq. 2.21 can be written as:

퐽푙 푐 −푐 푃 = = 1 0 퐷 (Eq. 2.22) 푝1−푝0 푝1−푝0

The permeability of a gas can hence be described as the thickness- and pressure-normalized flux through a film [11]. For upstream concentration and pressure significantly greater than its downstream values, Eq. 2.22 can be simplified to:

푐1 푃 = 퐷 (Eq. 2.23) 푝1

Since S can be defined as:

푐 푆 = (Eq. 2.24) 푝

Eq. 2.23 can be rearranged to define the permeability coefficient of a polymer:

푃 = 퐷푆 (Eq. 2.25)

Hence, permeability through a membrane is a function of two terms: (1) a kinetic term, D, that describes the movement of the gas molecules through the polymer, and; (2) a thermodynamic term, S, that represents the sorption of gas molecules in the polymer. The equation explains how high permeability can be achieved by either diffusivity D (e.g., He and H2), large sorption S (e.g., CO2) or a combination of both, e.g., H2O.

Permeability can be described in SI units of mol m m-2 s-1 Pa-1 but is more commonly denoted in barrer (where, 1 barrer = 10-10 cm3 (STP) cm (cm-2 s-1 cmHg-1)) in honor of

Richard Barrer’s pioneering work in the field. Permeability can show variations of up to

10 orders of magnitude depending on the penetrant and membrane material. A large number of techniques have been developed to measure P, D and S in polymers which are discussed in detail elsewhere [12]. The most common methods involve measuring coefficients on thick isotropic films (~50 to 100 µm). The most widely used method involves measuring steady-state permeability directly while simultaneously measuring D before the steady-state is achieved, known as the time-lag method. S is then calculated mathematically [13,14]. Similarly, given direct P measurements, steady-state S can be measured separately using barometric or gravimetric sorption, and D can be calculated

[15–17].

2.1.3. Characterizing Transport

As discussed, P, D and S measurements are generally made on thick isotropic films with a thickness on the order of ~50 to 100 µm. However, commercial membranes are generally

TFC membranes with a thin active layer on the order of < 1 µm and more realistically ~100 nm. Furthermore, coupled with complex and heterogeneous surface morphology effects, it

79 has proven difficult to make accurate thickness measurements of the active layer of TFCs

[18,19] despite numerous advances in imaging and characterization technologies. In such cases, transport is characterized in terms of permeance (P/l), i.e., steady-state pressure- normalized flux. It is defined in SI units as: mol m-2 s-1 Pa-1 or more commonly in Gas

Permeation Units (GPU) where, 1 GPU = 10-6 cm3 (STP) cm-2 s-1 cmHg-1. 1 GPU defines the pressure-normalized flux for a 1 µm thick membrane with a permeability of 1 barrer.

The second main performance characterization parameter used for evaluating membranes is selectivity (α). Selectivity can be explained generically as the capability of a membrane to separate component A from B. Ideal selectivity can be defined mathematically as:

푃퐴 훼퐴퐵 = (Eq. 2.26) 푃퐵

Combining Eq. 2.25 and Eq. 2.26 gives:

퐷퐴 푆퐴 퐷 푆 훼퐴퐵 = ( ) ( ) = 훼퐴퐵훼퐴퐵 (Eq. 2.27) 퐷퐵 푃퐵

D S α AB and α AB are within themselves essential terms as they help to independently analyze the effect of the two components on the selectivity of the membrane as well as help classify potential membrane materials.

Pure-gas permeance can be used to determine the ‘ideal’ selectivity as:

푄퐴 훼퐴퐵 = ( ) (Eq. 2.28) 푄퐵

Ideal selectivity is obtained using pure-gas tests, i.e., each gas is permeated separately through the membrane. In real mixed-gas cases, in which binary or tertiary mixtures are

80 permeated together, the transport of one gas can significantly affect the transport of the other(s). In such systems the ‘actual’ separation factor for a binary mixture can be calculated as follows:

∗ 푦퐴 푥퐴 훼퐴퐵 = ( ) / ( ) (Eq. 2.29) 푦퐵 푥퐵 where, yA and yB are mole fractions of components A and B downstream, and xA and xB are mole fractions of components A and B upstream of the membrane.

The separation factor can depend significantly on process parameters as well as the properties of the gas-polymer system, e.g., upstream and downstream pressure and temperature (discussed later).

2.1.4. Permeability/Selectivity Trade-Offs

After decades of research on polymers for membrane technology, relationships were established to characterize the overall permselectivity performance of polymers relative to one another. It became evident that the selectivity achieved for gas pairs was inversely correlated to the permeability of the faster gas and Robeson was one of the first to compile data for permselectivity of a large number of polymeric materials creating the ‘upper- bound” [20]. The method gained popularity as observations could be explained by simple considerations based on transition-state or free-volume models, and the performance was mapped in plots such as those in Figure 2.4 [20–24].

The upper bound consists of the permeability plotted against the selectivity, and the relationship can be explained by:

훽 훼 = ( 퐴퐵 ) (Eq. 2.30) 퐴퐵 휆퐴퐵 푃퐴

where, λAΒ and βAB are model parameters, and PA is the permeability of the faster gas. The slope of the graph can be estimated using the kinetic diameters of the gas pairs under consideration using the equation:

2 푑퐵 휆퐴퐵 = ( ) − 1 (Eq. 2.31) 푑퐴

similarly, βAB can be related to the solubility as:

1+휆퐴퐵 푆퐴 1 − 푎 훽퐴퐵 = 푒푥푝 [−휆퐴퐵 (푏 − 푓 { })] (Eq. 2.32) 푆퐵 푅푇 where, a, b and f are fixed constants.

It can be seen from Figure 2.4 that over the last three decades significant progress has been made towards high-performance polymers. Robeson [24], 2001, and recently Swaidan et al. [23] revisited the upper bound to include emerging materials including a large number of PIM-based materials which have helped redefine the potential performance achievable using polymers.

2.1.5. Process Considerations

2.1.5.1. Effect of Pressure on Transport

Pressure can have significant effects on both D and S and ultimately P in a polymer-gas system. Remembering that S=c/p, the effects of pressure on sorption are shown in Figure

2.5.

Figure 2.5. Variation of gas sorption with absolute and relative pressure: (a) O2 in PDMS [25]; (b) acetone in PDMS [26]; (c) argon in PSF [27]; (d) vinyl chloride monomer in PVC [28]. Reprinted with permission from Matteucci et al. [11] (Copyright © 2006 John Wiley & Sons).

Figure 2.5a shows the most straightforward behavior, depicting the sorption of light, non- interacting gases in rubbery polymers. Henry’s Law describes this relationship as:

푐 = 푘퐷푝 (Eq. 2.33) where, kD = Henry’s Law parameter.

For gases with higher condensability (e.g., CO2), rubbery polymers show deviation from

Henry's law at high relative pressures. S increases with increasing relative pressure and the relationships are described by the Flory-Huggins equation:

푝 2 푙푛 푎 = 푙푛 ( ) = 푙푛 휙푉 + (1 − 휙푉) + ꭓ(1 − 휙푉) (Eq. 2.34) 푝푠푎푡

where, psat = saturation vapor pressure, фV = volume fraction of penetrant in the polymer and  = the Flory-Huggins interaction parameter which depends on the temperature and concentration of the penetrant.

The next common isotherm observed is shown in Figure 2.5c, commonly known as the dual-mode sorption [29] isotherm for glassy polymers described by the equation:

푏푝 푐 = 푘 푝 + 퐶′ (Eq. 2.35) 퐷 퐻 1 + 푏푝

where, C’H is the Langmuir sorption capacity, and b is the Langmuir affinity parameter.

The model explains sorption in two states in equilibrium with one another. Here kD describes the sorption in the densely packed regions in the polymers while C’H describes species sorbed in the non-equilibrium excess volume. At very low pressures, the curve of the graph is described by finite gas solubility according to the equation:

′ 푆 = 푘퐷 + 퐶퐻푏 (Eq. 2.36)

At low to moderate pressure, sorption is dominated by the C’H term. The isotherm then transitions, at p = 1/b, to Henry’s Law at higher pressures. Non-equilibrium excess volume is an inherent characteristic of glassy polymers, below their Tg, and is generally expected to increase with T.

Figure 2.5d shows sorption in a glassy polymer with a relatively high solute concentration.

In such cases, Henry's Law cannot model the sorption behavior, and the behavior is explained by combining Flory-Huggins and Langmuir contributions. The exponential increase in sorption at higher relative pressures is described as plasticization and is commonly observed for highly sorbing gases such as CO2.

The effect of pressure on D is shown in Figure 2.6.

Figure 2.6a shows the simplest behavior for low sorbing gases in polymers where D is ideally independent of the penetrant concentration. If the penetrant is highly sorbing,

Figure 2.6b, a linear (or even exponential) increase in D is observed for increasing

87 concentration. This is explained by the sorption of the condensable components increasing the free volume of the polymer-gas system. This, in turn, increases chain mobility causing an increase in D. Figure 2.6c shows clustering phenomena in which penetrants can aggregate resulting in an increase in their effective size explaining the reduction of D.

Figure 2.6d shows typical behavior for sorbing gases in glassy polymers described by the dual-mode sorption mode The effect of pressure on P can be extrapolated from its impact on D and S and is depicted in Figure 2.7.

For low sorbing gases, P remains relatively constant with pressure, Figure 2.7a, but if the component is highly sorbing it can induce plasticization in the polymer and P increases with increasing pressure. In rigid, glassy polymers, the dual-mode sorption model predicts a decrease in P with increasing pressure, Figure 2.7c. However, at high concentration of

highly sorbing components like CO2, plasticization can occur, Figure 2.7d, resulting in significant increases in P.

2.1.5.2. Effect of Temperature on Transport

Process temperature can have a significant impact on gas transport properties. The relationship is described with Arrhenius-van ‘t Hoff equations as below:

−퐸퐷/푅푇 퐷 = 퐷0푒 (Eq. 2.37)

−퐸푃/푅푇 푃 = 푃0푒 (Eq. 2.38)

where, ED and EP are activation energies of diffusion and permeation. Sorption dependence on temperature is previously discussed, and HS is the enthalpy of sorption. EP can be described as:

퐸푃 = 퐸퐷 + ∆퐻푠 (Eq. 2.39)

For size sieving membranes, ED > |HS| and hence, a positive EP means P increases with increasing temperature. In such cases, EP and ED are independent of temperature. On the other hand, EP can be negative for reverse selective materials, e.g., PDMS and EP and ED can be temperature dependent.

2.1.5.3. Pressure Ratio

Pressure ratio (ф), defined as pfeed/ppermeate, is an often overlooked but is an essential process variable for the design of industrial gas separation systems. It determines the maximum enrichment possible in a permeate stream from a feed stream regardless of membrane

90 permselectivity. For a given membrane system, for a penetrant, i, to flow from upstream

(feed) to downstream (permeate):

푝푓 ≥ 푝푝 (Eq. 2.40)

Partial pressures of i can be defined in terms of total pressure (p) as:

푦푓푃푓 ≥ 푦푝푃푝 (Eq. 2.41)

Rearranging the equation to:

푦푝/푦푓 ≤ 휑 (Eq. 2.42)

It can be seen from the equation that the maximum enrichment (yp/yf) obtainable is determined by the ratio of absolute pressures used upstream and downstream of the process regardless of the membrane permselectivity performance. Furthermore, membrane selectivity, pressure ratio, and enrichment can be equated via:

휑 1 1 1 1 2 4훼훾 훾 = [훾 + + − √(훾 + + ) − 푓 ] (Eq. 2.43) 푝 2 푓 휑 훼 − 1 푓 휑 훼 − 1 (훼 − 1)휑

The equation can be used to plot maximum enrichment based on membrane selectivity and pressure ratio as shown in Figure 2.8. The figure illustrates that enrichment increases as selectivity increases but as selectivity reaches values similar to ф, no enrichment increments can be obtained and increasing selectivity beyond this point results in diminishing returns, i.e., the process becomes pressure-ratio limited. Selectivity only limits the enrichment when ф is significantly higher than selectivity, but this is rarely the case as low downstream pressure adds to compression cost which can potentially be the most

91 substantial costs associated with membrane separations. Therefore, process parameters like pressure-ratio must be considered when designing membrane systems for specific processes.

2.2. Liquid Transport through Membranes

2.2.1. Solution-Diffusion

Similar to gas separation, solution-diffusion has prevailed as the dominant model to explain the transport of liquids through dense polymer membranes [9]. Here, reverse osmosis (RO) will be used as a model liquid separation process to explain membrane transport in liquid systems. If a semi-permeable membrane (i.e., water permeability >> salt permeability) is used to separate pure water and a concentrated salt solution, water moves from the pure- water side to the concentrated salt side due to osmotic pressure (π). The osmotic pressure can be defined as:

휋 = 푖푀푅푇 (Eq. 2.44) where, M = molar concentration of solute (mol.L-1) and i = the van ‘t Hoff factor.

The process is known as osmosis (or normal osmosis for clarity). If hydraulic pressure is applied and increased on the salt side, the flow of water from the pure-water side is retarded and eventually stops when the applied pressure equals the osmotic pressure. As pressure is further increased, the flow of water is reversed, i.e., water begins to move from the salt solution towards the pure-water side. This is known as reverse osmosis (RO) and is currently the dominant process for producing potable water from seawater.

The assumptions and general principles of the model remain the same for gas and liquid systems and principles of operation can be derived from previously established ideas. In

SD, the chemical potential on the feed side bulk and feed side membrane interface are equal, so:

휇표 = 휇표(푚) (Eq. 2.45)

Eq. 2.9 can be substituted in the above equation to give:

표 휇 + 푅푇푙푛(훾표푐표) + 푣푖(푝표 − 푝푠푎푡) (Eq. 2.46) 표 = 휇 + 푅푇푙푛(훾표(푚)푐표(푚)) + 푣푖(푝표 − 푝푖.푠푎푡) which can be written as:

푙푛(훾표푐표) = 푙푛(훾표(푚)푐표(푚)) (Eq. 2.47) and can be rearranged as:

훾표 푐표(푚) = 푐표 (Eq. 2.48) 훾표(푚)

The ratio yo/yo(m) is described as the sorption coefficient, S, which can be added to the equation to give:

푐1(푚) = 푆푐1 (Eq. 2.49)

For the downstream side, there is a pressure difference between the downstream membrane interface and the permeate. The chemical potential on this interface can be described as:

휇1 = 휇1(푚) (Eq. 2.50)

Eq. 2.9 can be substituted above to give:

표 휇 + 푅푇푙푛(훾1푐1) + 푣푖(푝1 − 푝푠푎푡) (Eq. 2.51) 표 = 휇 + 푅푇푙푛(훾1(푚)푐1(푚)) + 푣(푝표 − 푝푠푎푡) which can be written as:

푣(푝 − 푝 ) 푙푛(훾 푐 ) = 푙푛(훾 푐 ) + 표 푙 (Eq. 2.52) 1 1 1(푚) 1(푚) 푅푇

The equation can be rearranged while integrating the sorption coefficient as:

−푣(푝 − 푝 ) 푐 = 푆푐 푒푥푝 ( 표 푙 ) (Eq. 2.53) 1 1 푅푇

Above equations 2.55 and 2.59 can be substituted into the Fick’s law derived Eq. 2.1.2.9 to give:

−푣(푝 − 푝 ) 퐽 = 퐷푆 [푐 − 푐 푒푥푝 ( 표 푙 )] (Eq. 2.54) 표 1 푅푇

The above equation can be used to describe water and salt flux through an RO membrane.

It can be further simplified by considering an upstream hydrostatic pressure that equals the osmotic pressure of the salt solution which can be written as:

퐷푆 −푣(∆휋) 퐽 = 0 = [푐 − 푐 푒푥푝 ( )] (Eq. 2.55) 푖 푙 표 1 푅푇 and can be rearranged as:

푣(∆휋) 푐 = 푐 푒푥푝 ( ) (Eq. 2.56) 1 표 푅푇

This can be substituted in Eq. 2.61 to give:

퐷푆푐 −푣(훥푝 − ∆휋) 퐽 = 표 [1 − 푒푥푝 ( )] (Eq. 2.57) 푖 푙 푅푇

where, ΔP = (p0-p1). It has been observed that the term –v(Δp-Δπ)/RT is very small under experimental conditions. Hence, the equation can be simplified as:

퐷푆푐 푣(훥푝 − ∆휋) 퐽 = 표 (Eq. 2.58) 푖 푙푅푇

Similarly, salt transport can be explained by Eq. 2.60 for the salt component (j) as:

퐷 푆 −푣 (푝 − 푝 ) 퐽 = 푗 푗 [푐 − 푐 푒푥푝 ( 푗 표 1 )] (Eq. 2.59) 푗 푙 푗표 푗1 푅푇

As said, –vj(Δp-Δπ)/RT is small, so the equation reduces to:

퐷 푆 퐽 = 푗 푗 (푐 − 푐 ) (Eq. 2.60) 푗 푙 푗표 푗1

The above equations form the basis of transport of components through a dense RO membrane.

2.2.2. Characterizing Transport

Similar to the case described of characterizing gas transport, it is difficult to determine the active layer membrane thickness for commercial liquid separating membranes. Hence, performance is characterized in terms of permeance with SI units of mol m-2 s-1 Pa-1.

Although the permeance is generally described in more common units of liters per square meter hour (l m-2 h-1 or LMH conventionally) for given operating conditions. Furthermore, performance through the membrane is generally described in terms of the of the water

(component i) permeability coefficient, A, which is described as DiSici0vi/lRT. So, Eq. 2.64 becomes:

퐽푖 = 퐴(훥푝 − ∆휋) (Eq. 2.61)

Similarly, salt transport is described by the B parameter where B = DjSj/l. So, Eq. 2.66 becomes:

퐽푗 = 퐵(푐푗표 − 푐푗1) (Eq. 2.62)

For practical reasons, the ability of a membrane to separate salt and water is described as the rejection R:

(1 − 푐 ) 푅 = 푗1 . 100% (Eq. 2.63) 푐푗표

2.2.3. Permeability/Selectivity Trade-Offs

Similar to gas separation, attempts have been made to establish permeability/selectivity trade-off relations for liquid primarily desalination applications. Figure 2.9 shows the trade-off developed by Geise et al. [37]. However, unlike gas separation trade-offs, the slope of the trade-off (λ) and the shift factor (β) could not be backed by fundamental theoretical physical findings. Although a fundamental basis of the trade-off must exist, similar to gas separations, establishing such relationships is significantly more complex due to the impact of complicated polymer/penetrant interactions in the liquid state as well as chemical properties of the membranes such as ionic charge and surface hydrophilicity.

2.2.4. Process Considerations

2.2.4.1. Concentration Polarization

In the RO desalination process, water and salt co-permeate through the membrane to the permeate side. A useful RO membrane allows only a very small amount of salt to permeate, i.e., water permeability >> salt permeability. Salt is primarily retained on the feed side of the membrane while the water permeates which can cause salt to accumulate on the feed side of the membrane resulting in a higher concentration of salt at the membrane interface compared to the bulk salt feed solution. This can result in increased salt permeability (due to increased concentrative driving force) and can result in misinterpretation of membrane properties as well as seriously hinder the performance of membranes in industrial systems.

Concentration polarization can be divided into concentrative or dilutive polarization as the phenomenon can both result in an increase or decrease of salt concentration at the membrane surface respectively. Furthermore, concentration polarization divided into external and internal concentration polarization, i.e., ECP and ICP respectively.

Concentrative ECP results from convection pulling solute from the bulk to the membrane surface and can be defined as:

푐 퐽 푚 = 푒푥푝 ( 푤) (Eq. 2.64) 푐푏 푘

where, cm = concentration at the membrane surface, cb = concentration in bulk feed solution and k is the mass transfer coefficient which is highly dependent on flow conditions including the flow profile, channel geometry, etc.

Alternatively, if convection pulls the solute away from the membrane surface, it is known as dilutive ECP and is defined by:

푐 퐽 푚 = 푒푥푝 (− 푤) (Eq. 2.65) 푐푏 푘

When the solute concentration builds in the porous support rather than the membrane surface the phenomenon is known as ICP. This is of little concern in processes employing co-current flows, e.g., RO but can contribute significantly for processes like forward osmosis (FO). ICP describes the resistance of the solute to diffuse in the support material and can be described as:

휏푡 퐾 = ( ) (Eq. 2.66) 퐷푠휀 where, τ = support tortuosity, t = support thickness and  = support porosity. Ds is the solute diffusion coefficient in water.

2.3. Interfacial Polymerization

2.3.1. Introduction

Interfacial polymerization (IP) refers to a polycondensation reaction between two (or more) monomers in two (relatively) immiscible solvents. Miscibility is defined by Jackson and

Drury [38] qualitatively as the complete solubility of either solvent in the other. It is characterized by initially mixing a solvent pair under stirring followed by allowing the mixture to settle. Immiscible solvents phase separate with the formation of an interfacial meniscus as well as showing no change in volume for either solvent.

Figure 2.10. Nylon 6,10 via interfacial polymerization.

In IP, a polar (aqueous) phase containing reactant A and a non-polar (organic) phase containing reactant B are mixed in a vessel. The monomers react at the interface to produce an ultrathin polymeric film. Reaction kinetics are extremely fast estimated between 102-

106 L mol-1s-1. The IP process for producing Nylon 6, 10 using hexamethyl diamine and sebacoyl chloride is shown in Figure 2.16.

100

Figure 2. 11. Fabrication of interfacially polymerized TFCs.

IP can be used for the fabrication of ultra-thin polymeric active layers (Figure 2.11) which can be deposited on organic or inorganic supports to form TFC membranes [39]. For the preparation of polymeric TFC membranes, shown in Figure 2.12 and 2.13, a mesoporous

UF support (roughly 50 µm in thickness resting on a thicker macroporous polymer base) is generally impregnated with one of the monomers via immersion in a water-based solution. This is followed by removing the excess water from the top of the support and immersion in the organic phase containing the secondary reactant. This produces a thin, selective polymer layer to which post-treatment steps are commonly applied. TFC membrane fabrication is shown schematically in Figure 2.11.

101

Figure 2.12. (a) Cross-section of a generic TFC.

Morgan [40] provided a detailed review of the early development of the interfacial polymerization for membrane technology. Cadotte [41] was one of the pioneers to apply

IP for fabrication of membranes for desalination processes. Cadotte discovered that reacting m-phenylene diamine (MPD) and trimesoyl chloride (TMC) on a polymeric support resulted in TFCs with very high rejection (> 99.3%) of sodium chloride from seawater (Figure 2.13). The TFCs were branded as Filmtec FT-30 membranes, and they offered considerably higher water flux compared to asymmetric cellulose acetate membranes which enabled cost-effective seawater desalination. Since then, the process has been a revelation, dominating the membrane desalination market due to its ability to fabricate reproducible and inexpensive high-performance membranes for desalination on an industrial scale. Dow/Filmtec FT-30 membranes were the early industry standard and are still the most commonly studied and well-cited IP membrane type [42–46].

102

Figure 2.13. Scanning electron micrograph of an MPDTMC-based interfacially polymerized TFC.

IP has been used to manufacture a wide range of polymers, including polycarbonates, polyurethanes, polyamides (PA), polyimides, polyesters and polysiloxanes among others

[47]. Polyamides were and remain the dominant materials used in the IP process. This is because polyamides offer intrinsic properties such as exceptional chemical stability, thermal resistance, and mechanical stability [48]. Despite the materials generally exhibiting low free volumes, high flux can be obtained due to the extremely thin nature of active layers formed using IP. Most common examples of polyamide TFC membranes fabricated using IP include numerous small variations of the FT-30 chemistry and fabrication parameters to produce TFCs for brackish water desalination as well as nanofiltration processes. Polyamide TFCs produced via using piperazine (PIP) and TMC have also gained considerable popularity in water treatment processes due to their extremely high water permeance (> 6 LMH/bar) and high rejection for divalent ions including MgSO4, MgCl2 and Na2SO4 [49]. Recently, more diverse polyamide chemistries are being explored

103 especially bio-inspired poly(bio-amide)s which integrate a biologically active component in the IP fabrication process. Examples of such TFCs include those prepared using aquaporin, pepsin and bovine serum albumin by substitution of a conventional IP monomer or via copolymerization [50–52]. A detailed discussion of poly(bio-amide)s is provided in a recent review by Raaijmakers and Benes et al. [47].

Fabrication of polyester-based TFCs is gaining attention in the membrane separation industry. The applications of such TFCs has been explored for gas separation [53], reverse osmosis [54] but primarily nanofiltration applications [55–57]. Recently, interfacially polymerized polyester TFCs were prepared by integrating kinked building blocks which showed potential for both liquid and gas separations [58,59]. Generally, polyesters tend to demonstrate significantly lower separation performance compared to their polyamide counterparts. Furthermore, such materials are prone to hydrolytic degradation at both high and low pH conditions which presents a significant challenge to their commercial relevance. Detailed studies on the impacts of fouling and chemical stability with cleaning solvents of polyester TFCs can help determine the true separation potential for such TFCs on an industrial scale.

Over the last two decades, polyimides have gained significant attention in the membrane industry. The materials have been used traditionally to manufacture high-performance commercial optical materials, plastics, and adhesives due to their exceptional thermal and chemical stabilities [60]. Recently, isotropic polyimides have been studied for a diverse range of gas separations [14,61–66] and have helped redefine conventional permeability/selectivity trade-offs for a wide range of gas pairs including H2/N2, H2/CH4,

O2/N2 and CO2/CH4 [23]. Polyimide TFCs prepared using IP show reasonable promise for

104 vapor and gas separations [67–69], but their limited stability in water and polar aprotic solvents can present a challenge for liquid separations.

Polyurethanes and polyureas have been prepared by the IP process using amines and isocyanate building blocks, but their main applications are limited to fabrication of microcapsules [70]. Similarly, polyamines fabricated using diamines and triazine chlorides have shown potential for nanofiltration applications due to their chemical stability [71,72], but studies have been limited. A detailed list of monomers used in the IP process is listed in the appendices.

2.3.2. Mechanism of Thin-Film Formation

The IP polycondensation reaction occurs at the interface of two immiscible solvents employing a nucleophile and electrophile species dissolved in each of the two solvent phases. Low inter-phase solubility (miscibility) is considered a primary requirement for the

IP process. Early studies on the process [40] revealed that even though initial film formation can be observed in miscible systems, the polymer soon spreads through the entire mixture. This results in the formation of very weak films with poor separation properties.

The electrophilic species used in IP (e.g., TMC) can react with water and hence, are generally dissolved in the organic solvent phase (e.g., n-hexane). Meanwhile, the nucleophilic species (e.g., MPD) is dissolved in a polar solvent (generally water). The two species initially react to form the polymeric chain at the interface, and as the reaction continues, the monomers encounter the polymeric chain at the interface rather than unreacted monomers resulting in a very high molecular weight active layer compared to bulk polymerization [40]. The process can be used to fabricate both linear and network

105 active layers depending on the number of reactive functional groups on the monomers used.

Typically, di-functional acyl chloride (e.g., isophthaloyl chloride, terephthaloyl chloride) are used in the organic phase to prepare linear polymeric chains while acyl chloride with 3 or more functional groups (e.g., trimesoyl chloride) are used for the preparation of network structures. Transport properties of linear polymeric structures highly depend on the chemical nature of the monomers which determine chain-chain interaction and chain packing. Meanwhile, in network polymers, the chain packing and free-volume are determined primarily by the interconnection (crosslinking) of the network structure.

Network polymers fabricated using the IP process display extremely poor solubility in most known polar and organic solvents as well as limited to no crystallinity and are potentially of semi-infinite molecular weight [47].

For the m-phenylene diamine and trimesoyl chloride reaction (Figure 2.14), it has been suggested by numerous studies [18,40,73] that the polyamide active layer (the structure of which is shown in Figure 2.15) is formed by IP on the organic side of the interface. This is because of the high solubility of the diamine in the organic phase compared to that of the acyl chloride in the polar phase. The amine diffusing to the organic phase reacts with the excess acyl chloride to form low molecular weight (MW) polyamide oligomers with acyl terminations. Subsequent diffusion and reaction of the amine with the oligomers causes the increase in polymer MW, which results in precipitation. Both heat and hydrogen chloride evolve during the reaction. Once a polymer layer forms, it hinders the diffusion of the amine into the organic phase reaction zone causing termination [74].

106

Figure 2.14. MPD-TMC reaction at the interface.

Song et al. [75] and Freger et al. [76] used a two-step growth mechanism to describe thin layer formation by IP. The initial phase involves the formation of a thin dense layer at the interface which grows towards the organic phase due to reasons mentioned earlier. The second phase is brought on by a dramatic decrease in amine diffusion rate due to the PA barrier. This decreases the amine transfer to the reaction zone causing a reduction in the reaction rate. This stage is critical for the formation of the network polymer necessary for low MW separations as the membrane densifies and crosslinks. The third stage involves reaction termination due to the thick PA layer limiting diffusion of the polar phase monomer.

107

Figure 2.15. Aromatic polyamide structure via IP reaction between MPD-TMC.

