SUPRAMOLECULAR APPROACHES TO SELECTIVE MERCURY CATION BINDING AND CRYSTAL ENGINEERING OF COVALENT CRYSTALLINE MATERIALS By Karrie M. Manes A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Chemistry 2011 ABSTRACT SUPRAMOLECULAR APPROACHES TO SELECTIVE MERCURY CATION BINDING AND CRYSTAL ENGINEERING OF COVALENT CRYSTALLINE MATERIALS By Karrie M. Manes The first part of this dissertation focuses on the preparation, characterization and 2+ Hg binding efficiency of a microparticulate sorbent based on a novel p-xylyl- peraza[2.2.2]cryptand-based polymer. The polymer was characterized in terms of surface morphology, size distribution and porosity. Sorption kinetics were measured and fitted to a homogeneous surface diffusion model whereby intraparticle diffusion parameters were determined. In order to gain insight into the mechanism(s) of binding, adsorption isotherms were constructed and sorbing capacity and specificity of a similar but nonmacrocycle-containing polymer were investigated. Desorption studies were also carried out and the regenerability of the sorbent was demonstrated. The mercury capacity and selectivity of the sorbent in the presence of calcium ions was compared to the nonmacrocycle-based polymer as well as a commercially available ion-exchange resin used for mercury removal. The second part of this dissertation involves the use of non-covalent intermolecular interactions in the purposeful design of dihydrogen-bonded (hydridic-toprotonic hydrogen bonding) crystalline molecular solids in the pursuit of extended covalent crystalline materials. The field of crystal engineering seeks to understand and utilize the intermolecular interactions that organize molecular crystals. The properties of covalent solids depend strongly on their crystallinity. Topochemical transformation of dihydrogen bonds to covalent bonds (M-H…H-X → M-X + H2) in crystalline solids potentially merges these areas by converting molecular to covalent crystals. Retaining the long-range order through this type of transformation faces two challenges: (1) geometry change on M-X bond formation and (2) gas release within the lattice. In dihydrogenbonded salts where either the cationic or anionic partner is much larger than the other, the packing of the large partners determines the lattice dimensions. These structures can tolerate bond reorganization and gas release, maintaining their crystalline order. In an effort to extend previous demonstrations of this approach based on large cations and simple borohydride anions, the present effort expands the range of large cations studied and explores the use of bulky 2-substituted benzimidazole boranes, zwitterionic structures that contain both the bulky cation and hydridic partner within the same molecule. DEDICATION This dissertation is dedication to my family and friends who have supported me through my 12 years of post-secondary education. I am forever in your debt. iv ACKNOWLEDGEMENTS This work would not have been possible without the support and assistance of many dedicated individuals both inside and outside the university. First and foremost I would like to express my personal and professional gratitude to my advisor Dr. James ‘Ned’ Jackson who supervised the projects reported in this dissertation with invaluable enthusiasm. His excellent advice and insightful, constructive criticism were of the utmost significance over the course of my studies at MSU. My experience would not have been the same without him. I am also incredible indebted to Dr. Richard Staples for the countless hours of help and instruction obtaining X-ray data, solving structures (or not solving them as the case may be) and excellent advice, as well as Dr. Daniel Holmes for his patience and willingness to try new NMR experiments even when they didn’t prove fruitful. My thanks also go out to Dr. Volodymyr Tarabara, Dr. Julian Taurozzi and Dr. Misha Redko for their roles in the mercury cation binding project. I appreciate the environmental engineering perspective I gained throughout the duration of that project. Also, thanks go out to Dr. Lars Peereboom and Ambareesh Murkute for BET measurements, Dr. Kathy Severin for help and instruction with flame AA techniques and Dr. Kirk Stuart from the Animal Toxicology Lab at MSU for running cold vapor AA. Many other people have played a personal role in my success, including Dr. Simona Marincean, Dr. Partha Nandi, Dr. Kevin Walker, Mr. Brandon Dutcher and Mr. Matthew Nethercott all of whom were excellent sources for help and advice throughout my time at MSU. Also, many thanks go to Dr. Gary Blanchard and Dr. Greg Baker for their help and support v throughout my time in the graduate program. I’d also like to thank Dr. William Wulff and Dr. Mitch Smith for agreeing to be committee members. Finally I’d like to thank my friends and family who have had to put up with many things during my time in graduate school, including, but not limited to, foul moods, middle of the night phone calls, food runs and coffee runs, severely delayed emails/text/return phone calls, along with missed birthdays holidays and other various celebrations. Your love and support ultimately made this journey possible. I am eternally grateful. vi TABLE OF CONTENTS LIST OF TABLES ix LIST OF FIGURES x LIST OF SCHEMES xiv CHAPTER 1 INTRODUCTION 1.1 The Chelate, Macrocyclic and Macrobicyclic Effects 1.2 Cryptands and Mercury Cation Binding 1 4 8 CHAPTER 2 ENVIRONMENTAL MERCURY REMEDIATION 2.1 Mercury Remediation 2.2 Mercury Remediation – Current Industrial Technology 2.3 Mercury Remediation – Current Areas of Research 2.3.1 Polyaza- and polythio-macrocycles 2+ 2.4 Using Peraza[2.2.2]cryptand to Selectively Capture Hg Cations 2.4.1 Polymer Preparation & Characterization 2.4.2 Size and Morphology of the Polymer 2.4.3 Adsorption Kinetics Measurements 2.4.4 Adsorption Isotherm Measurements 2.4.5 Adsorption Isotherm Modeling 2.4.6 Mercury Adsorption in the Presence of a Competing Divalent Ion and Comparison to a Commercially Available Material 2.4.7 Regeneration Study 2.4.8 Comparison to a Non-macrocycle-containing Cross-linked Polymer CHAPTER 3 DIHYDROGEN BONDING 3.1 Dihydrogen Bonding and Crystal Engineering 3.2 Applications of Extended Covalent Crystalline Solids – Hydrogen Storage via Metal Borohydrides and Amines 3.3 Crystal Engineering Using Various Dihydrogen bonding donors and acceptors 3.3.1 Preparation and X-ray Crystal Structures of + [Na@peraza[2.2.2]cryptand] BH4 3.3.2 Sodium [µ-(cyano-κC:κN]hexahydrodiborate 3.3.3 2-Substituted Benzimidazole Boranes 3.3.4 Other Amine-based and Hydroxyl-based Dihydrogen Bond Acceptors 3.3.5 Computational Study vii 17 18 20 23 34 40 41 43 45 49 51 56 58 61 66 67 73 79 79 89 91 99 104 CHAPTER 4 EXPERIMENTAL SECTION 4.1 Mercury Binding Study 4.2 Dihydrogen Bonding Study 113 116 126 APPENDIX A X-RAY CRYSTAL STRUCTURE LATTICE PARAMETERS 132 APPENDIX B REPRODUCTION PERMISSIONS 135 REFERENCES 146 viii LIST OF TABLES Table 1.1 Stability Constants for Alkali and Alkaline Earth Metal-Cryptand Complexes 13 Table 1.2 Stability Constants of Transition Metal-Cryptand Complexes 13 Table 2.1 Summary of Environmental Standards for Mercury 19 Table 2.2 Summary of Current Mercury Removal Methods 22 Table 2.3 Polyethylene imine- and polypyridine-based Sorbents for Hg Adsorption. Table 2.4 Thiol Functionalized Sorbents for Hg 2+ 2+ Adsorption Table 2.5 Simulated Coal-Fired Scrubber Waste Effluent Composition Table 2.6 Ion-exchange polymers for Hg 2+ adsorption Table 2.7 Logarithms of the Stability Constants for the Formation of Complexes of (10), (11) and (12) Table 2.8 Protonation and Hg 2+ Stability Constants for Various Ligands 24 28 29 33 36 39 Table 2.9 N2 Adsorption of the p-Xylyl Cross-linked peraza[2.2.2]cryptand 45 Table 2.10 Sorbent Regeneration Using Na2S in a Packed Column 60 Table 3.1 Selected Extensively Studied Hydrogen Storage Materials 75 Table 3.2 Hydrogen bonding interactions within the cavity of 3+ [H2O@H3peraza[2.2.2]cryptand] 2I- NaI2 (H2O)3 85 - Table 3.3 Bond angles in NaI2 (H2O)3 clusters 86 Table 3.4 Selected Characterization Parameters for 2-Substitued Benzimidazole Boranes 94 Table 3.5 Calculated Dihydrogen Bond Distances and Stabilization Energies of the Hydrated Borohydride Anion 109 ix LIST OF FIGURES Figure 1.1 Examples of supramolecular hosts and representative X-ray crystal + structures containing K ions. For interpretation of the references to color in this and all other figures, the reader is referred to the electronic version of this dissertation. 3 Figure 1.2 The chelate, macrocyclic and macrobicyclic effects. 6 Figure 1.3 [2.2.2]cryptand and peraza[2.2.2]cryptand. 9 Figure 1.4 Cryptands used in toxic heavy metal binding studies by Lehn & Montavon. 10 Figure 1.5 Aza-macrocycles by Grondahl et al. Sar (8), (NH2)2-sar (9). 14 Figure 2.1 Polyaza ligands investigated as mercury scavengers by Garcia-España et al. 35 Figure 2.2 (A) ORTEP drawing of the neutral [Hg2(10)Cl4] complex. (B) ORTEP 58 drawing showing the coordination sites of both mercury atoms. Thermal ellipsoids are drawn at the 30% probability level. Reprinted with permission from ref 58. Copyright 1996 American Chemical Society. Figure 2.3 Mixed macrocyclic donors whose Hg Grondahl and coworkers. 2+ 37 complexes were studied by 38 Figure 2.4 p-Xylyl cross-linked peraza[2.2.2]cryptand sorbent. 41 Figure 2.5 Particle size distribution of p-xylyl cross-linked peraza[2.2.2.]cryptand polymer. 43 Figure 2.6 SEM images of the p-xylyl cross-linked polymer showing spherules within the porous aggregates. 44 2+ Figure 2.7 Adsorption kinetics of Hg adsorption by p-xylyl cross-linked peraza[2.2.2]cryptand polymer. 47 Figure 2.8 Adsorption isotherm measurements over a set of six independent experiments. The dashed line represents the theoretical adsorption capacity by weight. 50 x C e versus Ce and linear regression fit for the isotherm q adsorption data. 53 Figure 2.10 € Plot of log (q)versus log(Ce) and linear regression fit for the isotherm adsorption data. 54 Figure 2.11 Plot of adsorption isotherm experimental data and the curve obtained using the calculated Langmuir parameters. 55 Figure 2.9 Plot of Figure 2.12 2+ Hg sorption in the absence and in the presence of a six-fold excess 2+ of Ca by peraza[2.2.2]cryptand-based sorbent and a commercially available sorbent (commercial resin). 2+ Figure 2.13 Hg Figure 2.14 Hg 2+ uptake by PEI-based and peraza[2.2.2]cryptand-based polymers. 63 binding by PEI-based polymers in the presence of a six-fold 2+ excess of Ca , Figure 3.1 58 64 NaBH4·THEC complex showing dihydrogen bonding to a 80 borohydride anion. Adapted with permission from ref 80. Copyright 2000 Wiley VCH. Figure 3.2 Figure 3.3 71 The X-ray crystal structure and crystal packing of NaBH4·THEC. Adapted with permission ref 80. Copyright 2000 Wiley-VCH. 72 (a) Crystal packing of Li2(BH4)2 • NH3BH3. (b) Crystal packing of + Ca(BH4)2 • (NH3BH3)2. (c) Coordination environment of Li . 2+ (d) Coordination environment of Ca . Reprinted with permission from ref 86. Copyright 2010 Royal Society of Chemistry. Figure 3.4 77 TPD results for H2 release for Li2(BH4)2 • NH3BH3 and Ca(BH4)2 • 2NH3BH3 with 2°C/min heating rate to 450°C along with pure NH3BH3, Li2(BH4)2 and Ca(BH4)2. The gas released from Li2(BH4)2 • NH3BH3 and Ca(BH)4 • 2NH3BH3 (top) is normalized to nH2 g/mol NH3BH3. Reprinted with permission from reference 86. Copyright 2010 Royal Society of Chemistry. Figure 3.5 Illustration of a large globular cation (large spheres) occupying the bulk of the space in a crystal lattice leaving tetrahedral ‘holes’ xi 78 (small solid circles) for small anionic partners. 80 + - Figure 3.6 X-ray crystal structure of [Na@peraza[2.2.2]cryptand] BH4 • H2O. Figure 3.7 Crystal packing of [Na@peraza[2.2.2]cryptand] BH4 • H2O. Figure 3.8 ORTEP depiction of [H2O@H3peraza[2.2.2]cryptand] + - 82 - 3+ 93 2I 82 - • NaI2 (H2O)3. Figure 3.9 83 ORTEP representation of hydrogen bonding interaction within the 3+ cavity of [H2O@H3peraza[2.2.2]cryptand] 2I NaI2 (H2O)3. 84 - Figure 3.10 ORTEP depiction of NaI2 (H2O)3 clusters. Figure 3.11 Crystal structures of water encapsulated by peraza[2.2.2]cryptand 92 4+ by Bowman-James et al. [H2O@H4peraza[2.2.2]cryptand] 4I • 4H2O (top left) and McKee et al. 93 86 [H2O@peraza[2.2.2]cryptand] 3+ - 3Cl •6H2O (top right). Reproduced with permission from references 92 & 93. Copyright 2002 Elsevier of Crystallography. cryptand] 3+ - 93 - 92 and Copyright 2002 International Union Crystal structure of [H2O@H3peraza[2.2.2] 2I NaI2 (H2O)3 for comparison (bottom). 87 + Figure 3.12 ORTEP representation of [Na@peraza[2.2.2]cryptand] BH4- formed by mixing 1:1 ratio of peraza[2.2.2]cryptand with NaBH4 (left) and by mixing a 1:1 ratio of peraza[2.2.2]cryptand with sodium Sodium [µ-(cyano-κC:κN]hexahydrodiborate (right). 89 Figure 3.13 ORTEP representation of the single crystal formed by mixing peraza[2.2.2]cryptand with NaBH4 in a 1:1 ratio (top). ORTEP representation of the single crystal formed by mixing peraza[2.2.2]cryptand with p-1,4-bis(benzimidazol-2-yl)benzene borane in a 1:1 ratio (bottom). 95 Figure 3.14 ORTEP depiction of 2-methylbenzimidazole borane. 97 Figure 3.15 Crystal packing of 2-methylbenzimidazole borane into 2-D sheets showing close N-H---H-B distances of 1.758Å. 98 Figure 3.16 Powder X-ray diffraction data of (21b) between 60°C and 130°C. 99 Figure 3.17 Commercially available (29 – 34) and synthesized (35 – 36) xii dihydrogen bond acceptors. 100 Figure 3.18 ORTEP representation of (37). 102 Figure 3.19 ORTEP representation of the crystal packing in (36). Figure 3.20 Optimized geometry (DFT) for Cl (H2O)3. Reprinted with permission. Copyright 1995 American Chemical Society. Figure 3.21 Figure 3.22 103 - 103 113 Graphical representation of the lowest energy conformations of nBH4 where n = 1 – 4. Graphical representation of the lowest energy conformations of nBH4 where n = 5 – 6. 108 - 110 - 111 Figure AB-1 Reproduction permission letter for reference 58. 136 Figure AB-2 Reproduction permission letter for reference 76. 137 Figure AB-3 Reproduction permission letter for reference 80. 130 Figure AB-4 Reproduction permission letter for reference 86. 139-140 Figure AB-5 Reproduction permission letter for reference 92. 141-142 Figure AB-6 Reproduction permission letter for reference 93. 143-144 Figure AB-7 Reproduction permission letter for reference 114. xiii 145 LIST OF SCHEMES Scheme 1.1 Representation of the formation of inclusion metal-cryptate complexes. 11 Scheme 2.1 Synthesis of p-xylyl cross-linked peraza[2.2.2]cryptand polymer (17). 42 Scheme 2.2 Synthesis of polyethylene imine polymer cross-linked with p-xylyl units. 61 Scheme 3.1 Closed-loop strategy for topochemical control of H2 loss in 79 dihydrogen-bonded systems. Adapted with permission from reference 79. Copyright 2001 American Chemical Society. Scheme 3.2 69 Divergent strategy for topochemical control of H2 loss in 79 dihydrogen-bonded systems. Adapted with permission from reference 79. Copyright 2001 American Chemical Society. Scheme 3.3 Scheme 3.4 69 Formation of dihydrogen-bonded (only one borohydride/macrocycle shown for clarity) peraza[2.2.2]cryptand and loss of H2. 81 Synthesis of Sodium [µ-(cyano-κC:κN]hexahydrodiborate by Wade et al. Scheme 3.5 Scheme 3.6 94 89 Possible disproportonation of sodium [µ-(cyano-κC:κN] hexahydrodiborate 90 99,100 General synthesis of 2-substitued benzimidazoles (21a – 25a), 101 their borane derivatives, (21b – 25b) and diphenylamine borane (26). For (27) the 2 equivalents of o-phenylenediamine react on each side of a p-terephthalic acid. 92 100 Scheme 3.7 Synthesis of 1,3,5-tris(benzimidazol-2-yl)benzene. Scheme 3.8 Synthesis of tri-substituted triazine adapted from Dilliban et al. Scheme 3.9 Disubstituted triazine resulting from the adapted synthesis of 102 Dilliban et al. 93 102 101 101 Scheme 3.10 Synthesis of 2,6-bis(N,N-bis-2-hydroxyethylamino)-4-(N-phenylamino)1,3,5-triazine (39). 104 xiv CHAPTER 1 INTRODUCTION 1 Supramolecular chemistry has been defined by Jean-Marie Lehn as the ‘chemistry of 1 molecular assemblies and of the intermolecular bond’. Lehn, along with Donald Cram and Charles Pederson, won the Nobel Prize in Chemistry for their ground-breaking work in this area in 1987. Supramolecular chemistry is often thought of as the ‘chemistry beyond the molecule’ or the ‘chemistry of non-covalent interactions; terms such as ‘host-guest chemistry’, ‘molecular 1 recognition’ and ‘self-assembly’ are also often used. In this context, however one chooses to describe the field, it is certain that over the last several decades it has significantly contributed to understanding and manipulating the subtleties of intermolecular relationships. Though supramolecular chemistry is considered a relatively young discipline, an early example of a chemical system defined by its non-covalent interactions can be dated back to 1811 when Sir Humphrey Davy discovered chlorine hydrate (chlorine trapped in the solid matrix of hydrogenbonded water). It wasn’t until 1823, however, that the molecular formula (Cl2 • 10H2O) was reported for chlorine hydrate by one of Humphrey’s students, Michael Faraday. Later studies found the actual composition to be closer to Cl2 • 8H2O but water’s remarkable affinity for itself had been noticed, without any understanding if the mechanism. It was in 1936 that Maurice 2 Huggins published two papers regarding hydrogen bonding in ice and liquid water and in 3 organic compounds when hydrogen bonds were finally seen as a structurally and energetically defined phenomenon. The importance of hydrogen bonding was further recognized when, in the 1950’s, Linus Pauling, James Watson, Francis Crick and others were modeling the structure of DNA using X-ray crystallography. Although it has been over 70 years since they were recognized, hydrogen bonding interactions remain a central theme in supramolecular research. 2 O O O O OH O O O O OH OH OH N N O O O O O O O O O O N O OH O OH OH N OH O O O O . Calixarene Corand Cryptand Corand Calixarene Cryptand (18-crown-6) (p-tert-butylcalix-4-arene) (p-tert-butylcalix-4-arene) ([2.2.2.]cryptand) (18-crown-6) ([2.2.2.]cryptand) Figure 1.1 Examples of supramolecular hosts and representative X-ray crystal structures + 4-6 containing K ions. For interpretation of the references to color in this and all other figures, the reader is referred to the electronic version of this dissertation. Several classes of molecules have become most notable for their use as supramolecules or ‘hosts’. Some examples include crown ethers, calixarenes and cryptands (Fig. 1.1). 3 Cryptands, which are polycyclic multi-dentate ligands, are of significant interest in our labs and were first introduced by Jean-Marie Lehn in the late 1960’s. Common supramolecular interactions include ion-ion, ion-dipole, dipole-dipole, hydrogen bonding and more recently, dihydrogen bonding (hydridic-to-protonic hydrogen bonding). These types of stabilizing interactions can range from 1 (i.e. O-H⋅⋅⋅π interactions) to 1 approximately 30 kcal/mol (i.e. gas phase dimers of strong acids) per occurance, but are commonly considered to be within the range of 2 -7 kcal/mol. The three interactions that are exploited in this work are ion-dipole, ion-pairing and dihydrogen bonding. The research described in this dissertation uses several forms of spectroscopy, X-ray crystallography and theoretical calculations to understand two very diverse chemical systems. Described first is the use of peraza[2.2.2]cryptand in a cross-linked polymer for selective Hg 2+ ion binding. This study is followed by the use of dihydrogen bonding (hydridic-to-protonic hydrogen bonding) in the pursuit of rational designs leading to covalent crystalline solids. 1.1 The Chelate, Macrocyclic and Macrobicyclic Effects Since supramolecular interactions are usually far weaker than covalent interactions, the design of complexes uses the idea of preorganizing multiple sites for non-covalent interactions. When summed, these interactions give greater overall stabilization to the complex. This extra stabilization is based on the chelate, macrocyclic and macrobicyclic effects. 4 Ligands, in coordination chemistry, are ions or molecules that bind to a central metal atom to form a coordination complex, and a ligand’s ‘denticity’ refers to the number of atoms in a single ligand that bind to the central atom in that complex. For example, ammonia, NH3, would be a monodentate ligand, whereas ethylenediamine, H2NCH2CH2NH2, would be bidentate. The term chelation is used when a ligand forms coordinate bonds with a central atom. The word ‘chelation’ is derived from the Greek chelé, meaning, “claw” like the claws of a lobster. All polydentate ligands exhibit what is termed the chelate effect. The macrocyclic and macrobicyclic effects are terms also used in relation to the chelation effect and are used to describe the relative increase in binding selectivity of chelating ligands with increasing preorganzation of the binding cavity in the ligand. In general, as the preorganization of a binding cavity in the host increases so does the binding selectivity. This is due to the fact that the cavity in which the guest can reside becomes increasingly pre-defined and therefore the host molecule does not undergo a significant conformational change upon guest binding (Fig. 1.2). Not taking into account the effects of solvation, the process of host-guest binding can be divided into two conceptual stages. Activation is the stage in which the host undergoes a conformational adjustment in order to accommodate the binding needs of the guest while minimizing the unfavorable interactions among the chelating sites on the host. This stage is energetically unfavorable while the host remains in the host-guest conformation since its energy of distortion has not been repaid. 5 H2N H N N H O Mn+ N H Chelate effect and Macrocyclic effect NH2 O O Mn+ O O O N NH2 Podand (triethylenetetraamine) O NH2 H Chelate effect O Corand (12-crown-4) N O O O O O O Chelate effect and Macrobicyclic effect N N O O O Mn+ O O O N Cryptand ([2.2.2]cryptand) Figure 1.2 The chelate, macrocyclic and macrobicyclic effects. Upon binding, the ligand remains in the energetically unfavorable conformation, therefore the energy of its distortion must be compensated by binding, the stage that follows, which is energetically favorable due to the enthalpically stabilizing attraction between complementary binding sites of the host and guest. The overall free energy of complexation is the difference in energy between the two stages, when solvation is ignored. When the 6 reorganization energy is large the overall free energy of complexation may be small and the complex is destabilized. Conversely, if the host is easily preorganized (i.e. the reorganization energy is minimal) and the binding is strong, the overall free energy of complexation is large and therefore the complex is stable. Of course solvation can also play a significant role in the complex formation process, especially when stronger supramolecular interactions such as hydrogen bonding are prevalent. During the activation step, in addition to conformational adjustments of the ligand, solvent reorganization must also take place, which may come at a high entropic cost depending on the extent of the reorganization required. However, upon binding, solvent molecules are released from the central atom as well as from the ligand’s chelating sites, each having been replaced by coordinate bonds, causing an overall increase in entropy and thus tipping the free energy of binding in a favorable direction. Regardless of the energetically favorable or unfavorable status of a preorganized host, the act of binding a guest can be kinetically difficult. Rigidly preorganized hosts can have significant difficulty in passing through a complexation transition state and therefore tend to exhibit slow binding kinetics. More conformationally mobile hosts are able to adjust more rapidly to changing conditions so both complexation and decomplexation are rapid. The least pre-organized class of chelating ligands are the podands, which are flexible, acyclic multi-dentate molecules such as polyethylene imine chains. This class of ligand requires the greatest amount of reorganization energy to wrap itself around a central guest. Corands, such 7 as the crown ethers, have a higher degree of preorganization due to the cavity formed by their cyclic shape. The size of the cavity formed determines the selectivity for a specific guest. For + example, it is well known that 18-crown-6 is extremely selective for K ions. While there is some degree of reorganization required for the crown to bind the metal guest, it is comparatively less than what is required by a podand ligand. Finally, the class of ligands with the highest degree of preorganization are the cryptands as in [2.2.2]cryptand. The bicyclic nature of this type of ligand further decreases the amount of reorganization energy required to encapsulate a guest when compared to podands and corands. 