Crystallization of glycine at liquid liquid interface an experimental perspective

CRYSTALLIZATION OF GLYCINE AT LIQUID INTERFACE - AN EXPERIMENTAL PERSPECTIVE RENO ANTONY LOUIS LEON (B.Tech., Bharathidasan University) A THESIS SUBMITTED FOR THE DEGREE OF MASTER OF ENGINEERING DEPARTMENT OF CHEMICAL AND BIOMOLECULAR ENGINEERING NATIONAL UNIVERSITY OF SINGAPORE 2011 **** i Acknowledgements My experience at NUS as a graduate student has been one of the most exciting and interesting aspects of my life. It was possible only because of the wonderful people who have accompanied me through this two year journey and I feel a great sense of gratitude towards each and every one of them. Foremost of all, I am highly indebted to Professor Raj Rajagopalan for his guidance, encouragement and supervision. He definitely has been my role model for his kind and supportive nature and is the most considerate at difficult times. I owe a great deal to Asst. Professor Saif A. Khan, a source of inspiration and enthusiasm. His invaluable advice which yearns for creative thinking has been the ultimate driving force for the project. I am thankful to Zhang Chunyan and Toldy Arpad Istvan who have collaborated with me in this project. Mutual support and hard work are two major lessons I have learnt from them and the fond memories of our working together will always remain intact. It has been a wonderful learning experience for me with the past and present members of both the lab groups of Prof. Raj and Dr. Khan. Sincere thanks to Dr. Soeren Enemark and Pravien Parthiban for sharing their valuable technical insights. I am thankful to Zahra, Kat, Dhawal and Sophia for their assistance and timely suggestions. Thanks to Dominik, Suhanya, Swekun, Josu, Prasanna, Dr.Rahman, Vigneshwar and Nicholas for all the fun times as wonderful lab mates. I am thankful to all the amazing people in my life outside NUS, my dad, mom and sister for their love and affection, Naviyn, Subash and Ashok who always encourage me to follow my dreams. Finally, I thank God Almighty who instills strength in me to face the highs and lows of life each day. i Table of Contents Acknowledgements ................................................................................................................ i Table of Contents .................................................................................................................. ii Summary ............................................................................................................................. iv List of Tables ........................................................................................................................ v List of Figures...................................................................................................................... vi List of Symbols .................................................................................................................... ix 1. Introduction .................................................................................................................. 1 1.1 Thesis objectives and layout .................................................................................... 1 2. Crystallization ............................................................................................................... 3 2.1 The Advent of Crystallization .................................................................................. 3 2.2 Crystallization - a dual process ................................................................................ 5 2.2.1Nucleation....................................................................................................... 5 2.2.2Crystal Growth ................................................................................................ 8 2.3 Polymorphism in organic compounds .................................................................... 16 2.4 Effect of supersaturation ........................................................................................ 18 3. Polymorphism of Glycine............................................................................................ 20 3.1 Glycine polymorphs .............................................................................................. 20 3.2 Effect of solvents on glycine polymorphism .......................................................... 22 3.3 Glycine polymorphic selection in water-methanol/ethanol systems ........................ 23 3.3.1β-glycine precipitation................................................................................... 24 3.3.2γ-glycine inhibition ....................................................................................... 25 3.3.3α-glycine formation ....................................................................................... 26 4. Influence of surface/interface on crystallization ........................................................... 28 4.1 Templated crystallization in nature ........................................................................ 28 4.2 Interaction between crystals and surfaces ............................................................... 29 4.3 Crystal interaction with liquid interfaces in emulsion based crystallization ............. 31 5. Experimental............................................................................................................... 34 5.1 Materials and Methods .......................................................................................... 34 5.1.1X-Ray Powder Diffraction (XRD) ................................................................. 35 ii 5.1.2Semi-Quantitative analysis of XRD data........................................................ 36 5.1.3Scanning Electron Microscopy (SEM) ........................................................... 38 5.2 Batch crystallization experiments .......................................................................... 40 5.2.1Data Analysis for batch crystallization........................................................... 43 5.3 Direct perfusion..................................................................................................... 45 5.3.1Data Analysis for direct perfusion experiment ............................................... 49 5.4 Crystallization at Liquid Interface.......................................................................... 52 5.4.1Data Analysis for interfacial crystallization experiment ................................. 57 5.5 Interfacial crystallization for zero ethanol flow rate ............................................... 59 6. Discussion .................................................................................................................. 61 6.1 Glycine crystal morphology................................................................................... 61 6.2 Dynamic observations of crystallization process .................................................... 62 6.3 Crystallization is localized at the liquid-liquid interface ......................................... 63 6.4 Disengagement of interface from crystal network .................................................. 65 6.5 Dynamic crystal morphology change ..................................................................... 67 6.6 Effect of Anti-solvent Perfusion Rate on Crystal Appearance ................................ 69 6.7 Concomitant polymorphism and switch in dominant polymorphic form ................. 70 7. Summary and Outlook ................................................................................................ 73 8. Bibliography ............................................................................................................... 76 Appendix A ........................................................................................................................ 86 MATLAB code for integrated intensity analysis .......................................................... 86 Appendix B ........................................................................................................................ 89 Calculation of mass of glycine crystallized .................................................................. 89 iii Summary In an attempt to improve the current understanding of polymorphic nucleation and crystal growth in organic crystallization processes, the effect of solvent on polymorphic selection has gained tremendous interest in recent years; especially in studies of crystallization at the vicinity of liquid-liquid interfaces, which is a complex and poorly understood phenomenon involving a host of interacting processes such as interfacial molecular recognition, exchange of material across phases and phase transformation. Conventional experimental tools that employ stirred tanks or flasks for both production or research and development, often involve heterogeneous spatial and temporal distribution of process parameters. This largely hinders the control and resolution required to acquire mechanistic insight into the process. This thesis reports a simple and effective method to probe anti-solvent induced crystal nucleation and growth at liquid-liquid interfaces at good spatial and temporal resolution coupled with realtime high speed optical microscopy. Glycine is used as the model solute for crystallization. It is observed that the liquid interface serves as a potential site where crystals first appear and grow. The size and morphology of the formed crystals are closely related to the anti-solvent addition rate. Furthermore, the dominant polymorphic form might also be influenced by the rate of addition of anti-solvent (ethanol). Thus we report that a change in the processing condition coupled with the presence of an interface might causes a polymorphic shift in the dominant crystal form, in this case from α-glycine to γ-glycine. Dynamic changes in crystal morphology and concomitant polymorphism of glycine have also been addressed. This mode of crystallization could enable better observation and improved understanding of crystallization at liquid-liquid interfaces. iv List of Tables Table 2-1 Properties affected by crystallization and their relationship with product characteristics [13]. .............................................................................................. 4 Table 5-1 Dominant peak positions for different polymorphs of Glycine ............................. 37 v List of Figures Figure 2-1 Topographic features on a growing crystal face, illustrating terrace, step and kink sites (Rosenberger 1986) .............................................................................. 9 Figure 2-2 Development of a spiral from a screw dislocation (Sunagawa 2005) ................... 12 Figure 2-3 γ-glycine as viewed down the b-axis. The capped face (0 3) exposes NH3+ while the flat face (00 ) exposes CO2- [36] ....................................................... 14 Figure 2-4 Scheme for ‘Relay Type’ Growth Mechanism [40] ............................................. 15 Figure 2-5 Polymorphs of ROY[39] .................................................................................... 16 Figure 3-1 A zwitterionic glycine molecule. ........................................................................ 20 Figure 3-2 α-glycine crystal structure displaying hydrogen bonded bilayers [68] ................. 21 Figure 3-3 β-glycine crystal structure [68] ........................................................................... 21 Figure 3-4 γ-glycine crystal structure displaying polar helical hydrogen bond structure [68] 22 Figure 3-5 Packing arrangement of β-glycine. The (010) "azure" and (0 0) "pink" surfaces are exposed at the interface [5]. .......................................................................... 24 Figure 3-6 γ-glycine crystals obtained from 1:1 water ethanol solution [5]. .......................... 26 Figure 3-7 Packing arrangement of γ-glycine showing the pockets of the fast growing (00 ) face that are poisoned by the adsorption of ethanol and methanol molecules (shown as ‘balls and sticks’)[5].......................................................... 26 Figure 3-8 Packing arrangements of α-glycine (a) exposing weak binding C-H bonds to the solution at (010) surface (azure) or (b) exposing strong solvent-binding NH bonds to the solution at the (010) surface (pink) [5]. ....................................... 27 Figure 3-9 α-glycine crystals obtained from a 9:1 water-ethanol solution [5]. ..................... 27 Figure 4-1 Coccolith formation of E. Huxleyi ...................................................................... 28 Figure 4-2 Self Assembled structures of Glycine on a glass surface. .................................... 29 Figure 4-3 (a) Bifunctional SAM pattern on surfaces; (b) Glycine crystals on patterned surface[88]......................................................................................................... 30 Figure 4-4 Schematic of surfactant-monolayer-templated nucleation and crystallization[87] .......................................................................................................................... 33 Figure 5-1 Raw XRD pattern displaying peaks corresponding to background Aluminium along with glycine peaks. ................................................................................... 35 vi Figure 5-2 Analysis of XRD Pattern for Glycine obtained from interfacial crystallization experiment with an ethanol perfusion rate of 10µL/min. ..................................... 38 Figure 5-3 SEM images of two different glycine crystal morphologies. ............................... 39 Figure 5-4 Optical microscopy images of different glycine crystal morphologies. ................ 39 Figure 5-5 Schematic for batch crystallization with different ethanol volume fractions (1080%). ................................................................................................................. 40 Figure 5-6 Optical microscopy images of crystals from (i) 40%vol (ii) 80%vol (Scale-bar size 300µm). ...................................................................................................... 41 Figure 5-7 Powdered glycine pressed into the cavity of XRD metal sample holder (top and orthographic) (Scale-bar size 1 cm) .................................................................... 41 Figure 5-8 XRD pattern overlay for glycine crystals from various ethanol fractions (4080%). ................................................................................................................. 42 Figure 5-9 Plot of polymorphic outcome of glycine for various ethanol volume fractions. ... 44 Figure 5-10 Schematic for crystallization by direct perfusion of ethanol (0.5-10 µL/min). ... 45 Figure 5-11 Optical microscopy image of crystals collected from 0.5 µL/min (top) and 7 µL/min (bottom) perfusion sets (Scale-bar size 300µm). .................................... 46 Figure 5-12 SEM images of crystals for (i) 2 µL/min (ii) 5 µL/min..................................... 47 Figure 5-13 XRD pattern overlay for two different flow rates of ethanol (3 and 7 µL/min). . 48 Figure 5-14 3D plot of XRD pattern for all perfusion rates of ethanol (0.5-10 µL/min).. ...... 48 Figure 5-15 Similarity in α-glycine morphology grown in (a) aqueous medium[111], (b) solvent mixture .................................................................................................. 50 Figure 5-16 Plot of polymorphic outcome of glycine for various ethanol perfusion rates (µL/min). ........................................................................................................... 51 Figure 5-17 Schematic for interfacial crystallization with perfusion of ethanol (0.5-10 µL/min) ............................................................................................................. 52 Figure 5-18 Glycine crystal network at the liquid interface with perfusion of ethanol at 10 µL/min (Scale-bar size 300µm) ....................................................................... 53 Figure 5-19 Optical micrograph of glycine crystals formed for different perfusion rates of ethanol (Scale-bar size 300µm). ...................................................................... 54 Figure 5-20 3D plot of XRD pattern for all flow regimes of ethanol (0.5-10 µL/min)........... 