Follow this and additional works at: Part of the Chemical Engineering Commons

Transcription

1 University of Colorado, Boulder CU Scholar Chemical & Biological Engineering Graduate Theses & Dissertations Chemical & Biological Engineering Spring Design, Development, and Evaluation of Thin-Film Composite Bicontinuous Cubic Lyotropic Liquid Crystal Polymer Membranes for Water Filtration Applications cost Blaine Michael Carter University of Colorado at Boulder, Follow this and additional works at: Part of the Chemical Engineering Commons Recommended Citation Carter, Blaine Michael, "Design, Development, and Evaluation of Thin-Film Composite Bicontinuous Cubic Lyotropic Liquid Crystal Polymer Membranes for Water Filtration Applications cost" (2014). Chemical & Biological Engineering Graduate Theses & Dissertations This Dissertation is brought to you for free and open access by Chemical & Biological Engineering at CU Scholar. It has been accepted for inclusion in Chemical & Biological Engineering Graduate Theses & Dissertations by an authorized administrator of CU Scholar. For more information, please contact

2 Design, Development, and Evaluation of Thin-Film Composite Bicontinuous Cubic Lyotropic Liquid Crystal Polymer Membranes for Water Filtration Applications by Blaine Michael Carter B.S., Northwest Nazarene University, 2009 A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirement for the degree of Doctor of Philosophy Department of Chemical and Biological Engineering 2014

3 This thesis entitled: Design, Development, and Evaluation of Thin-Film Composite Bicontinuous Cubic Lyotropic Liquid Crystal Polymer Membranes for Water Filtration Applications written by Blaine Michael Carter has been approved for the Department of Chemical and Biological Engineering Professor Douglas L. Gin Professor Richard D. Noble Date The final copy of this thesis has been examined by the signatories, and we Find that both the content and the form meet acceptable presentation standards Of scholarly work in the above mentioned discipline.

4 ABSTRACT Carter, Blaine Michael (Ph.D., Chemical Engineering, Department of Chemical and Biological Engineering) Design, development, and evaluation of thin-film composite bicontinuous cubic lyotropic liquid crystal polymer membranes for water filtration applications Thesis directed by Professor Douglas L. Gin and Professor Richard D. Noble The in situ cross-linking of reactive amphiphiles (i.e., surfactants) that self-organize in the presence of a solvent into a type I bicontinuous cubic (QI) lyotropic liquid crystal (LLC) phase have significant potential as a new type of membrane material. These QI-phase networks contain completely uniform pores on the nanometer scale and tunable pore chemistry that could offer significant improvements over conventional nanofiltration (NF) and reverse osmosis (RO) membranes. Despite their significant potential, cross-linked QI-phase networks have never been processed into thin-films necessary to demonstrate commercially relevant productivity and limited characterization has been performed on QI-phase membranes to understand their fundamental performance characteristics. The overall objective of this thesis research is to design and develop a method or approach to fabricate thin-film composite (TFC) QI membranes and perform additional fundamental filtration studies to better understand the transport characteristics of QI-phase membranes. In this thesis research, a new fundamental processing approach is presented to fabricate thin-films of cross-linked QI-phase materials. In this approach, the water typically used for LLC phase formation is replaced with the low volatility, polar, organic solvent glycerol. Due to the low vapor pressure of glycerol, it does readily evaporate during thin-film processing. This allows LLC monomer and glycerol to be dissolved in a volatile casting solvent and applied to a porous iii

5 support by solution-casting. Subsequent removal of the volatile casting solvent by gentle heating does not result in any appreciable evaporative loss of glycerol necessary to form the desired QIphase. The applied thin-film can then be photo-cross-linked at the required temperature to create a TFC QI membrane. Water filtration experiments on TFC QI membranes demonstrate they have similar rejection performance to previous QI-phase membranes, except the flux is ca. 10 times greater due to the much thinner active layer. Overall, TFC QI membranes have performance characteristics in-between conventional NF and RO membranes. They reject neutral solutes like a porous NF membrane and reject monovalent and divalent salts comparable to a RO membrane at brackish water feed concentrations. Additional water filtration experiments revealed that anion-exchange has a significant impact on the flux of TFC QI membranes. The flux of TFC QI membranes can be drastically tuned by exposing the membrane to feeds containing different anions with little to no change in the rejection performance. The unexpected performance changes in TFC QI membrane with anion-exchange prompted the design and synthesis of new, cross-linkable, zwitterionic amphiphiles in which ionexchange is no longer possible. A new zwitterionic LLC monomer system has been developed that contains benzimidazolium cationic headgroups with covalently tethered anionic sulfonate groups. These new, cross-linkable, zwitterionic amphiphiles are capable of forming LLC phases in water and glycerol and may readily afford cross-linked LLC assemblies. Zwitterionic LLC assemblies could offer a number of advantages over conventional cationic or anionic LLC networks for a number of applications, including water filtration. iv

6 DEDICATION I am not a self-made man, and the person I am today is the product of amazing people around me that have chosen to make an investment in me. This work is dedicated to everyone that has helped advance my education and/or supported me along the way. In particular, I would like to thank my wife, Arin, for her incredible love and support throughout this entire journey. I would not have made it without you. My parents, Mike and Tamara, you have provided me with all the opportunity in the world.

7 ACKNOWLEDGMENTS Financial support for this dissertation research from the National Science Foundation (grant: CBET ), the U.S. Bureau of Reclamation (grant: R13AC80040), and a National Renewable Energy Laboratory-sponsored grant (13-2) from the Membrane Science, Engineering and Technology (MAST) Center at the University of Colorado, Boulder is gratefully acknowledged. I also thank the U.S. Dept. of Education for a GAANN fellowship during the course of this work. I thank my two thesis advisors, Douglas Gin and Richard Noble, for their guidance, availability, and encouragement throughout my thesis research. I appreciate the creative and positive work environment you have both worked to develop. I also wish to thank Doug and Rich for all of the various opportunities they have provided over the years. I thank all of the past and current members of the Gin and Noble groups for their assistance, guidance, and companionship. In particular, I give a special thanks to Evan Hatakeyama and Brian Wiesenauer for mentoring me during my first year as a graduate student. I gratefully acknowledge Lee Miller, Trevor Carlisle, Will McDanel, Garret Nicodemus, and Matt Cowan for their friendship and advice over the years. You definitely made my graduate experience enjoyable and memorable. vi

8 TABLE OF CONTENTS LIST OF TABLES...xi LIST OF FIGURES...xii CHAPTER 1. Introduction and background The demand for clean water Overview of membrane technology for aqueous separations Separation mechanisms of water filtration membranes Current issues with nanofiltration and reverse osmosis membranes New materials for fabricating nanoporous membranes with uniform pores Cross-linkable lyotropic liquid crystals (LLCs) Thesis objectives: Design, development, and evaluation of thin-film composite bicontinuous cubic polymer membranes References...20 CHAPTER 2. Glycerol-based bicontinuous cubic lyotropic liquid crystal monomer system for the fabrication of thin-film membranes with uniform nanopores...26 Abstract Introduction Results and discussion Conclusions Supporting information Materials and general procedures Instrumentation New LLC monomer synthesis and characterization Bromopentadecanoic acid Bromopentadecanol Bromopentadecanal Bromotetradeca-1,3-diene Bromooctadeca-1,3-diene ,1 -(1,4-Butanediyl)bisimidazole ,1 -(1,6-Hexanediyl)bisimidazole ,1 -(Oxydi-2,1-ethanediyl)bisimidazole ,6-Bis(octadeca-15,17-dienylimidazolium)hexane dibromide (3) ,4-Bis(octadeca-15,17-dienylimidazolium)butane dibromide (4) ,1 -(Oxydi-2,1-ethanediyl)bis[3-(octadeca-15,17- dienyl)imidazolium] dibromide (5)...41 vii

9 ,6-Bis(tetradeca-11,13-dienylimidazolium)hexane dibromide (6) ,4-Bis(tetradeca-11,13-dienylimidazolium)butane dibromide (7) ,1 -(Oxydi-2,1-ethanediyl)bis[3-(tetraadeca-11,13- dienyl)imidazolium] dibromide (8) Ethylammonium nitrate Qualitative screening of LLC phase behavior with different solvents using the PLM-based penetration scan technique Preparation of LLC samples, determination of LLC phase behavior, and elucidation of LLC phase diagrams Preparation of bulk QI-phase films of 3/glycerol/HMP and crosslinking with LLC phase retention Comparison of the relative degree of order in the QI phases of 1/water, 2/water, and 3/glycerol by PXRD analysis Determination of degree of conversion for the radical photopolymerization of the 3/glycerol/HMP bicontinuous cubic (QI) phases Solution-casting and photo-cross-linking of a thin-film QI-phase of 3/glycerol/HMP on dense glass substrates to provide proof-ofconcept for successful thin-film formation and cross-linking with LLC phase retention Preparation of asymmetric, porous poly(ether sulfone) (PES) support membranes via phase-inversion processing Fabrication of PES-supported thin-film composite (TFC) QI polymer membranes of 3/glycerol/HMP General water nanofiltration testing procedure Permeate analysis Water nanofiltration testing of uncoated PES support membranes as controls Water nanofiltration testing of TFC QI polymer membranes made from 3/glycerol/HMP Estimation of the effective pore size by fitting the neutral solute rejection data with the Ferry equation Preliminary studies showing the beneficial effect of aq. NaCl pretreatment on the water flux of the TFC QI membranes for neutral solute filtrations without compromising rejection Comparison of the performance of TFC QI membranes with commercial Dow SW30HR (RO) and DOW NF-270 (NF) membranes Acknowledgments References...68 CHAPTER 3. Thin-film composite bicontinuous cubic lyotropic liquid crystal polymer membranes: Effects of anion-exchange on water filtration performance...71 viii

10 Abstract Introduction Experimental Materials and instrumentation Membrane preparation and characterization Anion-exchange of bulk cross-linked QI-phase films and TFC QI membranes Water filtration testing Results and discussion Characterization of TFC QI membranes made with a commercial UF support Effect of monovalent cation on water filtration performance Effect of monovalent anion on water filtration performance Effect of complete monovalent anion-exchange on water filtration performance Conclusions Acknowledgments References...99 CHAPTER 4. Design and synthesis of novel, cross-linkable, zwitterionic amphiphiles for the formation of cross-linked lyotropic liquid crystal assemblies Abstract Introduction Results and discussion Monomer design and synthesis LLC phase behavior of monomers 13 and Conclusions Experimental Materials Instrumentation Monomer synthesis and characterization ,3-bis(2-benzimidazolyl)propane (9) ,5-bis(2-benzimidazolyl)pentane (10) ,3-bis(1-(octadeca-15,17-dienyl)-2- benzimidazolyl)propane (11) ,5-bis(1-(octadeca-15,17-dienyl)-2- benzimidazolyl)pentane (12) Monomer Monomer Preparation of LLC samples and qualitative investigation of phase behavior Acknowledgments References ix

11 CHAPTER 5. Conclusions and recommendations Summary of thesis work Recommendations for future work Explore processing parameters and methods to prepare 0.5 μm thick TFC QI membranes Continue to investigate the effects of anion-exchange on TFC QI membrane performance Evaluate the chemical stability of TFC QI membranes based on Investigate the use of cross-linkable, zwitterionic amphiphiles to form LLC networks References CHAPTER 6. Bibliography x

12 LIST OF TABLES Table 2.1. Water filtration performance of TFC QI membranes made from 3/glycerol/HMP (3 µm thick active layer) S1. The gemini imidazolium bromide monomer homologues synthesized, and a summary of their qualitative Q phase formation behavior between 22 C and 95 C with different solvents from PLM penetration scan screening studies S2. Rejection and water flux data for individually tested TFC QI polymer membranes pre-soaked in a 2000 ppm NaCl solution before conducting neutral solute filtration experiments S3. Comparison of pure water flux and thickness-normalized pure water permeability values for NF-270, SW30HR, and the TFC QI membranes of 3/glycerol/HMP measured under the same dead-end filtration test conditions Flux and rejection data of the TFC QI membranes with different alkali bromide salts. All experiments performed in stirred dead-end filtration cells with an applied pressure of Pa (400 psi) Rejection of different sodium salts by the TFC QI membranes. All experiments performed in stirred dead-end filtration cells with an applied pressure of Pa (400 psi) Comparison of the pure water flux of completely anion-exchanged TFC QI membranes and TFC QI membranes partially anion-exchanged by exposure to 0.01 M sodium salt solutions Solvents and compositions investigated with monomers 13 and 14 from room temperature to 95 C for LLC behavior using PLM xi

13 LIST OF FIGURES Figure 1.1. Depiction of the size of different solutes removed by RO, NF, UF, and MF membranes Depiction of molecular transport through porous membranes and dense membranes described by the pore-flow and solution-diffusion model, respectively Schematic illustration of an anion-exchange membrane in equilibrium with a dilute electrolyte solution, the distribution of ions in this situation, and the difference in the potential between the anion-exchange membrane and the bulk solution Illustration of common LLC morphologies formed by the self-assembly of amphiphiles Schematic illustration of the process used to create TFC QI membranes. The reactive monomer and glycerol is dissolved in a casting solvent and the QI-phase solution is roll-cast onto a porous support. The casting solvent is slowly evaporated and the sample is heated to form the QI-phase and subsequently cross-linked. The final result is a cross-linked TFC QI membrane Monomer 3, its QI phase with glycerol, and the formation of cross-linked QI-phase TFC membranes PXRD profiles of bulk QI-phase films of 3/glycerol/HMP (79.7/19.8/0.5 (w/w/w)): (a) before, and (b) after photo-cross-linking. Inset: PLM optical textures (50x) (a) Cross-sectional SEM photo and (b) PXRD profile of cross-linked QI TFC membrane prepared via MeOH solution roll-casting of 3/glycerol/HMP (79.7/19.8/0.5 (w/w/w)) S1. Structures of prior gemini LLC monomers 1 and 2 used to form crosslinked QI-phase, melt-infused membranes for water purification that operate via a molecular sieving mechanism S2. General synthesis scheme for monomer 3 and several of its homologues that were tested for Q LLC phase formation in the presence of different solvents...37 xii

14 2.S3. Representative PLM images (magnification: 50x) of solvent penetration scans of 3 with (a) water at 68 C, (b) glycerol at 65 C, (c) formamide at 55 C, and (d) ethylammonium nitrate at 44 C. The black (pseudoisotropic) region between two bright, anisotropic LLC regions is indicative of the presence of a potential Q phase. The arrows in the PLM images point in the direction of increasing solvent concentration S4. Elucidated partial phase diagram highlighting the position of the QI phase for the 3/water/glycerol system at 65 C. (Other non-cubic LLC phases were also observed outside of the QI-phase region, but these have not been rigorously identified yet.) S5. PXRD profiles of the bulk cross-linked QI phases of the 1/water, 2/water, and 3/glycerol systems shown to the same 2θ (x-axis) scale S6. FT-IR spectra of 3/glycerol/HMP (79.7/19.8/0.5 (w/w/w)) QI-phase mixture: (a) before, and (b) after heating to 70 C and polymerizing with 365 nm UV light (1 mw cm -2 ) for 1 h. Attenuation of the 1004 cm -1 peak relative to the 1160 cm -1 internal reference peak suggests >90% 1,3-diene group conversion S7. FT-IR spectra of 3/glycerol/HMP (79.6/19.7/0.7 (w/w/w)) before (dried solution-cast monomer mixture) and after (thin-film cross-linked QI polymer) heating to 70 C and polymerizing with 365 nm UV light (1 mw cm -2 ) for 1 h. Attenuation of the 1004 cm -1 peak suggests >90% 1,3-diene group conversion S8. PLM optical texture and PXRD spectrum of a thin film of 3/glycerol/HMP (79.6/19.7/0.7 (w/w/w)) solution-cast from methanol onto glass, heated to 70 C to remove the methanol, and photo-cross-linked with 365 nm UV light (1 mw cm -2 ) for 1 h under Ar atmosphere S9. Cross-section and angled top-view SEM images of the prepared PES support membranes S10. Scheme for solution roll-casting and then cross-linking a thin QI-phase 3/glycerol/HMP top layer on porous PES supports to prepare nanoporous TFC QI polymer membranes S11. FT-IR spectra of solution-cast 3/glycerol/HMP (79.7/19.8/0.5 (w/w/w)) top layer (ca. 3 µm thick) on porous PES support before (blue trace) and after (red trace) heating to 70 C and polymerizing with 365 nm UV light (1 mw cm -2 ) for 1 h. Absorbance attenuation of the 1004 cm -1 peak suggests a high degree of 1,3-diene group conversion...57 xiii

15 2.S12. Cross-sectional SEM image of a typical cross-linked TFC QI membrane fabricated by roll-casting a methanol solution of 3/glycerol/HMP onto porous PES support S13. Picture of the final TFC QI polymer membrane prepared from 3/glycerol/HMP solution roll-cast on porous PES S14. Model for applying the Ferry equation for rejection performance of membranes with uniform circular pores to a QI-phase system with a uniform water layer manifold to determine layer gap spacing S15. Experimentally measured rejection data for the TFC QI membranes made from 3/glycerol/HMP for several non-charged solute molecules, and the Ferry equation plot with a uniform pore size (rpore) of 0.96 nm overlaid on these experimental data S16. Comparison of rejections of TFC QI, SW30HR, and NF-270 membranes (stirred dead-end filtration; 400 psi; 2000 ppm aq. feed solutions) Imidazolium-based monomer 3 that forms a QI phase with glycerol, and the formation of cross-linked TFC QI membranes PXRD spectra of the cross-linked TFC QI membrane and the uncoated GE UF PSf support Cross-sectional SEM photo of the TFC QI membrane made with the GE UF support Relative flux of the TFC QI membranes exposed to different feeds with the sequential order listed from left to right. The experiments were performed in stirred dead-end filtration cells with an applied pressure of Pa (400 psi) and the concentration of all sodium salt feeds was 0.01 M PXRD profiles of anion-exchanged bulk QI films UV-visible spectra of a bulk cross-linked QI film before, during, and after soaking in a 0.5 M aq. NaI solution UV-visible spectra of a bulk cross-linked QI film before, during, and after soaking in a 0.5 M aq. NaI solution adjusted to a ph of Rejection of completely anion-exchanged TFC QI membranes. All experiments performed in stirred dead-end filtration cells with an applied pressure of Pa (400 psi). The concentration of the sodium salt solutions was 0.01 M, and the concentration of the aqueous organic feed solutions was 2000 ppm...94 xiv

16 3.9. Flux of iodide-exchanged TFC QI membranes tested under alkaline (ph = 10) conditions as a function of time with pure water and 0.01 M aq. NaI feed solutions Synthesis scheme for gemini imidazolium LLC monomer 3 that forms QI phases with water and glycerol Original design concept for novel, cross-linkable, zwitterionic amphiphiles General synthesis scheme for new gemini zwitterionic amphiphilic monomers 13 and Representative PLM optical textures of LLC mixtures: (a) LLC phase consisting of 81.2/18.8 (w/w) 13/H2O at R.T. after annealing (b) LLC phase consisting of 90.3/9.7 (w/w) 14/H2O at R.T. upon mixing (c) LLC phase consisting of 69.9/30.1 (w/w) 14/H2O at R.T. upon mixing (d) LLC phase consisting of 60.3/39.7 (w/w) 14/H2O at R.T. after annealing (e) LLC phase consisting of 49.6/50.4 (w/w) 14/H2O at R.T. after annealing (f) LLC phase consisting of 89.9/10.1 (w/w) 14/glycerol at R.T. upon mixing H (top) and 13 C (bottom) NMR spectra of monomer 13 in CD2Cl H (top) and 13 C (bottom) NMR spectra of monomer 14 in CD2Cl Structures of other possible cross-linkable, gemini zwitterionic amphiphiles xv

17 CHAPTER 1 Introduction and background 1.1. The demand for clean water Despite the large abundance of water on earth, only a small fraction is drinkable or useful for agriculture. Most of Earth s water is saltwater (97%) found in the oceans and is unsuitable for human consumption or irrigation [1,2]. Only 2.5% of Earth s water is freshwater and the remaining percentage of water found on earth is brackish water (i.e., slightly salty water found in estuaries and underground aquifers) [1-3]. Most of Earth s freshwater is locked is locked in polar icecaps or permanent mountain snow cover, leaving less than one percent of all freshwater on earth usable by humans [1-3]. The general lack of usable freshwater and its uneven geographic distribution make clean water a scarce and vital resource. Inadequate access to clean freshwater is a well-recognized worldwide problem [2-6]. It is estimated that over 1 billion people lack access to safe drinking water and that over 2.3 billion people live in water-stressed areas [2,4,5]. Not only is clean water essential for survival, but it significantly impacts energy and food production, as well as industrial output and the quality of the environment [2,5,6]. Of the freshwater used by humans, 70% is used for agriculture, 20% is utilized by industry, and the remaining 10% is used for domestic purposes [2]. Consequently, adequate access to clean water is essential for sustaining human populations. As the global population continues to increase, the demand for clean water will intensify. The ability to efficiently and economically produce clean water from a variety of sources is imperative as the demand for clean water increases globally. 1

18 1.2. Overview of membrane technology for aqueous separations Membrane technology is an effective and efficient means to produce clean water from seawater, brackish water, or even contaminated freshwater supplies [5,7-9]. In these processes, the membrane acts as a semipermeable barrier or sieve, selectively allowing water to pass through the membrane while limiting the transport of solutes [10,11]. This process allows undesired ions or solutes in contaminated sources of water to be selectively removed to afford purified water. The transport of water through the membrane is driven by a difference in the chemical potential between the feed and permeate side, which is most commonly supplied by a pressure difference. Since no phase change is required in this process, there are significant energy savings compared to other water purification techniques such as distillation [3,10,11]. Not only is membrane technology useful for water purification, but water filtration membranes can also be used for a number of other important separations [7,10-13]. Water filtration membranes can be utilized to fractionate solutes of different size or isolate and purify valuable products. In addition, water filtration membranes are currently utilized in numerous applications in the food, chemical, pharmaceutical, and medical industries [10-13]. Aqueous separations via membrane technology employ a number of different types of membranes. They include microfiltration (MF), ultrafiltration (UF), nanofiltration (NF), and reverse osmosis (RO) membranes [2,7,9,14]. Figure 1.1 depicts the size of different solutes removed by these different types of water filtration membranes. MF membranes have the largest average pore diameter, typically rejecting solid or colloidal particles and various microorganisms in the approximate size range of 0.1 to 10 μm [7,9]. UF membranes have smaller pores, typically in the range of nm and can separate water and molecular-size solutes from macromolecules such as biomolecules and polymers [7,11]. In order to remove virtually all 2