1.2 Cryptands and Mercury Cation Binding The all-nitrogen macrobicyclic complexant 1,4,7,10,13,16,21,24 otaazabicyclo[8.8.8]hexacosane, herein known simply as peraza[2.2.2]cryptand was first 7 synthesized as the free ligand (Fig. 1.3) by Lehn and coworkers in 1989. Since then several 8 syntheses have been explored and in 2005 a one-pot preparation was developed by Redko et al. This streamlined synthesis provides an easy route to this very useful supramolecular building block. Since its first synthesis this ligand and several analogs have been used to encapsulate + + 2+ 2+ 2+ 2+ + 2+ + 2+ metal cations including Na , K , Mn , Ni , Cu , Zn , Ag , Cd , Ba , Pb and Hg 2+ 9,10 . This range illustrates the cryptand’s flexible nature in that it has the ability to accommodate ions 2+ 9 + 11 varying in radius from 0.78Å (Ni ) to 1.33Å (K ) . 8 N N O O O NH HN HN O O O NH HN HN N N (1) IUPAC Name "[2.2.2]cryptand" IUPAC Name "peraza[2.2.2]cryptand" Figure 1.3 [2.2.2]cryptand and peraza[2.2.2]cryptand. To ensure selectivity of one cation over another the number and geometry of stabilizing interactions must be maximized. Therefore increasing the number of interactions between the nitrogen donors and cationic acceptor may play an important role in selectivity. Assuming a 6+ coordinate environment, Na and Hg 2+ have approximately the same ionic radius (1.02Å). + - 12 Since the crystal structure of the [Na@peraza[2.2.2]cryptand] Cl is known, based on ionic size about the way Hg 2+ some predictions may bind inside the cryptand cage may be made. Other factors such as differences in ‘hardness’ between Na+ and Hg 2+ may play a role as well. In the 1970’s Lehn and coworkers published a series of papers concerned with cation binding by polymacrocycles. One paper in particular studied the stability and selectivity of toxic 9 heavy metal cation inclusion complexes of polyoxa- and polyaza-macrobicyclic ligands. In this 9 study, the metal-ligand complex stability constants as well as ion selectivity of six mixed oxygen- and nitrogen-containing cryptands (Figure 1.4.) were investigated. N O N O O O O O O N H3C N N O O N H 3C O O O O N H 3C N O N O N H3C N H 3C N N H3C O N CH3 3 N CHO O N O N CH 3 N CH 3 N O O N 3 N O 4 (5) O O O O O O O O O O O N O N (7)6 5 (6) 4 N O N N CH3 (4) O O N 6 5 Figure 1.4 Cryptands used in toxic heavy metal binding studies by Lehn & Montavon. 9 All of these ligands are considered inclusion cryptands meaning binding of metal cations would take place inside the cavity of the ligand as seen below in Scheme 1.2. Using potentiometric titrations the stability constants of each ligand with alkali, alkaline earth and several transition 2+ 2+ metals including toxic metals such as Hg , Cd 2+ and Pb were determined. The stability constants for the metal-ligand complexes are shown in Tables 1.1 and 1.2. Upon examining 10 CH3 3 O O N N N N O NO N H3C 2 (3) 2 N O N CH3 N O N CH3 N O N H3C N 1 (2) 1 N O N O H 3C H3C these data several trends can be seen with respect to O vs. N binding of alkali and alkaline earth metals vs. transition metals. In the case of alkali and alkaline earth metals, in general, the stability constants decrease with sequential replacement of nitrogen for oxygen. n+ N N R R O N N + O N R O O N O O O Mn+ O R N N Scheme 1.2 Representation of the formation of inclusion metal-cryptate complexes. This trend, according to the authors, is thought to be for two main reasons: First, the replacement of O by N-CH3 reduces the electrostatic interaction between the cation and the ligand. Nitrogen has a smaller dipole moment than that of oxygen and is therefore a ‘softer’ Lewis base/donor, whereas the alkali and alkaline earth metals are generally considered ‘hard’ Lewis acids/acceptors. Secondly, the larger van der Waals radius of N (1.5Å) compared to O (1.4Å) as well as longer C-N bond length (~1.50Å) compared to C-O (~1.43Å) reduces the size of the intramolecular cavity. Next, molecular models show that the methyl groups of the N-CH3 moieties increase the ‘thickness’ of the ligand, shielding the cavity (and the complexed cation) from stabilizing solvent interactions and thus lowering the overall stability of the complexes. Finally, the differences in the hydration of nitrogen and oxygen atoms in the ligands may play a part in lowering the stability of the complexes when comparing O to NH to N-CH3 sites. Free 11 energies of ligand hydration follow the sequence Et2O < Et3N 4.0 > 4.0 > 4.0 4.0 3.8 3.5 > 3.8 3.2 2.5 < 1.0 > 5.0 4.2 - Ligand <6 2.0 2.6 1.8 3.9 > 8.0 Ligand 7 5.5 > 6.0 7.58 3.2 6.1 + 1.33 4.2 2.7 1.7 5.4 < 2.0 K + > 5.0 4.2 > 8.0 2.3 K (M) + 1.49 3.0 2.3 4.3 < 2.0 Rb + > 4.0 > 4.0 > 6.0 1.9 Rb (M) + 1.65 < 2.0 < 2.0 < 2.0 < 2.0 Cs + 3.8 3.3 4.4 < 2.0 Cs (M) 2+ 0.78 1.9 2.6 2.4 < 2.0 ~ 2.5 Mg 2+ 1.06 4.6 4.3 1.5 2.2 4.4 2.5 Ca 2+ 1.27 7.4 6.1 1.5 8.0 < 2.0 Sr 2+ 1.43 9.0 6.7 3.7 9.5 < 2.0 Ba *All measurements taken in aqueous solution unless otherwise stated. M = methanol, M/W = methanol/water 95:5. 9 Table 1.2 Stability Constants of Transition Metal-Cryptand Complexes. Ionic Ligand Ligand Ligand Ligand Ligand Cation Radius 2 3 4 5 6 + 1.13 10.8 11.5 13.0 12.7 9.6 Ag + 1.49 6.3 5.5 4.1 3.9 6.3 Tl 2+ 0.82 5.2 4.9 5.2 9.9 ≤ 2.0 Co 2+ 0.78 5.0 5.1 5.7 10.0 ≤ 3.5 Ni 2+ 0.92 9.7 12.7 12.5 16.0 6.8 Cu 2+ 0.83 6.3 6.0 6.8 11.2 ≤ 2.5 Zn 2+ 1.03 9.7 12.0 10.7 12.4 7.1 Cd 2+ 1.12 21.7 24.9 26.1 26.6 18.2 Hg 2+ Pb 1.32 14.1 15.3 15.5 13 - 12.7 Ligand 7 8.5 ≤ 4.7 ≤ 4.5 7.8 ≤ 5.3 ≤ 5.5 7.9 2+ Further evidence for the stability of Hg -azacrpytand complexes is found in a study by Grondahl and coworkers. 13 This study looked at the mercury complexes of two aza-based cryptands (Fig. 1.5) probing thermodynamic stability and complexation and decomplexation mechanisms. Using NMR spectroscopy and UV-spectrophotometry, equilibrium constants were calculated to be log K = 28.1 (8) and 26.4 (9). These constants imply very stable complexes and are comparable to values found for other mercury(II) amine complexes. In order to understand the mechanism of decomplexation of the metal-ligand complexes, binding competition experiments of [Hg(8)] 2+ - - using Cl , I and (9) as competitive binding ligands and [HgH2(9)] 4+ - - (bridgehead NH2 moieties are protonated) using Cl , I and (8) as competitors, as well as varying the pH, were performed. NH2 NH HN HN NH HN HN NH HN NH HN HN NH HN HN NH HN NH2 1 (8) (9) Figure 1.5 Aza-macrocycles by Grondahl et al. Sar (8), (NH2)2-sar (9). 14 3 2 13 In these experiments the mechanisms and rates of dissociation of Hg 2+ from each of the ligands - (8 and 9) were ascertained. In the presence of strongly coordinating I , both ligand-metal complexes are labile and liberation of Hg 2+ + which were studied independently, the dissociation of Hg + - occurs quickly. In the case of H and Cl addition, 2+ from (8) is clearly dependent on - 4+ [H ] and to a lesser extent on [Cl ] up to ~0.5 M. In [HgH2(9)] , the dependence of Hg - 2+ + dissociation on [Cl ] is present to a higher extent than that for [H ]. Using these data the authors made two main observations regarding Hg 2+ decomplexation mechanisms with respect to ligand structure: 1) For both ligand-metal complexes, when all secondary amine sites are coordinated to the metal center, they are effectively quaternary and not available for protonation. Therefore, in acid-assisted dissociation, protonation of the amine site must occur due to spontaneous Hg-N bond breaking. Once liberated, the secondary amine site can be protonated and becomes unavailable to recombine with the metal center allowing a competing ligand to take its place. Successive Hg-N bond ruptures and protonation eventually lead to metal liberation. 2) Since the Hg-N and Hg-water or Hg-Cl bonds are somewhat labile, the rates of dissociation and recombination are very fast. However, when a competing ligand is involved, the amount of time the amine site is free is relatively longer than the time it is not free because the binding site on the metal center has been occupied by the competing ligand and protonation is more likely. Thus the rate of metal dissociation is enhanced by the presence of a competing ligand. When comparing the [Hg(8)] 2+ complex with that of the [HgH2(9)] liberation due to competing ligand addition in [HgH2(9)] 15 4+ 4+ complex, the rate of Hg 2+ is affected to a greater extent. This is thought to be due to the protonation of the external primary amine sites first, making additional protonation of the secondary amine sites more difficult. So, without the competing ligand allowing the site to be free for a longer period of time, recombination of the Hg-N bond is faster than protonation of the free amine site. These data confirm mechanism that has been invoked in several other metal-amine complexes. 14 Given the results of both the Lehn et al. and Grondahl et al. studies it is clear that azabased cryptands are excellent contenders for selective and specific binding of Hg 2+ ions. Given this knowledge, a general idea about what types of cryptand ligands may be able to bind mercury most strongly, yet easily release mercury upon addition of acid or some other decomplexation agent such as chloride, iodide or a sulfur-containing ligand, can be obtained. 16 CHAPTER 2 ENVIRONMENTAL MERCURY REMEDIATION 17 2.1 Mercury Remediation Heavy metals pose serious health hazards when introduced into the food chain via human activity. 15 Among the many industrial heavy metal wastes, mercury can have a severe impact on the environment and is known to bio-accumulate and bio-magnify in aquatic life. 16 This metal has a tendency to bind to proteins and mainly affects the renal and nervous systems causing reproductive failure, damage to intestines, stomach disruption, DNA damage and kidney failure. 17 Elemental mercury in emissions from coal-burning power plants or waste incinerators can exist in the atmosphere for up to a year and is deposited in lakes and streams, in many cases far from the source, where it can be oxidized and converted to inorganic mercury salts or into organic forms by microorganisms. Compounds of mercury tend to be more toxic than the element itself, with organic derivatives being the most toxic. Methylmercury and mercury chloride are thought to cause cancer in humans and dimethyl mercury is a potent neurotoxin that is lethal in very small amounts. Exposure to mercury vapor can also result in brain damage. With the serious health hazards that result from mercury exposure it is clear that remediation of mercury from wastewater streams to benign levels is an important challenge. The United States Environmental Protection Agency regulates mercury in pesticides and mercury releases into the environment through air and water and sets land disposal limits. There are two main types of regulations that influence the overall picture of mercury in the environment, which include useor release-related regulations and environmental management standards. Use- or release-related regulations involve the costs and/or conditions associated with using and releasing mercury during production or disposal. These regulations encourage pollution prevention. 18 18 Table 2.1 Summary of Environmental Standards for Mercury Media Mercury Standard Ambient • 0.144 ppb for ingestion of both water Water and aquatic organisms • 0.146 ppb for ingestion of only aquatic organism • 2.4 ppb for freshwater acute exposure • 12 ppt for freshwater chronic exposure • 2.1 ppb for marine acute exposure • 25 ppt for marine chronic exposure Drinking • Maximum contaminant level = 2 ppb Water Air Sludge Compost • No ambient standard Limits: • 17 ppm (dry wt) and 17 kg/hectare cumulative loading for sludge applied on agricultural, forest and publicly accessible lands • 17 ppm (dry wt) and 0.85 kg/hectare annual loading rate for sludge sold or distributed for application to a lawn or home garden • 57 ppm (dry wt) for sludge sold or distributed for other types of land disposal • 100 parts per thousand (dry wt) for sludge disposed in lined or unlined facilities • No federal standards Fish • 1 ppm Groundwater Bottled Water Hazardous Waste • • Toxicity Characteristic Leaching Procedure (TCLP) = 0.2 ppm • Minnesota sets mercury concentration standards FDA action level for methylmercury 2 ppb 2 ppb • Explanation • Established by Clean Water Act § 304(a) • Ambient water criteria varies by state (may change with the Great Lakes Initiative) • Maximum contaminant level for mercury established under Safe Drinking Water Act • 19 • Land disposal prohibited unless leachate contains <0.2 ppm Environmental management standards specify maximum acceptable mercury concentrations for different media based on scientific and risk-based criteria. Mercury standards exist for water, sludge, fish tissue, drinking water and several other media. These standards, summarized in Table 2.1, provide an assessment of the efficacy of mercury release regulations 2.2 Mercury Remediation – Current Industrial Technology Improvement on current technologies for mercury removal has been an area of research that has been of interest for the last few decades. As the population increases so does the demand for electricity. Between 1990 and 2007, world energy consumption increased from 3.55 17 x 10 to 4.95 x 10 17 Btu and is projected to increase to 7.39 x 10 17 Btu by 2035. 19 With this increased demand the number and output of coal-fired power plants must increase assuming slow market penetration by alternative/renewables-based technology. It follows that technologies that reduce emissions and current mercury levels must be improved. The main physicochemical methods that are currently employed to remove mercury from wastewater are chemical precipitation, ion exchange, and liquid phase sorption (Table 2.2). Precipitation involves formation of insoluble compounds where mercury is either bound by sulfur-based ligands or coprecipitated with aluminum or ferric hydroxide. The advantage of chemical precipitation is that it can be used to treat wastewater with very high concentrations of mercury. However, removal by precipitation is not mercury-selective and in addition, the resulting materials can be unstable over time or at low (< 3) or high (> 9) pH so that the precipitates can decompose to re-release toxics. 17-19 20 Another significant limitation of the precipitation-based removal is cost; it requires at least stoichiometric amounts of complexant (or co-precipitant) to remove mercury to the levels below the EPA standard for drinking water (2 µg/L). 20 - 24 Two additional mercury removal methods include ion exchange resins and liquid phase sorption. These methods can be more cost-efficient since they are regenerable. Ionexchange methods are based on binding mercury by terminal sulfur-containing groups in a resin. These resins have higher loading capacities than carbon-based sorbents, selectivity is low 27-30 20,25,26 but their due to the low binding specificity of sulfur ligands. Sorption-based methods employ a variety of sorbent materials. Activated carbon is widely used for non-selective separation of mercury from liquid phases. proposed such as whey proteins, 33 wool, 29 20,23,31 xanthate, 29 Other sorption-based strategies have been keratin, 29 polyelectrolytes, 32 industrial waste and others. Table 2.2 details the relative benefits and drawbacks of several mercury removal methods. There have been a number of reports on synthetic materials 24,34-39 exhibiting high affinity and/or selectivity for mercury cations. In many cases important issues in mercury sorption were not addressed. These issues include, complex stability, sorbent capacity, selectivity for mercury vs. other common divalent cations (Ca2+, Cu2+, etc.), which is an issue of particular importance for the nitrogen-containing materials, sorbent regeneration, sorbent kinetics and efficacy of binding in relevant pH win 21 Table 2.2 Summary of Current Mercury Removal Methods Method Sorption Ion Exchange Materials Selective? Whey, xanthate, keratin, polyelectrolytes, industrial waste wool, activated and sulfur impregnated carbon, chelating polymer Tannery hair, chelating fibers, starch xanthate and cationic polymer, triazole polymer, polymersupported thiazacrowns, thiourea-glutaraldehyde chitosan resin Rohm and Haas (TMR), Srafion Anion exchange, shredded rubber, reduced MoS2 and WS2 Regenerable? References - - 20, 29, 32, 23, 34, 37 -  26, 34, 36, 37, 40, 41 -  25,27-31, 38 Precipitation Microbial demercurization, membrane electrodialysis, foaming surfactant SDTC, STC, TMT, PyDETH2, BDETH2 - - 20-22 See References - - 38-40 Functionalized porous matrices Thiol functionalized membrane, thiol functionalized mesoporous silica substrate   40, 41 TMR = Total Mercury Recovery; SDTC = sodium dimethyldithiocarbamate; STC = sodium thiocarbonate; TMT =2,4,6=trimercaptotriazine, trisodium salt; PyDETH2 = 2,6-pyridinediamineethanethiol; BDETH2 = 1,3benzenediamidothanethiol 22 2.3 Mercury Remediation – Current Areas of Research There are several categories of chemical systems that have been explored for mercury cation binding. They include: Type 1 – Polyethylene imine (PEI)- and polypyridine-based systems; Type 2 - Thiol functionalized systems; Type 3 - Ion-exchange polymers; Type 4 Polyaza- and polythio-macrocycles. A summary of sorbents of Type 1, Type 2 and Type 3 are presented in Tables 2.3, 2.4 and 2.6 respectively. Type 4 sorbents are addressed in more detail. 2+ Where possible, each individual sorbent is analyzed based on stability of the Hg -sorbent complex, sorption kinetics, sorbent capacity, effect of competing ions if any, sensitivity of binding to pH and regenerability of sorbent. With respect to competing ions, it should be noted that in the sorbents presented in Table 2.3, sorbents 1 and 5 and in Table 2.4, sorbents 8, 9 and 10 were tested for adsorption efficiency of other cations, and in all cases each of the sorbents showed a higher affinity for Hg 2+ than the other ions. Of the Type 1 sorbents presented in Table 2.3, sorbent 3, polyethylene imine (PEI) polymer-enhanced ultrafiltration (PEUF), 44 appears to have the largest sorbent capacity at 1000 2+ mg Hg /g sorbent. This method differs from the other sorbent systems in that it uses water 2+ soluble PEI to bind metal ions, in this case Hg , and rather than the complex precipitating from solution, the PEI/Hg 2+ mixture is simply put through a filter that has a pore size smaller than that of the polymer. Factors that could possibly affect the efficiency and therefore the feasibility of this system such as permeate flux variations due to polymer and metal concentrations, feed flow 23 rate and pressure were analyzed and found to be of little consequence to the overall success of this UF (ultrafiltration) technique. While this method seems promising, factors such as regeneration and reuse of the PEI polymer and whether filters can be washed and reused were not addressed. Table 2.3 Polyethyleneimine- and Polypyridine-based Sorbents for Hg 2+ 2+ Adsorption. Sorbent Number Sorbent Name Hg / Sorbent Complex Stability (logKs) 42 Polymin® 16.06 - - 4-9 - 43 PEI-Cellulose 12.9 288 - Any - - 1000 - 5 - - 435 (DMA), 156 (DEA), 556 (PA) 10 min > 3.3 - - 217 4 hr PEI ComplexationUltrafiltration - - - poly(4vinylpyridine) - 166 21 hrs 1 2 44 3 45 4 46 5 47 6 48 7 PEI Polymerenhanced Ultrafiltration Polystyrene functionalized beads. Amines: -NH2 (PA), -N(CH3)2 (DMA), N(CH2CH2OH)2 (DEA) Polyvinyl alcohol/PEI Maximum Sorbent Capacity (mg/g) Time to Equilibrium pH Regener -able? 24 Yes (EDTA) Capacity Step reduced 1: 6-7 by 50% Step after 1 2: 3-4 cycle Yes (0.5M ~3 HNO3) 2-3 According to their relative Hg 2+ capacities, sorbent 3 is followed by sorbent 4: polystyrene beads functionalized with a primary amine (-NH2 = PA), dimethyl amine (N(CH3)2 45 = DMA) and diethanolamine (N(CH2CH2OH)2 = DEA). The PA functionalized beads 2+ performed the best of the three at 556 mg Hg /g sorbent followed closely by DMA 2+ functionalized beads at 435 mg Hg /g sorbent. These materials also seem promising having an equilibration time of approximately 10 minutes and a working pH range of above 3.3. Regeneration of the sorbents, however, was not addressed. The capacity of sorbents 2 (porous cellulose carrier modified with PEI)43 and 5 (polyvinyl alcohol/PEI membrane) 46 2+ 2+ follow next at 288 mg Hg /g sorbent and 217 mg Hg /g sorbent respectively. While sorbent 2 is claimed to be effective at any pH due to the high 2+ stability of the Hg /PEI complex (logKs = 12.9), sorbent 5 only proved effective in the pH range of 2 – 3 because at higher pHs, the authors claim precipitation of Hg(OH)2 fouled the membrane. However, since Hg(NO3)2 was used as the source for Hg pH values it is likely the precipitate was either Hg(OH)(NO3) 50 unknown 49 2+ in this study, at higher or HgO since Hg(OH)2 is and is theorized only to exist as a transient species. The poly(4-vinylpyridine) (P4VP) membrane, 2+ 42 sorbent 7, was able to bind 166 mg Hg /g sorbent. This membrane reached maximum (equilibrium) capacity after approximately 25 21 hours, which is very slow compared to other sorbents. Also, mercury retention is only efficient at a pH of 3. Sorbent 7 is fully regenerable, however, only using 1M HNO3. Sorbent 6, another PEI polymer enhanced ultrafiltration system 47 was not evaluated for maximum sorption capacity. Experiments were run at 1.4:1 (v/v) polymer:metal loading ratio in - the presence of 0.2M Cl . The absence of chloride resulted in the sorption of less mercury under the same loading and pH conditions. This method involves two steps as previously mentioned. First is mixing of the PEI and Hg 2+ feed (complexation) and second is filtering and recycling of the PEI (regeneration). Step 1 was run in the pH range of 6 – 7 and step 2 in the range of pH 3 4. In the complexation step, as the concentration of the PEI increases, so does Hg 2+ retention - but in the regeneration step (addition of 0.6M Cl at pH 3.5), high PEI content runs were only able to be regenerated to ~ 50% of the original capacity. 40 Finally, sorbent 1, the Polymin® system , which is a polymer similar to PEI but with one primary, one secondary and one tertiary amine site per monomer unit, has the lowest Hg 2+ capacity of this type of system. This study did not address maximum loading capacity of the 2+ polymer but did find an Hg /Polymin® complex stability constant of log K = 16.06 via potentiometry. Also this study found that the working pH range of this sorbent is 4 – 9. 26 Sorbents 1 and 5 were also tested for their ability to bind other metals. Sorbent 1 2+ (Polymin®) was tested for Cd 2+ and Pb binding and it was found that the stability constants 2+ 2+ 42 for M + L  ML were Cd , log K = 8.57 and Pb , log K = 12.53. 2+ 2+ membrane) was tested for Pb , Cd and Cu 2+ Sorbent 5 (PVA/PEI binding and these ions were found to bind at 151 mg/g sorbent, 59 g/g sorbent and 44 g/g sorbent respectively, much lower values than that of 2+ Hg (>251 mg/g sorbent). 46 2+ While each appears to have a higher affinity for Hg , these experiments were not run in a competitive manner. That is, the sorbents were not placed in mixtures of divalent metal ions, but subjected to each individually. Sorbent 5 was tested for 2+ Hg binding in the presence of Ca 2+ ions. At low concentrations of Ca 2+ (20 – 40 ppm), Hg 2+ retention was decreased by 45 – 66%. At higher concentrations (200 – 300 ppm) retention was decreased by 18 – 35%. The authors attribute the effect at high versus low concentration to be due to ionic strength; however, at these concentrations this seems like an unlikely explanation. 