55 Figure 5-21 Shift in dominant peak position with increasing flow rate of ethanol from 3 to 7 µL/min............................................................................................................ 56 vii Figure 5-22 Glycine crystallization process for 5 µL/min ethanol perfusion (Scale-bar size 1.5mm) .............................................................................................................. 57 Figure 5-23 Effect of ethanol perfusion rate on Glycine polymorphism ............................... 58 Figure 5-24Experimental Set-up for Diffusion-Controlled Crystallization............................ 59 Figure 5-25 XRD pattern overlay for multiple repeats of interfacial crystallization experiment for zero ethanol flow rate. ................................................................ 60 Figure 6-1 Various morphologies of glycine observed (a)bipyramidal, (b) plate, (c) rod, (d) needle (Scale-bar size 300µm) ........................................................................ 61 Figure 6-2 Formation of Glycine crystals during perfusion experiments (a) in cuvette at the liquid-liquid interface; (b)zoomed image of crystals at the interface; (c) in Cuvette without Hexane layer for an interface; (d) zoomed image of crystals in bulk (Scale-bar size 300µm) .............................................................................. 63 Figure 6-3 Optical micrograph of flow patterns for ethanol + rhodamine B with varying perfusion rates in the interfacial crystallization system, (a)-(d) bright field; (e) fluorescence under UV....................................................................................... 64 Figure 6-4 Plots for (a) Glycine solubility in water-ethanol mixture (b) Amount of glycine crystallized for different volumes of ethanol addition ......................................... 66 Figure 6-5 Dynamic crystal morphology change for different ethanol perfusion rates,(a) 0.5µL/min ; (b) 7µL/min (Scale-bar size 300µm). ............................................... 67 Figure6-6 Optical micrograph of glycine crystals obtained from different ethanol perfusion rates, (a)&(b) in cuvette for 0.5 µL/min; (c),(d)&(e) crystals from 0.5 µL/min;(f)&(g) in cuvette for 10 µL/min; (h) Crystals from 10 µL/min (Scalebar size 300µm). ................................................................................................ 69 Figure 6-7 Effect of Ethanol perfusion rate on Glycine polymorphism;(top) Comparison of trends in polymorphic outcome for perfusion experiments with and without Hexane, (bottom) Comparison of XRD peak patterns for perfusion experiments (3,7µL/min) with and without Hexane respectively ........................................... 71 viii List of Symbols ΔG Gibbs free energy difference µold chemical potential of the solution phase µnew chemical potential of the crystalline phase Δµ free energy difference between the old and the new phase per molecule ai activity of solute i ai,e activity at which the solute and the condensate are in phase equilibrium Ci concentration of species i Ci,s solubility limit of species i Vl volume of liquid phase Al surface area associated with the newly formed liquid phase kB Boltzmann constant T temperature υsolute molecular volume of the solute γ interfacial tension between the new phase and the solution J number of nuclei formed per unit of time per unit of volume ΔGhet Gibbs free energy change in heterogeneous crystallization ΔGhom Gibbs free energy change in homogeneous crystallization ƒ geometric correction factor θ contact angle (0 ≤ θ ≤ 180) tind induction period tT relaxation time required to reach a quasi-steady state distribution of molecular cluster tn nucleation time tg time required for a stable nucleus to grow to a detectable size v linear growth rate v0 step velocity h step height ix y0 interstep spacing a surface entropy factor ξ anisotropic factor ΔHf heat of fusion xs solubility Eslice horizontal bond energy between two adjacent crystal blocks Eer total crystallization or lattice energy kR rate constant of rough surface mechanism kMN rate constant of mononuclear growth model g shape factor γE edge surface tension kPN rate constant of polynuclear model kBS rate constant of birth and spread model kSN rate constant of surface nucleation k1, k2, k3 rate constant of polynuclear model σ Supersaturation x CHAPTER 1 1. INTRODUCTION Introduction Crystallization as a process plays a vital role in the pharmaceutical industry in the case of separation, purification and formulation of Active Pharmaceutical Ingredients (API). Crystal size, shape and polymorphic form are some of the important factors that govern a range of crystal properties such as solubility, stability, hardness, color, melting point and reactivity [1]. Among these, crystal polymorphism is a characteristic of prime interest as it has direct implications on the process sustainability as well as the physiological uptake [2]. For instance, different polymorphs have different stability, solubility and dissolution rates; stability determines the shelf-life of a crystalline product while solubility and dissolution kinetics are key factors in governing drug bioavailability [3]. Thus a key challenge in pharmaceutical crystallization is to efficiently produce crystals of a specific polymorphic from, an issue which is often approached by trial and error experiments. Although a multitude of methods such as seeded crystallization, cooling of melts, de-sublimation, spray drying and use of additives and mixed-solvents [4] does exist, emulsion based crystallization represents an attractive platform to simultaneously control both nucleation of a specific API morphology, while producing crystals of desired size and shape that greatly accelerates product formulation and eliminates costly downstream processing like dry milling or grinding. In an emulsion based crystallization process, the API is initially dissolved in solvent and mixed with an anti-solvent. Keeping this in mind, we focus on the face of the droplet which first sees the anti-solvent across the liquid-liquid interface and can serve as a site for nucleation and growth of the solute. In this work we study polymorphic nucleation and crystal growth in the vicinity of a liquid-liquid interface from an experimental perspective. 1.1 Thesis objectives and layout The purpose of this thesis is to gain insight into the process of crystallization at liquid-liquid interfaces. Glycine in water has been chosen as a model system while ethanol has been used as the anti-solvent and an aqueous-organic interface which constitutes of glycine-water for the 1 CHAPTER 1 INTRODUCTION aqueous region and hexane for the organic region serves as the liquid-liquid interface. It was observed that glycine crystallizes in its least stable forms when precipitated from watermethanol/ethanol solutions [5]. This study investigates similar aspects of crystallization in the glycine-water-ethanol mixture at the vicinity of a liquid-liquid interface. The liquid-liquid interfacial platform developed for this study presents a tool for gaining insight into the fundamentals of crystallization at liquid-liquid interface. This thesis is organized in six chapters. Chapter 2 discusses the process of crystallization, nucleation and crystal growth phenomena and theories associated with them. The concept of polymorphism in crystal forms and the role of supersaturation played are discussed as well. Chapter 3 presents a detailed report on the model system Glycine, its various polymorphic forms and discussed the effect of water and ethanol as solvent or anti-solvent mixtures on glycine as per reported in literature so far. In Chapter 4, the role of surfaces or interfaces and how they can influence crystal properties is explained. Chapter 5 demonstrates the crystallization experimental setups in which batch crystallization and crystallization at liquidliquid interface is studied. In addition the usage of optical microscopy for real time imaging and analysis, alongside Scanning Electron Microscopy (SEM) and X-Ray Powder Diffraction (XRD) Analysis techniques for crystal characterization is explained. We also discuss the various challenges and practical issues of the experimental setup here. It also elaborates on the various observations and data analysis of the experimental work. We hypothesize that the nature of the liquid-liquid interface coupled with specific process conditions, plays a role in the polymorphic selection of the crystal outcome and we discuss the results in light of the proposed hypothesis. The results can be proven by substituting hexane which was used to form the liquid-liquid interface, for another liquid e.g. Dodecane. Chapter 6 comprises of the discussions pertaining to the various observations and results of the experimental. Chapter 7 concludes the thesis with a summary of the study, significant understandings from the work and some directions for future work. 2 CHAPTER 2 2. Crystallization 2.1 The Advent of Crystallization CRYSTALLIZATION The process of crystallization is ubiquitous and has been utilized for thousands of years [6]. The sheer applications of crystals in the 19th century and earlier were as precious stones for their fascinating properties: transparency and color, refractive index and optical dispersion, symmetry and facets [7]. However, with the advent of scientific developments of the 20th century, crystallization has become an important process for numerous modern technologies, for a number of applications such as separation, concentration, purification and solidification. Crystallization is utilized in the petrochemical industry for separation and purification of solids. It is an important process in the specialty chemicals industry for manufacturing household products and cosmetics. In addition, crystallization finds applications in new areas such as understanding surfactant behavior [8]. It is also utilized in the food industry for controlling stability and texture of food products [9]. Other industries that have requirements for crystallization include microelectronics [10], pigments [11], and most importantly, pharmaceuticals [12]. Crystallization finds a variety of applications in the pharmaceutical industry, including isolation and synthesis of Active Pharmaceutical Ingredients (API), co-crystals, excipients and separation of chiral isomers. More than 90% of all pharmaceutical products, such as tablets, capsules, aerosols, suspensions and suppositories contain the API in particulate, mainly crystalline form [13, 14], and almost all small molecular weight pharmaceuticals are formulated in particulate, generally crystalline form [15]. This process defines the purity and other solid state properties of the drug such as the crystal habit, size and polymorphic form, as well as drug product stability and performance. Consequently, reproducibility issue with crystallization process causes a wide range of pharmaceutical formulation problems, such as bioavailability, as well as occurrence of varied chemical and physical forms. The most 3 CHAPTER 2 CRYSTALLIZATION important solid-state and drug delivery characteristics, affected by crystallization, are summarized in Table 2-1 [13]. Table 2-1 Properties affected by crystallization and their relationship with product characteristics [13]. Crystallization may be defined as a phase change in which a solid product with short and long range order of atoms and molecules in a fixed lattice arrangement is obtained from a solution. It is a deceptively complex process and the final outcome of this process results from the interplay of solution thermodynamics and kinetics, as well as other factors such as mass and heat transfer, fluid dynamics and molecular recognition phenomena. The lack of understanding of crystallization process and the various underlying phenomena lead to unwanted or previously unknown nucleation events that threaten the development of a pharmaceutical product. Dunitz and Bernstein have provided examples of such incidents, demonstrating the poor control persistent in crystallization practices [16]. One of the challenges associated is its sensitivity to the process parameters. The sudden appearance of a 4 CHAPTER 2 CRYSTALLIZATION new crystalline structure, different from the existing form of the HIV drug, Ritonavir, illustrates the sensitivity of the crystallization process to synthesis conditions [17]. 2.2 Crystallization - a dual process Crystallization is considered as a dual or two step process. The first step is nucleation, the birth of a stable crystal nucleus and the formation of a new solid phase. Nucleation is followed by crystal growth in the second step. 2.2.1 Nucleation The rate and the mechanism of crystal formation, can be affected by supersaturation, rate of supersaturation generation and desupersaturation, diffusivity, temperature, impurities and the reactivity of surfaces towards nucleation [18]. Nucleation can be either primary, which occurs in the absence of any crystal surface, or secondary, which requires the presence of a crystal surface in order to generate further nuclei. Primary nucleation can be either homogeneous nucleation, when the nuclei or pre-nucleation clusters form without being in contact with no phases or molecular species other than the old phase, where as in heterogeneous nucleation, the nucleation phenomenon or pre-nucleation cluster formation takes place while being in contact with other phases or molecular species [19, 20]. Homogeneous nucleation The most well developed theory for homogeneous nucleation is the capillary or classical theory of nucleation. It dates back to the works of Gibbs [21], Volmer [22], and others [20, 23]. The classical nucleation theory proposes successive addition of solute units to form the critical cluster. The free energy of a nucleus of critical size, l, at some supersaturation, σ, is given by the balance of the energy gained but the volume of new phase, V1, and the energy lost to form the surface area associated with the new phase, A1. 5 CHAPTER 2 G   V1  solute CRYSTALLIZATION kbT  Al  2-1 Where υsolute is the molecular volume of the solute and γ is the interfacial tension between the new phase and the solution. Consequently, the free energy, ΔG, increases with the cluster size, l, until a maximum is reached at l *, which marks the nucleation event. As the new phase grows larger than l*, the free energy decreases without bound. Nucleation is a probabilistic event, with the chances for occurrence depending on the free energy barrier relative to kBT. As the supersaturation increases, the free energy barrier, as well as the critical size, l*, decreases, resulting in faster nucleation. Due to the activated nature of the nucleation process, the rate of homogeneous nucleation can be expressed classically in the form of the Arrhenius equation:  BJ  3  J  K J exp   3 2   T   2-2 Where J is the number of nuclei formed per unit of time per unit of volume. Equation 1-2 summarizes the effects of supersaturation, temperature and interfacial tension on the nucleation rate. At lower supersaturation, the interfacial tension dominates and there is insufficient free energy to create a new surface. As supersaturation increases, the nucleation rate increases exponentially, eventually reaching a maximum. The nucleation theory predicts the transition behaviour from the metastable zone to the labile zone fairly accurately. However, the classical nucleation theory is limited by the assumptions it requires [24]. The most critical assumption in this theory is the capillary approximation, wherein the small critical cluster of the new phase is assumed to represent macroscopic regions. This assumption becomes questionable for nuclei that may contain only tens of molecules. In addition, the macroscopic values of the interfacial tension used in classical model, which is for an infinite planar surface, may not be an adequate representation of the actual interfacial 6 CHAPTER 2 CRYSTALLIZATION tension of the new phase, particularly at the critical size [25]. Moreover, it is assumed in the classical nucleation theory that the Gibbs free energy due to the addition of a new species to the nucleus depends only on the nucleus size. However, the free energy may also depend on the configuration and the site of attachment of the new species. The assumption of equilibrium between the critical nucleus and the surrounding is also unrealistic for small critical nuclei, since the exchange between the two phases even at equilibrium is associated with the fluctuations