19 dissolved salt ions and small organics from water, RO membranes can be employed. The separating layer (referred to as the active layer ) in RO membranes is typically dense and does not contain well-defined or discrete pores. Water passes through RO membranes by moving through the 2 5 Å gaps and voids between the polymer chains [7,12]. NF membranes have very small discrete pores on the order of 1 nm and possess filtration properties in between those of UF and RO membranes [8,12,15]. NF membranes reject most divalent ions and most organic solutes but only partially reject monovalent ions and very small organics [8,12,15]. Figure 1.1. Depiction of the size of different solutes removed by RO, NF, UF, and MF membranes. Partially adapted from references [2, 7]. For the different types of filtration membranes, the average pore diameter directly impacts their productivity and required operating pressure. The flux of a membrane is directly related to the thickness, the number of pores, and the pore diameter. A simple description of a membrane as a series of cylindrical pores and use of Poiseuille s law shows the permeance (i.e., pressure-normalized flux) of membranes of similar thickness and pore area significantly decreases with a decrease in the pore diameter [7,9]. As a result, MF membranes have permeance 3

20 values significantly higher than UF membranes and vastly higher than RO membranes [7,9]. Membranes that remove large contaminants typically require low operating pressures, and membranes with small pore diameters that can remove molecular-size solutes typically require much higher operating pressures. The difference in the rejection performance and relative permeance of water filtration membranes significantly impacts their use and application Separation mechanisms of water filtration membranes The purpose of a membrane is to selectively allow one component to pass through the membrane while limiting the transport of other components to achieve a separation. The mechanism through which the separation is achieved can be different depending on the type of membrane. There are two general models used to describe transport through water filtration membranes: the solution-diffusion model and the pore-flow model as depicted in Figure 1.2 [7,16,17]. Figure 1.2. Depiction of molecular transport through porous membranes and dense membranes described by the pore-flow and solution-diffusion models, respectively. Adapted from reference [7]. 4

21 The solution-diffusion model describes transport through dense materials that do not have discrete pores [2,7]. In this model, solutes and water in the feed dissolve in the active layer of the membrane and then diffuse through the membrane. Solutes are rejected by the membrane due to the differences in the solubility and mobility of the feed components in the dense active layer. Transport through RO membranes is typically described by the solution-diffusion model, since the active layer in RO membranes is completely dense and does not contain discrete pores [2,7]. The second model used to describe transport through water filtration membranes is the pore-flow model where transport is described by pressure-driven convective flow through tiny pores [7,9]. Unlike the solution-diffusion model, which is a unified theory used to describe transport through all dense membrane materials, there is no equivalent unified theory to describe transport through porous membranes. Since porous membranes can be processed with such a wide range of structures and morphologies, it is difficult characterize the exact mechanism of separation, whether it be size exclusion, adsorption, or both. For screen filter membranes that consist of a finely porous surface layer and a more open porous substructure, size exclusion is the predominant mechanism for achieving a separation [7,9]. Solutes larger than the surface pore diameter cannot pass through the membrane and feed components smaller than the pore diameter begin to pass through the membrane. Size-exclusion is relatively easy to describe mathematically and can utilize rejection measurements to estimate the average pore size of a membrane [7]. UF membranes are typically described by a size-exclusion pore-flow model, as well as some screenfilter type MF membranes [7,9]. While RO membranes are described using the solution-diffusion model, and UF membranes are described using size-exclusion pore-flow models, NF membranes have pores around 1 nm in diameter, and the membrane is often partially charged [2,12,15]. At this length 5

22 scale, macroscopic descriptions of hydrodynamics begin to fail. This puts NF membranes at a transition region between pore-flow and solution-diffusion transport models. Even though it is difficult to describe the transport at this length scale, the rejection of neutral solutes by NF membranes is still well described by a pore-flow model where separation is achieved via molecular size-exclusion [2,12,15]. However, due to the extremely small pore size of NF membranes, any surface charge on the membrane can result in appreciable rejection of ions that would normally pass through based solely on size. As a result, describing transport through NF membranes typically involves space-charge models, where both size-exclusion and charge interactions are taken into account [12,15]. In addition to the two general models used to describe the mechanism of permeation through water filtration membranes, membrane charge significantly impacts membrane permeability and the rejection of charged solutes such as salt ions [2,7,18]. The rejection of charged solutes by a charged membrane is often qualitatively described by the Donnan exclusion mechanism [7,12,18,19]. For a membrane that contains fixed charges, every fixed positive or negative charge is counter-balanced by a mobile oppositely charged ion (i.e., counterion) to satisfy electroneutrality on the macroscopic scale. When an ion-exchange membrane is in equilibrium with a dilute electrolyte solution, the counterion concentration is significantly lower in the solution than in the membrane, and the co-ion (mobile ions with the same sign charge as the fixed membrane charge) concentration is significantly higher in the solution than in the membrane. This is depicted in Figure 1.3 for an anion-exchange membrane. The steep gradients in the concentration differences between the ion-exchange membrane and the bulk solution are counter-balanced by an electrochemical potential difference called the Donnan potential. The Donnan potential illustrates that work must be done to move co-ions from the bulk solution into 6

23 the membrane phase. Since co-ions are repelled by the charged membrane, counterions are retained as well due to the requirement of electroneutrality on a macroscopic scale. This general mechanism of rejection is referred to as Donnan exclusion. The rejection of ions via Donnan exclusion strongly depends on the charge density of the membrane and the concentration of ions in solution. Donnan exclusion is most commonly used qualitatively to describe the rejection of ions by charged ion-exchange membranes. Figure 1.3. Schematic illustration of an anion-exchange membrane in equilibrium with a dilute electrolyte solution, the distribution of ions in this situation, and the difference in the potential between the anion-exchange membrane and the bulk solution. Adapted from reference [19]. 7

24 1.4. Current issues with nanofiltration and reverse osmosis membranes Current commercial NF and RO membranes are capable of effectively and efficiently producing large amounts of high purity water and are also used in a number of different industries for various aqueous separations [7,12]. Despite their heavy commercial use, current NF and RO membranes face a number of challenges, such as a lack of control over the average pore size and pore size distribution [10,12,20,21], limited resistance to harsh chemicals [8,22-25], limited control over membrane charge character and charge density [2,12,26-29], and limited resistance to fouling from various sources [12,30-33]. The development of novel NF and RO membranes that can overcome these limitations would offer significant advantages over current commercial membranes. One of the inherent difficulties in the construction of NF and RO membranes is the inability to control the average size and distribution of the effective pores or interstitial voids on the molecular size scale in the active layer [10,12,20,21]. A wide pore size distribution limits the rejection selectivity by allowing solutes that are larger than the average pore diameter to pass through the membrane. A wide pore size distribution also makes it much more difficult to cleanly fractionate solutes of varying size. The inability to control the pore size and pore size distribution in current NF and RO membranes is a result of the processing techniques that are used to create membranes with an extremely thin active layer necessary for high throughput. The two primary production techniques used to fabricate NF and RO membranes are phase inversion and interfacial polymerization [7,8,10]. Phase inversion membranes are typically made by taking a polymer dissolved in a water-miscible organic solvent and casting a thick film onto a fabric backing and then immediately immersing it into a water bath [7,8]. The water rapidly precipitates the top surface of the cast polymer film and this selective skin layer slows the 8

25 entry of water into the rest of the polymer solution, allowing a porous substructure to slowly form. The resulting anisotropic membranes have a macroporous substructure that supports a very thin, active separation layer on the top surface. The second production method used to create NF and RO membranes is interfacial polymerization [7,10]. In this technique, a microporous support membrane is soaked in a water solution containing a reactive monomer, most commonly multifunctional amines. The support soaked with the reactive amine solution is then immersed in a water-immiscible solvent containing a different reactive monomer, most commonly multifunctional acid chlorides. Upon contact, a step-growth polymerization reaction occurs between the two reactive monomers at the interface of the two immiscible solvents on the surface of the support membrane. The result is a thin-film composite (TFC) membrane that contains a highly cross-linked active layer supported by the microporous support. Both of these techniques produce membranes that contain an extremely thin active layer that is on the order of 0.1 μm thick [9]; however, due to the crude nature of these fabrication techniques there is very limited control over the pore size and pore size distribution in the active layer [10,12,20,21]. Fundamentally different processing techniques or materials are needed to produce membranes with uniform pores on the molecular size scale. Phase inversion and interfacial polymerization production methods currently only result in membranes with two major chemistries [8,23,34]. Commercial NF and RO membranes made by the phase inversion process are typically cellulose acetate-based materials, and membranes made by interfacial polymerization are typically aromatic polyamides [8,23,34]. Cellulose acetate-based membranes are limited for use in a relatively narrow ph range before they begin to hydrolyze, and they are also susceptible to microbial attack [23,24]. Membranes based on aromatic polyamide chemistry are extremely sensitive to oxidizing agents such as aq. chlorine 9

26 (e.g., ClO ), a common oxidizing reagent used for disinfection [23,35]. Membrane manufacturers typically recommend that polyamide-based membranes are only exposed to feeds containing less that 0.1 ppm of chlorine [23]. Due to the limited chemical resistance of current commercial NF and RO membranes, strong ph feeds must be neutralized and chlorine must be removed to prevent permanent damage to the membrane [35]. The development of NF and RO membranes that are resistant to these common pretreatment chemicals and other harsh feeds could eliminate expensive pretreatment steps and allow a wider range of operating conditions than current membranes based on cellulose acetate and polyamide chemistries. Not only do cellulose acetate- and polyamide-based membranes have limited resistance to common water treatment chemicals, but they also do not offer control over the charge character and charge density of the membrane [8,28,36-38]. Polyamide-based NF and RO membranes are commonly charged depending on the ph of the solution, but this merely is an artifact of unreacted functional groups left over from the interfacial polymerization synthesis process [36]. There is little to no control over the actual charge density of these groups on the membrane surface. It is well known that membrane charge significantly impacts membrane permeability and rejection performance [18,27,29]. Membrane materials with tunable charge characteristics could be modified to increase membrane permeability and the rejection of charged solutes. A final area where current NF and RO membranes could be improved is their resistance to fouling. Fouling can originate from a number of different sources and results in a decrease of the performance of the membrane that can only be remedied by cleaning or cartridge replacement [2,12,39]. The most common sources of fouling include organic fouling, scaling, colloidal and particulate fouling, and biofouling [3,12,40,41]. Although the mechanism of 10

27 fouling depends on a number of factors, efforts to reduce fouling commonly focus on surface modification of the membrane [12,42]. By modifying the membrane to create a more hydrophilic surface, water strongly bound to the surface prevents interactions with potential foulants [43]. Modifying the surface of the membrane after fabrication typically adds mass-transfer limitations that result in a decrease in water permeance [12]. The development of membrane materials with tunable functional groups and charge characteristics could significantly increase membrane resistance to fouling without a decrease in permeance. Although current NF and RO membranes are effectively utilized in a number of applications, they face a number of challenges as described in detail above: (1) Current phase inversion and interfacial polymerization production methods have little to no control over the effective pore size and pore size distribution in the active layer of the membrane [10,12,20,21]. (2) Conventional cellulose acetate and polyamide chemistries used to construct current NF and RO membranes afford materials with limited chemical resistance to common water treatment chemicals [8,22-25]. (3) These membrane synthesis chemistries offer very limited control over the charge characteristics and charge density of the membrane that could be utilized to improve permeability and rejection performance and or help reduce fouling tendencies [2,12,26-33]. Consequently, the study and development of new materials that can create NF and RO membranes with completely uniform molecular-size pores, tunable charge characteristics, and improved resistance to harsh chemicals and fouling would offer significant advantages over conventional NF and RO membranes. 11

28 1.5. New materials for fabricating nanoporous membranes with uniform pores The fabrication of conventional NF and RO membranes is a well-developed technology and no significant improvements have been made in about the last 40 years [10,44,45]. New materials or processes are necessary to produce membranes with completely uniform pores on the nanometer scale for improved rejection selectivity [46]. New materials chemistry must be utilized in order to optimize charge characteristics and improve resistance to harsh chemical feeds and fouling. A variety of materials have been explored to create membranes with monodisperse pores on the nanometer scale. Materials such as zeolites [34,45,47], carbon nanotubes [34,48,49], aquaporins [50], peptide nanotubes [51], molecular squares [52], macrocylic surfactants [53,54], block copolymers [55-57], colloidal crystal assemblies [58], microemulsions [59,60], thermotropic liquid crystals [61,62], and lyotropic liquid crystals (LLCs) [14,63,64] have been explored. Although most of these materials have ideal structures for membrane transport, many of them suffer from high synthetic cost and or unfeasible/elaborate processing. Only a few of the materials above have been processed into films thin enough to achieve high fluxes [46]. Successful development of the next generation of NF and RO membranes not only requires materials with improved control over pore architecture and resistance to harsh chemicals feeds and fouling, but materials with processability that allows the production of membranes with thin active layers for industrially relevant productivity. This thesis research will focus on the development and processing of cross-linkable LLCs to afford thin-film membranes with uniform, ionic nanopores and investigating how the charge characteristics and chemistry of these materials impact filtration performance. 12

29 1.6. Cross-linkable lyotropic liquid crystals (LLCs) LLCs are amphiphilic (i.e., surfactant) molecules that self-assemble in the presence of a solvent (e.g., most commonly water) to form ordered, phase-separated assemblies that contain uniform, nanometer-scale solvent domains [65-68]. The fluid, yet ordered assemblies typically exhibit morphologies such as lamellar (L), hexagonal (H), bicontinuous cubic (Q) (e.g., gyroid, diamond, primitive), and discontinuous cubic (I) (Figure 1.4) [65,68]. Except for lamellar phases (which have zero mean curvature), LLC phases are further classified as being type I or type II depending on the curvature of the surfactant/solvent interface. Phases that curve around the hydrophobic domains are classified as type I (i.e., normal) and contain excess solvent compared to the lamellar phase [68]. Phases that curve around the hydrophilic domains are classified as type II (i.e., inverted) and have lower solvation levels than the lamellar phase [68]. The specific phase that is formed by various amphiphiles strongly depends on the concentration of surfactant and solvent, as well as the temperature and pressure of the system [65]. 13

30 Figure 1.4. Illustration of common LLC morphologies formed by the self-assembly of amphiphiles. Partially reproduced from reference [69]. Even though LLCs form nanometer-scale morphologies ideal for transport applications, they are ordered materials with fluid-like mechanical properties. When various stresses are applied, LLCs will flow and deform. In addition to lacking mechanical stability, phase changes can occur with variations in composition, temperature, or pressure. All of these variables must 14

31 remain within specific limits in order to maintain the desired phase and morphology. Without modifications, these attributes of LLCs make them unsuitable as a membrane material. By designing and synthesizing reactive, cross-linkable amphiphilic compounds, the mechanical stability issues associated with LLCs can be overcome. Cross-linkable surfactant monomers still self-assemble into ordered LLC phases, but upon organization a polymerization reaction can be induced (e.g., photo-initiated radical polymerization) to create a cross-linked polymer network that is mechanically robust [66,70]. This cross-linked polymer network retains the nanostructure of the LLC phase and will no longer change phases with variations in composition, temperature, or pressure and will not readily deform under significant stresses. Cross-linked LLC networks have been utilized in a number of applications including heterogeneous catalysis, ion transport, and membrane separations [69,71]. For membrane applications, cross-linked Q phase LLC networks have proven to be the most desirable [63,69]. In addition to a high pore density, Q phases have 3D-interconnected nanopores that do not require phase alignment, unlike lower dimensionality hexagonal and lamellar phases. Eliminating phase alignment significantly reduces the processing complexity of these systems. The only drawback to Q phase systems is that a limited number of amphiphiles can form this phase and the Q phase only typically exists in limited concentration and temperature phase windows [14,72,73]. Previously, our research team demonstrated cross-linked type I bicontinuous cubic (QI) phases could be made into supported membranes and filtration studies demonstrated they possessed performance characteristics in-between conventional NF and RO membranes [64,71]. These initial investigations demonstrated QI-phase materials have significant potential as a new type of uniform, nanoporous membrane material; however processing limitations resulted in 15

32 membranes with thick active layers and low fluxes [74]. The only viable method found for membrane fabrication involved heating and pressing the monomer/water mixture into a porous support and subsequently photo-cross-linking to covalently lock in the QI nanostructure. In this processing approach, the support becomes fully infused with QI-phase material, and so the active layer of the membrane is as thick as the support itself (ca. 40 μm). The water flux of supported QI-phase membranes is far too low for any practical application or extensive testing due to the extremely thick active layer. Even though the flux of the melt-infused QI-phase membranes is too low for any application and significantly lower than commercial NF and RO membranes, the thickness-normalized pure water permeance is comparable to that of a commercial RO membrane [74]. This means it should be possible for QI-phase membranes to have a permeance comparable to a commercial RO membrane if the active layer of QI-phase membranes can be reduced to the same approximate 0.1 μm active separation layer thickness of commercial membranes [74]. Initial attempts to make thinner QI films using a variety of approaches have been unsuccessful. The major difficulty in trying to prepare a cross-linked thin-film of a QI-phase on top of a porous support is that QI phases are particularly sensitive to changes in composition and temperature. A specific ratio of reactive monomer and water must be applied as a thin film on top of a support and this composition must be maintained while the sample is heated to the required temperature and polymerized. Any evaporative loss of water during this process will disrupt the QI-phase assembly and lead to undesired phase changes before the nanostructure is covalently locked in place. The most straightforward and facile method to apply a thin film of a QI-phase mixture will utilize solution-based processing. The QI-phase monomer mixture will be dissolved in a suitable casting solvent and then applied to the support by common coating 16

33 techniques. However, any solution-based processing approach is not feasible for water-based QIphase mixtures, since it is not possible to remove casting solvent without removing some of the water required to form the QI phase. Until an approach is developed to create thin films of QI polymer on top of a porous support, the full potential and characterization of cross-linked QIphase assemblies as a new type membrane material cannot be realized Thesis objectives: Design, development, and evaluation of thin-film composite bicontinuous cubic polymer membranes As discussed above, cross-linked QI-phase networks have enormous potential as a membrane material with completely uniform pores on the nanometer scale and tunable chemistry that could offer significant improvements over conventional NF and RO membranes. Despite their significant potential, cross-linked QI-phase networks have never been processed into thinfilms necessary to demonstrate industrially relevant productivity. Current water-based QI-phase systems show no promise for being processed into TFC QI membranes. Additionally, limited characterization has been performed on QI-phase membranes to understand their fundamental performance characteristics. More thorough characterization of the rejection properties of QIphase membranes are needed to better understand aqueous transport behavior in these materials. More extensive filtration studies on QI-phase materials will also better gauge the performance of QI-phase membranes relative to commercial NF and RO membranes. Until the fabrication of TFC QI membranes can be demonstrated and additional fundamental filtration studies can be completed, cross-linked QI materials remain distant from commercial application. The overall objective of this thesis work is to design and develop a method or approach to fabricate ionic TFC QI membranes and to perform additional fundamental filtration studies to 17

34 better understand the transport characteristics of QI-phase membranes with respect to pore size and charge effects. Since current water-based QI-phase monomer systems cannot be processed into thin-films using solution-based processing, new methods or approaches must be investigated. Chapter 2 explores a new fundamental approach for processing QI-phase materials into cross-linked thin films. In this approach, the water typically used for LLC phase formation is replaced with the low volatility, polar organic solvent glycerol. Given glycerol s high boiling point and low vapor pressure, it was demonstrated that a casting solvent used for thin-film solution processing can readily be removed with minimal evaporative loss of glycerol necessary for QI-phase formation. This allows a solution of the monomer and glycerol to be applied to a support via solution-casting and after mild heating to remove the casting solvent, the applied film can be photo-cross-linked to create a TFC QI membrane (Figure 1.5). 18

35 Figure 1.5. Schematic illustration of the process used to create TFC QI membranes. The reactive monomer and glycerol is dissolved in a casting solvent and the QI-phase solution is roll-cast onto a porous support. The casting solvent is slowly evaporated and the sample is heated to form the QI-phase and subsequently cross-linked. The final result is a cross-linked TFC QI membrane. Chapter 3 investigates in more detail the filtration performance TFC QI membranes and in particular, how anion-exchange in the QI-phase material impacts performance. In these filtrations studies, it was found that anion-exchange significantly impacts the permeability of TFC QI membranes, but has little to no effect on the rejection performance. These results demonstrate the flux of TFC QI membranes can be tuned by exposing the membrane to a feed with a specific anion with little to no change in the rejection performance. The unique performance characteristics of TFC QI membranes may offer advantages over conventional NF and RO membranes for a number of different separations. 19