27 Table 2.4 Thiol Functionalized Sorbents for Hg 2+ 2+ Sorbent Number Sorbent Name Thiol Functionalized mesoporous silica microspheres Thiol functionalized organoceramic composite Supramolecular attachment of thiols to mesoporous silica substrates Thiol functionalized magnetic silica nanocomposite 51 8 52 9 53 10 54 11 Adsorption. Hg / Sorbent Complex Stability (logKs) Maximum Sorbent Capacity (mg/g) - 321 7 min - - - 726 50 min 3-5 Yes (12M HCl) - - 2 hrs - Yes (CHCl3 - 19.8 - - Yes Time to Regenerable pH Equilibrium ? Research in the area of thiol-functionalized systems for Hg 2+ & C5H12) removal from waste effluents has continued since Pacific Northwest National Laboratories’ (PNNL) and Mobil Corporation’s thiol-functionalized Self-Assembled Monolayers on Mesoporous Silica 55 (SAMMS) was commercialized. Most of the research focuses on this class of materials, mainly on increasing surface area to promote maximum binding of Hg 28 2+ and other metals. Of the sorbents in Table 2.4, the thiol-functionalized organoceramic composite, 52 sorbent 2+ 9, has the highest maximum sorbent capacity at 726 mg Hg /g sorbent. This composite is synthesized by co-condensating a thiol-functionalized dialkoxysilanol-functionalized precursor with a thiol-functionalized alkyoxylsilanediol cross-linking agent. After condensation, triethylamine (TEA) is added to the resulting polymer and gelation/formation of an organoceramic composite adsorbent is the result. This is called a SOL-AD process. This sorbent has a relatively quick time to equilibrium of 50 minutes, a working pH range of 3 – 5 and can be regenerated, although under very strongly acidic conditions (12M HCl). This sorbent was tested 2+ for its Hg binding efficacy using a simulated coal-fired scrubber waste effluent whose composition is listed in Table 2.5 below. 52 Table 2.5 Simulated Coal-Fired Scrubber Waste Effluent Composition. Ion mg/L mol/L Ion mg/L 2+ + 350 0.00873 206 Ca NH4 2+ Mg Zn 2+ + Na 50 30 714 - 0.00206 Cl 0.00046 SO4 0.03104 mol/L 0.01144 1100 0.03102 2- 550 0.00572 - 1338 0.02158 NO3 The experiment was run at pH 3- 5 (expected pH range of coal-fired waste effluents) since the 2+ Hg species present at each of these pH values may vary. It was found that in this pH range, 2+ there was little deviation from the maximum capacity of 726 mg Hg /g sorbent. 29 51 Sorbent 8, thiol-functionalized mesoporous silica microspheres , follows with a maximum Hg 2+ 2+ adsorption capacity of 321 mg Hg /g sorbent after 7 minutes. These microspheres were generated by using tetraethoxysilane (TEOS) and 3mercaptopropyltrimethoxysilane (MPTMS) in the presence of a surfactant (Triton-X 100), HCl and NaF. These surfactant-assembled mesoporous oxides (likened to molecular sieves) have long-range order and have pore channels that are typically uniform in diameter and can range from 2 – 10 nm, which theoretically means there is a large amount of surface area. At low Hg 2+ concentrations, the ions are initially very slow to enter the pore channels and thus, despite the favorable binding thermodynamics between Hg and S, accessibility of the thiol groups is hindered. The authors suggest that this may be due to 1) slow diffusion into the pore channels because of electrostatic repulsion between the approaching Hg 2+ ions and unbound Hg 2+ ions at the pore opening or inside channel and/or 2) the initial formation of positively charged Hg-S + complexes within the channel. The authors also suggest that this slow diffusion into the pore channels coupled with lower thermodynamic stability of the sulfur-metal bonds may be the reason that these types of mesostructures have been reported to be unable to adsorb metal ions other than Hg 2+ 2+ 2+ 2+ 3+ 3+ 2+ including Cd , Pb , Zn , Co , Fe , Cu 2+ 52 and Ni . Regeneration of this sorbent is not addressed. 54 Sorbent 11, thiol-functionalized magnetic silica nanocomposites , have a maximum 2+ Hg 2+ capacity of 19.8 mg Hg /g sorbent. These nanocomposites were synthesized by coating CoFe2O4 nanoparticles with silica and subsequently functionalizing the silica with 330 mercaptopropyl groups. The Hg by placing 0.9 g into Hg 2+ 2+ adsorption capacity of the magnetic nanoparticles was tested solutions of varying concentrations including 6.65, 73 and 560 ppb at pH 5.5 for 1 hour. The adsorption doesn’t increase significantly with greater concentrations of mercury and the highest adsorption is more than an order of magnitude less than thiol-SAMMS. While this system retains less Hg at this time it has the advantage of easy recovery of bound 2+ Hg by using the proper magnets followed by chemical separation techniques. It should be noted that the authors make no mention of what ‘proper magnets’ would work nor has any further work been done to address the regeneration question. Finally, sorbent 10, supramolecular attachment of thiols to mesoporous silica substrates, 53 was created by functionalizing MCM-41 (a porous silicate) with varying densities of phenyl monolayers. These monolayers served as the supramolecular scaffolding to insert either benzylmercaptan (BM), 1,3-bis(mercaptomethyl)benzene (1,3-BMMB) and 1,4bis(mercaptomethyl)benzene (1,4-BMMB) via π-stacking interactions. This sorbent showed that 2+ at a metal:ligand ratio of 1:1, the log KD value is approximately 5.0 for Hg . At the same + 2+ metal:ligand ratio the log KD values for Ag , Pb and Cd 2+ are approximately 7.0, 6.5 and 7.0 2+ respectively. At a 1:2 metal:ligand ratio, in all cases except for Pb , log KD values decreased indicating that the most stable complexes are formed when there is a 1:1 metal:ligand ratio. Regeneration of this sorbent is possible washing with chloroform or pentane solvents; however, use of these types of solvents is not ideal since their use creates new waste. 31 Ion-exchange polymers have also been investigated. Sulfur-based polymers of this type 2+ are currently used as a method of Hg extraction. However, they are not common given the lack of selectivity. The work summarized in Table 2.6 describes ion-exchange polymers that are nitrogen-based. These studies address Hg 2+ binding in general but do not discuss selectivity. Since it is not discussed presently and is a major concern in practical applications, ideally future work on these types of sorbents will address the binding selectivity of Hg 2+ in the presence of competing cations. Ion-exchange polymers, sorbents 12 (polycysteine on pure cellulose membrane) (poly(sodium 4-styrene sulfonate) cross-linked with N,Nʹ′-methylene-bisacrylamide), 2+ 56 55 and 13 perform 2+ similarly having a maximum sorbent capacity of 360 mg Hg /g sorbent and 241 mg Hg /g sorbent respectively. The working pH for sorbent 12 is 6 while that of sorbent 13 is much lower - at pH 2. The maximum capacity of sorbent 12 is reduced 7-fold when Cl ions are present in a - 2+ 2:1 ratio of Cl :Hg . Finally, sorbent 12 is 65% regenerable using 1M HNO3 while sorbent 13 is not regenerable. 32 Table 2.6 Ion-exchange polymers for Hg 2+ Sorbent Number 56 12 57 13 Sorbent Name Polycysteine on pure cellulose membrane poly(sodium 4styrene sulfonate) cross-linked with N,Nʹ′methylenebisacrylamide 2+ adsorption. Hg / Sorbent Complex Stability (logKs) Maximum Sorbent Capacity (mg/g) 14.5* 360 - 6 65% (1M HNO3) - 241 - 2 No 2+ *Stability constant is of cysteine-Hg Time to Regenerable pH Equilibrium ? complex While sulfur-based technologies appear to be relatively successful when compared to other types of sorbents for mercury removal, there are several caveats to using sulfur. While the ‘soft’ nature of mercury cations leads to their tendency to form very stable complexes with thiols, sulfur can be easily oxidized on exposure to air and water to SO2, SO3 and H2SO4 as well as other species, rendering them inefficient mercury complexants. Their susceptibility to oxidation greatly reduces the binding ability in practical applications. Another caveat is that, while able to reduce mercury concentrations significantly, sulfur-based sorbents tend to be non-reusable given the stability of the Hg-S bonds. This fact leaves limited and often undesirable options when faced with disposal of such materials after mercury absorption. Once such option is combustion, which releases mercury back into the environment. 33 Mercury binding for the purposes of remediation is a very broad area of research and the types of sorbents presented thus far are as varied in their Hg 2+ binding efficacy and potential downfalls as they are in chemical types and methods of measuring that efficacy. Given the drawbacks of sulfur-based and other types of sorbents discussed, our research focuses on one type of system, nitrogen-based macrobicyclic complexants, for Hg 2+ binding. A review of several studies done by others using this type of sorbent is presented below. 2.3.1 Polyaza- and Polythia-macrocycles Several studies have been performed on macrocycles for their use in mercury removal. 2+ These studies focused on determination of macrocycle-Hg equilibrium stability constants. Three representative studies are summarized below. In 1996 Garcia-España et al investigated three potential polyaza ligands as selective mercury scavengers. 58 The three ligands studied are shown below in Figure 2.1. The stability constants were measured via potentiometric titration and metal-ligand stoichiometry was measured via UV/Vis spectroscopy and X-ray crystallography. 34 Figure 2.1 Polyaza ligands investigated as mercury scavengers by Garcia-España et al. The stability constants of all three ligands with Hg 2+ 58 are summarized in Table 2.7. The - high stability of the resulting complexes necessitated the use of a competing ligand (Cl ) since + hydrogen ions (H ) do not compete significantly with Hg 2+ in the pH range of 2-10, thus using acid-assisted desorption was not an option. The high values of these constants for the macrocycles in this study suggest that these ligands form stable complexes with mercury. Metal-ligand complexation ratios were also studied by Garcia-España et al. via UV/Vis spectroscopy for (11) and (12) and by X-ray crystallography for (10). For (11) the free ligand produced two absorption bands at 204 and 221 nm in the UV spectrum at pH 8.5. Upon addition 2+ of Hg the UV spectrum showed a single broad band centered around 210 nm. The molar absorptivity increased until the metal to ligand stoichiometry reached 1:1 and remained constant 35 even upon addition of additional Hg 2+ suggesting a preferential 1:1 metal to ligand stoichiometry. For (12) the free ligand produced two UV bands at 211 and 220 nm at pH 6.5, 8.5 and 10.5. Upon addition of Hg 2+ one UV band was seen at 217 nm. Molar absorptivities increased until a maximum was attained at a 1:1 ligand to metal stoichiometry. Ligand (10) behaves similarly to (11) and (12), however, in the presence of Hg 2+ concentrations above a 1:1 metal:ligand ratio, a precipitate was formed. Upon recrystallization of the precipitate, an X-ray crystal structure of an Hg2(10)Cl4 complex was obtained and can be seen in Figure 2.2 showing a 1:2 ligand to metal stoichiometry. 2+ Table 2.7 Logarithms of the Stability Constants for the Formation of Hg Complexes of (10), (11) and (12). Determined via potentiometry at 298 K in 0.15 mol/L NaCl. Charges 58 omitted for clarity. Reaction 10 11 12 Hg + L + 2H + 2Cl  [HgH2LCl2] - 36.88 38.20 Hg + L + H + Cl  [HgHLCl] 30.60 26.70 29.4 Hg + L + Cl  [HgLCl] 22.63 21.88 - Hg + L  [HgL] - - 21.60 Hg + L + H2O  [HgL(OH)] + H 10.9 12.08 12.13 36 Figure 2.2 (A) ORTEP drawings of the neutral [Hg2(10)Cl4] complex. (B) ORTEP drawing showing the coordination geometries about both mercury atoms. Thermal 58 ellipsoids are drawn at the 30% probability level. Reprinted with permission from ref 58. Copyright 1996 American Chemical Society. In 2001 Grondahl and coworkers investigated the stability constants of a series of mixed donor macrobicyclic mercury complexes. 59 The macrocycles studied and used for comparison are shown in Figure 2.3. Protonation and Hg 2+ complex stability constants were determined for (13) and (14) using potentiometric titration and compared to several other ligands. These data are summarized in Table 2.8. Comparing the log Ks values of the Hg 37 2+ complexes of the mixed N/S donor macrobicycles (13) and (14) to that of the all-nitrogen macrobicycles (8) and (9) it is evident that the all-nitrogen donors form more stable complexes with Hg 2+ having stability constants several orders of magnitude higher than the mixed donors. In fact, all other ligands mentioned formed more stable Hg 2+ complex than the mixed donors. H2N NH NH S NH HN HN HN HN S NH HN HN HN HN NH S HN HN H2N (8) (9) S NH NH (13) NH HN HN HN S HN NH S HN NH HN (15) (14) Figure 2.3 Mixed macrocyclic and macrobicyclic donors whose Hg by Grondahl and coworkers. 59 38 (16) 2+ complexes were studied Table 2.8 Protonation and Hg (8) (9) Species c c 2+ 59 Stability Constants for Various Ligands. (13) (14) (15) (16) a,b a,b b b EDTA b HL 11.95 11.44 9.70 10.16 11.4 9.53 10.37 H2 L 10.33 9.64 5.74 7.50 10.27 8.15 6.13 H3 L 7.17 6.49 2.5 4.78 1.6 - 2.69 H4 L ~0 5.48 <2.0 <2.0 0.9 - 2.00 H5 L - - - <2.0 - - 1.5 H6 L - - - - - - 0.0 [HgL] 28.1 26.4 17.7 19.5 23.0 24.32 21.5 [HgHL] - - 5.3 5.9 - - 3.2 a b c d This work. I = 0.1 mol/L; 298 K. I = 1.0 mol/L; 298 K. I = 0.2 mol/L; 298 K. The authors mention that the large difference in the stability constants between the mixed azathiamacrocycles and the azamacrocycles is not typically seen in other macrocycles where a secondary amine donor is replaced with a thiaether. For example logKs [Hg@1,4,7,10tetraazacyclododecane])]] 2+ 60 = 25.5 compared to logKs [Hg(16)] [Hg@1,4,7,10.13-pentaazacyclohexadecane] tetraaza-13-thiocyclohexadecane] 2+ 2+ 60 = 27.4 2+ compared to logKs [Hg@1,4,7,10- 60 2+ and that the magnitude of the stability constant is a reflection of the number of secondary amines bound to the metal center. [Hg(15)] 2+ and logKs = 25.15 . Kimura et al. suggested that the macrocyclic effect is limited for large cations such as Hg the enhanced stability of the [Hg(8)] 60 = 24.32 2+ and [Hg(9)] 2+ 61 This suggests that complexes in comparison to the complex is merely due to the two additional secondary amine sites and not due to the 39 macrobicyclic effect. 61 Therefore the most obvious explanation for the decrease in stability constants for the mixed donors is how the Hg 2+ ion interacts with the ligand. It is likely that with the addition of the sulfur atoms the cryptand may be unable to accommodate the Hg within the cage due to the difference in Hg-N (~ 2.3 to 2.5Å) 62 2+ ion 63 and Hg-S (~ 2.5 to 2.7Å)6 bond lengths and, though the authors do not mention it, likely also due to the difference in C-N and C-S bond distances. The authors suggest that this is not the case given the symmetrical nature of the complex shown by 13 C NMR experiments but admit that it cannot be ruled out. 2.4 Using Peraza[2.2.2]cryptand to Selectively Capture Hg 2+ Cations This work was done in collaboration with Dr. Volodymyr Tarabara and Dr. Julian Taurozzi in the Department of Environmental Engineering at Michigan State University and Dr. Misha Redko in the Department of Chemistry. characterization and Hg 2+ 64 What follows describes the preparation, cation binding of a polymer-based microparticulate sorbent based on p-xylyl-cross-linked peraza[2.2.2.]cryptand (Figure 2.4). This polymer has shown high selectivity and affinity for mercury cations. Surface morphology, size distribution and porosity of the polymer were assessed and its performance in Hg 2+ binding was evaluated in batch reactor experiments. Based on the measured mercury adsorption kinetics and isotherms and competition studies using a model polyethylene imine-based polymer, the mechanism(s) of adsorption by this novel sorbent are proposed and relevant adsorption parameters have been determined. The regenerability of the sorbent was assessed and the performance of the polymer 40 was compared to that of a commercial ion exchange resin in the presence of Ca 2+ ions, which are prevalent in waste effluents (see Table 2.5). N RN RN NR R= RN RN NR N (17) Figure 2.4 p-Xylyl cross-linked peraza[2.2.2]cryptand sorbent. 2.4.1 Polymer Preparation & Characterization The microporous resin was synthesized following Scheme 2.1. Since this polymer is insoluble, common methods of characterization to ascertain its exact chemical structure were limited. It was intended that, on average, each original secondary amine site in the peraza[2.2.2]cryptand was bound to one end of a p-xylyl cross-linker. However, elemental analysis results (see Experimental section) suggested that this was not the case. Presumably not all available secondary nitrogen sites are cross-linked leaving ‘dangling’ α-chloride ends within the polymer causing the relative C and N content to be much lower than expected. The experimental elemental analysis suggests that on average approximately 2.5 secondary nitrogen sites are not fully cross-linked. One other explanation could be the presence of residual polyethylene glycol, though, given the low oxygen content (as measured by elemental analysis) 41 it is more likely there are waters of hydration present. Infrared spectroscopy confirmed the presence of the expected functional groups C-N, aromatic C=C etc. as well as water and possibly polyethylene glycol. 1) 6 N HN HN NH HO N N O 4 H2O HN HN NH N (1) 3 Cl O n OH Cl 120oC, 20 min 2) 200oC, 30 min, N2 then cool to R.T 3) Wash iPrOH, H2O, 1M NaOH then H2O, iPrOH, hexane NR RN RN NR RN RN N R= (17) Scheme 2.1 Synthesis of p-xylyl cross-linked peraza[2.2.2]cryptand polymer (17). An initial estimate was made on the Hg 2+ adsorption capacity. This estimate was made on a by weight basis. Assuming each original secondary amine site is cross-linked with a p-xylyl moiety, the mass of one repeating unit of the polymer is ~ 677 g/mol. If one equivalent of HgCl2 salt is added per repeat unit, then the mass increases to ~1071.3 g/mol. The ratio of the masses 2+ of Hg 2+ to Hg /Complex then equals 200.6g /mol = 0.187x100% = 18.7% . This is the 1071.3g /mol theoretical ideal capacity of the polymer. This number may be reduced by the inability of cations € to penetrate the innermost binding sites of the polymer once the outer sites are occupied. In 42 - - contrast, it may be increased by formation of anionic complexes such as HgCl2 , HgCl3 , 2- HgCl4 etc. because of the possibility of ion-pair formation {[Hg(17)] 2+ 2- •HgCl4 }. 2.4.2 Size and Morphology of the Polymer Figure 2.5 shows the particle size distributions as measured using light diffraction, in suspensions of 1) polymer after synthesis (dashed line), 2) polymer after sonication for 18 hours (solid line) and 3) filtered and re-suspended polymer particles (dotted line). Figure 2.5 Particle size distribution of p-xylyl cross-linked peraza[2.2.2.]cryptand polymer. 43 SEM imaging of the polymer (Figure 2.6) corroborated the light diffraction measurements, revealing that this polymer exists as porous aggregates which, on average, are ~ 10 µm in diameter. These aggregates consist of primary particles (spherules) that are ~1 µm in diameter. The aggregates had porosity (ratio of the pore volume to the aggregate (total) volume) of 0.38 ± 0.01, which was calculated based on the results of N2 adsorption measurements (Table 2.9). 20µm 5µm 20µm Figure 2.6 SEM images of the p-xylyl cross-linked polymer showing spherules within the porous aggregates. 44 Nitrogen adsorption analysis of the dry polymer showed that the spherules had low micropore volume and area, which is consistent with the results of the SEM images. The measured value of the BET surface area of polymer particles (3.66 m²/g) is as expected when compared to the value predicted by making the assumption that 60% of the surface of 1 µm spherules, with a density of 1 g/mL, is able to adsorb N2. Table 2.9 N2 Adsorption of the p-Xylyl Cross-linked peraza[2.2.2]cryptand Property Units Value BET surface area 3.66 ± 0.13 2 Micropore area 0.32 ± 0.40 (m /g) External surface area 3.34 -5 Micropore volume 6.50 x 10 3 Volume of pores (cm /g) 1.21 x 10-3 (17 to 3000) Ǻ in diameter Total pore volume 4.5 2.4.3 Adsorption Kinetics Measurements The adsorption process is complex and considerable use is made of mathematical models to describe the possible rate-controlling mechanisms. Once such model, the homogenous surface diffusion model (HSDM), is often used to predict or simulate fixed-bed adsorber performance. This model provides a mathematical description of the main physicochemical mechanisms recognized to occur in fixed-bed systems. These mechanisms include axial flow with dispersion, local equilibrium at the particle surface, mass transfer resistance across a hydrodynamic surface boundary (film) layer surrounding the particle, intraparticle diffusion 45 65 along the pore surfaces and through pore liquid within the particle, intraparticle pore diffusion is considered negligible compared to surface diffusion and that equilibrium behavior can be described by the Freundlich isotherm. 65 An absorption isotherm is used to quantify equilibria and is a measure of the amount of adsorbate on the adsorbent as a function of adsorbate concentration at a constant temperature. The Freundlich isotherm is a very common and often used adsorption isotherm. If the collected kinetic data fits to the HSDM prediction, then a binding mechanism whereby the adsorbent surfaces are penetrated by the adsorbate via binding site ‘hopping’ can be inferred. That is to say, an adsorbate molecule on the topmost surface of the adsorbent further moves into the pore by being displaced by another adsorbate molecule. To determine how the Aza222 polymer behaves under equilibrium conditions, adsorption kinetics were measured to determine time to equilibrium and the data were compared to the 2+ HSDM. The Hg ion uptake kinetics predicted by numerical modeling were compared to experimental data using the intra-particle diffusion coefficient Ds , which is a measure of pore surface penetration as a function of time and the Freundlich coefficient, n , as fitting parameters. The Freundlich isotherm is discussed in more detail in the ‘Adsorption Isotherm Measurements’ section. In order to simplify the interpretation of data, the variations in feed concentration of mercury and polymer loading between experiments were taken into account by expressing the results of three adsorption experiments in terms of a normalized relative adsorption capacity Rn : R(t) C(t) − Ce R (t) = = n R(0) C(0) − C e 46 € (1) where C (0) = C (t = 0) is the initial feed concentration. R(t ) is the relative adsorption capacity remaining at time, t, and given by the (normalized) difference between the concentration of Hg 2+ in the supernatant, C (t ) , at a given time and the equilibrium concentration, Ce , in the supernatant. Ce is achieved when the adsorbent removal capacity becomes zero: R(t) = C(t) − C e ⋅ 100% C e (2) Absolute adsorption capacity of the material (wt %) was calculated based on the amount of polymer added, the € concentration of mercury in the feed, and the concentration of mercury at equilibrium (after 6 hrs). 