comparable with the nuclei size [24]. Heterogeneous nucleation The nucleation process is usually enhanced by the presence of impurities of particles, ions or foreign surfaces. In industrial crystallizers, nucleation is mostly heterogeneous, with concomitant secondary nucleation. This is because the foreign surface lowers the nucleation energy barrier. The wetting properties of the foreign substance and their atomic packing arrangements are known to affect the heterogeneous nucleation. The Gibbs free energy of the critical nucleus that forms through the heterogeneous nucleation, ΔGhet, is expressed as: Gc ,het  fGc ,hom f  (2  cos )(1  cos ) 2 4 2-3 Where ΔGhom is the Gibbs free energy of the critical nucleus in the heterogeneous process, f is a geometric correction factor, and θ is the constant angle (0 ≤ θ ≤ 180o). If the foreign surface is not flat, which is the case of nano-sized foreign particles, the size of the foreign material also has to be taken into account. If the contact angle is 180o, as in the case of nonwetting situation, then f is 1, representing homogeneous nucleation. If the contact angle is between 0 and 180o, then f is smaller than 1 and the nucleation energy barrier is reduced [26, 27]. Secondary nucleation can also occur by a number of different mechanisms, originating either from the parent crystal or from the loosely ordered solute molecules near the crystal surface. Initial breeding occurs when the tiny crystallites, loosely bound to the pre-existing crystals, 7 CHAPTER 2 CRYSTALLIZATION act as nucleation sites. When dendritic crystals fragment serve as nucleation sites, the process is called needle breeding. Mincoabrasion of crystals at high stirring speeds can also produce fragments that act as nucleation sites, which is called collision or attrition breeding [27]. Induction time In the case of unseeded crystallization, the kinetics of nucleation can be measured with the induction time, defined as the time elapsed between the creation of supersaturation and the formation of the new phase. Induction time is a function of the solution temperature and supersaturation [27]. The induction period, tind is generally considered to be made of three parts: a relaxation time, tr, for the system to reach a quasi-steady state distribution of molecular clusters; nucleation time, tn; and the time a stable nucleus needs to grow to a detectable size, tg [23]. tind  t r  t n  t g 2-4 2.2.2 Crystal Growth In the dual process of crystallization nucleation is followed by crystal growth which occurs through a 2-dimensional molecular self assembling, where the solute molecules from the supersaturated solution are added to the solid phase. Solute molecules migrate from the bulk solution to the crystal surface, adsorb, diffuse around the surface to find a suitable site and are finally integrated into the crystal lattice. When the mass transfer is not limiting, surface integration is the rate limiting step of crystal growth. Crystal growth is known to take place in a layer-by-layer fashion with the linear growth velocity of a facet defined in a direction normal to it. Figure 2-1 illustrates a model, revealing the different sites for adsorption of growth units: terrace, step and kink [28]. Since a growth unit is attached to three surfaces in a kink site, it is the most energetically favourable. As growth units are added to a kink site, the kink moves along a step and eventually a full layer on a facet is completed. Thus, the linear 8 CHAPTER 2 CRYSTALLIZATION growth rate, v, of the face can be expressed in terms of the step velocity, v0, the step height, h and the interstep spacing, y0: v v0 h y0 2-5 Figure 2-1 Topographic features on a growing crystal face, illustrating terrace, step and kink sites (Rosenberger 1986) The different theories on the different growth mechanisms have been thoroughly reviewed by O’Hara and Reid [28] and Stirckland-Constable [29]. They can be classified into three main categories: continuous growth mechanism, birth and spread mechanism and screw dislocation mechanism. Gilmer and Bennema have shown with simulations that for a simple Kossel model, the mechanism of crystal growth is determined by the surface entropy factor, α [30], which can be approximated as:  H f       ln xs   RT  2-6 In equation 1-6, ξ is the anisotropic factor, ΔHf is the heat of fusion and xs is the solubility. The anisotropy factor describes the intermolecular interactions in the crystal surface and can be approximated as:  Eslice Ecr 2-7 9 CHAPTER 2 CRYSTALLIZATION Where Eslice is the horizontal bond energy between two adjacent blocks, and Ecr is the total crystallization or lattice energy [31]. It has been proposed that when α is less than 3, the interface is rough and growth occurs through continuous or normal mechanism. When α is between 3 and 4, the interface is better defined and growth may occur with surface nucleation method. As α increases above 4, the interface becomes smoother and growth at low supersaturation occurs at the steps generated by defects [32, 33]. Rough surface mechanism In the case of a rough interface, there are many kink sites on the crystal surface. Thus, addition of growth units into the crystal lattice occurs easily and continuously. In this case, a continuous growth equation has been suggested: G  kR 2-8 In equation 2-8, G is the growth rate, kR is the rate constant and σ is the supersaturation as defined previously. Surface nucleation mechanism In the scenario of a smoother interface, 3 ≤ α ≤ 4, the growth occurs through the formation of a 2-D nucleus on the crystal surface and its subsequent spread to complete the layer. The nucleus is conceived of growing at a finite rate, independent of size, by the incorporation of growth units at the steps. Three types of surface nucleation mechanisms have been widely used: mononuclear model, polynuclear model, birth and spread model. Mononuclear model depicts layer-by-layer growth, where only one nucleus exists at a time and the next nucleus forms only after the completion of the previous layer. In this case, the growth rate is expressed by equation 2-9 G  k MN 1/ 2 2  g E exp  2  3(k BT )      2-9 10 CHAPTER 2 CRYSTALLIZATION Where kMN is the rate constant of growth, γE is the edge surface tension, g is a shape factor and the exponential term represents the activation energy required for the formation of a critical 2D nuclei. For circular nuclei, g equals to pi. Polynuclear model is the other extreme configuration of surface nucleation mechanism. In this case, multiple nuclei can exist on the same surface at a time and the rate is limited by the spread of the layer instead of nuclei formation. Hence, the completion of a layer occurs mostly through the formation of nuclei. The growth rate for polynuclear model is expressed as: G  k PN  3 / 2 2  g E  exp  2  3(k BT )      2-10 Where kPN is the rate constant of growth. In between the two extreme cases, is the birth and spread model, also known as the nucleiabove-nuclei model. In this case, it is assumed that nuclei can possibly form on incomplete layers and grow at constant step advancement, independent of each other. The rate limiting process in this case is the formation of nuclei. The growth rate for birth and spread model can be expressed as: 2  g E G  k BS 5 / 6 exp  2  3(k BT )      2-11 Where kBS is the rate constant of this model. This model fails to account for the observed growth rates at very low supersaturation, where the driving force necessary for surface nucleation is not achieved [29]. Surface nucleation models exhibit a strong dependency of growth on supersaturation since the time elapsing between two nucleation events decreases rapidly with an increase in supersaturation [34]. 11 CHAPTER 2 CRYSTALLIZATION Screw dislocation mechanism At higher α value, the intermolecular interaction in the plane of the interface is enhanced, resulting in a much smoother surface to grow on. In this case, the crystal growth may occur mostly through steps. BCF (Barton-Cabrera-Frank) theory proposes a mechanism in which steps are self-perpetrating. Once a screw dislocation has formed, that provides a way for the continuous growth of the steps, similar to a spiral staircase as shown in Figure 2-2. Figure 2-2 Development of a spiral from a screw dislocation (Sunagawa 2005) Surface diffusion is assumed to be the rate limiting step in this mechanism and the growth rate is expressed as: k G  k1 2 tanh  2      2-12 Where k1 and k2 are constants. However, when the supersaturation, σ, is much greater than k2, the growth rate assumes a linear relationship with supersaturation, similar to rough mechanism. G  k3 2-13 where k3= k1k2. On the other hand, when σ is much smaller than k2, the growth rate assumes a parabolic relationship with supersaturation. 12 CHAPTER 2 CRYSTALLIZATION G  k1 2 2-14 Crystal growth rates may vary in a number of ways, making it difficult to interpret the growth kinetics data. Two main phenomena that can cause such variation are distinguished as size dependent growth and growth rate dispersion. Apart from mass transfer limited case, size dependent growth can also result from the Gibbs-Thomson effect for small crystals, less than a few micrometers, with a decreasing growth rate for decreased crystal size [35], as well as from size dependent integration kinetics. Dislocation density within a crystal may increase as a function of size, due to higher mechanical stress and increased incorporation of impurity atoms [19, 35]. In addition, there has been substantial experimental evidence that even when exposed to identical process conditions, different crystals of a given material and size may grow at different rates, simply because they experience different random fluctuations in defect density [19]. Relay mechanism for growth Although a list of such theories does exist, there seems to be an interesting case with an entirely different approach towards crystal growth from solutions when it comes to the growth of polar organic crystals like γ-glycine and (R-S) alanine. These two crystals have similar packing features and only the growth of γ-glycine is discussed here. γ-glycine in solutions has a flat (00 ) face perpendicular to the polar c-axis at one end and capped faces at the opposite end as shown in Figure 2-3 [36]. According to crystal growth and etching experiments [37, 38] the CO2- groups are exposed at the (00 ) face, the “flat –c end”, while the NH3+ amino groups are exposed at the +c capped end [37, 38]. Various experiments involving the comparison of relative rates of growth and dissolution of the crystals of (R-S) alanine and γ-form of glycine indicate that in aqueous solutions the –c carboxylated end of the crystals grow and dissolve faster than the +c amino end. 13 CHAPTER 2 CRYSTALLIZATION Figure 2-3 γ-glycine as viewed down the b-axis. The capped face (0 3) exposes NH3+ while the flat face (00 ) exposes CO2- [36] Inspection of the packing arrangement of γ-glycine reveals that (00 ) carboxylated faces comprise regular pockets on a molecular level and can be regarded as corrugated in two dimensions. The binding water in these pockets can be qualitatively explained looking at the water-glycine interactions in the pocket. The water molecules inside the pocket can essentially take two different orientations; one orientation comprises of O-H ....O hydrogen bond and two O ....O lone-pair-lone-pair repulsions and the other two O-H ....O bonds and one O ....O lone pair repulsion. Consequently introduction of water yields repulsive or at best weakly attractive interactions. The pocket will therefore be unhydrated or slightly hydrated and relatively easily accessible to approaching solute molecules. In contrast, the water molecule may be strongly bound to the outermost layer of CO2- groups via O-H .... O (carboxylate) hydrogen bonds [39]. As glycine molecules are incorporated into adjoining pockets, the CO2- groups of newly added substrate molecules will expel the water bound on the outermost surface, thereby generating new unsolvated pocket on the crystal surface. This relay process of solvent water binding and expulsion helps growth and dissolution by both desolvating the surface and perpetuating the natural corrugation of the surface, at a molecular level. 14 CHAPTER 2 CRYSTALLIZATION (a) (b) Figure 2-4 Scheme for ‘Relay Type’ Growth Mechanism [40] A general relay type mechanism is depicted in Figure 2.4. The difference between two types of sites (A & B) is emphasized by assuming a corrugated surface such that the A-type site is a cavity and the B-type site is on the outside upper surface of the cavity. Figure 2.4(a) shows the B-type sites blocked by solvent S and the A-type sites unsolvated. Thus solute molecules can easily fit into A-type sites. But once docked into position as seen in Figure 2.4(b) the roles of the A and B-type sites are essentially reversed and the solvent molecules which originally were bound to B-type sites would be repelled since they now occupy A-type sites. This cyclic process can lead to fast growth. In such a situation described here, where desolvation is rate limiting, it is implicitly indicated that the free energy of incorporation of a solute molecule helps to displace bound solvent. These models help in understanding the polymorph selection and morphology control of the final crystal. They also are helpful in explaining the inhibition of certain crystal forms even though they are not the most stable under the given experimental conditions. These vast variances in nucleation and crystal growth phenomena which eventually constitute the crystallization process lead to the formation of crystals with varied distribution of size, shape, colour and polymorphic form. 15 CHAPTER 2 2.3 CRYSTALLIZATION Polymorphism in organic compounds Polymorphism is an innate characteristic of a compound wherein the molecules can stack into different molecular conformations while retaining the same chemical composition [23]. In the pharmaceutical industry, a very large number of compounds exhibit the phenomenon of polymorphism. 70% of barbiturates, 60% of sulphonamides and 23% of steroids exist in different polymorphic forms [41]. The existence of polymorphism in the case of antiviral drug Ritonavir has had a dramatic commercial effect on pharmaceuticals. The manufacture of Norvir (commercial name for Ritonavir) semi-solid capsules formulation involved the preparation of a hydroalcoholic solution of Ritonavir which, although not saturated with respect to form I was 400% supersaturated with respect to form II. The sudden appearance and dominance of this dramatically less soluble crystal form made the formulation not manufacturable [42]. It was necessary to immediately reformulate Norvir. These factors combined to limit inventory and seriously threatened the supply of this life saving treatment for AIDS. Another classic example is polymorphism exhibited by compound ROY [39]. Six solvent free polymorphs of ROY are shown in Figure 2-5. Glycine, a simple amino acid, can pack itself in three different crystal structures. Figure 2-5 Polymorphs of ROY[39] Polymorphism and polymorphic transformations of organic systems have been studied extensively since development of X-ray diffraction techniques. Earlier studies mainly focused on characterization of crystal structures of various organic compounds and methods of manufacturing these polymorphs [43]. Recent studies typically examine conditions under 16 CHAPTER 2 CRYSTALLIZATION which polymorphic transformations occur including humidity, pressure, solvents, additives and other process induced transformations[44-48]. Early computational investigations typically focussed on ab initio polymorph prediction of various organic systems [49]. Recent studies on atomistic simulations of solutions and organic crystal interfaces provide molecular level insights into polymorphic selection [50, 51]. The influence of solvents on organic crystal polymorphism has gained importance because the use of solution phases as media for homogenization and crystallization for the subsequent assembly processes are common [52]. Understanding the mechanistic role of solvents in polymorph selection is of great importance. Solvent is an important consideration in solution crystallization, which affects the morphology, size distribution, downstream processing, as well as polymorphism of the final product. In pharmaceutical industries, solvent screening is the first and the most important step in polymorph study. Despite this, surprisingly, little is known about the molecular self assembly processes that surround the nucleation event and in particular the link between solution speciation, molecular aggregation and the nature of intermolecular interactions in the resulting crystals. This gap in understanding is particularly evident in systems which exhibit crystal polymorphism, when small changes in solvent choice and crystallization conditions can yield a new crystal structure. Several studies, both experimental and computational, have been conducted to gain insights into the effect of solvents on crystallization [50, 51, 53]. As discussed earlier, the process of crystallization from solutions too can be divided into two stages: the nucleation event to form an embryo and the growth of the embryo into a crystal. Solvent plays a major role during both these stages and hence influences the final crystal structure. The above stays as common ground, but crystallization can occur in a number of ways, the only criterion being the induction of supesaturation which is the driving force for crystallization. 