Freger et al. [18] used transmission electron microscopy to study the nanoscale heterogeneity of the MPDTMC polyamide thin-film and identified two distinct morphologies: a negatively charged carboxylic-rich sublayer facing the support (polar) side and a positively charged carboxyl-free layer on the top surface layer (facing the organic side during the reaction). The layers show further differences in thickness and surface roughness, with a central dense nodular polyamide layer with more loose structures on either side. Pacheco et al. [77] followed up with imaging studies and was able to confirm the two distinct layer model predicted by modeling [78] and above mentioned experimental data.

2.3.3. Significant Fabrication Parameters

TFCs produced employing the interfacial polymerization fabrication process are highly dependent on the fabrication conditions. Despite the inherent visual and industrial simplicity of the process (allowing commercialization), a large number of fabrication parameters can be manipulated to achieve desired permselectivity performance [79–84].

Some potential fabrication parameters are illustrated in Figure 2.16. The ability to

108 manipulate performance using fabrication conditions serves as a major benefit of the IP process as a wide range of performances can be obtained without varying active layer chemistry. However, optimization of the membranes can be a tedious process.

MPDTMC chemistry has been widely explored with hundreds, if not thousands, of fabrication parameter combinations tested. Klaysom et al. [85] prepared MPDTMC TFCs on polyacrylonitrile supports rather than hydrophobic polysulfone supports which are conventionally used for interfacially polymerized commercial membranes. They showed that the substitution of the hydrophobic polysulfone support with a relatively hydrophilic polyacrylonitrile support resulted in a smoother top surface of the polyamide active layer along with a decrease in sodium chloride rejection. The effect was similar when the pore size of the polymeric support was increased. They attributed this effect to penetration of the polyamide active layer in the support pores which can significantly affect the overall properties of a TFC as described by Henis and Tripodi [86].

109

Figure 2.16. Fabrication parameters involved in interfacial polymerization.

Ramon et al. [87] studied the effect of the support surface porosity and pore size on the evolution of the polyamide active layer structure in MPDTMC TFCs. They concluded that low surface porosity and large pore sizes resulted in high rejecting membranes. Meanwhile, increased surface porosity and smaller pores caused elevated flux through the TFCs. Zhang et al. [88] observed that low surface porosity resulted in MPDTMC TFCs with lower crosslinking, lower surface roughness and lower thickness of the active layer compared to supports with high surface porosity. Wijmans and Hao [89] emphasized how the support

110 transport resistance can bottleneck flux through a TFC membrane. Ramon et al. [90] confirmed the phenomenon for RO and NF membranes and demonstrated the effect is significant for high flux membranes which could pose a challenge for the fabrication of next-generation high-performance membranes.

Generally, water is used as the polar phase for interfacial polymerization, and very limited studies exist on alternative solvents as well as polar phase temperature. Meanwhile, a fairly large number of polar phase monomers have been explored which are listed in the appendices. For the MPDTMC chemistry, some parameters, e.g., MPD concentration in the polar phase, TMC concentration in the organic phase and IP reaction time have been studied extensively for desalination applications and qualitative relationships have been established. For example, lowering reaction time and TMC concentration results in thinner selective layers with higher water permeance but this is generally is coupled with lower salt rejection [80,83]. Increasing MPD concentration was shown to improve TFC sodium chloride rejection up to ~2 wt/vol% by numerous studies [83,91,92].

Several organic solvents have been investigated including hexane, isoparaffins, cyclohexane, heptane, xylene, and toluene as the organic phase for IP [79,93]. It was observed that changing the organic solvent can have a significant impact on both the structure and performance of RO membranes. Attempts were made to relate solvent properties such as viscosity, solubility and boiling point to the observed differences but only qualitative conclusions were made [79,93]. Kwak et al. used dimethyl sulfoxide

(DMSO) and observed a similar effect [45].

111

Ghosh and Hoek et al. [79] and Yu et al. [94] were two of the few investigators to study the impact of varying organic phase temperature on separation properties of MPDTMC RO

TFCs. They observed that as the temperature was increased from 8 to 38 °C, sodium chloride rejection decreased from ~98% to 95% while membrane flux increased by 50%.

Reaction time for interfacial polymerization has also been studied for liquid separation studies, but no clear conclusions could be reached [83,95]. Meanwhile, post-fabrication curing was shown to improve both membrane permeance as well as rejection for a number of salts until an optimum temperature depending primarily on the support material characteristics due to the impeccable thermal stability of the polyamide active layers

[79,96].

Louie et al. studied the effect of alcohol activation (immersion in or flushing with an organic solvent) on commercial RO membranes and demonstrated the ability to improve both permeance and rejection with the correct choice of solvents [97]. They attributed the improved performance on the removal of loose polyamide from the TFC structure.

Solomon et al. [98] confirmed this effect by using more aggressive polar aprotic solvents.

A large number of additives have been studied for IP, especially for the MPDTMC chemistry. Kong et al. [82,99] used acetone as a co-solvent and observed significant improvements in flux with little loss of salt rejection. They explained it by the formation of a thinner, denser active layer due to increased miscibility of the two phases. Klaysom et al. [85] showed that integration of triethyl amine (TEA) and sodium dodecyl sulfate (SDS) in the IP fabrication process could improve both water permeance and salt rejection of RO membranes.

112

Similarly, diverse fillers have been explored for integration in the IP fabrication process.

Duan et al. [100] added porous ZIF-8 fillers to the MPDTMC chemistry and systematically evaluated the effect of increasing filler concentration. The study showed improved water permeance by ~150% with no loss in sodium chloride rejection showing the potential of improving permselectivity by the addition of fillers. Likewise, further studies were conducted using multiple fillers including carbon nanotubes and silicates, but no conclusions could be made on mechanisms of improved performance [101–104].

Karan et al. [95,105] showed that by carefully controlling the reaction rate, major changes in active layer structure could be induced. They were able to produce extremely thin smooth polyamide layers with reasonably high permeance and salt rejection performance.

Ghosh and Hoek et al. [79] performed a detailed study on numerous parameters involved in MPDTMC thin-film fabrication and successfully demonstrated correlations of performance with physical characteristics of the reaction system. Combination of such experimental and modeling studies have laid the framework for fabrication of the next- generation of thin-film composites for separation applications.

2.3.4. Interfacially polymerized TFCs for gas separations

Despite the success of interfacially polymerized TFC membranes for liquid separations, particularly desalination, the application of the technology for fabricating TFCs for gas separation applications has been very limited. A primary reason for this is the presence of defects in the dry state in fully aromatic TFC polyamide membranes. A defect or pinhole can be defined for dense films/membranes as a region in the membrane area not covered

113 entirely with the polymeric active layer. For gas separations and desalination applications, a defect can arbitrarily be defined as a pore with a diameter of 1 nm or larger.

Interfacially polymerized commercial RO polyamide TFCs were first studied for gas separation performance by Louie et al. who observed that MPDTMC TFCs exhibited

Knudsen diffusion-based gas selectivities for a number of gas pairs, e.g., He/CO2 and H2/N2 gas pairs showed selectivity of ~4. The data implied that commercial RO polyamide TFCs contain defects in the dry-state with little to no useful gas separation properties [106].

“Caulking” layers, as discussed before, to plug defects in thin active layers were a revolutionary breakthrough for membrane science to enable defect-free-like properties in defective TFC membranes [86]. Louie et al. [106] used this approach and coated

Hydranautics SWC4 commercial RO membranes with a polyether-polyamide block copolymer (PEBAX-1657). The coating significantly reduced gas permeance but improved gas selectivities, e.g., He/CO2 improved to ~13, and H2/N2 selectivity increased to ~11.

This was followed by studies from Albo et al. [107] who demonstrated a post-fabrication solvent treatment and high-temperature curing method to improve gas separation properties of MPDTMC-based TFCs. Their method allowed for He/CO2 selectivity of ~7 while H2/N2 selectivity soared to ~100.

Although efforts to fabricate in-situ fully aromatic defect-free TFC polyamide membranes have shown limited success, some reports exist for gas separating interfacially polymerized

TFCs when one or both aromatic monomers were substituted by linear counterparts.

Sridhar et al. [108] prepared gas-tight TFCs using isophthaloyl chloride and MPD employing interfacial polymerization. They demonstrated reasonable gas separation properties for natural gas separations with CO2 permeance of ~17 GPU with CO2/CH4

114

selectivity of ~13 as well as for air separations with O2 permeance of ~6 GPU with O2/N2 selectivity of ~5. Zhao et al. [109] fabricated IP TFCs for CO2 separations using the linear triethylene tetramine with TMC and were able to obtain good CO2/CH4 selectivity of ~40 with CO2 permeance of ~10 GPU at low-pressure testing. Yu et al. [110] prepared TFCs using 3,3-diamino-N-methyldipropylamine and TMC and demonstrated CO2/N2 selectivity of ~80 with CO2 permeance ranging between 60 to 120 GPU. Yuan et al. [111] showed the ability to bypass the Robeson 2008 upper bound for CO2/N2 separations using N- methyldiethanolamine and TMC TFCs and further validated the ability to control membrane permeance and selectivity using fabrication parameters.

Another interesting observation made by Louie et al. [106] was that coating SWC4 with

PEBAX resulted in a 40% decrease in water permeance for the coated TFCs which disagreed with predictions made by the “series resistance model” proposed by Henis and

Tripodi [86]. This implied that the back-calculated gas transport properties of the polyamide active layer might not accurately explain the pristine properties of the polyamide active layer. Furthermore since liquid-separation properties of fully aromatic polyamide TFCs are significantly superior to those prepared using semi-aromatic or linear counterparts [112], the ability to manufacture defect-free fully aromatic polyamide active layers might elucidate structure-function-performance relationships in such membranes.

This can potentially enable the rational fabrication of novel materials for the fabrication of high-performance interfacially polymerized TFCs as well as improve understanding of gas and liquid transport for currently available commercial interfacially polymerized membranes for gas- and liquid-separations.

115

2.4. References

[1] R.W. Baker, Membrane Technology and Applications, 3rd ed, Wiley, Chichester, UK., 2012.

[2] M. Mulder, Basic Principles of Membrane Technology, Springer Netherlands, Dordrecht, 1996.

[3] R. Swaidan, Intrinsically microporous polymer membranes for high performance gas separation, Ph.D. thesis, King Abdullah University of Science and Technology (KAUST), 2014.

[4] E. Drioli, L. Giorno, Encyclopedia of Membranes, Springer Berlin Heidelberg, Berlin, Heidelberg, 2016.

[5] P. W. Atkins, Physical Chemistry, 4th Edition, Wiley-Blackwell, Oxford, 1990.

[6] E. Mason, A. Malinauskas, Gas Transport in Porous Media: the dusty-gas model, Vol 17, Elsevier Science Ltd., Amsterdam, 1983.

[7] T. Graham, XVIII. On the absorption and dialytic separation of gases by colloid septa, Philos. Trans. R. Soc. London 156 (2006) 399–439.

[8] T. Graham, XI. On the law of the diffusion of gases, Trans. R. Soc. Edinburgh 12 (1834) 222–258.

[9] J.G. Wijmans, R.W. Baker, The solution-diffusion model: a review, J. Membr. Sci. 107 (1995) 1–21.

[10] G.S. Park, Transport principles—solution, diffusion and permeation in polymer membranes, in: Synth. Membr. Sci. Eng. Appl., Springer Netherlands, Dordrecht, 2012: pp. 57–107.

[11] S. Matteucci, Y. Yampolskii, B.D. Freeman, I. Pinnau, Transport of gases and vapors in glassy and rubbery polymers, in: Y. Yampolskii, I. Pinnau, B. Freeman (Eds.), Materials Science of Membranes for Gas and Vapor Separation, John Wiley & Sons, Ltd, Chichester, UK, 2006: pp. 1–47.

[12] R.M. Felder, G.S. Huvard, Permeation, diffusion, and sorption of gases and vapors, in: Methods Exp. Phys., 1980: pp. 315–377.

[13] S.M. Allen, M. Fujii, V. Stannett, H.B. Hopfenberg, J.L. Williams, The barrier properties of polyacrylonitrile, J. Membr. Sci. 2 (1977) 153–163.

[14] R. Swaidan, M. Al-Saeedi, B. Ghanem, E. Litwiller, I. Pinnau, Rational design of intrinsically ultramicroporous polyimides containing bridgehead-substituted triptycene for highly selective and permeable gas separation membranes,

116

Macromolecules 47 (2014) 5104–5114.

[15] V.I. Bondar, B.D. Freeman, I. Pinnau, Gas transport properties of poly(ether-b- amide) segmented block copolymers, J. Polym. Sci. Part B Polym. Phys. 38 (2000) 2051–2062.

[16] V.I. Bonder, B.D. Freeman, Y.P. Yampolskii, Sorption of gases and vapors in an amorphous glassy perfluorodioxole copolymer, Macromolecules 32 (1999) 6163– 6171.

[17] S.N. Dhoot, B.D. Freeman, M.E. Stewart, A.J. Hill, Sorption and transport of linear alkane hydrocarbons in biaxially oriented polyethylene terephthalate, J. Polym. Sci. Part B Polym. Phys. 39 (2001) 1160–1172.

[18] V. Freger, Nanoscale heterogeneity of polyamide membranes formed by interfacial polymerization, Langmuir 19 (2003) 4791–4797.

[19] F. Pacheco, R. Sougrat, M. Reinhard, J.O. Leckie, I. Pinnau, 3D visualization of the internal nanostructure of polyamide thin films in RO membranes, J. Membr. Sci. 501 (2016) 33–44.

[20] L.M. Robeson, Correlation of separation factor versus permeability for polymeric membranes, J. Membr. Sci. 62 (1991) 165–185.

[21] A.Y. Alentiev, Y.. Yampolskii, Free volume model and tradeoff relations of gas permeability and selectivity in glassy polymers, J. Membr. Sci. 165 (2000) 201–216.

[22] B.D. Freeman, Basis of permeability/selectivity tradeoff relations in polymeric gas separation membranes, Macromolecules 32 (1999) 375–380.

[23] R. Swaidan, B. Ghanem, I. Pinnau, Fine-tuned intrinsically ultramicroporous polymers redefine the permeability/selectivity upper bounds of membrane-based air and hydrogen separations, ACS Macro Lett. 4 (2015) 947–951.

[24] L.M. Robeson, The upper bound revisited, J. Membr. Sci. 320 (2008) 390–400.

[25] T.C. Merkel, V.I. Bondar, K. Nagai, B.D. Freeman, I. Pinnau, Gas sorption, diffusion, and permeation in poly(dimethylsiloxane), J. Polym. Sci. Part B Polym. Phys. 38 (2000) 415–434.

[26] A. Singh, B.D. Freeman, I. Pinnau, Pure and mixed gas acetone/nitrogen permeation properties of polydimethylsiloxane [PDMS], J. Polym. Sci. Part B Polym. Phys. 36 (1998) 289–301.

[27] V.I. Bondar, Y. Kamiya, Y.P. Yampol’skii, On pressure dependence of the parameters of the dual-mode sorption model, J. Polym. Sci. Part B Polym. Phys. 34 (1996) 369–378.

117

[28] A.R. Berens, The solubility of vinyl chloride in poly(vinyl chloride), Die Angew. Makromol. Chemie 47 (1975) 97–110.

[29] D.R. Paul, Gas sorption and transport in glassy polymers, Berichte der Bunsengesellschaft/Physical Chem. Chem. Phys. 83 (1979) 294–302.

[30] 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.

[31] H. Lin, B.D. Freeman, Gas and vapor solubility in cross-linked polyethylene glycol diacrylate, Macromolecules 38 (2005) 8394–8407.

[32] A.A. Armstrong, V. Stannett, Temperature effects during the sorption and desorption of water vapor in high polymers. I. Fibers with particular reference to wool, Die Makromol. Chemie 90 (1966) 145–160.

[33] W.J. Koros, D.R. Paul, A.A. Rocha, Carbon dioxide sorption and transport in polycarbonate, J. Polym. Sci. Polym. Phys. Ed. 14 (1976) 687–702.

[34] 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.

[35] V. Stannett, The transport of gases in synthetic polymeric membranes — an historic perspective, J. Membr. Sci. 3 (1978) 97–115.

[36] R.T. Chern, C.N. Provan, Gas-induced plasticization and the permselectivity of poly(tetrabromophenolphthalein terephthalate) to a mixture of carbon dioxide and methane, Macromolecules 24 (1991) 2203–2207.

[37] G.M. Geise, H.B. Park, A.C. Sagle, B.D. Freeman, J.E. McGrath, Water permeability and water/salt selectivity tradeoff in polymers for desalination, J. Membr. Sci. 369 (2011) 130–138.

[38] W.M. Jackson, J.S. Drury, Miscibility of organic solvent pairs, Ind. Eng. Chem. 51 (1959) 1491–1493.

[39] B. Yuan, C. Jiang, P. Li, H. Sun, P. Li, T. Yuan, H. Sun, Q.J. Niu, Ultrathin polyamide membrane with decreased porosity designed for outstanding water- softening performance and superior antifouling properties, ACS Appl. Mater. Interfaces 10 (2018) 43057–43067.

[40] P.W. Morgan, Condensation polymers: By interfacial and solution methods, Interscience Publishers, 1965.

[41] J.E. Cadotte, R.J. Petersen, R.E. Larson, E.E. Erickson, A new thin-film composite seawater reverse osmosis membrane, Desalination 32 (1980) 25–31.

118

[42] P. Lipp, R. Gimbel, F.H. Frimmel, Parameters influencing the rejection properties of FT30 membranes, J. Membr. Sci. 95 (1994) 185–197.

[43] M.J. Kotelyanskii, N.J. Wagner, M.E. Paulaitis, Atomistic simulation of water and salt transport in the reverse osmosis membrane FT-30, J. Membr. Sci. 139 (1998) 1–16.

[44] R.E. Larson, J.E. Cadotte, R.J. Petersen, The FT-30 seawater reverse osmosis membrane element test results, Desalination 38 (1981) 473–483.

[45] S.Y. Kwak, S.G. Jung, S.H. Kim, Structure-motion-performance relationship of flux-enhanced reverse osmosis (RO) membranes composed of aromatic polyamide thin films, Environ. Sci. Technol. 35 (2001) 4334–4340.

[46] J.E. Cadotte, R.J. Petersen, Synthetic Membranes, Vol, 153, ACS Symposium Series, American Chemical Society, Washington, D.C., 1981.

[47] M.J.T. Raaijmakers, N.E. Benes, Current trends in interfacial polymerization chemistry, Prog. Polym. Sci. 63 (2016) 86–142.

[48] J.M. García, F.C. García, F. Serna, J.L. de la Peña, High-performance aromatic polyamides, Prog. Polym. Sci. 35 (2010) 623–686.

[49] L. Li, S. Zhang, X. Zhang, Preparation and characterization of poly(piperazineamide) composite nanofiltration membrane by interfacial polymerization of 3,3′,5,5′-biphenyl tetraacyl chloride and piperazine, J. Membr. Sci. 335 (2009) 133–139.

[50] M.J.T. Raaijmakers, T. Schmidt, M. Barth, M. Tutus, N.E. Benes, M. Wessling, Enzymatically active ultrathin pepsin membranes, Angew. Chemie - Int. Ed. 54 (2015) 5910–5914.

[51] Y. Zhao, C. Qiu, X. Li, A. Vararattanavech, W. Shen, J. Torres, C. Hélix-Nielsen, R. Wang, X. Hu, A.G. Fane, C.Y. Tang, Synthesis of robust and high-performance aquaporin-based biomimetic membranes by interfacial polymerization-membrane preparation and RO performance characterization, J. Membr. Sci. 423–424 (2012) 422–428.

[52] J. Zhao, Y. Zhang, Y. Su, J. Liu, X. Zhao, J. Peng, Z. Jiang, Cross-linked bovine serum albumin composite membranes prepared by interfacial polymerization with stimuli-response properties, J. Membr. Sci. 445 (2013) 1–7.

[53] M.I. Loría-Bastarrachea, M. Aguilar-Vega, Synthesis and gas transport properties of polyesters and a copolyester obtained from 4,4′-(9-fluorenylidene) bisphenol and 4,4′-(1-phenylethylidene) bisphenol, Ind. Eng. Chem. Res. 49 (2010) 12060–12066.

[54] S.Y. Kwak, M.O. Yeom, I.J. Roh, D.Y. Kim, J.J. Kim, Correlations of chemical structure, atomic force microscopy (AFM) morphology, and reverse osmosis (RO)

119

characteristics in aromatic polyester high-flux RO membranes, J. Membr. Sci. 132 (1997) 183–191.

[55] X.-Z. Wei, L.-P. Zhu, H.-Y. Deng, Y.-Y. Xu, B.-K. Zhu, Z.-M. Huang, New type of nanofiltration membrane based on crosslinked hyperbranched polymers, J. Membr. Sci. 323 (2008) 278–287.

[56] M.N.A. Seman, M. Khayet, N. Hilal, Nanofiltration thin-film composite polyester polyethersulfone-based membranes prepared by interfacial polymerization, J. Membr. Sci. 348 (2010) 109–116.

[57] X. Wei, X. Kong, J. Yang, G. Zhang, J. Chen, J. Wang, Structure influence of hyperbranched polyester on structure and properties of synthesized nanofiltration membranes, J. Membr. Sci. 440 (2013) 67–76.

[58] S. Yu, S. Li, Y. Liu, S. Cui, X. Shen, High-performance microporous polymer membranes prepared by interfacial polymerization for gas separation, J. Membr. Sci. 573 (2018) 425–438.

[59] M.F. Jimenez-Solomon, Q. Song, K.E. Jelfs, M. Munoz-Ibanez, A.G. Livingston, Polymer nanofilms with enhanced microporosity by interfacial polymerization, Nat. Mater. 15 (2016) 760–767.

[60] S. Hong, I.C. Kim, T. Tak, Y.N. Kwon, Interfacially synthesized chlorine-resistant polyimide thin film composite (TFC) reverse osmosis (RO) membranes, Desalination 309 (2013) 18–26.

[61] M.A. Abdulhamid, H.W.H. Lai, Y. Wang, Z. Jin, Y.C. Teo, X. Ma, I. Pinnau, Y. Xia, Microporous polyimides from ladder diamines synthesized by facile catalytic arene-norbornene annulation as high-performance membranes for gas separation, Chem. Mater. (2019) acs.chemmater.8b05359.

[62] X. Ma, M. Abdulhamid, X. Miao, I. Pinnau, Facile synthesis of a hydroxyl- functionalized Tröger’s base diamine: A new building block for high-performance polyimide gas separation membranes, Macromolecules 50 (2017) 9569–9576.

[63] N. Alaslai, X. Ma, B. Ghanem, Y. Wang, F. Alghunaimi, I. Pinnau, Synthesis and characterization of a novel microporous dihydroxyl-functionalized triptycene- diamine-based polyimide for natural gas membrane separation, Macromol. Rapid Commun. 38 (2017) 1700303.

[64] B.S. Ghanem, N.B. McKeown, P.M. Budd, J.D. Selbie, D. Fritsch, High- Performance Membranes from Polyimides with Intrinsic Microporosity, Adv. Mater. 20 (2008) 2766–2771.

[65] B.S. Ghanem, R. Swaidan, E. Litwiller, I. Pinnau, Ultra-microporous triptycene- based polyimide membranes for high-performance gas separation, Adv. Mater. 26 (2014) 3688–3692.

120

[66] R. Swaidan, B. Ghanem, M. Al-Saeedi, E. Litwiller, I. Pinnau, Role of intrachain rigidity in the plasticization of intrinsically microporous triptycene-based polyimide membranes in mixed-gas CO2/CH4 separations, Macromolecules 47 (2014) 7453– 7462.

[67] M.J.T. Raaijmakers, M. Wessling, A. Nijmeijer, N.E. Benes, Hybrid polyhedral oligomeric silsesquioxanes–imides with tailored intercage spacing for sieving of hot gases, Chem. Mater. 26 (2014) 3660–3664.

[68] M.J.T. Raaijmakers, M.A. Hempenius, P.M. Schön, G.J. Vancso, A. Nijmeijer, M. Wessling, N.E. Benes, Sieving of hot gases by hyper-cross-linked nanoscale-hybrid membranes, J. Am. Chem. Soc. 136 (2014) 330–335.

[69] J.-H. Kim, K.-H. Lee, S.Y. Kim, Pervaporation separation of water from ethanol through polyimide composite membranes, J. Membr. Sci. 169 (2000) 81–93.

[70] A. Khosravi, M. Sadeghi, Separation performance of poly(urethane–urea) membranes in the separation of C2 and C3 hydrocarbons from methane, J. Membr. Sci. 434 (2013) 171–183.

[71] S.K. Das, P. Manchanda, K.-V. Peinemann, Solvent-resistant triazine-piperazine linked porous covalent organic polymer thin-film nanofiltration membrane, Sep. Purif. Technol. (2018).

[72] K.P. Lee, J. Zheng, G. Bargeman, A.J.B. Kemperman, N.E. Benes, PH stable thin film composite polyamine nanofiltration membranes by interfacial polymerisation, J. Membr. Sci. 478 (2015) 75–84.

[73] J.E. Cadotte, R.J. Petersen, Thin-film composite reverse-osmosis membranes: Origin, development, and recent advances, ACS Symp. Ser. 153 (1981) 305–326.

[74] J. Duan, E. Litwiller, I. Pinnau, Solution-diffusion with defects model for pressure- assisted forward osmosis, J. Membr. Sci. 470 (2014) 323–333.

[75] M.A. Kuehne, R.Q. Song, N.N. Li, R.J. Petersen, Flux enhancement in TFC RO membranes, Environ. Prog. 20 (2001) 23–26.

[76] V. Freger, Kinetics of film formation by interfacial polycondensation, Langmuir 21 (2005) 1884–1894.

[77] F.A. Pacheco, I. Pinnau, M. Reinhard, J.O. Leckie, Characterization of isolated polyamide thin films of RO and NF membranes using novel TEM techniques, J. Membr. Sci. 358 (2010) 51–59.

[78] V. Freger, S. Srebnik, Mathematical model of charge and density distributions in interfacial polymerization of thin films, J. Appl. Polym. Sci. 88 (2003) 1162–1169.

[79] A.K. Ghosh, B.H. Jeong, X. Huang, E.M.V. Hoek, Impacts of reaction and curing

121

conditions on polyamide composite reverse osmosis membrane properties, J. Membr. Sci. 311 (2008) 34–45.

[80] A.L. Ahmad, B.S. Ooi, Properties-performance of thin film composites membrane: Study on trimesoyl chloride content and polymerization time, J. Membr. Sci. 255 (2005) 67–77.

[81] S.H. Maruf, A.R. Greenberg, Y. Ding, Influence of substrate processing and interfacial polymerization conditions on the surface topography and permselective properties of surface-patterned thin-film composite membranes, J. Membr. Sci. 512 (2016) 50–60.

[82] C. Kong, T. Shintani, T. Kamada, V. Freger, T. Tsuru, Co-solvent-mediated synthesis of thin polyamide membranes, J. Membr. Sci. 384 (2011) 10–16.

[83] A.M. El-aassar, Polyamide thin film composite membranes using interfacial polymerization : synthesis , characterization and reverse osmosis performance for water desalination, Aust. J. Basic Appl. Sci. 6 (2012) 382–391.

[84] D.J. Mohan, L. Kullová, A study on the relationship between preparation condition and properties/performance of polyamide TFC membrane by IR, DSC, TGA, and SEM techniques, Desalin. Water Treat. 51 (2013) 586–596.

[85] C. Klaysom, S. Hermans, A. Gahlaut, S. Van Craenenbroeck, I.F.J. Vankelecom, Polyamide/Polyacrylonitrile (PA/PAN) thin film composite osmosis membranes: Film optimization, characterization and performance evaluation, J. Membr. Sci. 445 (2013) 25–33.

[86] J.M.S. Henis, M.K. Tripodi, Composite hollow fiber membranes for gas separation: the resistance model approach, J. Membr. Sci. 8 (1981) 233–246.

[87] A.K. Ghosh, E.M. V. Hoek, Impacts of support membrane structure and chemistry on polyamide-polysulfone interfacial composite membranes, J. Membr. Sci. 336 (2009) 140–148.

[88] Q. Zhang, Z. Zhang, L. Dai, H. Wang, S. Li, S. Zhang, Novel insights into the interplay between support and active layer in the thin film composite polyamide membranes, J. Membr. Sci. 537 (2017) 372–383.

[89] J.G. Wijmans, P. Hao, Influence of the porous support on diffusion in composite membranes, J. Membr. Sci. 494 (2015) 78–85.

[90] G.Z. Ramon, M.C.Y. Wong, E.M. V. Hoek, Transport through composite membrane, Part 1: Is there an optimal support membrane?, J. Membr. Sci. 415–416 (2012) 298–305.

[91] I.J. Roh, A.R. Greenberg, V.P. Khare, Synthesis and characterization of interfacially polymerized polyamide thin films, Desalination 191 (2006) 279–290.