2+ Figure 2.7 Adsorption kinetics of Hg adsorption by p-xylyl cross-linked peraza[2.2.2]cryptand polymer predicted by the homogeneous surface diffusion model and experimental data. 47 Using data collected during the kinetics experiments, the experimentally measured Ce values gave an absolute polymer adsorption capacity of 18.6 % wt/wt or 186 mg of dissolved mercury per 1 g of polymer (186 mg/g). This value corresponds well to the initial theoretically estimated 18.7% wt/wt value. Figure 2.7 shows the kinetic data from experiment versus the model prediction. The fitted surface diffusion coefficient within the solid phase, Ds , was found to be 3.75 x 10 -9 is well within the wide reported range (1 x 10 to 8 x 10 -17 coefficients of organic compounds in soils and sediments -13 2 cm /s, which 2 ) cm /s for solid phase diffusion 63-65 . The local Freundlich coefficient, n , was found to be 0.0604 indicating favorable sorption ( n < 1 ). Although the values of Ds and n 2 were obtained with a coefficient of determination (r ) value of only 0.97, they may be used as coarse estimates given the assumptions within the HSDM. The accuracy of the values of Ds and n may have been affected by the finite width of the particle size distribution used since the measured particle size distribution (see Fig. 2.5) is extremely wide in comparison and chemical complexation (rather than Freundlich physisorption) of mercury. Once equilibrium was achieved, a calculated Rn value based on thermodynamic experimental data (see next section) was higher than the theoretical value obtained using the kinetic data, which may be due to one or more factors. First, transfer losses during weighing of the 48 polymer recovered from the filter cake may have occurred. Second, there may have been small differences in the amount of polymer removed for AA analysis for each sample, which resulted in varying amounts of polymer left over in the batch reactor to adsorb mercury. Finally, the time it took to filter the samples from the batch reactor was up to ~ 2 minutes, which was comparable with adsorption times for the first samples withdrawn from the batch reactor and may have caused fluctuations in the concentration of mercury in the filtrate. This is because as the suspension is filtered the adsorption of mercury into the suspended polymer particles and those already on the filter surface continues inside the filter-fitted syringe that was used to extract samples. This interference is also a possible explanation of the highly scattered data observed in earlier runs of the adsorption experiment. To account for the additional mercury adsorption during ~1 min of filtration time, a time offset of 1 min was added to each recorded batch adsorption time value. 2.4.4 Adsorption Isotherm Measurements There are two common adsorption isotherms that are used to quantify equilibria. They are the Freundlich (used above in the HSDM) and the Langmuir isotherms. One important assumption that both isotherms make is that binding only takes place on the surface of the pores (monolayer adsorption). The main difference between the two isotherms is that the Langmuir assumes that each binding site has the same relative energy and is not affected by the state of surrounding binding sites (i.e. bound, unbound etc.). This assumption makes the Langmuir isotherm a more restricted form of the Freundlich. Both are mathematical models of equilibrium adsorption behavior and are used to gain an understanding of adsorbate binding in an adsorbent. 49 Six independent adsorption experiments were conducted, which covered a wide range of equilibrium mercury concentrations in the liquid phase (Figure 2.8). This isotherm can be broken down into two segments. The first, consisting of lower values of C (0 to ~1,200 mg/L), can be e considered a high affinity isotherm indicating adsorption so strong that mercury in the qe (mg Hg/g Aza222) € supernatant is undetectable until the surface of the polymer is essentially saturated. Ce (mg/L) Figure 2.8 Adsorption isotherm measurements over a set of six independent experiments. The dashed line represents the theoretical adsorption capacity by weight. 50 The plateau of the high affinity segment may describe the binding of Hg 2+ within the peraza[2.2.2]cryptand cage since its value is consistent with the theoretically estimated 20% wt/wt stoichiometric adsorption capacity. The second segment, recorded for higher values of feed concentrations (> ~1,200 mg/L), 2+ corresponds to Hg binding above the theoretical 20% capacity and continues to increase for higher values of C . This portion of the overall isotherm may describe non-specific e physisorption that occurs after all available cryptand sites are occupied. This may be due to 2+ € electrostatic binding (ion-pairing) of the positively charged [Hg@peraza[2.2.2]cryptand] cages 2- with HgI4 feed anions. This fact is of environmental significance because this demonstrates the ability of the polymer to be useful in situations with high mercury feed loadings. This kind of behavior indicates the polymer’s ability to tolerate spikes in feed mercury concentrations, though the threshold where this type of non-specific physisorption occurs (~1,500 mg/L) is much higher than typical environmental mercury concentrations. 2.4.5 Adsorption Isotherm Modeling The isotherm data (excluding the physisorption concentrations) was fitted using the Langmuir and Freundlich equations, which are the most common empirical models for adsorption at a surface boundary at equilibrium as mentioned previously. This type of mathematical modeling helps to accurately calculate the maximum adsorbate capacity for a 51 specific adsorbent. The procedure used to fit data was performed by rearranging the Langmuir and Freundlich equations, which allowed for a linear regression analysis. The Langmuir equation is expressed as QbC e q= 1+ bC e where q is mass of adsorbed Hg 2+ (4) 2+ per unit mass of polymer at equilibrium (mg Hg /g € 2+ polymer), Q is the maximum loading capacity of the polymer at equilibrium (mg Hg /g polymer), b is the relative energy of adsorption parameter (L/mg), and Ce is the concentration of 2+ Hg in the liquid phase at equilibrium (mg/L). It should be noted that b is not truly a measure of energy but the ratio of the rate association constant (which is based on a second order reaction 2+ dependent on the [Hg ] and [sorbent] with units of L/mol*s) and the rate dissociation constant 2+ (which is based on a first order reaction dependent on [Hg /sorbent complex] with units of 1/s), thus the result is not unitless (L/mol), which can be easily converted to L/mg. Rearranging (4) for ratio of the concentration of Hg 2+ left in solution (Ce) to the mass of Hg adsorbed by the polymer per unit mass of polymer (q) gives: C C e = 1 + e q Qb Q € 52 (5) 2+ Plotting versus Ce gives a linear plot with a slope of 1/Q and y-intercept of 1/(Qb). This allows for the calculation of these two parameters. The Freundlich equation is written as q = KC1/n e (6) where K is the adsorption capacity at unit concentration and 1/n is an adsorption intensity parameter (between 0 and 1). € Taking the logarithm of both sides of (5) gives log(q) = log(K) + n log(C ) e (7) A plot of log(q) versus log(Ce) gives a linear plot with a slope of n and y-intercept of log(K). € This allows for the calculation of these two constants. The plots obtained for the rearranged € Langmuir and Freundlich equations are shown in Figures 2.9 and 2.10 respectively. 53 Figure 2.9 Plot of C e versus C and linear regression fit for the isotherm adsorption data e q € Figure 2.10 Plot of log(q) versus log(Ce) and linear regression fit for the isotherm adsorption data. € Regression analysis shows that the Langmuir model gives a better fit and thus appears to offer a 2 more accurate representation of the isotherm adsorption behavior (R = 0.986). Figure 2.11 illustrates the isotherm adsorption experimental data, and the plot of the Langmuir curve using the parameters obtained from the fitting procedure. 54 qe (mg/g) Ce (ppm) Figure 2.11 Plot of adsorption isotherm experimental data and the curve obtained using the calculated Langmuir parameters. The values for the maximum equilibrium loading on the sorbent (Q) and relative sorption energy 2+ (b) were 192.3 mg Hg /g polymer and 0.075 L/mg, respectively. The relative sorption energy 67 value (b) is below that of activated carbon, which has a b = 0.375 L/mg) , however, the polymer can be compared to other Hg adsorbing materials such as natural and modified clays (b = 1.2 to -5 67 8.1 x10 L/mg) , chelating copolymers (b = 0.0035 L/mg) 68 and membrane supported Hg 44 complexants (b = 0.0162 L/mg) . In order to test the polymer’s performance under low concentration conditions, 0.2 mg of the polymer was subjected to 25 mL of a ~2.4 ppm solution of Na2HgI4 at room temperature with constant stirring. After six hours, a 15 mL sample of the supernatant was taken and filtered 55 three times as in the high concentration experiments. Cold vapor atomic adsorption (AA) measurements of the initial feed solution as well as the supernatant were taken. The data revealed that the polymer’s performance under low Hg 2+ concentrations is similar to high 2+ concentrations, absorbing 167.5 mg Hg /g of polymer compared to 192.3 mg/g calculated during adsorption isotherm modeling. 2.4.6 Mercury Adsorption in the Presence of a Competing Divalent Ion and Comparison with a Commercially-available Material. 2+ Calcium is an environmentally relevant potential interference ion for Hg adsorption given its high concentrations in waste effluents as a major component in hard water. Water is considered ‘hard’ if the Ca between 0 and 60 ppm. excess to Hg 2+ 69 2+ ion concentration exceeds 60 ppm and ‘soft’ at concentrations Based on this definition, even soft water can contain Ca levels (on average 1 ppb). Therefore Ca 2+ 2+ ions in great was chosen as a relevant competitive +2 ion for binding in the polymer. The peraza[2.2.2]cryptand-based polymer did not adsorb Ca in detectable quantities. A sulfur-based commercial resin (name excluded at the request of the manufacturer) was selected as an appropriate comparative reference as it combined the benefits of the developed peraza[2.2.2]cryptand-based sorbent including high capacity, regenerability, and selectivity with respect to calcium. 70 2+ The capacity of the cryptand-based polymer for Hg was higher (~ 20 mg/g) than the commercial resin in the absence, and more so in the presence of (~40 mg/g) 56 calcium ions (Figure 2.12). The slight increase in Hg 2+ uptake in the presence of Ca 2+ by the cryptand-based polymer was unexpected. Given the complex nature of the mixture, a definite reason for the increase in Hg 2+ uptake in the presence of Ca(NO3)2 at these concentrations is difficult to flesh out. It was originally thought that one possible explanation may be that in the course of synthesizing the polymer (see Scheme 2.1) some of the polyethylene glycol (PEG) used to control cross-linking density remained ‘stuck’ in the interstices despite thorough rinsing with various solvents. This residual PEG could have been blocking possible Hg 2+ but once Ca 2+ binding sites, was introduced, the PEG bound the calcium cations, freeing those sites for Hg 2+ binding. Since oxygen atoms are considered ‘hard’ bases, the PEG would likely have a higher affinity for ‘hard’ Ca 2+ ions over ‘soft’ Hg 2+ ions. This explanation is unlikely though given the oxygen content by elemental analysis. Since the capacity was calculated based on the difference in the Hg 2+ feed concentration and that of the Hg 2+ content in the supernatant (by flame or cold vapor AA), precipitation of other insoluble mercury complexes such as Hg(OH)(NO3) may have led to higher mercury binding values. The formation of a precipitate was not visually observed but at low concentrations such as these, visual detection may be difficult. Therefore, the reason for the increase in Hg 2+ uptake in the presence of Ca 57 2+ by the Aza222 polymer remains unclear. 200   200   175   158   Capacity (mg ion/g sorbent) 180   156   160   140   120   Aza222   100   Commercial   80   60   40   20   0   2+ 2+ Figure 2.12 Hg sorption in the absence and in the presence of a six-fold excess of Ca by peraza[2.2.2]cyryptand-based sorbent and a commercially available sorbent (Commercial resin) 2.4.7 Regeneration Study Initially regeneration was attempted in a column by using acid to protonate the amine sites thus allowing the Hg 2+ to escape, but upon addition of the acid, the polymer swelled and liquid could not pass through. Therefore, regenerability of the polymer was studied under column and batch conditions using Na2S to extract Hg 42.0). 71 2+ from the polymer (log Ks for HgS = The initial column-based regeneration study was performed 72 using a HgCl2 stock solution , which contained an amount of mercury in 5 mL (14.3 mg Hg, 53 µmoles) that was below the estimated capacity of the resin (~ 26.5 mg HgCl2 or 19.6 mg Hg). A saturated solution 58 of Na2S regenerant was used. The amount of sulfur contained in the saturated solution far exceeded the amount necessary to bind mercury to ensure maximum possible regeneration and also to avoid precipitation of HgS by formation of other soluble species such as Na2HgS2. The amount of mercury in the eluent was measured and the mercury remaining in the polymer calculated. This data is presented in Table 2.10. The first and second cycles afforded ~ 75% of the original capacity while the third was regenerated to ~ 90% capacity and the final cycle resulted in ~ 99% available capacity. The results demonstrate that the column absorbed virtually all the Hg 2+ passed through. Also shown is the ability of the column to be regenerated to ~ 90% of the original calculated capacity. In order to obtain a more comprehensive and understanding of the polymer’s regeneration behavior, batch studies were performed. In these studies, both the feed and filtrate solutions were analyzed in both the adsorption from the aqueous phase and the Na2S regeneration steps. The analysis showed that upon regeneration using a saturated aqueous Na2S solution, the polymer released 85.5 % (256.6 ppm assuming all 300 ppm of the feed was adsorbed) of the bound mercury and the polymer digest, which theoretically contained the remaining 14.5% (43.5 ppm), showed no detectable mercury content though the detection limit (3 ppm) of the atomic absorption technique used in the analysis is much lower than that value. This suggests that there were possible mass losses during the polymer digestion process using HNO3. 59 Table 2.10 Sorbent Regeneration Using Na2S in a Packed Column. 72 Hg into column (mg) Hg in eluent (mg) Hg remaining in column (mg) 5 mL (~350 ppm) HgCl2 through 12.9±1.3 0 12.9±1.3 10 mL H2O rinse 0 0 12.9±1.3 H2 O 0 9.7±1.0 3.2±1.6 5 mL (~350 ppm) HgCl2 through 12.9±1.3 0 16.1±2.1 10 mL H2O rinse 0 0 16.1±2.1 5 mL 0.39% Na2S soln 0 10.6±1.0 5.5±2.3 10 mL H2O rinse 0 0.9±0.1 4.6±2.3 5 mL (~350 ppm) HgCl2 through 12.9±1.3 0 17.5±2.6 10 mL H2O rinse 0 0 17.5±2.6 5 mL 0.39% Na2S soln 0 16±1.6 1.5±3.1 10 mL H2O rinse 0 0 1.5±3.1 5 mL (~350 ppm) HgCl2 through 12.9±1.3 0 14.4±3.4 10 mL H2O rinse 0 0 14.4±3.4 5 mL 0.39% Na2S soln 0 13.9±1.4 0.5±3.6 10 mL H2O rinse 0 0.4±0.04 0.1±3.6 Cycle Sample Info 1 5 mL HgCl2 stock (~3500 ppm) to 100 mL 6 mL 0.39% Na2S soln + 7 mL 2 3 4 These data suggest that the Aza222 polymer may be suitable for environmental applications since it follows the sequence of adsorption, separation and regeneration, which is typical for currently used systems such as powdered activated carbon. 60 2.4.8 Comparison to a Non-macrocycle-containing Cross-linked Polymer To discover whether the presence of the macrocyclic cage in the cryptand-based polymer is crucial in uptake and selectivity of the system a polymer that did not include the macrocycle was synthesized under similar conditions. In this case, polyethylene imine, rather than the peraza[2.2.2] cryptand, was cross-linked using p-xylyl linkers (Scheme 2.2). Two batches were synthesized with different cross-linking densities. The cross-linking density appeared to be a function of heating time. Less heating time resulted in lower cross-linking density based on the C:N ratios in elemental analysis of the two polymers. In the case of (18), elemental analysis implies that one of the two cross-linking units per PEI unit depicted in Scheme 2, may be a ‘dangling’ unit attached to one end of the PEI. H N 4 O 2 HO H N N H + O + n Cl 2 N (24 hrs) OH Cl + N H n 1) 130oC, 4hrs or 24 hrs 2) Wash with H2O, EtOH, CH2Cl2 N N H N (18) N H n OR N N H (4 hrs) H N (19) N H n Scheme 2.2 Synthesis of polyethylene imine polymer cross-linked with p-xylyl units. 61 Both polymers were light yellow; however, (18) was powder-like whereas (19) had a spongy texture. Infrared spectra of the two polymers were essentially identical and similar to that of the cryptand-based polymer. Both batches were tested for binding and selectivity and prepared by sonicating for 18 hours and then grinding using a mortar and pestle. After preparation particle size measurements were taken. (18) and (19) had larger average particle sizes after grinding and sonicating (~80 µm for (19) and ~60 µm for (18)) than the cryptandbased polymer (~ 10 µm). Since these values were within the wide range of particle size distribution of the cryptand-based polymer (0.1 to 100 µm) this was not viewed as a major difference. Also, BET analysis of (18) showed a surface area similar to that of the p-xylyl cross2 linked peraza[2.2.2]cryptand sorbent. The BET surface area of (18) was found to be 3.08 m /g 2 compared to 3.66 m /g for the peraza[2.2.2]cryptand-based sorbent. A chemical formula of C20H31N4Cl • 3 H2O for one repeating unit in (18) is consistent with elemental analysis suggesting that there is one fully cross-linked and one partially cross-linked p-xylene per repeating unit. For (19) a chemical formula of C12H23N4 • 1.25 H2O per per repeating unit suggesting one full cross-linked p-xylene unit and, on average, 1.25 waters. Each polymer was subjected to a solution containing a high concentration of Hg 2+ ions (~ 560 ppm) to compare the binding of each polymer to that of the cryptand-based polymer. 2+ Figure 2.13 shows that (18) performs quite well, sorbing 1.65 g Hg /g after 24 hours while (19) 2+ 2+ performs poorly, only sorbing 77 mg Hg /g. Assuming two Hg ions per repeating unit, on a 2+ wt/wt basis as calculated for the Aza22 polymer, (18) had an estimated capacity of 1.1 g Hg /g 62 2+ and for (19) 1.8 g Hg /g. Based on these estimated values, (18) adsorbed 1.5 times more than expected and (19) adsorbed ~ 23 times less than expected. The difference in binding capability between (18) and (19) suggests that cross-linking density plays a significant role in Hg binding. This data suggest that higher cross-linking density results in much higher Hg 2+ 2+ binding with respect to the PEI-based polymers. One explanation for the significant difference may be + that ion-pairing interactions between positively charged [N-Hg] sites and the negatively 2- charged HgI4 from the feed may be much more prevalent in (18) than (19) since (18) is more densely cross-linked and therefore there may be less ‘free’ nitrogen binding sites adjacent to each other for [N-Hg-N] binding to take place. Another possibility is that the presence of a ‘dangling’ cross-linking unit may play a role in binding. 1800   2+ Capacity (mg Hg /g sorbent) 1600   1400   1200   1000   800   600   400   200   0   (18)   2+ Figure 2.13 Hg (19)   Aza222   uptake by PEI-based and peraza[2.2.2]cryptand-based polymers. 63 Both (18) and (19) were subjected to solutions containing Hg 2+ whether Ca would interfere with Hg 2+ 2+ 2+ and Ca to ascertain binding. As in the cryptand-based experiment, each PEI polymer was mixed with a solution of Hg Ca 2+ 2+ 2+ and Ca ions containing a six-fold excess of ions. The data are presented in Figure 2.14. The performance of (18) was significantly 2+ 2+ changed, reducing the capacity of the polymer from 1.65g Hg /g to 56 mg Hg /g in the 2+ Capacity (mg ion/g sorbent) presence of Ca , 2+ Figure 2.14 Hg 2+ binding by PEI-based polymers and Aza222 in the presence of a six-fold excess of Ca . 2+ a 97% decrease. The binding of (19) was reduced to only 3 mg Hg /g as opposed to 77 mg 2+ Hg /g in the absence of Ca 2+ ions, a 96% decrease. This is in sharp contrast to the performance discussed earlier of the Aza222 cryptand-based polymer’s slight enhancement in binding in the 64 2+ presence of Ca ions. These results suggest that the macrocyclic cage in the peraza[2.2.2]cryptand-based polymer plays a significant role in Hg 2+ of Ca 2+ selectivity in the presence which, can be present in high concentrations when compared to Hg 2+ ion content in waste effluents. This research represents the first polymer-based aza-cryptand system for the removal of mercury from waste effluents. Experimental results show that the novel p-xylylperaza[2.2.2]cryptand-based polymer exhibits selectivity over common hard water cations (Ca + + and indirectly Na since the feed used for many experiments was Na2HgI4 and extremely high affinity for mercury cations. The calculated stability constant of the 73 [Hg@peraza[2.2.2]cryptand] complex, log Ks = ~ 28.5 polymer. 65 was implied by the selectivity of the CHAPTER 3 DIHYDROGEN BONDING 66 3.1 Dihydrogen Bonding and Crystal Engineering Hydrogen bonding is fundamental in many areas of chemistry including molecular recognition and supramolecular synthesis. It is one of the most biologically important interactions between bio-molecules, playing a key role in establishing the structure of DNA and proteins. Typically this type of interaction is between a positively charged hydrogen of an A-H (A = O, N, halogen, C) proton donor and the lone pair of an electronegative element, the πelectrons of a multiple bond or aromatic ring, or a transition metal, which all represent the proton acceptor. 74 A different type of hydrogen bond, a ‘dihydrogen’ bond, was named as such by 75 Crabtree et al. in 1995. In this type of hydrogen bond, a σ M-H bond (where M is less electronegative than H) is the electron donor. This type of hydrogen-hydrogen bonding has strength and directionality comparable to those found in conventional hydrogen bonding and consequently may have applications in catalysis, crystal engineering and materials chemistry given that it can influence structure, reactivity and selectivity in solution and the solid state. 78 In the field of crystal engineering, extended crystalline covalent solids are generally elusive because it is extremely difficult to perform bond-making chemistry with reagents when the substance occupies 3-dimensional space. This is because of limited access to reacting sites. The methodical synthesis of complexes held together via intermolecular forces, such as dihydrogen bonding, essentially “loads” a reaction in a crystalline form and then “fires” upon a stimulus such as heat. Dihydrogen bonding offers an opportunity for a rational methodology to construct these types of compounds by using topochemical control via conversion of dihydrogen 67 to covalent bonding. Two challenges to retaining the crystalline order and thus the possibility of reversibility/reusability of the system then arise: (1) large geometry changes upon bond reorganization and (2) product release within the lattice. 78 For example, in the context of hydrogen storage, the product released would be H2 (g). Jackson and Custelcean proposed two strategies to address the topological problems. 76 The first is the design of cations that form a closed-loop structure using dihydrogen bonding, where, upon release of H2, the lattice distortion would not be cumulative (Scheme 3.1) and the second is the use of globular cations large enough so that the overall lattice structure is determined by their size and packing (Scheme 3.2), leaving room for other partners in the interstices of the large components’ lattice. Then, upon release of products (e.g. H2), the overall volume change would be minimal and the H2 could diffuse through and out of the crystal. Several closed-loop structures have been reported by Jackson and Custelcean where triethanolamine (TEA) and borohydrides MBH3X (M = Na, Li; X = H, CN) were utilized. 68 76-78 Y+ H 4 H2 X H BH4H -H Y+ X -H B 2 4B -H +Y X H 2B +Y X Scheme 3.1 Closed-loop strategy for topochemical control of H2 loss in dihydrogen-bonded 76 systems. Adapted with permission from ref 76. Copyright 2001 American Chemical Society. +X X+ H BH4 H - 3 H2 +X H HB X+ X+ X+ Scheme 3.2 Divergent strategy for topochemical control of H2 loss in dihydrogen-bonded 76 systems. Adapted with permission from ref 76. Copyright 2001 American Chemical Society. 69 The NaBH4·TEA complex self-assembles in two-dimensional hydrogen bonded layers in the solid state, and upon topological decomposition yielded a polymeric trialkoxyborohydride. 79 Unfortunately the crystallinity of the covalently-bonded product was poor. The less basic NaBH3CN·TEA analogue decomposed at 100°C above its melting point (i.e. in a liquid state) highlighting the point that to enable reaction in the solid state (i.e. below melting point temperature), the acidity/basicity matching of the proton and hydride partners requires tuning of the reactivity of these dihydrogen bonded systems. 