36 Chapter 4 describes the design, synthesis, and phase behavior of new bis(benzimidazolium)-based LLC monomers containing covalently tethered counterions that form LLC phases. The effects of anion-exchange on TFC QI membrane performance detailed in Chapter 3 showed an unexpected liability of polymer membranes made from ionic LLC monomers with free counterions. This prompted the design and synthesis of new cross-linkable zwitterionic amphiphiles in which counterion-exchange is no longer possible. Such an unprecedented LLC monomer design is important for future water purification applications in order to achieve more stable water permeance if feeds with different anions are present References [1] Shiklomanov, I. A. Water in Crisis: A Guide to the World's Fresh Water Resources; Oxford University Press: New York, [2] Geise, G. M.; Lee, H.-S.; Miller, D. J.; Freeman, B. D.; McGrath, J. E.; Paul, D. R. "Water purification by membranes: The role of polymer science." J. Polym. Sci., Part B: Polym. Phys. 2010, 48, [3] Greenlee, L. F.; Lawler, D. F.; Freeman, B. D.; Marrot, B.; Moulin, P. "Reverse osmosis desalination: Water sources, technology, and today's challenges." Water Res. 2009, 43, [4] Service, R. F. "Desalination Freshens Up." Science 2006, 313, [5] Shannon, M. A.; Bohn, P. W.; Elimelech, M.; Georgiadis, J. G.; Marinas, B. J.; Mayes, A. M. "Science and technology for water purification in the coming decades." Nature 2008, 452, [6] Oki, T.; Kanae, S. "Global Hydrological Cycles and World Water Resources." Science 2006, 313, [7] Baker, R. W. Membrane Technology and Applications; McGraw-Hill: New York, [8] Petersen, R. J. "Composite reverse osmosis and nanofiltration membranes." J. Membr. Sci. 1993, 83, [9] Noble, R. D.; Stern, S. A. Membrane Separations Technology: Principles and Applications; Elsevier: Amsterdam,

37 [10] Lonsdale, H. K. "The growth of membrane technology." J. Membr. Sci. 1982, 10, [11] Ho, W. S. W.; Sirkar, K. K. Membrane Handbook; Chapman & Hall: New York, [12] Schäfer, A. L.; Fane, A. G.; Waite, T. D. Nanofiltration: Principles and Applications; Elsevier: New York, [13] van Reis, R.; Zydney, A. "Bioprocess membrane technology." J. Membr. Sci. 2007, 297, [14] Gin, D. L.; Bara, J. E.; Noble, R. D.; Elliott, B. J. "Polymerized Lyotropic Liquid Crystal Assemblies for Membrane Applications." Macromol. Rapid Commun. 2008, 29, [15] Bowen, W. R.; Welfoot, J. S. "Modelling the performance of membrane nanofiltration-- critical assessment and model development." Chem. Eng. Sci. 2002, 57, [16] Wijmans, J. G.; Baker, R. W. "The solution-diffusion model: a review." J. Membr. Sci. 1995, 107, [17] Paul, D. R. "Reformulation of the solution-diffusion theory of reverse osmosis." J. Membr. Sci. 2004, 241, [18] Bhattacharya, A.; Ghosh, P. "Nanofiltration and Reverse Osmosis Membranes: Theory and Application in Separation of Electrolytes." Rev. Chem. Eng. 2004, 20, [19] Strathmann, H. Ion-Exchange Membrane Separation Processes; Elsevier: Amsterdam, [20] Ulbricht, M. "Advanced functional polymer membranes." Polymer 2006, 47, [21] Aimar, P.; Meireles, M.; Sanchez, V. "A contribution to the translation of retention curves into pore size distributions for sieving membranes." J. Membr. Sci. 1990, 54, [22] Glater, J.; Hong, S.-k.; Elimelech, M. "The search for a chlorine-resistant reverse osmosis membrane." Desalination 1994, 95, [23] Park, H. B.; Freeman, B. D.; Zhang, Z.-B.; Sankir, M.; McGrath, J. E. "Highly Chlorine- Tolerant Polymers for Desalination." Angew. Chem. 2008, 120, [24] Li, D.; Wang, H. "Recent developments in reverse osmosis desalination membranes." J. Mater. Chem. 2010, 20, [25] Cadotte, J. E.; Petersen, R. J.; Larson, R. E.; Erickson, E. E. "A new thin-film composite seawater reverse osmosis membrane." Desalination 1980, 32, [26] Childress, A. E.; Elimelech, M. "Effect of solution chemistry on the surface charge of polymeric reverse osmosis and nanofiltration membranes." J. Membr. Sci. 1996, 119,

38 [27] Childress, A. E.; Elimelech, M. "Relating Nanofiltration Membrane Performance to Membrane Charge (Electrokinetic) Characteristics." Environ. Sci. Technol. 2000, 34, [28] Paul, M.; Park, H. B.; Freeman, B. D.; Roy, A.; McGrath, J. E.; Riffle, J. S. "Synthesis and crosslinking of partially disulfonated poly(arylene ether sulfone) random copolymers as candidates for chlorine resistant reverse osmosis membranes." Polymer 2008, 49, [29] Yaroshchuk, A.; Staude, E. "Charged membranes for low pressure reverse osmosis properties and applications." Desalination 1992, 86, [30] Hatakeyama, E. S.; Ju, H.; Gabriel, C. J.; Lohr, J. L.; Bara, J. E.; Noble, R. D.; Freeman, B. D.; Gin, D. L. "New protein-resistant coatings for water filtration membranes based on quaternary ammonium and phosphonium polymers." J. Membr. Sci. 2009, 330, [31] Ju, H.; McCloskey, B. D.; Sagle, A. C.; Wu, Y.-H.; Kusuma, V. A.; Freeman, B. D. "Crosslinked poly(ethylene oxide) fouling resistant coating materials for oil/water separation." J. Membr. Sci. 2008, 307, [32] Vrijenhoek, E. M.; Hong, S.; Elimelech, M. "Influence of membrane surface properties on initial rate of colloidal fouling of reverse osmosis and nanofiltration membranes." J. Membr. Sci. 2001, 188, [33] Ang, W. S.; Lee, S.; Elimelech, M. "Chemical and physical aspects of cleaning of organic-fouled reverse osmosis membranes." J. Membr. Sci. 2006, 272, [34] Lee, K. P.; Arnot, T. C.; Mattia, D. "A review of reverse osmosis membrane materials for desalination--development to date and future potential." J. Membr. Sci. 2011, 370, [35] Elimelech, M.; Phillip, W. A. "The Future of Seawater Desalination: Energy, Technology, and the Environment." Science 2011, 333, [36] Coronell, O.; Mariñas, B. J.; Zhang, X.; Cahill, D. G. "Quantification of Functional Groups and Modeling of Their Ionization Behavior in the Active Layer of FT30 Reverse Osmosis Membrane." Environ. Sci. Technol. 2008, 42, [37] Schaep, J.; Vandecasteele, C. "Evaluating the charge of nanofiltration membranes." J. Membr. Sci. 2001, 188, [38] Teixeira, M. R.; Rosa, M. J.; Nyström, M. "The role of membrane charge on nanofiltration performance." J. Membr. Sci. 2005, 265, [39] Koros, W. J.; Ma, Y. H.; Shimidzu, T. "Terminology for membranes and membrane processes (IUPAC Recommendation 1996)." J. Membr. Sci. 1996, 120, [40] Amiri, M. C.; Samiei, M. "Enhancing permeate flux in a RO plant by controlling membrane fouling." Desalination 2007, 207,

39 [41] Baker, J. S.; Dudley, L. Y. "Biofouling in membrane systems -- A review." Desalination 1998, 118, [42] Rana, D.; Matsuura, T. "Surface Modifications for Antifouling Membranes." Chem. Rev. 2010, 110, [43] Kane, R. S.; Deschatelets, P.; Whitesides, G. M. "Kosmotropes Form the Basis of Protein-Resistant Surfaces." Langmuir 2003, 19, [44] Lonsdale, H. K. "The evolution of ultrathin synthetic membranes." J. Membr. Sci. 1987, 33, [45] Jeong, B.-H.; Hoek, E. M. V.; Yan, Y.; Subramani, A.; Huang, X.; Hurwitz, G.; Ghosh, A. K.; Jawor, A. "Interfacial polymerization of thin film nanocomposites: A new concept for reverse osmosis membranes." J. Membr. Sci. 2007, 294, 1-7. [46] Gin, D. L.; Noble, R. D. "Designing the Next Generation of Chemical Separation Membranes." Science 2011, 332, [47] Tavolaro, A.; Drioli, E. "Zeolite Membranes." Adv. Mater. 1999, 11, [48] Hinds, B. J.; Chopra, N.; Rantell, T.; Andrews, R.; Gavalas, V.; Bachas, L. G. "Aligned Multiwalled Carbon Nanotube Membranes." Science 2004, 303, [49] Chan, W.-F.; Chen, H.-y.; Surapathi, A.; Taylor, M. G.; Shao, X.; Marand, E.; Johnson, J. K. "Zwitterion Functionalized Carbon Nanotube/Polyamide Nanocomposite Membranes for Water Desalination." ACS Nano 2013, 7, [50] Kumar, M.; Grzelakowski, M.; Zilles, J.; Clark, M.; Meier, W. "Highly permeable polymeric membranes based on the incorporation of the functional water channel protein Aquaporin Z." PNAS 2007, 104, [51] Xu, T.; Zhao, N.; Ren, F.; Hourani, R.; Lee, M. T.; Shu, J. Y.; Mao, S.; Helms, B. A. "Subnanometer Porous Thin Films by the Co-assembly of Nanotube Subunits and Block Copolymers." ACS Nano 2011, 5, [52] Czaplewski, K. F.; Hupp, J. T.; Snurr, R. Q. "Molecular Squares as Molecular Sieves: Size-Selective Transport Through Porous-Membrane-Supported Thin-Film Materials." Adv. Mater. 2001, 13, [53] Yan, X.; Janout, V.; Hsu, J. T.; Regen, S. L. "A Polymerized Calix[6]arene Monolayer Having Gas Permeation Selectivity that Exceeds Knudsen Diffusion." J. Am. Chem. Soc. 2002, 124, [54] Yan, X.; Janout, V.; Hsu, J. T.; Regen, S. L. "The Gluing of a Langmuir Blodgett Bilayer." J. Am. Chem. Soc. 2003, 125,

40 [55] Hillmyer, M. In Block Copolymers II; Abetz, V., Ed.; Springer Berlin Heidelberg: 2005; Vol. 190, p [56] Olson, D. A.; Chen, L.; Hillmyer, M. A. "Templating Nanoporous Polymers with Ordered Block Copolymers." Chem. Mater. 2007, 20, [57] Peinemann, K.-V.; Abetz, V.; Simon, P. F. W. "Asymmetric superstructure formed in a block copolymer via phase separation." Nat. Mater. 2007, 6, [58] Newton, M. R.; Bohaty, A. K.; White, H. S.; Zharov, I. "Chemically Modified Opals as Thin Permselective Nanoporous Membranes." J. Am. Chem. Soc. 2005, 127, [59] Li, T. D.; Gan, L. M.; Chew, C. H.; Teo, W. K.; Gan, L. H. "Preparation of Ultrafiltration Membranes by Direct Microemulsion Polymerization Using Polymerizable Surfactants." Langmuir 1996, 12, [60] Liu, J.; Teo, W. K.; Chew, C. H.; Gan, L. M. "Nanofiltration membranes prepared by direct microemulsion copolymerization using poly(ethylene oxide) macromonomer as a polymerizable surfactant." J. Appl. Polym. Sci. 2000, 77, [61] Yoshio, M.; Mukai, T.; Ohno, H.; Kato, T. "One-Dimensional Ion Transport in Self- Organized Columnar Ionic Liquids." J. Am. Chem. Soc. 2004, 126, [62] Ichikawa, T.; Yoshio, M.; Hamasaki, A.; Kagimoto, J.; Ohno, H.; Kato, T. "3D Interconnected Ionic Nano-Channels Formed in Polymer Films: Self-Organization and Polymerization of Thermotropic Bicontinuous Cubic Liquid Crystals." J. Am. Chem. Soc. 2011, 133, [63] Gin, D. L.; Lu, X.; Nemade, P. R.; Pecinovsky, C. S.; Xu, Y.; Zhou, M. "Recent Advances in the Design of Polymerizable Lyotropic Liquid-Crystal Assemblies for Heterogeneous Catalysis and Selective Separations." Adv. Funct. Mater. 2006, 16, [64] Zhou, M. J.; Nemade, P. R.; Lu, X. Y.; Zeng, X. H.; Hatakeyama, E. S.; Noble, R. D.; Gin, D. L. "New type of membrane material for water desalination based on a crosslinked bicontinuous cubic lyotropic liquid crystal assembly." J. Am. Chem. Soc. 2007, 129, [65] Tiddy, G. J. T. "Surfactant-water liquid crystal phases." Phys. Rep. 1980, 57, [66] Gin, D. L.; Gu, W.; Pindzola, B. A.; Zhou, W.-J. "Polymerized Lyotropic Liquid Crystal Assemblies for Materials Applications." Acc. Chem. Res. 2001, 34, [67] Seddon, J. M. "Structure of the inverted hexagonal (HII) phase, and non-lamellar phase transitions of lipids." Biochim. Biophys. Acta 1990, 1031, [68] Tate, M. W.; Eikenberry, E. F.; Turner, D. C.; Shyamsunder, E.; Gruner, S. M. "Nonbilayer phases of membrane lipids." Chem. Phys. Lipids 1991, 57,

41 [69] Wiesenauer, B. R.; Gin, D. L. "Nanoporous polymer materials based on self-organized, bicontinuous cubic lyotropic liquid crystal assemblies and their applications." Polym. J. 2012, 44, [70] Mueller, A.; O'Brien, D. F. "Supramolecular Materials via Polymerization of Mesophases of Hydrated Amphiphiles." Chem. Rev. 2002, 102, [71] Hatakeyama, E. S.; Wiesenauer, B. R.; Gabriel, C. J.; Noble, R. D.; Gin, D. L. "Nanoporous, Bicontinuous Cubic Lyotropic Liquid Crystal Networks via Polymerizable Gemini Ammonium Surfactants." Chem. Mater. 2010, 22, [72] Fontell, K. "Cubic phases in surfactant and surfactant-like lipid systems." Colloid Polym. Sci. 1990, 268, [73] Sorenson, G. P.; Coppage, K. L.; Mahanthappa, M. K. "Unusually Stable Aqueous Lyotropic Gyroid Phases from Gemini Dicarboxylate Surfactants." J. Am. Chem. Soc. 2011, 133, [74] Hatakeyama, E. S.; Gabriel, C. J.; Wiesenauer, B. R.; Lohr, J. L.; Zhou, M.; Noble, R. D.; Gin, D. L. "Water filtration performance of a lyotropic liquid crystal polymer membrane with uniform, sub-1-nm pores." J. Membr. Sci. 2011, 366,

42 CHAPTER 2 Glycerol-based bicontinuous cubic lyotropic liquid crystal monomer system for the fabrication of thin-film membranes with uniform nanopores (Manuscript published under the same title in Chemistry of Materials 2012, 24, , coauthored with Wiesenauer, B. W.; Hatakeyama, E. S.; Barton, J. L.; Noble, R. D.; Gin, D. L. ) Abstract A glycerol-based, type I bicontinuous cubic lyotropic liquid crystal monomer system has been developed that allows facile solution fabrication of thin-film composite membranes with 3D-interconnected, ionic nanopores. These membranes have a ca. 3-µm-thick top layer with uniform, 0.96-nm-wide, cationic pores that size-exclude uncharged solutes from water. They also reject smaller hydrated salt ions near or at the level of reverse osmosis (RO) membranes (>98%) due to (cationic pore metal ion) repulsive interactions, making this material quite unique in terms of its water purification performance. It also has a thickness-normalized pure water permeability comparable to active layer materials in RO membranes Introduction Lyotropic liquid crystal (LLC) networks are nanoporous polymer materials formed by the in situ cross-linking of reactive amphiphiles (i.e., surfactants) that self-organize in water or other polar solvents into ordered, phase-separated assemblies [1-3]. They contain periodic, uniformsize, nanoscale solvent regions/pores (i.e., 1 to 10 nm) that have different possible geometries and are lined by the amphiphile headgroups, thus allowing for pore control and functionalization 26

43 [1-3]. Because of these features, LLC networks have been shown to be valuable for many applications [1,3]. One important application area where LLC networks have been found to have great potential is membrane-based water purification/desalination [4]. The ability of cross-linkable LLC systems to generate polymers with monodisperse, sub-1-nm, aqueous pores enables the removal of molecular contaminants from water (0.27 nm diameter) with high selectivity [5]. Also, the ability of LLCs to access different nanopore geometries and produce materials with high pore density is essential for high water filtration throughput (i.e., flux) [5]. In particular, bicontinuous cubic (Q) LLC networks have been found to be the most desirable for this application [4]. In addition to high pore density, Q phases have 3D-interconnected nanopores that do not require alignment for high throughput, unlike those in lower-dimensionality LLC phases (i.e., the 1D cylindrical hexagonal (H) and 2D lamellar (L) phases) [4]. Q phases are classified as type I (i.e., normal) or type II (i.e., inverted) depending on whether the hydrophilichydrophobic interface curves away or towards the water regions [1-3]. Recently, we showed that water nanofiltration (NF) membranes based on cross-linked type I bicontinuous cubic (QI) phases can be formed. The first system was based on a gemini phosphonium monomer (1) with water (0.75 nm pores) [6,7], while a second system was based on a more easily synthesized, gemini ammonium monomer with water (2) (0.86 nm pores) [8]. Both materials exhibit molecular-sieving capabilities; however, membrane fabrication was only possible via heating and pressing the QI monomer/water mixtures into a porous support, followed by radical photo-cross-linking to lock-in the phase [6,8]. This produces supported membranes with an active separation layer that is completely infused in the support material and thick as the support itself (ca. 40 µm [6-8] vs. the 0.1-µm-thick active layers of commercial reverse osmosis 27

44 (RO) [9] and NF membranes [10]), resulting in water fluxes too low to be practical. A thin separation layer is needed for high fluxes, since thicker layers have more resistance to flow. Attempts to make thinner QI films via solution-casting and related methods were unsuccessful due to the compositional sensitivity of the Q phases [6]. Evaporative water loss during processing and/or retention of residual casting solvent led to phase changes [6]. As with composite RO and NF membranes, the ideal configuration for this type of system (especially for commercial application) is a thin-film composite (TFC) architecture (i.e., a very thin separation layer on top of a more porous support). This configuration yields the highest water flux while providing the mechanical stability needed to withstand the pressures required for passage through the separation layer [9]. TFC membranes have recently been formed using cubic thermotropic (i.e., solvent-free) LC monomers [11]; however, only one example of a TFC LLC membrane has been reported (i.e., a HII system with non-aligned cylindrical nanopores and low flux) [12]. In order to get the best LLC membrane performance, what is needed is a Q monomer phase with 3D nanopores that can be easily fabricated into a TFC configuration with retention of the desired phase composition and structure. Herein, we present a new imidazolium-based gemini LLC monomer (3) that forms a cross-linkable QI-phase with the low volatility and environmentally benign solvent, glycerol, instead of water. This 3/glycerol QI monomer phase can be readily fabricated into nanoporous TFC membranes via solution-casting from MeOH to form defect-free thin films on porous supports, with minimal glycerol loss and retention of the QI composition and nanostructure after MeOH evaporation and photopolymerization (Figure 2.1). After pre-filtration to exchange the glycerol in the pores with water, these TFC membranes were found to reject uncharged molecular solutes consistent with size-exclusion through uniform 0.96 nm pores but with 28

45 absolute water fluxes ca. 10 times greater than prior melt-infused QI membranes. The rejection of smaller hydrated salt ions was found to be at the high level of RO membranes ( 98%) due to (charged pore charged solute) repulsive interactions. Figure 2.1. Monomer 3, its QI phase with glycerol, and the formation of cross-linked QI-phase TFC membranes Results and discussion To overcome the processing problems encountered in the earlier QI monomer systems, our approach was to substitute the water used for LLC phase formation with a lower volatility solvent. This alternative solvent should also be water-miscible (to enable facile flush-out) and non-toxic (in case traces remain in the membranes). Several polar organic solvents (formamide and its derivatives, N-methylsydnone, ethylene glycol (EG), glycerol, propylene carbonate) [13,14] and ionic liquids (ILs) [15,16] have been reported to form LLC phases (including Q phases) with select surfactants. Unfortunately, attempts to form LLC phases using 1 and 2 with 29

46 many of these solvents were unsuccessful. To obtain better compatibility/llc phase behavior with these solvents, a new gemini monomer (3) containing larger, more polarizable (i.e., softer ) imidazolium headgroups was designed and synthesized (see the Supp. Info. for synthesis details). Initially, a series of six gemini imidazolium monomers with different 1,3- diene tails and headgroup spacers was synthesized and screened for LLC behavior in different solvents using the solvent penetration scan technique with polarized light microscopy (PLM) [17]. Monomer 3 in this series was found to form Q phases with the broadest range of nonaqueous solvents (formamide, glycerol, and ethylammonium nitrate), in addition to pure water (see Supp. Info.). Since glycerol is a water-miscible, non-toxic solvent with low volatility (normal bp: 290 C, 20 C: <1 torr), the phase behavior of 3 with glycerol and water was studied in more detail. These studies showed that 3 forms a QI phase with pure glycerol, pure water, or a range of glycerol/water mixtures at 65 C (see Supp. Info. for phase diagram). The QI phase appears as a viscous, optically transparent gel located on the solvent-rich side of the L phase of the phase diagram, with a black PLM optical texture and powder X-ray diffraction (PXRD) d-spacings in the ratio 1/ 6 : 1 8, etc [6,8]. Bulk film photopolymerization studies using a water-free QI phase consisting of 79.7/19.8/0.5 (w/w/w) 3/glycerol/2-hydroxy-2-methylpropriophenone (HMP, a radical photo-initiator) showed that this system can be cross-linked in the QI phase, as confirmed by retention of the PLM and PXRD features (Figure 2.2). The black PLM textures were unchanged, and the characteristic 1/ 6 and 1 8 d-spacings showed very little change except for a slight increase in position after cross-linking, indicating a small unit cell expansion. FT-IR studies on the pre- and post-polymerized films showed >90% diene conversion (see Supp. Info.) [6]. 30