122

[92] S.-J. Park, W.-G. Ahn, W. Choi, S.-H. Park, J.S. Lee, H.W. Jung, J.-H. Lee, A facile and scalable fabrication method of thin film composite reverse osmosis membranes: Dual-layer slot coating, J. Mater. Chem. A (2017).

[93] S.-J. Park, S.J. Kwon, H.-E. Kwon, M.G. Shin, S.-H. Park, H. Park, Y.-I. Park, S.- E. Nam, J.-H. Lee, Aromatic solvent-assisted interfacial polymerization to prepare high performance thin film composite reverse osmosis membranes based on hydrophilic supports, Polymer 144 (2018) 159–167.

[94] S. Yu, M. Liu, X. Liu, C. Gao, Performance enhancement in interfacially synthesized thin-film composite polyamide-urethane reverse osmosis membrane for seawater desalination, J. Membr. Sci. 342 (2009) 313–320.

[95] S. Karan, Z. Jiang, A.G. Livingston, Sub-10 nm polyamide nanofilms with ultrafast solvent transport for molecular separation, Science 348 (2015) 1347–1351.

[96] D. Wu, Y. Huang, S. Yu, D. Lawless, X. Feng, Thin film composite nanofiltration membranes assembled layer-by-layer via interfacial polymerization from polyethylenimine and trimesoyl chloride, J. Membr. Sci. 472 (2014) 141–153.

[97] J.S. Louie, I. Pinnau, M. Reinhard, Effects of surface coating process conditions on the water permeation and salt rejection properties of composite polyamide reverse osmosis membranes, J. Membr. Sci. 367 (2011) 249–255.

[98] M.F. Jimenez Solomon, Y. Bhole, A.G. Livingston, High flux hydrophobic membranes for organic solvent nanofiltration (OSN)-Interfacial polymerization, surface modification and solvent activation, J. Membr. Sci. 434 (2013) 193–203.

[99] C. Kong, M. Kanezashi, T. Yamomoto, T. Shintani, T. Tsuru, Controlled synthesis of high performance polyamide membrane with thin dense layer for water desalination, J. Membr. Sci. 362 (2010) 76–80.

[100] J. Duan, Y. Pan, F. Pacheco, E. Litwiller, Z. Lai, I. Pinnau, High-performance polyamide thin-film-nanocomposite reverse osmosis membranes containing hydrophobic zeolitic imidazolate framework-8, J. Membr. Sci. 476 (2015) 303–310.

[101] H.D. Lee, H.W. Kim, Y.H. Cho, H.B. Park, Experimental evidence of rapid water transport through carbon nanotubes embedded in polymeric desalination membranes, Small 10 (2014) 2653–2660.

[102] L. Wang, M. Fang, J. Liu, J. He, J. Li, J. Lei, Layer-by-layer fabrication of high- performance polyamide/ZIF-8 nanocomposite mmbrane for nanofiltration applications, ACS Appl. Mater. Interfaces 7 (2015) 24082–24093.

[103] M.L. Lind, A.K. Ghosh, A. Jawor, X. Huang, W. Hou, Y. Yang, E.M. V. Hoek, Influence of zeolite crystal size on zeolite-polyamide thin film nanocomposite membranes, Langmuir 25 (2009) 10139–10145.

123

[104] H. Wu, B. Tang, P. Wu, Optimizing polyamide thin film composite membrane covalently bonded with modified mesoporous silica nanoparticles, J. Membr. Sci. 428 (2013) 341–348.

[105] Z. Jiang, S. Karan, A.G. Livingston, Water transport through ultrathin polyamide nanofilms used for reverse osmosis, Adv. Mater. 30 (2018) 1870107.

[106] J.S. Louie, I. Pinnau, M. Reinhard, Gas and liquid permeation properties of modified interfacial composite reverse osmosis membranes, J. Membr. Sci. 325 (2008) 793– 800.

[107] J. Albo, J. Wang, T. Tsuru, Gas transport properties of interfacially polymerized polyamide composite membranes under different pre-treatments and temperatures, J. Membr. Sci. 449 (2014) 109–118.

[108] S. Sridhar, B. Smitha, S. Mayor, B. Prathab, T.M. Aminabhavi, Gas permeation properties of polyamide membrane prepared by interfacial polymerization, J. Mater. Sci. 42 (2007) 9392–9401.

[109] J. Zhao, Z. Wang, J. Wang, S. Wang, Influence of heat-treatment on CO2 separation performance of novel fixed carrier composite membranes prepared by interfacial polymerization, J. Membr. Sci. 283 (2006) 346–356.

[110] X. Yu, Z. Wang, Z. Wei, S. Yuan, J. Zhao, J. Wang, S. Wang, Novel tertiary amino containing thin film composite membranes prepared by interfacial polymerization for CO2 capture, J. Membr. Sci. 362 (2010) 265–278.

[111] F. Yuan, Z. Wang, S. Li, J. Wang, S. Wang, Formation-structure-performance correlation of thin film composite membranes prepared by interfacial polymerization for gas separation, J. Membr. Sci. 421–422 (2012) 327–341.

[112] I.J. Roh, V.P. Khare, Investigation of the specific role of chemical structure on the material and permeation properties of ultrathin aromatic polyamides, J. Mater. Chem. 12 (2002) 2334–2338.

124

Chapter 3. Materials and Methods

3.1. Materials

Piperazine (PIP), 99% purity, m-phenylenediamine (MPD), 99% purity, p- phenylenediamine (PPD), 99% purity, and trimesoyl chloride (TMC), 98% purity, were purchased from Aldrich. Polysulfone (PS) and polyacrylonitrile (PAN) ultrafiltration supports were provided by Sepro Membranes Inc. (Carlsbad, CA, USA). The support was composed of a 50-µm-thick polysulfone or polyacrylonitrile membrane resting on a thick

(100 µm) macroporous polyester layer. Isoparaffin G (Isopar®) was obtained from

ExxonMobil and stored with 4 Å molecular sieves to prevent dissolution of atmospheric moisture. Toluene, > 99.9%, was obtained from Honeywell. Isopropanol, 99.5+% ACS reagent, methanol, 99%, and ethanol, 99%, were purchased from Sigma Aldrich. Deionised water (DIW) was obtained from a Millipore Advantage A10 system. Commercial TFC RO membranes were purchased from different RO membrane manufacturers. Boric acid and sodium bicarbonate (Na2CO3) were purchased from Sigma.

Test gases i.e. helium, hydrogen, oxygen, nitrogen, methane and carbon dioxide were obtained from Specialty Gas Center (SGC), with claimed purities > 99.99%.

NaCl, > 99.5%, MgSO4, > 99.5%, Na2SO4, 99.5%, and MgCl2, > 99.9%, were obtained from Fisher. Dyes Brilliant Blue R (826 g mol-1, absorption ~588 nm, negatively charged) and Sudan Orange G (214 g mol-1, absorption ~388 nm, > 97% HPLC) were purchased from Sigma Aldrich. Anthranilic acid, aluminum chloride, m-xylene, isopentyl nitrite and thionyl chloride were obtained from Aldrich and used as received. 1,3,6,8-

125 tetramethylanthracene (TMA) was prepared according to a previously reported procedure

[1].

3.1.1. Synthesis of Triptycene-1,3,6,8- Tetraacetyl Chloride (TripTaC)

The TripTaC monomer was kindly prepared by Dr. Bader Ghanem from the Functional

Polymer Membranes Group, King Abdullah University of Science and Technology.

The TripTac monomer was synthesized by a simple four-step synthetic procedure starting from m-xylene. The key intermediate 1,3,6,8-tetramethylanthracene was obtained from the

Friedel-Crafts alkylation reaction of m-xylene with dichloromethane in presence of aluminum chloride using the procedure reported by Ellison et al. [2]. The Diels-Alder reaction of the tetramethylanthracene with the diazonium salt of 2-aminobenzoic acid yielded 1,3,6,8-tetramethyltriptycene. Oxidation of triptycene-1,3,6,8- tetracarboxylic acid using potassium permanganate and followed by reaction with thionyl chloride yielded the tetraacetyl chloride product. The purity and molecular structures of the products were confirmed by standard characterization techniques, 1H- and 13C NMR, FTIR. This new protocol used for the preparation of TripTac monomer has advantages of using cheaper starting materials and simpler synthetic chemistry.

1H NMR and 13C NMR spectra for all monomers were recorded with a Bruker AVANCE-

III spectrometer at a frequency of 400 or 500 MHz in deuterated chloroform (CDCl3) or dimethylsulfoxide (DMSO-d6). Chemical shifts () are reported in parts per million (ppm) and referenced to tetramethylsilane. Fourier transform infrared (FTIR) measurements were performed using a Varian 670-IR FTIR spectrometer.

126

3.1.2. Synthesis of 1,3,6,8- Tetramethylanthracene (TMA).

1,3,6,8-Tetramethylanthracene was prepared according to the previously reported procedure by Ellison and Ghanem et al. [2,3].

3.1.3. Synthesis of 1,3,6,8-Tetramethyltriptycene (TMT).

1,3,6,8-tetramethyltriptycene (TMT) was prepared as white powder according to the previously reported procedure by Ghanem et al. [3].

3.1.4. Synthesis of Triptycene-1,3,6,8- tetracarboxylic Acid (TTC)

To a stirred reflux mixture of 1,3,6,8-tetramethyltriptycene (1.27 g, 4.09 mmole) in 64 ml pyridine and 38 ml water was added 23.37 g of KMnO4 (9 eq. per each CH3 group) over a period of 36 h. After cooling, MnO2 was filtered of and washed with 150 ml KOH (1%) and the filtrate was concentrated to one third of its volume and acidified with 3 M HCl.

The white precipitate was collected, washed with cold water and dried at 60 °C under vacuum to give the tetracarboxylic acid in 82% yield, which was used in the following reaction without further purification. 1H NMR (400 MHz, DMSO-d6, ): 6.16 (s, 1H), 7.08

(m, 2H), 7.47 (m, 1H), 7.54 (m, 1H), 7.98 (s, 1H), 8.12 (d, 2H), 8.22 (d, 2H). FTIR

(powder, ν, cm-1): 2718-3679 (carboxylic acid proton stretch), 3045 (aromatic C-H stretch),

2947 (aliphatic C-H stretch), 1690 (C=O stretch), 1234 (C-O-C stretch).

3.1.5. Synthesis of Triptycene-1,3,6,8- Tetraacetyl Chloride (TripTaC)

Thionyl chloride (20 ml) was added dropwise to triptycene-1,3,6,8- tetracarboxylic acid

(2.49 g) at 0 °C. after adding 3 drops of DMF, the mixture was stirred for 30 minutes and the temperature was raised to 60 °C under continuous stirring for 4 h. Excess thionyl chloride was stripped off by vacuum distillation and then anhydrous diethyl ether was

127 added. The resulting powder was collected and recrystallized from chloroform/petroleum.

1 ether to give the desired product a pale yellow powder. H NMR (400 MHz, CDCl3, ):

5.84 (s, 1H), 7.16 (m, 2H), 7.52 (m, 1H), 7.61 (m, 1H), 7.66 (s, 1H), 8.37 (d, 2H), 8.66 (d,

13 2H). C NMR (100 MHz, CDCl3, ): 46.6, 52.2, 124.4, 125.1, 126.0, 126.1, 127.6, 128.1,

128.2, 128.3, 142.7, 144.9, 147.7, 149.4, 166.3, 167.2. FTIR (powder, ν, cm-1): 3026

(aromatic C-H stretch), 2922 (aliphatic C-H stretch), a strong band at 1738 (C=O, stretch of acyl chloride).

3.2. Membrane Fabrication

Membranes were fabricated varying three parameters: i) reaction time; ii) organic monomer concentration and; iii) organic phase temperature. Porous support layers (11.5 x

15.5 cm) were immersed in tap water for 24 hours followed by immersion in aqueous monomer dissolved in distilled water. The support was then passed through a rubber roller to remove any excess droplets on the surface, and fixed in a sealed Teflon frame. The organic solvent was heated to the desired temperature and TMC was dissolved in a specific concentration. The organic monomer solution was then poured on the support surface, initiating the reaction. After the specified reaction time, the excess solution was poured off.

The membrane was immediately washed in the frame three times with 30 ml of the organic solvent used for the reaction, followed by isopropanol. Finally, it was dried at room temperature for 48 h and stored in a desiccator until testing.

The membrane designation in Chapter 3 and 4 is defined by: xs (reaction time between organic TMC and aqueous diamine phases in seconds); yTMC (TMC concentration in weight/volume percent); zC (organic phase temperature in °C) unless otherwise stated. In

128

Chapter 6, membrane samples were prepared with 300 s reaction time and 0.0038 mol/dm3 organic monomer concentration and are designated with the polymer name (i.e., MPDTMC or MPDTrip) with organic solution temperature as the suffix (e.g., MPDTMC-100 or

MPDTrip-20). In Chapter 7, membrane samples are designated with the polymer name

(i.e., MPDTMC or PPDTMC) with either SRO (referring to reaction time 10 s, organic monomer concentration 0.0038 mol/dm3 and organic solution temperature 20 °C) or KRO1

(referring to reaction time 300 s, organic monomer concentration 0.0038 mol/dm3 and organic solution temperature 100 °C) as the suffix.

Data for at least 3 samples are reported for each test.

3.3. Polymer Powder Preparation

0.093 mol/dm3 diamine (100 ml in distilled water) and 0.0038 mol/dm3 of acyl chloride

(300 ml in organic solvent) were prepared separately and poured in a 1 L vial to begin the polycondensation reaction. The vial was rotated gently to ensure continuous interface generation. After 30 min, the polymer was removed and washed with 500 ml of the organic solvent used followed by vacuum filtration. The washing was repeated once with the organic solvent used followed twice with distilled water and twice with ethanol, consecutively. After the final filtration, the polymer was dried under vacuum at 120 °C for

20 h. Finally, the polymer was stored in a desiccator until further testing.

129

3.4. Gas Permeation

3.4.1. Pure-Gas Permeation

Figure 3.1 shows the custom-made thin-film pure-gas permeation setup used. The system is based on the constant pressure/variable volume method. A Millipore stainless steel cell

(active area 13.6 cm2) was connected to a feed, permeate and retentate line. Membrane coupons were cut using an EPILOG mini laser cutter and sealed in the cell. Standard tests were performed at 22 °C.

Pressure To vacuum regulator P

Bubble Gas flowmeter Cell cylinder

Permeate Retentate/ Relief To vacuum

Figure 3.1. Constant pressure/variable volume pure-gas permeation system.

Before the permeation test, both upstream and downstream were evacuated for 10 minutes.

The feed gas was then loaded at 7.9 bar (100 psig). The permeate side of the membrane was exposed to atmospheric pressure (1 bar). Flow rates were measured using soap-bubble flow meters, and the system was purged before the measurement with the respective test gas.

130

Permeance was calculated using the following equation:

273푝푎푡푚푑푉 푃푒푟푚푒푎푛푐푒 = (Eq. 3.1) 퐴(푝푓푒푒푑 − 푝푝푒푟푚)푇76푑푡

where, patm, pfeed, and pperm are atmospheric, feed and permeate pressures (cmHg), respectively, dV/dt is the volumetric flow rate (cm3 s-1), and T is the measurement temperature (K). Permeance is calculated in GPU where 1 GPU = 10-6 cm3 (STP) cm-2 s-1 cmHg-1.

Pure-gas selectivity (α) for each gas pair was calculated using Equation 2.28.

Prior to the specific gas permeation tests, compressed air was permeated through the samples at 7.9 bar (100 psig) for 48 hours to allow for potential membrane compaction.

Gas permeation properties were measured in the order helium, hydrogen, oxygen, nitrogen, methane, and carbon dioxide. Between each gas permeation test, the membrane cell was evacuated for 10 minutes.

3.4.2. Mixed-Gas Permeation

Figure 3.2 shows the apparatus used. Initially, the non-flammable gas was permeated through the system for 30 minutes to ensure complete removal of atmospheric oxygen in the lines. The preheat coil and cell heating elements were heated to the desired temperature.

Secondary gas feed was then initiated. Both gases were fed at 500 ml/min totaling to a cross-flow rate of 1000 ml/min with a gas ratio of 50:50. The flow rate through the membrane was measured using a bubble flow meter and permeate composition was

131 measured continuously using an Agilent Technologies 490 Micro gas chromatograph. The permeate was collected at room temperature.

P Pre-heat coil MFC 1 T MFC 2

Bubble Cell flowmeter Needle valve Heating tape Retentate Permeate GC

Figure 3.2. Constant pressure/variable volume mixed-gas permeation system.

Separation factor was calculated using Equation 2.29.

Mixed-gas permeance for component x was calculated as:

푥푝푒푟푚273푝푎푡푚푑푉 푃푒푟푚푒푎푛푐푒 = (Eq. 3.2) 퐴[(푥푓푒푒푑 푝푓푒푒푑) − (푥푝푒푟푚 푝푝푒푟푚)]푇76푑푡

where, xperm and xfeed are molar fractions of component x in permeate and feed, respectively.

3.4.3. Temperature Dependence Measurements

For temperature dependence measurements, the membrane cell was heated using heating tape at the desired test temperature until equilibration. Permeate was collected at room

132 temperature. Pure-gas temperature dependence was conducted between 22 - 140 °C at a feed pressure of 7.9 bar.

3.5. Liquid Permeation

3.5.1. Water Desalination

A custom-made RO cross-flow system shown in Figure 3.3 was used to determine the salt rejections. The feed solution was prepared by dissolving either 2000 or 35000 ppm sodium chloride in 12 liters of DI water for brackish water desalination (BWRO) and seawater desalination (SWRO) tests, respectively. Pressure- and feed flow rates were adjusted by the pump and relief valve. Spacers were used in the feed chamber to increase turbulence, thus reducing concentration polarization near the membrane surface. Permeate was recirculated back to the feed tank. A heat exchanger was used to maintain the solution temperature at a set level (23 °C ± 1). The system was operated at 15.5 bar or 55 bar, for

BWRO and SWRO respectively, at a crossflow rate of 1 L/min. The system was allowed to equilibrate for 24 to 48 hours. Afterward, m gram of permeate was collected for a period of time t with a membrane area A.

A conductivity meter (calibrated to correspond to specific salt concentration) was used to measure salt concentration in the permeate and retentate. At least three coupons were tested for each membrane type.

133

Figure 3.3. Schematic diagram of salt rejection permeation setup [4].

3.5.2. Rejection and Water/Solute Flux Calculations

Rejection was calculated using Equation 2.63.

-2 -1 Water flux Jw (L m h ) can be calculated using Equation 2.61.

For sodium chloride solutions, Δπ is given by;

2훥푐 · 푅 · 푇 훥휋 = 푠 (Eq. 3.3) 푀푊

3 where, MW = molecular mass sodium chloride (g/mol), R = universal gas constant (cm bar

-1 -1 -3 K mol ), T = temperature (K) and Δcs (g cm ) = differential mass concentration of sodium chloride in water which is defined as:

훥푐푠 = 푐푓푒푒푑 − 푐푝푒푟푚푒푎푡푒 (Eq. 3.4)

134

3 -2 -1 -1 Normalized permeance of water PW/l (cm cm s or cm s ) can be calculated using a previously reported method [5,6] as:

푃 퐴 · 푅 · 푇 푊 = (Eq. 3.5) 푙 푀푊

Eq. 3.5 is used to convert pressure-normalized flux to concentration-normalized flux to ensure fair comparison between water and salt flux.

-2 -1 The salt flux Js (kg m h ) through the membrane is defined as;

퐽푠 = 퐽푤. 푐푝푒푟푚푒푎푡푒 (Eq. 3.6)

Salt permeance B (m3 m-2 h-1 or m h-1) can then be defined as;

퐽 퐵 = 푠 (Eq. 3.7) 훥푐푠

Normalized salt permeance PS/l equals B.

Water/sodium chloride selectivity αWater/NaCl is calculated as

푃푊/푙 푃푊 훼푊푎푡푒푟/푁푎퐶푙 = = (Eq.3.8) 푃푠/푙 푃푠

Boron (and other solute) flux, permeance and water/boron selectivity (αWater/Boron) were also calculated by applying Eqs. 3.5 to 3.8.

3.6. Modelling

Modelling was conducted by Dr. Hakkim Vovusha and Prof. Udo Schwingenschlogl.

Molecular dynamics (MD) simulations were performed using Material Studio 8.0 with

COMPASS force field. Amorphous cells of 10 units of MPDTMC and 10 units of

135

MPDTRIP have been built separately using amorphous cell module, energy minimization was performed using molecular mechanics methods in the Forcite module. The systems were stabilized by inter molecular hydrogen bonds (-CO···NH, -COOH···CO) and stacking interactions (between aromatic rings). Both NPT and NVT ensemble have been used throughout the MD simulation (2000 ps) with 295K temperature and 8 bar pressure.

The temperature and pressure was maintained using the Nose-Hoover thermostat and the

Berendsen barostat respectively. The velocity Verlet algorithm was used to integrate the equations of motion, with a time step of 1 fs. The non-bonded terms and the Coulombic interactions were evaluated by the Ewald sum method. The Lennard-Jones interactions were calculated by the atom-based method with a cutoff of 12.5Å.

3.7. Sorption

3.7.1. Low-Pressure Gas Sorption

The Brunauer, Emmett, and Teller (BET) technique was used to determine the nature of microporosity in the bulk polymer using a Micrometrics ASAP™ 2020 system. A schematic of the system is shown in Figure 3.4. N2 and CO2 were chosen as the probing molecules for measuring uptake and specific surface area at 77 and 273K, respectively between absolute pressures of 0 – 1 atm. CO2 uptake was also measured at 298K. Non-localized density function theory (NLDFT) was used to obtain qualitative pore size distributions assuming carbon slit-like pores.

136

3.7.2. High-Pressure Gas Sorption

High-pressure sorption was conducted using a Hiden Intelligent Gravimetric Analyzer

(IGA) from Hiden Isochema, UK, by Dr. Giuseppe Genduso. The system can test sorption between vacuum to 20 bar and cryogenic to 1000 °C temperatures. Instrument schematics are shown in Figure 3.5. Sorption measurements were made between 1 to 20 bar at a temperature of 25 °C.

137

Figure 3.5. Schematic of Hiden IGA high-pressure sorption system. Adopted from Swaidan et al. [7].

3.7.3. Vapor Sorption

Water and organic solvent uptakes were measured using a gravimetric sorption analyzer.

The instrument measures the mass variation in the polymer with variance in water (or organic) relative pressure.

138

The relative pressure was controlled by varying the flow of saturated and dry pure helium gas in a temperature-controlled chamber. Isotherms were measured at 25 °C between relative pressure values of 0.02 – 0.98.

3.8. Extended Characterization

3.8.1. Fourier Transform Infrared (FTIR) Spectroscopy

To confirm the presence of relevant functional groups on the surface of the TFCs, Fourier transform infrared (FTIR) spectroscopy was conducted using a Thermo Scientific Nicolet iS10 spectrometer. A germanium crystal was employed at an angle of 45° to obtain spectra between 4000 - 400 cm-1.

3.8.2. Scanning Electron Microscopy (SEM)

SEM micrographs were obtained by Dr. Federico Pacheco. FEI Nova NanoSEM (Scanning

Electron Microscope) was used for imaging of the surface and cross-sections of the membranes to examine structural features and layer homogeneity. Samples were sputter coated with iridium to improve conductivity. Samples for cross-sectional images were obtained by breaking the membrane following immersion in liquid N2.

3.8.3. Atomic Force Microscopy (AFM)

Atomic force microscopy (AFM) was used to characterize the surface morphology of the membranes specifically surface roughness. Measurements were conducted for 8-points on each sample using a Bruker Dimension Icon SPM AFM, and average values were reported.

A FESPA (Bruker) AFM probe with spring constant of 2.8 N/m was used employing a tapping mode in air imaging mode with a scanning speed of 1 Hz.

139

3.8.4. Wide-Angle X-ray diffraction (XRD)

Wide-angle X-ray diffraction (XRD) was performed by Dr. Yingge Wang. XRD patterns of the polymer powder samples were collected on a Bruker D8 Advance diffractometer using Bruker zero background sample holder at a scanning rate of 1o min-1, 0.02o step size with 2 ranging from 7o to 40o and the average chain spacing was calculated using Bragg’s law.

3.8.5. X-ray Photoelectron Spectroscopy (XPS)

X-ray photoelectron spectroscopy (XPS) studies were carried out in a Kratos Axis Supra spectrometer equipped with a monochromatic Al Kα X-ray source (hν = 1486.6 eV) operating at 150 W, a multi-channel plate and delay line detector under a vacuum of ~10-9 mbar. All spectra were recorded using an aperture slot of 300 μm x 700 μm. Survey spectra were collected using a pass energy of 160 eV and a step size of 1 eV. A pass energy of 20 eV and a step size of 0.1 eV were used for the high-resolution spectra. Samples were mounted in the floating mode to avoid differential charging. Charge neutralization was required for all samples. Binding energies were referenced to the C 1s binding energy of

Sp2 hybridized carbon taken to be 284.4 eV. The data were analyzed with commercially available software, CASAXPS.

3.8.6. Activation Energy of Permeation

The permeance of a gas can be described as:

−퐸 푃 = 푃 푒푥푝 ( 푝) (Eq. 3.9) 0 푅푇

140

-1 where, P0 is a constant, Ep is activation energy of permeation (J mol ), R is the universal gas constant (8.314 J mol-1 K-1), and T is the temperature (K).

Ep for each gas was calculated using the slope of log P plotted versus 1/T.

3.8.7. Thermogravimetric Analysis

A TGA Q5000 was used to carry out the Thermal gravimetric analysis (TGA). Helium was used as the carrier gas. The samples were kept isothermal at 100 °C to remove any absorbed water. Afterward, the temperature was ramped to 800 °C at a rate of 5 °C/min.

3.8.8. Membrane Surface Charge Measurements

Membrane surface charge was measured using the Anton Paar/SurPASS electrokinetic analyzer. The cell gap height was adjusted to approximately 100 μm. A 0.001M NaCl solution was used as an electrolyte, and the pH was adjusted manually using 0.1M HCl and

0.1M NaOH. The surface charge (or zeta potential) was calculated using the Helmholtz-

Smoluchowski equation.

3.8.9. Ellipsometry

Ellipsometry measurements were conducted by Dr. Wojciech Ogieglo. A spectroscopic ellipsometer M-2000 UI operating in a wavelength range 192 – 1690 nm coupled with light focusing optics (300 microns, short axis) and modeling software CompleteEASE v.5.24 was used to perform polyamide layer thickness measurements. The procedure is described in detail elsewhere [8,9]. In brief, the ellipsometric spectra were recorded over a square sample in equal steps in x and y directions to produce 4 x 4 mm maps. Afterward, the raw data, represented by ellipsometric angles psi and delta, were converted into IP film

141 thickness and its refractive index by numerical fitting using a single layer, isotropic optical model. The properties of the bare substrate were recorded separately and subsequently fixed in the optical model of the IP membrane.

142

3.9. References

[1] N. Alaslai, X. Ma, B. Ghanem, Y. Wang, F. Alghunaimi, I. Pinnau, Synthesis and characterization of a novel microporous dihydroxyl-functionalized triptycene- diamine-based polyimide for natural gas membrane separation, Macromol. Rapid Commun. 38 (2017) 1700303.

[2] H. Ellison, D.H. Hey, The action of benzaldehyde on o-, m-, and p-xylene in the presence of aluminium chloride, J. Chem. Soc. 0 (1938) 1847.

[3] B.S. Ghanem, F. Alghunaimi, Y. Wang, G. Genduso, I. Pinnau, Synthesis of highly gas-permeable polyimides of intrinsic microporosity serived from 1,3,6,8- tetramethyl-2,7-diaminotriptycene, ACS Omega 3 (2018) 11874–11882.

[4] J. Duan, Liquid and gas permeation studies on the structure and properties of polyamide thin-film composite membranes, Ph.D. thesis, King Abdullah University of Science and Technology, 2014.

[5] Z. Tan, S. Chen, X. Peng, L. Zhang, C. Gao, Polyamide membranes with nanoscale Turing structures for water purification, Science 360 (2018) 518–521.

[6] G.M. Geise, H.B. Park, A.C. Sagle, B.D. Freeman, J.E. McGrath, Water permeability and water/salt selectivity tradeoff in polymers for desalination, J. Membr. Sci. 369 (2011) 130–138.

[7] R. Swaidan, Intrinsically microporous polymer membranes for high performance gas separation, Ph.D. thesis, King Abdullah University of Science and Technology (KAUST), 2014.

[8] W. Ogieglo, I. Pinnau, M. Wessling, In-situ non-invasive imaging of liquid- immersed thin film composite membranes, J. Membr. Sci. 546 (2018) 206–214.

[9] M. Barth, M. Wiese, W. Ogieglo, D. Go, A.J.C. Kuehne, M. Wessling, Monolayer microgel composite membranes with tunable permeability, J. Membr. Sci. 555 (2018) 473–482.