76 In contrast, the LiBH4·TEA complex self-assembled into one-dimensional dihydrogenbonded ribbons and was topochemically decomposed in the solid state yielding a polymeric trialkoxyborohydride, which was stiochiometrically more well defined than the Na analogue. Further annealing of this complex at 120°C resulted in total loss of order suggesting the metastable nature of the system. 79 Another system modeled after the closed-loop and globular approaches was NaBH4·THEC (THEC = N,N´,N´´,N´´´- tetrakis-(2-hydroxyethyl)cyclen) (Figure 3.1). 70 80 80 Figure 3.1 NaBH4·THEC complex showing dihydrogen bonding to a borohydride anion. Adapted with permission from ref 80. Copyright 2000 Wiley-VCH. This compound’s crystal structure consists of four conventional O···H···O hydrogen bonds and four O···H···H···B dihydrogen bonds forming a crystal with D2 symmetry. The X-ray crystal structure and crystal packing diagrams confirm the self-assembly of the dimers (Figure 3.2). Heating this complex in the solid state led to complete loss of H2 and the formation of a crystalline product as evidenced by the well-defined peaks in the powder X-ray diffraction 80 plot. The decomposition is homogeneous and a 9% volume shrinkage of the unit cell during the process was shown by X-ray diffraction and polarizing microscopy. The importance of this system was that it demonstrated that the local crystal stoichiometry dictated the extent of the - reaction at the BH4 centers. 71 Figure 3.2 The X-ray crystal structure and crystal packing of NaBH4·THEC. permission from ref 80. Copyright 2000 Wiley-VCH. 72 80 Reprinted with 3.2 Applications of Extended Covalent Crystalline Solids - Hydrogen Storage via Metal Borohydrides and Amines The demand for environmentally benign fuels has increased in recent years given the volatility in the cost of oil based on economic and/or political reasons and concerns about the sustainability of a fossil-based hydrocarbon fuel economy. This has spurred research in the design of materials for hydrogen storage. The ‘holy grail’ material for this purpose must meet several criteria including: 81 1) it must have a high mass and volumetric storage capacity of 6.5% by wt production of hydrogen and at least 65 g/L of available hydrogen. 2) it must have a decomposition temperature of 60 - 120°C. 3) its thermal adsorption/desorption process must be reversible including low temperature desorption and low pressure adsorption. 4) its application must be cost effective. 5) and be environmentally benign, non-explosive and ideally inert to water and oxygen. Several different types of materials are being studied as hydrogen storage materials, including metal hydrides, metal clusters, nanostructured materials, carbon nanostructures, complex chemical hydrides and metal organic frameworks. 82 Metal hydrides are reversible under ambient conditions but too heavy and expensive. Metal cluster materials show promise according to theoretical calculations but are difficult to synthesize. Nanostructured materials are an emerging area of research but are not yet practical in application according to a recent review by P. Jena. 82 Carbon nanostructures and metal organic frameworks offer good reversibility but require very low temperatures, while complex chemical hydrides have high hydride densities, are reversible at reasonable temperatures but overall have poor reversibility. The promise shown in 73 a study by Autrey et al. in 2005 using ammonia borane hydrogen storage has sparked interest in this system for use as a method of on-board hydrogen storage. 83 There are materials that meet several of the above ‘holy grail’ criteria and the most extensively studied are summarized in Table 3.1. A review of these materials and several others that do not appear in the table allowed several considerations to be taken into account in the rational design of possible hydrogen storage systems: 81 1) The thermal decomposition o temperature for binary hydrides, MHn, correlates well with the standard redox potential, E , for n+ 0 the M /M couple. 2) The M-H bonds may be stabilized or destabilized, and the Tdec value may thus be tuned by the proper choice of metal, M, second element, E, and the stoichiometry of the ternary hydride (eg. (EHm)a(MHn)b). 3) The larger the a/b ratio and the softer the E m+ n+ cation the larger the thermal stability of the complex (for hexacoordinated M ). 4) Hydrides - may be stabilized by partial substitution of H by electron donating ligands such as I , or by molecules with an available lone pair for bonding. 74 81,83,85 Table 3.1 Selected Extensively Studied Hydrogen Storage Materials. Practical Kinetic Tdecomposition wt % Material Notes reversibility (°C) hydrogen Criteria Met PdH0.6 0.6 Excellent Ambient $1000/oz. 2,3,5 Mg2NiH4 3.6 Very good Ambient Fails to meet DOE wt % criterion 2,3,4,5 NaAlH4 • 5.5 Good 125 Fails to meet wt % criterion 2,3,4,5 MgH2 7.6 Very poor 330 Cheap Mg metal 1,4,5 LiAlH4 • 9.0 As of 2004 irreversible 200 - 400 LiAlH4 is expensive 5 NaBH4/H2O 9.2 Irreversible Ambient Expensive Ru-containing catalyst 1,2,4,5 AlH3 10.0 Irreversible 150 Very cheap Al metal 1,4,5 11.1 Risk of explosion of HSM if H & nonmetal stored in liquid or gas phase Thermal activation difficult Environmentally friendly 1,3,4,5 12.5 Irreversible Thermal activation difficult Toxic liquid 1,4 11.5 Reversible 250 -450 - 1,4 TiO2 TiO2 H2 O MeOH LiBH4 + MgH2 ½ Given the aforementioned criteria, the use of aminoborohydride- or amine-borane-based systems in hydrogen storage applications appears to be a worthwhile pursuit. In 2010 Wu and coworkers developed a new class of ammonia borane complexes for use in hydrogen storage. 86 The materials included Li2(BH4)2NH3BH3 and Ca(BH4)2(NH3BH3)2. 75 These complexes were synthesized by ball milling stiochiometric ratios of either LiBH4 or Ca(BH4)2 with NH3BH3 powders under 1 bar He. Since these complexes are moisture sensitive, all experiments were performed in an inert atmosphere. X-ray crystal structures were determined for each material using powder X-ray diffraction. These structures are shown below in Figure 3.3. In the Li2(BH4)2NH3BH3 crystal the H3B-H---HN distances are 2.248Å and 2.254Å, however there is little dihydrogen bonding interaction between NH3BH3 layers, which are separated by a contact distance of 2.439 Å, longer than the sum of the hydrogen van der Waals radii. The BH---HN distance within the amine-borane layers in Ca(BH4)2(NH3BH3)2 is 1.735Å, which is shorter than that in AB (2.0 – 89 2.2Å) suggesting strong dihydrogen bonding. The H3B-H---HN distances are 1.986 and 2.037Å. Dehydrogenation of Li2(BH4)2NH3BH3 and Ca(BH4)2(NH3BH3)2 using temperatureprogrammed desorption (TPD) was performed by Wu et al. and the data are shown in Figure 3.4. These data shows that Li2(BH4)2NH3BH3 initially decomposed between 105 and 135°C with an additional decomposition step at 360°C. Ca(BH4)2(NH3BH3)2 shows similar results beginning decomposition between 125 and 155°C and completing the final decomposition step at 325°C. Referring to the XRD patterns collected on Li2(BH4)2NH3BH3 and Ca(BH4)2(NH3BH3)2 samples after the first desorption step it appeared that the H2 release in this step was mainly due 76 to NH3BH3 units. Pure NH3BH3 releases 1.9 equivalents of H2/mol at 129°C and 160°C, which are temperatures higher than that of Li2(BH4)2NH3BH3 and Ca(BH4)2(NH3BH3)2 materials. Li2(BH4)2NH3BH3 and Ca(BH4)2(NH3BH3)2 are able to release ~ 5.5 and ~ 4.5 equivalents of H2 suggesting the second desorption results from the decomposition of the borohydride units. After complete desorption only amorphous patterns were shown via powder XRD. Figure 3.3 (a) Crystal packing of Li2(BH4)2NH3BH3. (b) Crystal packing of Ca(BH4)2 + 2+ 86 (NH3BH3)2. (c) Coordination environment of Li . (d) Coordination environment of Ca . Reprinted with permission from ref 86. Copyright 2010 Royal Society of Chemistry. 77 Figure 3.4 TPD results for H2 release for Li2(BH4)2NH3BH3 and Ca(BH4)2(NH3BH3)2 with 2°C/min heating rate to 450°C along with pure NH3BH3, LiBH4 and Ca(BH4)2. The gas released from Li2(BH4)2NH3BH3 and Ca(BH4)2(NH3BH3)2 (top) is normalized to nH2 g/mol 86 NH3BH3. Reprinted with permission from ref 86. Copyright 2010 Royal Society of Chemistry. 78 After dehydrogenation, rehydrogenation was attempted in order to probe the regenerability of the materials, however, neither 2LiBH • AB nor CaBH • 2AB could successfully be rehydrogenated at 50 bar H2 and 300°C. Partial rehydrogenation was possible at - 82 bar H2 and 400°C, which is thought to be due to BH4 reconstitution, though these conditions are well outside the ideal low-pressure adsorption criterion stated earlier. Upon rehydrogenation the materials remained amorphous. 3.3 Crystal Engineering Using Various Dihydrogen bonding donors and acceptors 3.3.1 + Preparation and X-ray Crystal Structures of [Na@peraza[2.2.2]cryptand] BH4 - Explorations began with NaBH4 and peraza[2.2.2] cryptand. This system was chosen for several reasons: 1) Familiarity with the chemistry of sodium borohydride via previous studies of dihydrogen bonding-assisted reductions of carbonyls. 87-88 2) Known crystal structures from + other work proved that the peraza[2.2.2]cryptand cage could accommodate the Na cation. 12 3) Sodium borohydride is readily available and inexpensive in comparison with other borohydrides such as LiBH4. 4) The use of a large cation such as that formed by a metalperaza[2.2.2]cryptand complex and a small counter anion like borohydride could ensure small changes in the crystalline lattice upon evolution of H2 (Figure 3.5). 5) Given that peraza[2.2.2]cryptand has six secondary amine sites allowing for multiple amine-borohydride 79 interactions, it has the possibility for multiple moles of H2 to be formed upon heating per mole of cryptand. The proposed synthesis is shown Figure 3.5 Illustration of a large globular cation (large spheres) occupying the bulk of the space in a crystal lattice leaving tetrahedral ‘holes’ (small solid circles) for small anionic partners. below in Scheme 3.3. Reaction mixtures were made in a 1:1, 2:1 and 3:1 ratio of borohydride:peraza cryptand • 4H2O using sodium borohydride as well as sodium cyanoborohyride. Sodium cyanoborohydride was investigated due to its decreased reactivity in the presence of water and other protic solvents. In all cases, a significant upfield shift was seen in the assigned N-H peak from ~3.3 ppm in the free ligand ~1.5 ppm. An upfield shift in the NH peak is expected since the borohydride will provide additional shielding to the proton on the nitrogen. IR studies were also performed on each of the products. KBr pellets of each of the 80 complexes were made but the results were inconclusive given that the peraza ligand has a large -1 water band centered at ~3000 cm . Synthesizing complexes that were completely water-free, even when using sublimed peraza[2.2.2]cryptand and other water-free starting materials under inert atmosphere (N2 stream and dry-box), proved extremely difficult. N HN HN NH + NaBH4 HN HN NH N Solvent NH N HN N Na+ HN N NH N H H H B H H ! H N BH3HN N NH + 2H2 Na+ NH HN HN N (1) Scheme 3.3 Formation of dihydrogen-bonded (only one borohydride/macrocycle shown for clarity) peraza[2.2.2]cryptand and loss of H2. Single crystals suitable for X-ray crystallography were obtained by slow diffusion of benzene into a solution of (1) in a 10:1 THF:DCM solution. The lattice parameters of all crystal structures reported in this work can be seen in Appendix A. The crystal structure is shown in Figure 3.6. 89 The complex exists in the highly symmetric rhombohedral space group R3c with + the Na ion bound in the cage in a distorted octahedron (Na---N = 2.97Å). The boron of the € borohydride anion is disordered about a three-fold axis (Figure 3.7), making it impossible to assign the positions of the hydrogens bonded to the boron or the oxygen. However, given the boron’s proximity to the oxygen (2.78Å) and the fact that the closest nitrogen is tertiary and is at - a distance of 4.070Å, is was assumed that the BH4 ion is not participating in dihydrogen 81 bonding interactions with the macrocyclic cage. Instead, it appears to be interacting with the water molecule. + - Figure 3.6 X-ray crystal structure of [Na@peraza[2.2.2]cryptand] BH4 • H2O. + - Figure 3.7 Crystal packing of [Na@peraza[2.2.2]cryptand] BH4 • H2O. 82 89 89 Attempts to produce X-ray quality crystals without the inclusion of water molecules proved unsuccessful. The use of sublimed peraza[2.2.2]cryptand, water-free borohydrides (dried and stored under vacuum), distilled solvents and inert reaction environments (under N2 and in a dry box) only produced amorphous solids. Heating the water-containing crystal resulted in an amorphous solid at ~ 55°C with no change in the crystal lattice prior to loss of crystallinity. Figure 3.8 ORTEP depiction of [H2O@H3peraza[2.2.2]cryptand] - 3+ - - 90 2I • NaI2 (H2O)3. - - Since the BH4 ion is approximately the same size as I and assuming BH4 is spherical, +- attempts were made to produce a single crystal of [Na@peraza[2.2.2]cryptand] I . This venture produced an unforeseen crystal structure, which is shown in Figure 3.8. 83 90 Single crystals suitable for X-ray crystallography were obtained by slow evaporation of water. The complex exists in the P2/c space group with one H2O molecule hydrogen bonded in a distorted octahedron within the cavity. Hydrogen bonding interactions in the cavity can be seen in Figure 3.9 and are summarized in Table 3.2. One H2O molecule is outside the cavity and hydrogen bonded to one of the three protonated secondary amines in the cryptand (NH---O = 1.94Å ) and to one of the three iodides (OH---I – 2.98Å). Figure 3.9 ORTEP representation of hydrogen bonding interaction within the cavity of 3+ 90 [H2O@H3peraza[2.2.2]cryptand] 2I NaI2 (H2O)3. 84 Table 3.2 Hydrogen bonding interactions within the cavity of 3+ [H2O@H3peraza[2.2.2]cryptand] 2I NaI2 (H2O)3. Atoms Distance (Å) O1S---H2B 1.940 O1S---H3 2.272 O1S---H5B 2.215 O1S---H7A 2.103 N4---H1S 1.944 N6---H2S 1.829 - What’s most interesting is the formation of NaI2 • 3H2O clusters in the structure (Figure 3.10) . The two iodides (Na---I = 3.28Å) and three waters (2 at Na---OH = 2.27Å and 1 at Na---OH = 2.45Å), are arranged around the sodium ion in a distorted trigonal bipyramidal geometry (Fig. 3.10). The Na---I bond lengths are comparable with that of the NaI hydrate crystal (3.19 – 91 3.22Å) - and therefore it is believed that the NaI2 • 3H2O anion is a reasonable structure. The - - set of 4 bond angles for the NaI2 • 3H2O are summarized in Table 3.3. The iodides of NaI2 • 3H2O appear to be hydrogen bonded to one of the three protonated secondary amine sites, each to a different peraza[2.2.2]cryptand cage (NH---I distances of 2.66Å). 85 O1 O2 I2 Na O3 I1 - Figure 3.10 ORTEP depiction of NaI2 (H2O)3 clusters. 90 - Table 3.3 Bond angles in NaI2 (H2O)3 clusters. Atoms Angle Atoms Angle I1 – Na- I2 110.97° I2 – Na – O2 124.52° I1 – Na – O1 103.12° I2 – Na – O3 103.12° I1 – Na – O2 124.52° O1 – Na – O2 77.80° I1 – Na – O3 90.71° O1 – Na – O3 155.61° I2 – Na – O1 90.71° O2 – Na – O3 77.80° With respect to the encapsulated water, two similar crystal structures have been reported (Fig. 3.11). 92,93 The first was reported by Bowman-James et al. of a tetra-protonated 86 peraza[2.2.2]cryptand ([H2O@ H4peraza[2.2.2]cryptand] water in the cavity. 92 4+ - 4I • 4H2O) cage that included a Compared to [H2O@H3peraza[2.2.2]cryptand] 3+ - 2I • NaI2(H2O)3, this structure a had similar NH---O distance (1.95Å compared to 1.94Å) and OH---N distances of 2.10Å and 1.66Å corresponding to the water encapsulated within the cryptand cavity. The unit cell was completed with four additional waters and four iodide anions outside the cryptand, which had OH---I distances ranging from 2.36 to 2.61Å, indicating hydrogen bonding interactions. Figure 3.11 Crystal structures of water encapsulated by peraza[2.2.2]cryptand by Bowman92 4+ 93 James et al. [H2O@H4peraza[2.2.2]cryptand] 4I • 4H2O (top left) and McKee et al. 3+ - [H2O@H3peraza[2.2.2]cryptand] 3Cl •6H2O (top right). Reproduced with permission from refs 92 & 93. Copyright 2002 Elsevier (ref 92) and Copyright 2005 International Union of 3+ Crystallography (ref 93). Crystal structure of [H2O@H3peraza[2.2.2]cryptand] 2I NaI2 (H2O)3 for comparison (bottom). 87 The second structure, reported by McKee et al. 93 also showed NH---O bond distances of 1.90Å, 2.19Å and 2.21Å and OH---N distances of 1.84Å and 1.86Å. The unit cell was completed with 6 additional waters and 3 chloride ions outside the cryptand, which had OH---Cl distances of ~ 2.30Å. A search of the Cambridge Structural Database did not turn up any other occurrences - of the NaI2 anion as in the current crystal structure; those structures containing sodium bonded to two (or more) iodides were structures where the iodide atoms were bridging two metal atoms, though its existence was speculated in 1940 when Cheeseman et al. published a study on sodium polyiodides synthesized by mixing NaI with I2 in water at room temperature and allowing the 94 solution to equilibrate at 0°C. One of the products present at equilibrium was thought to be - NaI2 • 3H2O, whose presence was inferred from analysis of the solution after several months of equilibration time. Our inability to produce water-free crystalline material of [Na@peraza[2.2.2]cryptand] - + - BH4 led to the hypothesis that perhaps the BH4 anion was not large enough to span the large voids in the interstices of the crystal lattice and therefore water was necessary for crystallization. Based on this idea, new directions were pursued that included: 1) Use of hydroxyl groups as dihydrogen bond acceptors rather than amines. 2) Synthesis of larger hydridic partners for the peraza[2.2.2]cryptand. 3) Reversal of the large cation – small anion scheme to a large anion – - small cation scenario and deviation from the spherical nature of the BH4 . 4) Use of more rigid frameworks to possibly provide more control and directionality to the dihydrogen bonds. The implementation and results of a combination of each of these ideas follows. 88 3.3.2 Sodium [µ-(cyano-κC:κN]hexahydrodiborate To synthesize a larger, more rigid dihydrogen bond donor for peraza[2.2.2]cryptand, NaH3BCNBH3 was synthesized following the procedure of Wade et al. 95 as seen in Scheme 3.4 below. X-ray quality single crystals of this dihydrogen bond donor paired with peraza[2.2.2]cryptand were obtained from slow evaporation of THF but did not produce the expected result. 95 NaCN + 2 BH3 THF 00C to R.T., 5 hrs Na+H3BCNBH3(20) Scheme 3.4 Synthesis of Sodium [µ-(cyano-κC:κN]hexahydrodiborate by Wade et al. + - 95 Figure 3.12 ORTEP representation of [Na@peraza[2.2.2]cryptand] BH4 • H2O formed by mixing 1:1 ratio of peraza[2.2.2]cryptand with NaBH4 (left) and by mixing a 1:1 ratio of 96 peraza[2.2.2]cryptand with sodium [µ-(cyano-κC:κN]hexahydrodiborate (right) . 89 The X-ray structure (Fig. 3.12) 96 was essentially identical to that obtained by mixing peraza[2.2.2]cryptand with NaBH4 directly. As in the previous structure, this complex exists in the R3c space group and the geometry about the sodium is a distorted octahedron with Na---N distances of 2.97Å. The B---O distance is 2.79Å compared to 2.78Å in the previous structure € implying interaction between the water and the boron. The B---N distance was 4.10Å compared to 4.07Å in the previous structure. Unfortunately the time during mixing and single crystal formation apparently allowed the sodium cyanobisborane to disproportionate yielding NaBH4 and possibly CNBH2 ⋅ THF. The presence of NaBH4, as opposed to other boron species such as borates, was presumed given the fact that there is a sodium encapsulated within the peraza[2.2.2]cryptand cage and no other anion - was present in the structure. Therefore the presence of BH4 was the simplest explanation given the composition of the reaction mixture during crystal growth. It has been shown that in the 97 presence of HCl, a THF solution of NaCNBH3 at 0°C forms oligomers of BH2CN. H Na+ B H H H C N B H H H H Na+ B H C + N Na+ H B H H H H B H H + B C H H Scheme 3.5 Possible disproportionation of sodium [µ-(cyano-κC:κN]hexahydrodiborate. 90 N In a study by Gyori et al, 98 several amines including aniline, piperidine and pyrazole were reacted with these oligomers at room temperature in dimethylsulfoxide solution. Each yielded only (CNBH2)n – amine adduct. However, one can imagine another route to NaH3BCNBH3 disproportionation as depicted in Scheme 3.5. 3.3.3 2-Substituted Benzimidazole Boranes Another attempt was made at not only creating larger, more rigid dihydrogen bond donors but also to reverse the big cation – little anion scheme. For this, the synthesis of 2substituted benzimidazole boranes was employed. This framework was chosen for three reasons: 1) Ease of synthesis of the amine-borane and access to multi-gram quantities of several derivatives of the benzimidazole molecule (Scheme 3.6). 2) The flat, aromatic framework provided (a) directionality to the BH3 moiety giving more control over directed dihydrogen bonding and (b) an acidic N-H proton to possibly provide added structural stability to the overall complex through traditional hydrogen bonding. This led to the synthesis and characterization of several 2-substitued benzimidazole boranes. 99,100 Those benzimidazole boranes that could be 1 purified and fully characterized are summarized in Table 3.2. H NMR upfield N-H peak shifts from free benzimidazole and shifts to lower frequencies in the IR spectra suggest interactions -1 between borane hydrogens with amine protons. Also, three bands in the 2200 cm region of the IR spectra and are present. 11 B NMR peaks in the range of ~ -20 to -10 ppm indicate that N-BH3 moieties 101,102 Lastly, direct probe mass spectrometry showed mass peaks corresponding to 91 the m/z of the 2-substituted benzimidazole borane or diphenylamine borane and parent ion peaks that corresponded with the loss of the BH3 group. This class of boranes (excluding the diphenylamine borane) are air stable and relatively unreactive in protic solvents. An attempt was made to synthesize 1,3,5-tris(benzimidazol-2-yl)benzene-trisborane (30) (Scheme 3.7), however, confirmation of pure product was not obtained. BH3 NH2 + nR NH2 O N 1) o-phosphoric acid, ! OH 2) sat . NaHCO3 (aq) R N H N mBH3-THF 0oC to RT R N X (21-25b, 26) (21-25a) (21) n = 1; R = CH3; m = 1; X = H (22) n = 1; R = CH3; m = 2; X = BH3(23) n = 1; R = Ph; m = 1; X = H (24) n = 1; R = Ph; m = 2; X = BH3(25) n = 1; R = COOH ; m = 2; 2X = H (26) Commerical diphenylamine; m = 1 Scheme 3.6 General synthesis of 2-substituted benzimidazoles (21a – 25a) 101 borane derivatives (21b – 25b) 99,100 and their and diphenylamine borane 26. n = equivalents of o- phenylenediamine; m = equivalents of BH3 • THF. For (25) the 2 equivalents of ophenylenediamine react on each side of a p-terephthalic acid. 92 BH3 N N HN N H HN N H N N BH3 3 BH3 THF, 0oC --> R.T. N H3 B NH N NH (28) (27) Scheme 3.7 Synthesis of 1,3,5-tris(benzimidazol-2-yl)benzene-tris-borate. 100 Attempts to grow single crystals of the peraza[2.2.2]cryptand – 2-substituted benzimidazole boranes produced several suitable for X-ray analysis. The mixture of peraza[2.2.2]cryptand and 1,4-bis(benzimidazol-2-yl)benzene-borane produced an X-ray structure similar to that of the + - [Na@peraza[2.2.2]cryptand] BH4 • H2O, in that a disordered boron is centered around a three+ fold axis with one water interacting with the boron atom. In the place of the Na ion (for charge balance) was a protonated secondary amine (Figure 3.13). 93 103 Table 3.4 Selected Characterization Parameters for 2-Substitued Benzimidazole Boranes. 