17 CHAPTER 2 2.4 CRYSTALLIZATION Effect of supersaturation Supersaturation is the driving force for crystallization and is directly related to the chemical potential difference between the old and new phases.  (  old   new )   k bT k bT 2-15 Where µold is the chemical potential of the solution phase and µnew is the chemical potential of the crystalline phase. Physically, Δµ represents the gain in free energy per molecule associated with the passage of the old phase to the new phase with lower Gibbs free energy [54]. In terms of measurable quantities, supersaturation can be defines as a function of the activity coefficient or concentration of some crystallizing species, i [31].  a  C    ln  i   ln  i   ai , e   Ci ,s  2-16 Where ai is the activity of solute i, ai ,e is the activity at which the solute and the condensate are in phase equilibrium. Ci is the actual concentration of i and Ci,s is the solubility limit of i [55]. A supersaturated solution, although in a thermal equilibrium state, is not at a thermodynamic equilibrium state. Concentration fluctuations in the solution cause the solute molecules to form clusters. On a microscopic level, a dynamic situation exists where clusters, in the form of dimmers, trimmers, tetramers and longer chains, are continuously formed and broken up. Eventually, when the cluster size reaches a critical number, a stable nucleus is born [56]. Among the many factors that affect crystal nucleation and growth kinetics, supersaturation is a prime factor, directly influencing the number, size, habit, polymorphic form and the structure of the final products. Supersaturation can be achieved in a number of ways: by cooling solution, by evaporating solvent, by adding an anti-solvent, by changing the pH, by chemical reaction or by a combination of the above. 18 CHAPTER 2 a) CRYSTALLIZATION Cooling: The solubility of a solute is a function of temperature, usually decreasing as the temperature is lowered. It is the most widely used method for generating supersaturation. This type of temperature-dependent solubility data is often used in industry to determine the cooling profile in a crystallizer [56, 57]. b) Evaporation: If the solvent is evaporated, either by heating the solution to its boiling point or by convective mass transport or by vacuum suction, the solute concentration is increased, generating supersaturation. c) Addition of Anti-solvent: Adding an anti-solvent, in which the solute solubility is low, to a solvent in which the solute solubility is high, generates supersaturation since the solute solubility in the final solvent composition becomes lower than the initial solubility in the solvent. The degree of supersaturation that can be achieved with this technique depends on the difference in solute solubility in the solvent and the anti-solvent. d) Changing pH: the solubility of a solute can also depend on the pH of the solution. In those cases, the pH can be modulated to decrease the solubility and generate supersaturation. e) Chemical Reaction: A chemical reaction can be used to decrease the solubility of the dissolved solute, creating supersaturation. For example, reactive species such as ions that precipitate in the precipitation of the solute can be added for this purpose. In this study, crystallization by the addition of anti-solvent has been used as a platform to fundamentally understand the process. 19 CHAPTER 3 POLYMORPHISM OF GLYCINE 3. Polymorphism of Glycine 3.1 Glycine polymorphs Three crystalline polymorphs were described for glycine: two monoclinic (α, space group. P21 [40, 58, 59] ) and (β, space group. P21/n [5, 60]) and one trigonal (γ, space group. P31 / P32 [37, 61, 62]). The three polymorphs differ in the way how NH3+.CH2.COO- Zwitterions as shown in Figure 3-1, are linked together via hydrogen-bond networks. In the α-polymorph zwitterions are linked by hydrogen bonds in double antiparallel layers, the interactions between these double layers being purely van der Waals. In the β-polymorph individual parallel polar layers are linked by hydrogen bonds in a three-dimensional network. In the γpolymorph consists of polar helixes linked with each other in a three-dimensional polar network. The first attempt to study α-glycine by means of X-ray diffraction were undertaken by Bernal [63] and by Hengstenberg and Lenel [64] independently. The explicit crystalline structure was refined by Albrecht and Corey [65], and a precise refinement of not only the heavier atoms but also locating of the hydrogen atoms for α-modification was carried out by Marsh [66]. Figure 3-1 A zwitterionic glycine molecule. Label: Red-Oxygen, Blue-Nitrogen, Cyan-Carbon, White-Hydrogen In the α-polymorph the zwitterions are linked by hydrogen bonds in double anti-parallel layers, the interactions between these double layers being purely van der Waals as shown in Figure 3-2. β-glycine has already been obtained and described by Fischer [67] at the 20 CHAPTER 3 POLYMORPHISM OF GLYCINE beginning of the century, whereas X-ray examination was performed quite a long time later [60]. It was suspected that this fact was due to the general low stability of this phase and possibly due to irreversible transformation into α or γ-glycine forms. Figure 3-2 α-glycine crystal structure displaying hydrogen bonded bilayers [68] In the β-polymorph individual parallel polar layers are linked by hydrogen bonds in a three dimensional network as seen in Figure 3-3. Figure 3-3 β-glycine crystal structure [68] The structure of γ-glycine was first revealed and resolved by X-ray diffraction by Iitaka [61, 62]. Moreover, he analyzed not only the packing of the molecules in the crystal lattice, but also the role of the hydrogen bonds in the formation of the framework and the substructures of the hydrogen bonds. Investigations of γ-glycine at 298 and 83 K were carried out by means of neutron diffraction by Kvick [69] in order to determine electron density changes of the molecules with temperature. The γ-polymorph consists of polar helices linked with each other in a three-dimensional polar network as seen in Figure 3-4. 21 CHAPTER 3 POLYMORPHISM OF GLYCINE Figure 3-4 γ-glycine crystal structure displaying polar helical hydrogen bond structure [68] The γ-polymorph is the most stable form at ambient conditions, although the α-form crystallizes much more readily, and the α-form (with rare exceptions) was not observed to transform into the γ-form at these conditions. With increasing temperature, the order of stability inverts, the α-form becomes the most stable one above ~440 K, and a γ α polymorph transition is observed when the γ-form is heated. On subsequent cooling, the αform does not transform back to the γ-form, presumably due to kinetic reasons. The β-form is obviously unstable at all temperatures [63]. 3.2 Effect of solvents on glycine polymorphism Experimentally, the unusual feature of glycine is that crystallization from aqueous solution at neutral isoelectric pH (5.97) always gives α polymorph and it seems that under these conditions γ never appears, despite its thermodynamic stability. It has been previously concluded that the nucleation and apparent ‘stability’ of the metastable α form at pHs close to the isoelectric point is a reflection of the presence of centrosymmetric dimer ‘growth units’ in solution. On the basis of solution [70], interfacial and solid-state chemistry [71] it was suggested that at and around the isoelectric point, glycine is dimerized in solution as centrosymmetric pairs of zwitterions. Myerson and Lo predicted that glycine exists mostly as dimers in supersaturated solutions by measuring the diffusion coefficient for supersaturated aqueous solution of glycine [72]. Using small-angle X-ray scattering, Chattopadhyay et al. have directly studied the nucleation of glycine from its aqueous supersaturated solution and indicated that glycine molecules exist as dimers in the supersaturated solution [73]. It is then apparent that nucleation from such solution could lead directly and spontaneously to the 22 CHAPTER 3 POLYMORPHISM OF GLYCINE metastable α structure [48]. This is further supported by the experiments involving S-control (supersaturation control) by Chew et al. where they have shown that α-glycine grows about 500 times faster than γ-glycine in neutral aqueous solutions. This difference in growth rate corresponds to a difference in activation energy for growth of ~15kJ mol-1 calculated from the Arrhenius equation. This large difference in activation energy is attributed with the dissociation of glycine dimer in solution prior to growth of γ-glycine, but their preservation in the α-glycine crystal structure [74]. The next interesting issue is why γ-glycine forms at low pH aqueous solutions. In each of its polymorphic form glycine molecules pack as zwitterions. As discussed above, the ‘stability’ of the metastable α form at pH close to the isoelectric point is a reflection of the presence of centrosymmetric dimer growth units in solutions. The effect of moving the pH away from the isoelectric point is in reducing the proportion of the α-form growth unit (because singly charged glycine molecules will not form cyclic dimmers). This would increase the proportion of monomeric zwitterions available to form the polar chain structure of the γ-polymorph. In this work, we examine selective polymorphism of γ-glycine from ethanol-aqueous solutions in the presence of an aqueous-organic interface. The experimental results and underlying hypothesis (based on literature) is discussed in the following sections. 3.3 Glycine polymorphic selection in water-methanol/ethanol systems As discussed previously, α-glycine crystallizes primarily in aqueous solutions through hydrogen-bonded cyclic dimer growth units. The precipitation of γ-glycine at low or high pH aqueous solutions had been explained successfully by Davey et al. [75]. Another interesting case of polymorphic selection of glycine is the precipitation of β-glycine in alcohol-water solutions. The conundrum that the most thermodynamically stable γ-glycine polymorph does not generally precipitate in aqueous solutions containing methanol or ethanol under the specified experimental conditions was addressed from the growth kinetics of the three polymorphs of glycine coupled with an analysis of the action of solvent at the various crystal faces by Weissbuch et al. [5]. 23 CHAPTER 3 POLYMORPHISM OF GLYCINE 3.3.1 β-glycine precipitation The first crystallization of β-glycine from water-alcohol solutions was reported by Fischer [67]. The crystal structure [60] is polar (space group P21/n) and comprises hydrogen-bonded layers, which are similar to those observed in the α form, but which are interlinked by NH-O and CH-O interactions through a twofold screw-symmetric axis perpendicular to the layer plane (Figure 2-3). The addition of alcohol reduces the solubility of glycine from 25.0g/100 ml water (25 oC) to 2.65 g/100 ml solvent in 50.1 %( v/v) ethanol-water mixtures. It was hypothesized that this reduced solubility would result in an increased concentration of solvated glycine monomers relative to that of hydrogen-bonded cyclic dimers. Such behaviour is apparently consistent with the preferred precipitation of β-glycine from alcoholwater solutions because crystal structure consists of hydrogen-bonded monomer units, as opposed to α-glycine which comprises cyclic hydrogen-bonded pairs. Long needles of β-glycine were grown in water-ethanol mixtures containing 50, 26.1 and 10% (v/v) ethanol and also from 1:1 water-methanol mixtures containing 4.0, 19.0, 35.9 and 5.0g glycine/100 ml solvent, respectively. Growth kinetics measurements of single β-glycine crystals in 1:1 water-ethanol solutions at 25 oC reveal a fast growth at one pole of the needle and a very slow growth at the opposite end. The absolute polarity [76] of β-glycine was determined by employing “tailor-made” additives [44], in this case racemic tryptophan (Trp). It was concluded that β-glycine grows faster at the side with exposed C-H bonds (coloured pink; Figure 3-5). Figure 3-5 Packing arrangement of β-glycine. The (010) “azure” and ( exposed at the interface [5]. “pink” surfaces are 24 CHAPTER 3 POLYMORPHISM OF GLYCINE Previous studies have shown that the relative rates of growth at the opposite ends of polar crystals in polar solvents can be correlated directly with the relative rates by which solvent molecules are stripped from the opposite ends [36, 77-79]. The faster growth rate at the βglycine pole with expose C-H bonds is in agreement with this model; the water or alcohol solvent molecules can be attached more effectively to the slow growing glycine surface with exposed N-H bonds through strong OHsol .... Ogly- and NHgly+ .... Osol than to the fast growing βglycine pole with exposed C-H bonds with strong OHsol .... Ogly- interactions but only weak CHgly .... Osol interactions. 3.3.2 γ-glycine inhibition The absence of the stable γ-glycine form in crystals formed in alcohol-water solutions is explained by examination of its growth properties (refer section 2.2.2). The polar crystal structure of γ-glycine (space group P31; Figure 3-7), which is not composed of cyclic glycine pairs, is delineated by a ( ) face at which CO2- groups emerge and capped crystal faces at the opposite end that exposes NH3+ groups. Previous studies [80] have shown that γ-glycine, grown in aqueous solutions and in the presence of auxiliaries e.g. racemic hexafluorovaline, that inhibit the crystallization of α-glycine, appear as (001) needles that grow along the polar c axis much faster at the end of the crystal with the CO2- groups than at the opposite capped end. This unidirectional growth was interpreted in terms of a “relay” mechanism. However, ethanol and methanol solvent molecules can reside within the pockets through OHsol .... Oglyand CHalcohol .... Ogly- hydrogen-bonding interactions, thus inhibiting growth at the CO2- end of the crystal (Figure 3-7). In several of the crystallization experiments carried out in waterethanol mixtures, the few γ-glycine crystals that were observed exhibited morphology in keeping with the proposed inhibition by ethanol or methanol of growth along the otherwise fast growing CO2- end of the crystal (Figure 3-6). 25 CHAPTER 3 POLYMORPHISM OF GLYCINE Figure 3-6 γ-glycine crystals obtained from 1:1 water ethanol solution [5]. Figure 3-7 Packing arrangement of γ-glycine showing the pockets of the fast growing ( face that are poisoned by the adsorption of ethanol and methanol molecules (shown as ‘balls and sticks’)[5]. 3.3.3 α-glycine formation The surface of the fast and slow growing ends of α-glycine are very similar in structure to the (010) surface of β-glycine with either exposed C-H or N-H bonds, as shown in Figure 3-8 (a) and (b) respectively. On the basis of assumption, which is supported by experimental evidence [71, 81], that glycine molecules in aqueous solution dock onto the crystal surface primarily as hydrogen-bonded cyclic glycine pairs, it is thought that a (010) face will expose the faster growing surface with exposed C-H bonds to a much larger extent than the slower growing surface with exposed N-H bonds. 