143

Chapter 4. Ultra-Selective Defect-Free Interfacially Polymerized

Molecular Sieve Thin-Film Composite Membranes for H2 Purification

4.1. Abstract

Purification is a major bottleneck towards generating low-cost commercial hydrogen. In this work, inexpensive high-performance H2 separating membranes were fabricated by modifying the commercially successful reverse osmosis membrane production method.

Defect-free thin-film composite membranes were formed that demonstrate unprecedented mixed-gas H2/CO2 selectivity of ~50 at 140 °C with H2 permeance of 350 GPU, surpassing the permeance/selectivity upper bound of all known polymer membranes by a wide margin.

The combination of exceptional separation performance and low manufacturing cost makes

144 them excellent candidates for cost-effective hydrogen purification from steam cracking and similar processes. Permeance and selectivity measurements imply an ultrathin polyamide layer with apparent thickness of ~10 to 20 nm.

————————————————————————————————————

This chapter was published as:

Z. Ali, F. Pacheco, E. Litwiller, Y. Wang, Y. Han, I. Pinnau, Ultra-selective defect-free interfacially polymerized molecular sieve thin-film composite membranes for H2 purification, J. Mater. Chem. A. 6 (2017) 30–35.

Reproduced with permission from © The Royal Society of Chemistry 2018.

145

4.2. Introduction

The transport sector consumes between 30 - 50% of global energy with demands continuing to increase annually [1]. Coupled with established correlations between anthropogenic greenhouse gas emissions and global climate change, the need for energy efficient, environmentally friendly fuels is greater than ever [2]. Hydrogen offers huge potential as an alternative fuel of the future due to its high energy storage capacity (119

MJ/kg) and zero-emissions combustion (produces only water) [3–5]. Approximately 8.3 x

1011 m3 of hydrogen – carrying 6 x 1012 MJ of energy – is produced annually, with over

90% obtained from fossil fuels (mainly natural gas and coal) and biomass or its derivatives.

A much smaller fraction is produced using water electrolysis [6].

During steam cracking of natural gas to produce hydrogen (steam-methane reforming,

SMR), methane and water are reformed to CO and H2 at ∼800 °C. The H2/CO mixture is then converted at about 350 °C into H2 and CO2. Composition of output streams can vary depending on the specific method employed. A typical SMR plant produces a 75/20

H2/CO2 ratio with 5% methane and <1% of other impurities [6]. Integrated Gasification

Combined Cycle (IGCC) plants using biomass or coal feedstock can produce H2/CO2 ratios of 60/40 [7]. Currently about half of globally synthesized hydrogen is used for the production of ammonia employed as fertilizer by the Haber process, while the remaining half is utilized in hydrocracking i.e. breaking large hydrocarbons into smaller ones for use as fuel [8]. Smaller quantities are used for production of methanol, plastics, pharmaceuticals, hydrogenation of oils, desulfurization of fuels, etc. [8]. Hydrogen production is currently growing at 10% annually, but it is estimated that availability of lower-cost hydrogen could immediately boost its use by 5- to 10-fold [9].

146

To date, chemical separation processes account for 10-15% of global energy consumption

[10]. The state-of-the-art technologies for H2 purification, i.e. cryogenic distillation and pressure-swing adsorption, are extremely energy intensive, accounting for around 30% of total plant capital and operating cost [11,12]. Membrane-based H2/CO2 separation offers a potential path to reduce process costs and debottleneck H2 purification.

Table 4.1 lists the United States Department of Energy (USDOE) membrane performance targets for hydrogen purification from syngas mixtures [13,14]. A number of materials are being considered for such separations, including inorganics such as carbon molecular sieves, zeolites, and metal membranes, as well as glassy polymers such as polybenzimidazole and polyimides with and without nanoparticles [15–25]. The economic and environmental benefits of using membranes for H2/CO2 separations have been discussed by Merkel et. al., who argued that H2/CO2 selectivities greater than 10 can significantly reduce hydrogen production cost [7,14,26]. Proteus™ by Membrane

Technology & Research Inc. is a proprietary commercial membrane offering H2/CO2

-6 3 selectivity of approximately 11 with H2 permeance of 500 GPU (1 GPU = 10 cm (STP) cm-2 s-1 cmHg-1) at 150 °C mixed-gas operation [14].

Table 4.1. USDOE specified requirements for H2/CO2 membranes [6,13,14].

Low fabrication costs: approximately 100 USD/ft2 or lower

Ability to manufacture large membrane areas and modules

High operating temperature: 120 – 150 °C and above

High pressure operability: 7 bar and above

High hydrogen purity and recovery

High durability: around 5 years

147

Performance: H2 permeance > 200 GPU

Mixed-gas H2/CO2 selectivity at 150 °C > 12 (IGCC operation)

Thin-film composite (TFC) reverse osmosis (RO) membranes constitute the most successful implementation of membrane technology in large-scale industrial separation processes to date due to their unmatched combination of high water flux and salt rejection.

Their high water flux results from the extremely thin polyamide selective layers, made possible by interfacial polymerization (IP). Figure 4.1 shows the structure of the partially crosslinked fully aromatic polyamide layer fabricated by reacting m-phenylenediamine

(MPD) and trimesoyl chloride (TMC) on polymeric supports, pioneered by Cadotte and commercially named FT-30 [27]. This TFC is currently employed in more than 15,000 desalination plants, accounting for 90% of the global market [28]. In commercial settings, the FT-30-type RO membranes are produced by impregnating (via dipping or spraying) a highly porous support material (usually polysulfone) with MPD dissolved in water.

Typically, excess solution is removed from the surface by using an air knife or a rubber roller. The diamine-soaked polysulfone is then exposed to TMC dissolved in a hydrocarbon solvent (n-hexane or Isopar®) typically between 1 to 60 seconds [27]. Most commonly, solutions in the IP process are applied at room temperature (20 - 25 °C). The membrane is then immediately exposed to high temperatures (80 - 100 °C) for drying and curing of the polyamide. All such membranes have been laboriously studied and reported in the literature with no useable gas separation properties for commercial separation processes. Gas permeation studies of dry FT-30-type RO membranes [29–32] established that they exhibited Knudsen diffusion, implying presence of mesoporous surface defects. Louie et al. demonstrated that plugging the surface defects by coating FT-30-type membranes with

148 a rubbery polyether–polyamide block copolymer (PEBAX® 1657) showed potential for

H2/CO2 separations [31].

Figure 4.1. Aromatic polyamide structure via interfacial polymerization reaction between MPD-TMC [30].

In this work, the successful fabrication of highly crosslinked, ultra-selective, defect-free

MPD-TMC polyamide thin-film composite molecular sieve membranes is reported for the first time. Pure- and mixed-gas H2/CO2 permeation was measured across a range of temperatures. The TFCs were further characterized using scanning electron microscopy

(SEM), X-ray photoelectron spectroscopy (XPS), X-ray diffraction (XRD) and Fourier transform infrared spectroscopy (FTIR). The performance of these inexpensive high- performance membranes exceeds the USDOE targets and the Robeson permeability/selectivity upper bound for H2/CO2 separation by a wide margin.

149

4.3. Results and Discussion

Table 4.2. Membrane formation variables and sample information. Concentration = TMC concentration, temperature = organic phase temperature and ‘m’ = crosslinking degree. (N.M = not measured).

Reaction time Concentration Temperature Membrane m (s) (wt/vol%) (°C)

FT-30 variant RO4 Proprietary (N.M)

10s-0.1TMC-20C 10 0.1 20 (N.M)

60s-0.1TMC-20C 60 0.1 20 (N.M)

300s-0.1TMC-20C 300 0.1 20 0.63

600s-0.1TMC-20C 600 0.1 20 (N.M)

300s-0.1TMC-60C 300 0.1 60 0.66

300s-1TMC-60C 300 1 60 0.55

300s-10TMC-60C 300 10 60 0.39

300s-0.1TMC- 300 0.1 100 0.89 100C

Table 4.2 lists the TFCs prepared. Commercially produced dry FT-30 membranes are known to contain micropores larger than the dimensions of gas molecules, as Knudsen selectivity has been measured in a variety of FT-30-type products [31,32]. Although specific production information is proprietary, it is widely known that these membranes are made with IP reaction times under one minute. Figure 4.2a and b show how defect-free polyamide layer characteristics, as indicated by significantly increased selectivity, start to emerge at longer reaction times. Permeance for H2 and He decreased 10-fold while an average decrease of at least 100-fold was observed for larger gases as reaction time was

150 increased from 10 to 300 s. Longer reaction times presumably allow more MPD to penetrate into the reaction zone forming additional polyamide and thus closing any defects in the ultrathin-film by a diffusion-driven self-healing process, as schematically shown in

Figure 4.3.

During this process, the mode of transport shifted from Knudsen flow to solution/diffusion, and gas pair selectivity already increased at 1 min, reaching an optimum at 5 minutes reaction time. Numerical performance data are summarized in Table 4.3 and 4.4. The 10s-

0.1TMC-20C membrane demonstrated identical gas permeation properties to a commercial

FT-30-type (Sepro RO4) and was used as the reference for comparing the performance of other TFCs in this work.

151

Figure 4.3. Proposed in-situ pore plugging process during interfacial polymerization of thin-film composite membrane.

Table 4.3. Permeance data for prepared TFCs.

300s- 10s- 60s- 300s- 600s- 300s- 300s- 300s- FT-30- 0.1TM Gas 0.1TM 0.1TM 0.1TM 0.1TM 0.1TM 1TMC 10TM type C- C-20C C-20C C-20C C-20C C-60C -60C C-60C 100C He 240.4 226.7 38.70 41.20 33.10 25.50 22.30 34.20 30.90

H2 295.4 203.7 38.60 32.90 28.70 20.50 17.40 27.70 25.80

CO2 106.2 67.6 7.60 4.30 4.60 1.10 1.50 2.90 1.80

152

O2 87.4 60.0 4.10 1.10 1.40 0.30 1.00 2.00 0.35

N2 91.0 69.2 3.90 0.60 0.90 0.06 0.80 1.80 0.04

CH4 118.2 109.9 4.90 0.70 1.10 0.05 1.20 2.50 0.02

Table 4.4. Selectivity data for prepared TFCs.

300s- 10s- 60s- 300s- 600s- 300s- 300s- 300s- Gas FT-30- 0.1TM 0.1TM 0.1TM 0.1TM 0.1TM 0.1TM 1TMC 10TM pair type C- C-20C C-20C C-20C C-20C C-60C -60C C-60C 100C

H2/CO2 3.20 2.50 5.60 7.60 6.20 20.0 13.50 2.50 14.3

H2/N2 3.00 4.30 13.80 63.10 30.60 29.6 29.60 4.30 644.4

H2/CH4 2.00 2.50 9.40 47.10 25.70 578.4 19.92 2.50 1592.4

O2/N2 0.90 0.80 1.10 2.10 1.50 5.10 1.40 1.20 8.70 CO /C 2 0.60 1.10 1.70 6.30 4.10 25.70 1.40 1.50 111.8 H4

N2/CH4 0.70 0.50 0.70 0.80 0.80 1.40 0.70 0.80 2.50

CO2/N2 1.00 1.90 2.40 8.40 4.90 18.50 2.10 1.90 45.1

Figure 4c and 4e show the effects of varying TMC concentration and temperature. A clear trend of decreasing permeance started to emerge for gases larger than hydrogen (kinetic diameter (kd) > 2.89 Å) [33] but no significant variation was observed for helium and hydrogen for either parameter. FTIR spectra (Figure 4.4) demonstrated no visible difference in polyamide chemistry for different fabricated samples compared to the 10s-

0.1TMC-20C reference membrane. After the interfacial polymerization (IP) reaction, three new peaks appear. The peaks at 1545 cm-1 and 1660 cm-1 confirm the presence of amide groups on the surface of the composite membrane. The former relates to the C=O stretching while the latter corresponds to N-H bending in the amide linkage. The peak at 1610 cm-1 is associated with aromatic ring breathing [34–36]. Because the penetration depth of the IR

153 beam was > 0.3 µm, the spectra of the polysulfone was clearly visible even after the support was coated with interfacially polymerized polyamide [37].

As TMC concentration decreased, the ratio of amine to acyl chloride functional groups increased as demonstrated by XPS analysis. As a result, permeance for CO2 and larger gases decreased due to increased crosslinking, as evidenced in Table 4.2. This resulted in tightening of the polyamide network, consequently hindering transport for larger gas molecules while no significant effect was observed on smaller gases (He and H2), which resulted in significant selectivity boosts. Similarly, increase of organic-phase temperature also increased crosslinking, which affected the permeance of gases larger than H2 thereby significantly enhancing selectivity. Presumably, increase in reaction-zone temperature increased the overall reaction rate as well as solubility and diffusivity of MPD in the organic phase (reaction-zone), resulting in increased formation of amide linkages and, hence, increased crosslinking [38].

154

Figure 4.4. FTIR spectra for polysulfone and TFCs in this study.

Figure 4.2d and 4.2f show the same performance results expressed in terms of selectivity.

High selectivity for H2/CO2 and negligible selectivity for He/H2 implies a primary molecular-sieve-like cut-off around 3 Å. This is clearly displayed in the XRD spectrum for the MPD-TMC powder in Figure 4.5, showing a single amorphous peak with average chain d-spacing centered around 3.55 Å. As ‘m’ increased from 0.39 to 0.66, selectivity of hydrogen over CO2, O2, N2 and CH4 increased, implying a decrease in the fraction of free volume elements with chain spacing larger than 3 Å (i.e., increased ultramicroporosity).

As crosslinking increased further from 0.66 to 0.89, N2 and CH4 permeance decreased (kd for N2 and CH4 are 3.64 Å and 3.80 Å, respectively) but CO2 and O2 permeances remained unaffected. This result can be interpreted as a reduction in the fraction of free volume elements with chain spacing larger than 3.5 Å. As a direct consequence, O2/N2, CO2/N2

155

and CO2/CH4 selectivities increased (Figure 4.6). These are all significant industrial gas separation applications for implementation of membrane technology.

Figure 4.5. XRD data for polyamide powder prepared by interfacial polymerization of trimesoyl chloride and m-phenylene diamine.

156

Figure 4.6. Alternative gas pair selectivity data for fabricated TFCs: a) reaction time variation, b) TMC concentration variation and c) organic phase temperature variation.

157

158

Figure 4.7. Top surface (a, c, e, g, i, k) and cross-section (b, d, f, h, j, l) SEM images of fabricated TFCs.

159

Figure 4.8. High magnification cross-section SEM images of fabricated TFCs for estimating polyamide layer thickness.

SEM images, Figure 4.7a-l and 4.8a-d, depict the fabricated TFCs show average visual polyamide ridge-and-valley-based film thickness of approximately 100 - 300 nm [39].

However, it has been debated that there is an appreciable difference between observed average cross-sectional thickness and actual effective thickness of the barrier layer. The apparent visual thickness has been conventionally considered the true thickness of the polyamide barrier layer [38,40]; however, more recent research has indicated that the effective thickness of the separation layer lies around the order of only 10 - 20 nm

160

[39,41,42]. Weinkauf, Kim and Paul [43] reported the gas permeation properties of thick films of a similar linear aromatic polyamide made from p-phenylenediamine and terephthaloyl chloride. Figure 4.9 depicts coupling selectivity data for 300s-0.1TMC-60C and poly(p-phenylene terephthalamide) unveils almost identical gas selectivities for both polymers. Assuming similar permeability for both polymers, the thickness of the 300s-

0.1TMC-60C film can be estimated as ~10 to 20 nm. Because the films exhibit a rough structure, the active surface area of the separation layer may be underestimated [38]. The cross-sections of a FT-30-type reference membrane (10s-0.1TMC-20C) and the defect- free, highly gas-selective polyamide TFC of this work (300s-0.1TMC-100C) are shown in

Figure 4.8. Although it is difficult to clearly assign a thickness to the selective polyamide barrier layer of both membrane types, it is clear that the PA layer is thicker and more tightly packed in the membrane made with longer reaction time and higher reaction temperature.

a b

Figure 4.9. Comparison of 300s-0.1TMC-60C and poly(p-phenylene terephthalamide) [43]: a). Pure-gas selectivity, and b) permeability data. 300s-0.1TMC-60C thickness estimated as 10 nm.

161

Despite barrier-type polyamides showing moderate to high selectivity for a number of gas separations, they exhibit particularly low gas permeabilities [44] and have subsequently been overlooked for gas separation processes. However, as demonstrated here, this disadvantage can be overcome by fabricating ultra-thin films allowing the exploitation of highly selective barrier materials with industrially useable performance characteristics.

4.3.1. Pure- and mixed-gas temperature dependence

a b

Figure 4.10. a) 300s-0.1TMC-100C pure-gas temperature dependence for H2 and CO2, and b) Robeson plot for performance comparison of membrane studied here (300s-0.1TMC- 100C). Adapted from Robeson (2008) upper-bound assuming 1 µm films [44]. State-of- the-art data for USDOE requirements, PEBAX-coated SWC4 (PEBAX-SWC4), ZIF- 8/PBI, MTR Proteus™ plotted separately from sources [14,20,31].

The membranes fabricated in this study showed excellent potential for syngas separations at 22 °C. However, the most important requirements for membranes in H2 production from syngas are stability at high feed temperatures (between 120 - 150 °C) and high pressures

(> 7 bar).

Figure 4.10a shows the performance of the 300s-0.1TMC-100C membrane as a function of temperature, using pure-gas H2 and CO2 measurements. Permeance, for

162

both gases, showed excellent Arrhenius regression with temperature. H2 experienced a larger increase in permeance compared to CO2 presumably due to reduced sorption of CO2 at higher temperatures. Activation energies for H2 and CO2 were calculated

-1 as 8.50 and 1.20 kJ mol , respectively. At 140 °C, H2 permeance increased to 275

± 4 GPU with H2/CO2 selectivity of 95.5 ± 5, the highest reported pure-gas selectivity to date of any polymer membrane.

Mixed-gas separation was conducted using a 1:1 H2/CO2 feed at 140 °C to verify realistic performance data in industrial systems. Figure 4.10b shows pure- and mixed-gas data for

300s-0.1TMC-100C compared to state-of-the-art membranes on the H2/CO2 Robeson plot

[44]. Average stabilized H2 permeate concentration of 98% was achieved, translating to an unprecedented mixed-gas separation factor of 50 ± 4 with hydrogen permeance of 350 ±

15 GPU at 140 °C. Hence, our mixed-gas permeation results clearly demonstrated unparalleled performance of the defect-free polyamide TFCs for H2/CO2 separation with properties located well above all state-of-the-art polymers when tested under industrially relevant conditions.

4.4. Conclusions

The growing need for cleaner energy has dramatically increased interest in separations using membrane systems. Highly crosslinked, ultra-selective, defect-free MPD-TMC membranes were successfully fabricated in this study showing tremendous potential for

H2/CO2 separation in syngas applications as well as a number of other challenging gas separations. These membranes exhibited unprecedented H2/CO2 selectivity, surpassing all other reported polymers and lying well above the 2008 Robeson upper bound.

163

Coupled with excellent H2/CH4 separation properties, given the targets specified by the

USDOE [13,14], the membranes are excellent candidates for pilot-scale testing aimed at hydrogen purification from syngas. Fortuitously, these ultra-high-performance membranes can be produced by making only small changes to existing commercial membrane manufacturing processes. Therefore, their fabrication cost should be similar to standard RO membranes – only ~1 – 2 USD ft-2, which would lower the membrane cost by 50- to 100- fold based on the USDOE target value of 100 USD ft-2 [45]. This study demonstrated that varying fabrication parameters can tune permselectivity to meet the needs of specific processes. A few simple modifications to a time-tested commercial membrane fabrication process can produce membranes that meet a key industrial need.

These membranes also demonstrated remarkable selectivity for O2/N2, CO2/CH4, H2/N2 and CO2/N2 separations. With rapidly developing economic and environmental pressures to increase efficiencies for separation processes, such highly-selective, low-cost, commercial barrier materials fabricated as ultrathin-films show potential for a paradigm shift to streamline industrial use of membranes for a large number of gas separation applications, specifically hydrogen separations.

164

4.5. References

[1] R.P. Lively, D.S. Sholl, From water to organics in membrane separations, Nat. Mater. 16 (2017) 276–279.

[2] J. Hansen, R. Ruedy, M. Sato, K. Lo, Global surface temperature change, Rev. Geophys. 48 (2010) 1–29.

[3] D. Teichmann, W. Arlt, P. Wasserscheid, Liquid organic hydrogen carriers as an efficient vector for the transport and storage of renewable energy, Int. J. Hydrogen Energy 37 (2012) 18118–18132.

[4] USA Congress, EPACT (Energy Policy ACT); Public law number. 102-486, Washington, 1992.

[5] J.A. Turner, Sustainable hydrogen production, Science 305 (2013) 972–974.

[6] N.W. Ockwig, T.M. Nenoff, Membranes for hydrogen separation, Chem. Rev. 107 (2007) 4078–4110.

[7] T.C. Merkel, M. Zhou, R.W. Baker, Carbon dioxide capture with membranes at an IGCC power plant, J. Membr. Sci. 389 (2012) 441–450.

[8] P. Häussinger, R. Lohmüller, A.M. Watson, Hydrogen, 2. Production, in: Ullmann’s Encycl. Ind. Chem., Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, Germany, 2011.

[9] G. Conzelmann, M. Petri, C. Forsberg, B. Yildiz, ORNL, Configuration and technology implications of potential nuclear hydrogen system applications, Chicago, 2005.

[10] D.S. Sholl, R.P. Lively, Seven chemical separations to change the world, Nature 532 (2016) 6–8.

[11] Thomas L. Buchanan, Michael G. Klett, Ronald L. Schoff, Capital and operating cost of hydrogen production from coal gasification, Pittsburgh, 2003.

[12] R. Spillman, Economics of gas separation membrane processes, in: R.D. Noble, S.A. Stern (Eds.), Membr. Sci. Technol., 2nd ed., Elsevier B.V., 1995: pp. 589–667.

[13] United States Department of Energy, Basic research needs for the hydrogen economy. Workshop on hydrogen production, storage and use., Lemont, 2003.

[14] United States Department of Energy, Advanced carbon dioxide capture R&D program: Technology update: Pre-combustion membranes, Pittsburgh, 2013.

[15] K.J. Bryden, J.Y. Ying, Nanostructured palladium–iron membranes for hydrogen

165

separation and membrane hydrogenation reactions, J. Membr. Sci. 203 (2002) 29– 42.

[16] Y.H. Ma, Hydrogen separation membranes, Adv. Membr. Technol. Appl. (2008) 671–684.

[17] S.C. Kumbharkar, K. Li, Structurally modified polybenzimidazole hollow fibre membranes with enhanced gas permeation properties, J. Membr. Sci. 415–416 (2012) 793–800.

[18] D.R. Pesiri, B. Jorgensen, R.C. Dye, Thermal optimization of polybenzimidazole meniscus membranes for the separation of hydrogen, methane, and carbon dioxide, J. Membr. Sci. 218 (2003) 11–18.

[19] K.A. Berchtold, R.P. Singh, J.S. Young, K.W. Dudeck, Polybenzimidazole composite membranes for high temperature synthesis gas separations, J. Membr. Sci. 415–416 (2012) 265–270.

[20] T. Yang, T.-S. Chung, High performance ZIF-8/PBI nano-composite membranes for high temperature hydrogen separation consisting of carbon monoxide and water vapor, Int. J. Hydrogen Energy 38 (2013) 229–239.

[21] S.S. Hosseini, N. Peng, T.S. Chung, Gas separation membranes developed through integration of polymer blending and dual-layer hollow fiber spinning process for hydrogen and natural gas enrichments, J. Membr. Sci. 349 (2010) 156–166.

[22] S. Japip, K.S. Liao, T.S. Chung, Molecularly tuned free volume of vapor crosslinked 6FDA-Durene/ZIF-71 MMMs for H2/CO2 separation at 150 °C, Adv. Mater. 29 (2017) 1–6.

[23] P. Li, Z. Wang, Z. Qiao, Y. Liu, X. Cao, W. Li, J. Wang, S. Wang, Recent developments in membranes for efficient hydrogen purification, J. Membr. Sci. 495 (2015) 130–168.

[24] X. Li, R.P. Singh, K.W. Dudeck, K.A. Berchtold, B.C. Benicewicz, Influence of polybenzimidazole main chain structure on H2/CO2 separation at elevated temperatures, J. Membr. Sci. 461 (2014) 59–68.

[25] S.C. Kumbharkar, Y. Liu, K. Li, High performance polybenzimidazole based asymmetric hollow fibre membranes for H2/CO2 separation, J. Membr. Sci. 375 (2011) 231–240.

[26] United States Department of Energy, Cost and performance baseline for fossil energy plants: Report Number DOE/NETL-2007/1281, Pittsburgh, 2007.

[27] J.E. Cadotte, Interfacially synthesized reverse osmosis membrane, US 4,277,344 A, 1981.

166

[28] S.S. Shenvi, A.M. Isloor, A.F. Ismail, A review on RO membrane technology: Developments and challenges, Desalination 368 (2015) 10–26.

[29] J.S. Louie, I. Pinnau, M. Reinhard, Effects of surface coating process conditions on the water permeation and salt rejection properties of composite polyamide reverse osmosis membranes, J. Membr. Sci. 367 (2011) 249–255.

[30] J. Duan, PhD thesis: Liquid and gas permeation studies on the structure and properties of polyamide thin-film composite membranes, King Abdullah University of Science and Technology, 2014.

[31] J.S. Louie, I. Pinnau, M. Reinhard, Gas and liquid permeation properties of modified interfacial composite reverse osmosis membranes, J. Membr. Sci. 325 (2008) 793– 800.

[32] J. Albo, J. Wang, T. Tsuru, Gas transport properties of interfacially polymerized polyamide composite membranes under different pre-treatments and temperatures, J. Membr. Sci. 449 (2014) 109–118.

[33] N. Mehio, S. Dai, D.E. Jiang, Quantum mechanical basis for kinetic diameters of small gaseous molecules, J. Phys. Chem. A 118 (2014) 1150–1154.

[34] B. Khorshidi, T. Thundat, B.A. Fleck, M. Sadrzadeh, Thin film composite polyamide membranes: parametric study on the influence of synthesis conditions, RSC Adv. 5 (2015) 54985–54997.

[35] M. Liu, D. Wu, S. Yu, C. Gao, Influence of the polyacyl chloride structure on the reverse osmosis performance, surface properties and chlorine stability of the thin- film composite polyamide membranes, J. Membr. Sci. 326 (2009) 205–214.

[36] Y. Jin, W. Wang, Z. Su, Spectroscopic study on water diffusion in aromatic polyamide thin film, J. Membr. Sci. 379 (2011) 121–130.

[37] A.P. Rao, S. V. Joshi, J.J. Trivedi, C. V. Devmurari, V.J. Shah, Structure- performance correlation of polyamide thin film composite membranes: Effect of coating conditions on film formation, J. Membr. Sci. 211 (2003) 13–24.

[38] A.K. Ghosh, B.H. Jeong, X. Huang, E.M.V. Hoek, Impacts of reaction and curing conditions on polyamide composite reverse osmosis membrane properties, J. Membr. Sci. 311 (2008) 34–45.

[39] F. Pacheco, R. Sougrat, M. Reinhard, J.O. Leckie, I. Pinnau, 3D visualization of the internal nanostructure of polyamide thin films in RO membranes, J. Membr. Sci. 501 (2016) 33–44.

[40] L. Lin, C. Feng, R. Lopez, O. Coronell, Identifying facile and accurate methods to measure the thickness of the active layers of thin-film composite membranes - A comparison of seven characterization techniques, J. Membr. Sci. 498 (2016) 167–

167

179.

[41] F.A. Pacheco, I. Pinnau, M. Reinhard, J.O. Leckie, Characterization of isolated polyamide thin films of RO and NF membranes using novel TEM techniques, J. Membr. Sci. 358 (2010) 51–59.

[42] S. Karan, Z. Jiang, A.G. Livingston, Sub-10 nm polyamide nanofilms with ultrafast solvent transport for molecular separation, Science 348 (2015) 1347–1351.

[43] D.H. Weinkauf, H.D. Kim, D.R. Paul, Gas transport properties of liquid crystalline poly(p-phenyleneterephthalamide), Macromolecules 25 (1992) 788–796.

[44] L.M. Robeson, The upper bound revisited, J. Membr. Sci. 320 (2008) 390–400.

[45] X. Chen, Z. Zhang, L. Liu, R. Cheng, L. Shi, X. Zheng, RO applications in China: History, current status, and driving forces, Desalination 397 (2016) 185–193.

168

Chapter 5. Defect-Free Highly Selective Polyamide Thin-Film

Composite Membranes for Desalination and Boron Removal

5.1. Abstract

Removal of boron from saline water sources has presented a major challenge for commercially available reverse osmosis desalination membranes. In this work, we report the boron and sodium chloride separation properties of truly defect-free, highly selective, interfacially polymerized aromatic polyamide thin-film composite membranes. The fabricated membranes show potential for separating sodium chloride with a maximum rejection of 99.6% obtained for the optimized film-forming protocol under lab-scale

169 brackish water desalination conditions. This translated into promising boron rejection performance with rejections of up to 99% at pH 10, higher than a number of commercially available reverse osmosis membranes tested in-house. Comprehensive characterization including X-ray photoelectron spectroscopy, atomic force microscopy, scanning electron microscopy, ellipsometry, and surface charge measurements revealed intimate insights on interfacially polymerized polyamide membrane structure-property relationships. Increased membrane crosslinking was shown to be the primary determining factor for membrane permselectivity performance. Furthermore, relationships were established between microstructural properties such as crosslinking and morphological characteristics like surface roughness, highlighting an intricate and complex structure formation mechanism.