11 m/z B-H B Δν + + Δδ N-H stretch [M ], [M Amine-borane -1 NMR (ppm) -1 (cm ) (cm ) (ppm) - BH3] +1.3 110 -17.3 2400, 2307, 2242 146.2, 132.6 208.1 101 (207.4) 194.1 101 (194.3) +1.0 110 -15.7 101 (-22.0) 2396, 2295, 2251 +1.1 106 -15.9 2316, 2228, 2242 310 -21 2391, 2295, 2259 169 +0.012 9 All NMR taken in d6-DMSO 94 H2O Figure 3.13 ORTEP representation of the single crystal formed by mixing peraza[2.2.2]cryptand with NaBH4 in a 1:1 ratio (top). ORTEP representation of the single crystal formed by mixing peraza[2.2.2]cryptand with p-1,4-bis(benzimidazol-2-yl)benzene borane in a 1:1 ratio 103 (bottom) . This complex exists, as with the other two peraza[2.2.2]cryptand/borohydride structures, in the highly symmetric R3c space group. The encapsulated water is bound inside the cage in a distorted octahedron. It appears that this water is bound in the center of the cage given that the € oxygen is essentially equidistant from all secondary nitrogens, at a distance of 2.96Å, and 3.29Å from each tertiary nitrogen. Originally it was thought that the oxygen was a sodium ion given that they have the same number of electrons but refinement of the structure with a sodium ion in 2 the center gave poor R values. Since there is uncertainty with respect to the placement of the 95 encapsulated water molecule, O-H hydrogens are not shown. However, there is one hydrogen bonding interaction shown between the protonated cage nitrogen and the water oxygen (N-H---O 2.05Å. Though not shown due to the inability to solve for the hydrogens isotropically, there are likely several other hydrogen bonding interactions between the cryptand and the encapsulated water. Finally, given the distance of 2.80Å between the boron and the water outside the cage, - there is clearly interaction between the water and the BH4 . Following the reversal of the large cation- small anion scheme, attempts were made to grow single crystals of the 1,4-bis(2-benzimidazolyl)benzene-borane with guanidinium carbonate, guanidinium iodide, urea, ethylenediamine, N,N, N’,N’-tetrakis(2hydroxyethyl)ethylenediamine, tris(hydroxymethyl)aminomethane, ethyelene glycol and propylene glycol but proved unsuccessful. Single crystals suitable for X-ray analysis were obtained of the 2-methylbenzimidazole borane via slow evaporation of methanol. This compound exists in the Pnma space group. Intermolecular dihydrogen bonding was observed in the crystal structure where the 2methylbenzimidazole borane acted as both the dihydrogen bond donor (N-H) and acceptor (BH3). This structure is shown in Figure 3.11 and the crystal packing is shown in Figure 104 3.14. Short N-H---H-B distances of 1.758Å indicate strong dihydrogen bonding interactions between adjacent rings. In this structure, hydrogen atoms were solved for isotropically, which resulted in a short N-H bond length of 0.87Å. This value is common for 2-substituted benzimidazole structures according to a search of the Cambridge Structural Database. 96 100 The crystal packing shows (Figure 3.15) that 2-methylbenzimidazole boranes arrange in twodimensional sheets and there is no interaction between sheets. This poses an issue in assuring the retention of the 3-D structure and thus the crystallinity upon heating and H2 loss. Despite this concern the single crystal was heated and remained stable and intact until 80°C at which point the crystal would not stay attached to the mount. Due to this difficulty the solid was analyzed by powder XRD 105 to 130°C, which produced an amorphous solid (Figure 3.16) with no change in the powder XRD pattern until reaching an amorphous state. Figure 3.14 ORTEP depiction of 2-methylbenzimidazole borane. 97 104 Figure 3.15 Crystal packing of 2-methylbenzimidazole borane into 2-D sheets showing close N104 H---H-B distances of 1.758Å. 98 130°C 120°C 110°C Counts 100°C 80°C 60°C 40°C 2ϑ 105 Figure 3.16 Powder X-ray diffraction data of (21b) between 60°C and 130°C. Data for single crystals of the 2-phenylbenzimidazole borane were collected both here and at the Argonne Advanced Photon Source (APS) at Argonne National Laboratory in Chicago, IL, but the structure has been unsolvable within acceptable values of uncertainty thus far. 3.3.4 Other Amine-based and Hydroxyl-based Dihydrogen Bond Acceptors Commercially available N,N, N’,N’-tetrakis(2-hydroxyethyl)ethylenediamine (29), tris(hydroxymethyl)aminomethane (30), 3-methyl-1H-pyrazole (31), 2,5-dimethylhexan-2,5-diol (32), 2,4-dimethylpentan-2,4-diol (33) and (1R,2R,3S,4S,5S,6S)-cyclohexane-1,2,3,4,5,6-hexaol 99 (34), and two other dihydrogen bond acceptors were synthesized, 2,6-bis(benzimidazol-2yl)pyridine (35) and 2,6-bis(N,N-bis-2-hydroxyethylamino)-4-chloro-1,3,5-triazine (36) (Figure 3.17). HO OH H N NH2 N N N OH HO OH HO OH (30) (29) HO HO OH HO OH OH OH OH HO (32) (31) OH (33) (34) OH N N H HN N HO N OH N N N N N Cl (36) (35) Figure 3.17 Commercially available (29 - 34) and synthesized (35 – 36) dihydrogen bond acceptors. Compound 36 was originally intended to be a trisubstituted triazine following the adapted procedure of Dilliban et al. 106 seen in Scheme 3.8. 100 OH OH N Cl N Cl 1) NaHCO3, acetone/isopropanol HO 3 HO dropwise over 1 hr to keep temperautre between 60 - 80oC OH N H N N 70oC, 2) 5 hr 3) 80oC, 1hr Cl N N N HO OH N N HO OH Scheme 3.8 Synthesis of tri-substituted triazine adapted from Dilliban et al. 106 Despite many modifications, this synthesis only resulted in the disubstituted product (Scheme 3.9). Single crystals suitable for X-ray diffraction were obtained through slow evaporation of isopropanol solution of the disubstituted product. The resulting structure is seen in Figure 107 3.18. This compound exists in the Pnma space group and hydrogen bonding interactions are seen in the crystal packing diagram in Figure 3.19. Each hydroxyethyl ‘arm’ on the N,N-2hydroxylethyl moieties is hydrogen bonded to another on an adjacent arm of the molecule. O-H--H distances are all equal at 1.927Å. OH 1) NaHCO3, acetone/isopropanol N Cl N Cl N Cl HO 3 HO dropwise over 1 hr to keep temperautre between 60 - 80oC OH N H 70oC, 2) 5 hr 3) 80oC, 1hr HO N N N N OH N Cl (36) Scheme 3.9 Disubstituted triazine resulting from the adapted synthesis of Dilliban et al. 101 106 Figure 3.18 ORTEP representation of (36). 102 107 Figure 3.19 ORTEP representation of the crystal packing in (36). 107 An attempt was made to derivatize (36) by adding an N-phenylamine group in the 4 position of the triazine ring to yield (37) (Scheme 3.10) with limited success. While the desired 1 compound was visible via H and 13 C NMR, purification of the final product proved to be difficult. Attempts at recrystallization and column chromatography proved unsuccessful in purifying the compound. 103 OH NH2 1) Glyme, 1 hr NaH + 2) HO N N N OH HO N N N OH HO N N Cl (36) HO N OH N NH reflux, 24 hr (37) Scheme 3.10 Synthesis of 2,6-bis(N,N-bis-2-hydroxyethylamino)-4-(N-phenylamino)-1,3,5triazine (39). Dihydrogen bond acceptors 29 – 36 were mixed with various dihydrogen bond donors including LiBH4, NaBH4, KBH4, NaCNBH3, NaH3BCNBH3, (Me)4NBH4 and (Bu)4NBH4. After approximately 2 minutes of stirring 29 in the presence of NaBH4, a solid, white precipitate 1 - was formed. The H NMR showed the presence of both (29) and BH4 , however, single crystals of the solid were not obtained. Despite many attempts and several different techniques, all other donor – acceptor combinations failed to produce any precipitate or single crystals. 3.3.5 Computational Study The curious results of the 2-subsituted benzimidazole borane – peraza[2.2.2]cryptand combination forming very similar crystal structures as NaBH4 spurred a computational 104 - investigation into the stepwise hydration of the BH4 anion. While water-borohydride interactions are not what was originally intended it was thought that perhaps we could modify the original large cation – small anion approach. The modified approach relied on the large n+ n+ cation ([M @peraza[2.2.2]cryptand] ) dictating the overall crystal lattice structure, but the interstices formed between the cations, rather than accommodating a hydride-containing anion, - may simply provide small reaction spaces. Using the assumption that the BH4 anion is spherical - and approximately the same size as I , the prediction was made that the first solvation sphere of - - BH4 may contain as many as 6 water molecules. Various studies on the hydration of I have been performed using photoelectron spectroscopy (PES) 108-110 and neutron diffraction 109 . All - are in agreement that the first hydration sphere of I contains approximately 6 water molecules. The hydration of the borohydride anion was studied at the B3LYP level of theory using 111 the 6-311G++(d,p) basis set in Gaussian98 . Lowest energy conformations obtained for BH4 - • nH2O where n = 1 – 6 were investigated and the stabilization energies due to dihydrogen bonding (DHB) were calculated relative to the sum of the lowest energy conformations of water and borohydride at the same level of calculation. The lowest conformations seen below were the lowest energies obtained after several different starting points of the water molecule’s position - relative to the borohydride. For example, for the first BH4 /H2O calculation, starting points included first starting with one hydrogen on the water molecule pointing directly toward one - BH4 hydride within the sum of van der Waals radii. The second starting point was nearly 105 identical but the distance between the two hydrogens was double the van der Waals radii. The third starting point was where the hydrogen atom on the water was placed directly between two - of the BH4 hydride hydrogens within van der Waals radii and then outside van der Waals radii. Finally, the last starting point was having both hydrogens on the water point away from the - hydride hydrogens on the BH4 . Similar step were taken with each sequential addition of a water molecule. The results of these calculations are summarized in Table 3.5 and graphical representations of each of the six structures are presented in Figure 3.17. With the addition of one water molecule, a single dihydrogen bond is formed with a length of 1.694Å and the borohydride anion is stabilized by ~ 11 kcal/mol. This value is slightly higher than the average heat of interaction of a dihydrogen bond of ~ 4 – 7 kcal/mol value of 10.3 kcal/mol obtained by Meijer and Heuft. 112 77 but closer to the experimental The sequential addition of waters up to 4 added approximately 9.2 – 9.5 kcal/mol stabilization per water according to the authors. 112 In this study, upon the addition of second water molecule to the borohydride anion the stabilization - 112 energy increases to 21 kcal/mol (~20 for I ) and there are four DHB interactions with distances of 1.722Å, 1.727Å, 2.326Å and 2.370Å. On average each DHB interaction provides ~ 5.2 kcal/mol stabilization, though shorter contacts imply stronger DHB interactions and likely provide more stabilization to the overall structure than the longer, weaker DHB interactions. The addition of a third water molecule increases the stabilization to ~31 kcal/mol (~29 kcal/mol for 106 - I ) and five DHB interactions whose lengths are 1.670Å, 1.743Å, 1.929Å, 2.340Å and 2.372Å. This scenario includes a traditional hydrogen bond between two water molecules. Accounting for an average traditional hydrogen bond of ~ 6 kcal/mol, each DB provides ~ 4.9 kcal/mol stabilization. When a fourth water molecule is added the stabilization energy increases to ~41 - kcal/mol (~38 kcal/mol for I ) but the DHB interactions remain at five and traditional hydrogen bond interactions increase to two. The length of each DHB is 1.752Å, 1.767Å, 1.773Å, 1.786Å and 2.386Å. Again, accounting for each traditional hydrogen bond, each DHB provides ~5.7 kcal/mol of stabilization. This larger per DHB stabilization energy value is higher than that of both the three and four water molecule scenarios likely because there are a larger number of closer DHB contacts. Five water molecules increases the stabilization to 48.47 kcal/mol, the DHB contacts to six and the traditional hydrogen bond contacts to three. The DHB lengths are 1.683Å, 1.799Å, 1.802Å, 1.877Å, 2.238Å and 2.337Å and on average contribute ~ 5.1 kcal/mol stabilization. Finally the addition of a sixth water molecule increases the stabilization to 55.01 kcal/mol. This increase in stabilization is largely due to the addition of the additional traditional hydrogen bond as the sixth water molecule does not participate in any DHB interactions. Six DHB interactions remain interact at 1.799Å, 1.814Å, 1.819Å, 1.934Å, 2.197Å and 2.374Å. From these results it appears as though each borohydride anion can accommodate six DHB interactions with water. - The formation of a cyclic water trimer in the structures seen below of BH4 (H2O)n where n = 3 – 6 is very common. 113-115 These types of clusters have been found around halogen atoms in X-ray scattering experiments 113 as well as in computational studies using density 107 CI-H=2.16 A CI-0=3.15 A CIHO=168.66 HOH=101.39 39.85 32.40 66.00 23.30 59.50 113 Experimental functional methods nd 13. and ab initio MO methods 114 - . What’s interesting in the BH4 (H2O)n case is that only one water trimer forms where n = 6. ctronic energy he energy for e cluster. The n Table 4. We d by 0.45 eV ates that 0.122 ectrons on the molecule. This gy (-71 k c d t of the highly ss of electron ransfer within owing similar Figure 4 Optimized geometries (DIT)114 Cl-(H20) and Cl-(H20)3 . for drogen atoms, Figure 3.20 Optimized geometry (DFT) for Cl (H2O)3.represented by dotted lines as Reprinted with permission. clusters. The Cl--HZO interactions are cluster by cu. Copyrightwell as the hydrogen bonding between the water molecules. 1995 American Chemical Society. y of formation e is, however, would expect, based on studies of halide anion – water clusters, thatlocated below shape (sp3 hybridization), each water molecule is when there are waters One value of 23.3 and in between two lone pairs of electrons. That is, each water in close proximity to a interacts strongly with at conformation should be where there molecule spherical anion, the lowest energy least two lone pairs of y of F-(H20),, electrons. Of course, the lone pair of electrons located on top are H----X hydrogen bonds anion donor hydrogen least with the water molecules. the f F- with the of the fluoride as one interacts the per water, stabilizing the anion, and that the hydrogen This is reflected in the analysis of the second-order perturbation remaining hydrogens would form these hydrogen-bonded trimers. Figure 3.20 shows the ed to the other interaction energy terms originating from the CT process. For emains strong optimizedinstance,(DFT) of a chlorideof thewith three water molecules.on thepresence of geometry the interaction anion lone pair of electrons In the anion with water monomer (2) produces a stabilization energy due to X = halides) three additional waters one may expect another water trimerkcdmol;the top to OH*the F-(H20)3 CT as follows: LP1 to OH*(2) -1.53 to form on LP2 face of the uced by 0.1 8, (2), -15.5 kcdmol, LP3 to OH*(2), -1.27 kcdmol; and LP4 chloride anion. In the case of BH4 (H2O)6 this is not the case. Many attempts at trying to d by using ab to OH*(2), -13.47 kcal/mol. This implies that the water tically affects a lower energy structurelocated between LP2 and LP4. Similar results obtain monomer (2) is where there are two water trimers surrounding the borohydride 0 interaction are obtained for the other two water units. anion This Chloride-Water Clusters. Next, we studied not same een the water failed. 3.6. leads to one of two conclusions: 1) The structure obtained isthe a true ce is increased type of interactions between a less electronegative anion C1minimum and that, despite many attempts, the double water trimer is the lowest energy and a water molecule. Also the size of this anion is cu. 0.5 A study. These these types of larger than the fluoride anion (ionic radii). In Figure 4,we 108 o the hydrogen present the optimized geometry obtained at the DFT level for ong hydrogen the cluster Cl-(H20). Similar to the F-(H20) cluster the C1-of the water - conformation of the BH4 (H2O)6 cluster. 2) Since dihydrogen bonding (and hydrogen bonding - in general) is a directional bond, perhaps the assumption that the BH4 anion is spherical and should exhibit similar bonding as seen in a hydrated iodide anion not entirely correct. Table 3.5 Calculated Dihydrogen Bond Distances and Stabilization Energies of the Hydrated Borohydride Anion Number of H2O Molecules Stabilization* (kcal/mol) H---H Bond Length 1 H3 - H7 1.694Å 11.10 2 H2 – H11 H5 – H11 2.326Å 1.722Å H3 – H7 H4 – H7 1.727Å 2.370Å 20.98 3 H4 – H8 H3 – H11 H4 – H14 1.929Å 1.670Å 2.372Å H5 – H14 H2 – H11 1.743Å 2.340Å 30.59 4 H5 – H14 H3 – H14 H4 – H8 1.752Å 2.386Å 1.767Å H2 – H16 H3 – H11 1.773Å 1.786Å 40.60 5 H5 – H14 H3 – H14 H4 – H20 1.683Å 2.337Å 2.238Å H4 – H8 H2 – H16 H3 – H11 1.877Å 1.802Å 1.799Å 48.47 6 H3 – H11 H4 – H8 H2 – H16 1.819Å 1.934Å 1.814Å H3 – H14 H5 – H14 H5 - H20 2.374Å 1.799Å 2.197Å 55.01 109 - Figure 3.21 Graphical representation of the lowest energy conformations of nBH4 where n = 1 – 4. 110 - Figure 3.22 Graphical representation of the lowest energy conformations of nBH4 where n = 5 – 6. 111 While successful H2 storage for all materials presented here proved unattainable, the discovery of several crystal structures that have not been reported to date were made and the obtained 2-methylbenzimidazole borane crystal structure shows a promising direction to explore. The ease of synthesis, crystallization and thermal stability up to ~ 110°C are positive attributes of 2-methylbenzimidazole borane. While the crystal structure of this dual dihydrogen bond donor/acceptor shows strong hydrogen bonding interactions between adjacent molecules, there is no evidence for supramolecular interaction between layers of dihydrogen-bonded sheets. Designing a system that has a similar chemical nature and thus potentially similar thermal stability, but provides more three-dimensional stability, is a new direction that should be pursued. 112 CHAPTER 4 EXPERIMENTAL SECTION 113 General: All reagents were obtained from commercial suppliers and used without further 1 purification unless stated otherwise. All products were characterized via H, 13 C and 11 B NMR, 1 FT-IR, direct probe MS, elemental analysis and X-ray crystallography. Solution-based H, and 11 13 C B (500 MHz only) NMR were obtained on either on 300 MHz (Varian Inova-300 Spectrometer operating at 300.103 MHz) or 500 MHz (Varian Inova-500 (up500) Spectrometer operating at 499.955 MHz). Solid state 11 B NMR was collected using a Varian Infinity-Plus 400 spectrometer operating at 128.4 MHz. FT-IR spectra were collected on a Perkin Elmer Spectrum 2000 FT-IR using KBr pellets. Direct probe MS were taken on a VG MASSLAB LTD Trio-1 (E1 & C1) Spectrometer operating , CHN/O analyses were run on a Series II CHN/O Analyzer 2400 and all X-ray crystal structures were obtained on an Apex II CCD Single Crystal Diffractometer running on copper radiation (Kα = 1.54 nm) using an Oxford Cryostream 700 Series low temperature device. Powder X-ray diffraction was obtained using a Bruker Davinci Diffractometer. Preparation of 1,4,7,10,13,16,19,22,25-octaazabicyclo[8.8.8]hexacosane (1). The following is 8 an adapted synthesis reported by Redko et al. To a 3 L 3-necked round bottom flask equipped with a mechanical stirrer and an N2 inlet was added 20.6 mL (0.14 mol) tris-(2aminoethyl)amine, 50 mL 99.5% triethylamine and 1L 99% 2-propanol. The mixture was cooled to -78°C and a slow, continuous stream of N2 gas was allowed to flow into the vessel. Then a 114 solution of 24.1 mL (0.21 mol) of 40% glyoxal solution in H2O diluted to 250 mL in 2-propanol was added dropwise (~2 drops/sec) to the reaction flask. Once all glyoxal solution was added, the reaction mixture was allowed to slowly warm to room temperature with stirring for 48 hours. Then, the resulting light yellow reaction mixture was cooled to -78°C and approximately 200 mL anhydrous NH3 was condensed into the solution. Condensation was interrupted while 49 g of Na metal, cut into 1 cm x 1 cm x 1 cm cubes, was added to the reaction flask. With vigorous stirring, the condensation of NH3 gas was resumed until a homogeneous dark blue solution with golden flecks appeared (~500 mL NH3), after which the stirring was slowed and the reaction was allowed to slowly warm to room temperature and stirred overnight. Then 300 mL of toluene and 200 mL distilled water were added SLOWLY to re-dissolve the resulting off-white solid product (WARNING: Unreacted Na metal reacts violently with H2O forming H2 gas and may possibly ignite). The mixture was heated in a water bath for approximately 1-2 hours to drive off excess NH3 gas. Then the product solution was added to a 2 L separatory funnel and allowed to settle for 1 hour. The organic layer was extracted and dried. The product was separated from the resulting brown solid by adding a minimal amount (~100 mL) of ice-cold water. The white to off-white tetrahydrate product precipitated from solution. Purification of the solid was attained by sublimation of the tetrahdydrate at 0.1 torr at 100°C. Yield (hydrate) 60%. Melting point 1 (hydrate) 104°C. H NMR (500MHz, d6-DMSO) δ 2.56 (m (overlapping peaks), 24H), 2.36 (t, 12H), 1.60 (bs, 6H). 13 C NMR (125 MHz, d6-DMSO) δ 52.1, 49.7, 46.7. MS calcd m/z for + ˜ [C18H42N8] ([M ]) 370.5, found MS m/z 370.3. IR (KBr pellet) υ = 3530, 3259, 2946, 2883, € 115 2812, 1459, 1336, 1295, 1222, 1110, 1067, 958, 742, 569, 493. Anal. (C18H42N8): calcd C 58.33% H 11.42% N 30.24%; found C 59.55% H 11.95% N 29.26%. 4.1 Mercury Binding Study 64 Preparation of Poly(peraza[2.2.2]cryptand/α,αʹ′-dichloro-p-xylene) (19). A mixture of peraza[2.2.2]crypand•4H2O (10.45 g, 23.6 moles), α, αʹ′-dichloro-p-xylene (12.41 g, 71 mmol), dicyclohexylmethylamine (30.00 g, 154 mmol) and polyethylene glycol (PEG) (33.0 g) were placed into a 100 mL flask, stirred and heated to 120 °C under a nitrogen atmosphere for ~ 20 min. The flask was cooled to room temperature and the resin product was ground into ~ 1 cm 3 pieces. Then the ground resin was heated to 200 °C for 30 min under a nitrogen atmosphere and then cooled to room temperature for 3 hrs. The resin was removed from the flask, manually ground using a mortar and pestle, placed onto a fritted glass filter and washed with 200 mL isopropanol, 200 mL of deionized water, 500 mL 1M aqueous NaOH solution, 500 mL water, 200 mL isopropanol, 200 mL hexane. The resulting product was left on the filter to dry for 1 hr and then kept under vacuum for 24 hrs at room temperature. Yield: 9.18 g (57%). Anal. (C42H60N8Cl2.5 ⋅ 1.5 H2O): calcd C 66.21%, H 7.99%, N 12.87%, O 2.75%, Cl 10.18; found C: 65.90%, H: 8.20%, N: 13.71%, O 2.66% Cl 9.53% (C48H66N8Cl2.5 • 1.5 H2O). IR (KBr pellet) ˜ υ = 3412, 2930, 2810, 1646, 1457, 1345, 1130. € Preparation of Poly[polyethylene imine/α,αʹ′-dichloro-p-xylene] polymer (18) & (19). a 250 mL round bottom flask equiped with a magnetic stir bar and a reflux condensor, 116 64 To polyethyleneimine (21 g), polyethylene glycol (10 g), dicyclohexylmethylamine (12.4 g) and α,αʹ′-dichloro-p-xylene (5 g) were added. The solution was heated to 130°C and heated for 4 hrs (19) or 24 hrs (18). The resulting solid was cooled and rinsed with 500 mL deionized water, 500 mL ethanol and 250 mL dichloromethane. Both resulting yellow solids were placed under vacuum at room temperature for 24 hrs. Yield: 8.6 g (18), 4.3 g (19). (18): Anal. (C20H30N4Cl • 3H2O); calcd C 57.74%, H 8.72%, N 13.47%, O 11.54%, Cl 8.52%; found C 58.63%, H ˜ 7.53%, N 12.72%. IR (KBr pellet) υ = 3450, 2935, 2796, 1654. (19): Anal. (C12H23N4 • 1.25 H2O); calcd C 58.61%, H 10.45%, N 22.79%, O 8.13; found C 59.57%, H 10.22%, N 21.25%, O € ˜ 8.96%. IR (KBr pellet) υ = 3450, 3045, 2953, 1445. Preparation of Poly[peraza[2.2.2]cryptand//α,αʹ′-dichloro-p-xylene] polymer (17) € particles. 64 Particles of (19) were prepared by adding approximately 2.0 g of the polymer source material to 100 mL of ultrapure water. The mixture was sonicated for 30 h using a water bath sonicator (model 50T, VWR Aquasonic). The resulting opaque suspension was then filtered through a polycarbonate hydrophilic screen filter (Isopore, Millipore) with 0.22 µm nominal pore size using a pressurized stainless steel filtration cell (HP4750, Sterlitech Corp.) The cake of particles collected from the filter surface was dried for 24 h and then was ground in a ceramic crucible. To assess the resuspension capacity of particles, a portion of the finely ground dry powder was resuspended in water and sonicated for 1 min. Aqueous suspensions of Aza222 particles used in all experiments were prepared by sonicating this fine powder in water for ca. 1 min. 117 SEM characterization for (17). 