47 Figure 2.2. PXRD profiles of bulk QI-phase films of 3/glycerol/HMP (79.7/19.8/0.5 (w/w/w)): (a) before, and (b) after photo-cross-linking. Inset: PLM optical textures (50x). Given glycerol s properties, there should be minimal evaporative loss during thin-film solution processing if an appropriate volatile casting solvent is used. To prepare TFC membranes, a MeOH solution containing 60 wt % [79.7/19.8/0.5 (w/w/w)] 3/glycerol/HMP was roll-cast onto the surface of porous, asymmetric poly(ether sulfone) (PES) support films using a wire-wound rod. After mild heating to evaporate off the MeOH and photo-cross-linking at 70 C (see the Supp. Info. for details), scanning electron microscopy (SEM) indicated that the resulting membranes had defect-free coatings with an average thickness of 3 µm (Figure 2.3a). PXRD (Figure 2.3b) and FT-IR data (see the Supp. Info.) confirmed that the TFC membranes had top layers with a QI structure and a high degree of polymerization. 31

48 Figure 2.3. (a) Cross-sectional SEM photo and (b) PXRD profile of cross-linked QI TFC membrane prepared via MeOH solution roll-casting of 3/glycerol/HMP (79.7/19.8/0.5 (w/w/w)). After pre-filtering with pure water to remove the glycerol in the pores, the water filtration performance of these new TFC QI membranes was evaluated using dead-end filtration cells operating at 400 psi (27.6 bar) and aqueous solutions containing uncharged solutes and salts of varying molecular size (Table 2.1). Based on the rejection values of the uncharged organic solutes, the pore size of the TFC QI membranes was determined to be 0.96 nm using the Ferry pore size-exclusion model previously employed for QI membranes (see Supp. Info.) [6]. Even though the pore size of the TFC QI membranes of 3 is larger than that of the prior melt-infused QI membranes of 1 and 2, [6,8] its salt rejection performance is comparable or higher. This high salt rejection may be the result of the repulsive interactions between the new/different cationic headgroups in the pore walls and the approaching salt cations, or due to the new TFC configuration where the QI phase is a pure top layer instead of infused in the support. 32

49 Table 2.1. Water filtration performance of TFC QI membranes made from 3/glycerol/HMP (3 µm thick active layer). Test solute Diameter Rejection Flux f (nm) (%) e (L m 2 h 1 ) e sucrose 0.94 b 97 ± ± 0.01 glucose 0.73 b 87 ± ± 0.02 glycerol 0.36 c 45 ± ± 0.03 EG 0.32 c 24 ± ± 0.03 NaCl Na + (aq): 0.72 d 98 ± ± 0.05 MgCl2 Mg 2+ (aq): 0.86 d 99 ± ± 0.03 a 25-mm-I.D. stirred dead-end filtration cells; 400 psi; 2000 ppm aqueous feed solutions; 1 pass. b Ref. [18]. c Ref. [19]. d Ref. [20], diameter of Cl (aq) = 0.66 nm. e Average of 3 independent runs with std. dev. error ranges. f Pure water flux = 0.61 ± 0.05 L m 2 h 1 at the end of the filtration run. The TFC QI membranes also have a pure water flux ca. 10 times higher than the prior melt-infused QI membranes [6,8] as a result of the much thinner active layer. Interestingly, the salt filtration fluxes are ca times greater than the neutral solute filtrations (Table 2.1). We have found that aq. NaCl pre-treatment of the TFC QI membranes affords the same increased fluxes for neutral solute filtrations without altering rejection (see Supp. Info.). This beneficial effect is likely due to some sort of anion exchange or interaction between the salts and the ionic membrane pores/surface. We are currently investigating this phenomenon more fully. Compared to commercial RO (Dow SW30HR) and NF (Dow NF-270) membranes tested under the same conditions [7], the TFC QI membranes were found to have uncharged solute rejections higher than the NF membrane, but lower than the RO membrane (see the Supp. Info.). However, the salt rejection of the TFC QI membranes is much higher than that of NF-270 and on par with that of SW30HR. This dual-mode rejection behavior makes these new membranes quite unique in terms of their water purification performance. The thickness-normalized pure water permeability of the TFC QI membranes was found to be (6.6 ± 0.5) x 10 2 L m 2 h 1 bar 1 µm 33

50 based on an average 3-µm-thick active layer. This value is comparable to that measured for SW30HR under the same conditions (8.0 x 10 2 L m 2 h 1 bar 1 µm) [7], as well as to ranges reported for RO membranes [ca. (5.0 55) x 10 2 L m 2 h 1 bar 1 µm] [21] Conclusions In summary, a new glycerol-based LLC monomer system has been developed that enables facile fabrication of unprecedented TFC QI membranes that have molecular sieving capabilities, high salt rejection, and good water permeability. We are currently exploring methods to reduce the thickness of the 3/glycerol layers to 0.3 µm and increase flux by optimizing roll-casting and support parameters, as well as by using other solution processing techniques (e.g., dip-, spray-, and spin-coating). We are also exploring methods for varying the QI pore size, such as the use of co-surfactants, different anions, and mixtures of LLC solvents Supporting information Materials and general procedures Chromium (IV) oxide, pyridine, allyltrimethylsilane (98%), sec-butyl lithium (1.4 M in cyclohexane), aluminum oxide (activated, basic), 2-isopropoxy-4,4,5,5-tetramethyl-1,3,2- dioxaborolane (98%), hydrogen bromide (48% wt. % in H2O), borane-tetrahydrofuran complex solution (1.0 M in THF), p-toluenesulfonyl chloride, diethylene glycol, 1,4-dibromobutane (99%), 1,6-dibromohexane (96%), sodium hydride (60 wt. % dispersion in mineral oil), imidazole, ω-pentadecalactone (98%), 11-bromo-1-undecanol (98%), 2-hydroxy-2- methylpropiophenone (HMP, 97%), ethylamine (70 wt. % in H2O), nitric acid, formamide, and ethylene glycol (all ACS Reagents unless specified otherwise) were purchased from the Sigma- 34

51 Aldrich Chemical Co., and used as received. Glycerol (ACS Reagent) was purchased from Mallinckrodt, and used as received. Sulfuric acid (ACS Reagent) was purchased from VWR, and used as received. Sodium hydroxide, sodium chloride, and magnesium chloride (all ACS Reagents) were purchased from Fisher Scientific, and used as received. Filtration through silica gel was performed using mesh, normal-phase silica gel purchased from Sorbent Technologies. The water used in LLC phase formulation and water nanofiltration experiments was de-ionized, and had a resistivity of >12 MΩ cm 1. Poly(ether sulfone) (PES) polymer (Ultrason E 6020 P) was donated by BASF. Polyester fabric (Hollytex 3329) was generously donated by Ahlstrom Filtration, LLC Instrumentation 1 H NMR spectra were obtained using a Bruker 300 Ultrashield TM (300 MHz for 1 H) spectrometer. Chemical shifts are reported in ppm relative to deuterated solvent. Fouriertransform infrared spectroscopy (FT-IR) measurements of all of the new monomers synthesized were performed using a Matteson Satellite series spectrometer, as thin films on Ge crystals. All other FT-IR measurements were performed using a Thermo Scientific Nicolet 6700 spectrometer equipped with a PIKE MIRacle TM single-reflection horizontal ATR accessory with a diamond crystal. HRMS analysis was performed by the Central Analytical Facility in the Dept. of Chemistry and Biochemistry at the University of Colorado, Boulder. Powder X-ray diffraction (PXRD) spectra were collected using an Inel CPS 120 diffraction system with a monochromated Cu Kα radiation source. This apparatus was equipped with a film holder to analyze membrane samples. All PXRD spectra were calibrated using a silver behenate diffraction standard (d100 = 58.4 ± 0.1 Å). PXRD measurements were all performed at ambient temperature (22 ± 1 C), 35

52 unless noted otherwise. Polarized light microscopy (PLM) was performed using a Leica DMRXP polarizing light microscope equipped with a Linkam LTS 350 thermal stage, a Linkam CI 94 temperature controller, and a Q-Imaging MicroPublisher 3.3 RTV digital camera. Automatic temperature profiling and image capture were accomplished using Linkam Linksys32 software. Radical photopolymerizations were conducted using a Spectroline XX-15A 365 nm UV lamp (1 mw cm -2 at the sample surface). UV light fluxes at the sample surface were measured using a Spectroline DCR-100X digital radiometer equipped with a DIX-365 UV-A sensor. Water filtration studies were performed using custom-designed, stainless-steel, stirred, dead-end filtration cells for 2.5-cm-diameter membrane samples. The ion conductivity of permeate solutions was measured using a VWR International electrical conductivity meter model 2052-B. Total organic carbon (TOC) analysis of aqueous permeate solutions containing organic solutes was conducted using a Test N Tube TOC kit (Hach Company), a COD reactor (DRM 200, Hach Company), and an Agilent 8453 UV-visible spectrophotometer. The wire-wound, stainless-steel applicator rods used to solution roll-cast the TFC QI membranes were purchased from the Paul N. Gardner Company. 36

53 Figure 2.S1. Structures of prior gemini LLC monomers 1 and 2 used to form cross-linked QIphase, melt-infused membranes for water purification that operate via a molecular sieving mechanism [6,8] New LLC monomer synthesis and characterization Monomer 3 and five other homologues (4 8) were prepared as shown in Figure 2.S2 below. Detailed synthesis and structural characterization information on each of these monomers are also listed below. Figure 2.S2. General synthesis scheme for monomer 3 and several of its homologues that were tested for Q LLC phase formation in the presence of different solvents. 37

54 Bromopentadecanoic acid [22] ω-pentadecalactone (20.69 g, mmol, 100 mol %) was stirred in 48% HBr (98 ml, mmol, 1250 mol %) in a 250-mL round-bottom flask equipped with a stir bar and reflux condenser. H2SO4 (13 ml, mmol, 283 mol %) was added dropwise to minimize exothermic activity, liquefying the solid lactone. The yellow emulsion was heated to 115 C and stirred for 40 h. The resultant brown solution was extracted into CHCl3 (200 ml), washed with de-ionized H2O (3 x 100 ml) and brine (3 x 100 ml), dried over anhydrous MgSO4. The resulting organic solution was reduced under rotary vacuum to afford a light yellow solid, which was then recrystallized from hot CHCl3 and washed with cold hexanes to afford the product as a white, crystalline solid. Yield: 20.9 g (76%). Spectroscopic characterization and purity data for this compound matched published data [22] Bromopentadecanol [22] Borane-THF complex solution (83 ml, 83 mmol, 210 mol %) was measured out into a 500-mL Schlenk flask equipped with a stir bar. 15-Bromo-1-pentadecanoic acid (12.70 g, mmol, 100 mol%) was added slowly to minimize bubbling, and the light yellow solution turned clear over the course of 40 h of stirring. The reaction solution was then quenched with DI H2O (10 ml) dropwise, extracted into Et2O (50 ml), and washed with de-ionized H2O (3 x 50 ml) and brine (3 x 50 ml), dried over anhydrous MgSO4, and evaporated to afford the product as a white solid. Yield: 39.4 g (99%). Spectroscopic characterization and purity data for this compound matched published data [22]. 38

55 Bromopentadecanal [22] 15-Bromopentanol (6.69 g, mmol, 100 mol %) was dissolved in CH2Cl2 (200 ml) in a 500-mL round-bottom flask equipped with a stir bar. To the clear, slightly yellow solution was added PCC on alumina (39.76 g, mmol, 172 mol %) with vigorous stirring. The slurry was stirred at room temperature for 40 h. Reduction of the CH2Cl2 via rotary vacuum produced a dark brown solid that was stirred in diethyl ether and filtered through a pad of SiO2, washing with diethyl ether (700 ml). Concentration of the diethyl ether under rotary evaporation afforded the product as a white solid. Yield: 21.1 g (92%). Spectroscopic characterization and purity data for this compound matched published data [22] Bromotetradeca-1,3-diene [22] Synthesized as described in the literature [22]. Spectroscopic characterization and purity data for this compound matched published data [22] Bromooctadeca-1,3-diene [22] Synthesized as described in the literature [22]. Spectroscopic characterization and purity data for this compound matched published data [22] ,1 -(1,4-Butanediyl)bisimidazole [23] Synthesized as described in the literature [23]. Spectroscopic characterization and purity data for this compound matched published data [23]. 39

56 ,1 -(1,6-Hexanediyl)bisimidazole [23] Synthesized as described in the literature [23]. Spectroscopic characterization and purity data for this compound matched published data [23] ,1 -(Oxydi-2,1-ethanediyl)bisimidazole [23] Synthesized as described in the literature [23]. Spectroscopic characterization and purity data for this compound matched published data [23] ,6-Bis(octadeca-15,17-dienylimidazolium)hexane dibromide (3) 18-Bromooctadeca-1,3-diene (4.25 g, mmol, 204 mol %) and 1,1 -(1,6- hexanediyl)bisimidazole (1.38 g, 6.32 mmol, 100 mol %) were dissolved in acetonitrile (70 ml) in a 250-mL round-bottom flask equipped with a stir bar and reflux condenser. The clear, light yellow solution was stirred at 84 C for 100 h. Concentration of the reaction solvent via rotary evaporation produced an off-white solid, which was stirred in hexanes (3 x 200 ml) and filtered to afford the product as a white, crystalline solid. Yield: 5.2 g (93%). 1 H NMR (300 MHz, CDCl3): (s, 2H), 7.98 (s, 2H), 7.25 (s, 2H), 6.30 (dt, J = 10.1, 16.8, 2H), 6.03 (dd, J = 10.4, 15.1, 2H), 5.71 (dd, J = 7.3, 14.8, 2H), 5.07 (d, J = 16.7, 2H), 4.94 (d, J = 10.1, 2H), 4.46 (t, J = 7.3, 4H), 4.28 (t, J = 7.4, 4H), 2.00 (m, 12H), 1.32 (m, 58H). 13 C NMR (300 MHz, DMSO-d6): 24.78, 25.50, 28.35, 28.58, 28.60, 28.84, 28.94, 28.99, 29.03, 29.31, 31.87, 48.62, 48.81, , , , , , IR (thin film, MeOH): 3073, 3004, 2920, 2851, 1651, 1563, 1467, 1168, 1003, 949, 919 cm -1. HRMS (ES) calcd. for C48H84BrN4 (M + M + Br ): ; observed:

57 ,4-Bis(octadeca-15,17-dienylimidazolium)butane dibromide (4) 1,1 -(1,4-Butanediyl)bisimidazole (0.40 g, 2.10 mmol, 100 mol %) and 18- bromooctadeca-1,3-diene (1.47 g, 4.45 mmol, 212 mol %) were dissolved in acetonitrile (20 ml) and toluene (5 ml) in a 100 ml round bottom flask equipped with a stir bar and reflux condenser. The clear, light yellow solution was stirred at 86 C for 196 h. Upon cooling to room temperature, the mixture was concentrated via rotary vacuum to afford a crude, off-white solid, which was stirred in hexanes (4 x 100 ml) and filtered to afford the product as a white, crystalline solid. Yield: 1.5 g (82%). 1 H NMR (300 MHz, DMSO-d6): 9.25 (s, 2H), 7.81 (m, 4H), 6.28 (m, 2H), 6.03 (ddd, J = 0.6, 10.7, 11.7, 2H), 5.72 (dd, J = 7.3, 14.8, 2H), 5.08 (m, 2H), 4.95 (dd, J = 1.9, 10.1, 2H), 4.17 (m, 8H), 2.04 (q, J = 7.0, 4H), 1.77 (m, 8H), 1.25 (m, 44H). 13 C NMR (300 MHz, DMSO-d6): 25.54, 26.06, 28.39, 28.58, 28.61, 28.85, 28.97, 28.99, 29.04, 29.33, 31.88, 48.06, 48.88, , , , , , , IR (thin film, MeOH): 3070, 3005, 2920, 2851, 1656, 1568, 1468, 1338, 1165, 1004, cm -1. HRMS (ES) calcd. for C46H80BrN4 (M + M + Br): ; observed: ,1 -(Oxydi-2,1-ethanediyl)bis[3-(octadeca-15,17-dienyl)imidazolium] dibromide (5) 18-Bromooctadeca-1,3-diene (1.05 g, 3.18 mmol, 214 mol %) and 1,1-(oxydi-2,1- ethanediyl)bisimidazole (0.31 g, 1.49 mmol, 100 mol %) were dissolved in acetonitrile (20 ml) and toluene (10 ml) in a 50-mL round-bottom flask equipped with a stir bar and reflux condenser. The clear, light yellow solution was stirred at 86 C for 100 h. Upon cooling to room temperature the reaction was concentrated under rotary vacuum to produce an off-white solid, which was stirred in hexanes (3 x 100 ml) and filtered to afford the product as a white, 41

58 crystalline solid. Yield: 1.0 g (78%). 1 H NMR (300 MHz, DMSO-d6): 9.24 (s, 2H), 7.79 (m, 2H), 7.71 (m, 2H), 6.29 (dt, J = 10.2, 16.8, 2H), 6.03 (dd, J = 10.5, 15.3, 2H), 5.72 (dd, J = 7.4, 14.7, 2H), 5.08 (d, J = 16.9, 2H), 4.95 (d, J = 10.1, 2H), 4.36 (d, J = 4.1, 4H), 4.18 (d, J = 7.6, 4H), 3.78 (s, 4H), 2.03 (t, J = 7.0, 4H), 1.77 (m, 4H), 1.23 (m, 44H). 13 C NMR (300 MHz, DMSO-d6): 25.56, 28.43, 28.59, 28.61, 28.86, 28.89, 29.01, 29.05, 29.45, 31.89, 48.66, 48.81, 68.01, , , , , , , IR (thin film, MeOH): 3078, 3010, 2922, 2850, 1658, 1565, 1468, 1163, 1124, 1003, 950 cm -1. HRMS (ES) calcd. for C46H80BrN4 (M + M + Br ): ; observed: ,6-Bis(tetradeca-11,13-dienylimidazolium)hexane dibromide (6) 1,1-(1,6-Hexanediyl)bisimidazole (1.07 g, 4.89 mmol, 100 mol %) was dissolved in acetonitrile (15 ml) in a 100-mL round-bottom flask equipped with a stir bar and reflux condenser. Addition of 14-bromotetradeda-1,3-diene (2.70 g, 9.90 mmol, 203 mol %) produced a clear, light yellow solution that was stirred at 86 C for 96 h. Upon cooling to room temperature, solvent was concentrated under rotary vacuum to afford an off-white, waxy solid that was stirred in hexanes (3 x 50 ml) and filtered to afford the product as a white, crystalline solid. Yield: 3.1 g (82%). 1 H NMR (300 MHz, DMSO-d6): 9.25 (s, 2H), 7.81 (ds, J = 1.4, 4H), 6.29 (dt, J = 10.2, 16.9, 2H), 6.03 (dd, J = 10.4, 15.2, 2H), 5.71 (m, 2H), 5.09 (dd, J = 1.7, 17.0, 2H), 4.95 (dd, 2H), 4.16 (t, J = 7.1, 8H), 2.04 (q, J = 6.7, 4H), 1.78 (m, 8H), 1.24 (m, 32H). 13 C NMR (300 MHz, DMSO-d6): 25.29, 25.96, 28.80, 29.05, 29.08, 29.28, 29.34, 29.53, 29.79, 32.35, 49.12, 49.30, , , , , , IR (thin film, MeOH): 3081, 3008, 2928, 2858, 1647, 1567, 1463, 1164, 1003, 953, 896 cm -1. HRMS (ES) calcd. for C40H68BrN4 (M + M + Br ): ; observed:

59 ,4-Bis(tetradeca-11,13-dienylimidazolium)butane dibromide (7) 14-Bromo-tetradeac-1,3-diene (0.912 g, 3.36 mmol, 213 mol %) and 1,1 -(1,4- butanediyl)bisimidazole (0.30 g, 1.58 mmol, 100 mol %) were dissolved in acetonitrile (25 ml) and toluene (10 ml) in a 100-mL round-bottom flask equipped with a stir bar and reflux condenser. The clear, light yellow solution was stirred at 86 C for 70 h. Cooling to room temperature and concentration of the solvent under rotary evaporation afforded a light brown solid which was stirred in diethyl ether (4x50 ml) and filtered to afford the product as a white, crystalline solid. Yield: 0.7 g (57%). 1 H NMR (300 MHz, CDCl3): (s, 2H), 8.09 (s, 2H), 7.19 (s, 2H), 6.31 (dt, J = 10.2, 17.0, 2H), 6.04 (dd, J = 10.4, 15.2, 2H), 5.70 (m, 2H), 5.08 (dd, J = 1.2, 16.9, 2H), 4.95 (d, J = 10.1, 2H), 4.61 (t, 4H), 4.25 (m, 4H), 2.23 (t, 4H), 2.07 (q, J = 6.9, 4H), (m, 4H), 1.30 (d, J = 20.8, 28H). 13 C NMR (300 MHz, CDCl3): 26.43, 26.63, 29.08, 29.29, 29.46, 29.53, 30.33, 32.67, 48.98, 50.38, , , , , , , IR (thin film, MeOH): 3070, 3012, 2924, 2851, 1652, 1564, 1464, 1165, 1004, 897 cm -1. HRMS (ES) calcd. for C38H64BrN4 (M + M + Br ): ; observed: ,1 -(Oxydi-2,1-ethanediyl)bis[3-(tetraadeca-11,13-dienyl)imidazolium] dibromide (8) [24] Synthesized as described in the patent literature [24]. Spectroscopic characterization and purity data for this compound matched published data [24]. 43

60 Ethylammonium nitrate Synthesized as described in the literature [25]. Spectroscopic characterization and purity data for this compound matched published data [25] Qualitative screening of LLC phase behavior with different solvents using the PLMbased penetration scan technique [17] To quickly and qualitatively determine the LLC phase behavior of homologues of monomer 3 (see Table 2.S1) with a specific solvent, the PLM penetration scan technique was employed [17]. This technique is a solvent-amphiphile gradient assay using PLM that quickly (i.e., in minutes) determines qualitatively what phases can be formed by a certain amphiphile and solvent pair at a specific temperature. This technique was performed by taking the solid monomer and pressing it in-between a microscope slide a cover slip. The sample was then placed on the PLM thermal stage and annealed to its melting temperature or up to 90 C, whichever came first. The sample was then slowly cooled back down to room temperature. A small amount (<30 µl) of the chosen solvent was added to the edge of the cover slip and the solvent was drawn via capillary action into contact with the solid monomer, creating a concentration gradient. The specimen was then heated to 95 C at a rate of 5 C/min with digital image capture. The differences in optical texture and intensity were used to determine the potential LLC phases formed. Since QI phases are black (i.e., isotropic) under PLM and are typically found between birefringent lamellar (L) and hexagonal (HI) phases, a dark isotropic band between two birefringent LLC phases indicates a potential QI phase. The six gemini imidazolium bromide monomer homologues were tested with water, glycerol, formamide, ethylene glycol, and ethylammonium nitrate. Out of the six homologues analyzed, only 44