————————————————————————————————————

This chapter was published as:

Zain Ali, Yasmeen Al Sunbul, Federico Pacheco, Wojciech Ogieglo, Yingge Wang, Giuseppe Genduso and Ingo Pinnau, Defect-free highly selective polyamide thin-film composite membranes for desalination and boron removal, Journal of Membrane Science. 578 (2019) 85-94.

170

5.2. Introduction

Over one billion people worldwide have extremely limited availability of safe drinking and irrigation water and by 2030 the demand is predicted to further increase by 50% [1,2]. This requires the development of more energy-efficient desalination processes. Over the past decades, reverse osmosis (RO) has established itself as a low-cost, high-efficiency method for water desalination. Currently, over 60% of the total world desalination capacity is provided by RO-based desalination plants [3]. The success of the process can be accredited to development of Filmtec’s revolutionary aromatic polyamide thin-film composite (TFC) reverse osmosis (RO) FT-30 membranes made by interfacial polymerization of m- phenylene diamine (MPD) and trimesoyl chloride (TMC), which exhibit high water flux and salt rejection [4]. Such membranes are fabricated using interfacial polymerization (IP), an easily scalable fabrication process [5]. Despite their tremendous success for desalination with > 99% salt rejection, FT-30-type membranes tend to have lower rejection towards small uncharged solutes such as alcohols, ketones, esters and boric acid.

Boron originated from either natural and/or anthropogenic sources is widely present in surface water, groundwater, and oceans with a wide range of concentrations. However, the majority of boron is found in seawater with an average concentration of about 4.5 mg L-1 [6]. The World

Health Organization (WHO) recommends boron concentration of 2.4 mg L-1 for drinking water while for irrigation purposes concentrations of < 1 mg L-1 are advised for agricultural crops including wheat, carrots and potatoes among others [7,8]. In plants, high boron concentration can reduce root division, root growth and inhibit photosynthesis [9] whereas in humans and animals, it is correlated with diseases in the alimentary, cardiac and nervous systems [10].

Therefore, it is essential to regulate boron concentration but boron species can easily diffuse

171 through RO membranes due to their small size and their uncharged nature in the form of boric acid in seawater with pH around 8 [11,12]. Boron rejection of several commercially available

FT-30-type RO membranes is shown in Table 1. Because of the relatively low boron rejection of FT-30 type membranes, a two-stage RO process is typically employed in desalination plants to decrease boron concentration to acceptable limits. This adds both capital and operating costs to the desalination process [13,14].

Table 5.1. Boron rejection for commercially available FT-30-type membranes.

Temperature Boron Year Membrane type pH Experimental conditions Ref. (° C) rejection (%)

Hydranautics 15.5 bar, 5 ppm boron, no 2000 6.5 - 40 [15] 4040-LHACPA2 NaCla Toray 15.5 bar, 4 ppm boron, no 2001 9.5 - 60 [16] SUL-C10 NaCla

Toray 55 bar, 4 ppm boron, 35,000 2001 - 25-30 75-90 [17] UTC-80-LB ppm NaClb

DOW 48 bar, 5.1 ppm boron, 2008 SW-30-HR 8.2 23 88 [18] 38,000 ppm NaCla

GE Osmonics 20 bar, 40 ppm boron, no 2008 AG 11 34 95 [19] NaClb

Toray 55 bar 5 ppm boron, 40,200 2011 TM-820A-400 8 25 93 [20] ppm NaClc

Hydranautics 55 bar, 5 ppm boron, 32,000 2018 SWC5 max 7 - 59 [21] ppm NaClb

Hydranautics 55 bar, 5 ppm boron, 32,000 2018 SWC4B 7 - 68 [21] ppm NaClb

Toray 55 bar, 5 ppm boron, 32000 2019 TM820S-400 8 25 60-85 [22] ppm NaCla

Toray 55 bar, 5 ppm boron, 32000 2019 8 25 85-90 [22] TM820L-440 ppm NaCla Note: a Pilot/plant scale study; b lab study; c manufacturer data.

Most research aimed at increasing boron rejection of membranes has investigated the effects of desalination process operating conditions such as pH, temperature, salinity and

172 pressure [23,24]. Other efforts include: (1) process flow optimization approaches [13,25];

(2) modification of commercially available membranes as well as using novel materials

[12,26,27] to reduce either boron affinity or pore size of the membranes [21]. Recently,

Shultz et al. used hydrophobic aliphatic amines as a “caulking” agent to tighten the pore structure and simultaneously altered membrane surface chemistry, both favorable for enhanced boron rejection. Successful scale-up of such approach was later demonstrated in a spiral-wound seawater RO pilot study with increased boron rejection, a moderate decrease in water permeability, and essentially no change in salt rejection due to the effective sealing of large defects in the membranes using “hydrophobic molecular plugs”

[21,28].

It is well established that FT-30-type membranes contain micro- and mesoporous defects in the dry state, clearly indicated by Knudsen selectivity for small gas molecules [29,30].

In the dry state, these defects have a significant impact on the gas separation capabilities of such membranes. Recently, our group fabricated defect-free, ultra-selective FT-30-type

RO membranes with excellent gas separation performance especially for hydrogen separation from mixtures containing CO2 [31]. Our study showed that “in-situ” plugging of defects in FT-30-type RO membranes, by increasing reaction time and reaction temperature, changed penetrant transport mode from Knudsen diffusion to solution- diffusion as illustrated in Figure 5.1. Even further performance enhancement was achieved by increasing the degree of membrane crosslinking by varying organic solution temperature and organic monomer concentration. The gas separation properties showed a significant performance boost, e.g., H2/N2 pure-gas selectivity was increased from ~4 for

FT-30-type fabrication conditions to ~600 with our optimized fabrication procedure. If gas

173 separation properties are any indicator for separation performance in the liquid state, the membranes should also demonstrate improvement in permselectivity performance over currently available RO membranes.

Figure 5.1. Proposed in-situ pore plugging process during interfacial polymerization of thin-film composite membranes.

In this work, the performance of defect-free, highly selective MPD-TMC polyamide thin- film composite membranes are reported for boron and NaCl RO separation under brackish water operating conditions. Boron rejection was measured at three different pH values and compared to a number of commercially available sea- and brackish water type RO membranes in both dead-end and crossflow modes. The membranes were characterized using X-ray photoelectron spectroscopy (XPS), scanning electron microscopy (SEM), ellipsometry, atomic force microscopy (AFM) and surface charge measurements. The

174 defect-free membranes show potential for improving boron and sodium chloride rejection of traditional TFC membranes through in-situ plugging and/or self-healing using modified interfacial polymerization conditions. Various characterization techniques combined with boron and NaCl separation performance of the modified membranes reveal structure- transport property relationships for FT-30-type RO membranes.

5.3. Results and Discussion

Table 5.2. Fabricated membrane sample types.

Membrane

FT-30 variant RO4 10s-0.1TMC-20C 60s-0.1TMC-20C 300s-0.1TMC-20C 600s-0.1TMC-20C 300s-0.1TMC-60C 300s-0.1TMC-100C (KRO-1) 300s-1TMC-60C 300s-10TMC-60C

The fabricated TFCs are listed in Table 5.1. It has been well established that for FT-30- type RO membranes very high rejection for sodium chloride translates to high boron rejection [32]. In this study, we used sodium chloride rejection as a screening tool to identify membranes with potential for high boron rejection. Commercially manufactured

FT-30-type RO membranes exhibit micro- and mesoporous defects in the dry state but are known for their high rejection of sodium chloride in the wet state [33]. It has been proposed that under RO conditions the polyamide layer swells, possibly self-plugging the majority of defects in the ultrathin polyamide layer [34]. Transport through such membranes has

175 been defined by ‘solution-diffusion’ and ‘solution-diffusion with defects’ models

[22,35,36]. In commercial production of FT-30-type RO membranes a short reaction time is used, with the idea to minimize the thickness of the barrier layer and hence, maximize water flux. Sodium chloride rejection of the tested membranes is tabulated in Table 5.3.

Similar flux and rejection performance for FT-30 variant Sepro RO4 compared to in-house fabricated 10s-0.1TMC-20C implies similar fabrication conditions. Figure 5.2a shows how defect-free properties, as illustrated by higher rejection, start to develop at longer reaction times. As reaction time was increased from 10 to 300 s, the pure water permeance decreased from 1.65 to 0.83 LMH/bar showing a linearly decreasing correlation with the reaction time. In this study, SEM and ellipsometry were used to qualitatively characterize polyamide thickness. Ellipsometry data, Table 5.4, shows an increase of film thickness from 107 to 151 nm as reaction time was increased from 10 to 300 s, which can explain the decreasing water permeance. However, when the reaction time was further increased to 600 s, an increase in water permeance was observed despite increasing thickness. This increase in permeance can possibly be explained by the increase in effective surface area of the membrane due to increased roughness, described by the roughness ratio (RR) in

Table 5.4 [37].

Flux Membrane A (LMH/bar) Rejection (%) Pw/Ps (LMH)

Sepro RO4 25.1 1.73 97.9 5053

10s-0.1TMC- 23.9 1.65 97.9 4881 20C

176

60s-0.1TMC- 21.7 1.50 98.9 9213 20C

300s-0.1TMC- 12.0 0.83 99.2 12849 20C

600s-0.1TMC- 15.3 1.06 99.2 14299 20C

300s-0.1TMC- 10.3 0.71 99.2 13009 60C

300s-0.1TMC- 10.9 0.75 99.6 25804 100C (KRO-1*)

300s-1TMC-60C 9.9 0.68 98.9 9131

300s-10TMC- 11.2 0.77 94.7 1898 60C *KRO-1 = Optimized KAUST TFC membrane made under the following conditions: reaction time: 300 s; 0.1 wt/vol% TMC in Isopar G; 100 °C organic solution temperature.

Roughness Crosslinking Thickness Charge Membrane ratio (RR) N/O ratio (nm) (mV)

Sepro RO4 1.5 (N.M) (N.M) -2.1 10s-0.1TMC-20C 1.9 0.78 107 -18.3 60s-0.1TMC-20C 1.8 0.76 122 -20.5 300s-0.1TMC-20C 2.0 0.78 151 -13.9 600s-0.1TMC-20C 2.2 0.72 165 -15.3 300s-0.1TMC-60C 2.1 0.79 231 -16.1 300s-0.1TMC-100C 3.2 0.93 777 -11.5 (KRO-1)

300s-1TMC-60C 1.9 0.74 139 -12.2

177

300s-10TMC-60C 1.1 0.66 20 -21.3

Meanwhile as reaction time was increased from 10 to 300 s, the rejection increased from

97.9 to 99.2%. Increasing overall membrane thickness with increasing reaction time ensures plugging of defects in the polyamide layer as MPD continues to diffuse through the defects to react with TMC. This manifests itself as an increase in membrane rejection potentially due to change of transport from ‘solution-diffusion with defects’ to “pure” solution-diffusion mechanism [22,36].

Figure 5.2. a) Effect on water permeance and salt rejection of: a) reaction time at fixed organic solution temperature of 20 °C, and; b) organic solution temperature at fixed reaction time of 300 s. TMC concentration 0.1wt/vol% in Isopar G.

From our previous work [31] we gained understanding that further performance enhancements can be achieved by increasing organic solution temperature. As illustrated in Figure 5.2b, a small decrease of water permeance from 0.83 to 0.71 LMH/bar was observed as organic solution temperature was increased from 20 to 60 °C. When the temperature was further increased to 100 °C, the water permeance increased slightly. It can be seen from Figure 5.3 that as organic solution temperature was increased, a significant increase in membrane roughness was observed. Normalizing the water permeance with the

178 roughness ratio RR (A/RR) demonstrates a clear trend of linearly decreasing water permeance with increasing temperature. Meanwhile, rejection showed no variation when temperature was increased from 20 to 60 °C. When temperature was further increased to

100 °C, rejection significantly improved from 99.2 to 99.6%.

Figure 5.3. Surface AFM images of in-house fabricated polyamide thin-film composite membranes. RMS refers to root-mean-square roughness.

The decrease in water flux with increasing organic solution temperature can be explained by an increase in membrane thickness, e.g., membrane thickness increased from 151 to 231

179 nm as organic solution temperature was increased from 20 to 60 °C. Increasing the organic solution temperature decreased the solution viscosity, density and surface tension [38].

This increases the diffusivity of MPD through the reaction zone for further growth of the polyamide layer, which leads to formation of a thicker barrier layer with negligible changes in the polymer microstructure. Changes in polymer microstructure can be quantified using the degree of crosslinking represented by the N/O ratio deduced by XPS measurements. A fully crosslinked polymer has an N/O ratio of 1 whereas a fully linear polymer (no crosslinking) has an N/O ratio of 0.50 [39]. Increasing the temperature from 20 to 60 °C had a negligible effect on the crosslinking ratio, as shown in Table 5.3, confirming minimal microstructural changes. This explains the small variation in membrane rejection.

On the other hand, increasing organic solution temperature to 100 °C resulted in a further decrease in water permeance (roughness ratio normalized) but rejection increased significantly to 99.6%. As temperature was increased to 100 °C, the crosslinking ratio increased from 0.79 to 0.93. Increased polymer crosslinking resulted in decreased water- induced swelling of the thin polyamide layer due to the reduced mobility of the polymer chains leading to a boost in rejection [40–42]. Increasing the organic solution temperature presumably increases the reaction zone temperature. This, in turn, increases the reaction rate for the formation of amide linkages between the polymer chains resulting in increased crosslinking [31]. However, this change was not observed when the organic solution temperature was increased from 20 to 60 °C. Recently, attempts were made to measure reaction zone temperature for the MPD-TMC reaction and it was estimated that the reaction zone reaches ~80 °C during the polyamide formation processes [43]. To enable heat flow from the bulk organic solution to the reaction zone, a temperature of > 80 °C is required

180 for the organic phase which can explain a lack of variation in crosslinking and rejection at

60 °C. The overall effect of increasing reaction time and organic solution temperature on membrane morphology agrees qualitatively with direct observations obtained from SEM analysis as highlighted in Figure 5.4.

Because no further performance enhancement was observed by varying TMC concentration, the defect-free thin-film composite membrane made by the optimized 300s-

0.1TMC-100C protocol (henceforth referred to as KRO-1) was chosen for boron separation studies due to its superb sodium chloride rejection. The pH dependent boron rejection of the membrane was explored and compared to some commercially available membranes for boron removal from water. The dead-end rejection data for the membranes are plotted in

181

Figure 5.5a. Boron rejection for all tested membranes increased with increasing pH.

Moderate enhancement in boron rejection was observed by increasing pH from 6 to 8, whereas significant improvement resulted as the pH was increased from 8 to 10. Under dead-end filtration conditions, defect-free KRO-1 membranes consistently demonstrated higher boron rejection than four tested commercial RO membranes at three pH values, and reached a high boron rejection of 94% at pH 10. The Toray UTC-80LB membrane showed the second highest rejection of 88% followed by UTC-80B membrane (84%), whereas

Sepro RO4 showed a lower rejection of 80% at pH 10.

Figure 5.5. Boron rejection as function of pH for KRO-1 and some commercial RO membranes. a) Dead-end filtration and b) crossflow filtration. The highlighted area is - B(OH)4 ion dominated region.

At low concentrations (< 200 mg L-1), boron is present in solution as mononuclear species,

- comprising boric acid B(OH)3 and borate ions B(OH)4 [36]. Such solutions have a pKa of

~9.2 at 25 °C and distribution of boron in boric acid and borate ions is highly dependent on the pH of the solution. At pH < 9.2, boron is primarily present in the solution as boric acid while at pH > 9.2, boric acid rapidly disassociates into larger borate ions. Boric acid has a

182 trigonal structure, is non-polar and smaller than the polar borate ion [14]. Boron rejection has previously been correlated with several parameters especially the membrane surface charge due to increased boron rejection at higher pH values and the polar nature of borate ions [2,36,44]. Surface charge of the tested membranes is tabulated in Table 5.5. As pH was increased, the surface charge of all tested membranes became more negative. For the commercial membranes, Toray UTC-80LB showed the highest negative surface charge (-

48 mV) at pH 10 followed by Toray UTC-80B (-41 mV), Sepro RO4 (-16 mV) and Koch

XR (-12 mV). The boron rejection of the commercial membranes displayed some dependence on the surface charge as the boron rejection increased with increasing surface charge in the order of Koch XR < Sepro RO4 < Toray UTC-80B < Toray UTC-80LB, as shown in Figure 5.6a. Such correlation diminished as pH decreased to 8 and 6, as shown in Figure 5.7. This was expected because boric acid is the dominant boron species at lower pH, thus boron rejection should not show much charge dependence in this case. However, the defect-free KRO-1 consistently exhibited much higher boron rejection despite having a much less negative surface charge across all pH conditions tested compared to both Toray

UTC-80B and UTC-80LB membranes. Furthermore, KRO-1 retained the best boron rejection of all membranes in this test series even at pH 6 regardless of its moderate negative surface charge, suggesting a tighter membrane structure in accordance with its defect-free nature.

Table 5.5. Membrane characterization data for commercial membranes. RR measurements were determined using AFM.

Charge (mV) Membrane Roughness ratio (RR) pH 6 pH 8 pH 10

183

300s-0.1TMC-100C (KRO-1) -11.5 -17.0 -19.8 3.2

RO4 -2.1 -8.1 -15.8 1.5

UTC-80LB -20.1 -38.8 -48.2 1.8

UTC-80B -15.3 -28.0 -41.4 1.7

XR -11.0 -11.8 -12.1 1.1

184

Figure 5.7. Variation of surface charge and boron selectivity at: a) pH 8, and; b) pH 6.

Crossflow separation experiments performed on the best performing commercial membrane in this study, Toray UTC-80LB, and the optimized in-house fabricated membrane highlight the separation capabilities of the defect-free KRO-1 with 99% rejection for boron at pH 10, as shown in Figure 5.5b. On the other hand, the Toray UTC-

80LB membrane showed a maximum rejection of 95% at pH 10. This 4% increase in rejection compared to UTC-80LB represents a significant increase in permselectivity performance. Similar to trends observed in the dead-end testing mode, the KRO-1 TFC membrane consistently demonstrated higher boron rejection in crossflow mode compared to Toray UTC-80LB membrane at the lower pH 6 and 8. For both types of membranes, crossflow mode displayed higher boron rejection compared to the dead-end mode testing, as illustrated in Figure 5.5a and 5.5b. Increase of rejection in crossflow mode compared to dead-end mode is well-known and results from the reduction of concentration polarization on the membrane surface. Performance can potentially be further enhanced when membranes are tested in commercial spiral wound modules due to flow optimizations when compared to lab-scale crossflow systems [45]. Furthermore, it is noteworthy that pKa of

185 boric acid decreases with increasing salt concentration in seawater feed. The higher ionic strength of the feed solution could lead to enhanced boron rejection due to the increased amount of borate ions under the same pH condition. However, the impact of increasing

NaCl concentration in the feed on boron rejection can be highly dependent on the testing conditions [46,47]. In this study, we measured sodium chloride and boron rejection independently to decouple such effects. Meanwhile, increasing operating pressure always results in increased boron rejection [24] which must be considered when comparing membranes tested in this study under brackish conditions to those tested at seawater RO conditions (Table 5.2).

Rejection is typically used as a performance indicator for RO membranes but it is often misleading in characterizing the intrinsic salt transport properties. Here, similar to gas permeation studies, selectivity of water over boron (PW/PBoron) was used to characterize intrinsic separation performance of membrane materials as illustrated in Figure 5.6b-d.

It is apparent that the defect-free KRO-1 membrane exhibited the highest selectivity for water over boron compared to all tested commercial membranes across all pH values and under both dead-end and crossflow testing modes as shown in Figure 5.6b-d. Defect-free

KRO-1 TFC displayed an average of two to five times higher selectivity for water over boron compared to commercial membranes in dead-end operation at pH 10. In crossflow mode, defect-free KRO-1 demonstrated water over boron selectivity of 8300 compared to

1700 for the highest performing commercial membrane (Toray UTC-80LB) at pH 10. At lower pH conditions, the differences of water over boron selectivity among the defect-free

KRO-1 TFC membrane and four commercial membranes were less pronounced but the water over boron selectivity of defect-free KRO-1 TFC remained the highest even at pH 6.

186

The influence of membrane surface charge would have minimal effects on boron rejection at pH 6 due to the predominant presence of boric acid, therefore, the highest water over boron selectivity indicates that the defect-free KRO-1 TFC membrane exhibits very strong size-sieving properties in line with its excellent gas separation properties reported in our previous study [31]. Notably, the crossflow mode showed significant increase in water over boron selectivity compared to dead-end mode as discussed previously. On the other hand, the increased water/boron selectivity of the defect-free membrane was coupled with a three-fold drop in water flux relative to that of the Toray UTC-80LB. This trend follows the water permeance/selectivity trade-off for desalination membranes as proposed by Geise et al. in Freeman’s group [48]. Because the chemical composition and fabrication methods for the commercial membranes are proprietary, direct conclusions cannot be made between the structure-property relationships of all tested commercial membranes.

Figure 5.8. Effect of TMC concentration on water permeance and rejection.

To gain further insight into structure property relationships for FT-30-type MPD-TMC membranes, an extra fabrication variable, TMC concentration, was examined. As the TMC

187 concentration was increased from 0.1 to 1 wt/vol%, a very small decrease was observed in the water permeance, from 0.71 to 0.68 LMH/bar, and sodium chloride rejection, from

99.2% to 98.9%, as shown in Figure 5.8. As TMC concentration was increased to 10 wt/vol%, water permeance increased slightly to 0.77 LMH/bar but with a significant drop in rejection to 95%. It is important to note that as TMC concentration was increased, a significant decrease in roughness ratio was observed, as shown in Table 5.4. This implies an underestimation of the calculated permeance for increasing TMC concentration.

Increasing the TMC concentration in the organic phase increases the ratio of the reactive

COCl groups in the interfacial reaction zone compared to reactive NH2 groups. This excess of COCl groups promotes the formation of more linear, lesser crosslinked polyamide chains, which is confirmed by the crosslinking ratio [38]. The crosslinking ratio decreased from 0.79 to 0.74 and 0.66 as TMC concentration was increased from 0.1 to 1 and 10 wt/vol%, respectively. The decrease in crosslinking ratio with increasing TMC concentration was coupled with an expected decrease in rejection [40–42].

188

A significant change in surface morphology was observed by varying the TMC concentration, as illustrated in Figure 5.9. As TMC concentration was increased from 0.1 to 1 wt/vol %, membrane roughness decreased, i.e., roughness ratio was reduced from 2.1 to 1.9. The surface of the membrane started to flatten and ellipsometry data showed a significant decrease in membrane thickness from 231 to 139 nm. Thin "veils" of polyamide, which are almost transparent, started to appear over the initial thin film. These veils are only a few nm in thickness and very fragile (i.e., sensitive to the electron beam). As the concentration was increased to 10 wt/vol%, this effect appeared further exaggerated. Top surface images show almost a flat structure, presumably due to the growth of the veils, partially covered by thin-walled ring-like structures. The apparent membrane thickness further decreased to about 20 nm.

189

It was observed that, for fixed reaction time, a clear trend starts to emerge for membrane roughness with respect to the crosslinking N/O ratio, as shown in Figure 5.10a. RR shows a linearly increasing trend with increasing crosslinking ratio. MPD-TMC-based polyamide membranes are typically heterogeneous in nature with the polyamide layer consisting of large voids surrounded by thin polyamide layers [49,50]. The layers are formed by MPD diffusing from the aqueous into the organic phase and reacting with TMC molecules.

Furthermore, formation of the MPD-TMC polyamide is an exothermic reaction [43].

Hence, as the reaction progresses to produce and crosslink the polyamide layer, heat is added to the reaction zone. The heat promotes turbulence at the interface due to increase in molecular kinetic energy. This leads to aggressive diffusion of MPD into the organic phase, which ultimately reacts with TMC to form the polyamide. This can explain the correlation between surface roughness and increased degree of crosslinking. Increasing the organic solution temperature adds further heat to the reaction zone which forms the extremely rough membranes produced at 100 °C organic solution temperature. As a result,

190 the KRO-1 (300s-0.1TMC-100C) membrane displayed the best salt and boron rejection of all tested membranes.

Figure 5.11. Surface SEM images of commercial membranes.

191

Figure 5.12. Surface AFM images of commercial membranes.

In this study, water permeance showed no clear correlation with the crosslinking ratio. It is well understood that membrane permeance varies significantly with membrane thickness as well as the membrane roughness. Because only qualitative estimates of thickness were made in this study, no quantitative conclusions can be drawn about variation of intrinsic A with the crosslinking N/O ratio. On the other hand, it is well known from gas permeation studies that for defect-free membranes, selective layer thickness and surface area have little impact on membrane selectivity. Plotting Pw/Ps vs. crosslinking N/O ratio, as shown in

Figure 5.10b, confirms a clear linear trend between increasing degree of crosslinking and

192 membrane water over sodium chloride selectivity. An increase in crosslinking ratio results in restriction of the polymer chain mobility, which, in turn, decreases water-induced polymer swelling and, thereby, enhances the size sieving properties of the polyamide layer.

Figure 5.13. Roughness ratio vs. water over boron selectivity (Pw/PBoron) for KRO-1 and commercial RO membranes tested at pH 10.

Surface roughness of membranes tested for boron rejection experiments is illustrated in

Figure 5.11-5.12 and quantified in Table 5.4. It can be seen, in Figure 5.13, that membrane roughness ratio correlates much stronger with the membrane PW/PBoron compared to the membrane surface charge. This implies that the highly rejecting commercial RO membranes may have a higher crosslinking ratio compared to lower rejecting membranes.

Furthermore, recent research [50] showed that the pH dependent performance of seawater polyamide RO membranes cannot be correlated with the pH dependence of the membrane charge. Similarly, in this study the brackish water type Koch XR membrane showed only

193 a small increase in surface charge, from -11 to -12 mV, as pH was increased from 6 to 10 while its rejection increased significantly from 55 to 77%. The lowest boron rejection performance of the XR membrane was also found to be associated with its smoother surface roughness feature, as shown in Figure 5.12. Here it is more likely that the increase of boron rejection with increasing pH can be attributed primarily to the increased number of much larger borate ions, which are rejected significantly more by highly crosslinked membranes with rougher surfaces and stronger size sieving capabilities such as the defect- free KRO-1 (300s-0.1TMC-100C) membrane reported in this study [18,21,46].

5.4. Conclusions

Defect-free, highly selective MPD-TMC membranes were successfully fabricated showing potential for removal of boron and sodium chloride from water. The membranes exhibited improved boron rejection capabilities in crossflow operation, i.e., up to 99% rejection at pH 10, higher than the commercial membranes tested in this study for comparison highlighting the impact of healing defects in FT-30-type RO membranes. The best performing commercial membrane in this study, Toray UTC-80LB, showed a maximum rejection of 95% under identical test conditions at pH 10. The membranes also exhibited improved sodium chloride rejection, reaching a maximum of 99.6% at lab-scale brackish water RO test conditions, compared to the commercial FT-30 variant Sepro RO4 membrane with a maximum rejection of 97.9%.

It was demonstrated that in-situ plugging of defects in FT-30-type RO membranes can significantly improve performance by simply controlling the reaction time, e.g., rejection improved from 97.9 to 99.2% when the reaction time was increased from 10 to 300 s. No

194 change was observed in the degree of crosslinking of the polyamide layer as function of reaction time, which implied a pore plugging mechanism at play rather than changes in the intrinsic polymer microstructure. Further benefits were achieved by increasing the membrane crosslinking ratio, e.g., increasing organic solution temperature to 100 °C resulted in maximum sodium chloride rejection of 99.6%. Although, the defect-free high boron rejection membranes displayed lower water flux compared to commercially available desalination membranes, the improved selectivity addresses a critical need for future development of such membranes [51]. Importantly, these membranes can be fabricated with small variations to the commercially used RO membrane manufacturing process enabling low-cost production as well as rapid scalability and reliability.

Furthermore, the correlation found in this study between the high crosslinking ratio with rougher and tighter membrane structure and high water over boron selectivity provides an improved understanding of the structure-property relationship of FT-30-type RO membranes and guidance toward development of high rejection RO membranes.

195

5.5. References

[1] M.A. Shannon, P.W. Bohn, M. Elimelech, J.G. Georgiadis, B.J. Mariñas, A.M. Mayes, Science and technology for water purification in the coming decades, Nature 452 (2008) 301–310.