64 SEM micrographs of the (19) particles were recorded using a Hitachi S-4700II field emission scanning electron microscope operated in an ultra high resolution mode. The SEM samples were prepared by distributing the particles over the surface of the carbon coated adhesive tape mounted on an aluminum stub. The samples were coated with pure osmium for 20 s at a current of 10 mA using an osmium plasma coater (NEOC-AT, Meiwa Shoji Co., Osaka, Japan). 64 Particle size distribution measurement for (17), (18) & (19). Particle size distribution and particle fractal dimension were measured using a light diffraction apparatus (Mastersizer 2000, Malvern Instruments, Worcestershire, UK). The suspension was introduced into the optical cell using a sample dispersion unit (Hydro 2000 SM, Malvern Instruments, Worcestershire, UK). In this method a "halo" of diffracted light is produced when a laser beam passes through a dispersion of particles in a liquid. The angle of diffraction increases as particle size decreases. Nitrogen adsorption test for (17) & (18). 64 The total surface area of sorbent particles was determined by BET nitrogen physisorption (Micromeritics ASAP 2010, Micromeritics Instrument, Norcross, GA) at 78 K over the (0.0 to 0.2) range of the relative pressure, P/P0. Surface area was calculated from the BET equation. Micropore volume for (17) was determined using the t-plot method and total pore volume was characterized as volume adsorbed at the maximum relative pressure of 0.99. 118 Determination of particle porosity for (17). 64 The porosity of the sorbent particles was calculated from the results of N2 adsorption and settling experiments using the assumption that both the aggregates and the primary particles are spherical and that the primary particles are monodisperse. Atomic absorption analysis for (17), (18) & (19). 64 The aqueous concentration of mercury (II) in sorption and desorption (regeneration) studies was measured using Varian SpectrAA-200 flame atomic absorption (AA) spectrometer. Solutions of HgCl2 in 1 mM aqueous HNO3 containing 1, 3, 10, 30, 100, 300, and 1000 mg(Hg)/L were used as calibration standards. Low mercury (II) concentration analysis (ppb) was measured using cold vapor atomic absorption with a Cetac M6000A using high purity N2 carrier gas. Commercially available solutions (Specpure plasma standard solutions, Alfa Aesar) containing 0.025, 0.100 and 0.500 ppb mercury (II) were used as calibration standards. Adsorption kinetics measurements for (17). 64 Completely mixed batch reactor tests were carried out to determine the kinetics of mercury sorption by (17) sorbent particles. Kinetic adsorption experiments were done using a Hg feed value of approximately 150 ppm. This value was chosen to be at least a 3-fold excess of the theoretical Hg capacity of the polymer in the batch. The feed value was still well within the specific sorption (flat) portion of the isotherm, and the excess was chosen to ensure that all polymer sorption sites were occupied by mercury. Fixed amounts (~ 1.0 g) of HgCl2 were added to 150 mL beakers each containing 120 mL of 0.1M NaI 2- (a), which was done to form the highly stable HgI4 complex to ensure that this would be the 119 predominant species present in solution This approach avoided complex mixtures of mercuric chloride ions, hydrolyzed species and protons. Then 20 mL of the solution from each beaker was separated for determination of mercury content by AA. A fixed mass of dry polymer powder was added to the remaining 100 mL of feed solution. Immediately upon the addition of the sorbent, the suspension was sonicated to disaggregate the particles. The reaction beakers were then stirred continuously and 6 mL suspension samples were periodically withdrawn from each reactor. All samples were filtered through polyvinylpyrrolidone-coated polycarbonate tracketched membrane filters (Whatman, Nuclepore) with a 0.1 µm nominal pore size and the filtrates were analyzed by AA. Each adsorption kinetics experiment was performed in triplicate. A similar method was used for low concentration adsorption experiments. Adsorption kinetics modeling for (17). 64 Numerical modeling was employed to evaluate local (surface adsorption parameters by comparing model predictions and experimentally observed kinetics. The method employed for the numerical modeling was the homogeneous surface diffusion model (HSDM) as adopted from studies by Hand et al. on suspended sorbents in completely mixed batch reactors. 35 In a completely mixed batch system with dissolved adsorbate and suspended particles of microporous sorbent, the adsorbate is removed from the liquid phase as it partitions into and diffuses within the porous matrix of the sorbent particles. Assumptions regarding the mass transfer process were developed, which allowed for a simplification of the numerical algorithm: 1) there is only monolayer binding, that is only on the surface of the pores, 2) surface diffusion can be described by Fick’s law, which postulates that the flux of particles in a solution goes from high concentration to regions of low concentration, with a magnitude that is proportional to the concentration gradient, 3) the porosity of sorbent particles has the same value 120 for all particles and for all locations within each particle, 4) adsorbate mass flux within the liquid phase is driven by the difference between the liquid bulk concentration and the concentration at the sorbent’s surface, 5) local adsorption equilibrium occurs at the solid-liquid interface, 6) the local equilibrium at the liquid-solid interface can be described by a Freundlich isotherm that relates the adsorption capacity of a sorbent as a function of adsorbate concentration. Adsorption isotherm measurements for (17). 64 Adsorption isotherm measurements were conducted encompassing a wide concentration range, from 1 to 2000 ppm, in order to have a comprehensive knowledge of the adsorption behavior of the material at both low and very high concentrations. The study was done to gain a better understanding of the adsorption mechanisms at a wide range of concentrations, rather than emulating a particular feed water type at a particular effluent concentration. Two sets of adsorption isotherm measurements were carried out. In the first set mercury was in excess with respect to (17) while in the second set (17) was in excess with respect to mercury. In the first set of experiments, the feed solutions were prepared by dissolving HgCl2 in 70 mL of a 1M NaI (a) solution, which was used to maintain the Hg/I molar ratio used in the kinetic experiments. 20 mL of the feed solution was separated for flame AA analysis. Five Teflon-capped stirred scintillation vials were filled with 10 mL of the feed solution and different amounts of the dry sorbent. The suspensions were sonicated to disaggregate the sorbent particles. After 24 h of continuous stirring, 1 mL of suspension was separated and added to a vial with 9 mL of ultrapure water. The resulting diluted suspensions were filtered through a 0.1 µm polycarbonate filter and the filtrate was analyzed using flame AA. The amount of adsorbed mercury was calculated as the dilution-corrected difference between the initial (feed) concentration and the measured equilibrium concentration of mercury in each vial. 121 In the second set of experiments, the feed solutions were prepared by dissolving HgCl2 in a 1M NaI solution. Four Teflon-capped stirred scintillation vials with different HgCl2 feed concentrations were filled with approximately the same amount of the dry sorbent. The rest of the procedure was the same as in the experiments from the first set. The amount of mercury adsorbed by the particles ( ) was expressed in mg/g as mass of adsorbate (mercury) per mass of sorbent ([α,α’-dichloro-p-xylene/polyethylene imine] polymer). 64 Adsorption isotherm modeling for (17). The obtained isotherm data was fitted using the Langmuir and Freundlich models. The non-specific physisorption portion of the isotherm was excluded from the calculations as both Langmuir and Freundlich models assume all sorption sites have similar adsorption energies. 64 Low concentration adsorption of (17). In order to test the polymer’s performance under low concentration conditions, 0.2 mg of the polymer was subjected to 25 mL of a ~2.5 ppm solution of Na2HgI4, for six hours at room temperature with constant stirring. The feed solution was prepared by dissolving 2 mg NaI in 35 mL of a 2.5 ppm HgCl2 (a) solution for 1 hr. Approximately 10 mL of the feed was separated for cold vapor AA analysis. Once equilibrium was attained, a 15 mL sample of the supernatant was taken and filtered through a 1 µg polycarbonate filter. Cold vapor atomic adsorption (AA) measurements of the initial feed solution as well as the supernatant were taken. 122 Hg 2+ 64 adsorption by (18) and (19). As in the experiments with (19), mercury was present in excess with respect to (18) and (19). The feed solutions were prepared by dissolving HgCl2 in 70 mL of a 1M NaI (a) solution, which was used to maintain the Hg/I molar ratio used in the kinetic experiments. 20 mL of the feed solution was separated for flame AA analysis. Two (one for each polymer) Teflon-capped scintillation vials were filled with 10 mL of the feed solution and 0.300 g of either (18) or (19). The suspensions were sonicated to allow for disaggregation of sorbent particles. The suspensions were filtered through a 0.1 µm polycarbonate filter three times to ensure no polymer particles were contained in the filtrate. The filtrate was analyzed using flame AA. The amount of adsorbed mercury was calculated as the difference between the initial (feed) concentration and the measured equilibrium concentration of mercury in each vial. 64 Batch sorbent regeneration study for (17). To assess the regenerability of the sorbent, a fixed amount of (17) particles were dispersed with sonication in a HgCl2 solution prepared to stoichiometrically saturate the polymer’s estimated sorption capacity (20% of its mass) with no mercury excess so that the regeneration was for chemisorbed mercury ions only (not physisorbed). For this reason, regeneration studies were performed using a Hg concentration of 300 ppm. While concentrations of mercury in the ppb range could have been used, this would required using lower amounts of polymer (to maintain the stoichiometric relation), and this would have been experimentally difficult to measure with reasonable accuracy. The dispersion was followed by 24 h of continuous stirring to achieve equilibrium between liquid and adsorbed phases. The sorbent particles were filtered out from the suspension using a 0.1 µm polycarbonate 123 filter. The filtered sorbent was then rinsed three times with ultrapure water to remove all nonsorbed mercury. The effectiveness of the rinsing procedure was confirmed by AA analysis, as mercury levels in the rinsate were below the detection limit (3 ppm) after the third rinsing step. The obtained filtrate and the HgCl2 feed solution were both analyzed by AA, to quantify the amount of mercury removed by adsorption onto the polymer from the feed solution. The filtered polymer was dried and then weighed to account for possible mass losses during the filtration step. The polymer particles were subsequently redispersed in a saturated Na2S solution to allow for the release of the polymer-bound mercury into the liquid phase. This mercury desorption study was conducted under continuous stirring for 24 h, after which the polymer was filtered, rinsed, and dried using the same procedure as during the adsorption step. The filtrate was analyzed by flame AA to quantify the amount of desorbed mercury. The regenerated polymer was then digested using concentrated H2O2 and HNO3 until the sorbent particles were completely dissolved. The digestion was accompanied by the emission of dark vapor characteristic of nitric acid digestion. The digest solution was analyzed by flame AA to quantify the amount of residual sorbed mercury. Column regeneration study for (17). 72 A column was loaded with 66 mg of the microporous [α, α'-dichloro-p-xylene/polyethylene imine] polymer particles. Solutions of HgCl2 and Na2S (5 mL 0.39% aqueous HgCl2 and variable volumes of saturated aqueous Na2S) were passed through it in cycles. Water was passed between mercury and sulfide washes to rinse out the excess electrolytes and thus prevent formation of HgS from cross mixing. The eluted solutions 124 were diluted to 100 mL in a volumetric flask and analyzed by flame AA. After the end of the experiment the resin was digested in aqua regia and analyzed for mercury content. Column breakthrough was measured by mixing 62 mg of (17) with 0.54 g Celite-545 (Sigma-Aldrich) using a mortar and pestle. The obtained mixture was loaded into a column and a solution of ~ 400 ppm HgCl2 was passed through it in 10 mL portions. The eluent was collected in 20 mL vials and their mercury content determined by flame AA. 64 Mercury sorption in the presence of a competing divalent ion for (17), (18) and (19). To study the potential mitigation of the sorbent’s Hg capacity in the presence of competing divalent ions, a series of adsorption studies were conducted by adding Ca(NO3)2 salt to the HgCl2 feed solution. Calcium nitrate was added at a 6:1 (Ca:Hg) molar excess ratio to give a calcium concentration of ~ 200 ppm and a mercury concentration to stoichiometrically saturate the available sorbent. The excess was chosen to maximize the competing effect of Ca 2+ 2+ over Hg . A fixed amount of polymer particles (0.026 g) was dispersed in the feed solution with the aid of sonication. The dispersion was followed by 24 h of continuous stirring to achieve adsorption equilibrium. The sorbent particles were filtered out from the suspension using a 0.1 µm polycarbonate filter. The filtrate and the feed solutions were analyzed by AA to quantify the amount of mercury removed from the feed solution by adsorption onto the polymer. Identical control experiments were performed for calcium-free solutions. The performance of the polymer sorbent was compared with that of an existing commercial Hg sorbent on an equal mass loading basis and with (18) and (19) as well. 125 4.2 Dihydrogen Bonding Study Sodium [µ-(cyano-κC:κN]hexahydrodiborate (20): 96 To 0.35 g (7.14 mmol) sodium cyanide, dried in vacuo at 60°C for 36 hours and kept under an inert atmosphere, 20 mL of 1 M BH3 ⋅ THF was added with an oven-dried glass syringe and the solution was stirred under N2 for 5 hours in a room temperature water bath. The solution was filtered and the solvent removed from the filtrate at 50°C. The resulting white solid was stored under N2. Yield: 62.1%. 11 B NMR ˜ (solid state) (128.4 MHz) δ -36.8 (m, 3H), -21.4 (q, 3H). IR (KBr pellet) υ = 2354 (B-H stretch), 2272 (B-H stretch), 2184 (CN stretch) -36.8. € Preparation of 2-Methylbenzimidazole (21a). 97 To a 250 mL round bottom flask, 2.16 g (20 mmol) o-phenylenediamine, 1.8mL (30 mmol) glacial acetic acid and 50 mL of 4M HCl (a) were added. The mixture was heated to reflux and stirred for 1 hour. Upon cooling of the mixture, NH3 gas was bubbled into the solution causing the product to precipitate from solution. The solution was filtered and the beige solid product was washed with distilled water. Yield 86%. 97 mp 173°C (lit 1 174 – 177°C) H NMR (500 MHz, d6-DMSO) δ 12.14 (bs, 1H), 7.41(bs, 2H), 7.08 (m, 2H), 2.48 (s, 3H). 13 C NMR (125 MHz, d6-DMSO) δ 150.6, 120.7 117.3, 110.0. 14.1. + MS calcd m/z for [C8H8N2] ([M ]) 132.16, found MS m/z 132.9. Anal. (C8H8N2): calcd C 72.7% H 6.1% N 21.2%; found C 70.26, H 5.04% N 23.77%. 126 2-Phenylbenzimidazole (24a/25a). 97 To a 250 mL round bottom flask, 2.60 g (24 mmol) o- phenylenediamine, 2.44 g (20 mmol) benzoic acid and 8.5 g polyphosphoric acid (3-4 times by weight of benzoic acid) were added. The mixture was heated to 175°C and stirred for 1.5 h. Upon cooling of the mixture, 7% aqueous NH4OH was added to the blue-green solid with stirring upon which the light brown product formed a suspension. The solution was filtered and 97 the solid product was washed with 7% aqueous NH4OH solution. Yield 78.1%. mp 295°C (lit 1 293 - 296°C). H NMR (300 MHz, d6-DMSO) δ12.9 (bs, 1H), 8.16, (m, 2H, J=8Hz), 7.607.7.48 (5H), 7.19 (m, 2H, J=6Hz). 13 C NMR (125 MHz, d6-DMSO) δ150.7, 129.6, 129.3, 128.4, + 125.9, 122.0, 121.1, 118.3, 110.8. MS calcd m/z for [C13H10N2] ([M ]) 194.23, found MS m/z 194.4. Anal. (C7H6N2): calcd C 80.38% H 5.19% N 24.43%; found C 70.26, H 5.04% N 23.77%. Preparation of 1,4-bis(benzimidazol-2-yl)benzene (26a). To a 100 mL round bottom flask, 2.16 g (20 mmol) o-phenylenediamine and 1.66 g (10 mmol) terephthalic acid were combined with a minimum amount of 86% o-phosphoric acid. The mixture was stirred and heated up to 210°C for 4 hours. The resulting sky blue solution was poured into 100 mL distilled water upon which the solution became darker blue. To this solution, 5% sodium carbonate solution was added until no more precipitate formed. The dark teal-green solid was recrystallized from 1 distilled water and then dried in a 120°C oven. Yield: 78%. H NMR (500 MHz, d6-DMSO) δ 127 13.05 (bs, 2H), 8.34 (s, 4H), 7.63 (bs, 4H), 7.23 (dd, 4H). 13 C NMR (125 MHz, d6-DMSO) δ + 150.5, 130.9, 126.9, 122.4. MS calcd m/z for [C20H14N4] ([M )] 310.35, found MS m/z 311.1. Preparation of 1,3,5-tris(benzimidazol-2-yl)benzene (27). 100 To a 100 mL round bottom flask equipped with a magnetic stir bard and water-cooled reflux condenser, 0.21 g (1 mmol) trimesic acid, 0.324 g (3 mmol) and 10 mL 86% o-phosphoric acid were added. The mixture was heated to 210°C and stirred for 4 hours. Then, the reaction mixture was allowed to cool to near room temperature and then poured into 100 mL of distilled ice water whereupon the solution turned deep green in color. The reaction mixture was neutralized using solid K2CO3 until there was no more grayish-purple solid precipitate formed. The solution was filtered and 100 the solid dried in vacuo and upon drying turned grayish-brown. Yield: 68%. Mp > 240°C (lit 1 > 240°C) H NMR (300 MHz, d6-DMSO) δ 13.93 (bs, 3H), 9.30 (m, 2H), 9.20 (m 1H), 7.73 (m, 6H), 7.24 (m, 6H). 13 C NMR (125 MHz, d6-DMSO) δ 150.3, 135.2, 131.8, 125.4, 122.4, 115.9. + MS calcd m/z for [C30H25N6] ([M ]) 469.2, found MS m/z 469.1 (lit. 469.2). 99 Amine-Borane Complexes (22b, 23b, 24b, 25b, 26b, 27): 100 General Procedure. To an oven- dried round bottom flask equipped with a magnetic stir bar and a rubber septa, 1 mmol of amine was added and the flask subsequently cooled to 0°C in an ice bath. Then, 1 - 2 mmol BH3•THF (depending on number of N-BH3 substitutions) was added via an oven-dried glass syringe with 128 vigorous stirring. The reaction mixture was allowed to stir at 0°C for 15-20 min. and the solvent was removed in vacuo. 1 2-methylbenzimidazole borane (22b) H NMR (500 MHz, d6-DMSO) δ 13.40 (bs, 1H), 7.74 (m, 1H), 7.55 (m, 1H), 7.34 (m, 2H), 2.64 (s, 3H). 137.0, 131.0, 123.8, 123.1, 115.6, 112.0, 12.5. 11 13 C NMR (125 MHz, d6-DMSO) δ 151.3, B NMR (128.4 MHz, d6-DMSO) δ -17.3. MS + + + calcd m/z for [C8H11N2B] [M ] 146.1 found MS m/z 146.2 (24) [M ], 132.6 (100) [M - BH3]. ˜ IR (KBr pellet) υ = 3264, 2348, 2308, 2259, 1461, 742. € 99 1 2-phenylbenzimidazole borane (23b) H NMR (500 MHz, d6-DMSO) δ 13.91 (bs, 1H), 8.02 (m, 2H), 7.89 (m, 1H), 7.66 (m, 4H), 7.45 (m, 2H), 2.64. 13 C NMR (125 MHz, d6-DMSO) δ150.5, 137.9, 131.7, 131.1, 130.3, 128.2, 126.2, 124.7, 123.8, 116.7, 112.4. 11 B NMR (128.4 + MHz, d6-DMSO) δ -15.7. MS calcd m/z for [C13H13N2B] ([M ]) 208.1, found MS m/z 208.1 + + ˜ (26) ([M ]), 194.1 (100) ([M - BH3]). IR (KBr pellet) υ = 3283, 2922, 2372, 2308, 2253, 1453, 739. € Preparation of 2,6-bis(benzimidazol-2-yl)pyridine (35). To a 250 mL round bottom flask equipped with a magnetic stir bar and water-cooled reflux condenser, 3.34 g (20 mmol) 2,6pyridinedicarboxcylic acid, 4.32 g (40 mmol) o-phenylenediamine and 40 mL 86% o-phosphoric acid were added and heated and stirred at reflux for 24 hours. The reaction mixture was allowed 129 to cool to near room temperature and added to 200 mL of distilled ice water whereupon the mixture turned blue. The solution was neutralized with Na2CO3. The supernatant turned pinkish-orange as the same color precipitate formed. Addition of Na2CO3 ceased when no more precipitate formed. The mixture was filtered and the resulting pink solid was recrystallized in 1 methanol and dried in vacuo. Yield: 77.3%. H NMR (500 MHz, d6-DMSO) δ 8.33 (d, 3H), 8.15 (t, 1H), 7.45 (m, 4H), 7.29 (m, 4H). 13 C NMR (125 MHz, d6-DMSO) δ . MS calcd m/z for + [C19H13N5] ([M )] 311.34, found MS m/z 311.2. 101 Preparation of 2,6-bis(N,N-bis-2-hydroxyethylamino)-4-chloro-1,3,5-triazine (36). To a 250 mL 3-necked round bottom flask equipped with a magnetic stir bar, thermometer, addition apparatus and water-cooled reflux condenser, 9.22 g (50 mmol) cyanuric chloride (CAUTION! Reacts violently with water, toxic if inhaled, burns on contact.), 3.98 g (37.5 mmol) sodium bicarbonate, 15 mL isopropanol and 20 mL acetone were added. 15.77 g ethanolamine (150 mmol) was added dropwise to keep the reaction temperature between 60 - 80°C. When the temperature began to drop below 60°C after the addition of all the ethanolamine, the reaction was heated to 70°C for 5 hours after which the solution turned a light beige color. Then the solvent was removed and the resulting solid was redissolved in isopropanol and heated to 80°C for 1 hour. The solution was filtered and a light-brown solid resulted. The solvent was removed from the filtrate and the solid residue was combined with the first batch. The product was 1 recrystallized from isopropanol, which resulted in a white solid. Yield: 54.1% H NMR (300 MHz, d6-DMSO) δ 4.6 (bs, 4H), 3.5 – 3.5 (m, 16H). 130 13 C NMR (300 MHz, d6-DMSO) δ 168.1, ˜ 164.0, 58.6, 58.5, 50.7, 50.3. IR (KBr pellet) υ = 3401, 3043, 2952, 2912, 1587, 1463, 1163, + 1128. MS calcd m/z for [C11H20ClN5O4)] ([M )] 321.8, found MS m/z 321.1. € Preparation of 2,6-bis(N,N-bis-2-hydroxyethylamino)-4-(N-phenylamino)-1,3,5-triazine (37). To a clean, dry 100 mL round bottom flask 0.08 g (3.1 mmol) sodium hydride was added to 0.30 g (3.1 mmol) aniline in 25 mL glyme. The mixture was stirred at room temperature for 1 hour. Then, 1.0 g (3.1 mmol) (40) was added and the mixture was heated to reflux and stirred overnight. The reaction was removed from heat, allowed to cool to room temperature and the solvent removed. The resulting white solid was washed with isopropanol and acetone and then 1 dried in vacuo. Yield: 42.7% H NMR (500 MHz, d6-DMSO) δ 6.98 (t, 2H), 6.53 (d, 2H), 6.46 (t, 1H), 3.56 (bm, 11H), 3.41 (s, 1H), 3.34 (bs, 3H), 3.22 (s, 1H). DMSO) δ 168.0, 164.0, 148.6, 128.7, 115.5, 113.8, 57.9, 50.6. 131 13 C NMR (500 MHz, d6- APPENDIX A X-RAY CRYSTAL STRUCTURE LATTICE PARAMETERS 132 + - [Na@peraza[2.2.2]cryptand] BH4 • H2O Space Group: R3c Cell Lengths: a 8.5412(2) b 8.5412(2) c 59.7393(14) € Cell Angles: α 90.00 β 90.00 γ 120.00 Cell Volume: 3774.23 R-Factor(%): 8.07 + - [H2O@peraza[2.2.2]cryptand] NaI2 • 3H2O Space Group: P2/c Cell Lengths: a 16.1232(12) b 11.1522(9) Cell Angles: α 90.00 β 98.867 c 18.5870(14) γ 90.00 Cell Volume: 3303.06 R-Factor(%): 1.9 + - [Na@peraza[2.2.2]cryptand] BH4 • H2O from peraza[2.2.2]cryptand + NaH3BCNBH3 Space Group: R3c Cell Lengths: a 8.5437(10) b 8.5437(10) c 59.8211(9) € Cell Angles: α 90.00 β 90.00 γ 120.00 Cell Volume: 3781.61 R-Factor(%): 9.94 133 + - [H2O@peraza[2.2.2]cryptand] BH4 • H2O from peraza[2.2.2]cryptand + 1,4bis(benzimidazol-2-yl) borane Space Group: R3c Cell Lengths: a 8.5404(2) b 8.5404(2) c 59.6985(17) € Cell Angles: α 90.00 β 90.00 γ 120.00 Cell Volume: 3770.95 R-Factor(%): 6.54 2-methylbenzimidazole borane Space Group: Pnma Cell Lengths: a 12.546(5) b 6.798(3) c 9.687(4) Cell Angles: α 90.00 β 90.00 γ 90.00 Cell Volume: 826.182 R-Factor(%): 4.43 2,6-bis(N,N-bis-2-hydroxyethylamino)-4-chloro-1,3,5-triazine Space Group: Pnma Cell Lengths: a 15.026(2) b 19.985(3) c 4.7043(7) Cell Angles: α 90.00 β 90.00 γ 90.00 Cell Volume: 1412.68 R-Factor(%): 4.13 134 APPENDIX B REPRODUCTION PERMISSIONS 135 Figure AB-1. Reproduction permission letter for reference 58. 