61 monomer 3 had clear potential QI phases with multiple non-aqueous solvents tested. From qualitative analysis of the penetration scan PLM optical textures, monomer 3 showed evidence of QI phase formation with the non-aqueous solvents glycerol, formamide, and ethylammonium nitrate (e.g., see Figure 2.S3). Based on this initial study, only monomer 3 was chosen for more detailed LLC phase behavior characterization with glycerol as the LLC solvent. Table 2.S1. The gemini imidazolium bromide monomer homologues synthesized, and a summary of their qualitative Q phase formation behavior between 22 C and 95 C with different solvents from PLM penetration scan screening studies. 45

62 Figure 2.S3. Representative PLM images (magnification: 50x) of solvent penetration scans of 3 with (a) water at 68 C, (b) glycerol at 65 C, (c) formamide at 55 C, and (d) ethylammonium nitrate at 44 C. The black (pseudo-isotropic) region between two bright, anisotropic LLC regions is indicative of the presence of a potential Q phase. The arrows in the PLM images point in the direction of increasing solvent concentration Preparation of LLC samples, determination of LLC phase behavior, and elucidation of LLC phase diagrams LLC samples of specific composition were made by adding an appropriate amount of monomer and solvent to custom-made glass vials. A photo-initiator, 2-hydroxy-2- methylpropiophenone (HMP), was added if required, and the vials were sealed with Parafilm. LLC samples were mixed by alternately hand-mixing and centrifuging (2800 rpm) until completely homogenous. It should be noted that the LLC samples are sensitive to water loss or gain, depending on the solvent system. Special attention was taken to keep the samples sealed as much as possible during sample mixing and transferring to minimize composition drift. For samples with low-viscosity solvent systems (e.g., water), LLC samples of specific composition were prepared by adding the desired amount of monomer to custom-made glass 46

63 vials, followed by the addition of an appropriate amount of solvent via pipette. Photo-initiator was then added if required. The vials were sealed with Parafilm and centrifuged at 2800 rpm. Samples were then alternately hand-mixed and centrifuged until homogeneous. For samples with high viscosity solvent systems (e.g., glycerol), LLC samples of specific composition were prepared by adding the desired amount of solvent to custom-made glass vials, followed by the addition of an appropriate amount of monomer via spatula. Photo-initiator was then added if required. The vials were sealed with Parafilm and centrifuged at 2800 rpm. Samples were then alternately hand-mixed and centrifuged until homogeneous. The composition and temperature range of the LLC phase was determined using variabletemperature PLM. Specimens at various concentrations were prepared and then pressed between a microscope slide and microscope cover-slip. The assembly was then placed on the PLM thermal stage and annealed past its isotropic temperature or up to 90 C (whichever came first). The sample was slowly cooled and allowed to come back to its room temperature phase. The sample was then heated to 95 C at a rate of 5 C/min with digital image capture and continuous recording of the light intensity. Images were captured at 50x magnification. Changes in optical texture and light intensity were used to determine changes in the LLC phase of the mixture. The identity of each observed phase was then confirmed with PXRD by analyzing a point in each distinct phase region as elucidated by PLM analysis. PXRD spectra of the samples were taken either by using a film holder apparatus for room temperature spectra or with a heated stage for higher temperature spectra. In the film holder, a sample was placed between Mylar sheets with an appropriate spacer, annealed, placed in the film holder, and then examined. On the heated stage, a sample was placed in an aluminum PXRD pan and a piece of Mylar was used to cover the sample. The d spacing pattern of the PXRD peaks is used to determine the LLC phase. 47

64 Using the combined PLM and PXRD data, a partial phase diagram of the 3/glycerol/water system was created as a function of system composition and temperature. The minimum temperature for QI phase formation in this ternary system was found to be 65 C. Above 65 C, the QI phase typically remained up to 95 C. A partial phase diagram of 3/glycerol/water at 65 C is shown in Figure 2.S4. Figure 2.S4. Elucidated partial phase diagram highlighting the position of the QI phase for the 3/water/glycerol system at 65 C. (Other non-cubic LLC phases were also observed outside of the QI-phase region, but these have not been rigorously identified yet.) 48

65 Preparation of bulk Q I -phase films of 3/glycerol/HMP and cross-linking with LLC phase retention Bulk LLC samples of the desired composition with HMP photo-initiator were prepared as described in the "Preparation of LLC samples" section. The QI phase monomer mixtures were then sandwiched between two pre-cut pieces of Mylar. The samples were clamped in-between two quartz plates and heated to 70 C to form the QI phase. After 5 min, the samples were irradiated with 365 nm light (1 mw cm -2 ) for 1 h. The quartz plates and Mylar sheets were carefully removed to afford thick, free-standing QI polymer films. PLM and PXRD analysis of the polymerized samples show that the QI-phase nanostructure is retained (see Figure 2.2 in main manuscript) Comparison of the relative degree of order in the Q I phases of 1/water, 2/water, and 3/glycerol by PXRD analysis Comparison of the PXRD profiles of the bulk cross-linked QI phases of 1/water [26], 2/water [8], and 3/glycerol systems set to the same x-axis (i.e., 2θ (degrees)) range shows that the three QI phase systems have a similar degree of overall order. As can be seen in Figure 2.S5, all three have 2 principal PXRD d-spacing peaks: a very pronounced first diffraction peak that indexes to 1/ 6, and a secondary, less intense peak/shoulder that indexes to 1/ 8. A weak, very broad peak can also be seen in some of the spectra that corresponds to either the 1/ 20 [26] or 1 22 [8] d-spacing peaks of a QI phase [27,28]. The possible 1/ 20 and 1 22 peaks do not correspond to hexagonal or lamellar phase PXRD peaks [17]. The 1/ 8 peak for the QI phase of 1/water is a bit better resolved from the main peak than the other two systems, but this may be due to the presence of additional heavier atoms (i.e., phosphorus) and thus more electron 49

66 density contrast at the water/organic interface in the 1/water system compared to the other two. With their PXRD profiles scaled to the same 2θ x-axis range, the three different cross-linked QI systems have similar widths for their principal (i.e., 1/ 6) diffraction peak. The 1/ 6 diffraction peak full width at half max for all three systems is approximately 0.5 degrees. The only major difference between the cross-linked QI phases of the three systems is the position of the d- spacings, indicating different unit cell sizes. Figure 2.S5. PXRD profiles of the bulk cross-linked QI phases of the 1/water [26], 2/water [8], and 3/glycerol systems shown to the same 2θ (x-axis) scale. The PXRD profile of the bulk cross-linked QI phase of 1/water is taken from Ref. [26], Figure 9c. The data for the PXRD profile for the bulk cross-linked QI phase of 2/water is from Ref. [8]. 50

67 Determination of degree of conversion for the radical photopolymerization of the 3/glycerol/HMP bicontinuous cubic (Q I ) phases Bulk samples of 3/glycerol/HMP with the desired composition were prepared that form a QI phase when heated above 65 C as described in previous sections. The QI phase monomer mixture was then placed on the ATR plate of the FT-TR spectrometer, and FT-IR spectra were collected. The QI phase monomer mixture was then sandwiched between two pre-cut pieces of Mylar. The sample was clamped in-between two quartz plates and heated to 70 C to form the QI phase. After 5 min, the samples were irradiated with 365 nm light (1 mw cm -2 ) for 1 h. The quartz plates and Mylar sheets were carefully removed and the QI polymer film was placed on the ATR plate and the post-polymerization FT-IR spectra was collected. The 1004 cm -1 absorption band in the monomer samples comes from the C H out-of-plane wagging from the 1,3-diene units located on the end of the tails of the monomer [6,26]. As seen in Figure 2.S6, the attenuation of the 1004 cm -1 peak in the post-polymerized sample suggests >90% conversion of the 1,3-diene units for the QI-phase LLC sample based on a static internal reference peak at 1160 cm -1. PLM and PXRD analysis of the polymerized samples show that the QI-phase nanostructure is retained. 51

68 Figure 2.S6. FT-IR spectra of 3/glycerol/HMP (79.7/19.8/0.5 (w/w/w)) QI-phase mixture: (a) before, and (b) after heating to 70 C and polymerizing with 365 nm UV light (1 mw cm -2 ) for 1 h. Attenuation of the 1004 cm -1 peak relative to the 1160 cm -1 internal reference peak suggests >90% 1,3-diene group conversion Solution-casting and photo-cross-linking of a thin-film Q I -phase of 3/glycerol/HMP on dense glass substrates to provide proof-of-concept for successful thin-film formation and cross-linking with LLC phase retention A trial casting solution was prepared by adding appropriate amounts of monomer 3, glycerol, and HMP to a 10-mL glass vial and dissolving the components in the desired amount of methanol. A small amount of the casting solution (<50 µl) was then pipette onto a glass slide and allowed to air dry for 10 min. After placing the glass slide on a temperature-controlled hot stage, the stage was warmed to 25 C, and the temperature was gradually increased to 75 C by heating at a rate of 5 C/min and holding the temperature for 2 min every 10 C up to 75 C. This heating step removed excess methanol casting solvent from the mixture. A small portion of the sample was then scraped off the glass slide, placed on the ATR plate of the FT-IR spectrometer, and a pre-polymerization FT-IR spectrum was collected. The sample was then cooled to room temperature and placed in a specially designed photo-polymerization chamber 52

69 with an aluminum base and a Pyrex glass plate cover. The atmospheric O2 in the chamber was removed by alternating vacuum (2000 mtorr) and Ar purge cycles three times. The Ar-filled polymerization chamber was then warmed to 70 C by a hot stage with a temperature controller, and the chamber was irradiated with a 365 nm UV lamp (1 mw cm -2 ) for 1 h. The glass slide was then removed from the chamber, and the sample was scraped from the slide with a razor blade. The resulting thin, free-standing film (ca. 10 µm thick) was placed on the ATR plate of the FT-IR spectrometer, and a post-polymerization FT-IR spectrum was collected. As seen in Figure 2.S7, the absorbance attenuation of the peak at 1004 cm -1 suggests >90% conversion for the solution-cast thin film QI phase. As shown in Figure 2.S8, PLM and PXRD analysis of the polymerized thin-film sample show retention of the desired QI-phase structure. Figure 2.S7. FT-IR spectra of 3/glycerol/HMP (79.6/19.7/0.7 (w/w/w)) before (dried solutioncast monomer mixture) and after (thin-film cross-linked QI polymer) heating to 70 C and polymerizing with 365 nm UV light (1 mw cm -2 ) for 1 h. Attenuation of the 1004 cm -1 peak suggests >90% 1,3-diene group conversion. 53

70 Figure 2.S8. PLM optical texture and PXRD spectrum of a thin film of 3/glycerol/HMP (79.6/19.7/0.7 (w/w/w)) solution-cast from methanol onto glass, heated to 70 C to remove the methanol, and photo-cross-linked with 365 nm UV light (1 mw cm -2 ) for 1 h under Ar atmosphere Preparation of asymmetric, porous poly(ether sulfone) (PES) support membranes via phase-inversion processing The PES support membranes were prepared following guidelines described in the literature [29]. A 15 wt % solution of PES in dimethylformamide (DMF) was prepared by adding PES flakes to a small round-bottom flask containing a stir bar and the appropriate amount of DMF. The solution was stirred for 24 h under ambient conditions and then allowed to sit with no stirring for another 24 h to remove any bubbles from the solution. A piece of Hollytex 3329 fabric was cut and taped to a glass plate. A custom-made stainless-steel blade caster with a 250- µm gap was placed on the fabric swatch, and the PES solution was poured into the caster. The caster was then quickly drawn down the plate, and the plate was immediately transferred to a deionized water bath. The PES polymer immediately precipitated. After 30 min, the plate was transferred to a fresh water bath and allowed to soak overnight. The support membranes were then allowed to air dry for 24 h before use. As seen in Figure 2.S9, cross-section and top-view 54

71 scanning electron microscope (SEM) images of the support membranes demonstrate the desired asymmetric porous structure and a defect-free top surface. Figure 2.S9. Cross-section and angled top-view SEM images of the prepared PES support membranes Fabrication of PES-supported thin-film composite (TFC) Q I polymer membranes of 3/glycerol/HMP (see Figure 2.S10) A QI-phase casting solution was prepared by adding appropriate amounts of monomer 3, glycerol, and HMP to a 10-mL glass vial and dissolving the components in the desired amount of methanol. A PES support membrane was taped to a small glass plate along two edges and a small piece of aluminum foil was placed on top of one of the taped edges. A #3 wire-wound rod from the Paul N. Gardner Company was placed in the aluminum trough. The casting solution was pipette along the length of the trough and the rod was then immediately drawn at a slow constant speed across the support membrane. The membrane sample was allowed to air dry for 10 min and then placed on a temperature-controlled hot stage. The stage was warmed to 25 C, and then the temperature was gradually increased to 75 C by heating at a rate of 5 C/min and 55

72 holding the temperature for 2 min every 10 C up to 75 C. This gradual heating process is performed to removes excess methanol casting solvent. The membrane sample was then cooled to room temperature. A portion of the sample was cut off and placed on the ATR plate of the FT- IR spectrometer and a pre-polymerization FT-IR spectrum of the solution-cast QI-phase monomer mixture on the PES support was collected. The membrane sample was then placed in a specially designed photo-polymerization chamber with an aluminum base and a Pyrex glass plate cover. The atmospheric O2 in the chamber was removed by alternating vacuum (2000 mtorr) and Ar purge cycles three times. The Ar-filled polymerization chamber was then warmed to 70 C by a temperature controller on a hot stage, and the chamber was irradiated with a 365 nm UV lamp (1 mw cm -2 ) for 1 h. A portion of the membrane sample was cut off and placed on the ATR plate of the FT-IR spectrometer and a post-polymerization FT-IR spectrum of the final TFC QI membrane was collected. As seen in Figure 2.S11, the absorbance attenuation of the FT- IR band at 1004 cm -1 suggests a high degree of conversion for the TFC QI membrane. 56

73 Figure 2.S10. Scheme for solution roll-casting and then cross-linking a thin QI-phase 3/glycerol/HMP top layer on porous PES supports to prepare nanoporous TFC QI polymer membranes. Figure 2.S11. FT-IR spectra of solution-cast 3/glycerol/HMP (79.7/19.8/0.5 (w/w/w)) top layer (ca. 3 µm thick) on porous PES support before (blue trace) and after (red trace) heating to 70 C and polymerizing with 365 nm UV light (1 mw cm -2 ) for 1 h. Absorbance attenuation of the 1004 cm -1 peak suggests a high degree of 1,3-diene group conversion. 57

74 As seen in Figure 2.S12, a cross-section SEM image shows an average ca. 3-µm-thick coating on top of the porous PES support film. Retention of the desired QI-phase nanostructure of the coating was verified by PXRD analysis (see Figure 2.3 in the main manuscript). A picture of the resulting TFC QI polymer membrane made from 3/glycerol/HMP is shown in Figure 2.S13. Figure 2.S12. Cross-sectional SEM image of a typical cross-linked TFC QI membrane fabricated by roll-casting a methanol solution of 3/glycerol/HMP onto porous PES support. Figure 2.S13. Picture of the final TFC QI polymer membrane prepared from 3/glycerol/HMP solution roll-cast on porous PES. 58

75 General water nanofiltration testing procedure Water nanofiltration experiments were performed using the same custom-made, stainless steel, stirred dead-end filtration cells used in previous studies [6,8]. The membrane holder of each cell has a 2.5 cm outer diameter, but only an effective filtration area of 3.8 cm 2 with the O- ring configuration. The nanofiltration experiments were performed using aqueous feed solutions containing a single solute at a concentration of 2000 ppm. Each dead-end filtration cell was loaded with 25 ml of the feed solution and pressurized until 4 5 ml of permeate was collected to wash out any of the previous permeate. The cell was then reloaded with the same feed solution and the next 3-4 ml of permeate was collected and analyzed to determine the flux and rejection. For all of the filtration studies, the flux (J) was calculated as follows using Equation 2.1: J = V A t (2.1) where ΔV is the permeate volume, A is the surface area of the membrane (3.8 cm 2 ), and Δt is the time needed to collect the permeate. The % rejection (R) was calculated using Equation 2.2: R = 1 C permeate C feed 100 (2.2) where Cpermeate and Cfeed are the concentration of the solute in the permeate and feed, respectively. All of the fluxes and rejections are averages of three different membrane samples in separate experiments. The reported error is the standard deviation from the three different membranes in separate experiments. 59

76 Permeate analysis The concentrations of all the neutral organic solutions were determined using a TOC digestion kit with a modified procedure based on Hach method and subsequent UVvisible analysis. A calibration curve was made with standard solutions prior to each analysis. The concentrations of NaCl and MgCl2 in the permeate solution were determined using an electrical conductivity meter. A calibration curve was made for each salt using aqueous standard solutions of each salt Water nanofiltration testing of uncoated PES support membranes as controls Membrane discs (2.5 cm in diameter) of the prepared porous PES support membrane were punched out from sheets using a sharpened circular die. The membrane discs were soaked in de-ionized water for 30 min at ambient temperature and then carefully loaded into the custommade, stainless-steel, stirred dead-end filtration cells. The cells were filled with 25 ml of deionized water and pressurized to 50 psi (3.4 x 10 5 Pa) with N2 pressure as the driving force at ambient temperature (22 ± 1 C). The first 5 ml of permeate from each test cell was collected and discarded. This first filtration was to ensure the integrity of the membranes and clean out any contaminants that might remain in the membranes after processing. The general water filtration procedure was then followed. Analysis of all of the permeates showed the porous PES support membranes had <10% rejection for all of the solutes tested (sucrose, glucose, glycerol, ethylene glycol, NaCl, and MgCl2). This is to be expected, since the PES support membrane should have comparable performance to a conventional ultrafiltration (UF) membrane. The pure water flux of the uncoated PES support membrane was measured to be (1.4 ± 0.1) x 10 2 L m -2 h

77 Water nanofiltration testing of TFC Q I polymer membranes made from 3/glycerol/HMP Membrane discs (2.5 cm in diameter) of the TFC QI polymer membranes of 3/glycerol/HMP were punched out from small sheets using a sharpened circular die. Membrane discs were then soaked in de-ionized water for 1 hour at ambient temperature. After soaking, the membrane discs were carefully loaded into the custom-made, stainless-steel, stirred, dead-end filtration cells. All nanofiltration experiments were performed with 400 psi (2.76 x 10 6 Pa) of N2 pressure as the driving force at ambient temperature (22 ± 1 C). The cells were then filled with de-ionized water that was filtered through the membranes until at least 5 ml of permeate was collected in each cell. This first filtration was to ensure the integrity of the membranes and clean out any unreacted monomer, glycerol, or other contaminants that might remain in the membranes after processing. The cells were then reloaded with de-ionized water and filtered until another 5 ml of permeate was collected and analyzed for any residual organic carbon. All permeate samples had <15 ppm of total organic carbon (detection limit of analysis), confirming virtually all of the glycerol had been removed from the TFC QI membrane pores. The general water nanofiltration procedure was then followed, with the following order of solutes tested: sucrose, glucose, glycerol, ethylene glycol, NaCl, and MgCl2. A final filtration was performed with pure water to determine the pure water flux at the end of the experiments Estimation of the effective pore size by fitting the neutral solute rejection data with the Ferry equation The original Ferry equation describes particle rejection through uniform-size pores as a function of effective solute particle size (rsolute) and effect pore size (rpore) [30]. This simple steric 61

78 pore model assumes that the solutes are spherical and the membrane pores are uniform cylinders [30]. The Ferry equation has been used to describe a variety of porous membranes. The Ferry equation is shown as follows in Equation 2.3: R = 1 1 r 2 2 solute 100 (2.3) r pore where R is the rejection in %, rsolute is the solute diameter, and rpore is the pore diameter. The Ferry equation has also been recently found to be also valid for earlier versions of QI-phase LLC polymer membranes with uniform slit pores with a uniform slit width that is equivalent to rpore. (Figure 2.S14) [6]. Figure 2.S14. Model for applying the Ferry equation for rejection performance of membranes with uniform circular pores to a QI-phase system with a uniform water layer manifold to determine layer gap spacing [6]. 62

79 The observed rejection data for the non-charged solutes of the TFC QI polymer membranes made from 3/glycerol/HMP were fitted to the Ferry equation to estimate the effective pore size in the absence of pore wall and solute particle charge-charge effects. The neutral solute diameters used in this study are the same as the values used in two previous studies [6,8]. From fitting the uncharged solute rejection data to the Ferry model, it was found that the effective pore size of the TFC QI polymer membranes of 3/glycerol/HMP is 0.96 nm. The Ferry Equation plot set to a uniform pore size of 0.96 nm is in excellent agreement with the experimental rejection data for the tested non-charged solutes (Figure 2.S15). Figure 2.S15. Experimentally measured rejection data for the TFC QI membranes made from 3/glycerol/HMP for several non-charged solute molecules, and the Ferry equation plot with a uniform pore size (rpore) of 0.96 nm overlaid on these experimental data Preliminary studies showing the beneficial effect of aq. NaCl pre-treatment on the water flux of the TFC Q I membranes for neutral solute filtrations without compromising rejection Membrane discs (2.5 cm in diameter) of the TFC QI polymer membranes made from 3/glycerol/HMP were punched out from small sheets using a sharpened circular die. Membrane 63