[2] J. Kim, H. Hyung, M. Wilf, J.-S. Park, J. Brown, Boron rejection by reverse osmosis membranes: National reconaissance and mechanism study, Denver, Colorado, 2009.

[3] N. Misdan, W.J. Lau, A.F. Ismail, Seawater reverse osmosis (SWRO) desalination by thin-film composite membrane—Current development, challenges and future prospects, Desalination 287 (2012) 228–237.

[4] J.E. Cadotte, Interfacially synthesized reverse osmosis membrane, US 4,277,344 A, 1981.

[5] M.J.T. Raaijmakers, N.E. Benes, Current trends in interfacial polymerization chemistry, Prog. Polym. Sci. 63 (2016) 86–142.

[6] F.S. Kot, Boron in the Environment, in: Boron Sep. Process., Elsevier B.V., 2015: pp. 1–33.

[7] J.K. Fawell, Boron in drinking-water. Background document for development of WHO guidelines for drinking water quality (GDWQ), Geneva, 2009.

[8] E.V. Grieve, C.M., Grattan, S.R., Maas, Plant salt tolerance in agricultural salinity assessment and management, Am. Soc. Civ. Eng. (2012) 405–459.

[9] R.O. Nable, G.S. Bañuelos, J.G. Paull, Boron toxicity, Plant Soil 193 (1997) 181– 198.

[10] C.D. Hunt, C. Benjamin, Encyclopedia of Food Sciences and Nutrition, Oxford Academic Press, Oxford, 2003.

[11] R. Nightingale, Phenomenological theory of ion solvation, Effective radii of hydrated ions, J. Phys. Chem. 63 (1959) 1381–1387.

[12] R. Bernstein, S. Belfer, V. Freger, Toward improved boron removal in RO by membrane modification: feasibility and challenges., Environ. Sci. Technol. 45 (2011) 3613–20.

[13] L.A. Mel’nik, The current state of the problem of removing boron from the sea and brackish waters in the process of reverse osmosis desalination, J. Water Chem. Technol. 32 (2010) 311–318.

[14] V.S. Freger, H. Shemer, A.A. Sagiv, R.R. Semiat, Boron Removal Using Membranes, in: Boron Sep. Process., Elsevier, Haifa, Israel, 2015: pp. 199–217.

196

[15] D. Prats, M.F. Chillon-Arias, M. Rodriguez-Pastor, Analysis of the influence of pH and pressure on the elimination of boron in reverse osmosis, Desalination 128 (2000) 269–273.

[16] M. Rodríguez Pastor, A. Ferrándiz Ruiz, M.F. Chillón, D. Prats Rico, Influence of pH in the elimination of boron by means of reverse osmosis, Desalination 140 (2001) 145–152.

[17] M. Taniguchi, M. Kurihara, S. Kimura, Boron reduction performance of reverse osmosis seawater desalination process, J. Membr. Sci. 183 (2001) 259–267.

[18] H. Koseoglu, N. Kabay, M. Yüksel, S. Sarp, Ö. Arar, M. Kitis, Boron removal from seawater using high rejection SWRO membranes - impact of pH, feed concentration, pressure, and cross-flow velocity, Desalination 227 (2008) 253–263.

[19] Y. Cengeloglu, G. Arslan, A. Tor, I. Kocak, N. Dursun, Removal of boron from water by using reverse osmosis, Sep. Purif. Technol. 64 (2008) 141–146.

[20] C. Dominguez-Tagle, V.J. Romero-Ternero, A.M. Delgado-Torres, Boron removal efficiency in small seawater reverse osmosis systems, Desalination 265 (2011) 43– 48.

[21] S. Shultz, M. Bass, R. Semiat, V. Freger, Modification of polyamide membranes by hydrophobic molecular plugs for improved boron rejection, J. Membr. Sci. 546 (2018) 165–172.

[22] A. Ruiz-García, F.A. León, A. Ramos-Martín, Different boron rejection behavior in two RO membranes installed in the same full-scale SWRO desalination plant, Desalination 449 (2019) 131–138.

[23] M. Taniguchi, Y. Fusaoka, T. Nishikawa, M. Kurihara, Boron removal in RO seawater desalination, Desalination 167 (2004) 419–426.

[24] K.L. Tu, L.D. Nghiem, A.R. Chivas, Boron removal by reverse osmosis membranes in seawater desalination applications, Sep. Purif. Technol. 75 (2010) 87–101.

[25] J. Redondo, M. Busch, J.-P. De Witte, Boron removal from seawater using FILMTEC high rejection SWRO membranes, Desalination 156 (2003) 229–238.

[26] Y. He, Y.P. Tang, T.S. Chung, Concurrent removal of selenium and arsenic from water using polyhedral oligomeric silsesquioxane (POSS)-polyamide thin-film nanocomposite nanofiltration membranes, Ind. Eng. Chem. Res. 55 (2016) 12929– 12938.

[27] S. Wang, Y. Zhou, C. Gao, Novel high boron removal polyamide reverse osmosis membranes, J. Membr. Sci. 554 (2018) 244–252.

[28] S. Shultz, V. Freger, In situ modification of membrane elements for improved boron

197

rejection in RO desalination, Desalination 431 (2018) 66–72.

[29] J.S. Louie, I. Pinnau, M. Reinhard, Gas and liquid permeation properties of modified interfacial composite reverse osmosis membranes, J. Membr. Sci. 325 (2008) 793– 800.

[30] J. Albo, J. Wang, T. Tsuru, Gas transport properties of interfacially polymerized polyamide composite membranes under different pre-treatments and temperatures, J. Membr. Sci. 449 (2014) 109–118.

[31] Z. Ali, F. Pacheco, E. Litwiller, Y. Wang, Y. Han, I. Pinnau, Ultra-selective defect- free interfacially polymerized molecular sieve thin-film composite membranes for H2 purification, J. Mater. Chem. A 6 (2017) 30–35.

[32] H. Hyung, J.H. Kim, A mechanistic study on boron rejection by sea water reverse osmosis membranes, J. Membr. Sci. 286 (2006) 269–278.

[33] J.S. Louie, I. Pinnau, M. Reinhard, Effects of surface coating process conditions on the water permeation and salt rejection properties of composite polyamide reverse osmosis membranes, J. Membr. Sci. 367 (2011) 249–255.

[34] Q. Zhang, Z. Zhang, L. Dai, H. Wang, S. Li, S. Zhang, Novel insights into the interplay between support and active layer in the thin film composite polyamide membranes, J. Membr. Sci. 537 (2017) 372–383.

[35] J. Duan, E. Litwiller, I. Pinnau, Solution-diffusion with defects model for pressure- assisted forward osmosis, J. Membr. Sci. 470 (2014) 323–333.

[36] P.V.X. Hung, S.-H. Cho, S.-H. Moon, Prediction of boron transport through seawater reverse osmosis membranes using solution–diffusion model, Desalination 247 (2009) 33–44.

[37] M. Hirose, H. Ito, Y. Kamiyama, Effect of skin layer surface structures on the flux behaviour of RO membranes, J. Membr. Sci. 121 (1996) 209–215.

[38] A.K. Ghosh, B.H. Jeong, X. Huang, E.M.V. Hoek, Impacts of reaction and curing conditions on polyamide composite reverse osmosis membrane properties, J. Membr. Sci. 311 (2008) 34–45.

[39] S.H. Kim, S.Y. Kwak, T. Suzuki, Positron annihilation spectroscopic evidence to demonstrate the flux-enhancement mechanism in morphology-controlled thin-film- composite (TFC) membrane, Environ. Sci. Technol. 39 (2005) 1764–1770.

[40] I.J. Roh, Influence of rupture strength of interfacially polymerized thin-film structure on the performance of polyamide composite membranes, J. Membr. Sci. 198 (2002) 63–74.

[41] I.J. Roh, V.P. Khare, Investigation of the specific role of chemical structure on the

198

material and permeation properties of ultrathin aromatic polyamides, J. Mater. Chem. 12 (2002) 2334–2338.

[42] I.J. Roh, J.J. Kim, S.Y. Park, Mechanical properties and reverse osmosis performance of interfacially polymerized polyamide thin films, J. Membr. Sci. 197 (2002) 199–210.

[43] B. Ukrainsky, G.Z. Ramon, Temperature measurement of the reaction zone during interfacial polymerization, J. Membr. Sci. 566 (2018) 329–335.

[44] A.E. Childress, M. Elimelech, Effect of solution chemistry on the surface charge of polymeric reverse osmosis and nanofiltration membranes, J. Membr. Sci. 119 (1996) 253–268.

[45] R.W. Baker, Membrane Technology and Applications, 3rd ed, Wiley, Chichester, UK., 2012.

[46] M.H. Oo, S.L. Ong, Implication of zeta potential at different salinities on boron removal by RO membranes, J. Membr. Sci. 352 (2010) 1–6.

[47] M.H. Oo, S. Lianfa, Effect of pH and ionic strength on boron removal by RO membranes, Desalination 246 (2009) 605–612.

[48] G.M. Geise, H.B. Park, A.C. Sagle, B.D. Freeman, J.E. McGrath, Water permeability and water/salt selectivity tradeoff in polymers for desalination, J. Membr. Sci. 369 (2011) 130–138.

[49] F.A. Pacheco, I. Pinnau, M. Reinhard, J.O. Leckie, Characterization of isolated polyamide thin films of RO and NF membranes using novel TEM techniques, J. Membr. Sci. 358 (2010) 51–59.

[50] N. Fridman-Bishop, V. Freger, What makes aromatic polyamide membranes superior: New insights into ion transport and membrane structure, J. Membr. Sci. 540 (2017) 120–128.

[51] J.R. Werber, A. Deshmukh, M. Elimelech, The critical need for increased selectivity, not increased water permeability, for desalination membranes, Environ. Sci. Technol. Lett. 3 (2016) 112–120.

199

Chapter 6. Triptycene-based Interfacially Polymerized Thin-Film Composite Polyamide Membranes for Liquid- and Gas Separations

The chapter deals with the fabrication of interfacially polymerized thin-film composite membranes using a highly kinked building block, similar to the strategy taken to fabricate polymers of intrinsic microporosity.

6.1. Abstract

Substitution of traditionally used monomers in interfacial polymerization with highly contorted, kinked building blocks (similar to those used for preparation of polymers of intrinsic microporosity) can help improve permeance performance of interfacially polymerized thin-film composite (TFC) membranes. In this work, we report the successful fabrication of interfacially polymerized TFC polyamide membranes integrating a novel bridged-bicyclic triptycene-based tetra-acyl chloride (TripTaC) building block to replace the conventional trimesoyl chloride (TMC) moiety. Membranes prepared with the TripTaC building block (labeled MPDTrip) showed improved permeance performance in both liquid- and gas separation systems. Integration of the TripTaC moiety significantly increased liquid permeance, e.g., MPDTrip-20 showed water and methanol permeance of

9.2 and 9.1 LMH/bar, respectively, compared to 3.6 and 4.8 LMH/bar for MPDTMC-20.

Both MPDTrip TFCs showed elevated performance for separating small solutes from methanol, i.e., both membranes exhibit > 98% rejection for Sudan Orange G dye in methanol surpassing a number of previously reported state-of-the-art TFCs. The fabricated

TFCs show promising data for use in the reverse osmosis process with ~96% sodium chloride rejection and ~99% rejection for divalent salts under brackish water conditions.

200

For gas separation, MPDTrip-100 exhibited ~3-fold higher permeance of H2 (~55 GPU) compared to MPDTMC-based TFCs (~20 GPU) with reasonable pure-gas H2/CO2 selectivity of ~9 at 22 °C. Broad characterization of TFCs and polymer powders using free- volume modeling, the Brunauer, Emmett, and Teller (BET) technique, X-ray photoelectron spectroscopy, scanning electron microscopy, atomic force microscopy, ellipsometry, wide- angle X-ray diffraction, vapor sorption and gas sorption revealed good agreement of TFC performance with physical characteristics for both gas and liquid transport.

————————————————————————————————————

This chapter is under preparation as:

Z. Ali, B. Ghanem, Y. Wang, F. Pacheco, W. Ogieglo, G. Genduso, H, Vovusha, U. Schwingenschlogl, Y. Han and I. Pinnau, Triptycene-based Interfacially Polymerized Thin-Film Composite Polyamide Membranes for Liquid- and Gas Separations.

201

6.2. Introduction

Industrial chemical separation processes account for ~10-15% of the world’s total energy consumption [1]. Diminishing fossil fuel resources and growing environmental concerns have pushed the world to seek sustainable and energy efficient alternative solutions.

Implementation of membrane technology provides a major opportunity to perform less energy-intensive liquid- and gas separations compared to conventional technologies due to its simple continuous operation with modular design and small footprint. This is evidenced by widespread successful commercial applications of reverse osmosis for desalination and membrane-based gas separations. However, lack of development of high-performance membranes has presented a major hurdle for membrane technology to further compete with conventional separation technologies [2].

Aromatic crosslinked polyamides are classified as high-performance materials for a wide range of applications due their exceptional mechanical strength, thermal resistance and chemical stability [3]. Thin-film composite polyamides have shaped the modern membrane desalination industry but development of next generation of polyamide composites has been limited [4]. Polymers of intrinsic microporosity (PIMs) are an emerging group of amorphous microporous materials gaining significant attention in membrane-based gas separation due to their ability to transcend the conventional permeability/selectivity tradeoff relationships [5]. Such materials exhibit high free volumes as a result of inefficient chain packing due to the integration of highly rigid and contorted building blocks including

Tröger’s base, ethanoanthracene, spirobisindane and triptycene [6–9]. Triptycene, a member of the iptycene family, has been established as a remarkable building block for

PIMs owing to its 3D paddlewheel-like robust, rigid and contorted structure [10–12].

202

Integration of the triptycene moiety in polymers has led to the design of high-performance

PIMs including KAUST-PIs, TPIMs, and PIM-Trip-TB, defying the 2015 polymer upper bound for a number of important gas pairs such as O2/N2, H2/CH4, and H2/N2 [10,13,14].

However, technical challenges associated with scale-up of PIM synthesis, and, most importantly, fabricating defect-free, inexpensive, thin-film composite membranes have limited their industrial use. Thin-film fabrication using interfacial polymerization (IP) offers a potential solution for such limitations due to its ability to produce highly crosslinked, reproducible, ultrathin-films employing an industrially scalable, low-cost, roll-to-roll process [15]. Highly crosslinked, chemically and thermally stable network polyamides can directly be fabricated on customizable porous supports using interfacial polymerization to merge polymer synthesis, crosslinking and thin-films fabrication into a simple one-step, scalable, membrane production process.

In this work, we report the first successful fabrication of interfacially polymerized, defect- free polyamide thin-film composite membranes using a novel bridged-bicyclic triptycene- based tetra-acyl chloride (TripTaC) building block with m-phenylene diamine (MPD) as well as the conventional m-phenylene diamine/trimesoyl chloride (TMC) polyamide chemistry for comparison. Solvent permeation, aqueous and organic dye rejection, salt desalination employing reverse osmosis (RO) and, gas separation performance are reported. The samples were characterized using scanning electron microscopy (SEM), ellipsometry, atomic force microscopy (AFM), molecular dynamics simulations, X-ray photoelectron spectroscopy (XPS), vapor sorption, X-ray diffraction (XRD), and Fourier transform infrared spectroscopy (FTIR).

203

6.3. Results and Discussion

6.3.1. Interfacial Polymerization using a triptycene building block

204

TripTaC as 4700 Å3. Blue color: accessible surface at probe radius of 1 Å. (d) Sorption behavior using CO2 BET measurements at 273 K up to 1 bar for MPDTMC and MPDTrip powder. (e) XRD spectra of MPDTMC and MPDTrip powders with average d-spacing obtained using Bragg’s Law.

Figure 6.1a shows the reaction scheme between the triptycene-1,3,6,8-tetraacetyl chloride

(TripTaC) monomer with MPD to fabricate a fully aromatic, highly crosslinked network polyamide (polyaramid). The 3D structure of the triptycene moiety, comprising of a

[2,2,2]-tricyclic ring system with three benzene groups fused at 120°, entraps internal free volume (IFV) when compared to the planar trimesoyl chloride moiety. Molecular dynamics simulations (Figure 6.1b-c) demonstrated that the substitution of the TMC with the triptycene moiety permitted a significant increase of the polymer fractional free volume

(from ~14 to 24%) and a higher interconnectivity between the free volume units.

Figure 6.2. Pore size distribution for MPDTMC and MPDTrip powder obtained from CO2 isotherms employing the non-linear density function theory (NLDFT). TGA analysis of MPDTMC and MPDTrip powder.

BET surface area and pore size distribution of MPDTMC and MPDTrip polymer powders was probed using CO2 sorption at 273 K which showed an increase in BET surface area by

2 -1 2-fold (from 49 to 92 m g based on CO2 sorption at 273 K) when employing the TripTaC monomer in place of TMC (Figure 6.1d). Furthermore, pore size distribution based on non-

205 linear density function theory (NLDFT) showed increased portions of ultramicropores in

MPDTrip polyamide powder compared to its MPDTMC counterpart (Figure 6.2a). XRD spectra (Figure 6.1e) demonstrated that the average chain spacing of the main amorphous peak increased from 3.55 Å (MPDTMC) to 3.9 Å (MPDTrip), confirming the trend observed in BET surface area measurements based on CO2 sorption and fractional free volume and density calculation from molecular dynamic simulations. Thermogravimetric analysis of MPDTrip powder (Figure 6.2b) demonstrated good thermal stability up to 300

°C, comparable to MPTMC polyamide, implying potential to develop high performance

TFC membranes with chemical and thermal stability for high temperature applications.

Table 6.1. Experimental conditions and properties of thin-film composite (TFC) membranes prepared in this study.

Table 6.1 summarizes the conditions for membrane preparation and their physical properties discussed in this work. All membrane types were tested with FTIR spectroscopy to confirm the formation of polyamide layers (Figure 6.3). Figure 6.4a shows that

MPDTrip-100 exhibited a ridge-and-valley structure commonly observed for conventional

MPDTMC network polymers. Figure 6.4b shows the loose structure of the top surface of

MPDTrip-100 membranes that is composed of large voids trapped within thin polyamide layers [16]. Membranes fabricated at 20 °C displayed partial surface coverage by relatively

206 smooth network polymer powder-like features on the top of the ridge-and-valley assemblies (Figure 6.4c i-iv). On both the micron- (AFM characterization, Figure 6.4d) and millimeter scales (via ellipsometry, Figure 6.4e) MPDTrip-20 displayed higher surface roughness as compared to MPDTMC-20 because of the presence of thicker and more rounded features on the top surface.

Figure 6.3. FTIR spectra for PAN support and TFCs fabricated in this study.

207

Figure 6.4. Imaging of interfacially polymerized TFC membranes. (a) Cross-section SEM image of defect-free MPDTrip-100 membrane highlighting structural features. (b) High magnification SEM image of MPDTrip-100 TFC emphasizing predicted actual active layer thickness and support/selective layer interface. (c) SEM top surface images depicting the TFCs produced. Samples produced using an organic solution temperature of 20 °C show two distinct top layer structures as highlighted by insets i-iv. (d) 3D visualization of MPDTMC-20 and MPDTrip-20 surface using AFM. (e) Ellipsometry images highlighting variance in surface morphology over larger sample areas, i.e., 5x5 mm for MPDTMC-20 and MPDTrip-20.

208

6.3.2. Fluid Transport Behavior

The substitution of TMC with the TripTaC monomer enhanced liquid permeance for all tested solvents as shown in Table 6.2 (in good agreement with the variation of free volume described previously). For membranes prepared at 20 °C, the replacement of TMC with

TripTaC increased the water permeance from 3.4 to 9.2 LMH/bar (~170% increase) and ethanol permeance from 1.0 to 2.6 LMH/bar (~160% increase), as shown in Figure 6.5a.

Methanol permeance increased from 4.8 to 9.1 LMH/bar (90% increase) whereas isopropanol permeance soared from 0.05 to 0.36 LMH/bar (600% increase). At a relative pressure of 0.95, the organic vapor sorption of bulk powders revealed 30% increase in water uptake (from 13 to 17.4 mmol g-1) in the MPDTrip polymer compared to MPDTMC as shown in Figure 6.6. Meanwhile, MPDTrip methanol uptake decreased by 25% (7.7 to

5.8 mmol g-1) compared to MPDTMC. Hence, the 170% and 90% increase in water and methanol permeance, respectively, is potentially due to increased diffusivity of the solvents through the MPDTrip membranes.

Table 6.2. Pure solvent permeation data for fabricated TFCs in LMH/bar.

Sample H2O MeOH EtOH IPA MPDTMC-20 3.40 4.82 0.99 0.05 MPDTMC-100 2.36 2.37 0.10 0.02 MPDTrip-20 9.15 9.12 2.57 0.36 MPDTrip-100 4.08 8.72 2.38 0.34

209

Figure 6.5. Liquid separation properties of the membranes tested in this study. (a) Pure- solvent permeance through fabricated TFCs vs. the kinetic diameters of the tested solvents. (b) Rejection of Brilliant Blue R (826 g mol-1) through fabricated TFCs in water, methanol, ethanol and isopropanol. (c) Rejection of Sudan Orange (216 g mol-1) through fabricated TFCs in water, methanol, ethanol and isopropanol. (d) Permeance of methanol vs selectivity for solutes between 210-250 g mol-1. Red stars denote MPDTrip-20 membranes from this study using Sudan Orange (216 g mol-1), blue circles denote MPDTMC membranes from this study using Sudan Orange (216 g mol-1), yellow squares denote data obtained from Solomon et al. [17] using Chrysoidine G (249 g mol-1), green triangles denote data obtained from Karan et al. [18] using 6-hydroxy- 2-naphthalenesulfonic acid sodium salt (HNSA) dye (246 g mol-1), light blue triangle denotes data obtained from Yang et al. [19] using Chrysoidine G (249 g mol-1) and gray triangles denote data obtained from Sorribas et al. [20] using polystyrene oligomers (230 g mol-1). (e) Sodium chloride rejection comparison of MPDTrip-20 fabricated in this study with state-of-the-art membranes.

210

Figure 6.6. Vapor uptake for chosen solvents for: (a) MPDTMC powder, and (b) MPDTrip powder.

All prepared TFCs in this study completely rejected Brilliant Blue R (BB) dye (826 g mol-

1) in water (Table 6.3). When the same experiment was carried out employing alcohols,

BB rejection by the TFCs decreased as the carbon-content of the solvent increased. In particular, MPDTMC-20 rejection of BB in methanol slightly decreased to about 99.3% whereas in ethanol and isopropanol the rejection decreased to ~98.8% (Figure 6.5b).

Similarly, MPDTrip-20 BB rejection decreased to 98% in methanol while maintaining a high rejection of ~99% in both ethanol and isopropanol. Sudan Orange G (SO) dye (214 g mol-1) rejections were also tested in organic solvents (Table 6.4). Compared to MPDTMC-

20, MPDTrip-20 exhibited higher rejections in both methanol and isopropanol, i.e., 99.1 vs. 97.9% and 99.2 vs. 98.7%, respectively (Figure 6.5c). Trends for both dyes were consistent for membranes fabricated at 100 °C.

211

Table 6.3. Brilliant Blue R (826 g mol-1) rejection data for fabricated TFCs in %.

Sample H2O MeOH EtOH IPA MPDTMC-20 99.9 99.3 99.2 98.1 MPDTMC-100 99.9 94.2 85.6 80.2 MPDTrip-20 99.9 98.0 99.2 99.2 MPDTrip-100 99.9 98.2 98.4 98.7

Table 6.4. Sudan Orange G (216 g mol-1) rejection data for fabricated TFCs in %.

Sample MeOH EtOH IPA MPDTMC-20 97.9 98.9 98.7 MPDTMC-100 87.5 92.5 93.1 MPDTrip-20 99.1 98.9 99.2 MPDTrip-100 98.2 97.9 98.5

In comparison with MPDTMC, MPDTrip membranes performed particularly well when methanol was used as the solvent to reject the SO dye. Substitution of TMC with TripTaC improved both methanol permeance and reduced SO flux for both fabrications conditions

— for example, MPDTrip-20 showed 90% increased methanol flux and 50% lower SO flux compared to MPDTMC-20. This could be an effect of increased amide linkages and/or carboxyl groups in the MPDTrip membranes compared to MPDTMC which can form hydrogen bonds with the resonation forms of the SO dye (see 6.5c inset), thus promoting absorption of the dye in the polyamide layer and improving rejection. Filtration performance of MPDTrip-20 membranes with methanol can be evaluated via the permeance/selectivity trade-off (displayed in Figure 6.5d) in comparison to state-of-the-art membranes. MPDTrip polyamide membranes exceeded the selectivity performance of ultrathin MPDTMC nanofilms (< 10 nm) synthesized on cadmium hydroxide nanostrands supports while matching their permeance [18]. Furthermore, MPDTrip-20 membranes

212 showed significantly higher permselectivity compared to MOF-integrated TFC polyamides as well as previously fabricated high-free-volume thin-film polyarylates (intended for organic nanofiltration applications) that currently remain the most successful efforts to introduce kinked building blocks in the interfacial polymerization membrane fabrication process [17,21].

Table 6.5. Salt rejection data for fabricated TFCs in %.

Sample NaCl MgSO4 MgCl2 Na2SO4 MPDTMC-20 98.4 99.5 99.5 99.7 MPDTMC-100 97.9 98.9 99.1 99.0 MPDTrip-20 95.5 99.2 97.3 98.5 MPDTrip-100 95.5 99.2 97.2 99.4

MPDTrip membranes displayed excellent stability and separation performance in water/salt separations. MPDTrip-20 showed reasonably high sodium chloride rejection of

~95.5% with a water permeance 9.2 LMH/bar respectively. Comparatively, Dow Filmtec

NF270 and recently fabricated triazine-piperazine membranes exhibit water permeance between 9 - 10 LMH/bar with sodium chloride rejection between 30 to 40% (Figure 6.5e)

[22,23]. MPDTrip membranes also demonstrated exceptional performance for rejection of divalent salts as shown in Table 6.5.

213

Figure 6.7. (a) Pure-gas permeance of fabricated TFC membranes with respective to the kinetic diameters of the tested gases. (b) Isosteric heat of sorption obtained from CO2 sorption isotherms at 273 K and 298 K.

MPDTrip-100 demonstrated 3-fold increase in H2 permeance (56 vs. ~20 GPU) compared to MPDTMC membranes (Figure 6.7a) as well as enhanced H2/CO2 selectivity (9 vs. 7) compared to MPDTMC-20 but lower than that of MPDTMC-100. Increased CO2 flux through MPDTrip-100 was attributed to increased interaction between the MPDTrip polyamide and CO2 molecules as highlighted by Figure 6.7b. The improved performance of MPDTrip membranes across gas, organic solvent and water systems implies great potential of using highly rigid and contorted PIM-type monomers in interfacial polymerization to fabricate high-performance TFC membranes for future liquid- and gas separations.

6.4. Conclusions

Triptycene-based interfacially polymerized polyamide TFCs were successfully fabricated on polymeric supports showing improved performance for a wide range of membrane- based liquid- and gas separations. MPDTrip membranes demonstrated high permselectivity for removal of small and large solutes from water and organic solvents such as methanol,

214 ethanol and isopropanol. MPDTrip-20 showed methanol flux of ~9 LMH/bar with a high rejection for Sudan Orange G dye of 99.1% under dead-end filtration conditions. This translated to an exceptionally high separation factor (~10,300 for MPDTrip-20) with permselectivity performance superior than state-of-the-art membranes fabricated for nanofiltration applications. The membranes also exhibited a water flux of 9 LMH/bar, comparable to high flux nanofiltration membranes but with ~96% rejection of sodium chloride under BWRO conditions, significantly higher than its nanofiltration counterparts.

This was coupled with exceptional divalent ion rejection (e.g. MgSO4 rejection > 99%) in water. Furthermore, MPDTrip-100 demonstrated 3-fold higher H2 permeance compared to its MPDTMC counterparts with a H2/CO2 selectivity of 9.

The improved liquid- and gas separation performance exhibited by MPDTrip membranes coupled with the ability to fabricate such membranes directly on polymeric supports, with minimal variations to the traditional RO manufacturing process highlights their promising potential to fabricate the next generation of high-performance membranes for a wide range of industrial separations including, but not limited to, organic solvent reverse osmosis, nanofiltration, desalination and gas separation.

215

6.5. References

[1] D.S. Sholl, R.P. Lively, Seven chemical separations to change the world, Nature 532 (2016) 6–8.

[2] I. Pinnau, F.A. Pacheco, E. Litwiller, D. Jintang, Elucidation of the microstructure of interfacially polymerized reverse osmosis membranes, in: 8th Sino-US Jt. Conf. Chem. Eng., Shanghai, China, 2015.

[3] J.M. García, F.C. García, F. Serna, J.L. de la Peña, High-performance aromatic polyamides, Prog. Polym. Sci. 35 (2010) 623–686.

[4] N. Fridman-Bishop, V. Freger, What makes aromatic polyamide membranes superior: New insights into ion transport and membrane structure, J. Membr. Sci. 540 (2017) 120–128.