136 Rightslink® by Copyright Clearance Center 8/26/11 6:22 PM Title: Author: Dihydrogen Bonding:  Structures, Energetics, and Dynamics Radu Custelcean et al. Logged in as: Karrie Manes Account #: 3000425971 Publication: Chemical Reviews Publisher: American Chemical Society Date: Jul 1, 2001 Copyright © 2001, American Chemical Society No charge permission and attribution Permission for this particular request is granted for print and electronic formats at no charge. Figures and tables may be modified. Appropriate credit should be given. Please print this page for your records and provide a copy to your publisher. Requests for up to 4 figures require only this record. Five or more figures will generate a printout of additional terms and conditions. Appropriate credit should read: "Reprinted with permission from {COMPLETE REFERENCE CITATION}. Copyright {YEAR} American Chemical Society." Insert appropriate information in place of the capitalized words. This permission does not apply to images that are credited to publications other than ACS journals. For images credited to non-ACS journal publications, you will need to obtain permission from the journal referenced in the Table/Figure/Micrograph legend or credit line before making any use of the image(s) or table(s). Copyright © 2011 Copyright Clearance Center, Inc. All Rights Reserved. Privacy statement. Comments? We would like to hear from you. E-mail us at customercare@copyright.com Figure AB-2. Reproduction permission letter for reference 76. https://s100.copyright.com/AppDispatchServlet Page 1 of 1 137 Figure AB-3. Reproduction permission letter for reference 80. 138 Dear Dr Manes     The Royal Society of Chemistry hereby grants permission for the use of the material specified below in the work described and in all subsequent editions of the work for distribution throughout the world, in all media including electronic and microfilm. You may use the material in conjunction with computer-based electronic and information retrieval systems, grant permissions for photocopying, reproductions and reprints, translate the material and to publish the translation, and authorize document delivery and abstracting and indexing services. The Royal Society of Chemistry is a signatory to the STM Guidelines on Permissions (available on request).     Please note that if the material specified below or any part of it appears with credit or acknowledgement to a third party then you must also secure permission from that third party before reproducing that material.     Please ensure that the published article carries a credit to The Royal Society of Chemistry in the following format:     [Original citation] – Reproduced by permission of The Royal Society of Chemistry     and that any electronic version of the work includes a hyperlink to the article on the Royal Society of Chemistry website. The recommended form for the hyperlink is http://dx.doi.org/10.1039/DOI suffix, for example in the link http://dx.doi.org/10.1039/b110420a the DOI suffix is ‘b110420a’. To find the relevant DOI suffix for the RSC paper in question, go to the Journals section of the website and locate your paper in the list of papers for the volume and issue of your specific journal. You will find the DOI suffix quoted there.     Regards     Gill Cockhead   Contracts & Copyright Executive     Gill Cockhead (Mrs), Contracts & Copyright Executive   Royal Society of Chemistry, Thomas Graham House   Science Park, Milton Road, Cambridge CB4 0WF, UK   Tel +44 (0) 1223 432134, Fax +44 (0) 1223 423623   http://www.rsc.org I am preparing the following work for publication: 139 Article/Chapter Title : Dihydrogen Bonding Journal/Book Title : Supramolecular Approaches to Selective Mercury Cation Binding and Crystal Engineering of Covalent Crystalline Materials Editor/Author(s) : Karrie M. Manes Publisher : Michigan State University I would very much appreciate your permission to use the following material: Journal/Book Title : Journal of Materials Chemistry Editor/Author(s) : Hui Wu, Wei Zhou, Frederick E. Pinkerton, Martin S. Meyer, Gadipelli Srinivas, Taner Yildirim, Terrence J. Udovic and John J. Rush Volume Number : 20 Year of Publication : 2010 Description of Material : Figures 1 & 3 Page(s) : 6552, 6553   Figure AB-4. Reproduction permission letter for reference 86. 140 141 Figure AB-5. Reproduction permission letter for reference 92. 142 Dear Karrie Thank you for your email. Permission is hereby granted, on behalf of the IUCr, for you to reproduce the material specified below, subject to the following conditions: 1. Reproduction is intended in a primary journal, secondary journal, CD-ROM, book or thesis. 2. The original article in which the material appeared is cited. 3. IUCr's copyright permission is indicated next to the Figure in print. In electronic form, this acknowledgement must be visible at the same time as the Figure, and must be hyperlinked to the article (http://dx.doi.org/10.1107/S1600536804032088). Best wishes Peter Strickland Managing Editor IUCr Journals ______________________________________________________________________ IUCr Editorial Office, 5 Abbey Square, Chester CH1 2HU, England Phone: 44 1244 342878 Fax: 44 1244 314888 Email: ps@iucr.org Ftp: ftp.iucr.org WWW: http://journals.iucr.org/ -------- Forwarded Message -------From: dh@iucr.org To: gh@iucr.org Subject: Permission to reproduce Figure re: Nelson Date: Thu, 07 Jul 2011 15:25:40 -0000 email: maneskar@msu.edu 143 Subject: Permission to reproduce Figure re: Nelson Enquiry type: Journals Comment: I would like to reproduce Figure 1 as printed from: Farrell, D.; McKee, V.; Nelson. J.; Acta Crystallogr. Section E, 2005, E61, 081-083. This figure will be reproduced my dissertation entitled: Supramolecular Approaches to Selective Mercury Cation Binding and Crystal Engineering of Covalent Crystalline Solids. Thank you, Karrie M Manes 505 Chemistry Building Michigan State University East Lansing, MI 48824 313-319-4789 maneskar@msu.edu Figure AB-6. Reproduction permission letter for reference 93. 144 Rightslink® by Copyright Clearance Center 8/25/11 12:50 PM Title: Author: Density Functional Study of Short-Range Interaction Forces between Ions and Water Molecules Logged in as: Karrie Manes Account #: 3000425971 Jaime E. Combariza et al. Publication: The Journal of Physical Chemistry Publisher: American Chemical Society Date: Mar 1, 1995 Copyright © 1995, American Chemical Society No charge permission and attribution Permission for this particular request is granted for print and electronic formats at no charge. Figures and tables may be modified. Appropriate credit should be given. Please print this page for your records and provide a copy to your publisher. Requests for up to 4 figures require only this record. Five or more figures will generate a printout of additional terms and conditions. Appropriate credit should read: "Reprinted with permission from {COMPLETE REFERENCE CITATION}. Copyright {YEAR} American Chemical Society." Insert appropriate information in place of the capitalized words. This permission does not apply to images that are credited to publications other than ACS journals. For images credited to non-ACS journal publications, you will need to obtain permission from the journal referenced in the Table/Figure/Micrograph legend or credit line before making any use of the image(s) or table(s). Copyright © 2011 Copyright Clearance Center, Inc. All Rights Reserved. Privacy statement. Comments? We would like to hear from you. E-mail us at customercare@copyright.com Figure AB-7. Reproduction permission letter for reference 114. https://s100.copyright.com/AppDispatchServlet Page 1 of 1 145 REFERENCES 146 REFERENCES 1. Steed, J.W.; Atwood, J.L. Supramolecular Chemistry, John Wiley & Sons, Ltd: West Sussex, England, 2000. 2. Huggins, M.L. J. Phys. Chem. 1936, 40, 723. 3. Huggins, M.L. J. Org. Chem. 1936, 1, 407. 4. Seiler, P.; Dobler, M.; Dunitz, J.D. Acta Crystallogr., Sect. B: Struct. Crystallogr. Chem. 1974, 30, 2744. 5. Burns, R.C.; Corbett, J.D. J. Am. Chem. Soc. 1982, 104, 2804. 6. Clague, N.P.; Clegg, W.; Coles, S.J.; Crane, J.D.; Moreton, D.J.; Sinn, E.; Teat, S.J.; Young, N.A. Chem. Commun. 1999, 379. 7. Lehn, J-M.; Dietrich, B.; Guilhem, J.; Pascard, C. Tetrahedron Lett. 1989, 30, 4125. 8. Redko, M.Y.; Huang, R.; Dye, J.L.; Jackson, J.E. Synthesis 2006, 5, 759-761. 9. Lehn, J-M.; Montavon, F. Helvetica Chimica Acta 1978, 61, 67. 10. Hunter, J.; Nelson, J.; Harding, C.; McCann, M.; McKee, V. J. Chem. Soc., Chem. Commun. 1990, 1148. 11. Morf, W.E.; Simon, W.S. Helv. 1971, 54, 794. 12. Manes, K.M. Unpublished results. 13. Grøndahl, L.; Hammershøi, A.; Sargeson, A.M.; Thom, V.J. Inorg. Chem. 1997, 36, 5396. 14. Kodama, M.; Kimura, E. J. Chem. Soc., Dalton Trans. 1976, 2335. 15. Uzun, L.; Kara, A.; Tuzmen, N.; Karabakan, A.; Besirli, N.; Denizli, A. J. Appl. Polym. Sci. 2006, 102, 4276. 16. Yavuz, E.; Senkal, B.F. J. Appl. Polym. Sci. 2006, 101, 348. 17. Uludag, Y.; Ozbelge, H.O.; Yilmaz, L. J. Membr. Sci. 1997, 129, 93-99. Centers for Disease Control and Prevention. 18. Great Lakes Binational Toxics Strategy, Stakeholder Forum 1998. http://www.epa.gov/glnpo/bnsdocs/mercsrce/mercreg.html (Accessed 5 March 2011). 147 19. Independent Statistic & Analysis, U.S. Energy Information Administration. International Energy Outlook 2010. http://www.eia.gov/oiaf/ieo/world.html (Accessed 5 July 2011). 20. Matlock, M. M.; Henke, K. R.; Atwood, D. A. J. Hazard. Mater. 2002, 92, 129. 21. Matlock, M. M.; Howerton, B. S.; Atwood, D. A. J. Hazard. Mater. 2001, 84, 73-82. 22. Matlock, M. M.; Howerton, B. S.; Henke, K. R.; Atwood, D. A. J. Hazard. Mater. 2001, 82, 55. 23. Tassel, F.; Rubio, J.; Misra, M.; Jena, B. C. Miner. Eng. 1997, 10, 803. 24. Gash, A. E.; Spain, A. L.; Dysleski, L. M.; Flaschenriem, C. J.; Kalaveshi, A.; Dorhout, P. K.; Strauss, S. H. Environ. Sci. Technol. 1998, 32, 1007. 25. Hollerman, W.; Holland, L.; Ila, D.; Hensley, J.; Southworth, G.; Klasson, T.; Taylor, P.; Johnston, J.; Turner, R. J. Hazard. Mater. 1999, 68, 193-203. 26. Calmon, C.; Gold, H. Ion exchange for pollution control. In; CRC Press, 1979; pp 201206. 27. Dujardin, M. C.; Cazé, C.; Vroman, I. React. Funct. Polym. 2000, 43, 123-132. 28. Michelsen, D. L.; Gideon, J. A.; Griffith, G. P.; Pace, J. E.; Kutat, H. L. "Report PB244890," NTIS, 1975. 29. Russell, E. R. "Report DP-1395," NTIS, 1975. 30. Buckley, L. P.; Vijayan, S.; McConeghy, G. J.; Maves, S. R.; Martin, J. F. "Report AECL-10174," At. Energy Can. Ltd., 1990. 31. Reddy, K. H.; Reddy, A. R. J. Appl. Polym. Sci. 2003, 88, 414. 32. Tratnyek, J. P.; Arthur, D. L. "PB Report No. 211128," U. S. Nat. Tech. Inform. Serv., 1972. 33. Baumann, T. F.; Reynolds, J. G.; Fox, G. A. React. Funct. Polym. 2000, 44, 111. 34. Donia, A. M.; Atia, A. A.; Elwakeel, K. Z. J. Hazard. Mater. 2008, 151, 372. 34. Reddy, K. H.; Reddy, A. R. J. Appl. Polym. Sci. 2003, 88, 414. 35. Senkal, B. F.; Yavuz, E. J. Appl. Polym. Sci. 2006, 101, 348. 36. Yavuz, E.; Senkal, B. F.; Bicak, N. React. Funct. Polym. 2005, 65, 121. 148 37. Hand, D. W.; Crittenden, J. C.; Thacker, W. E. J. Environ. Eng. 1983, 109, 82. 38. Guerra, D. L.; Santos, M.; Airoldi, C. J. Braz. Chem. Soc. 2009, 20, 594. 39. Li, X. G.; Feng, H.; Huang, M. R. Chemistry-A European Journal 2009, 15, 4573. 40. Bessbousse, H.; Rhlalou, T.; Verchere, J. F.; Lebrun, L. J. Membr. Sci. 2008, 325, 997. 41. Nichols, L. D.; Obermayer, A. S.; Allen, M. B.; Cekala, C. J. "Report OWRT-C-200009R(2410)(1), OWRT-RU-84/6; Order No. PB84-228212," NTIS, 1983. 42. Wagener, J.M.; Jarvis. N.V Talanta 1995, 42, 219. 43. Matsumura, M.; Navarro, R.R.; Sumi, K.; Fuji, N. Wat. Res. 1996, 30, 2488. 44. Yilmaz, L.; Uludag, Y.; Onder, H.; Ozbelge, O. J. Membr. Sci. 1997, 129, 93. 45. Alexandratos, S.D.; Zhu, X. Ind. Eng. Chem. Res. 2005, 44, 8605. 46. Lebrun, L.; Bessbousse, H.; Rhlalou,T.; Verchere, J.-F. J. Membr. Sci. 2008, 325, 997. 47. Laborie, S.; Barron-Zambrano, J.; Viers, P.; Durand, R.G. Desalination 2002, 144, 201. 48. Lebrun, L.; Bessbousse, H.; Rhlalou, T.; Verchere, J.-F. J. Phys. Chem. B 2009, 113, 8588. 49. Meyer, G.; Nolte, M.; Pantenburg, I. Z. Anorg. Allg. Chem. 2006, 632, 111. 50. Cotton, F.A.; Wilkinson, G.; Murillo, C.A.; Bochmann, M. Advanced Inorganic th Chemistry, 6 ed. ; Wiley: New York, 1999. 51. Mercier, L.; Bibby, A. Chem. Mater. 2002, 14, 1591. 52. Tavlarides, L.L.; Nam K. H.; Gomez-Salazar, S. Ind. Eng. Chem. Res. 2003, 42, 1955. 53. Carter, T.G.; Yantasee, W.; Sangvanich, T.; Fryxell, G.E.; Johnson, D.W.; Addleman, R.S. Chem. Commun. 2008, 5583. 54. Song, B.Y.; Eom, Y.; Lee, T.G. Appl. Surf. Sci., 2011, 257, 4754. 55. SAMMS Technical Summary, rev. Feb. 2009, Pacific Northwest National Laboratory. http://samms.pnl.gov/sammstech_summary.pdf (Accessed 11 March 2011). 56. Ritchie, S.M.C.; Kissick, K.E.; Bachas, L.E.; Sikdar, S.K.; Parikh, C.; Bhattacharyya, D. Environ. Sci. Technol. 2001, 35, 3252. 149 57. Magosso, H.A.; Panteleimonov, A.V.; Kholin, Y.V.; Gushikem, Y. Coll. Interface Sci., 2006, 303, 18. 58. Garcia-España, E.; Latorre, J.; Luis, S.V.; Miravet, J.F.; Pozuelo, P.E.; Ramirez, J.A.; Soriano, C. Inorg. Chem. 1996, 35, 4591 59. Sharrad, C.A.; Grøndahl, L.; Gahan, LR. J. Chem. Soc., Dalton Trans. 2001, 2937. 60. Smith, R.M.; Martell, A.E.; Motekaitis, R.J. NIST Critically Selected Stability Constants of Metal Complexes Database, Version 4.0, 1997, U.S. Department of Commerce, Gaithersburg, MD, USA. 61. Kodama, M.; Koike, T.; Hoshiga, N.; Machida, R.; Kimura, E. J. Chem. Soc. Dalton Trans., 1984, 673. 62. Blake, A.J.; Reid, G.; Schroder, M. Polyhedron, 1990, 9, 2931. 63. Byriel, K.A.; Gahan, L.R.; Kennard, C.H.L.; Sunderland, C.J. J. Chem. Soc., Dalton Trans., 1993, 625. 64. Taurozzi, J.S.; Manes, K.M.; Redko, M.Y.; Tarabara, V.V.; Jackson, J.E. Microsized particles of Aza222 polymer as a regenerable sorbent for the removal of mercury from aqueous solutions. ACS Appl. Mater. Interfaces. Submitted for publication. 65. Weber, Jr., W.J.; Carter, M.C. Environ. Sci. Technol. 1994, 28, 614. 66. Afshar, S.; Marcus, S.T.; Gahan, L.R.; Hambley, T.W. Aust. J. Chem. 1999, 52, 1. 67. Heinzel, U.; Mattes, R. Polyhedron, 1992, 11, 597. 68. Niedermaier, C.A.; Loehr, R.C. J. Environ. Eng. 2005, 131, 943. 69. United States Geological Survey Water Quality Information. http://water.usgs.gov/owq/hardness-alkalinity.html#hardness (Accessed March 2, 2011). 70. SolmeteX: The Power of Applied Science, http://www.solmetex.com/water/technologies/keylex.html. (Accessed 11 March 2011). 71. Dyrssen, D. Marine Chem. 1988, 24, 143. 72. Redko, M.Y. Unpublished results. 73. Redko, M.Y. Unpublished results. 74. Jeffrey, G.A.; Saenger, W. Hydrogen Bonding in Biological Structures; Springer-Verlag: Berlin, 1991. 150 75. Richardson, T.B.; de Gala, S.; Crabtree, R.H. J. Am. Chem. Soc. 1995, 117, 12875. 76. Custelcean, R.; Jackson, J.E. Chem. Rev. 2001, 101, 1963. Reprinted with permission. Copyright 2001 American Chemical Society. 77. Custelcean, R.; Jackson, J.E. J. Am. Chem. Soc. 1998, 120, 12935 78. Custelcean, R.; Jackson, J.E.; Angew. Chem. Int. Ed. 1999, 38, 1661. 79. Custelcean, R.; Jackson, J.E.; J. Am. Chem. Soc. 2000, 122, 5251. Reprinted with permission. Copyright 2000 American Chemical Society. 80. Custelcean, R.; Vlassa, M.; Jackson, J.E. Angew. Chem. Int. Ed. 2000, 39, 3299. Reprinted with permission. Copyright 1999 Wiley-VCH. 81. Grochala, W.; Edwards, P.P. Chem. Rev. 2004, 104, 1283. 82. Jena, P. J. Phys. Chem. Lett. 2011, 2, 206. 83. Gutowska, A.; Li, L.; Shin, Y.; Wang, C.M.; Li, X.S.; Linehan, J.C.; Smith, R.S.; Kay, B.D.; Schmid, B.; Shaw, W.; Gutowski, M.; Autrey, T Angew. Chem., Int. Ed., 2005, 44, 3578. 84. Jain, I.P.; Jain, P.; Jain, A. J. Alloys Compounds, 2010, 303. 85. Li, H-W.; Yan, Y; Orimo, S-I.; Zuttel, A.; Jensen, C.M. Energies, 2011, 4, 185. 86. Wu, H.; Zhou, W.; Pinkerton, F.E.; Meyer, M.S.; Srinivas, G.; Yildirim, T.; Udovic, T.J.; Rush, J.J. J. Mater. Chem. 2010, 20, 6550. Reprinted with permission from the Royal Society of Chemistry. 87. Gatling, S.; Jackson, J.E. J. Am. Chem. Soc. 1999, 121, 8655. 88. Marincean, S.; Jackson, J.E. J. Phys. Chem. A 2010, 114, 13376. 89. Data were collected using a Bruker CCD (charge coupled device) based diffractometer equipped with an Oxford Cryostream low-temperature apparatus operating at 173 K. Data were measured using omega and phi scans of 0.5° per frame for 30 s. The total number of images was based on results from the program COSMO where redundancy was expected to be 4.0 and completeness to 0.74 Å to 100%. Cell parameters were retrieved using APEX II software and refined using SAINT on all observed reflections. Data reduction was performed using the SAINT software, which corrects for Lp. Scaling and absorption corrections were applied using SADABS multi-scan technique, supplied by George Sheldrick. The structures are solved by the direct method using the SHELXS2 97 program and refined by least squares method on F , SHELXL- 97, which are 151 incorporated in SHELXTL-PC V 6.10.   90. Data were collected using a Bruker CCD (charge coupled device) based diffractometer equipped with an Oxford Cryostream low-temperature apparatus operating at 173 K. Data were measured using omega and phi scans of 0.5° per frame for 30 s. The total number of images was based on results from the program COSMO where redundancy was expected to be 4.0 and completeness to 0.74 Å to 100%. Cell parameters were retrieved using APEX II software and refined using SAINT on all observed reflections. Data reduction was performed using the SAINT software, which corrects for Lp. Scaling and absorption corrections were applied using SADABS multi-scan technique, supplied by George Sheldrick. The structures are solved by the direct method using the SHELXS2 97 program and refined by least squares method on F , SHELXL- 97, which are incorporated in SHELXTL-PC V 6.10. 91. Lyssenko, K.A.; Nelyubina, Y.V.; Antpin, M.Y. CrystEngComm 2007, 9, 632. 92. Bowman-James, K.; Hossain, M.A.; Llinares, J.M.; Alcock, N.W.; Powell, D. J.Supramol. Chem. 2002, 2, 143. 93. McKee, V.; Farrell, D.; Nelson, J. Acta. Cryst. 2005, 61, o81. 94. Cheeseman, G.H.; Duncan, D.R.; Harris, W.H. J. Chem. Soc. 1940, 837. 95. Wade, R.C.; Sullivan, E.A.; Berschied Jr., J.R.; Purcell, K.F. Inorg. Chem. 1970, 9, 2146. 96. Data were collected using a Bruker CCD (charge coupled device) based diffractometer equipped with an Oxford Cryostream low-temperature apparatus operating at 173 K. Data were measured using omega and phi scans of 0.5° per frame for 30 s. The total number of images was based on results from the program COSMO where redundancy was expected to be 4.0 and completeness to 0.74 Å to 100%. Cell parameters were retrieved using APEX II software and refined using SAINT on all observed reflections. Data reduction was performed using the SAINT software, which corrects for Lp. Scaling and absorption corrections were applied using SADABS multi-scan technique, supplied by George Sheldrick. The structures are solved by the direct method using the SHELXS2 97 program and refined by least squares method on F , SHELXL- 97, which are incorporated in SHELXTL-PC V 6.10 97. Uppal, S.S.; Kelly, H.C. Chem. Comm. 1970, 23, 1619. 98. Gyori, B.; Emri, J.; Feher, I. J. Organomet. Chem., 1983, 255, 17. 99. Hein, D.W.; Alheim, R.J.; Leavitt, J.J. J. Am. Chem. Soc. 1957, 79, 427. 100. Hiraoka, S.; Yi, T.; Shiro, M.; Shionoya, M Inorg. Chem. 2002, 124, 14510. 152 101. Padilla-Martinez, I.I.; Contreras, R.; Andrade-Lopez, N.; Gama-Goicochea, M.; Aguilar-Cruz, E.; Tlahuext, H. Heteroatom Chem., 1996, 7, 323. 102. Hermanek, S. Chem. Rev. 1992 92, 325. 103. Data were collected using a Bruker CCD (charge coupled device) based diffractometer equipped with an Oxford Cryostream low-temperature apparatus operating at 173 K. Data were measured using omega and phi scans of 0.5° per frame for 30 s. The total number of images was based on results from the program COSMO where redundancy was expected to be 4.0 and completeness to 0.74 Å to 100%. Cell parameters were retrieved using APEX II software and refined using SAINT on all observed reflections. Data reduction was performed using the SAINT software, which corrects for Lp. Scaling and absorption corrections were applied using SADABS multi-scan technique, supplied by George Sheldrick. The structures are solved by the direct method using the SHELXS-97 program and 2 refined by least squares method on F , SHELXL- 97, which are incorporated in SHELXTL-PC V 6.10. 104. Data were collected using a Bruker CCD (charge coupled device) based diffractometer equipped with an Oxford Cryostream low-temperature apparatus operating at 173 K. Data were measured using omega and phi scans of 0.5° per frame for 30 s. The total number of images was based on results from the program COSMO where redundancy was expected to be 4.0 and completeness to 0.74 Å to 100%. Cell parameters were retrieved using APEX II software and refined using SAINT on all observed reflections. Data reduction was performed using the SAINT software, which corrects for Lp. Scaling and absorption corrections were applied using SADABS multi-scan technique, supplied by George Sheldrick. The structures are solved by the direct method using the SHELXS-97 program and 2 refined by least squares method on F , SHELXL- 97, which are incorporated in SHELXTL-PC V 6.10. 105. X-ray diffraction (XRD) patterns were obtained on Bruker D8 Davinci diffractometer equipped with Cu X-ray radiation operating at 40 kV and 40 mA. Peak intensities were obtained by counting with the Lynxeye detector every 0.0153637° at sweep rates of 0.5° 2θ /min. The sample was ground and sieved (300 MESH) and loaded in a 0.3 mm capillary. The sample was spun at 5 degrees per minute. No background correction is applied to raw data. 106.   107. Dilliban, J. Macromol. Mat. Eng. 2006, 291, 137. Data were collected using a Bruker CCD (charge coupled device) based diffractometer equipped with an Oxford Cryostream low-temperature apparatus operating at 173 K. Data were measured using omega and phi scans of 0.5° per frame for 30 s. The total number of images was based on results from the program COSMO where redundancy was expected to be 4.0 and completeness to 0.74 Å to 100%. Cell parameters were retrieved using APEX II software and refined using 153 SAINT on all observed reflections. Data reduction was performed using the SAINT software, which corrects for Lp. Scaling and absorption corrections were applied using SADABS multi-scan technique, supplied by George Sheldrick. The structures are solved by the direct method using the SHELXS-97 program and 2 refined by least squares method on F , SHELXL- 97, which are incorporated in SHELXTL-PC V 6.10.   108. 109. Elola, M.D.; Laria, D. J. Chem. Phys. 2002, 117, 2238-2245. 110. Weckstrom, K.; Soper, A.K. Biophys. Chem., 2006, 124, 180-191. 111. Gaussian 98, Revision A.6, M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, V. G. Zakrzewski, J. A. Montgomery, Jr., R. E. Stratmann, J. C. Burant, S. Dapprich, J. M. Millam, A. D. Daniels, K. N. Kudin, M. C. Strain, O. Farkas, J. Tomasi, V. Barone, M. Cossi, R. Cammi, B. Mennucci, C. Pomelli, C. Adamo, S. Clifford, J. Ochterski, G. A. Petersson, P. Y. Ayala, Q. Cui, K. Morokuma, D. K. Malick, A. D. Rabuck, K. Raghavachari, J. B. Foresman, J. Cioslowski, J. V. Ortiz, B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R. Gomperts, R. L. Martin, D. J. Fox, T. Keith, M. A. AlLaham, C. Y. Peng, A. Nanayakkara, C. Gonzalez, M. Challacombe, P. M. W. Gill, B. Johnson, W. Chen, M. W. Wong, J. L. Andres, C. Gonzalez, M. HeadGordon, E. S. Replogle, and J. A. Pople, Gaussian, Inc., Pittsburgh PA, 1998. 112. Heuft, J.M.; Meijer, E.J. J. Chem. Phys. 2005 123, 094506. 113. Narten, A.H. J. Phys. Chem. 1970, 74, 765. 114.     Sagar, D.M.; Bain, C.D.; Verlet, J.R.R. J. Am. Chem. Soc. 2010, 132, 6917-6919. Combariza, J.E.; Kestner, N.R. J. Phys. Chem. 1995, 99, 2717. 115. Miyake, T.; Aida, M. Internet Journal of Molecular Design 2003, 2, 24. 154