80 discs were then soaked in a 2000 ppm NaCl solution for 1 h at ambient temperature. After soaking, the membrane discs were carefully loaded into the custom-made, stainless-steel, stirred dead-end filtration cells. All nanofiltration experiments were performed with 400 psi (2.76 x 10 6 Pa) of N2 pressure as the driving force at ambient temperature (22 ± 1 C). The cells were then filled with de-ionized water that was then filtered through the membranes until at least 5 ml of permeate was collected in each cell. The cells were then reloaded with de-ionized water and filtered until another 5 ml of permeate was collected and analyzed for any residual organic carbon. All permeate samples had <15 ppm of total organic carbon (detection limit of analysis), confirming that virtually all of the glycerol had been removed from the TFC QI membrane pores. The general water nanofiltration procedure was then followed. A final filtration was performed with de-ionized water to determine the pure water flux at the end of the experiment. As seen in Table 2.S2, the rejection performance of the aq. NaCl pre-treated TFC QI polymer membranes is the same (within error) as that of the untreated TFC QI polymer membranes that were just soaked in de-ionized water before filtration (see Table 2.1 in the main manuscript). However, the fluxes of the aq. NaCl pre-treated TFC QI polymer membranes are virtually identical (within error) for all of the test solutes, whereas the untreated TFC QI membranes only showed a significant increase in the flux after contact with salt feed solutions (see Table 2.1 in the main manuscript). 64

81 Table 2.S2. Rejection and water flux data for individually tested TFC QI polymer membranes pre-soaked in a 2000 ppm NaCl solution before conducting neutral solute filtration experiments. Test solute Hydrated diameter (nm) a Rejection (%) Flux (L m 2 h 1 ) sucrose ± ± 0.2 glucose ± ± 0.2 glycerol ± ± 0.2 EG ± ± 0.2 NaCl Na + (aq): ± ± 0.2 MgCl2 Mg 2+ (aq): ± ± 0.2 a Hydrated diameters of solute molecules and ions were obtained from the cited papers in Ref. [6]. PXRD analysis of the aq. NaCl pre-treated TFC QI membranes did not show any noticeable differences in the LLC nanostructure. We are currently investigating the nature of this beneficial effect on water flux (without compromising rejection performance) in more detail. More information on this phenomenon will be published in a more detailed follow-up publication Comparison of the performance of TFC Q I membranes with commercial Dow SW30HR (RO) and DOW NF-270 (NF) membranes The water nanofiltration testing of SW30HR and NF-270 membranes under the same test conditions as the TFC QI membranes using the same custom-made, stainless-steel, stirred deadend filtration cells is published in the literature [7]. Membrane discs (2.5 cm in diameter) were punched out from sheets using a sharpened circular die. The membrane discs were carefully loaded into separate custom-made, stainless-steel, stirred, dead-end filtration cells. All nanofiltration experiments were performed with 400 psi (2.76 x 10 6 Pa) of N2 pressure as the driving force at ambient temperature (22 ± 1 C). The cells were then filled with de-ionized 65

82 water that was filtered through the membranes until at least 5 ml of permeate was collected in each cell. This first filtration was to ensure the integrity of the membrane and clean out any contaminants that might remain in the membrane after processing. Filtration experiments were then carried out using aqueous feed solution containing a single solute at 2000 ppm. Each stirred dead-end filtration cell was loaded with 25 ml of the feed solution and then pressurized. The first 1 2 ml of permeate was discarded. The next 2 4 ml of permeate was then collected and examined to determine rejection and flux. The neutral organic solute rejection, salt rejection, and water flux measurements were made as described in the previous sections. As seen in Figure 2.S16, when compared to a high-performance RO membrane (Dow SW30HR) and a porous NF membrane with non-uniform pores (Dow NF-270) tested under the same conditions, the 3/glycerol/HMP TFC QI membranes were found to have neutral solute rejections higher than the NF membrane but lower than the RO membrane for all of the solutes tested. However, the salt rejection performance of the TFC QI membrane is much higher than that of the NF membrane and comparable to that of the RO membrane. These rejection properties make these new TFC QI membranes unique in terms of their water purification performance: They perform neutral solute separations better than a NF membrane (due to the uniform pore size distribution), but reject certain salts at a high level similar to that of a RO membrane. The TFC QI membranes operate in a performance regime between that of conventional NF and RO membranes. 66

83 Figure 2.S16. Comparison of rejections of TFC QI, SW30HR, and NF-270 membranes (stirred dead-end filtration; 400 psi; 2000 ppm aq. feed solutions) The pure water flux and thickness-normalized pure water permeability values of SW30HR, NF-270 and the TFC QI membranes measured under the same dead-end filtration conditions (25-mm-ID test cells; 400 psi applied pressure) are summarized in Table 2.S3 below. Table 2.S3. Comparison of pure water flux and thickness-normalized pure water permeability values for NF-270, SW30HR, and the TFC QI membranes of 3/glycerol/HMP measured under the same dead-end filtration test conditions. Membrane Active Layer Thickness (µm) Pure Water Flux (L m 2 h 1 ) Thicknessnormalized Pure Water Permeability (L m 2 h 1 bar 1 µm) NF a 250 a 0.91 a SW30HR 0.1 a 22 a a TFC QI membrane of 3/glycerol/HMP ± ± (a) Data from Ref. [7]. 67

84 As can be seen in Table 2.S3, the observed pure water flux of the TFC QI membranes is much lower than the commercial membranes. The low water flux can be attributed to the thicker active layer of the TFC QI membranes compared to SW30HR and NF-270 (i.e., 3 µm vs. ca. 0.1 µm). However, when normalized for active layer thickness, the nanoporous QI network separation material in the TFC QI membranes has a thickness-normalized pure water permeability that is comparable to that of the active material in SW30HR. This suggests if the QI separation material can be processed as thin as the commercial membranes, it may be possible to obtain fluxes comparable to that of an RO membrane Acknowledgments Financial support from the National Science Foundation (CBET ) is gratefully acknowledged. E.S.H. thanks the National Water Research Institute and American Membrane Technology Association for a fellowship. J.L.B. thanks the CU Boulder NSF Liquid Crystal Materials Research Center REU program for undergraduate summer research support. We also thank Dr. J. E. Bara for initial helpful discussions References [1] Gin, D. L.; Gu, W.; Pindzola, B. A.; Zhou, W.-J. "Polymerized Lyotropic Liquid Crystal Assemblies for Materials Applications." Acc. Chem. Res. 2001, 34, [2] Mueller, A.; O'Brien, D. F. "Supramolecular Materials via Polymerization of Mesophases of Hydrated Amphiphiles." Chem. Rev. 2002, 102, [3] Gin, D. L.; Lu, X.; Nemade, P. R.; Pecinovsky, C. S.; Xu, Y.; Zhou, M. "Recent Advances in the Design of Polymerizable Lyotropic Liquid-Crystal Assemblies for Heterogeneous Catalysis and Selective Separations." Adv. Funct. Mater. 2006, 16, [4] Gin, D. L.; Bara, J. E.; Noble, R. D.; Elliott, B. J. "Polymerized Lyotropic Liquid Crystal Assemblies for Membrane Applications." Macromol. Rapid Commun. 2008, 29,

85 [5] Gin, D. L.; Noble, R. D. "Designing the Next Generation of Chemical Separation Membranes." Science 2011, 332, [6] Zhou, M. J.; Nemade, P. R.; Lu, X. Y.; Zeng, X. H.; Hatakeyama, E. S.; Noble, R. D.; Gin, D. L. "New type of membrane material for water desalination based on a crosslinked bicontinuous cubic lyotropic liquid crystal assembly." J. Am. Chem. Soc. 2007, 129, [7] Hatakeyama, E. S.; Gabriel, C. J.; Wiesenauer, B. R.; Lohr, J. L.; Zhou, M.; Noble, R. D.; Gin, D. L. "Water filtration performance of a lyotropic liquid crystal polymer membrane with uniform, sub-1-nm pores." J. Membr. Sci. 2011, 366, [8] Hatakeyama, E. S.; Wiesenauer, B. R.; Gabriel, C. J.; Noble, R. D.; Gin, D. L. "Nanoporous, Bicontinuous Cubic Lyotropic Liquid Crystal Networks via Polymerizable Gemini Ammonium Surfactants." Chem. Mater. 2010, 22, [9] Noble, R. D.; Stern, S. A. Membrane Separations Technology: Principles and Applications; Elsevier: Amsterdam, [10] Bhattacharya, A.; Ghosh, P. "Nanofiltration and Reverse Osmosis Membranes: Theory and Application in Separation of Electrolytes." Rev. Chem. Eng. 2004, 20, [11] Henmi, M.; Nakatsuji, K.; Ichikawa, T.; Tomioka, H.; Sakamoto, T.; Yoshio, M.; Kato, T. "Self-Organized Liquid-Crystalline Nanostructured Membranes for Water Treatment: Selective Permeation of Ions." Adv. Mater. 2012, 24, [12] Zhou, M.; Kidd, T. J.; Noble, R. D.; Gin, D. L. "Supported Lyotropic Liquid-Crystal Polymer Membranes: Promising Materials for Molecular-Size-Selective Aqueous Nanofiltration." Adv. Mater. 2005, 17, [13] Auvray, X.; Perche, T.; Petipas, C.; Anthore, R.; Marti, M. J.; Rico, I.; Lattes, A. "Influence of solvent-headgroup interactions on the formation of lyotropic liquid crystal phases of surfactants in water and nonaqueous protic and aprotic solvents." Langmuir 1992, 8, [14] Kerr, R. L.; Miller, S. A.; Shoemaker, R. K.; Elliott, B. J.; Gin, D. L. "New Type of Li Ion Conductor with 3D Interconnected Nanopores via Polymerization of a Liquid Organic Electrolyte-Filled Lyotropic Liquid-Crystal Assembly." J. Am. Chem. Soc. 2009, 131, [15] Greaves, T. L.; Drummond, C. J. "Ionic liquids as amphiphile self-assembly media." Chem. Soc. Rev. 2008, 37, [16] Shimura, H.; Yoshio, M.; Hoshino, K.; Mukai, T.; Ohno, H.; Kato, T. "Noncovalent Approach to One-Dimensional Ion Conductors: Enhancement of Ionic Conductivities in Nanostructured Columnar Liquid Crystals." J. Am. Chem. Soc. 2008, 130, [17] Tiddy, G. J. T. "Surfactant-water liquid crystal phases." Phys. Rep. 1980, 57,

86 [18] Bowen, W. R.; Mohammad, A. W.; Hilal, N. "Characterisation of nanofiltration membranes for predictive purposes -- use of salts, uncharged solutes and atomic force microscopy." J. Membr. Sci. 1997, 126, [19] Kosutic, K.; Furac, L.; Sipos, L.; Kunst, B. "Removal of arsenic and pesticides from drinking water by nanofiltration membranes." Sep. Purif. Technol. 2005, 42, [20] Nightingale, E. R. "Phenomenological Theory of Ion Solvation. Effective Radii of Hydrated Ions." J. Phys. Chem. 1959, 63, [21] Ho, W. S. W.; Sirkar, K. K. Membrane Handbook; Chapman & Hall: New York, [22] Pindzola, B. A.; Hoag, B. P.; Gin, D. L. "Polymerization of a Phosphonium Diene Amphiphile in the Regular Hexagonal Phase with Retention of Mesostructure." J. Am. Chem. Soc. 2001, 123, [23] Bara, J. E.; Hatakeyama, E. S.; Wiesenauer, B. R.; Zeng, X.; Noble, R. D.; Gin, D. L. "Thermotropic liquid crystal behaviour of gemini imidazolium-based ionic amphiphiles." Liq. Cryst. 2010, 37, [24] Gin, D. L.; Zhou, M.; Noble, R. D.; Bara, J. E.; Wisenauer, B. R.; Kerr, R. L. "Lyotropic Liquid Crystal Membranes Based on Cross-linked Type I Bicontinuous Cubic Phases," U.S. Patent , [25] Greaves, T. L.; Weerawardena, A.; Fong, C.; Krodkiewska, I.; Drummond, C. J. "Protic Ionic Liquids: Solvents with Tunable Phase Behavior and Physicochemical Properties." J. Phys. Chem. B 2006, 110, [26] Pindzola, B. A.; Jin, J.; Gin, D. L. "Cross-Linked Normal Hexagonal and Bicontinuous Cubic Assemblies via Polymerizable Gemini Amphiphiles." J. Am. Chem. Soc. 2003, 125, [27] Fontell, K. "Cubic phases in surfactant and surfactant-like lipid systems." Colloid Polym. Sci. 1990, 268, [28] Mariani, P.; Luzzati, V.; Delacroix, H. "Cubic phases of lipid-containing systems : Structure analysis and biological implications." J. Mol. Biol. 1988, 204, [29] Petersen, R. J. "Composite reverse osmosis and nanofiltration membranes." J. Membr. Sci. 1993, 83, [30] Aimar, P.; Meireles, M.; Sanchez, V. "A contribution to the translation of retention curves into pore size distributions for sieving membranes." J. Membr. Sci. 1990, 54,

87 CHAPTER 3 Thin-film composite bicontinuous cubic lyotropic liquid crystal polymer membranes: Effects of anion-exchange on water filtration performance (Manuscript published under the same title in Journal of Membrane Science 2014, 455, , co-authored with Wiesenauer, B. W.; Noble, R. D.; Gin, D. L.) Abstract The effects of anion-exchange on the filtration performance of thin-film composite (TFC) membranes with an active layer consisting of a nanostructured, lyotropic (i.e., surfactant) liquid crystal (LLC) polymer were investigated. These TFC LLC membranes are made by the in situ cross-linking of reactive amphiphiles (i.e., surfactants) that self-organize in the presence of glycerol into a type I bicontinuous cubic (QI) phase that contains a uniform, 3D-interconnected pore network lined with tethered cationic moieties and free mobile anions in the pores. In this study, a systematic series of experiments were performed to independently investigate how monovalent cations and anions affect transport in these TFC QI membranes. TFC QI membranes exposed to feed solutions that contain different cations (i.e., Li + (aq), Na + (aq), and K + (aq)) but have the same monovalent anion as the free mobile anion in the membrane (i.e., Br (aq)) have a constant flux and a high rejection (>98%). When the cation is kept constant (i.e., Na + (aq)) and the anion in the feed is varied (i.e., Cl (aq), Br (aq), NO3 (aq), and I (aq)) and allowed to partially anionexchange with the membrane, a high rejection is maintained ( 96%), but the flux significantly changes depending on the anion in the feed solution. The flux of the TFC QI membranes can also be repeatedly be cycled by contacting the membranes with different anions. Control experiments with completely anion-exchanged TFC QI membranes (i.e., with Cl (aq), Br (aq), NO3 (aq), and I 71

88 (aq)) showed that the rejection of sodium salts and uncharged organic solutes was virtually the same for all of the completely anion-exchanged membranes. As a whole, these results demonstrate that the flux of these TFC QI membranes can be tuned by changing the anion with little to no change in the rejection performance. The unique performance characteristics of TFC QI membranes may offer advantages over conventional NF and RO membranes for water purification applications or other aqueous separations Introduction Inadequate access to clean water is a significant worldwide problem [1-5]. Over 1 billion people lack access to safe drinking water and this number is expected to increase [1]. Not only is clean water essential for survival, but it also significantly impacts food and energy production as well as industrial output and the quality of the environment [2-4]. As the demand for clean water intensifies, it is imperative to be able to efficiently remove salts, organic solutes, and other impurities from contaminated sources of water. Membrane technology is currently utilized in numerous processes to efficiently generate clean water [1-3,5]. In order to remove virtually all dissolved salt ions and small organics from water, reverse osmosis (RO) membranes can be employed. RO membranes are typically thin-film composite (TFC) or asymmetric membranes consisting of an ultrathin (ca µm thick) dense polymer layer (i.e., the active layer) on top of a porous support [6-8]. Because the active layer in RO membranes is dense and does not contain discrete pores, the transport is described by the solution-diffusion model [3,9]. In this model, solutes and water in the feed dissolve in the active layer and then diffuse through the membrane. Solutes are rejected by the membrane due to the differences in the solubility and mobility of the feed components in the active layer. The 72

89 permeability of dense RO membranes is typically much lower than that of membranes with discrete pores. Another type of membrane commonly used to purify water to a high degree is nanofiltration (NF) membranes [6,8,10]. These porous membranes can remove almost all divalent ions and large uncharged solutes from water but only partially reject monovalent ions and very small organics. NF membranes are also typically TFC or asymmetric membranes, but the active layer contains discrete pores that are about 1 nm in diameter [6,10,11]. Rejection of neutral solutes in NF membranes is well described by a pore flow model where separation is achieved by molecular size-exclusion [3,10,11]. However, due to the extremely small pore size, any surface charge on NF membranes can result in appreciable rejection of ions that would normally pass through based solely on size [3,11]. The transport of ions through a charged material is often qualitatively described by the Donnan exclusion mechanism [9,10,12,13]. Coions (same sign charge as the fixed membrane charge) are repelled by the charged membrane; and due to the requirement of electroneutrality on a macroscopic scale, their counterions are rejected as well. Although current NF and RO membranes are capable of efficiently producing large amounts of high purity water, they face a number of challenges. One of the inherent difficulties in the construction of NF and RO membranes is the inability to control the average size and distribution of the effective pores or interstitial polymer voids on the molecular size scale in the active layer [10,14,15]. Current production methods such as phase inversion and interfacial polymerization offer little control over the effective pore size and pore size distribution of the membrane [10,15]. A wide pore size distribution limits rejection selectivity and the ability to cleanly fractionate solutes of different size. Other inherent challenges that face conventional NF 73

90 and RO membranes based on polyamide or cellulose acetate chemistries are limited resistance to chlorine or strong ph feed solutions, as well as a lack of control over the charge character and charge density of the membrane [10,16-19]. Membrane charge significantly impacts and often improves rejection and permeability performance [10,17-21]. The study and development of polymer materials or processes to create polymer membranes with uniform pores on the molecular size scale, tunable charge characteristics, and improved resistance to harsh chemical feeds would offer significant improvements over conventional NF and RO membranes. Our research team previously developed a unique type of polymeric NF membrane material that contains uniform, sub-1-nm pores that are lined with tethered cationic moieties and free mobile anions [22-24]. This new NF membrane material is made by the in situ cross-linking of reactive amphiphiles (i.e., surfactants) that self-organize in the presence of a solvent (e.g., most commonly water) into ordered lyotropic liquid crystal (LLC) phases. The best LLC phase that has been utilized for this new type of NF material is a cross-linkable type I bicontinuous cubic (QI) phase (Figure 3.1). Consisting of an ordered (i.e., cubic symmetry) 3D-interconnected annular pore network, QI phases do not require any sort of phase alignment for the best transport characteristics, unlike lower dimensionality hexagonal (H) and lamellar (L) LLC phases. Water is transported through the QI phase by the interconnected, annulus-like water channels, but due to their small dimensions and charged character, larger solutes and ions are excluded. 74

91 Figure 3.1. Imidazolium-based monomer 3 that forms a QI phase with glycerol, and the formation of cross-linked TFC QI membranes. Initial studies of supported QI-phase polymer membranes have shown they have performance characteristics in-between those of conventional RO and NF membranes [24]. These QI-phase LLC membranes reject neutral solutes like a porous NF membrane but reject monovalent and divalent salts comparable to a RO membrane at brackish water feed concentrations. Although initial studies involving supported QI-phase membranes demonstrate these materials have significant potential as a new type of uniform, nanoporous NF membrane, the only viable method found for membrane fabrication involves heating and pressing the monomer/water mixture into a porous support and subsequently photo-cross-linking to covalently lock in the QI-phase. In this process, the support is fully infused with the QI-phase material and as a result the active layer is as thick as the support itself (ca. 40 µm). Due to the extremely thick active layer, the water flux of the supported QI-phase membranes is far too low for any extensive testing or practical application. Even though the flux of the melt-infused QIphase membranes is much lower than commercial NF and RO membranes, the thicknessnormalized pure water permeance is comparable to a commercial RO membrane [24]. This 75

92 suggests that it is possible for QI-phase membranes to obtain fluxes comparable to commercial RO membranes if the active layer of the QI-phase membranes can be reduced to the same approximate 0.1 µm thickness as commercial RO membranes. The major problem in trying to process thin films of QI phases (prior to cross-linking) on top of a support is the fact that these LLC phases are particularly sensitive to changes in composition and temperature [25]. A specific composition of reactive LLC monomer and water must be maintained at a specific elevated temperature range until the desired QI phase is completely polymerized. As a result, solution-based processing techniques are not possible for narrow-composition, water-based QI-phases, since any evaporative loss of water that occurs during processing will result in disruption of the QI-phase monomer assembly. To overcome the processing limitations of the previous water-based QI phases, our research team recently developed a new imidazolium-based monomer (3) that forms a crosslinkable QI-phase with glycerol instead of water (Figure 3.1) [26]. Due to the low volatility of glycerol (which is also water-miscible and non-toxic), minimal evaporative loss occurs during solution-based thin-film processing. As a result, a thin film of a cross-linked QI phase can be prepared by dissolving the monomer and glycerol in volatile solvent such as methanol and then roll-casting this solution over the top of a porous asymmetric membrane support. After mild heating to evaporate off the methanol and photo-cross-linking at 70 C, the result is a thin film of a cross-linked QI phase on top of the support (i.e., a TFC QI membrane). This new glycerolbased system and processing approach enabled TFC QI membranes to be prepared with an average active layer thickness of ca. 3 µm [26]. These TFC QI membranes showed similar rejection performance to the previous thick, melt-infused, supported QI-phase membranes, except the flux was ca. 10 times greater due to the much thinner active layer. Initial experiments 76