[5] R. Swaidan, B. Ghanem, I. Pinnau, Fine-tuned intrinsically ultramicroporous polymers redefine the permeability/selectivity upper bounds of membrane-based air and hydrogen separations, ACS Macro Lett. 4 (2015) 947–951.

[6] B.S. Ghanem, N.B. McKeown, P.M. Budd, J.D. Selbie, D. Fritsch, High- performance membranes from polyimides with intrinsic microporosity, Adv. Mater. 20 (2008) 2766–2771.

[7] N. Alaslai, B. Ghanem, F. Alghunaimi, I. Pinnau, High-performance intrinsically microporous dihydroxyl-functionalized triptycene-based polyimide for natural gas separation, Polymer (United Kingdom) 91 (2016) 128–135.

[8] F. Alghunaimi, B. Ghanem, N. Alaslai, R. Swaidan, E. Litwiller, I. Pinnau, Gas permeation and physical aging properties of iptycene diamine-based microporous polyimides, J. Membr. Sci. 490 (2015) 321–327.

[9] P.M. Budd, K.J. Msayib, C.E. Tattershall, B.S. Ghanem, K.J. Reynolds, N.B. McKeown, D. Fritsch, Gas separation membranes from polymers of intrinsic microporosity, J. Membr. Sci. 251 (2005) 263–269.

[10] B.S. Ghanem, R. Swaidan, X. Ma, E. Litwiller, I. Pinnau, Energy-efficient hydrogen separation by AB-type ladder-polymer molecular sieves, Adv. Mater. 26 (2014) 6696–6700.

[11] N.T. Tsui, A.J. Paraskos, L. Torun, T.M. Swager, E.L. Thomas, Minimization of internal molecular free volume: A mechanism for the simultaneous enhancement of polymer stiffness, strength, and ductility, Macromolecules 39 (2006) 3350– 3358.

[12] R. Bera, S. Mondal, N. Das, Nanoporous triptycene based network polyamides (TBPs) for selective CO2 uptake, Polymer 111 (2017) 275–284.

216

[13] R. Swaidan, M. Al-Saeedi, B. Ghanem, E. Litwiller, I. Pinnau, Rational design of intrinsically ultramicroporous polyimides containing bridgehead-substituted triptycene for highly selective and permeable gas separation membranes, Macromolecules 47 (2014) 5104–5114.

[14] G. Genduso, B. Ghanem, Y. Wang, I. Pinnau, G. Genduso, B.S. Ghanem, Y. Wang, I. Pinnau, Synthesis and gas-permeation characterization of a novel high- surface area polyamide derived from 1,3,6,8-Tetramethyl-2,7-diaminotriptycene: Towards polyamides of intrinsic microporosity (PIM-PAs), Polymers 11 (2019) 361.

[15] M.J.T. Raaijmakers, N.E. Benes, Current trends in interfacial polymerization chemistry, Prog. Polym. Sci. 63 (2016) 86–142.

[16] F. Pacheco, R. Sougrat, M. Reinhard, J.O. Leckie, I. Pinnau, 3D visualization of the internal nanostructure of polyamide thin films in RO membranes, J. Membr. Sci. 501 (2016) 33–44.

[17] M.F. Jimenez-Solomon, Q. Song, K.E. Jelfs, M. Munoz-Ibanez, A.G. Livingston, Polymer nanofilms with enhanced microporosity by interfacial polymerization, Nat. Mater. 15 (2016) 760–767.

[18] S. Karan, Z. Jiang, A.G. Livingston, Sub-10 nm polyamide nanofilms with ultrafast solvent transport for molecular separation, Science 348 (2015) 1347– 1351.

[19] Q. Yang, Y. Su, C. Chi, C.T. Cherian, K. Huang, V.G. Kravets, F.C. Wang, J.C. Zhang, A. Pratt, A.N. Grigorenko, F. Guinea, A.K. Geim, R.R. Nair, Ultrathin graphene-based membrane with precise molecular sieving and ultrafast solvent permeation, Nat. Mater. 16 (2017) 1198–1202.

[20] S. Sorribas, P. Gorgojo, C. Téllez, J. Coronas, A.G. Livingston, High flux thin film nanocomposite membranes based on metal-organic frameworks for organic solvent nanofiltration, J. Am. Chem. Soc. 135 (2013) 15201–15208.

[21] C. Echaide-Górriz, S. Sorribas, C. Téllez, J. Coronas, MOF nanoparticles of MIL- 68(Al), MIL-101(Cr) and ZIF-11 for thin film nanocomposite organic solvent nanofiltration membranes, RSC Adv. 6 (2016) 90417–90426.

[22] M. Dalwani, N.E. Benes, G. Bargeman, D. Stamatialis, M. Wessling, Effect of pH on the performance of polyamide/polyacrylonitrile based thin film composite membranes, J. Membr. Sci. 372 (2011) 228–238.

[23] S.K. Das, P. Manchanda, K.-V. Peinemann, Solvent-resistant Triazine-Piperazine Linked Porous Covalent Organic Polymer Thin-film Nanofiltration Membrane, Sep. Purif. Technol. (2018).

217

Chapter 7. A Generalized Method for Fabricating Ultra-Selective

Defect-Free Interfacially Polymerized Thin-Film Composite

Membranes for Gas- and Liquid Separations

The chapter deals with a generalized method for fabricating interfacially polymerized TFC polyamide membranes. The effect of varying water soluble monomers is studied and structure-property-function relationships are established.

7.1. Abstract

Presence of defects (in the dry state) has restricted the use of interfacially polymerized thin- film composite (TFC) network polyamide membranes in the gas separation industry. In this work, we report a generalized facile fabrication method for producing highly crosslinked, defect-free thin-film composite membranes for industrial gas- and liquid separations. The optimized TFC membranes demonstrate defect-free properties achieved by in-situ self-healing with excellent size sieving capabilities for hydrogen and air separations employing commercially available materials. The optimized protocol (KRO-

1) successfully produced TFC membranes with maximum H2/CO2 selectivity of ~16 and

O2/N2 selectivity of ~10. Furthermore, the protocol allowed for considerably enhanced performance in liquid separations compared to those prepared by industrially standard protocols particularly for desalination applications for both brackish and seawater reverse osmosis. Comprehensive characterization of membranes and bulk polymer powders using scanning electron microscopy, ellipsometry, X-ray photoelectron spectroscopy, wide-angle

X-ray diffraction, and the Brunauer, Emmett, and Teller (BET) technique disclosed good

218 agreement of physical properties with performance characteristics in both gas and liquid transport of components through such TFC membranes.

————————————————————————————————————

This chapter is under preparation as:

Zain Ali, Yingge Wang, Wojciech Ogieglo, Giuseppe Genduso, Federico Pacheco and Ingo Pinnau, A Generalized Method for Fabricating Ultra-Selective Defect-Free Interfacially Polymerized Thin-Film Composite Membranes for Gas- and Liquid Separations.

219

7.2. Introduction

Declining conventional fuel resources and mounting environmental concerns are driving the call for lowering the energy demands for chemical separation industries — currently accounting for 10 to 15% of global energy use [1]. Membrane technology offers a promising path to meet such demands due to its significantly lower energy requirements compared to conventional separation technologies as well as inherent advantages including lack of phase change required for separations and simple, continuous operation. Filmtec’s groundbreaking aromatic polyamide thin-film composite (TFC) FT-30 membranes provide an extremely relevant case-study to understand the potential of employing chemically, thermally and mechanically stable, commercially scalable, high-performance membranes to replace the traditional infrastructure for chemical separations.

FT-30 TFC membranes were pioneered by Cadotte and were prepared using inexpensive monomers, namely m-phenylene diamine (MPD) and trimesoyl chloride (TMC) [2]. The fabrication method, though, held the key to their commercial success. The membranes could be fabricated on a number of customizable polymeric and inorganic supports employing the interfacial polymerization (IP) process. The process allowed for the fabrication of crosslinked, ultrathin-films with excellent reproducibility applying an industrially scalable roll-to-roll process. The ultrathin nature of the membranes resulted in significantly higher fluxes than previously reported for asymmetric membranes as well as remarkable salt rejections (with the correct choice of monomers) for desalination applications. Coupled with that, the in-situ crosslinking resulted in excellent thermal and chemical resistance of the polyamide active layer opening the door for the FT-30-type

TFCs to be used in a wide-range of chemical environments [3]. Over time, the advantages

220 resulted in aggressive commercialization of the technology, and the FT-30-type TFCs currently account for 90% of the global desalination market (more than 15,000 desalination plants) using the reverse osmosis (RO) process [4]. This has led to the fabrication of a wide variety of interfacially polymerized TFC polyamide membranes with small variations in monomer chemistry and fabrication conditions. Perhaps the most important variation are

TFCs based on the semi-aromatic piperazine diamine used commercially for nanofiltration applications [5].

It is well known that interfacially polymerized FT-30-type RO membranes contain defects in the dry state [6,7]. The effects of defects in a TFC membrane have been described well by Henis and Tripodi [8] who used the analogy of an electrical circuit to explain transport of gases through a multi-layer TFC. They explained that defects offer very low resistance

(defined as the inverse of permeance, (P/l)-1) to the flow of gases through a TFC. Polymeric materials provide at least 100- to 1000-fold higher resistance to gas flow when compared to the air in the defects. Therefore similar to the case of current flow through an electric circuit, the presence of defects offers a very low resistance pathway (similar to a short circuit) to the flow of gases resulting in preferential gas passage through defects compared to the polymeric active layer. This presents two challenges to interfacially polymerized

TFC polyamide membranes: (1) such membranes cannot be used for gas separation applications, and; (2) gas transport through such membranes primarily depends on the nature of the defects rather than intrinsic separation properties of the material, making it tedious and unproductive to use gas transport as a characterization tool. This prevented a thorough understanding of fluid transport through such membranes.

221

In industrial settings, interfacially polymerized RO TFCs are typically fabricated with short reaction times (usually between 1 to 60 seconds) leading to comparatively thin layers (<

100 nm) to maximize water flux. Furthermore, typically the monomer solutions are used at room temperature (20 - 25 °C) [2]. Recently, our group fabricated defect-free interfacially polymerized TFC membranes by manipulating basic fabrication conditions such as IP reaction time and organic solution temperature. The membranes showed improved performance for both gas- and liquid separations as well as offered an enhanced understanding of transport mechanisms and structure-function relationships in such membranes [9,10]. Removal of defects significantly improved gas separation properties with remarkable potential for purifying H2 from mixtures containing CO2 displaying a molecular sieve-type separation mechanism [9]. Similarly, the optimized fabrication considerably improved the removal of both sodium chloride and boron under reverse osmosis conditions, again hinting towards a molecular sieve-like separation mechanism due to the removal of defects and improved crosslinking of the active layer [10].

In this work, we report a successful generic method for fabricating ultra-selective defect- free interfacially polymerized thin-film composite membranes using m-phenylene diamine

(MPD), p-phenylene diamine (PPD) and piperazine (PIP) with trimesoyl chloride (TMC).

Our optimized fabrication conditions are compared with the industrially standard ones.

Pure-gas permeation, as well as brackish and seawater reverse osmosis separation performance, is reported. The samples were characterized using Fourier transform infrared spectroscopy (FTIR), scanning electron microscopy (SEM), atomic force microscopy

(AFM), ellipsometry, X-ray photoelectron spectroscopy (XPS) and the Brunauer, Emmett, and Teller (BET) technique. The optimized TFCs show promising potential for improved

222 separation in both gas and liquid systems compared to traditional TFCs in addition to helping unravel structure-function-property relationships in interfacially polymerized TFC polyamide membranes.

7.3. Results and Discussion

Figure 7.1. Chemical structures of monomers used in this study.

Chemical structures of the monomers used in the study are shown in Figure 7.1.

223

Figure 7.2. Pure-gas permeance performance of commercial polyamide thin-film composite membranes.

Apparent Crosslinking N/O Roughness Sample thickness (nm) ratio ratio

MPD-SRO 79 0.78 1.91

MPD-KRO1 821 0.93 3.20

PPD-SRO 64 0.72 1.40

PPD-KRO1 122 0.77 2.19

PIP-SRO 71 0.82 1.03

PIP-KRO1 178 0.98 1.30

Commercially fabricated reverse osmosis membranes prepared using m-phenylene diamine and trimesoyl chloride are known to contain defects in the dry state [6,7]. Gas permeation

224 of commercial TFC membranes (as shown in Figure 7.2) revealed Knudsen transport mechanisms confirming defects in both Sepro RO4 (likely fabricated using m-phenylene diamine and trimesoyl chloride) and Dow Filmtec NF270 (likely fabricated using piperazine and trimesoyl chloride) membranes.

Figure 7.3. Pure-gas permeance performance of polyamide thin-film composite membranes using a) SRO and c) KRO-1 formulations. Pure-gas selectivity for chosen gas pairs for b) SRO and d) KRO1.

In-house fabricated TFCs using MPD and PIP monomers employing the SRO conditions agreed with Knudsen-type transport (Figure 7.3a). MPD-SRO exhibited gas permeances data almost identical to Sepro RO4 implying potential similarities in materials and fabrication conditions. On the other hand, PIP-SRO showed 3- to 5-fold increased gas

225 permeance compared to Dow Filmtec NF270 which could be a result of different fabrication conditions and/or materials used. PPD-SRO displayed the lowest permeance among the SRO membranes with He permeance of 100 GPU. Furthermore, it demonstrated slightly improved gas pair selectivities compared to Knudsen transport — for example,

PPD-SRO showed H2/CO2 and H2/N2 selectivities of ~7 compared to ~3 for MPD-SRO

(Figure 7.3b and 7.3d). The data implied reduced defects in the active layer of PPD-SRO and could be an effect of the higher reactivity of the PPD monomer as noted by Ichino et al. [11] and Roh et al. [12]. This potentially resulted in the faster formation and ultimately, increased defect plugging of the PPDTMC active layer. This improved gas pair selectivity beyond estimates made using Knudsen transport models.

Using the optimized KRO-1 fabrication formulation significantly decreased the permeance of gases (ranging between 10- to 1000-fold) through the fabricated TFC membranes

(Figure 7.3c). For example, switching from SRO to KRO-1 reduced the He and N2 permeance of piperazine-based membranes from 280 to 52 GPU and 105 to 0.12 GPU, respectively. PIP-KRO1 TFCs consistently demonstrated higher permeance values across all gases followed by PPD-KRO1 and MPD-KRO1 TFCs, respectively. This effect was complemented with improved selectivity for a number of gas pairs for all membranes when

KRO1 was used in place of SRO (Figure 7.3d). For example, H2/CO2 and O2/N2 selectivities increased from ~3 and ~1 for PIP-SRO to ~10 for PIP-KRO1. The overall impact of switching membrane formulations (from SRO to KRO-1) can be attributed to two unique phenomena. Firstly, the increased reaction time enabled in-situ plugging of defects due to continued diffusion of the amine to the reaction zone through potential defects in the membrane. Secondly, the higher organic solution temperature amplifies the

226 crosslinking in the active layer which was demonstrated by the increased crosslinking N/O ratio for all monomers used (as shown in Table 7.1). The two phenomena are discussed in detail elsewhere [9].

PPD-based membranes highlight the ability of the optimized formulation protocol (KRO1) to significantly remove very small defects in interfacially polymerized TFCs which would be inaccessible using the standard SRO protocol. As discussed, PPD-SRO shows relatively defect-free properties compared to MPD-SRO and PIP-SRO. Regardless, using the KRO-

1 formulation resulted in significant enhancement in gas pair selectivities of the PPD-based membranes, e.g., H2/N2 and H2/CH4 selectivities soared ~50-fold (from 7 to 309 and 5 to

379 respectively). This was possibly due to a considerable reduction in very small defects compared to PPD-SRO membranes as well as increased crosslinking of the active layer as crosslinking N/O ratio improved from 0.72 to 0.77.

227

Figure 7.4. Robeson plot for performance comparison of TFC membranes fabricated in this study. Adopted from Robeson (2008) upper-bound assuming 1 µm films [13]. Data for PEBAX coated SWC4 (PEBAX-SWC4), Polyamide + ZIF8 and MTR Proteus™ plotted separately from sources [6,14,15].

All TFCs fabricated using the KRO-1 protocol showed excellent separation properties for oxygen enrichment from air as well as hydrogen purification from CO2, N2 and CH4 containing mixtures. Hydrogen separation properties are particularly interesting since high

H2/CO2, H2/N2 and H2/CH4 selectivities indicate molecular sieve-like behavior for all

KRO-1 TFCs. Figure 7.4 shows the KRO-1 membranes compared to similar TFC membranes in literature. All membrane types successfully bypass the Robeson tradeoff emphasizing the excellent size sieving capabilities of interfacially polymerized TFC polyamide membranes.

228

CO2 sorption measurements (273K) on polymer powders (Figure 7.5a) demonstrated that

PIPTMC powder exhibited the highest surface area (78 m2 g-1) followed by PPDTMC (69 m2 g-1) and MPDTMC (m2 g-1). This agreed well with the gas permeance performance with

PIP-based membranes consistently demonstrating higher permeance values followed by

PPD-based membranes and finally MPD-based membranes. For the KRO-1 formulation,

selectivity for a number of gas pairs — including H2/CO2, H2/N2, H2/CH4, and CO2/CH4

— enhanced as the monomer was changed from PIP to PPD to MPD. XRD spectra (Figure

7.5b) demonstrated increased average chain spacing of the main amorphous peak (from

3.55 Å to 4.24 Å) as well as a broader peak (translating to wider pore size distribution) when MPD was replaced by the PIP monomer. It has been reported that as interchain distance increases beyond the optimum range for the separation of a gas pair, the permeance of both the slow and fast penetrant increase. The slower penetrant experiences a relatively larger rise in permeance resulting in decreased selectivity [16]. This can explain the elevated gas permeance and reduced gas pair selectivity for PIP-KRO1 compared to

MPD-KRO1. On the other hand, PPD- and MPD-based polymer powders both exhibited identical average chain spacing whereas MPD showed a narrower peak, implying a narrower pore size distribution, which can potentially explain lower permeance and higher selectivities of MPD-KRO1 membranes.

229

CO2/N2 selectivity exhibited a slightly different trend as selectivity decreased when MPD was substituted by PPD (from ~45 to 30). The CO2/N2 selectivity slightly improved when

PPD was substituted by the PIP monomer despite an increase in CO2 permeance in the direction MPD→PPD→PIP. O2/N2 selectivity, on the other hand, displayed an opposite trend to other gas pairs (mentioned above) with selectivity increasing from MPD to PPD to PIP. Although the trend previously observed for increasing permeance in the direction

MPD→PPD→PIP holds for both O2 and N2, careful observation revealed larger increments in O2 permeance compared to N2. For example, PPD-KRO1 exhibited ~187% increase in

O2 permeance while N2 increased by 166% compared to MPD-KRO1. Similarly, PIP-

KRO1 displayed a 31% increase in O2 permeance whereas N2 increased by only 14% compared to PPD-KRO1. As discussed above, if the interchain distance is increased beyond the optimum range for separation of a gas pair, the permeance enhancement of the slow penetrant is higher than the permeance increase of the fast penetrant resulting in decreased selectivity. Conversely, if the interchain distance is increased towards the

230 optimum size for the separation of a gas pair, the faster penetrant should experience an increase of permeance larger than the slower penetrant resulting in increased selectivity.

Although XRD measurements show MPDTMC powder average chain-spacing around 3.55

Å which is theoretically ideal for separating O2-N2 mixtures, it must be noted that powder

XRD measurements can only provide a valid qualitative comparison between the three polymers. Actual average chain spacing can be considerably different in thin-films and can significantly vary with fabrication conditions as discussed previously [9,10]. Realistically, the average chain spacing for highly crosslinked active layers fabricated using KRO-1 are lower than those measured for bulk polymer powders produced at mild conditions. Hence, it can be implied that replacing the monomer from MPD→PPD→PIP did, in fact, increase average chain spacing but potentially enlarged the chain spacing towards the ideal sieve size for O2-N2 separations.

Similar to gas permeation, switching the fabrication conditions from SRO to KRO-1 resulted in reduced water permeance and improved sodium chloride rejection as well as

231 enhanced water over sodium chloride separation factor in both brackish water reverse osmosis (BWRO) and seawater reverse osmosis (SWRO) modes of operation (Figure 7.6 and Table 7.2). For example for the PIP monomer, water permeance dropped from ~12.6 to 3.9 LMH/bar while rejection rose from ~24 to 75% for BWRO and from ~18 to 89% for

SWRO operation mode when KRO-1 was used instead of SRO. PIP-based TFCs consistently demonstrated the highest water flux and lowest salt rejection followed by PPD and MPD-based TFCs identical to gas permeation behavior. The improvement of salt rejection properties by switching to the KRO-1 formulation was consistent with our previous work [10] as well as the gas separation data reported above. This, again, is caused by both in-situ plugging of defects along with higher crosslinking of the polyamide active layer.

232

The difference in the position of the amine groups can affect polymer structural properties such as crosslinking, free carboxyl content and chain packing, resulting in different physical characteristics such as free volume and hydrophilicity [12]. Roh et al. [17] studied the salt rejection of membranes using multiple monomers and concluded that the positions of reacting groups in both the aqueous and organic phase monomers are the dominant characteristic determining salt rejection performance. They observed that high salt rejecting TFCs could be produced when functional groups on the aqueous and organic phase monomers lie at the same position on the aromatic ring. This agreed with the data is this study with MPD-based membranes producing significantly higher selectivity and rejection compared to PPD-based membranes.

Figure 7.7. Surface SEM images of in-house fabricated MPD (a-b), PPD (c-d) and PIP (e- f) TFC membranes using the SRO (a, c and e) and KRO1 (b, d, and f) formulation protocols.

233

Similar to gas permeation, the water flux exhibited good correlation with the BET surface area calculated using CO2 isotherms. PIP-based membranes showed the highest water flux followed by PPD-based and finally MPD-based membranes. This result emphasized the importance of chain packing in determining both liquid- and gas permeance through a TFC membrane. Likewise, average chain spacing correlated qualitatively well with salt rejections in both BWRO and SWRO modes. PIP-based membranes showed the lowest rejection which could be explained by the comparatively larger d-spacing of the PIPTMC polymer. This allowed greater transport of salt ions along with water molecules through the membrane; hence, PIP-KRO1 showed a water/sodium chloride selectivity of only ~120.

MPD-KRO1 showed the narrowest pore size distribution measured via XRD measurements which translated to the highest water/sodium chloride selectivity of ~26000.

PPDTMC powder showed identical average chain spacing to MPDTMC powder (as mentioned before), but the wider pore size distribution in PPDTMC powder could be a potential reason for decreased sodium chloride rejection of PPD-based membranes.

Although, both PPD-based TFCs demonstrated lower N/O crosslinking ratio in comparison to its MPD-counterparts. This is potentially due to the difference in reactivity of the two monomers and can contribute significantly to the lower rejections produced by the PPD- based membranes, as discussed previously [10].

234

Figure 7.8. Surface AFM images of in-house fabricated MPD (a-b), PPD (c-d) and PIP (e- f) TFC membranes using the SRO (a, c and e) and KRO1 (b, d, and f) formulation protocols.

Figure 7.9. Cross-sectional SEM images of in-house fabricated MPD (a-b), PPD (c-d) and PIP (e-f) TFC membranes using the SRO (a, c and e) and KRO1 (b, d, and f) formulation protocols.

235

Surface imaging of the fabricated membranes (Figure 7.7) exhibited an interesting trend in the surface structures of the TFCs. MPD-based membranes demonstrated the well-studied extremely rough ridge-and-valley structures. When the MPD monomer was replaced by

PPD using the SRO formulation, the nodules in the polyamide base become significantly more regular in shape and size. The chaotic ridge-and-valley structures is considerably less visible causing reduced surface roughness (Figure 7.8). This also resulted in decreased apparent cross-sectional thickness as shown in Figure 7.9. Further substitution of PPD with

PIP continued the trend as the surface roughness further dropped which can be seen by a visible reduction in nodule size. Cross-sectional images of PIP-SRO membranes revealed a seamless flat surface. When the SRO formulation was substituted by KRO-1, all membrane types experienced significantly increased surface roughness as well as polyamide layer thickness. For example, PIP-KRO1 demonstrated a thickness and surface roughness ratio of 178 nm and 1.30, respectively, compared to 71 nm and 1.03 for PIP-

SRO. The reasons for increased thickness and surface area were discussed previously [10] and resulted from increased turbulence at the aqueous/organic interface during the interfacial polymerization reaction due to the transfer of heat from the bulk organic solution to the reaction zone.

236

Figure 7.10. Crosslinking N/O ratio vs.: a) membrane surface roughness ratio, and; b) water over sodium chloride selectivity (Pw/PS). Red points denote data from this study. Blue points denote data from Ali et al. [10].

Surprisingly, the roughness of PPD-based membranes agreed well with correlations between surface roughness and crosslinking N/O ratio previously established [10] for

MPDTMC-based membranes, as shown in Figure 7.10a. This indicated a potential similar structure formation mechanism governing the surface roughness of both MPD and PPD- based membranes. However, PPD-based membranes significantly deviated from the previously established correlations between crosslinking N/O ratio and water/salt selectivity for MPDTMC membranes (Figure 7.10b) which highlights the differences in permselectivity performance by changing the position of the amine functional groups in phenyl diamines.

237

Plotting the gas and BWRO separation data for all the fabricated TFC membranes revealed a clear correlation between permeance of the faster components (He and water) with respect to the permeance of the slower components (CO2 and sodium chloride) as shown in Figure 7.11a-b. As the permeance of the faster component was improved, the permeance of the slower component increased according to the power law with good regression for both gas and liquid systems. Plotting the selectivity for He/CO2 (Figure 7.11c) and water/sodium chloride (Figure 7.11d) system on a Robeson-like tradeoff demonstrated

238 component selectivity decreasing as a function of increasing permeance of the faster component in-line with previously observed data [18]. Robeson defined the relationship between permeability and selectivity as:

푛 푃푥 = 푘훼푥푦

where, Px = permeability of component x, k = front factor, αxy is the selectivity of x over y and n is the slope of the log-log plot of the above equation [18].

Because of the logarithmic nature of the plot and limited differences in actual membrane thickness for all fabricated TFCs, the plot can be estimated using permeance in place of permeability. Using the above relationship, the n value can be estimated as -1.1 for the

He/CO2 system and -0.4 for the water/sodium chloride system which is directly related to the larger size difference between water molecules and hydrated sodium chloride ions compared to the size difference between He and CO2 molecules [13].

7.4. Conclusions

Removal of defects from interfacially polymerized TFC polyamide membranes can enable their use in the gas separation industry as well as improve their performance characteristics for liquid separations. Highly crosslinked, ultra-selective, defect-free TFC polyamide membranes were successfully fabricated in this study employing different commercial diamine monomers in the aqueous phase. The use of the industrially standard fabrication conditions was shown to induce defects in the interfacially polymerized TFC membranes.

The use of our optimized fabrication protocol (KRO-1) successfully plugged the defects in these TFCs through in-situ self-healing by simple manipulation of interfacial

239 polymerization reaction time. Furthermore, the protocol considerably enhanced the crosslinking of the active layers resulting in highly selective molecular sieve-like properties

[19].

All the membranes fabricated using the KRO-1 protocol demonstrated excellent potential for hydrogen separations especially from CO2 containing mixtures, which was illustrated by their permselectivity performance well above the 2008 Robeson upper bound. Using the

KRO-1 protocol resulted in a decrease of gas permeance — for example, PIP-SRO exhibited H2 permeance of 337 GPU vs. 38 GPU for PIP-KRO1 — with a significant boost in gas pair selectivity. For example, the selectivity for H2/N2 soared from ~3 for PIP-SRO to 310 for PIP-KRO1 (~100-fold increase). For the KRO-1 formulation, the permeance of all gases increased in the direction MPD→PPD→PIP (H2 permeance improved from 31 to

52 GPU) which was complemented with a selectivity drop for most gas pairs (H2/CO2 gas pair selectivity dropped from 14.3 to 10.2). O2/N2 selectivity showed the opposite trend with enhanced selectivity in the direction MPD→PPD→PIP (from 8.8 to 10.8).

Employing KRO-1 also improved both BWRO and SWRO separation performance for all membrane types. For example, PPD-SRO showed BWRO and SWRO rejections of 78 and

71%, respectively, compared to 98 and 85% for PPD-KRO1. Both gas and liquid permeance increased in the direction MPD to PPD to PIP. The increase of permeance was attributed to enhanced free volume (estimated using CO2 sorption employing the BET technique) while the decrease in gas pair selectivity and salt rejection were accredited to the enlarged interchain spacing (estimated using XRD measurements). Combined analysis of gas and liquid transport data exhibited permeance/selectivity trade-off behavior (similar to that reported by Robeson [18]) for both He/CO2 and water/sodium chloride systems.

240

The ability to fabricate highly crosslinked, ultra-selective TFC membranes using inexpensive commercially available monomers employing an industrially scalable roll-to- roll method shows promise in both creating the next-generation of highly stable, high- performance membranes for gas separations as well as improving the performance of currently used liquid separating membranes.