26 CHAPTER 3 POLYMORPHISM OF GLYCINE (a) (b) Figure 3-8 Packing arrangements of α-glycine (a) exposing weak binding C-H bonds to the solution at (010) surface (azure) or (b) exposing strong solvent-binding N-H bonds to the solution at the (010) surface (pink) [5]. It was anticipated that reduced solubility of glycine in solution caused by the presence of alcohol would lead to a higher proportion of solvated glycine monomer units docking onto the α-glycine (010) surface sites with exposed N-H bonds. Thus, the time required to strip the overlying solvent molecules, prior to formation of the glycine cyclic dimer growth units and propagation of the glycine bilayer with exposed C-H bonds on its (010) surface, would lead to an overall reduction in growth rate along the b directions of the α-glycine crystals. Indeed, the α-glycine crystals obtained from a 9:1 water-ethanol solution tended to display more well developed (010) faces (Figure 3-9) than crystals obtained from purely aqueous solutions. Therefore embryonic crystallites would expose slow-growing (010) surface at higher concentrations of the alcohol in contrast to β nuclei, which has only one slow growing polar end and so results in a preferred kinetic precipitation of the latter. Thus, water or alcohol as solvent impeded growth normally on the (h01) faces of the needle crystals as a result of strong solvent attachment on these faces through OHsol .... Ogly- and NHgly+ .... Osol hydrogen bonding interactions. Figure 3-9 α-glycine crystals obtained from a 9:1 water-ethanol solution [5]. 27 CHAPTER 4 4. INFLUENCE OF SURFACE/INTERFACE ON CRYSTALLIZATION Influence of surface/interface on crystallization In the word ‘polymorphism’, poly means ‘many’ and morph means ‘form’ in Greek. McCrone defines it as a solid crystalline phase of a given compound resulting from the possibility of at least two different arrangements of the molecules of that compound in the solid state[82]. Due to the difference in molecular arrangement, the physical and chemical properties of different polymorphs such as melting point, bulk density, solubility and dissolution rate deviate significantly from each other as well[83]. But the crystal morphology is inherently controlled by its polymorphism as well. Therefore, the control of crystal polymorphism is equally important as the control of crystal size and shape in API production. To control the physical and chemical properties of crystals, the effect of various process variables have been examined, such as crystallization temperature, mixture of solvent and additive type etc [84, 85]. Among all, the use of foreign surfaces to control crystal properties is gaining more attention. 4.1 Templated crystallization in nature What has recently gained focus has been utilized by nature for millions of years. Coccolithophores, a group of unicellular plant planktons, produce exoskeletons of minute calcite plates called coccoliths (shown in Figure 4-1), whose remarkable structural hierarchy is characterized by precise control of nucleation and growth of calcite at fluid interfaces using organic templates[86]. Figure 4-1 Coccolith formation of E. Huxleyi 28 CHAPTER 4 INFLUENCE OF SURFACE/INTERFACE ON CRYSTALLIZATION In essence, biominerals are actually composites of crystals separated by organic material. A concerted self-assembly process, all in the absence of complex organic additives or surface scaffolds yields a highly ordered arrangement of carbonate crystallites over micrometer length. Researchers in the field of API crystallization have noticed the important role that a surface could play in heterogeneous nucleation, and have started to use surfaces and interfaces to monitor crystallization. This very idea was demonstrated to understand crystallization using a surfactant self-assembly as templates for nucleation at a liquid-liquid interface [87]. 4.2 Interaction between crystals and surfaces Solid surfaces in general aid crystal nucleation due to two vital reasons, which are reducing nucleation activation energy and controlling molecular arrangement. As a result, the crystal polymorphism, morphology and crystal size can be tuned by tailoring surface. Moreover, the presence of a foreign surface in any crystallization environment has a vital role to play since crystals may interact with the surface by electrostatic interactions as well as by hydrophobichydrophobic interactions. For instance, glycine, that is known to have both positive and negative charges on the crystal facets, was found to interact with glass surface, as shown in Figure 4-2. Figure 4-2 Self Assembled structures of Glycine on a glass surface. 29 CHAPTER 4 INFLUENCE OF SURFACE/INTERFACE ON CRYSTALLIZATION This may result from electrostatic interaction and/or H-bonding with the hydroxyl group on glass. It is due to these interactions that a foreign surface helps overcome the energy barrier for nucleation. This scenario has been well exploited by the scientific community to study surface induced crystallization. In such studies surfaces have been modified with a number of methods wherein they have a wide range of surface properties which in turn influence the properties of crystals that nucleate and grow on them. A typical example is shown in Figure 4-3, where polymorphism and size of glycine crystals have been controlled using engineered bifunctional patterned surfaces[88], covered with hydrophilic SAM (aminopropyltriethoxysilane) and hydrophobic SAM(octadecyltrichlorosilane). (a) (b) Figure 4-3 (a) Bifunctional SAM pattern on surfaces; (b) Glycine crystals on patterned surface[88]. SAMs are ordered molecular assemblies spontaneously formed due to the adsorption of surfactants with a specific affinity of their head group to the substrate. Due to their uniform molecular orientation, SAMs could serve as a template and in turn direct the solute molecules to align orderly during nucleation. Lee et al. reported controlled crystallization of glycine on patterned gold square islands in 2006 [83]. In 2011, Kim et al. managed to obtain nanosized glycine crystals from silane-based bifunctional SAMs [88]. In the above studies, constrained small volumes were provided by the SAM islands and hence the size of crystals produced on each island was controlled. Initial glycine concentration, rate of cooling and rate of solvent evaporation were manipulated for generating different polymorphs. Besides SAMs, polymer 30 CHAPTER 4 INFLUENCE OF SURFACE/INTERFACE ON CRYSTALLIZATION substrates have also been studied for heterogeneous nucleation and crystallization [89, 90]. Diao et al. utilized polymer materials, such as PEGDA, poly (4-acryloylmorpholine) and poly (2-carboxyethyl acrylate) for aspirin crystallization and the crystallization kinetics and final crystal size was controlled by polymer surface porosity and pore size. There have been quite a few studies on surface modification techniques for glass and PDMS(Poly Dimethyl Siloxane), including exposure to energy such as oxygen plasma, UV light etc., covalent modification such as radiation induced graft polymerization, silanization etc., and physical deposition of other materials[91-93]. This is because these modifications have a potential to generate superhydrophobic surfaces. There has been a lot of interest in fabricating superhydrophobic surfaces because of its self-cleaning properties[94]. For crystallization a nonwetting surface can eliminate heterogeneous nucleation (see heterogeneous nucleation, Chapter 2). 4.3 Crystal interaction with liquid interfaces in emulsion based crystallization Emulsion Based Crystallization is a relatively novel approach used in API manufacturing studies. It is part of a wider set of techniques such as seeded crystallization, cooling of melts, de-sublimation, spray drying, mixed-solvents, use of additives[95] which are used to produce crystals with specific polymorphism. As emulsion-based crystallization utilizes liquid-liquid interfaces as the nucleation sites, the functional role that liquid interfaces could play in crystallization is highly significant. In emulsion-based crystallization, the crystal size is controlled as the volume of solute is partitioned by the dispersed droplets [96]. Surfactants added to the system could align orderly at the droplet interface due to amphibian interaction, which in turn directs an orderly orientation of solute molecules [87, 97, 98]. For instance, Ueno et al. noticed that the morphology of n-alkanes crystals changes complete when highmelting point surfactant was added into the oil-in-water emulsions [98], while Allain et al. found that the potassium hexacyanoferrate(III) crystals generated from water-in-oil emulsion appear much more regular with the presence of octadecylamine monolayer [87]. The liquid interface could also serve as a barrier for mass transfer in QESD, thus the rate of 31 CHAPTER 4 INFLUENCE OF SURFACE/INTERFACE ON CRYSTALLIZATION supersaturation generation is controlled [99, 100]. In addition, interfacial tension at the interface of droplets was shown to be able to influence the crystallization process as well [101-103]. But the challenge remains in not just controlling the morphology of crystals but also the polymorphism of API. As a resort to this issue, Emulsion based crystallization is seen an attractive process platform to simultaneously control both selective nucleation of a specific polymorph, while producing crystal agglomerates of desired size and shape that greatly accelerates product formulation and eliminate costly downstream processing like dry milling and grinding. To state an example as Quasi Emulsion Solvent Diffusion (QESD) crystallization, the API is initially dissolved in solvent and mixed with anti-solvent at a different temperature[104]. The ensuing phase separation, possibly transient, generates droplets in which the solute-solvent mixture constitutes the dispersed phase and anti-solvent is the continuous phase. The diffusion of anti-solvent into the droplet creates supersaturation, which in turn drives the crystallization of solute, thereby leading to nucleation and crystal growth within the droplet. Restriction of the crystallization environment within the droplet boundary causes the formation of spherical crystal agglomerates within the droplet. The applications of QESD in pharmaceutical crystallization have been demonstrated on a number of pharmaceutical solutes [104-106]. Mathematical model of mass transfer inside droplets in QESD was been developed[107], but has found limited use in predicting crystal size distribution due to lack of detailed experimental data. In a few other studies, specific polymorphic crystal forms of organic molecules such as glycine, L-glutamic acid HCL and ephedrine HCL have been obtained using water-in-oil emulsions using amphiphlic additives[96, 98, 108]. These additives self-assemble at the water-oil interface and presumably create a structure template that promotes nucleation of a specific polymorphic form. In spite of demonstrations that showcase emulsion based crystallization as a promising technique for precise control over particle size, morphology and polymorphism through engineering of emulsion interfaces and control of supersaturation within droplets, crystallization at or in the vicinity of a liquid-liquid interface, with or without templating additives, is a complex and poorly understood phenomenon. It involves a host of interacting 32 CHAPTER 4 INFLUENCE OF SURFACE/INTERFACE ON CRYSTALLIZATION processes such as interfacial molecular recognition, exchange of material across phases and phase transformation. In addition, the contributions of different physical processes that may interfere in the final crystal outcome are yet to be addressed. A typical example where the study aimed at disentangling the relevant major factors that contributed to crystallization at liquid-liquid interfaces was by Allain et al., where controlled nucleation of KFC in isolated microdroplets at liquid-liquid interface was performed as shown in Figure 4-4. Figure 4-4 Schematic of surfactant-monolayer-templated nucleation and crystallization[87] With lack of understanding of crystallization at liquid-liquid interfaces, the setup in addition has an added complication of a surfactant monolayer at the oil-water interface which exerts additional effects into the apparent mechanism of the process, which sets the understanding on unfirm ground. This is because only little is conclusively known about surfactant assemblies at the liquid-liquid interface compared to that at a liquid-vapour interface[87]. This calls for a simplistic scheme that can help study and understand some of the key aspects of crystallization at liquid interfaces. The above existing challenges serve as motivational tools for the undergone work in this thesis. We use glycine as a model system for the crystallization studies here. A detailed experimental study on glycine crystallization using solvent mixtures at the vicinity of a liquid-liquid interface is done to study their polymorphic outcome, morphology, size and shape as reported in the following chapters. 33 CHAPTER 5 5. EXPERIMENTAL Experimental This chapter describes the experimental work undergone starting from batch crystallization to interfacial crystallization experiments, crystal characterization using XRD and SEM and optical microscopy. The data analysis under each experimental section is also discussed in detail here. 5.1 Materials and Methods Glycine (≥ 99% pure), n-hexane (HPLC grade, 95%) and anhydrous ethanol (99.86%) was purchased from Sigma Aldrich. Ultrapure water was obtained from MilliQ water system. Commercial glycine was dissolved in water (25g/100ml water), heated and stirred for 5 hours at 23.5oC using a magnetic stirrer-heater system followed by cooling to room temperature. The prepared saturated glycine solution was filtered using a 0.45 µm non-sterile syringe filter purchased from Cole-Parmer. Hexane and ethanol were filtered using a syringe filter to avoid presence of dust and impurities.1000 µL single-channel pipettes and suitable pipette tips were purchased from Eppendorf Research and ensured that they are dust free before usage. 10 mL screw neck vials were used for the batch crystallization experiments. UV grade quartz cuvettes with light path of 10 mm were purchased from Hellma Analytics. The cuvettes were air blown and assured for absence of any foreign impurities along the inner walls so as to provide a pristine environment for crystallization. Harvard PHD 22/2000 series syringe pump was used for regulated flow at µL scales. Fittings purchased from Upchurch Scientific which include flangeless nut 1/16 in PEEK short, flangeless ferrule 1/16 and ¼ - 28 female to female Luer were used to connect the dispensing end of the syringe to the Teflon tubing (1/16 in OD, 0.010 in or 0.25 mm ID). 1cc sterile, non-toxic, non-pyrogenic, latex free Terumo syringes and sterile, non-toxic, non-pyrogenic single use Terumo needles (21 G * 1 ½”) were purchased from Terumo Corporation, Japan. Real-time imaging was performed using a Leica MZ16 stereo microscope/camera system that can capture digital image sequences at rates of 1000 frames per second. A Leica CLS 150 XE light source was used for precision view. 34 CHAPTER 5 EXPERIMENTAL Whatman- grade 1, 4.7 cm filter paper was used for sample filtering and collection. Vacuum of 0.08MPa was generated by Vacuumbrand ME2 vacuum pump for drying the filtered samples. Crystals were ground into a fine powder for XRD analysis using a porcelain mortar and pestle set. 5.1.1 X-Ray Powder Diffraction (XRD) The X-ray Powder Diffraction (XRD) Analysis for the generated sample was performed by a LabX XRD-6000 Shimadzu X-ray diffractometer using characteristic Cu-Kα radiation to investigate their polymorphism. A 0.3mm diameter collimator was used and data for each sample were collected over a 2 theta range with chi (tilt) of 5o. The sample powder pressed into the cavity of an aluminum die (metal sample holder) was mounted onto a motorized stage which was used to bring the sample into the exact focus of the beam. The X-ray diffractometer operated at 40kV, 30 mA and a scanning rate of 2 o/min over the range 2θ = 1060o, using Cu-Kα radiations. The sample was rotated in order to increase signal to noise ratio. The X-ray Powder diffraction pattern for the sample was obtained whose well defined peak patterns are characteristic of the various lattice planes of specific crystal form. The raw spectrum of all the samples was found to contain large background. This was a result of the aluminum die used to station the sample. However having known the peak positions of background aluminum they were easily subtracted in the X-ray data from that of the sample. Al Al Figure 5-1 Raw XRD pattern displaying peaks corresponding to background Aluminium along with glycine peaks. 