93 on these new TFC QI membranes based on 3/glycerol also revealed that the filtration fluxes of aq. NaCl solutions were significantly higher than that of similar test solutions containing uncharged molecular solutes. Interestingly, it was found that soaking the 3/glycerol TFC QI membranes in an aqueous NaCl solution prior to neutral solute filtrations afforded the same increased flux without any appreciable change in rejection. We hypothesized this was most likely the result of some sort of anion-exchange between the original bromide anions in the membrane and the chloride anions in the feed solution [26]. Herein, we investigate the effects of different anions and cations in the aqueous feed solution on the filtration performance of TFC QI membranes in order to acquire a more fundamental understanding of the ion transport and exchange in these membranes. A systematic series of experiments was performed to independently investigate how monovalent cations and anions impact transport. By varying the monovalent cation (i.e., Li + (aq), Na + (aq), and K + (aq)) and keeping the anion in solution the same as the free mobile anion in the membrane (i.e., Br (aq)), it was found that the TFC QI membranes of 3/glycerol maintain a high rejection ( 98%) and constant flux for all the alkali metal bromide salts tested. When the monovalent cation is kept constant (i.e., Na + (aq)) and the anion is varied (i.e., Cl (aq), Br (aq), NO3 (aq), and I (aq)), a high salt rejection is also maintained ( 96%), but the flux significantly changes depending on the anion in solution. Not only does the flux vary with the anion in solution, but the pure water flux of the membrane reflects the last anion it was contacted with. These changes in flux are reversible and can be repeatedly cycled by exposing the membrane to solutions containing different anions, even under conditions that do not afford complete anion-exchange in the membranes. Control experiments with completely anion-exchanged 3/glycerol TFC QI membranes (i.e., with Cl (aq), NO3 (aq), and I (aq)) showed that the flux of the completely anion-exchanged membranes is the 77

94 same as that of membranes only contacted with a dilute solution of the corresponding anion with virtually the same rejections for all of the test salts and uncharged organic solutes. Similar rejection of the uncharged organic solutes by the completely anion-exchanged membranes suggests the monovalent anions tested have little to no effect on the physical pore size of the membrane, even though the flux still significantly depends on the anion. This behavior is contrary to the conventional permeability and selectivity/rejection tradeoffs usually observed in membrane materials. These results demonstrate the performance characteristics of these TFC QI polymer membranes are very unique compared to conventional NF and RO membranes and have significant potential as a new type of nanoporous NF membrane for water purification applications or other aqueous separations Experimental Materials and instrumentation All reagents and solutes were purchased from Sigma-Aldrich and were ACS Reagent Grade or higher quality. Monomer 3 was prepared according to literature procedures, and characterization and purity data for the monomer was consistent with published data [26]. The water used in the preparation of solute solutions used in the water filtration experiments was deionized (DI) and had a resistivity of >10 MΩ cm 1. Mylar sheets were purchased from American Micro Industry, Inc. Commercial General Electric (GE) polysulfone (PSf) ultrafiltration (UF) membranes with a molecular-weight-cut-off of 30K (amu) were purchased from Sterlitech Corporation (YMERSP3001) and used as the support material for the TFC QI membranes. 78

95 Powder X-ray diffraction (PXRD) spectra were collected with an Inel CPS 120 detector with a monochromated Cu Kα radiation source. All PXRD spectra were calibrated using a silver behenate diffraction standard (d100 = 58.4 ± 0.1 Å). Polarized light microscopy (PLM) was performed using a Leica DMRXP polarizing light microscope equipped with a Linkam LTS 350 thermal stage, a Linkam CI 94 temperature controller, and a Q-Imaging MicroPublisher 3.3 RTV digital camera. Scanning electron microscopy (SEM) and energy-dispersive X-ray spectroscopy (EDS) was performed with a JEOL JSM-6480LV scanning electron microscope equipped with an EDS detector. Fourier-transform infrared spectroscopy (FT-IR) measurements were acquired using a Thermo Scientific Nicolet 6700 spectrometer equipped with a diamond crystal PIKE MIRacle TM single-reflection horizontal ATR accessory. UV-visible spectroscopy was performed with an Agilent 8453 spectrophotometer. Radical photo-polymerization/cross-linking was performed in a N2 purged polymer glove box from Coy Labs that was equipped with a Spectroline XX-15A 365 nm UV lamp (1 mw cm -2 at the sample surface), a temperature controlled Xtreme TM hot/cold plate from LR Technologies, and an Alpha Omega OXY-SEN Oxygen Monitor. UV light fluxes were measured using a UVP UVX digital radiometer equipped with a UVX-36 sensor. Wire-wound, stainless-steel applicator rods used to solution roll-cast the TFC QI membranes were purchased from the Paul N. Gardner Company. Water filtration studies were performed using custom-designed, stainless-steel, stirred dead-end filtration cells for 2.5- cm-diameter membrane test samples. The ion conductivity of the solutions was measured using a VWR International electrical conductivity meter model 2052-B. The ph of the solutions was measured using a Corning 320 ph meter. Total organic carbon (TOC) analysis of aqueous solutions containing organic solutes was conducted using a Test N Tube TOC kit (Hach Company) and a COD reactor (DRM 200, Hach Company). 79

96 Membrane preparation and characterization A cut section of a GE UF support membrane (6 cm 10 cm) was soaked in 300 ml of methanol for 20 min to remove any preservative or wetting agents. The support was then allowed to air dry for 30 min. A 30 wt% QI-phase casting solution in methanol was prepared by adding appropriate amounts of monomer 3, glycerol, and 2-hydroxy-2-methylpropiohenone (HMP, a radical photo-initiator) [79.7/19.8/0.5 (w/w/w)] to a 10-mL glass vial and dissolving the components in the required amount of methanol. The UF support membrane was taped to an aluminum plate along the two long edges and a small piece of aluminum foil was placed on top of one of the taped edges. A #3 wire-wound rod from the Paul N. Gardner Company was placed in the aluminum trough. Approximately 0.6 ml of the casting solution was pipette along the length of the trough and the rod was then immediately drawn across the support membrane at a slow constant speed. The membrane sample was allowed to air dry at ambient temperature for 30 min. This casting procedure was repeated by drawing the wire-wound rod in the opposite direction of the initial cast. The sample was then allowed to air dry for 30 min at ambient temperature. The aluminum plate with the membrane sample was then put in a N2 purged polymer glove box (<3% O2) and placed on a temperature-controlled hot stage. The stage was warmed to 25 C, and then the temperature was gradually increased to 75 C by heating at a rate of 5 C/min and holding the temperature for 2 min every 10 C up to a final temperature of 75 C. This gradual heating process was performed to remove excess methanol casting solvent and anneal the sample. The aluminum plate and membrane sample were then cooled to ambient temperature and allowed to sit for 20 min. The sample was placed back on the 75 C hot stage and an O-ring spacer and quartz plate were then placed on top of the membrane sample. The 80

97 sample was allowed to equilibrate for 30 min and then irradiated with a 365 nm UV lamp (1 mw cm -2 ) for 1 h while on the heated stage. The final TFC QI membrane sample was removed from the polymer glove box for characterization and filtration testing. The fabric backing of the support was delaminated and the membrane was analyzed using PXRD and SEM. A high degree of conversion of the TFC QI membrane was verified by FT-IR as described in a previous study [26] Anion-exchange of bulk cross-linked Q I -phase films and TFC Q I membranes Bulk cross-linked QI-phase films were prepared as described in a previous publication [26]. Homogeneous QI-phase monomer mixtures containing the proper amount of monomer 3, glycerol, and HMP were sandwiched between two pre-cut pieces of Mylar. The samples were clamped in-between two pre-heated quartz plates at 70 C to form the QI-phase and then placed on a temperature controlled stage maintained at 70 C. After 5 min, the samples were irradiated with 365 nm light (1 mw cm -2 ) for 1 h. The quartz plates and Mylar sheets were carefully removed to afford ca. 200 µm thick, free-standing QI polymer films. PLM and PXRD analysis of the polymerized samples confirmed that the QI-phase nanostructure was retained. The bulk QI polymer films were exchanged with different anions by soaking the films in concentrated aqueous salt solutions containing the desired anion (1 M NaNO3, 2 M NaCl) for 48 h with repeated replacement of the soaking solution every 12 h. The anion-exchanged bulk QI films were then soaked in DI water for 24 h with replacement of the rinsing solution at 12 h. The films were analyzed with PLM and PXRD. The films were then freeze-fractured in liquid N2 and their cross-sections were analyzed using EDS to confirm complete anion-exchange. 81

98 Anion-exchange of the bulk QI films to I was conducted under two different soaking conditions. A 0.5 M aq. NaI solution with a ph of 5.8 ± 0.5 was prepared using DI water equilibrated with atmospheric carbon dioxide, while a second 0.5 M aq. NaI solution was adjusted to a ph of 10 ± 0.5 with an aq. NaOH solution. Prior to soaking, UV-visible spectra of the bulk QI films were acquired. The bulk QI films were soaked in the two NaI solutions for 48 h with repeated replacement of the soaking solution every 12 h. After 48 h, a UV-visible spectrum of the films was acquired. The films were then soaked in DI water at the appropriate ph for 24 h, with replacement of the rinsing solution at 12 h. A final UV-visible spectrum was obtained on the QI films and the films were analyzed with PLM and PXRD. The films were then freezefractured in liquid N2, and their cross-sections were analyzed using EDS to confirm complete anion-exchange. TFC QI membranes discs were punched out from small sheets and exchanged with the desired anion using the same procedure outlined above for the bulk QI polymer films. EDS was attempted on the top surface of the TFC QI membranes, but the small amount of active layer material did not provide adequate signal to confirm anion-exchange. It was assumed the soaking conditions that completely exchanged the bulk QI films also completely exchanged the TFC QI membranes Water filtration testing TFC QI membrane discs (2.5 cm in diameter) were loaded into the custom-made, stainless-steel, stirred, dead-end filtration cells. The membrane holder of each cell has a 2.5 cm outer diameter, but only an effective filtration area of ca. 3.8 cm 2 with the O-ring configuration. All water filtration experiments were performed at ambient temperature (22 ± 1 C) with

99 10 6 Pa (400 psi) of N2 pressure as the driving force. The cells were filled with DI water and pressurized until at least 5 ml of permeate was collected in each cell and a steady-state flux was reached. This first filtration was to ensure the integrity of the membrane and clean out any contaminants that might remain in the membrane after processing. To wash out any of the previous permeate, the cells were filled with the desired feed solution and pressurized until 2 4 ml of permeate were collected. The cells were then repeatedly reloaded with the desired feed solution and 2 4 ml of permeate were collected until a steady-state flux was reached and the permeate was then analyzed to determine rejection. All of the reported fluxes and rejections are the average values measured from three or more individual membrane samples tested in separate experiments. The reported error is the standard deviation from the three or more individual membranes in separate experiments. For all of the filtration studies, the flux (J) was calculated using Equation 3.1: J = V A t (3.1), where ΔV is the permeate volume, A is the surface area of the membrane (3.8 cm 2 ), and Δt is the time required to collect the permeate. The rejection (R) was calculated as follows: R (%) = 1 C permeate C feed 100 (3.2), where Cp and Cf are the concentrations of solute in the permeate and feed, respectively. The salt concentration in the feed and permeate was measured using an electrical conductivity meter. A 83

100 calibration curve was made for each salt. For filtration experiments where the permeate and feed solutions can contain a mixture of anions due to anion-exchange with the membrane, it was assumed the conductivity is directly proportional to the concentration. For these experiments, the rejection is specified as total ion conductivity rejection. The concentration of aq. organic solutes was determined with a TOC reagent kit from Hach. A calibration curve with standard aq. organic solutions was made prior to each analysis Results and discussion Characterization of TFC Q I membranes made with a commercial UF support In a previous study, TFC QI membranes were made for the first time by roll-casting a glycerol-based QI-phase mixture of 3 dissolved in methanol on top of a porous, asymmetric polyethersulfone (PES) support [26]. These PES support membranes were made in-house on a small scale by phase-inversion [26]. To demonstrate the adaptability of this new processing approach and more rapidly and reliably fabricate TFC QI membranes for extensive study, a commercial UF membrane support was utilized. A methanol solution containing 30 wt% [79.7/19.8/0.5 (w/w/w)] 3/glycerol/HMP was roll-cast two times onto the surface of a GE UF PSf membrane. After mild heating to evaporate off the methanol, annealing, and photo-crosslinking at 75 C, PXRD analysis was performed. As can be seen in Figure 3.2, the presence of a peak at ca. 39 Å in the TFC membrane corresponds to the 1 6 d-spacing peak originally observed in the bulk QI material. This PXRD peak match confirms that the cross-linked active layer of the TFC membrane has the desired QI nanostructure [26]. SEM indicated that the resulting TFC QI membranes had a QI active layer with an average thickness of ca. 4 µm (Figure 3.3). FT-IR analysis confirmed that the TFC QI membranes had a high degree of 1,3-diene 84

101 conversion after photo-cross-linking. The highly cross-linked TFC QI membranes are mechanically stable and have been demonstrated to withstand over 50 days of continuous operation at 400 psi ( Pa) without loss of the QI-phase structure or degradation in water filtration performance. The TFC QI membranes made with the GE UF support were used in the subsequent studies. Figure 3.2. PXRD spectra of the cross-linked TFC QI membrane and the uncoated GE UF PSf support. Figure 3.3. Cross-sectional SEM photo of the TFC QI membrane made with the GE UF support. 85

102 Effect of monovalent cation on water filtration performance To investigate the effect of monovalent cations on water filtration performance, it was necessary to test the TFC QI membranes with bromide salts to eliminate any possible exchange of different anions between the membrane and feed solution. TFC QI membranes were tested with 0.01 M feeds of aq. LiBr, NaBr, and KBr solutions. As can be seen in Table 3.1, the TFC QI membranes exhibited a high rejection (ca. 98%) for all of the tested alkali bromide salts. Compared to a conventional porous NF membrane that only partially rejects monovalent ions [10], the TFC QI membranes almost completely reject monovalent ions due to the large number of tethered cationic groups in the pores that hinder ion transport by Donnan exclusion. The flux of the TFC QI membranes does not change upon contact with any of the bromide salt solutions. Compared to a previous study where the flux of the TFC QI membranes changed upon contact with a NaCl solution [26], this result suggests that anion-exchange between the original bromide anions in the membrane and different anions in solution is responsible for changing the flux of TFC QI membranes. Table 3.1. Flux and rejection data of the TFC QI membranes with different alkali bromide salts. All experiments performed in stirred dead-end filtration cells with an applied pressure of Pa (400 psi). Feed solution Rejection (%) Flux (L m -2 h -1 ) 0.01 M aq. LiBr 98.0 ± ± M aq. NaBr 98.4 ± ± M aq. KBr 98.2 ± ±

103 Effect of monovalent anion on water filtration performance The effect of the anion on the water filtration performance of TFC QI membranes was investigated by exposing membranes to 0.01 M aq. solutions of NaBr, NaCl, NaNO3, and NaI. Figure 3.4 shows the relative flux of the membranes with the sequential feeds tested compared to the initial pure water flux (Jo) measured for each membrane. As can be seen in Figure 3.4, initial treatment of the membranes with a NaBr solution had no impact on the flux of the membranes compared to the initial pure water flux. However, upon exposure to a NaCl solution, the relative flux of the membranes increased over 1.6 times. Not only did the relative flux increase while the membranes were exposed to the NaCl solution, but the relative pure water flux also increased the same amount after exposure. Treatment of the membranes with a NaBr solution restored the membranes to the initial pure water flux value. Contacting the TFC QI membranes with a NaNO3 solution slightly decreased the relative flux and pure water flux after exposure. Once again, treatment of the membranes with a NaBr solution restored the membranes to their original flux. Upon exposing the membranes to a NaI solution, the relative flux and pure water flux after exposure was reduced to less than 0.3 times the initial pure water flux. Contacting the membrane with a NaBr solution increased the relative flux, but it was still slightly lower than the original pure water flux. As seen in Table 3.2, the conductivity rejection of all the different sodium salt feeds remained in the similar range of %. 87

104 Figure 3.4. Relative flux of the TFC QI membranes exposed to different feeds with the sequential order listed from left to right. The experiments were performed in stirred dead-end filtration cells with an applied pressure of Pa (400 psi) and the concentration of all sodium salt feeds was 0.01 M. Table 3.2. Rejection of different sodium salts by the TFC QI membranes. All experiments performed in stirred dead-end filtration cells with an applied pressure of Pa (400 psi). Feed solution Conductivity rejection (%) 0.01 M aq. NaBr 97.6 ± M aq. NaCl 97.8 ± M aq. NaNO ± M aq. NaI 97.4 ± 0.4 It is clear from these experiments that different monovalent anions present in the feed can exchange with the original bromide anions in the TFC QI membranes and change the flux of the membranes without significantly altering the conductivity rejection. This anion-exchange process is reversible, and so the change in flux is reversible as well. The degree of anion- 88

105 exchange between the membrane and feed solution depends on the concentration of the different anions in the feed and the affinity of the different anions for the cationic membrane. Interestingly, the change in the relative flux by the different anions is analogous to the preferential take-up of the different anions (i.e., selectivity coefficients) commonly observed in quaternary ammonium anion-exchange materials [12]. In quaternary ammonium anion-exchange materials the selectivity coefficient sequence is typically: I > NO3 > Br > Cl [12]. The relative flux observed for the different anions increased in the sequence I < NO3 < Br < Cl. The anions that typically have a higher affinity and preferential absorption by quaternary ammonium materials appear to decrease the flux of the TFC QI membranes. It is important to consider that the dilute 0.01 M feed concentrations used in these experiments could not completely exchange all of the anions in the test membrane for the anions present in the feed, and so these experiments only represent the performance of partially anionexchanged TFC QI membranes. Continual exposure to a fresh feed solution with a specific anion would be required to completely anion-exchange the TFC QI membranes. Even though anions continue to partially exchange by repeated exposure to a dilute feed containing a specific anion, the TFC QI membranes in these experiments always immediately reached a new steady-state flux upon the initial exposure to a 0.01 M feed with a different anion and the performance did not appear to significantly change over time. This result suggests that the partially anion-exchanged TFC QI membranes may have similar performance characteristics to a completely anionexchanged membrane. In order to test this hypothesis, it was necessary to test completely anionexchanged TFC QI membranes and evaluate their flux and rejection performance. 89

106 Effect of complete monovalent anion-exchange on water filtration performance Complete anion-exchange was first investigated in bulk QI films to more easily identify the conditions required for complete anion-exchange and examine any changes in the nanostructure of the QI material. Bulk cross-linked QI films ca. 200 µm thick were soaked in concentrated aq. sodium salt solutions with the Cl (aq), NO3 (aq), and I (aq) anions. After repeated soaking in the salt solutions, the films were soaked in DI water, and cross-sections of the bulk QI films were analyzed with EDS to confirm complete anion-exchange. Figure 3.5 shows the PXRD profiles of the bulk QI films exchanged with different anions. As can be seen in Figure 3.5, all of the anion-exchanged films show a shift in the primary PXRD d-spacing peak upon replacing the original glycerol in the pores with water; however, there is no significant variation in the primary d-spacing value for the different anions. The shift in the primary d-spacing peak upon solvent replacement could be the result of a minor structural change in the QI material due to a change in the solvation environment and the relative ion stabilization, or the result of a small amount of water loss that can occur at ambient conditions due to the volatility of water. 90

107 Figure 3.5. PXRD profiles of anion-exchanged bulk QI films Interestingly, it was observed that bulk QI test films soaking in 0.5 M aq. NaI solutions turned a distinct brown and yellow color suggestive of triiodide (I3 ) formation. UV-visible analysis of the QI films (Figure 3.6) soaking in NaI solution confirmed the presence of triiodide by a characteristic absorption peak centered at ca. 355 nm [27]. Upon soaking these QI films in DI water, triiodide was no longer detectable by UV-visible analysis. Films soaking in dilute 0.01 M aq. NaI solutions also did not have any detectable amount of triiodide by UV-visible analysis. Although iodide (I ) can be oxidized by dissolved O2 in acidic solutions to form triiodide (I3 ), oxidation is usually insignificant in neutral solutions in the absence of heat, light, and metal ions [28]. The ph of the DI water used in these soaking experiments is slightly acidic due to equilibrium with atmospheric carbon dioxide. When the bulk QI films were soaked in a 0.5 M aq. NaI solution adjusted to a ph of 10 with aq. NaOH, no visible triiodide formation was observed by UV-visible analysis (Figure 3.7). These control experiments indicate that triiodide is only formed in the QI material while soaking in a concentrated NaI solution that is slightly acidic, 91

108 consistent with conditions known to convert I to I3 [28]. Further study is needed to completely characterize the redox reactions that can occur with the iodide anion in the nanoporous environment of the cross-linked QI material; however, these initial results indicate there should be no detectable amount of triiodide formed under the conditions used for water filtration testing. Figure 3.6. UV-visible spectra of a bulk cross-linked QI film before, during, and after soaking in a 0.5 M aq. NaI solution. Figure 3.7. UV-visible spectra of a bulk cross-linked QI film before, during, and after soaking in a 0.5 M aq. NaI solution adjusted to a ph of

109 TFC QI membranes were anion-exchanged using the same procedure used to quantitatively anion-exchange the bulk QI films. EDS was attempted on the top surface of the TFC QI membranes to confirm complete anion-exchange, but the small amount of active layer material did not provide adequate signal for analysis. Consequently, it was assumed the highly concentrated salt solutions that successfully anion-exchanged the thick bulk QI films also completely exchanged the TFC membranes containing a thin QI active layer. These completely anion-exchanged TFC QI membranes were then tested with 0.01 M aq. sodium salt solutions with the corresponding anion, as well as with 2000 ppm aq. feed solutions individually containing uncharged organic solutes of varying molecular size (sucrose, glucose, glycerol, and ethylene glycol (EG)) in order to evaluate the physical pore size of the membranes. As seen in Figure 3.8, the rejection of the sodium salts by the anion-exchanged membranes is about the same for all of the different anions and within error of the previously measured conductivity rejections of partially anion-exchanged TFC QI membranes (Table 3.2). The rejection of the uncharged organic solutes by the anion-exchanged membranes is also very similar for all of the different anions. The similar rejection values of the uncharged solutes by the anion-exchanged TFC membranes suggests there is little to no change in the physical pore size for the different monovalent anions tested, since only size-exclusion is ideally operating in this case. 93