241

7.5. References

[1] D.S. Sholl, R.P. Lively, Seven chemical separations to change the world, Nature 532 (2016) 6–8.

[2] J.E. Cadotte, Interfacially synthesized reverse osmosis membrane, US 4,277,344 A, 1981.

[3] J.E. Cadotte, R.J. Petersen, R.E. Larson, E.E. Erickson, A new thin-film composite seawater reverse osmosis membrane, Desalination 32 (1980) 25–31.

[4] S.S. Shenvi, A.M. Isloor, A.F. Ismail, A review on RO membrane technology: Developments and challenges, Desalination 368 (2015) 10–26.

[5] M.J.T. Raaijmakers, N.E. Benes, Current trends in interfacial polymerization chemistry, Prog. Polym. Sci. 63 (2016) 86–142.

[6] J.S. Louie, I. Pinnau, M. Reinhard, Gas and liquid permeation properties of modified interfacial composite reverse osmosis membranes, J. Membr. Sci. 325 (2008) 793– 800.

[7] J. Albo, J. Wang, T. Tsuru, Gas transport properties of interfacially polymerized polyamide composite membranes under different pre-treatments and temperatures, J. Membr. Sci. 449 (2014) 109–118.

[8] J.M.S. Henis, M.K. Tripodi, Composite hollow fiber membranes for gas separation: the resistance model approach, J. Membr. Sci. 8 (1981) 233–246.

[9] Z. Ali, F. Pacheco, E. Litwiller, Y. Wang, Y. Han, I. Pinnau, Ultra-selective defect- free interfacially polymerized molecular sieve thin-film composite membranes for H2 purification, J. Mater. Chem. A 6 (2017) 30–35.

[10] Z. Ali, Y. Al Sunbul, F. Pacheco, W. Ogieglo, Y. Wang, G. Genduso, I. Pinnau, Defect-free highly selective polyamide thin-film composite membranes for desalination and boron removal, J. Membr. Sci. (2019).

[11] T. Ichino, S. Sasaki, T. Matsuura, S. Nishi, Synthesis and properties of new polyimides containing flourinated alkoxy side chains, J. Polym. Sci. Part A Polym. Chem. 28 (1990) 323–331.

[12] I.J. Roh, Effect of the physicochemical properties on the permeation performance in fully aromatic crosslinked polyamide thin films, J. Appl. Polym. Sci. 87 (2003) 569– 576.

[13] L.M. Robeson, The upper bound revisited, J. Membr. Sci. 320 (2008) 390–400.

[14] United States Department of Energy, Advanced carbon dioxide capture R&D

242

program: Technology update: Pre-combustion membranes, Pittsburgh, 2013.

[15] J. Sánchez-Laínez, L. Paseta, M. Navarro, B. Zornoza, C. Téllez, J. Coronas, Ultrapermeable Thin Film ZIF-8/Polyamide Membrane for H2/CO2 Separation at High Temperature without Using Sweep Gas, Adv. Mater. Interfaces 5 (2018) 1800647.

[16] S. Matteucci, Y. Yampolskii, B.D. Freeman, I. Pinnau, Transport of gases and vapors in glassy and rubbery polymers, in: Y. Yampolskii, I. Pinnau, B. Freeman (Eds.), Materials Science of Membranes for Gas and Vapor Separation, John Wiley & Sons, Ltd, Chichester, UK, 2006: pp. 1–47.

[17] I.J. Roh, S.Y. Park, J.J. Kim, C.K. Kim, Effects of the polyamide molecular structure on the performance of reverse osmosis membranes, J. Polym. Sci. Part B Polym. Phys. 36 (1998) 1821–1830.

[18] L.M. Robeson, Correlation of separation factor versus permeability for polymeric membranes, J. Membr. Sci. 62 (1991) 165–185.

[19] J.R. Werber, A. Deshmukh, M. Elimelech, The critical need for increased selectivity, not increased water permeability, for desalination membranes, Environ. Sci. Technol. Lett. 3 (2016) 112–120.

243

Chapter 8: Conclusions and Recommendations

This research aimed at developing methods to fabricate defect-free interfacially polymerized thin-film composite polyamide membranes for improved performance in gas- and liquid separations. Chapters 4 to 7 discussed methods and effects of removing defects in interfacially polymerized membranes as well as efforts to establish structure-function- performance relationships by varying fabrication conditions and building blocks. This chapter provides a summary of conclusions reached as well as recommendations for future work in the area.

8.1. Conclusions

Gas permeation tests confirmed the ability to self-heal defects in interfacially polymerized

TFCs by simply tweaking the reaction/contact time to achieve defect-free characteristics.

The data revealed the transition of gas transport from Knudsen diffusion to a solution- diffusion-based sieving mechanism confirming removal of micro-defects. Increasing organic solution temperature further resulted in improvement of gas separation properties which was attributed to increased crosslinking of the active layer. With the optimized fabrication parameters, the prepared TFCs showed excellent potential for separation of H2 from CO2, N2, and CH4 containing mixtures. H2/CO2 selectivity increased from ~4 for commercially standard protocols to ~14 using the optimized method. The membranes were tested at high temperature/pressure conditions which revealed anomalous behavior of increasing H2/CO2 selectivity with temperature. Mixed-gas separation performance using

1:1 H2/CO2 mixtures at high temperature/pressure conditions showed unprecedented performance for such separation, with the membrane showing 350 GPU H2 permeance and

244

mixed-gas H2/CO2 selectivity of 50 at 140 °C, the highest reported polymer selectivity for the system with excellent membrane stability.

Following this, the impact of removing defects was studied on liquid separation properties

(Chapter 5). Crossflow brackish water reverse osmosis permeation data showed improvement of sodium chloride rejection with increasing reaction time, reaching 99.2% rejection compared to 97.9% for standard industrial fabrication conditions and the commercial Sepro RO4 TFC. Similar to gas permeation studies, increasing organic solution temperature further improved rejection reaching a maximum of 99.6% for the optimized fabricated conditions designated KRO-1. KRO-1 performance was tested for boron rejection (a conventionally difficult solute to remove during desalination) and compared to four commercial membranes. KRO-1 showed improved performance for boron across all pH tested (6-10) confirming higher sieving capabilities imparted from the optimized fabrication conditions. Furthermore, it was observed that once defect-free characteristics were achieved, crosslinking showed a linear correlation with both water/sodium chloride separation factor as well as the membrane roughness.

Once defect-healing protocols were established, the conventional organic phase monomer for interfacial polymerization (trimesoyl chloride) was replaced with a kinked triptycene-

1,3,6,8- tetraacetyl chloride (designated TripTaC) moiety. Similar to previous studies using linear polymers, the introduction of the contorted TripTaC moiety successfully increased the fractional free volume of the active layer polyamide. This led to boosts in permselectivity performance ranging from organic/aqueous dye rejection to gas separation.

Methanol permeance soared to ~9 LMH/bar for triptycene-based membranes prepared using the KRO-1 protocol compared to the standard MPDTMC polyamide chemistry. This

245 was coupled with excellent rejection for both Sudan Orange G dye as well as the Brilliant

Blue R dye. The high permselectivity performance allowed the membrane to bypass the state-of-the-art permeance/selectivity trade-off highlighting the rigid nature of triptycene- based interfacially polymerized PIM-PA TFCs. In water, replacement of TMC with

TripTaC allowed improved water flux from 3 to 9 LMH/bar with no loss in rejection of

Brilliant Blue R dye. Sodium chloride rejection decreased to ~96% for the triptycene-based membranes compared to 99% for the highest rejecting MPDTMC-based membranes.

MPDTrip membranes achieved extremely high rejection of divalent ions including MgSO4,

Na2SO4, and MgCl2. Gas permeance showed a similar increase in permeance performance with H2 flux increasing from 21 GPU to 56 GPU for defect-free MPDTrip-100 compared to MPDTMC-20. This was coupled with an increase in H2/CO2 selectivity from ~7 to ~9.

Activation of MPDTrip-100 using isopropanol leads to a rise in CO2 permeance. The research further highlighted the extremely rigid nature of interfacially polymerized network PIM-PA TFCs as well as supported the benefits that could be achieved using such

TFCs for a wide variety of membrane-based separation applications.

Finally, the KRO-1 protocol was applied to a number of water soluble diamines namely m- phenylene diamine (MPD), p-phenylene diamine (PPD) and piperazine (PIP) with trimesoyl chloride (TMC). The KRO-1 formulation scheme enabled defect-free properties in all TFCs fabricated using MPD, PPD and PIP monomers. The protocol resulted in significant reduction in gas permeance through the TFCs due to plugging of defects which was complemented with considerable improvements in selectivities for a number of gas pairs. For example, the selectivity for H2/N2 soared from ~3 for PIP-SRO to 310 for PIP-

KRO1 (~100-fold increase). All KRO-1 protocol based TFCs showed excellent potential

246

for H2/CO2 separations with performance data lying well above the Robeson 2008 tradeoff.

Furthermore, the protocol improved sodium chloride rejection performance for the given

TFCs in both brackish water reverse osmosis as well as sea water reverse osmosis modes of separation tested under crossflow conditions. For example, PPD-SRO showed BWRO and SWRO rejections of 78 and 71%, respectively, compared to 98 and 85% for PPD-

KRO1. The research fairly summarized the ability of the KRO-1 protocol to be used with a wide range of building blocks to produce defect-free TFCs for improved performance of interfacially polymerized polyamide TFCs for both gas- and liquid separations.

8.2. Recommendations

It was observed in chapter 6 that the chemical nature of the support can have a significant impact on the properties of TFCs. The hydrophilic nature of the polyacrylonitrile promoted the synthesis of the hydrophilic polyamide in the support pores. Furthermore, due to the ultrathin nature of active layers produced during the interfacial polymerization fabrication procedure, radial diffusion can play a significant role in causing underestimation of membrane permeances. This has been previously observed, but strong conclusions on designing and fabricating ideal supports remain elusive [1–4]. Additionally interfacially polymerized network polyamides show excellent stability in a diverse range of organic solvents along with excellent thermal and mechanical stabilities but their use is restricted by limited thermal and chemical stabilities of the support layers. With the recent understanding of mechanisms of transport in interfacially polymerized polyamide TFCs, development of state-of-the-art supports which offer high chemical and thermal stabilities along with narrow pore size distributions can significantly contribute to increasing TFC

247 performance in current applications as well as expand their use in new applications including applications using polar aprotic solvents.

Despite the use of binary mixed-gas systems in Chapter 4, performance evaluation of fabricated polyamide TFCs using multi-component systems can unravel the real potential of such membranes in industrial systems. Applications for hydrogen purification in refineries require membrane chemical and performance stability in the presence of a wide variety of components. Condensable components (specifically, C2 and higher alkanes/alkenes, methanol, ammonia) can have a significant impact on the final performance characteristics of TFC polymeric membranes. Evaluation of permselectivity performance in the presence of multi-components in the test streams can provide better insight into the industrial applicability of such membranes.

Development of linear polymers has taken significant steps over the last two decades to produce a wide variety of high-performance materials, especially polymers of intrinsic microporosity [5–9]. Relatively, the development of materials for fabrication of high free volume interfacially polymerized polyamide TFCs has been extremely limited. Primary examples of novel materials explored for interfacial polymerization are limited to fabrication of polyesters, currently at less than 10 research publications [10,11]. Extending the portfolio of materials to establish structure-property-performance relationships in high free-volume interfacially polymerized polyamide TFCs can add substantial knowledge to this emerging field of research and allow rational design of interfacially polymerized PIM polyamides for fabrication of membranes to solve the separation problems of tomorrow.

248

8.3. References

[1] H.R. Afshoun, M.P. Chenar, A.F. Ismail, T. Matsuura, Effect of support layer on gas permeation properties of composite polymeric membranes, Korean J. Chem. Eng. (2017) 1–7.

[2] N.W. Oh, J. Jegal, K.H. Lee, Preparation and characterization of nanofiltration composite membranes using polyacrylonitrile (PAN). I. Preparation and modification of PAN supports, J. Appl. Polym. Sci. 80 (2001) 1854–1862.

[3] W.-C. Chao, Y.-H. Huang, W.-S. Hung, Q. An, C.-C. Hu, K.-R. Lee, J.-Y. Lai, Effect of the surface property of poly(tetrafluoroethylene) support on the mechanism of polyamide active layer formation by interfacial polymerization, Soft Matter 8 (2012) 8998.

[4] A. Tiraferri, N.Y. Yip, W.A. Phillip, J.D. Schiffman, M. Elimelech, Relating performance of thin-film composite forward osmosis membranes to support layer formation and structure, J. Membr. Sci. 367 (2011) 340–352.

[5] X. Ma, M. Abdulhamid, X. Miao, I. Pinnau, Facile synthesis of a hydroxyl- functionalized Tröger’s base diamine: A new building block for high-performance polyimide gas separation membranes, Macromolecules 50 (2017) 9569–9576.

[6] R. Swaidan, B. Ghanem, M. Al-Saeedi, E. Litwiller, I. Pinnau, Role of intrachain rigidity in the plasticization of intrinsically microporous triptycene-based polyimide membranes in mixed-Gas CO2/CH4 separations, Macromolecules 47 (2014) 7453– 7462.

[7] W. Ogieglo, B.S. Ghanem, X.-H. Ma, M. Wessling, I. Pinnau, High Pressure CO2 sorption in polymers of intrinsic Mmty under ultra-thin film confinement, ACS Appl. Mater. Interfaces 10 (2018) acsami.8b01402.

[8] N. Alaslai, B. Ghanem, F. Alghunaimi, I. Pinnau, High-performance intrinsically microporous dihydroxyl-functionalized triptycene-based polyimide for natural gas separation, Polymer (United Kingdom) 91 (2016) 128–135.

[9] B.S. Ghanem, R. Swaidan, X. Ma, E. Litwiller, I. Pinnau, Energy-efficient hydrogen separation by AB-type ladder-polymer molecular sieves, Adv. Mater. 26 (2014) 6696–6700.

[10] M.F. Jimenez-Solomon, Q. Song, K.E. Jelfs, M. Munoz-Ibanez, A.G. Livingston, Polymer nanofilms with enhanced microporosity by interfacial polymerization, Nat. Mater. 15 (2016) 760–767.

[11] S. Yu, S. Li, Y. Liu, S. Cui, X. Shen, High-performance microporous polymer membranes prepared by interfacial polymerization for gas separation, J. Membr. Sci. 573 (2018) 425–438.

249

Appendices

Publications

The following publications are to be/were made during this Ph.D. work:

1. Zain Ali, Federico Pacheco, Eric Litwiller, Yingge Wang, Yu Han and Ingo Pinnau,

Ultra-Selective Defect-Free Interfacially Polymerized Molecular Sieve Thin-Film

Composite Membranes for H2 Purification, Journal of material chemistry A, 6

(2017) 30–35. “The publication arose from Chapter 4.”

2. Xiaohua Ma, Yingge Wang, Kexin Yao, Zain Ali, Yu Han and Ingo Pinnau, Pristine

and Carboxyl-Functionalized Tetraphenylethylene-Based Ladder Networks for

Gas Separation and Volatile Organic Vapor Adsorption, ACS Omega, 3 (2018)

15966–15974. “Contributed to characterization.”

3. Zain Ali, Yasmeen Al Sunbul, Federico Pacheco, Wojciech Ogieglo, Yingge

Wang, Giuseppe Genduso and Ingo Pinnau, Defect-Free Highly Selective

Polyamide Thin-Film Composite Membranes for Desalination and Boron Removal,

Journal of membrane science, 579 (2019) 85-94. “The publication arose from

Chapter 5.”

4. Zain Ali, Bader Ghanem, Yingge Wang, Federico Pacheco, Wojciech Ogieglo,

Giuseppe Genduso, Hakkim Vovusha, Udo Schwingenschlogl, Yu Han and Ingo

Pinnau, Triptycene-based Interfacially Polymerized Thin-Film Composite

Polyamide Membranes for Liquid- and Gas Separations. Manuscript in preparation.

“The publication will arise from Chapter 6.”

5. Zain Ali, Yingge Wang, Wojciech Ogieglo, Federico Pacheco, Giuseppe Genduso

and Ingo Pinnau, A Generalized Method for Fabricating Ultra-Selective Defect-

250

Free Interfacially Polymerized Thin-Film Composite Membranes. Manuscript in

preparation. “The publication will arise from Chapter 7.”

6. Zain Ali and Ingo Pinnau, Thin Film Composite Membranes For Fluid Separations.

International patent published. International publication number: WO 2018/224906

A1. “The publication arose from Chapter 4.”

Chapter 2

Monomers used for aqueous phase in interfacial polymerization

Monomer Structure Reference

m-Phenylenediamine (MPD) [1–4]

o-Phenylenediamine (OPDA) [5]

p-Phenylenediamine (PPDA) [5]

Piperazine (PIP) [6–8]

1,3-Cyclohexanebis(methyl- [6] amine) (CBMA)

251

Diethylenetriamine (DAP1) [6]

2,3-Diaminopyridine (DETA) [6]

Poly(m-aminostyrene) [5] (PmAS1)

Poly(m-aminostyrene) [9] (PmAS2)

N-methyl-m- phenylenediamine (N- [8] MMPD)

N,N’-dimethyl-m- phenylenediamine (N,N'- [8] DMMPD)

252

3,5-Diaminobenzoic acid [9] (DABA)

MPD-5-Sulfonic acid (SMPD) [10]

N,N'-Diamino-piperazine [11] (DAP)

1,4-Bis(3-aminopropyl)- [11] piperazine (DAPP) n-(2-Aminoethyl)-piperazine [11] (EAP)

Tetrahydroxy-3,3,3',3'-

tetramethyl-l,l'- [12]

spirobisindane (THTMSPD)

N-Methyldiethanolamine [13] (MEDA)

1,2-Dimethylenediamine [14] (DMDA)

253

1,6- Hexamethlenediamine(HMD [14]

1,9-Nonamethylenediamine [14] (NMDA)

Polyetheramine (PEA) [15]

3,5- Diaminobenzoylpiperazine [16] (3,5-DABP)

3,5-Diamino-N-(4- aminophenyl) benzamide [17] (DAAPB)

Dopamine [18]

N-aminoethyl piperazine [19] propane sulfone (AEPPS)

Bipiperidine [20]

254

N-Methyl-diethanolamine [21] (MDEOA)

Triethaolaine (TEOA) [21]

N,N-Dimethyl-1,3- phenylenediamine [22] dihydrochloride

4-Methyl-m- [23] phenylenediamine

4,6-Dimethyl-1,3- phenylenediamine [24] dihydrochloride

4-Methoxy-1,3- [24] phenylenediamine

255

2,3,5,6-Tetramethyl-1,4- [25] phenylene diamine

Benzidine [22]

3,3'-Dimethylbenzidine [26] dihydrochloride

3,3'-Dimethoxybenzdine [27]

1,5-Diaminonaphthalene [27]

Bis 4-aminophenyl methane diamine [26]

(Several variations)

256

Cyclen [28]

Bis-2,6-N,N-(2-hydroxyethyl) [29] diaminotoluene

2,6-Diaminotoluene [30]

2,4-Diaminotoluene [30]

3,4-Diaminotoluene [30]

2,4-Diaminoanisole [30]

257

4-Chloro-1,3-diaminobenzene [30]

N,N′- [30] Diphenylethylenediamine

Monomers used for organic phase in interfacial polymerization

Monomer Structure Reference

Trimesoyl chloride (TMC) [3,4,6,31,32]

Benzoyl chloride (BC) [5]

Phthaloyl chloride (PC) [5]

258

Isophthaloyl chloride [5] (IPC)

Terephthaloyl chloride [5] (TPC)

5-Isocyanato-isophthaloyl chloride (ICIC) [33]

5-Chloroformyloxy- isophthaloyl chloride (CFIC) [33]

1,6- hexamethylenedicarbonyl [14] (SC)

259

1,2,4,5-Benzene tetracarbonyl chloride [34] (BTC)

p-Toluoyl chloride [35]

3,4',5-Biphenyl triacyl [36] chloride (BTRC)

3,3',5,5'-Biphenyl tetraacyl [36] chloride (BTEC)

Adipoyl chloride [37]

Suberoyl chloride [36]

Sebacoyl chloride [36]

260

2,6-Pyridinedicarbonyl dichloride [36]

Succinyl chloride [36]

Fumaryl chloride [36]

Oxalyl chloride [37]

Additional references

[1] R.E. Larson, J.E. Cadotte, R.J. Petersen, The FT-30 seawater reverse osmosis membrane--element test results, Desalination 38 (1981) 473–483. [2] J.E. Cadotte, R.J. Petersen, R.E. Larson, E.E. Erickson, A new thin-film composite seawater reverse osmosis membrane, Desalination 32 (1980) 25–31. [3] A.K. Ghosh, B.H. Jeong, X. Huang, E.M.V. Hoek, Impacts of reaction and curing conditions on polyamide composite reverse osmosis membrane properties, J. Membr. Sci. 311 (2008) 34–45. [4] M.M. Pendergast, J.M. Nygaard, A.K. Ghosh, E.M. V. Hoek, Using nanocomposite materials technology to understand and control reverse osmosis membrane compaction, Desalination 261 (2010) 255–263. [5] C.. K. Kim, J.. H.J.. J. Kim, I.. I.J. Roh, J.. H.J.. J. Kim, The changes of membrane performance with polyamide molecular structure in the reverse osmosis process, J. Membr. Sci. 165 (2000) 189–199.

261

[6] D.J. Mohan, L. Kullová, A study on the relationship between preparation condition and properties/performance of polyamide TFC membrane by IR, DSC, TGA, and SEM techniques, Desalin. Water Treat. 51 (2013) 586–596. [7] J. Xiang, Z. Xie, M. Hoang, K. Zhang, Effect of amine salt surfactants on the performance of thin film composite poly(piperazine-amide) nanofiltration membranes, Desalination 315 (2013) 156–163. [8] T. Shintani, A. Shimazu, S. Yahagi, H. Matsuyama, Characterization of methyl- substituted polyamides used for reverse osmosis membranes by positron annihilation lifetime spectroscopy and MD simulation, J. Appl. Polym. Sci. 113 (2009) 1757–1762. [9] S.Y. Kwak, M.O. Yeom, I.J. Roh, D.Y. Kim, J.J. Kim, Correlations of chemical structure, atomic force microscopy (AFM) morphology, and reverse osmosis (RO) characteristics in aromatic polyester high-flux RO membranes, J. Membr. Sci. 132 (1997) 183–191.

[10] Z. Yong, Y. Sanchuan, L. Meihong, G. Congjie, Polyamide thin film composite membrane prepared from m-phenylenediamine and m-phenylenediamine-5-sulfonic acid, J. Membr. Sci. 270 (2006) 162–168. [11] S. Verissimo, K.-V. Peinemann, J. Bordado, Influence of the diamine structure on the nanofiltration performance, surface morphology and surface charge of the composite polyamide membranes, J. Membr. Sci. 279 (2006) 266–275. [12] A.G. Livingston, M.F.J. Solomon, Membranes for separation, US20140251897A1, 2013. [13] F. Yuan, Z. Wang, S. Li, J. Wang, S. Wang, Formation-structure-performance correlation of thin film composite membranes prepared by interfacial polymerization for gas separation, J. Membr. Sci. 421–422 (2012) 327–341. [14] I.J. Roh, V.P. Khare, Investigation of the specific role of chemical structure on the material and permeation properties of ultrathin aromatic polyamides, J. Mater. Chem. 12 (2002) 2334–2338. [15] A.A.M.M. Salih, C. Yi, H. Peng, B. Yang, L. Yin, W. Wang, Interfacially polymerized polyetheramine thin film composite membranes with PDMS inter-layer for CO2 separation, J. Membr. Sci. 472 (2014) 110–118. [16] L. Wang, D. Li, L. Cheng, L. Zhang, H. Chen, Preparation of thin film composite nanofiltration membrane by interfacial polymerization with 3,5- diaminobenzoylpiperazine and trimesoyl chloride, Chinese J. Chem. Eng. 19 (2011) 262–266. [17] H. Wang, L. Li, X. Zhang, S. Zhang, Polyamide thin-film composite membranes prepared from a novel triamine 3,5-diamino-N-(4-aminophenyl)-benzamide monomer and m-phenylenediamine, J. Membr. Sci. 353 (2010) 78–84.

262

[18] J. Zhao, Y. Su, X. He, X. Zhao, Y. Li, R. Zhang, Z. Jiang, Dopamine composite nanofiltration membranes prepared by self-polymerization and interfacial polymerization, J. Membr. Sci. 465 (2014) 41–48. [19] Q.F. An, W.D. Sun, Q. Zhao, Y.L. Ji, C.J. Gao, Study on a novel nanofiltration membrane prepared by interfacial polymerization with zwitterionic amine monomers, J. Membr. Sci. 431 (2013) 171–179. [20] John E. Tomaschke, Interfacially polymerized, bipiperidine-polyamide membranes for reverse osmosis and/or nanofiltration and process for making the same US 6464873, 1 (2002). [21] B. Tang, C. Zou, P. Wu, Study on a novel polyester composite nanofiltration membrane by interfacial polymerization. II. The role of lithium bromide in the performance and formation of composite membrane, J. Membr. Sci. 365 (2010) 276–285. [22] I.K. Nekrasov, T.V. Kudim, L.B. Sokolov, Effect of various factors on the molecular weight distribution of poly-(m-phenylene)isophthalamide in emulsion polymerization, Polym. Sci. U.S.S.R. 14 (1972) 877–883. [23] W.S. Hill Jr Harold Wayne, Louise Kwolek Stephanie, Wholly aromatic polyamides, US3094511A, 1963. [24] W.M.P. Hill Jr Harold Wayne, Louise Kwolek Stephanie, Polyamides from reaction of aromatic diacid halide dissolved in cyclic nonaromatic oxygenated organic solvent and an aromatic diamine, US3006899A, 1961. [25] E.M. Eugene, Process for preparing polyamides, US2831834A, 1958. [26] O.Y. Fedotova, M.L. Kerber, I.P. Losev, Some properties of aromatic and aryl aliphatic polyamides prepared by interfacial polycondensation-IX, Polym. Sci. U.S.S.R. 6 (1964) 502–509. [27] A.S. Maloshitskii, G.S. Kolesnikov, T.P. Malinovskaya, Carbochain polymers and copolymers—LIV. Polymerization of methyl methacrylate in the presence of n- butylboryl difluoride, Polym. Sci. U.S.S.R. 6 (1964) 1058–1062. [28] G.E. Chen, Y.J. Liu, Z.L. Xu, D. Hu, H.H. Huang, L. Sun, Preparation and characterization of a composite nanofiltration membrane from cyclen and trimesoyl chloride prepared by interfacial polymerization, J. Appl. Polym. Sci. 132 (2015) 42345. [29] Z. Zhang, S. Wang, H. Chen, Q. Liu, J. Wang, T. Wang, Preparation of polyamide membranes with improved chlorine resistance by bis-2,6-N,N-(2-hydroxyethyl) diaminotoluene and trimesoyl chloride, Desalination 331 (2013) 16–25. [30] C. Guo, L. Zhou, J. Lv, Effects of expandable graphite and modified ammonium polyphosphate on the flame-retardant and mechanical properties of wood flour- polypropylene composites, Polym. Polym. Compos. 21 (2013) 449–456.

263

[31] J. Duan, E. Litwiller, I. Pinnau, Preparation and water desalination properties of POSS-polyamide nanocomposite reverse osmosis membranes, J. Membr. Sci. 473 (2015) 157–164. [32] I.C. Kim, B.R. Jeong, S.J. Kim, K.H. Lee, Preparation of high flux thin film composite polyamide membrane: The effect of alkyl phosphate additives during interfacial polymerization, Desalination 308 (2013) 111–114. [33] W. Li, Z. Yang, G. Zhang, Z. Fan, Q. Meng, C. Shen, C. Gao, Stiff metal–organic framework–polyacrylonitrile hollow fiber composite membranes with high gas permeability, J. Mater. Chem. A 2 (2014) 2110. [34] S. Hong, I.C. Kim, T. Tak, Y.N. Kwon, Interfacially synthesized chlorine-resistant polyimide thin film composite (TFC) reverse osmosis (RO) membranes, Desalination 309 (2013) 18–26.

[35] I.J. Roh, S.Y. Park, J.J. Kim, C.K. Kim, Effects of the polyamide molecular structure on the performance of reverse osmosis membranes, J. Polym. Sci. Part B Polym. Phys. 36 (1998) 1821–1830. [36] L. Li, S. Zhang, X. Zhang, G. Zheng, Polyamide thin film composite membranes prepared from 3,4′,5-biphenyl triacyl chloride, 3,3′,5,5′-biphenyl tetraacyl chloride and m-phenylenediamine, J. Membr. Sci. 289 (2007) 258–267. [37] V.H. Hopff, A. Krieger, H. Hopff, A. Krieger, V.H. Hopff, A. Krieger, Über polyamide aus heterocyclischen dicarbonsäuren, Die Makromol. Chemie 47 (1961) 93–113.