35 CHAPTER 5 EXPERIMENTAL The raw XRD pattern is presented in Figure 5-1, showing background aluminum peaks at 2θ = 38, 39 and 45 o along with sample glycine peaks. The X-ray diffraction pattern from a solid results from the satisfaction of the Bragg condition (nλ = 2d sinθ), where λ is the wavelength of the X-ray radiation, and d is the particular spacing between individual parallel planes. The condition can be satisfied when the angle θ between the incident radiation and that set of planes results in constructive interference. The X-ray powder diffraction pattern of a solid is thus a plot of the diffraction intensity as a function of 2θ values (or equivalently, d spacings) and may be considered to be a fingerprint of that solid. The values of the d spacing reflect the dimensions of the unit cell, while the intensities are due to the contents of the unit cell and the way the atoms and molecules are arranged therein. As polymorphs comprise different solids with different unit cells and different arrangements of molecules within the unit cell they have different fingerprints most often as different as the X-ray powder patterns of two different compounds. Thus X-ray powder diffraction is probably the most definitive method for identifying polymorphs and distinguishing among them. 5.1.2 Semi-Quantitative analysis of XRD data Integrated Intensity Analysis The raw XRD patterns obtained after analysis require processing to retrieve information from them regarding the polymorphic outcome of glycine obtained under each sample set. The fraction of glycine polymorphs for crystals collected from various experiments was qualitatively estimated by performing an Integrated Intensity Analysis for the XRD pattern. For glycine, three polymorphs are reported so far which are α, γ, and β forms. As mentioned previously each polymorph has its characteristic peak positions in the XRD pattern which are finger prints to identify its occurrence. Based on their existence, specific peaks corresponding to specific polymorphs tend to be dominant over other peaks due to greater diffraction intensity by specific crystal lattices of the polymorph formed. There are more than one dominant peak positions for each polymorph of glycine on the XRD pattern. The dominating peak positions for each polymorph of glycine are listed in Table 5-1. 36 CHAPTER 5 Glycine Polymorphism Dominating Peak Position (2θ) EXPERIMENTAL α γ β 14.7 14.5 18.9 23.8 21.7 24.8 30.0 25.2 28.4 35 29 Table 5-1 Dominant peak positions for different polymorphs of Glycine After extracting away the background intensity, the area under each peak is computed by numerical integration using MATLAB. The fraction of each polymorph formed can be estimated by dividing the area under its characteristic peaks with the total area for all characteristic peaks of all existing polymorphs. For instance, the XRD pattern for glycine crystals obtained from the interfacial crystallization experiment (ethanol perfused at 10 µl/min into the hexane-glycine system) is shown Figure 5-2. According to the XRD pattern, the sample possesses the characteristic peaks for both α and γ polymorphs, while β glycine is absent in the sample. The areas under the peaks for each polymorph are computed and the total area under all characteristic peaks is computed too. The fraction of each glycine polymorph is then estimated by dividing the area for each polymorph present by the total area of all possible polymorphs expressed. MATLAB code for the above analysis can be referred from Appendix A. 37 CHAPTER 5 EXPERIMENTAL Figure 5-2 Analysis of XRD Pattern for Glycine obtained from interfacial crystallization experiment with an ethanol perfusion rate of 10µL/min. 5.1.3 Scanning Electron Microscopy (SEM) Scanning Electron Microscopy was performed using a JOEL JSM-5600 LV SEM setup with a resolution of 3.5 nm (magnification: ×25 to ×300,000) to study the morphology of crystals by generating high-resolution images. The system operates at accelerating voltages of between 130 kV. This is a high vacuum and partial vacuum (10 Pa-10-4 Pa) SEM with secondary electron detector based on the scintillator-photomultiplier design of Everhardt and Thornley. Also fitted are an INCA x-act SDD detector for high resolution imaging along with solid state backscattered electron detector for compositional and topographical information. The sample is mounted on double-sided adhesive tape adhered to the SEM stubs and is aligned into the sputter coater. The sputter coating is done with Pt to make the sample conductive before examination.Scanning electron microscopy (SEM) provides greater magnification than optical microscopy. In the study of polymorphs, it can be very useful in characterizing and understanding differences in the shape of polymorphs. The differences in 38 CHAPTER 5 EXPERIMENTAL morphology (bypiramidal and pinacoidal) which is clear from Figure 5-3 might not be obvious in optical microscopy for samples of small grain size (Figure 5-4). (a) (b) Figure 5-3 SEM images of two different glycine crystal morphologies. Figure 5-4 Optical microscopy images of different glycine crystal morphologies. Scanning electron microscopy is particularly useful for the investigation of the properties of surfaces. The characterization of habit and surface features, the structural symbiosis between two crystalline modifications can be readily studied by SEM. Such information can considerably aid in the understanding of the process of transformations among crystal modifications and the development of robust procedures for the selective preparation of a desired crystal modification. 39 CHAPTER 5 5.2 EXPERIMENTAL Batch crystallization experiments The crystallization experiments begin with batch crystallization of glycine in solvent mixtures using ethanol as an anti-solvent. Here we try to investigate the different morphologies and polymorphic outcomes that glycine can acquire with a change in the volume fraction of antisolvent ethanol in an ethanol-water mixture.1 mL ethanol and 1 mL saturated glycine solution were drawn using a 1000 µL Eppendorf single-channel pipette and dispensed into a 10 mL screw neck glass vial which was air blown using an air gun. The content was shaken rapidly using a VELP scientifica vortex mixer at 30 rpm for a few seconds. Instantaneous nucleation followed by growth of glycine crystals was observed. A tentative schematic of the experimental procedure is as shown in Figure 5-5. Figure 5-5 Schematic for batch crystallization with different ethanol volume fractions (10-80%). The formed crystals were harvested and filtered using a filter paper and dried under vacuum for 3-4 hours. The crystal samples were used for morphology analysis using Optical microscopy and Scanning Electron Microscopy (SEM). Optical microscopy was performed using the Leica stereo microscope/camera system. A Leica CLS 150 XE light source was used for precision view. The crystal samples were placed on a clean glass slide and positioned right beneath the eye piece within the field of view. The magnification was set to 7.1X at the eyepiece and still images of the sample were recorded. A few crystals from the sample set 40 CHAPTER 5 EXPERIMENTAL were used for image analysis using SEM. The optical microscopy images of samples collected from various batches are shown in Figure 5-6. (010) (a) (b) Figure 5-6 Optical microscopy images of crystals from (a) 40%vol (b) 80%vol (Scale-bar size 300µm). Having studied the external characteristics of the crystal such as morphology, habit and crystal size the next step was to characterize the polymorphic form of the crystal set. X-Ray Powder Diffraction (XRD) Analysis was chosen for polymorph characterization. The sample crystals to be characterized using XRD were first ground into a fine powder using a porcelain mortar and pestle. The finely ground powder was then pressed into a small pellet using a metal press into the cavity of an aluminium die. The surface of the pellet was smoothened by grazing the press over and over until it is horizontally in surface with the aluminium die as shown in Figure 5-7. The prepared die was placed in position within the XRD housing and the chamber was closed for x-ray generation and the resulting diffraction was obtained as the XRD pattern for the sample set. Figure 5-7 Powdered glycine pressed into the cavity of XRD metal sample holder (top and orthographic) (Scale-bar size 1 cm) 41 CHAPTER 5 EXPERIMENTAL The XRD pattern is a plot of the diffraction intensity against the grazing angle, 2θ. The peak positions are characteristic of a certain polymorph and the peak intensities are a measure of the occurrence of specific crystal planes in the polymorph. An analysis of such patterns is discussed in the following chapter. The above explained procedure was followed for a list of ethanol volume fractions (1080%vol) in water for crystallizing 1ml saturated glycine solution in the batch. There was no crystallization observed for the ethanol volume fractions of 10-30% vol. For the other batches from 40-80% vol, crystallization occurred in time span of few hours to a few seconds. The formed crystals were collected following the procedure as mentioned previously and image analysis was done followed by polymorph characterization using XRD. The XRD pattern overlay for glycine crystals grown in the ethanol volume fractions from 40-80% is shown in Figure 5-8. α γ Figure 5-8 XRD pattern overlay for glycine crystals from various ethanol fractions (40-80%). 42 CHAPTER 5 EXPERIMENTAL 5.2.1 Data Analysis for batch crystallization As mentioned, batch crystallization was performed using saturated glycine solution by varying the volume fraction of ethanol in water (10-80%vol). The batches for 10-30% ethanol volume fraction did not result in crystallization whereas the 40-80% volume batches resulted in crystallization of glycine. The formed crystals were characterized for their morphology and polymorphic form using optical microscopy, Scanning Electron Microscopy (SEM) and X-ray Powder Diffraction (XRD). The optical microscopy results for the various batches of crystals for different ethanol fractions from 40-80% vol were studied for their morphology and habit distribution. Most of the crystal samples in all of the batches had bypiramidal, pinacoidal and needle shaped morphologies. However the crystal samples from 40% vol ethanol fraction batch had mostly crystals of the bipyramidal shape and the crystal samples from the 80% vol ethanol fraction batch had mostly crystals of the pinacoidal shape. The difference in habit which is clear from Figure 5-6 can be attributed to level of supersaturation[109], presence of solvent mixtures which influence the crystal habit due to change in growth rate of certain crystal facets[110]. The presence of ethanol-water mixture in varying volume fractions cause a change in morphology as a result of which the 80% vol fraction batch shows a well developed (010) surface compared to the crystals in the 40% vol fraction batch. This clearly shows that ethanol concentration influences crystal morphology. These morphologies are very close to previous experimental observations for alcohol/water mixtures available in literature (refer Figure 3-9). The crystal samples after analysis for their morphology and shape were then analyzed using X-ray Powder Diffraction analysis for polymorph characterization. The explained procedure for XRD was followed as per mentioned in section 5.1.1. The resulting powder diffraction pattern for each sample set was individually obtained and analyzed. The definite peaks at various 2θ scales are reflective of the occurrence of the corresponding lattice planes in the crystal sample set. The dominant peak at 2θ=30o is indicative of the dominance of α-glycine over the β and γ forms. And this trend is observed in all the batches for the different ethanol 43 CHAPTER 5 EXPERIMENTAL fractions from 40-80% vol as shown in Figure 5-8. This clearly indicates that irrespective of the change in the volume fraction of ethanol the polymorphic outcome of the system is fixed with α-glycine being the dominant form of all, although the presence of β and γ forms of glycine in small amounts is inevitable. Using Integrated Intensity Analysis (refer section 5.1.2) the fraction of various polymorphs of glycine expressed were estimated for crystals obtained from all the ethanol volume fractions. The plot as shown in Figure 5-9 which is the fraction of glycine polymorph obtained as a function of ethanol volume fraction in water clearly displays the relative amounts of α-glycine expressed as compared to β and γ-glycine for each batch and is indicative of the dominance of α-glycine in water-ethanol mixtures. Although α-glycine is the dominant polymorph of occurrence, concomitant crystallization of all three polymorphs of glycine is also seen. Figure 5-9 Plot of polymorphic outcome of glycine for various ethanol volume fractions. 44 CHAPTER 5 5.3 EXPERIMENTAL Direct perfusion From the batch experimental sets one of the volume fraction of ethanol and water i.e. 50% vol ethanol in water was chosen as a sample for the direct perfusion experiments. In direct perfusion, with the ethanol to water ratio being fixed as 1:1, fixed volume of ethanol was perfused directly into the saturated glycine solution using a syringe pump. The experiment was undertaken to study the effect of perfusion rate on glycine polymorphism, morphology, crystal shape and size. Following a similar protocol as previously mentioned the following were performed. 1 ml of ethanol was drawn into a 1 cc Terumo syringe and ensured that it was devoid of any air bubbles within. 1mL saturated glycine solution was drawn using a pipette and dispensed into a quartz cuvette which was air blown using an air gun. Fittings purchased from Upchurch Scientific were used to connect the dispensing end of the syringe to one end of the connecting Teflon tubing. The other end of the tubing was positioned to perfuse ethanol into the cuvette containing saturated glycine solution. The syringe with the connecting tubing was mounted onto a Harvard syringe pump to induce flow at µL levels into the glycine solution. The settings on the pump such as syringe ID, infuse rate and target volume for the required flow rate of 0.5 µL/min were set and flow was started at t=0. The schematic for the direct perfusion experimental set-up is shown in Figure 5-10. Teflon Tubing UV cuvette Syringe Pump Figure 5-10 Schematic for crystallization by direct perfusion of ethanol (0.5-10 µL/min). The sequence of phenomena in the cuvette overtime during the perfusion experiment was recorded by Leica MZ16 stereo microscope/camera system. A Leica CLS 150 XE light source 45 CHAPTER 5 EXPERIMENTAL was used for precision view. Nucleation of glycine crystals followed by growth with buildup of supersaturation due to perfusion of ethanol was observed. The entire volume took a total time of t = 33.3 hours to be dispensed at a rate of 0.5 µL/min into the glycine solution. Once the flow had ceased the formed crystals within the cuvette were harvested and filtered using a filter paper and dried under vacuum of 0.08MPa for 3-4 hours. The crystals were used for morphology analysis using Optical microscopy and Scanning Electron Microscopy (SEM). The crystals were placed on a clean glass slide and positioned right beneath the eye piece of a Leica stereo microscope within the field of view. The magnification was set to 7.1X at the eyepiece and still images of the sample were recorded. Optical microscopy images of crystals samples from the perfusion experiment are shown in Figure 5-11. Figure 5-11 Optical microscopy image of crystals collected from 0.5 µL/min (top) and 7 µL/min (bottom) perfusion sets (Scale-bar size 300µm). A few crystals from the sample set were used for image analysis using SEM. Optical microscopy in this case was less reliable for morphology study due to the smaller size of the 46 CHAPTER 5 EXPERIMENTAL crystals formed especially for higher perfusion rates. A precise difference in crystal morphology was observed while using SEM. Some of the SEM images of collected crystals are shown in Figure 5-12. (a) (b) Figure 5-12 SEM images of crystals for (a) 2 µL/min, (b) 5 µL/min. Having studied the external characteristics of the crystal such as morphology, habit and crystal size the next step was to characterize the polymorphic form of the crystal set. X-Ray Powder Diffraction (XRD) Analysis was chosen as an analysis method for characterization. The sample crystals for XRD were first ground into a fine powder using a porcelain mortar and pestle. The finely ground powder was then pressed into a small pellet using a metal press into the cavity of an aluminum die. The prepared die was placed in position within the XRD housing and the chamber was closed for x-ray generation and the resulting scattering was obtained as the XRD pattern for the sample set. As shown in Figure 5-13, the XRD pattern is a plot of the diffraction intensity against the grazing angle, 2θ. 47 CHAPTER 5 EXPERIMENTAL α γ Figure 5-13 XRD pattern overlay for two different flow rates of ethanol (3 and 7 µL/min). The above perfusion experiment followed by crystal characterization was performed for a list of perfusion rates of anti-solvent (ethanol) ranging from 0.5-10 µL/min. Multiple repeats of experiments was done to confirm the results under each flow regime. Figure 5-14 shows a 3D plot of XRD pattern for all the flow regimes to give a wholesome picture of the trend in the polymorphic outcome observed. The pattern shows α-glycine to be the dominant form of occurrence. However trace of other polymorphs is also evident. Figure 5-14 3D plot of XRD pattern for all perfusion rates of ethanol (0.5-10 µL/min). 