110 Figure 3.8. Rejection of completely anion-exchanged TFC QI membranes. All experiments performed in stirred dead-end filtration cells with an applied pressure of Pa (400 psi). The concentration of the sodium salt solutions was 0.01 M, and the concentration of the aqueous organic feed solutions was 2000 ppm. Although the overall rejection performance of the completely anion-exchanged TFC QI membranes is similar for all of the different monovalent anions, the flux of the membranes still highly depends on the anion. As seen in Table 3.3, the pure water flux increased for the chlorideexchanged membranes, was about the same for the nitrate-exchanged membranes, and significantly decreased for the iodide-exchanged membranes compared to the original bromidecontaining membranes. 94

111 Table 3.3. Comparison of the pure water flux of completely anion-exchanged TFC QI membranes and TFC QI membranes partially anion-exchanged by exposure to 0.01 M sodium salt solutions. Anion I Br Cl NO 3 Pure water flux (L m -2 h -1 ) Completely anion-exchanged TFC QI membranes Partially anion-exchanged TFC QI membranes 0.08 ± ± ± ± ± ± ± ± 0.07 When the pure water flux of the completely anion-exchanged membranes is compared to TFC QI membranes that were only partially anion-exchanged via exposure to 0.01 M aq. sodium salt solutions (sub-stoichiometric exchange solutions) (Table 3.3), the resulting pure water flux is approximately the same. This indicates that there is no significant difference in the performance of TFC QI membranes that are completely anion-exchanged compared to membranes that are only partially anion-exchanged with the corresponding anion. It is clear from these experiments that TFC QI membranes possess very unique performance characteristics, since the flux can be tuned by exposing the membrane to different anions with little to no change in the overall rejection performance. We hypothesize that the changes in flux of the TFC QI membranes from the different anions is most likely due to (1) changes in how strongly the water molecules are bound in the nanopores resulting from interactions with the anions and tethered cations, or (2) changes in the relative amount of ion pairs and dissociated ions in the QI material and membrane surface. Further study is needed to completely understand the underlying cause of these changes 95

112 in flux with anion-exchange and what degree of anion-exchange is necessary to observe these effects. Additional control experiments were also performed with iodide-exchanged membranes to confirm that the triiodide formation that occurred during anion-exchange under slightly acidic conditions had no significant impact on the filtration performance. TFC QI membranes were exchanged to the iodide anion at a ph of 10, and all subsequent filtration studies were also performed at a ph of 10 to prevent any triiodide formation during the whole course of these control experiments. As can be seen in Figure 3.9, the flux of the TFC QI membranes exchanged to iodide at a ph of 10 (no I3 formation) was steady when exposed to a 0.01 M aq. NaI (ph =10) feed solution. When the iodide membranes were then contacted with DI water adjusted to ph 10, the flux slowly increased. This is most likely the result of hydroxide anions present in the feed slowly exchanging with the iodide anions in the membranes and changing their flux. As soon as the membranes were contacted with a NaI solution adjusted to ph 10, the flux returned to the original observed value. The conductivity rejection of the alkaline 0.01 M NaI solutions was the same observed for the iodide membranes exchanged under slightly acidic conditions. These control experiments suggest there is no significant difference in the performance of iodide membranes exchanged under slightly acidic or basic conditions that allow or prevent triiodide formation. The slow change in flux of iodide-exchanged membranes exposed to DI water at ph 10 seems to indicate the TFC QI membranes have a much higher affinity for the iodide anion compared to the hydroxide anion, which prevents rapid anion-exchange to create a new steady state flux as observed in the previous experiments. This is in agreement with the selectivity coefficients for OH and I commonly observed with other quaternary ammonium anionexchange materials [12]. Higher concentrations of the hydroxide anion would be required to 96

113 more rapidly anion-exchange the membrane and establish a new steady state flux. Additional studies are needed to completely understand how TFC QI membranes perform under alkaline feed conditions and how the membranes perform when exposed to concentrated NaI solutions that allow triiodide to form within the QI material. Figure 3.9. Flux of iodide-exchanged TFC QI membranes tested under alkaline (ph = 10) conditions as a function of time with pure water and 0.01 M aq. NaI feed solutions Conclusions The effects of different cations and anions in the feed solution on the filtration performance of TFC QI membranes were systematically investigated. The effect of the cation was investigated by exposing TFC QI membranes to feed solutions that contain different cations (i.e., Li + (aq), Na + (aq), and K + (aq)) but have the same monovalent anion as the free mobile anion in the membrane (i.e., Br (aq)). All of the alkali bromide salts tested had a high rejection (>98%) and a constant flux. The effect of the anion was investigated by exposing the TFC QI membranes to feed solutions containing different anions (i.e., Cl (aq), Br (aq), NO3 (aq), and I (aq)) but the same cation (i.e., Na + (aq)). The flux of TFC QI membranes significantly changes depending on the 97

114 anion in the feed solution, but the rejection remains high (>96%) for all of the sodium salts tested. Not only do the anions change the flux of TFC QI membranes during exposure, but the pure water flux is changed the same amount after exposure. The flux of the TFC QI membranes can repeatedly be altered by contacting the membrane with feeds containing different anions. These results demonstrate that partial anion-exchange between the feed solution and the TFC QI membrane significantly impacts the flux without altering the salt rejection. Control experiments with completely anion-exchanged TFC QI membranes (i.e., with Cl (aq), Br (aq), NO3 (aq), and I (aq)) showed the rejection of sodium salts with the corresponding anion and the rejection of uncharged aq. organic solutions was virtually the same for all of the completely anion-exchanged membranes. The similar rejection of uncharged organic solutes suggests the tested monovalent anions have little to no effect on the physical pore size of the membrane. It was also found that the flux of completely anion-exchanged TFC QI membranes was the same as membranes only partially anion-exchanged via exposure to dilute 0.01 M sodium salt solutions. These results indicate there is no significant difference in the performance of completely anion-exchanged TFC QI membranes compared to TFC QI membranes partially anion-exchanged via feed exposure. Collectively, these results demonstrate the flux of TFC QI membranes can be tuned by exposing the membrane to a feed with a specific anion with little to no change in the rejection performance. TFC QI membranes have very unique performance characteristics that may offer significant advantages over conventional NF and RO membranes for a number of different aqueous separations. Further study is needed to acquire a fundamental understanding of why different anions significantly change the flux of the TFC QI membranes and what degree of anion-exchange is necessary for this to occur. Future work also involves evaluating how the rejection of different salts varies with concentration and how multivalent anions impact the 98

115 filtration performance of TFC QI membranes. We are also currently exploring methods to vary the QI pore size that would allow custom membranes to be made with a desired uniform pore size for a specific separation Acknowledgments Primary financial support from the National Science Foundation (CBET ) is gratefully acknowledged. Partial support for this work from a National Renewable Energy Laboratory-sponsored grant (13-2) from the Membrane Science, Engineering, and Technology (MAST) Center at the University of Colorado, Boulder is also gratefully acknowledged References [1] Service, R. F. "Desalination Freshens Up." Science 2006, 313, [2] Shannon, M. A.; Bohn, P. W.; Elimelech, M.; Georgiadis, J. G.; Marinas, B. J.; Mayes, A. M. "Science and technology for water purification in the coming decades." Nature 2008, 452, [3] Geise, G. M.; Lee, H.-S.; Miller, D. J.; Freeman, B. D.; McGrath, J. E.; Paul, D. R. "Water purification by membranes: The role of polymer science." J. Polym. Sci., Part B: Polym. Phys. 2010, 48, [4] Oki, T.; Kanae, S. "Global Hydrological Cycles and World Water Resources." Science 2006, 313, [5] Greenlee, L. F.; Lawler, D. F.; Freeman, B. D.; Marrot, B.; Moulin, P. "Reverse osmosis desalination: Water sources, technology, and today's challenges." Water Res. 2009, 43, [6] Petersen, R. J. "Composite reverse osmosis and nanofiltration membranes." J. Membr. Sci. 1993, 83, [7] Noble, R. D.; Stern, S. A. Membrane Separations Technology: Principles and Applications; Elsevier: Amsterdam, [8] Ho, W. S. W.; Sirkar, K. K. Membrane Handbook; Chapman & Hall: New York, [9] Baker, R. W. Membrane Technology and Applications; McGraw-Hill: New York,

116 [10] Schäfer, A. L.; Fane, A. G.; Waite, T. D. Nanofiltration: Principles and Applications; Elsevier: New York, [11] Bowen, W. R.; Welfoot, J. S. "Modelling the performance of membrane nanofiltration-- critical assessment and model development." Chem. Eng. Sci. 2002, 57, [12] Strathmann, H. Ion-Exchange Membrane Separation Processes; Elsevier: Amsterdam, [13] Bhattacharya, A.; Ghosh, P. "Nanofiltration and Reverse Osmosis Membranes: Theory and Application in Separation of Electrolytes." Rev. Chem. Eng. 2004, 20, [14] Lonsdale, H. K. "The growth of membrane technology." J. Membr. Sci. 1982, 10, [15] Ulbricht, M. "Advanced functional polymer membranes." Polymer 2006, 47, [16] Glater, J.; Hong, S.-k.; Elimelech, M. "The search for a chlorine-resistant reverse osmosis membrane." Desalination 1994, 95, [17] Park, H. B.; Freeman, B. D.; Zhang, Z.-B.; Sankir, M.; McGrath, J. E. "Highly Chlorine- Tolerant Polymers for Desalination." Angew. Chem. 2008, 120, [18] Schaep, J.; Vandecasteele, C. "Evaluating the charge of nanofiltration membranes." J. Membr. Sci. 2001, 188, [19] Childress, A. E.; Elimelech, M. "Relating Nanofiltration Membrane Performance to Membrane Charge (Electrokinetic) Characteristics." Environ. Sci. Technol. 2000, 34, [20] Paul, M.; Park, H. B.; Freeman, B. D.; Roy, A.; McGrath, J. E.; Riffle, J. S. "Synthesis and crosslinking of partially disulfonated poly(arylene ether sulfone) random copolymers as candidates for chlorine resistant reverse osmosis membranes." Polymer 2008, 49, [21] Yaroshchuk, A.; Staude, E. "Charged membranes for low pressure reverse osmosis properties and applications." Desalination 1992, 86, [22] Zhou, M. J.; Nemade, P. R.; Lu, X. Y.; Zeng, X. H.; Hatakeyama, E. S.; Noble, R. D.; Gin, D. L. "New type of membrane material for water desalination based on a crosslinked bicontinuous cubic lyotropic liquid crystal assembly." J. Am. Chem. Soc. 2007, 129, [23] Hatakeyama, E. S.; Wiesenauer, B. R.; Gabriel, C. J.; Noble, R. D.; Gin, D. L. "Nanoporous, Bicontinuous Cubic Lyotropic Liquid Crystal Networks via Polymerizable Gemini Ammonium Surfactants." Chem. Mater. 2010, 22,

117 [24] Hatakeyama, E. S.; Gabriel, C. J.; Wiesenauer, B. R.; Lohr, J. L.; Zhou, M.; Noble, R. D.; Gin, D. L. "Water filtration performance of a lyotropic liquid crystal polymer membrane with uniform, sub-1-nm pores." J. Membr. Sci. 2011, 366, [25] Tiddy, G. J. T. "Surfactant-water liquid crystal phases." Phys. Rep. 1980, 57, [26] Carter, B. M.; Wiesenauer, B. R.; Hatakeyama, E. S.; Barton, J. L.; Noble, R. D.; Gin, D. L. "Glycerol-Based Bicontinuous Cubic Lyotropic Liquid Crystal Monomer System for the Fabrication of Thin-Film Membranes with Uniform Nanopores." Chem. Mater. 2012, 24, [27] Awtrey, A. D.; Connick, R. E. "The Absorption Spectra of I2, I3 -, I -, IO3 -, S4O6-2 and S2O3-2. Heat of the Reaction I3 - = I2 + I." J. Am. Chem. Soc. 1951, 73, [28] Harris, D. C. Quantitative Chemical Analysis; 2 nd ed.; W. H. Freeman and Company: New York,

118 CHAPTER 4 Design and synthesis of novel, cross-linkable, zwitterionic amphiphiles for the formation of cross-linked lyotropic liquid crystal assemblies Abstract The synthesis of intrinsically cross-linkable gemini zwitterionic surfactants is described. These amphiphilic monomers contain benzimidazolium cationic headgroups with covalently tethered anionic sulfonate groups and cross-linkable alkyl-1,3-diene tails. Preliminary investigation of the phase behavior of these amphiphilic monomers with polarized light microscopy showed these monomers are capable of forming lyotropic liquid crystal phases in water and glycerol Introduction Cross-linked lyotropic liquid crystal (LLC) assemblies are ordered, nanoporous polymer materials that contain periodic, uniform-size, nanometer-scale solvent domains/pores [1-3]. These polymer materials are formed by the in situ cross-linking of reactive amphiphiles (i.e., surfactants) that self-assemble into well-known ordered LLC morphologies such as lamellar (L), hexagonal (H), and bicontinuous cubic (Q) phases [4]. The solvent domains of these different geometries are lined by the amphiphile head-groups, and thus, the functionality or charge of the pore walls can readily be modified [1]. Because of these features, LLC networks have been shown to be useful for a number of important applications including heterogeneous catalysis [1], molecular-size-based membrane separations [5], and enhanced ion transport [6,7]. Q phase 102

119 networks are particularly interesting for transport-based applications since there is no need for pore alignment due to their 3D cubic symmetry. Recently, our research group showed that a new glycerol-containing LLC monomer system can be utilized to fabricate thin-film composite (TFC) membranes with an active separation layer containing a cross-linked type I bicontinuous cubic (QI) phase morphology [8,9]. Due to the low volatility of the glycerol used as the LLC phase-forming liquid medium, minimal evaporative loss occurs during solution-based thin-film processing. This allows a thin film of a cross-linked QI phase to be prepared by dissolving the monomer and glycerol in methanol and then roll-casting the solution over the top of a porous asymmetric membrane support. The methanol can then be removed by mild heating without any evaporative loss glycerol, and the monomer/glycerol mixture can subsequently be cross-linked at 70 C to form a thin film of the cross-linked QI-phase. This new glycerol-based QI-phase mixture was formed with a gemini imidazolium-based, cross-linkable LLC monomer (3). This gemini imidazolium-based monomer produces a QI-phase pore network that is lined with tethered cationic imidazolium moieties and associated free mobile anions. These cationic TFC QI membranes have molecular sieving capabilities, high salt rejection, and a water permeance approaching RO membranes [8,9]. Additional filtration studies on these TFC QI membranes also revealed that anion-exchange in the nanopores significantly alters the permeability of the membranes with little to no change in the rejection performance [9]. Further study is needed to acquire a fundamental understanding of this phenomenon; however, these initial results clearly demonstrate that anion-exchange can readily produce unwanted changes in water permeability depending on the nature of the salt ions in the feed. To avoid this potential problem in water purification applications, it is desirable to have charged TFC QI membranes that will readily reject salts but also maintain stable 103

120 performance regardless of the ions present in the feed. The design of such membranes will require new cross-linkable LLC monomers that produce charged QI networks that are not susceptible to ion-exchange. Zwitterions are electrically neutral compounds that contain a covalently tethered cation and a tethered anion [10]. Because the ionic substituents in zwitterionic molecules are all covalently linked, they do not undergo ion-exchange. Zwitterions are known for their low fouling properties [11-13], and many biological lipids are zwitterions [12,14].There are a number of reports using polymerizable zwitterionic amphiphiles to form stabilized LLC assemblies [2,15]. Unfortunately, most of these systems are based on polymerizable derivatives of naturally occurring phospholipids that require elaborate syntheses and expensive starting materials and/or mixtures of the reactive zwitterionic surfactants are required to obtain the desired phases [16]. There would be significant value in the design of new cross-linkable zwitterionic amphiphiles that can readily afford various cross-linked LLC assemblies in water or other non-aqueous solvents, since these materials would not undergo ion-exchange, and they have an intrinsic low fouling potential that could offer a number of advantages over conventional cationic or anionic LLC networks for various applications. Herein, we present the design and synthesis of novel, cross-linkable, zwitterionic amphiphilic monomers based on gemini surfactants that contain benzimidazolium cationic headgroups with covalently tethered anionic sulfonate groups. A preliminary investigation of the phase behavior of these monomers show they are capable of forming LLC phases in water and glycerol. 104

121 4.2. Results and discussion Monomer design and synthesis Designing a cross-linkable zwitterionic amphiphilic monomer requires that at least two chain-addition polymerizable moieties are incorporated into the molecule. The general approach that our research group has utilized in the past to create intrinsically cross-linkable monomers is to synthesize gemini surfactants [8,17,18]. Gemini surfactants consist of two conventional single-tail amphiphiles connected covalently at the headgroups by a spacer [17]. As seen in Figure 4.1, the current gemini imidazolium LLC monomer 3 used for TFC QI membrane fabrication contains two polymerizable tails and a headgroup spacer linked together by the nitrogen atoms in the imidazolium ring. The synthesis of this monomer only allows the spacer and the tails to be readily modified. An additional point of functionality is required on each cationic imidazolium unit in the headgroup to link together the two polymerizable tails and incorporate the zwitterionic character. Our approach to make cross-linkable zwitterionic monomers was to synthesize a headgroup that is similar to the current gemini imidazolium monomer 3 that includes an additional point of functionality that will allow the connection of tethered anionic groups to afford a zwitterionic configuration. 105

122 Figure 4.1. Synthesis scheme for gemini imidazolium LLC monomer 3 that form QI phases with water and glycerol. In order to design a gemini headgroup system with an additional available functional attachment point on the imidazolium rings for tethered anions, it was initially thought make a new bis(imidazole) headgroup precursor that connected the two imidazole units at their C2 positions on the rings with an alkyl spacer. This would allow one N atom on the imidazolium rings to be used for polymerizable tail attachment, and the other N atom to be used for tethering the associated counterion (Figure 4.2). Unfortunately, this approach was problematic. The use of n-butyllithium is required to link together two imidazole units at the C2 position and the synthesis of this headgroup would require protecting group chemistry and additional synthetic steps [19]. Attachment of the diene tails in any step prior to linking together two imidazole or imidazolium units is not possible since n-butyllithium will initiate polymerization of the diene tails. As a result, a more simplistic synthetic approach was sought. 106

123 Figure 4.2. Original design concept for novel, cross-linkable, gemini zwitterionic amphiphiles. A more facile and easily synthesized C2-C2 connected gemini imidazolium platform that allows N-atom connections of tails and tethered anions at both rings is a gemini bis(benzimidazolium) system. As seen in Figure 4.3, the reaction of o-phenylenediamine with diacids affords bis(benzimidazolyl) alkane compounds [20,21]. These bis(benzimidazolyl) alkanes can be readily alkylated with ω-bromoalkyl-1,3-diene tails in high yield. Reaction of the alkylated bis(benzimidazolyl) alkanes with 1,3-propanesultone then affords the desired zwitterionic monomers. Two new monomers (13, 14) containing different spacer lengths (x = 3 and 5) were synthesized in order to examine the potential of these new zwitterionic monomers to form LLC phases. The structure and chemical purity of these two gemini zwitterionic monomers synthesized as shown in Figure 4.2 were confirmed by 1 H and 13 C NMR analyses, and HRMS (see the Experimental). 107

124 Figure 4.3. General synthesis scheme for new gemini zwitterionic amphiphilic monomers 13 and LLC phase behavior of monomers 13 and 14 Monomers 13 and 14 did not have melting points below 95 C, so it was not possible to perform polarized light microscopy (PLM) penetration scans to quickly and qualitatively screen what LLC phases can be formed by these monomers with specific solvents over a range of temperatures [22]. As a result, partial phase diagrams of monomers 13 and 14 with water or glycerol were systematically mapped out as a function of different system composition using PLM to determine the potential of these monomers to form LLC phases. Primary focus was put 108

125 on the phase behavior of these monomers with water, since this is the traditional solvent studied with LLCs. A preliminary investigation of the phase behavior of monomer 13 was performed with water. As seen in Table 4.1, three compositions at ca. 90, 80 and 50 wt % monomer were examined. Out of the three compositions surveyed, only a mixture at ca. 80 wt % monomer formed a LLC phase with a mosaic-type PLM image (Figure 4.4a). This LLC phase existed from room temperature to 95 C. While the preliminary investigation of monomer 13 does not show elaborate LLC phase behavior, this monomer is capable of forming a LLC phase. A more complete phase diagram is needed to acquire a more comprehensive understanding of the LLC phase behavior of this monomer with water. 109

126 Table 4.1. Solvents and compositions investigated with monomers 13 and 14 from room temperature to 95 C for LLC behavior using PLM. Monomer Solvent Composition (wt % monomer) LLC behavior 13 H2O 90.2 Crystalline 13 H2O 81.2 LLC phase R.T. to 95 C 13 H2O 50.2 Crystalline 14 H2O 90.3 LLC phase R.T. to 95 C 14 H2O 80.0 LLC phase R.T. to 95 C 14 H2O 69.9 LLC phase R.T. to 95 C 14 H2O 60.3 LLC phase R.T. to 95 C 14 H2O 49.6 LLC phase R.T. to 95 C 14 Glycerol 89.9 LLC phase R.T. to 95 C 110

127 Figure 4.4. Representative PLM optical textures of LLC mixtures: (a) LLC phase consisting of 81.2/18.8 (w/w) 13/H2O at R.T. after annealing (b) LLC phase consisting of 90.3/9.7 (w/w) 14/H2O at R.T. upon mixing (c) LLC phase consisting of 69.9/30.1 (w/w) 14/H2O at R.T. upon mixing (d) LLC phase consisting of 60.3/39.7 (w/w) 14/H2O at R.T. after annealing (e) LLC phase consisting of 49.6/50.4 (w/w) 14/H2O at R.T. after annealing (f) LLC phase consisting of 89.9/10.1 (w/w) 14/glycerol at R.T. upon mixing. A more detailed study of the phase behavior of monomer 14 with water was completed. Five compositions of monomer 14 at ca. 90, 80, 70, 60, and 50 wt % monomer were investigated. As shown in Table 4.1, all five compositions investigated showed LLC phases by 111