48 CHAPTER 5 EXPERIMENTAL 5.3.1 Data Analysis for direct perfusion experiment The direct perfusion of ethanol carried out for a wide range of ethanol fractions from 10-80% vol where one of the volume fractions, 50% vol ethanol was chosen for direct perfusion. The experimental procedures being explained in the previous section, the results from optical microscopy, Scanning Electron Microscopy (SEM) and X-ray Powder Diffraction Analysis (XRD) are analyzed here. The images of crystal samples from optical microscopy and SEM for various perfusion rates of ethanol (0.5-10 µL/min) are shown in Figure 5-11 and 5-12 respectively. It appears that there exists a convincing deviation in the morphology of crystals that were grown under different perfusion rates. For example the crystal in Figure 5-12(a) has a bipyramidal shape whereas the crystal in Figure 5-12 (b) has a plate like morphology. It is to be noted that the crystal in Figure 5-12(a) was gown with ethanol perfusion rate of 2 µL/min which took a total time of t= 8.33 hours for the total volume of 1mL ethanol to be dispensed while the crystal in Figure 5-12(b) was grown with ethanol perfusion rate of 5 µL/min which took t= 3.33 hours for the total volume to be dispensed. A significant difference in the crystal morphologies is evident which conveys the message that not just the concentration of ethanol but also the perfusion rate plays a significant role in crystal morphology. The SEM image in Figure 5-12(a) is comparable to the α-glycine crystal grown from aqueous solution[111]. The bipyramidal crystal morphology displaying crystal planes like (011), (110), (120) and (020) can be seen as shown in Figure 5-15. 49 CHAPTER 5 EXPERIMENTAL (a) (b) Figure 5-15 Similarity in α-glycine morphology grown in (a) aqueous medium[111], (b) solvent mixture On the other hand the SEM image in Figure 5-12(b) is very much comparable to the α-glycine crystal morphology predicted by Bisker-Leib and Doherty[112]. Similar to the predicted morphology, the SEM image is rectangular in shape and shows a very well developed (010) face with (020) plate bounded by (110) and (011) faces. To have a firm understanding on the polymorphism of the formed crystals X-ray Powder Diffraction Analysis was done. Having known the signature peaks of α, β and γ glycine it can be confirmed that α-glycine is the dominant polymorphic form of occurrence in all the perfusion sets. A plot of the normalized integrated intensity indicative of the fraction of glycine polymorph formed against the perfusion rate shows the fraction of each polymorph (Figure 5-16). Although minimal occurrence of β and γ forms is observed, α-glycine turns out to be the dominant form with direct perfusion of ethanol in saturated glycine solution. 50 CHAPTER 5 EXPERIMENTAL Figure 5-16 Plot of polymorphic outcome of glycine for various ethanol perfusion rates (µL/min). At regular conditions and neutral pH, without the presence of any impurities or additives, the alpha[113] or the beta[114] form crystallizes spontaneously from solution. The alpha form is the metastable form,[115] and usually crystallizes as centrosymmetric bipyramids in a monoclinic space group(P21) at comparatively lower supersaturation[40, 58, 59]. The beta form is the unstable form and usually crystallizes as noncentrosymmetric, high aspect ratio habits such as needles in a monoclinic space group (P21/n), at higher supersaturation[5, 60]. 51 CHAPTER 5 5.4 EXPERIMENTAL Crystallization at Liquid Interface In order to study crystallization at liquid interfaces the following experimental set up was manually assembled as shown in Figure 5-17. The set up consists of a quartz cuvette which was air blown using an air gun. 1 mL of ethanol was drawn into a 1 cc Terumo syringe and ensured that it was devoid of any air bubbles within. 1mL of saturated glycine solution was drawn using a pipette and dispensed into the cuvette. Above the saturated glycine solution was added 1 mL of hexane whose density(654.8 kg/m3) is lower than that of the saturated glycine solution and hence it sits above the solution provided both the liquids being immiscible, results in formation of the necessary liquid-liquid interface. Fittings purchased from Upchurch Scientific were used to connect the dispensing end of the syringe to one end of the Teflon tubing. The other end of the tubing was positioned to perfuse ethanol into the hexane region in the cuvette. The syringe with the connecting tubing was mounted on a Harvard syringe pump. The settings on the pump such as syringe ID, infuse rate and target volume were set for the required rate of 0.5 µL/min. Figure 5-17 Schematic for interfacial crystallization with perfusion of ethanol (0.5-10 µL/min) The sequence of phenomena at the interface in the cuvette during the perfusion experiment was recorded by Leica MZ16 stereo microscope/camera system and Supereyes HD B003 52 CHAPTER 5 EXPERIMENTAL portable microscope/camera. A Leica CLS 150 XE light source was used for precision view. The eyepiece of the microscope was positioned exactly at the hexane-saturated glycine solution interface in order to record the phenomena in real time. Nucleation of glycine crystals followed by growth for increase in supersaturation with perfusion of ethanol was observed at the interface. Needle shaped crystals formed at the interface initially. With the buildup of supersaturation, several such needle like crystals form and group at the interface to form a crystal network which frames the liquid interface as shown in Figure 5-18. Figure 5-18 Glycine crystal network at the liquid interface with perfusion of ethanol at 10 µL/min (Scale-bar size 300µm) The crystal network at the interface rises with addition of ethanol, but when the volume of ethanol added reaches 300 µL the interfaces disengages from the network and rises above it. A few crystals form at the interface and drop to the bottom of the cuvette. The entire volume of ethanol (1 mL) took a total time of t = 33.3 hours to be dispensed at a rate of 0.5 µL/min. Once the flow had ceased the formed crystals within the cuvette were harvested and filtered using a filter paper and dried under vacuum of 0.08MPa for 3-4 hours to suffice collection of crystals in the morphology that they had acquired in the cuvette. The crystal samples were analyzed using optical microscopy and Scanning Electron Microscopy (SEM). The crystals were placed on a clean glass slide and positioned right beneath the eye piece of a Leica stereo 53 CHAPTER 5 EXPERIMENTAL microscope within the field of view. The magnification was set to 7.1X at the eyepiece and still images of the sample were recorded. Optical microscopy images of crystals obtained from 0.5 µL/min and 10 µL/min are shown in Figure 5-19. (a) Glycine crystals at the interface observed at the end of ethanol perfusion at 0.5 µL/min. (b) Glycine crystals collected from the 0.5 µL/min experimental set. (c) Glycine crystals observed at the end of ethanol perfusion at 10 µL/min. Figure 5-19 Optical micrograph of glycine crystals formed for different perfusion rates of ethanol (Scale-bar size 300µm). The crystals after morphology characterization were analyzed for their polymorphic form using XRD. The regular protocol for the sample preparation was followed (refer section 54 CHAPTER 5 EXPERIMENTAL 5.1.1). The prepared sample was placed in position within the XRD housing and the chamber was closed for x-ray generation and the resulting scattering was obtained as the XRD pattern for the sample set. The above perfusion experiment followed by crystal characterization was performed for a list of flow rates ranging from 0.5-10 µL/min. Multiple repeats of experiments was done to confirm the results under each flow regime. Figure 5-20 shows a 3D plot of XRD pattern for all the flow regimes to give a wholesome picture of the trend in the polymorphic outcome observed. Figure 5-20 3D plot of XRD pattern for all flow regimes of ethanol (0.5-10 µL/min) The XRD results for perfusion experiment with the interface show an interesting trend which greatly deviates from the XRD pattern of other experiments performed. From Figure 5-20 it can be seen that the dominating peak at 2θ=30º (corresponding to α-glycine) for ethanol flowrate of 0.5-3 µL/min (indigo and green) decreases tremendously when the ethanol perfusion rate shifts to 5-10 µL/min (red, blue and purple), during which the characteristic 55 CHAPTER 5 EXPERIMENTAL peak corresponding to γ-glycine (2θ=25º) becomes dominant at such flow rates. A better comparison of the trend is shown in Figure 5-21 where the peak position corresponding to αglycine has the highest intensity for the crystals collected from the 3 µL/min ethanol perfusion experimental set (Blue pattern). On the other hand, the peak position corresponding to γglycine has the highest intensity for the crystals collected from the 7 µL/min ethanol perfusion experimental set (Green pattern). Thus from the observed pattern, α-glycine is the dominating polymorphic form when the ethanol flowrate is maintained at 0.5-3 µL/min. When the ethanol flow rate is > 5µL/min, the dominant polymorphic form switches to γ-glycine. In either case, the β-form is also found in small fractions. α γ Figure 5-21 Shift in dominant peak position with increasing flow rate of ethanol from 3 to 7 µL/min. 56 CHAPTER 5 EXPERIMENTAL 5.4.1 Data Analysis for interfacial crystallization experiment The experimental procedure for the interfacial crystallization experiment was discussed in the previous section. The setup is similar to that of the direct perfusion experiments except for the presence of an immiscible aqueous – organic interface formed by saturated glycine solution and n-hexane. Studying crystallization at a liquid-liquid interface being a major aspect of this work, the crystal thus formed were studied and characterized for their shape, morphology, size and polymorphic form using optical microscopy, Scanning Electron Microscopy (SEM) and X-ray Powder Diffraction Analysis (XRD). The optical microscopy images of crystals formed for ethanol perfusion rate of 0.5 and 10 µL/min (Figure 5-19), reveal that the crystals collected from the lower perfusion rate experiments (0.5-3 µL/min) are regular in shape and are faceted. But the crystals collected form perfusion rates of > 5 µL/min deviate from regular crystal morphology. It can also be seen that the needle shaped crystals, tend to connect and form a network of crystals at the interface which remains intact until the end of the process. A measurement of position of the crystal network over time for a perfusion rate of 5µL/min clearly shows that the initially formed crystal network is stationary with just the level of the aqueous-organic interface rising with the increase in volume of ethanol (Figure 5-22). Figure 5-22 Glycine crystallization process for 5 µL/min ethanol perfusion (Scale-bar size 1.5mm) A few crystals which form at the interface tend to drop to the bottom of the cuvette. But as the interface rises above the crystal network they drop and settle on the network. It can also be observed that on addition of 100µL ethanol, the interface level drops as a result of the decline in bulk volume of glycine solution due to crystallization of the solute. The crystals thus formed were harvested and dried under vacuum for X-ray powder diffraction analysis to 57 CHAPTER 5 EXPERIMENTAL characterize their polymorphic form. The peak intensity patterns for two different perfusion rates in Figure 5-21 show that there is a shift in the dominant peak position with the change in the rate of perfusion. The XRD pattern in blue which is of the crystals collected from the 3µL/min perfusion set shows a dominant peak at 2θ=30º characteristic of α-glycine while the XRD pattern in green which is of the crystal collected from the 7µL/min perfusion set shows a dominant peak at 2θ=25º characteristic of γ-glycine. After multiple repeats of the experiments for each perfusion rate and their respective XRD patterns analyzed using the integrated intensity analysis, an estimate of each polymorph’s occurrence was obtained. A plot of the fraction of each polymorph obtained for the various perfusion rates of ethanol is shown in Figure 5-23. The plot shows that α-glycine is the dominant form for perfusion rates of 0.5-3µL/min. On increase in perfusion rate at and above 5µL/min, a switch in the dominant polymorphic form from α-glycine to γ-glycine can also be seen. The results a highly reproducible with each perfusion set being repeated thrice provided the error bars being included for each data point. Although such a switch in dominant polymorph is seen between α-glycine and γ-glycine, the least stable β-glycine is also present. Figure 5-23 Effect of ethanol perfusion rate on Glycine polymorphism 58 CHAPTER 5 5.5 EXPERIMENTAL Interfacial crystallization for zero ethanol flow rate The experimental set-up is demonstrated in Figure 5-24 below. 1ml of saturated glycine solution was contained in a quartz cuvette with 1ml of hexane above the glycine solution. 200 µL of Poly dimethyl siloxane (PDMS) was poured into a 1.5ml plastic vial and cured at room temperature over 12 hours. The bottom of the vial was cut using a scalpel in such a manner that the PDMS slab (about 2 mm thick) was exposed. 1ml of ethanol was transferred into vial above the PDMs slab and the content was capped. The vial was then positioned in such a way that the bottom of the PDMS slab was immersed into the hexane region of the cuvette. The entire set-up was then placed into a 20mL glass vial and tightly capped to prevent the evaporation of hexane. Ethanol could slowly diffuse through PDMS in about 6 hours. The entire set-up was left on shelf for about 12 hours and glycine crystals were harvested at the hexane-water interface afterwards. Figure 5-24Experimental Set-up for Diffusion-Controlled Crystallization Multiple repeats of experiments were done and glycine crystals collected from each set were characterized for their polymorphic form by XRD. Figure 5-25 is an overlay of the XRD patterns of the glycine crystals collected from multiple repeats of the above experiment. 59 CHAPTER 5 EXPERIMENTAL α γ Figure 5-25 XRD pattern overlay for multiple repeats of interfacial crystallization experiment for zero ethanol flow rate. The XRD patterns indicate that α-glycine is the dominant polymorphic form, while both βand γ-forms are kept at minimum. This observation agrees with the results obtained from the interfacial crystallization experiments for an ethanol perfusion rate < 3µl/min. The experimental set-up explores the lower extreme of slow anti-solvent addition rate, wherein diffusion is the major mode of ethanol transport. 60 CHAPTER 6 6. Discussion 6.1 Glycine crystal morphology DISCUSSION The crystal size and habit were characterized with optical microscopy and SEM for samples produced from different ethanol flow rates from 0.5 to 10µL/min. Typical images of the habits of glycine produced are presented in Figure 6-1. We use the term “bipyramidal” for the habits that displayed comparative growth in all three dimensions. The crystals typically ranged from 0.1 mm to 1 mm in size. The habits with comparative growth in two large dimensions are referred to as “plates”. The size of the plate-like crystals ranged from a few microns to tens of microns. The thickness was typically between ~10 to 100µm. The term “rod” is used for habits with considerable growth along one dimension and a low growth rate in all other dimensions. The typical width and length of the rods were found to be a few microns and tens of microns, respectively. The term “needle” is used for habits that have an extremely high growth in just one dimension. The thickness of needles was found to be similar to that of plates. Figure 6-1 Various morphologies of glycine observed (a)bipyramidal, (b) plate, (c) rod, (d) needle (Scale-bar size 300µm) The appearance of such varied habits may be attributed to the different levels of supersaturation generated due to variation in the ethanol perfusion rate. A low perfusion rate 61 CHAPTER 6 DISCUSSION means a low initial level of supersaturation for nucleation and hence we see crystal morphologies corresponding to that of α-glycine. But a higher perfusion rate implies a higher level of supersaturation initially for nucleation thus resulting in needle shaped crystals at the interface, characteristic of β-glycine. In addition, the two key steps of crystallization process, namely nucleation and growth, require a distinct set of optimum conditions for operation. For instance, to carry out well controlled growth process, the environment needs to be at a lower level of supersaturation for a longer time in order to observe detectable growth. This explains the observable differences in the crystal appearances where we see a comparative growth on all crystal facets and a selective growth of specific facets of glycine for the low and high ethanol flow rates respectively. 6.2 Dynamic observations of crystallization process As mentioned earlier, the interfacial crystallization process in the cuvette was observed and recorded using an optical microscope, from which a great amount of information was obtained. One of the most prominent observations is that, the first observable crystal for all perfusion rates always appears at the liquid-liquid interface. The glycine crystals emerge from the interface, grow bigger at the interface, and most of them stay at the position of the original interface while few of them drop to the bottom of the cuvette. This observation is shown in Figure 6-2(a) below. Figure 6-2(b) presents a zoomed-in view of crystals formed at the liquidliquid interface during the experiment. Interestingly, no observable crystals were identified elsewhere in the picture, except at the original liquid-liquid interface. 62 CHAPTER 6 DISCUSSION Figure 6-2 Formation of Glycine crystals during perfusion experiments (a) in cuvette at the liquid-liquid interface; (b)zoomed image of crystals at the interface; (c) in Cuvette without Hexane layer for an interface; (d) zoomed image of crystals in bulk (Scale-bar size 300µm) In comparison, the same experiments were repeated in a system without the hexane layer, and the result is demonstrated in Glycine crystals were observed to form in the bulk solution (Figure 6-2(c), (d)). The above observation can be explained by two possible reasons, which are nucleation induction time reduction and the local supersaturation elevation at the interface. Diao et al.[90] have studied the effect of surfaces on nucleation induction time and its influence on kinetics of nucleation. Nucleation induction time is defined as the time elapsed prior to formation of a detectable amount of crystalline phase [116]. The presence of a surface could dramatically reduce the kinetic barrier to nucleation and hence decrease the nucleation induction time, therefore nucleation takes place at the interface prior to other regions in the environment. Furthermore, crystallization is initiated at the interface, the concentration of glycine in cuvette would reduce and hence bulk nucleation could be inhibited. As a result of both effects, all glycine crystals were localized to the liquid-liquid interface. 6.3 Crystallization is localized at the liquid-liquid interface The formation of crystals at the liquid-liquid interface as seen from the interfacial experiments are explained by two possible reasons which are nucleation induction time reduction and the elevation of local supersaturation at the interface. Diao et al. [90] have studied the effect of surfaces on nucleation induction time and their influence on kinetics of 63 CHAPTER 6 DISCUSSION nucleation. Nucleation induction time is defined as the time elapsed prior to formation of a detectable amount of crystalline phase[116]. The presence of a surface could dramatically reduce the kinetic barrier for nucleation as a result of which the nucleation induction time might decrease. Hence we observe crystal formation at the interface prior to other positions. Furthermore, when crystallization is initiated at the interface, the bulk density of glycine in solution would reduce thereby inhibiting bulk nucleation. As a result of both these effects the crystals formed were localized to the liquid-liquid interface. Another possible explanation is that the concentration of ethanol at the liquid-liquid interface is much higher compared to the ethanol concentration in bulk glycine solution. To explain this, ethanol flow pattern for the interfacial experiments was investigating. Figure 6-3 shows the different flow patterns formed when ethanol mixed with Rhodamine B (0.2mM/ml) was perfused at flow rates from 0.5-10µL/min. All images shown here were taken at t = 3 minutes. It was observed that the color was much deeper at the hexane-water interface than in the bulk glycine solution, which indicates that ethanol was concentrated at the liquid-liquid interface. Considering that the density of ethanol is higher than hexane but lower than water, an ethanol rich zone might form at the interface. As a result of the high concentration of ethanol at the interface than in bulk, nucleation of glycine is likely to be initiated at the interface. Similarly, the initial nucleation at the interface reduces the bulk glycine concentration and thus prevents bulk crystallization. Hence crystallization is localized to the liquid-liquid interface only. Figure 6-3 Optical micrograph of flow patterns for ethanol + rhodamine B with varying perfusion rates in the interfacial crystallization system, (a)-(d) bright field; (e) fluorescence under UV. (Scale-bar size 1.5mm) 64 CHAPTER 6 6.4 DISCUSSION Disengagement of interface from crystal network Another interesting observation during the crystallization process was the formation of a crystal network at the interface. With ethanol being mutually miscible with water, the liquidliquid interface keeps rising throughout the entire experiment. Due to ethanol perfusion, crystals form and rise together with the interface. As they rise, the crystals grow and newly forming crystals add to the existing crystal volume at the interface as a result of which a crystal network is formed. Initially the liquid-liquid interface is overlapped by the crystal network, and as the interface level raises the crystal network rises with it as well. When approximately 300 µL of ethanol being perfused into the system, the liquid-liquid interface disengages from the crystal network and rises above the network afterwards. After the liquid interface disengages from the crystal network, few crystals still form at the interface. However, the number of crystals is too few to form a tight network similar to the existing network of crystals. The crystallization process was observed for different flow rates of ethanol and the exact point in time at which the liquid interface disengages from the crystal network was recorded. Interestingly, for all experiments in spite of the difference in perfusion rate, this disengagement phenomenon was observed when about 300µL of ethanol was added to the system. The above described process is demonstrated in Figure 5-22 for an ethanol perfusion rate of 5µL\min. The mass of glycine crystals that can form for the addition of 1 mL of ethanol to 1 mL of saturated glycine solution is shown here. Having known the solubility of glycine[117] for the various volume fractions of ethanol in water from the plot in Figure 6-4(a), the following is estimated. The solubility of α-glycine in 50%vol ethanol-water mixture is taken as basis considering the corresponding values of all the three forms of glycine are close(α-glycine: 0.022 g/g; β-glycine: 0.023 g/g; γ-glycine: 0.020 g/g). Since we use a saturated glycine solution (25g/100 mL water) for the experiments, 1mL of saturated glycine solution should have 0.25g of glycine in it. But for α-glycine, the saturation is at 0.234 g/g as per documented in the solubility data[117]. Thus we consider 1mL of saturated glycine solution to have 0.234 65 CHAPTER 6 DISCUSSION g of glycine in it. Hence for 50% vol ethanol-water mixture, the mass of glycine that would crystallize is 0.1746 g and the amount of glycine that remains dissolved in the mixture is 0.0594 g provided the total volume of ethanol is known. Therefore for the addition of 300µL ethanol to 1 mL saturated glycine solution, the solubility is 0.1139 g/g, the mass of glycine crystallized would be 0.093 g and the mass of glycine remaining in solution would be 0.1409 g. The amount of glycine that would crystallize for different volumes of ethanol addition from calculation is demonstrated in Figure 6-4(b) below. Details of calculation can be referred from Appendix B. Figure 6-4 Plots for (a) Glycine solubility in water-ethanol mixture (b) Amount of glycine crystallized for different volumes of ethanol addition From the observations and calculations, the phenomenon could be explained as follows. It is apparent from Figure 6-4(b) that the rate of crystallization of glycine is highest at the beginning which then slows down with further ethanol addition. With the total volume of ethanol added ≤ 300 µl (19.1 wt %), the rate of glycine crystal increase is almost linear, while the trend reaches a plateau when the volume of ethanol increases further. Taking into account that the volume of saturated glycine solution in the cuvette is consistent and the final volume of ethanol perfused is fixed, the rate at which crystal form is coherent with the rising liquidliquid interface for ethanol volume of ≤ 300µl. Thus the crystal network and the liquid-liquid interface overlap. However, with further addition of ethanol, the rate of crystal formation is slow and could not catch up with the rising interface, hence the disengagement of the liquid- 66 CHAPTER 6 DISCUSSION liquid interface from the crystal network. As a result, the liquid-liquid interface starts to detach from the crystal network for the same volume of ethanol added regardless of the different perfusion rates used in the interfacial crystallization experiments. 6.5 Dynamic crystal morphology change Besides the position of the crystal network, the change of crystal morphology throughout the experiment is also intriguing. For perfusion rate of anti-solvent (≥5µL/min), the initial observable crystals generated at the interface always possess a needle-like shape, while the morphology of these crystals changes to irregular shapes and eventually transform to irregular crystals with facets. The evolution of crystal morphology along the experiment is demonstrated in Figure 6-5. However, this phenomenon was not observed for the cases with perfusion rate of [...]... the anti-solvent across the liquid- liquid interface and can serve as a site for nucleation and growth of the solute In this work we study polymorphic nucleation and crystal growth in the vicinity of a liquid- liquid interface from an experimental perspective 1.1 Thesis objectives and layout The purpose of this thesis is to gain insight into the process of crystallization at liquid- liquid interfaces Glycine. .. investigates similar aspects of crystallization in the glycine- water-ethanol mixture at the vicinity of a liquid- liquid interface The liquid- liquid interfacial platform developed for this study presents a tool for gaining insight into the fundamentals of crystallization at liquid- liquid interface This thesis is organized in six chapters Chapter 2 discusses the process of crystallization, nucleation and... 2.2 Crystallization - a dual process Crystallization is considered as a dual or two step process The first step is nucleation, the birth of a stable crystal nucleus and the formation of a new solid phase Nucleation is followed by crystal growth in the second step 2.2.1 Nucleation The rate and the mechanism of crystal formation, can be affected by supersaturation, rate of supersaturation generation and... surfaces or interfaces and how they can influence crystal properties is explained Chapter 5 demonstrates the crystallization experimental setups in which batch crystallization and crystallization at liquidliquid interface is studied In addition the usage of optical microscopy for real time imaging and analysis, alongside Scanning Electron Microscopy (SEM) and X-Ray Powder Diffraction (XRD) Analysis techniques... factor ξ anisotropic factor ΔHf heat of fusion xs solubility Eslice horizontal bond energy between two adjacent crystal blocks Eer total crystallization or lattice energy kR rate constant of rough surface mechanism kMN rate constant of mononuclear growth model g shape factor γE edge surface tension kPN rate constant of polynuclear model kBS rate constant of birth and spread model kSN rate constant of surface... 2 Crystallization 2.1 The Advent of Crystallization CRYSTALLIZATION The process of crystallization is ubiquitous and has been utilized for thousands of years [6] The sheer applications of crystals in the 19th century and earlier were as precious stones for their fascinating properties: transparency and color, refractive index and optical dispersion, symmetry and facets [7] However, with the advent of. .. developments of the 20th century, crystallization has become an important process for numerous modern technologies, for a number of applications such as separation, concentration, purification and solidification Crystallization is utilized in the petrochemical industry for separation and purification of solids It is an important process in the specialty chemicals industry for manufacturing household products and... Glycine in water has been chosen as a model system while ethanol has been used as the anti-solvent and an aqueous-organic interface which constitutes of glycine- water for the 1 CHAPTER 1 INTRODUCTION aqueous region and hexane for the organic region serves as the liquid- liquid interface It was observed that glycine crystallizes in its least stable forms when precipitated from watermethanol/ethanol solutions... relative rates of growth and dissolution of the crystals of (R-S) alanine and γ-form of glycine indicate that in aqueous solutions the –c carboxylated end of the crystals grow and dissolve faster than the +c amino end 13 CHAPTER 2 CRYSTALLIZATION Figure 2-3 γ -glycine as viewed down the b-axis The capped face (0 3) exposes NH3+ while the flat face (00 ) exposes CO2- [36] Inspection of the packing arrangement... per unit of volume Equation 1-2 summarizes the effects of supersaturation, temperature and interfacial tension on the nucleation rate At lower supersaturation, the interfacial tension dominates and there is insufficient free energy to create a new surface As supersaturation increases, the nucleation rate increases exponentially, eventually reaching a maximum The nucleation theory predicts the transition ... Figure 4-4 Schematic of surfactant-monolayer-templated nucleation and crystallization[ 87] With lack of understanding of crystallization at liquid-liquid interfaces, the setup in addition has an added... comparison of relative rates of growth and dissolution of the crystals of (R-S) alanine and γ-form of glycine indicate that in aqueous solutions the –c carboxylated end of the crystals grow and dissolve... mode of crystallization could enable better observation and improved understanding of crystallization at liquid-liquid interfaces iv List of Tables Table 2-1 Properties affected by crystallization